Doxycycline is a broad-spectrum bacteriostatic agent (antibiotic) synthetically derived from a naturally occurring tetracycline produced by Streptomyces species bacteria known as oxytetracycline; it is a member of the tetracycline class of antibiotics (Kundu et al., 2015; Thillainayagam and Ramaiah, 2016). In the early 1960s Pfizer Inc. created and clinically developed doxycycline and began marketing it under the brand name Vibramycin® (Tan et al., 2011). Doxycycline has been approved by the Food and Drug Administration (FDA) for the prevention or treatment of specific conditions within each of the following categories: rickettsial infections, sexually transmitted infections, respiratory tract infections, bacterial infections, Lyme disease, ophthalmic infections, anthrax, acute intestinal amebiasis, traveler’s diarrhea, and severe acne (FDA, 2018a). It has also been investigated as a treatment for specific cancers because some studies suggest that doxycycline can inhibit cell proliferation and invasion and also induce apoptosis and block the gap phase (in which a cancerous cell grows and prepares to synthesize DNA) (Kundu et al., 2015).
While doxycycline can be used to treat a broad range of conditions, its use for malaria prophylaxis is the focus of this chapter. In 1992 Pfizer Inc. submitted a new drug application to FDA with an indication for malaria prophylaxis added to the product insert (Arguin and Magill, 2017); the indication was added in 1994 (Tan et al., 2011). The approved dosing regimen for malaria prophylaxis for adults is 100 mg per day 1–2 days before entering an endemic area, 100 mg per day while in the endemic area, and 100 mg daily for 28 days after leaving an endemic area. The FDA package insert states that this regimen is approved for up to 4 months (FDA, 2018a). Studies examining the long-term (≥4 months) use of doxycycline for malaria prophylaxis have offered mixed results on its tolerability.
Among Australian military personnel who were deployed to Cambodia (n = 600) for 12 months or Somalia (n = 900) for 4 months, doxycycline was well tolerated, and only 7 (0.6%) personnel in Cambodia and 15 (1.7%) personnel in Somalia discontinued the drug because of concurrent adverse events related to gastrointestinal symptoms or photosensitivity (Shanks et al., 1995a). However, a survey of 228 U.S. Peace Corps volunteers who had taken doxycycline on average for 19 months found that 45 (20%) of respondents reported changing medications due to severe concurrent adverse events, such as gastrointestinal symptoms, pruritic skin reactions, photosensitivity, and vaginal yeast infections (Korhonen et al., 2007).
Overall, studies suggest that adherence rates for doxycycline range from 70% to 84% when it is used for malaria prophylaxis, and adhering to the dosing regimen may be more challenging than adhering to dosing regimens for weekly prophylaxis medications (e.g., mefloquine, tafenoquine). One study found a 98% adherence for mefloquine but only an 81% adherence for doxycycline in U.S. troops in Somalia (Sánchez et al., 1993), and another study reported 70% adherence for a weekly prophylaxis regimen but only 50% adherence in people taking daily doxycycline (Watanasook et al., 1989). Furthermore, studies have shown that adherence to the doxycycline regimen decreases over time. In one study of Australian soldiers deployed to Cambodia, the adherence rate for taking doxycycline decreased from 60% at 2 months to 44% at 4 months (Shanks et al., 1995a). Studies have indicated that drugs with longer post-travel requirements tend to have worse adherence than drugs with shorter post-travel regimens (Overbosch et al., 2001); thus, doxycycline likely has poorer adherence during the post-travel period than a prophylaxis medication with a shorter post-travel regimen, such as atovaquone/proguanil (A/P).
Other studies have reported greater adherence in individuals prescribed once-daily prophylaxis than individuals prescribed once-weekly prophylaxis. Brisson and Brisson (2012) conducted an online survey in 1,200 military personnel deployed to Afghanistan between 2002 and 2012 to examine adherence rates for malaria prophylaxis in a combat zone. Of the 530 individuals who started the survey, 528 completed it (response rate of 44% to the initial survey distribution). The authors found that 3.6% of respondents were prescribed mefloquine, 90.1% received doxycycline, 0.9% received A/P, 0.2% received primaquine, and 4.4% were unsure which prophylactic drug they were prescribed. Of the individuals prescribed once-daily prophylaxis, 61% reported complete adherence; however, only 38% of individuals prescribed once-weekly prophylaxis (e.g., mefloquine) reported full adherence. Resistance of P. falciparum to doxycycline is not described, but breakthroughs in prophylaxis have been associated with inadequate doses, possibly inadequate serum levels, and poor adherence (Tan et al., 2011).
Beginning in the late 1980s prior to FDA’s approval for doxycycline use for malaria prophylaxis, the U.S. Army conducted several field and human challenge clinical trials demonstrating the efficacy of doxycycline as malaria prophylaxis (Arguin and Magill, 2017). A 2009 Department of Defense (DoD) memorandum
advised that doxycycline be used in preference to mefloquine in service members with a history of neurobehavioral disorders (DoD, 2009), and current DoD policy states that doxycycline is a first-line prophylactic agent for malaria. The U.S. military began using doxycycline as a primary agent for malaria prophylaxis after the anthrax attacks of September 2001, and it was used as a first-line agent for Operation Enduring Freedom (2001–2014) and operations Iraqi Freedom and New Dawn (2003–2011).1 Because doxycycline provides simultaneous protection from anthrax and malaria, it is attractive for use in military operations where these are potential threats.
Doxycycline reduces the incidence and severity of traveler’s diarrhea, and it has been shown to provide protection against traveler’s diarrhea in 60–85% of individuals, depending on enterotoxin-producing E. coli resistance (Sack et al., 1984). A study comparing diarrhea rates among British and Australian medical teams deployed to Iraq in support of the 1990–1991 Gulf War found that Australians who used doxycycline for malaria prophylaxis and who had instituted an enforced plate- and hand-washing routine experienced half the rate of diarrhea as the British (36% versus 69%, respectively) and that diarrhea illness, when it occurred, was both milder and of shorter duration (p < 0.001) (Rudland et al., 1996).
This chapter begins by describing the key changes that have been made to the doxycycline package insert and label since its approval for malaria prophylaxis in 1994, with particular emphasis on changes to the Contraindications, Warnings, and Precautions sections. This is followed by an overview of the pharmacokinetic properties of doxycycline. Known concurrent adverse events associated with use of doxycycline when used at the dose and interval as directed for malaria prophylaxis are summarized, followed by a presentation of detailed summaries and assessments of the seven identified epidemiologic studies that met the committee’s inclusion criteria and were able to contribute some information on persistent or latent health outcomes following the cessation of doxycycline. As in the other chapters, the epidemiologic studies are ordered by population: studies of military and veterans (U.S. followed by international forces), the U.S. Peace Corps, travelers, and endemic populations. Where available, studies of U.S. participants are presented first. A table that gives a high-level comparison of each of the seven epidemiologic studies that examined the use of doxycycline and that met the committee’s inclusion criteria is presented in Appendix C. Supplemental supporting evidence is then presented, including other identified studies of health outcomes in populations that used doxycycline for malaria prophylaxis but that did not meet the committee’s inclusion criteria regarding timing of follow-up. This is followed by case reports of persistent adverse events and, next, information about the adverse events of doxycycline use in specific groups, such as pregnant women and those
1 Personal communication to the committee, COL Andrew Wiesen, M.D., M.P.H., Director, Preventive Medicine, Health Readiness Policy and Oversight, Office of the Assistant Secretary of Defense (Health Affairs), April 16, 2019.
with chronic health conditions. After the primary and supplemental evidence in humans is presented, supporting literature from experimental animal and in vitro studies is then summarized. The chapter ends with a synthesis of all of the evidence presented, followed by the inferences and conclusions that can be made.
There have been numerous trade-name and generic formulations of doxycycline hyclate marketed in the United States. The FDA website offers a webpage for each formulation; each page lists the package insert updates, but downloadable package inserts are not available for all listed updates, and in older drugs this can mean that many years of updates are unavailable. This is also true of drugs that are currently on the market. For drugs that have been discontinued, often no downloadable package inserts are available. The oldest available package insert for doxycycline was dated 2005 (FDA, 2005a). Package inserts are listed on the webpage with an action date, but the date provided in the downloaded package insert document may occasionally differ from the action date posted on the webpage (e.g., a downloaded document listed as the 1989 mefloquine package insert was a July 2002 revision). Occasionally a downloaded document contained no date (e.g., the template’s “month/year” placeholder is not filled). Because package inserts dated prior to 2005 were not available, it could not always be determined whether a formulation had ever been indicated for malaria prophylaxis or, if it was, when the indication was added. For example, the Doryx® capsule (discontinued) was approved in 1985, but the earliest package insert available for download is 2005 (FDA, 2005b), 11 years after doxycycline was approved for malaria prophylaxis.
The design and formatting of package inserts has changed for individual drugs over time, so comparisons could not always be easily made. Moreover, there continue to be differences in information sourcing and section-labeling conventions among even the most recent package inserts, although they appear to provide the same information. For example, the Adverse Reactions section for the 2018 Doryx® tablet divides its content between Clinical Trial Experience and Postmarketing Experience, while the Adverse Reactions section for 2018 Vibramycin® capsules does not cite an information source, yet matches the Postmarketing Experience of the 2018 Doryx® tablet insert. Section headings and organization may differ also; in the 2018 package inserts, Doryx® contains a single Warnings and Precautions section (FDA, 2018a); the 2018 Vibramycin® insert has two separate sections but does not indicate the differences between the two levels of guidance (FDA, 2018c). The following text contains information related to the use of doxycycline for several symptoms, illnesses, or disorders (all its approved indications) and is not limited to the use of doxycycline for malaria prophylaxis unless otherwise stated.
Contraindications, Warnings, and Precautions
Doxycycline is contraindicated in individuals who have a known hypersensitivity to tetracyclines. The Warnings and Precautions sections of package inserts for doxycycline alert users to a number of risks: permanent tooth discoloration and enamel hypoplasia during tooth development in a child if taken in the last half of the mother’s pregnancy; Clostridium difficile–associated diarrhea and related morbidity and mortality; photosensitivity; potential overgrowth of nonsusceptible organisms, including fungi; severe skin reactions (exfoliative dermatitis, erythema multiforme, Stevens-Johnson syndrome, toxic epidermal necrolysis, and drug reaction with eosinophilia and systemic symptoms); intracranial hypertension; possible toxic effects on the developing fetus (often related to retardation of skeletal development) owing to the development of drug-resistant bacteria; the drug crossing the placenta; an increase in BUN2 due to anti-anabolic action; incomplete suppression of the asexual blood stages of malaria Plasmodium parasites; and an inability to suppress P. falciparum’s sexual blood stage, which allows person-to-mosquito transmission of infection. Periodic laboratory evaluation of organ systems, including hematopoietic, renal, and hepatic studies, should be made when using the drug long term (FDA, 2018a).
In individuals taking oral blood thinners, tetracyclines have been shown to intensify the anticoagulant effect of these medications by interfering with the use of prothrombin and reducing vitamin K production by intestinal bacteria. In a study conducted by Penning-van Beest et al. (2008), the investigators examined the anticoagulant–doxycycline interaction by analyzing the PHARMO Record Linkage System where patients were followed through the end of coumarin treatment, and they found a 2.5-fold increased risk for increased bleeding episodes in participants who were concurrently using doxycycline and acenocoumarol or phenprocoumon; these findings have been confirmed in other studies (Hasan, 2007).
Individuals using digoxin and oral antibiotics concomitantly with doxycycline may experience increased serum digoxin concentrations. These increased concentrations are a consequence of altered gut flora and reduced conversion of digoxin to inactive metabolites. The half-life of doxycycline is believed to be reduced when barbiturates, carbamazepine, and phenytoin induce microsomal enzyme activity. Any individuals receiving the oral typhoid vaccine are advised by most experts to not use doxycycline within the 24 hours immediately following the vaccine, as the vaccine effectiveness may be reduced (Tan et al., 2011).
The FDA label also states that “concurrent use of tetracycline may render oral contraceptives less effective” (FDA, 2018a); however, some studies have shown
2 BUN = blood urea nitrogen test; used to determine kidney function.
no significant association between the use of doxycycline and reduced efficacy of oral contraceptives (Dickinson et al., 2001). One study that examined the effect of doxycycline on another form of contraceptive, the subcutaneous implant, indicates that doxycycline may affect the pharmacodynamics of the levonorgestrel released from the implant; however, the results were inconclusive (Zhao et al., 2009). The use of additional contraceptive methods while taking doxycycline is advised.
Changes to the Doxycycline Package Insert Over Time
The committee established that Doryx® (tablet, capsule, and MPC formulations), Vibramycin®, Vibra-Tabs®, Acticlate® (tablet and capsule), and doxycycline hyclate are or had been indicated for malaria prophylaxis. The 2005 package inserts for the three Doryx® formulations (FDA, 2005a,b) showed no meaningful differences, nor did those for the 2007 Vibramycin® and Vibra-Tabs® formulations (FDA, 2007), nor the 2014 Acticlate® tablet (FDA, 2014) and 2016 Acticlate® capsule (FDA, 2016) formulations. Thus, the committee compared the earliest available package insert with the latest available package insert each for Doryx® tablets (FDA, 2005a, 2018a,b), Vibramycin® capsules (FDA, 2007, 2018c), and Acticlate® tablets (FDA, 2014, 2017) and summarized the major adverse-event-related updates. It is difficult to estimate how many adverse-reaction-related updates have been issued since 2005 to package inserts for doxycycline used for malaria prophylaxis. For example, during a specific period during which more than one drug formulation was being marketed for this indication, the number of label or package insert updates varied among the drugs, and it was unclear specifically which ones included changes to adverse events.
Between 2005 and 2018 several adverse reactions were added to the Warnings (or Warnings and Precautions) section for each of the formulations. One of the added adverse reactions was intracranial hypertension, in which clinical manifestations were stated to include headache, blurred vision, diplopia, vision loss, and papilledema via fundoscopy. The risk of intracranial hypertension is stated to be increased in women of childbearing age who are overweight or have a history of intracranial hypertension and in those using isotretinoin concomitantly. The warning also states that because intracranial pressure can remain elevated for weeks after drug cessation, patients should be monitored until they stabilize. Other adverse reactions added since 2005 were Stevens-Johnson syndrome, toxic epidermal necrolysis, erythema multiforme, and drug reaction with eosinophilia and systemic symptoms. The warning for Clostridium difficile‒associated diarrhea was expanded to say that this should be considered in all patients with diarrhea after antibiotic use as it can occur more than 2 months after drug cessation and can cause increased morbidity and mortality since these infections can be refractory to antimicrobial therapy and may require colectomy. Additions to other sections of the package inserts included the alert that patients can develop watery and bloody stools (with or without stomach cramps
and fever) as late as 2 or more months after antibiotic cessation, in which case patients should immediately contact a physician; reports of pancreatitis, exfoliative dermatitis, and discoloration (reversible) of adult teeth; and the advice that drugs should be taken with adequate amounts of fluid to reduce the risk of esophageal irritation and ulceration.
Compared with tetracycline, doxycycline has a longer half-life, better absorption, and a better safety profile (Shapiro et al., 1997; Thillainayagamam and Ramaiah, 2016). An oral dose of 100–200 mg of doxycycline is almost completely absorbed in the small bowel and is detectable in the blood 15–30 minutes after administration (Tan et al., 2011). Doxycycline is highly protein bound (93%), has a small volume of distribution (0.7 L/kg), and achieves relatively high blood levels (Schlagenhauf et al., 2019). Following a 200 mg oral dose of doxycycline, peak concentrations of about 2.6 μg/mL are reached at approximately 2 hours, but this may vary as gastrointestinal absorption rates differ among individuals. Doxycycline is readily transported across cell membranes, resulting in widespread distribution in body tissues and fluids. It localizes in the bone marrow, liver, and spleen; crosses the placenta; and is excreted in breast milk. Doxycycline and other tetracyclines form tetracycline-calcium orthophosphate complexes in sites of calcification such as developing teeth and bone, which may lead to permanent discoloration. The bioavailability of the monohydrate free base and of the hydrochloride salt (hyclate) forms of doxycycline has been shown to be equivalent (Tan et al., 2011).
Studies have shown that when medications composed of divalent or trivalent cations (such as antacids, laxatives, and oral iron preparations) are taken simultaneously with doxycycline, the absorption of doxycycline is decreased. Other types of medications that decrease the absorption of tetracyclines include antidiarrheal agents containing kaolin, pectin, or bismuth subsalicylate; these should be taken a few hours before ingesting doxycycline (Tan et al., 2011). Milk decreases the absorption of tetracyclines because of chelation of the calcium in the milk by the tetracyclines; however, the magnitude of this decrease varies between different tetracycline preparations, and the data for doxycycline are limited. According to FDA, the absorption of doxycycline “is not markedly influenced by simultaneous ingestion of food and milk,” despite the reduced absorption observed with other tetracyclines, and taking doxycycline with food is recommended to prevent concurrent adverse gastrointestinal events.
Unlike the case with other tetracyclines, the excretion of doxycycline occurs primarily by the gastrointestinal tract and to a much lesser extent by the kidneys. The serum half-life of doxycycline (15–25 hours) is not affected by impaired renal function or hemodialysis, and in patients with renal failure all excretion of doxycycline occurs by the gastrointestinal route. There are limited to no data on
sex, age, body weight, or race differences in the pharmacokinetics of doxycycline (Tan et al., 2011). One small study of healthy adult (aged 18–33 years) Vietnamese male (n = 14) and female (n = 14) volunteers found no differences in the pharmacokinetics of doxycycline by sex (Binh et al., 2009).
The following section contains a summary of the known concurrent adverse events associated with the use of doxycycline. Epidemiologic studies of persistent adverse events in which information was presented regarding the adverse events occurring at least 28 days post-doxycycline-cessation are then summarized, with the emphasis on reported results of persistent adverse events associated with the use of doxycycline, including results of studies in which other antimalarial drugs were used as a comparison group.
Concurrent Adverse Events
The FDA package insert for doxycycline uses information from trials that used doxycycline at dosages or for purposes other than for malaria prophylaxis (e.g., urogenital Chlamydia trachomatis infection), which are considered outside of the committee’s charge. All labels provide a lengthy list of adverse reactions that appear to be based on postmarketing experience, but they are not quantified, nor is the type of use of doxycycline always specified. The committee chose to use published information on concurrent adverse events from the Cochrane Database of Systematic Reviews and other reviews in which doxycycline was used for malaria prophylaxis.
The most commonly reported adverse events associated with the use of doxycycline are gastrointestinal symptoms and photosensitivity. Gastrointestinal symptoms are typically mild to moderate and include nausea (4–33%), abdominal pain (12–33%), vomiting (4–8%), and diarrhea (6–7.5%) (Tan et al., 2011). Studies show that nausea is more likely to occur if doxycycline is taken without food (Ohrt et al., 1997; Shanks et al., 1995b). Esophageal ulcers have also been reported and are more common in individuals who take doxycycline on an empty stomach or without liquid or who lie down within an hour after ingestion (Bott et al., 1987; Carlborg et al., 1983), but this effect was not specific to its use as malaria prophylaxis. In individuals who have taken doxycycline, exposing skin to sunlight may result in an erythematous rash, which has been reported in 7.3–21.2% of individuals, depending on the population (Rieckman et al., 1993; Sánchez et al., 1993; Wallace, 1996). Individuals with lighter complexions may be more prone to photosensitivity while taking doxycycline; these individuals should remain out of the sun or use appropriate protective measures if sun exposure cannot be avoided (Smith et al., 1995). Other adverse events that have been reported in association
with tetracyclines but they are less common or have not been reported during doxycycline use. These include onycholysis, benign intracranial hypertension, skin hyper-pigmentation, postinflammatory elastolysis, tooth discoloration, vertigo, ataxia, Clostridium difficile diarrhea, visual disturbances, and phlebitis (Klein and Cunha, 1995; Tan et al., 2011). Although Clostridium difficile infection is listed as an adverse event of doxycycline, current evidence suggests that doxycycline may actually provide protection against such infections (Tariq et al., 2018; Turner et al., 2014).
Tetracyclines are believed to suppress vaginal bacterial flora, resulting in an overgrowth of Candida albicans, and may enhance the virulence factors associated with the bacteria; the mechanism by which this occurs is unknown. Vaginitis has been reported with the use of doxycycline when taken for malaria prophylaxis, but estimates of the incidence are limited because vaginitis is often included under the category of non-specific skin reactions, and many of the early studies were conducted using populations of male military personnel. Women who are predisposed to or have a past history of candida vulvovaginitis or those taking oral contraceptives should consider carrying a self-treatment course of antifungals while taking doxycycline (Tan et al., 2011).
Tickell-Painter et al. (2017) performed a Cochrane systematic review in which adverse events were prespecified to include these disorders: psychiatric (abnormal dreams, insomnia, anxiety, depression, psychosis); nervous system (dizziness, headaches); ear and labyrinth (vertigo); eye (visual impairment); gastrointestinal (nausea, vomiting, abdominal pain, diarrhea, dyspepsia); and skin and subcutaneous tissue (pruritus, photosensitivity, vaginal candida). The purpose of the assessment was to summarize the efficacy and safety of mefloquine for malaria prophylaxis in adult, children, and pregnant women travelers compared with other antimalarials (including doxycycline), placebo, or no treatment. The dosages of mefloquine varied, as did the methods of collecting adverse event data. Therefore, the identified studies in this review were only those in which doxycycline was used as a comparator to mefloquine. The authors applied categories of certainty to the results based on the five GRADE considerations (risk of bias, consistency of effect, imprecision, indirectness, and publication bias) (Higgins et al., 2019). The committee recalculated the effect estimates presented below to directly compare doxycycline with mefloquine (instead of mefloquine with doxycycline).
When analyses were performed to compare doxycycline with mefloquine (4 trials totaling 1,317 participants and 20 cohort studies totaling 435,209 participants), no difference in the incidence of serious adverse events was found (RR = 0.65, 95%CI 0.10–4.35). Regarding neurologic adverse events, no differences were found for headache (RR = 0.83, 95%CI 0.34–2.90; 5 cohort studies, 3,322 participants) or dizziness (RR = 0.28, 95%CI 0.07–1.14; 5 cohort studies, 2,633
participants) when mefloquine users were compared with doxycycline users. However, for psychiatric adverse events reported in the cohort studies, doxycycline users were statistically significantly less likely than mefloquine users to report abnormal dreams, insomnia, anxiety, and depressed mood, although the pooled effect estimates were very imprecise. Whereas there were 15 episodes of abnormal thoughts and perceptions with mefloquine, no episodes were reported for doxycycline users in the cohort studies.
The 10 serious adverse events reported among doxycycline users were due to gastrointestinal disturbance (n = 6), anemia (n = 1), photosensitivity (n = 1), esophagitis (n = 1), and cough (n = 1). Among the included trials and cohort studies, there was no statistical difference in the number of discontinuations due to adverse events between mefloquine and doxycycline users. In the cohort studies reporting adverse events, doxycycline users were statistically significantly more likely to report nausea (RR = 2.70, 95%CI 2.22–3.33; 5 cohort studies, 2,683 participants), vomiting (RR = 5.55, 95%CI 3.70–8.33; 4 cohort studies, 5,071 participants), and diarrhea (RR = 3.57, 95%CI 1.37–9.09; 5 cohort studies, 5,104 participants), but in the single trial of military personnel that reported adverse events, no differences were demonstrated for these adverse gastrointestinal events.
Other symptoms were also included when available. In cohort studies reporting adverse events, photosensitivity (RR = 12.5, 95%CI 9.09–20.0; 2 cohort studies, 1,875 participants) and vaginal yeast infection in female participants (RR = 10.0, 95%CI 6.25–16.67; 1 cohort study, 1,761 participants) were more common in doxycycline users than mefloquine users. Based on two cohort studies that examined visual impairment, this adverse event was statistically significantly less commonly reported among doxycycline users than mefloquine users (RR = 0.42, 95%CI 0.25–0.71; 1,875 participants). A range of other adverse events were reported in individual cohort studies, including alopecia (hair loss), asthenia (physical weakness), balance disorder, decreased appetite, fatigue, hypoaesthesia (numbness), mouth ulcers, palpitations, and tinnitus, but for all these outcomes, there was either no difference or higher risks among mefloquine users. Risk of malaise was found to be statistically significantly higher among doxycycline users compared with mefloquine users.
Post-Cessation Adverse Events
A total of 5,672 abstracts or titles were identified by the committee’s literature search for doxycycline. After an initial evaluation of the types of citations captured, the committee determined that a large portion of the literature contained information related to alternative uses of doxycycline (e.g., acne, bacterial infections). Additional search terms related to prophylaxis and malaria were added, which reduced the number of captured citations to 2,406 titles or abstracts. After screening, 568 abstracts and titles remained, and the full text for each was retrieved and reviewed to determine whether it met the committee’s inclusion criteria, as
defined in Chapter 3. The committee reviewed each article and identified seven epidemiologic studies that included some mention of adverse events that occurred ≥28 days post-cessation of doxycycline (Andersen et al., 1998; Eick-Cost et al., 2017; Lee et al., 2013; Meier et al., 2004; Schneiderman et al., 2018; Schwartz and Regev-Yochay, 1999; Tan et al., 2017), and these are summarized next. A table that gives a high-level comparison (study design, population, exposure groups, and outcomes examined by body system) of each of these seven epidemiologic studies is presented in Appendix C.
Military and Veterans
Using DoD administrative databases, Eick-Cost et al. (2017) performed a retrospective cohort study among 367,840 active-duty service members who filled at least one prescription for an antimalarial drug between 2008 and 2013: 36,538 were prescribed mefloquine, 318,421 doxycycline, and 12,881 A/P. The primary study objective was to assess and compare the risk of incident and recurrent International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM)-coded neurologic and psychiatric outcomes (adjustment disorder, anxiety disorder, depressive disorder, posttraumatic stress disorder (PTSD), psychoses, suicide ideation, paranoia, confusion, tinnitus, vertigo, convulsions, hallucinations, insomnia, and death from suicide) that were reported at medical care visits during concurrent use plus 365 days after the end of the prescription for mefloquine, doxycycline, or A/P. Although the authors did not report results for the period of ≥28 days post-cessation of antimalarial drug use, they stated that they performed several sensitivity analyses, including one in which the risk period was restricted to 30 days post-prescription. The results of that analysis were summarized in the text as, “However, none of these analyses significantly changed the results of the study and are therefore not reported” (p. 161). This statement implies (but does not show directly) that similar findings to those reported would be seen if the data were restricted to the period of relevance to the committee’s definition of persistence (i.e., ≥28 days after cessation of exposure). The committee was unsure how to interpret the statement that the results did not change significantly (statistical significance, precision of effect estimates, number of diagnoses, etc.), but given that the authors performed sensitivity analyses and that the study had a number of methodologic strengths, including a strong measurement of relevant outcomes conducted in the target population, the committee chose to include the study, despite the ambiguity in the language. If an individual had multiple prescriptions over the follow-up period, the risk periods were merged. Doxycycline and A/P prescriptions were excluded if the service member had previously or concurrently received mefloquine. Mefloquine risk periods were censored if an individual received a prescription for a different antimalarial. Analyses were stratified by deployment and psychiatric history. Models were adjusted for age, sex, service, grade, and year of prescription start; analyses of deployed service members also
controlled for location and combat exposure. A majority of the doxycycline recipients had served in the Army (69%), many were junior enlisted (48%), and a large percentage had had their prescriptions filled after 2010 (78%). Among the deployed service members, more individuals who had received doxycycline than who had received the other antimalarial drugs reported combat exposure (43%, compared with 29% for mefloquine and 21% for A/P).
With few exceptions, the adjusted incident rates were higher among the deployed than among the nondeployed for doxycycline as well as for the other antimalarial drugs that were considered. Effect estimates of the neurologic and psychiatric outcomes for mefloquine and A/P are reported in the relevant chapters. For doxycycline users, the highest incident rates in both the deployed and nondeployed were for adjustment disorder (56.92 versus 44.35 per 1,000 person-years, respectively), insomnia (27.53 versus 22.46 per 1,000 person-years, respectively), and anxiety disorder (23.53 versus 18.47 per 1,000 person-years, respectively). Incident tinnitus (18.25 versus 15.17 per 1,000 person-years, respectively), depressive disorder (18.59 versus 18.24 per 1,000 person-years, respectively), suicide ideation (4.43 versus 4.23 per 1,000 person-years, respectively), and hallucinations (0.83 versus 0.70 per 1,000 person-years) were also higher among the deployed group. On the other hand, the incidence of vertigo (14.85 versus 15.75 per 1,000 person-years, respectively), convulsions (1.67 versus 2.16 per 1,000 person-years, respectively), paranoia (0.09 versus 0.13 per 1,000 person-years, respectively), death from suicide (0.03 versus 0.05 per 1,000 person-years, respectively), and confusion (0.03 versus 0.05 per 1,000 person-years, respectively) were higher among the nondeployed than the deployed group. Among those prescribed doxycycline, the incidence rate of PTSD was 15.55 per 1,000 person-years in the deployed group and 9.06 per 1,000 person-years in the nondeployed group. When adjusted incidence rate ratios (IRRs) were calculated comparing doxycycline to mefloquine by deployment status, the only statistically significant difference among the deployed between the two drugs was for anxiety disorder (IRR = 0.89, 95%CI 0.81–0.99). When doxycycline and mefloquine users among the nondeployed were compared, adjustment disorder (IRR = 1.44, 95%CI 1.25–1.67), insomnia (IRR = 1.49, 95%CI 1.23–1.79), anxiety disorder (IRR = 1.43, 95%CI 1.16–1.75), depressive disorder (IRR = 1.47, 95%CI 1.19–1.82), vertigo (IRR = 1.92, 95%CI 1.14–3.23), and PTSD (IRR = 1.45, 95%CI 1.10–1.92) all showed a statistically significant higher risk for doxycycline users. No other statistically significant differences were seen for the other outcomes examined, including psychosis, suicide ideation, and death by suicide. A subsequent analysis restricted the population to those individuals who were receiving their first mefloquine or doxycycline prescription and included individuals with a prior history of a neurologic or psychiatric diagnosis. The incidence rates and IRRs for each neurologic or psychiatric outcome were compared, stratified by those with and without a prior neurologic or psychiatric diagnosis. A diagnosis of PTSD was recorded for 2,671 (0.8%) of individuals in the doxycycline group and 131 (0.4%) of individuals in
the mefloquine group in the 365 days prior to their first antimalarial prescription. For both the doxycycline and the mefloquine groups, individuals with a neuropsychiatric diagnosis in the year preceding the prescription had statistically significantly elevated risks for a subsequent diagnosis of the same condition for all conditions reported (adjustment disorder, anxiety, insomnia, depressive disorder, PTSD, tinnitus, vertigo, and convulsions) compared with individuals without a diagnosis in the prior year. However, when the IRRs were used to compare doxycycline and mefloquine users within strata of those with and without prior neuropsychiatric diagnoses, there were no statistically significant differences between doxycycline and mefloquine for any of the conditions, including PTSD (bootstrap RRR = 0.88, 95%CI 0.61–1.28).
The committee found this study to be well designed; important factors that increased the study’s quality were its large size, its use of an administrative data source that provided some degree of objectivity, and its careful consideration of potential confounders including deployment and combat exposure. Because neurologic and psychiatric diagnoses occurring during current and recent use were analyzed together without distinguishing between events that occurred within 28 days of antimalarial use and those that occurred ≥28 days post-cessation, the study provides no quantitative information regarding the persistence of most events other than the comment in the text that the results did not change when restricted to the post-cessation period. The use of administrative data provided a standard, consistent method to capture filled prescriptions and medical diagnoses through the use of ICD-9-CM codes. However, filled prescriptions do not equate to adherence to the drug regimens. Moreover, if the antimalarials were provided to entire units as part of force health protection measures, the use of these drugs would not be coded in individual records. Whereas the use of medical diagnoses is likely to be more reliable for the identification of adverse events than self-report, the data are dependent on the accuracy of the coding, and there was no validation of the diagnoses recorded in the administrative databases, and symptoms or events that did not result in a medical visit or diagnosis would have been missed. For PTSD diagnoses, there was no information concerning when the index trauma occurred. Although the authors report higher risks for several outcomes among doxycycline users compared with mefloquine users, particularly among the nondeployed, the IRRs are not adjusted for history of neurologic or psychiatric outcomes. Given the evidence that such a history is a very strong predictor of subsequent events (as shown in table 7 of the article) and clear evidence that those with such a history were preferentially prescribed doxycycline rather than mefloquine (as shown in table 6 of the article), it is reasonable to presume that the findings of higher risk among doxycycline users in this study results from confounding by neuropsychiatric history.
Schneiderman et al. (2018) conducted a retrospective observational analysis of self-reported adverse events associated with use of antimalarial drugs in a cohort of U.S. veterans who had responded to the 2009–2011 National Health Study for
a New Generation of U.S. Veterans (referred to as the “NewGen Study”). The NewGen Study is a population-based survey that sampled 30,000 veterans who had been deployed to Iraq or Afghanistan between 2001 and 2008 and 30,000 nondeployed veterans who had served during the same time period. It included a 20% oversampling of women. The survey was conducted using mail, telephone, and web-based collection, yielding a response rate of only 34.3%; while the response rate was low, the respondents nonetheless constitute a large population. For this particular analysis, 19,487 participants were included who had self-reported their history of antimalarial medication use, and this use was grouped for analysis by drug (mefloquine, chloroquine, doxycycline, primaquine, mefloquine in combination with other drugs, other antimalarials, and not specified) or no antimalarial use. Health outcomes were self-reported using standardized instruments: the Medical Outcomes Study 12-item short form (SF-12) for general health status, the PTSD Checklist-Civilian version (PCL-C), and the Patient Health Questionnaire (PHQ). These instruments yielded scores that were dichotomized for analysis on composite physical health, composite mental health (above or below the U.S. mean), PTSD (above or below screening cutoff), thoughts of death or self-harm, other anxiety disorders, and major depression. Potential confounders included in the multivariable analysis were branch of service, sex, age, education, race/ethnicity, household income, employment status, marital status, and self-reported exposure to combat. Responses were weighted to account for survey non-response. Most veterans reported no antimalarial drug exposures (61.4%, n = 11,100), and these served as the referent group. When stratified by deployment status, among the deployed (n = 12,456), of those who reported the use of an antimalarial drug (n = 6,650), 1,315 (weighted 20.5%) veterans reported using only doxycycline, and 425 (weighted 6.0%) reported using mefloquine and another antimalarial, which may have included doxycycline. Among the nondeployed (n = 7,031), 1,737 (weighted 20.8%) reported using an antimalarial drug, and of this group, 141 (weighted 8.8%) reported the use of doxycycline alone, and 52 (weighted 2.8%) used mefloquine and another antimalarial, which may have included doxycycline. Because it is not clear how many people in the mefloquine-plus-another-antimalarial group may have also used doxycycline, the results of that group are not included in the committee’s assessment. The deployed doxycycline users reported increased frequencies of mental health diagnoses compared with nondeployed doxycycline users: PTSD (17.9% versus 11.1%), other anxiety disorders (11.2% versus 7.1%), major depression (9.8% versus 9.1%), and thoughts of death or self-harm (10.9% versus 10.5%), but no statistical comparisons were presented. In the adjusted logistic regression models with all covariates considered (including demographics, deployment, and combat exposure), the use of doxycycline was not associated with any of the adverse events compared with nonuse of antimalarial drugs: composite mental health score (OR = 0.96, 95%CI 0.83–1.09), composite physical health score (OR = 0.91, 95%CI 0.79–1.04), PTSD (OR = 0.96, 95%CI 0.79–1.15), thoughts of death or self-harm (OR = 0.87, 95%CI
0.69–1.09), other anxiety (OR = 0.85, 95%CI 0.67–1.07), and major depression (OR = 0.84, 95%CI 0.66–1.06). Results were similar to those of other antimalarials for analyses restricted to the deployed subset of veterans. An additional analysis was performed of the six health indicators or outcomes stratified by antimalarial exposure and a four-level measure of combat exposure intensity. The weighted prevalence estimates seem to indicate an increasing prevalence of disorders with increasing combat exposure intensity, but it is challenging to interpret the results or to compare across antimalarial exposures, given the small numbers in some cells and the lack of confidence intervals or hypothesis tests.
This analysis of the NewGen survey is highly relevant to the question of whether there are adverse events of doxycycline use that persist after the cessation of that use. The study is large enough to generate moderately precise measures of association, the specific drugs were assessed, the outcomes were based on standardized instruments (although not face-to-face diagnostic interviews), important covariates of deployment and combat exposure were considered in addition to demographics and other military characteristics, and the data were appropriately analyzed. It is noteworthy that adjusting for combat exposure consistently reduced the measures of association for adverse psychiatric events related to doxycycline use. Although the time period of drug use and the timing of adverse events were not directly addressed, given that the members of the populations were all veterans who had served between 2001 and 2008 and that the survey was not administered until 2009–2011, it is reasonable to assume that antimalarial drug use had ceased some time before the survey was conducted. Nonetheless, the study could not address explicitly the health experiences during use and in specific time intervals following the cessation of use. There are a number of methodologic concerns that limit the strength of this study’s findings. The low response rate of 34% raises the concern of non-response bias, but responses were weighted to account for non-response. Selective participation by both antimalarial drug use history and health status would be required to introduce bias. The accuracy of the self-reported antimalarial drug use is unknown. Although self-reported information has some advantages over studies based on prescribed drugs in that the individual recalls using the drug, the reported drug and information on adherence are not validated. Self-reported health experience is subject to the usual disadvantages of recall bias and bias of reporting subjective experience without independent expert assessment; however, the use of standardized assessment tools may have circumvented these biases to some extent.
Lee et al. (2013) conducted a cross-sectional, web-based survey in 2009 of current and past members of the Australian Federal Police Association (AFP; a population similar to the U.S. military, but 70% male) to study the associations between deployment and exposure to doxycycline (compared with nondeployment and no exposure) and a new onset of gastrointestinal disease. Of those invited to participate, 1,300 (34%) responded, and 1,167 were eligible for analysis after the exclusion of 133 who had pre-existing gastrointestinal disease. The survey col-
lected data on demographics, gastrointestinal health prior to joining AFP, the duration of employment, past and current overseas engagement, incident development of gastrointestinal illness, and the concurrent use of doxycycline. Adverse events were self-reported or inferred by the investigators based on reported symptoms and treatments. Respondents who reported a new gastrointestinal illness provided known diagnoses that were validated by supportive data such as appropriate investigations or appropriate treatments received for the given diagnosis; unknown diagnoses were inferred by investigators based on symptom descriptors, investigation, or treatment received. A diagnosis of inflammatory bowel disease (IBD) was supported by the description of an appropriate investigation or medication listed, and a diagnosis of irritable bowel syndrome (IBS) was either provided by the respondent, or inferred by the investigators based on the symptoms reported. Individuals were assigned to three illness groups: acute gastroenteritis, IBS, or IBD (including ulcerative colitis or Crohn’s disease). All doxycycline prescriptions taken by deployed individuals were assumed to be for malaria prophylaxis; in nondeployed individuals, the timing, duration, and indication for doxycycline were not collected. Any incident illness was linked to the relevant deployment status according to its temporal relationship to the onset of symptoms. Logistic regression analyses were performed for the three adverse gastrointestinal events, stratified by deployment status and by deployment location in developing or developed countries using the classification of the United Nations Statistics Division. Of 590 deployed AFP members, 171 (30%) reported doxycycline use compared with 18 (3%) of 577 not deployed, although 21 of the 171 deployed individuals were exposed to at least one other antimalarial drug. A total of 158 incident gastrointestinal illnesses were reported during AFP employment, including acute gastroenteritis (10% deployed versus 1% not deployed; p < 0.001), IBS (5% versus 2%; p < 0.001), and IBD (2% versus 1%; p > 0.05). Compared with nondeployed AFP members with no doxycycline exposure (reference for all comparisons), nondeployed individuals who had used doxycycline reported fewer events of gastroenteritis (0 versus 4 cases, respectively), IBS (1 versus 12 cases, respectively), and IBD (0 versus 8 cases, respectively). Regression models included covariates of gender, family history of gastrointestinal illness, deployment status, deployment destination, and the use of doxycycline. For gastroenteritis, compared with nondeployed AFP members with no doxycycline exposure, those AFP members deployed to developing countries had a statistically significantly (but very imprecise) increased risk with doxycycline use (OR = 31.94, 95%CI 10.95–93.19), but the risk was not shown to be increased with doxycycline use and deployment to developed countries (OR = 7.89, 95%CI 0.83–75.05). For both deployment to developed countries (OR = 6.93, 95%CI 1.4–34.39) and to developing countries (OR = 2.47, 95%CI 0.77–7.89), the risk of IBS was statistically significantly increased for doxycycline use relative to nondeployed, non-doxycycline-exposed AFP personnel. For both deployment to developed countries (OR = 8.75, 95%CI 1.67–45.86) and to developing countries (OR = 6.99, 95%CI 3.19–15.31) the risk
of IBD was statistically significantly increased for doxycycline use relative to nondeployed, non-doxycycline-exposed AFP personnel. The authors presented effect estimates for comparisons with and without the use of doxycycline between deployments to developed and developing countries relative to AFP personnel not exposed to doxycycline and not deployed, but they did not present effect estimates for comparisons between doxycycline exposed and unexposed individuals by any deployment stratum (e.g., with and without exposure to doxycycline for deployment to developed or developing countries).
This study provides a possible signal concerning persistent adverse gastrointestinal events (IBS and IBD) of doxycycline when used during deployment. It used a relatively weak study design and relied on a self-reported survey as the source of information on both exposures and outcomes, which could have occurred many years before. There was limited information on the timing of exposure, duration, and adherence, and all exposure among deployed personnel was assumed to be for malaria prophylaxis. This is potentially an important issue because, in addition to malaria prophylaxis, doxycycline may be used to prevent diarrhea and to treat a number of other infections, so the drug use may be associated with the gastrointestinal illnesses because it was prescribed to treat them. There was no information on the timing of the symptoms in relation to the exposure to doxycycline. The response rate was low (34%), raising questions about the representativeness of the respondents. There was little information on possible confounding variables other than occurrence and area of deployment and general demographic factors.
U.S. Peace Corps
Tan et al. (2017) conducted a retrospective observational Internet-based survey of 8,931 (11% response rate) returned Peace Corps volunteers (who had served during 1995–2014) to compare the prevalence of selected health conditions after Peace Corps service between those who reported taking malaria prophylaxis (n = 5,055, 56.6%) and those who did not. Reported initial antimalarial prophylactic prescriptions were mefloquine (n = 2,981; 59.0%), A/P (n = 183; 3.6%), chloroquine (n = 674; 13.3%), doxycycline (n = 831; 16.4%), and 386 (7.6%) “other” prophylactic medications. In addition to questions on malaria prophylaxis (type, regimen, duration, and adherence), the survey included questions about the country of service, the type of assignment, and whether malaria prophylaxis was required at the assigned site. Respondents were also asked to report medical diagnoses made by a health care provider before, during, and after service in the Peace Corps and to answer questions about medications used before, during, or after Peace Corps service; about family history of disease and psychiatric illness; about psychiatric history prior to exposure; and about alcohol consumption. In total, more than 40 disease outcomes were examined for associations with each antimalarial, including derived outcomes of major depressive disorder, bipolar disorder, anxiety disorder, insomnia, psychoses, and cancers. Outcomes were grouped
by system (neuropsychologic, cardiac, ophthalmologic, dermatologic, reproductive, and gastrointestinal) or class (infectious, hematologic/oncologic) and within each group several diagnoses were listed. “Any psychiatric outcome” included all reported psychiatric diagnoses both derived and those reported as individual diagnoses, including schizophrenia, obsessive-compulsive disorder, and “other.” Neuropsychologic disorders were presented as a category that separately included dementia, migraines, seizures, tinnitus, vestibular disorder, “other” neurologic disorder, and “any” neurologic disorder. Of the outcomes examined, the authors reported that insomnia was the only diagnosis statistically significantly more prevalent among those who used any doxycycline compared with those who did not (9.0% versus 5.4%, respectively; prevalence ratio = 1.27, 95%CI 1.02–1.59). There was no difference in the prevalence of insomnia between those with prolonged or prolonged exclusively doxycycline use and those with no doxycycline use. For exposed and unexposed groups, there were no differences in prevalence of several disease diagnoses extrapolated from adverse events derived from reported and feared adverse events with doxycycline use (i.e., recurrent yeast infections, allergic or contact dermatitis, and gastrointestinal diseases). There were no statistical results presented for adverse events related to doxycycline exposure.
The study had many limitations which stemmed primarily from its design as an Internet-based survey of people with email addresses on file. The response rate was low (11%), the authors relied on self-report for both exposure and outcome information and the timing of each, and for some participants the time between exposure and the survey was many years. Most comparisons were between those who had been exposed to a specific drug (i.e., mefloquine, chloroquine, doxycycline, A/P, other) and all of those who had not. Thus, the comparison group for each antimalarial was a mixture of those who did not report taking any antimalarials and those who reported taking antimalarial drugs other than the one being examined. Overall, there were few details of the limited analyses presented making it difficult to understand the groups that were being compared, how they differed with respect to important covariates, and what variables were included in the models. The reliance on self-report, often years (range 2–20 years) after exposure, introduces several potential biases (selection bias, recall bias, and confounding bias), with inadequate information available to determine the likely impact or direction of the potential biases acting in this study. While the use of self-reported diagnoses that were specified to be those made by a medical professional to ascertain diagnoses was arguably a better method than using a checklist of symptoms, the adverse events were not validated against any objective information.
Meier et al. (2004) conducted a retrospective observational study in travelers using data from the UK-based General Practice Research Database (GPRD)—which has since changed names to the Clinical Practice Research Datalink—to
assess the incidence of and compare the odds of developing first-time psychiatric disorders in individuals using mefloquine for malarial prophylaxis compared with individuals who used other antimalarial drugs. The Clinical Practice Research Datalink, which has now been active for more than 30 years, collects de-identified patient data from a network of general practitioner practices across the United Kingdom for use in public health research and clinical studies, which have included investigations of drug safety, the use of medications, health care delivery, and disease risk factors (CPRD, 2019). Using the GPRD, investigators identified individuals who had at least one prescription for mefloquine, A/P, doxycycline, or chloroquine and/or proguanil in the time period of interest and who had a pretravel consultation within 1 week of the date of the prescription that included specific codes indicating that the prescription was for malaria prophylaxis. The start of the follow-up was the date of receipt of the first prescription for an individual. Current use was defined as between the date that a prescription was started and 1 week after the end of the time period of the drug prescription. Current exposure time was calculated differently for each antimalarial drug because the regimen for each of the antimalarial drugs differs. Investigators based their assessment on the number of tablets recorded by the general practitioner and calculated the assumed exposure time for each of the antimalarial drugs being investigated. For doxycycline the current exposure time (in days) was the number of tablets plus 7 days. Investigators added 90 days to each exposure to capture events occurring during travel that came to the attention of the general practitioner after the individual returned to the United Kingdom; this timeframe was termed recent use. Recent use included periods both relevant to the committee’s charge (days 28–89) and time periods that the committee considered exclusionary (days 7–27). Past use started at day 90 and ended at a maximum of 540 days after the end of current exposure, reflecting a time period pertinent to the committee’s assessment. Non-exposed people served as controls and had no antimalarial prescription during the study period or during 540 days after their pre-travel consultation, which also served as the date of the start of their follow-up. Participants were required to have at least 12 months of information on prescribed drugs and medical diagnoses before the first prescription date for an antimalarial or their travel consultation for the non-exposed controls. An additional inclusion criterion required participants to have recorded medical activity (diagnoses or drug prescriptions) after receiving a prescription to ensure that only those individuals who returned to the United Kingdom were included. A nested case–control analysis was also performed for a subset of the population in which up to six controls (who did not develop an outcome of interest during follow-up) were randomly selected per case; controls were matched to cases on age, sex, general practice, and calendar time (by assigning each control to the same index date as their matched case).
Meier et al. (2004) used the GPRD to assess the incidence of depression (n = 505), psychosis (n = 16), panic attacks (n = 57), and death by suicide (n = 2) in recent users (90 days following current use) of doxycycline compared with both
current users (during active use) of mefloquine, proguanil, and/or chloroquine and past users (90–540 days) of any of these antimalarials. The study population consisted of 35,370 individuals aged 17–79 years who used antimalarials between January 1990 and December 2000: 16,491 mefloquine users, 16,129 chloroquine, and/or proguanil users, and 4,574 doxycycline users (some individuals used multiple drugs). Investigators calculated the incidence of the four prespecified psychiatric outcomes during current, recent, and past use (people with prior diagnoses of the four psychiatric outcomes or alcoholism were excluded), and they also performed a nested case–control analysis in which both cases and controls had no history of the outcomes of interest prior to the use of any antimalarial. The incidence rates of first-time diagnoses were calculated using person-years and adjusted for age, gender, and calendar year. In total, 14 diagnoses of depression, 0 diagnoses of psychosis, 1 diagnosis of panic attack, and 0 deaths by suicide were reported for doxycycline users. The incidence rate of a first-time depression diagnosis did not differ between recent doxycycline users and all past users of antimalarials (RR = 0.8, 95%CI 0.4–1.4). In the nested case–control analysis, there was no difference in the odds of depression for recent doxycycline users compared with all other users combined after adjustment for age, gender, year, general practice, smoking status, and body mass index (BMI) (OR = 0.7, 95%CI 0.1–1.6). Regarding panic attacks, the incidence rate of a first-time diagnosis was no different for recent use of doxycycline than for past users of antimalarials (RR = 1.1, 95%CI 0.2–8.2). In the nested case–control analysis, the odds of panic attack were higher but not statistically significantly so for recent doxycycline users compared with all users (OR = 2.0, 95%CI 0.2–19.0) after adjusting for smoking status and BMI. This was a large retrospective study that found no increase in depression associated with current or recent use of doxycycline compared with the use of mefloquine, proguanil, and/or chloroquine, or all past users of antimalarials. The sample size was limited for the study of panic attacks and psychosis, leading to very imprecise estimates for those outcomes. Because current and recent use were analyzed separately, persistent outcomes were difficult to determine.
Schwartz and Regev-Yochay (1999) performed a prospective observational study, and followed 158 Israeli male and female travelers aged 22–65 years who took part in rafting trips on the Omo River, Ethiopia, and who had visited a travel clinic to obtain malaria prophylaxis. Travelers were prescribed mefloquine, primaquine, doxycycline (100 mg daily), or hydroxychloroquine by travel group. The primary aim of the study was to assess incident malaria and to compare the effectiveness of these four antimalarial drugs against both P. falciparum and P. vivax. Travelers were followed from the time of their return to Israel for an average of 16.6 months (range 8–37 months) for incident malaria. Adherence to the prophylactic regimens and the occurrence of adverse events were also collected by survey. The authors reported that “no severe side effects” were reported in any of the travelers, and one traveler withdrew from doxycycline use due to development of a rash. No other adverse events or withdrawals were noted in the doxycycline
users. The strengths of this study include its design and the long duration of followup (an average of 16.6 months after return from a malaria-endemic country). It is limited by its small sample size, the nonrandomized design, and the lack of details on adverse events beyond reporting that no severe events occurred and only one withdrawal was reported among doxycycline users. As a result, this study provides limited information that can be used for inferences.
Andersen et al. (1998) conducted a double-blinded randomized placebo-controlled trial of azithromycin and doxycycline as prophylaxis for malaria among 232 semi-immune adults aged 18–55 years in Kenya from April through August 1995 in an area with high rates of endemic malaria. The study compared the prophylactic efficacy of three antibiotic regimens given for 10 weeks—azithromycin, 250 mg daily (n = 59); azithromycin, 1,000 mg weekly (n = 58); and doxycycline, 100 mg daily (n = 55)—versus a placebo (n = 60). Participants were determined to be in good health and were given quinine and doxycycline therapy over 7 days to clear pre-existing parasitemia. Volunteers with medical complaints, including possible cases of symptomatic malaria, were evaluated and treated at a research clinic. Pregnancy tests and enrollment blood tests were repeated after 5 and 10 weeks of drug administration. After the period of study drug administration, weekly blood smears were examined for an additional 4 weeks, and blood tests were repeated 4 weeks after the last dose of study drug. The safety and tolerability of the regimens were assessed by a daily symptom questionnaire, by review of research clinic records, and by interval hematology and serum chemistry tests. Because the timing of the adverse events was not clearly specified, the study provides little evidence as to whether adverse events were persistent following the cessation of doxycycline. There were no substantial differences between the groups in the results of the serum chemistry and hematology tests, including at the 4-week post-dosing time point, although detailed results were not presented.
Several studies of doxycycline use in service members from the United States (Arthur et al., 1990; Saunders et al., 2015), Australia (Kitchener et al., 2005; Rieckmann et al., 1993), France (Michel et al., 2010; Pages et al., 2002), Indonesia (Ohrt et al., 1997), Italy (Peragallo, 2001), Turkey (Sonmez et al., 2005), and the United Kingdom (Terrell et al., 2015; Tuck and Williams, 2016) were reviewed by the committee. However, because they did not follow the military cohorts after doxycycline prophylaxis was complete or did not report on adverse events that occurred post-doxycycline-cessation, these studies were not further considered.
Studies of other populations were also excluded from the final set of studies evaluated in depth because the follow-up was not at least 28 days post-doxycycline-cessation or because the authors did not distinguish between the timing of adverse events (less than or at least 28 days post-cessation). Such studies included Al-Mofarreh and Al Mofleh (2003); Bjellerup and Ljunggren (1994); Lobel et al. (2001); Meropol et al. (2008); Pang et al. (1988); Phillips and Kass (1996); Schlagenhauf et al. (2003, 2009); Shanks et al. (1995a,b); Sharafeldin et al. (2010); Story et al. (1991); Taylor et al. (2003); Vilkman et al. (2016); and Waner et al. (1999). Similarly, three studies that were designed to examine the safety or tolerability of doxycycline when used for long-term (>4 months) prophylaxis in different populations were excluded from further consideration because they did not report on adverse events or other outcomes post-cessation of doxycycline (Cunningham et al., 2014: Korhonen et al., 2007; Landman et al., 2014).
Upon full-text review and quality assessment, additional studies were excluded from further consideration. Some studies were excluded because of doxycycline’s use for alternative treatment or prevention regimens (e.g., post-surgery or bacterial infection), because the reason for use of doxycycline was unclear, or because the dosage (or lack of) was determined to be irrelevant by the committee in addressing their charge (Berger, 1988; Chaabane et al., 2018).
The committee reviewed 14 published studies, totaling 23 cases, of adverse events related to the use of doxycycline in malaria prophylaxis. Various adverse events experienced by patients taking doxycycline resolved with discontinuation of the medication, including irritable mood and suicidality (Atigari et al., 2013), esophagitis (Geschwind, 1984), hiccups or esophageal ulceration (Morris and Davis, 2000; Tzianetas et al., 1996), diarrhea (Golledge and Riley, 1995), skin and nail issues (Cavens, 1981), and intracranial hypertension resulting in loss of vision (Lochhead and Elston, 2003), but a small number of studies reported adverse events that persisted beyond 28 days or 1 month after the discontinuation of doxycycline. The literature review produced seven published reports (nine total cases) that reported data after at least 1 month following doxycycline discontinuation (Belousova et al., 2018; Böhm et al., 2012; Gventer and Bruneti, 1985; Lim and Triscott, 2003; Lochhead and Elston, 2003; Morris and Davis, 2000; Neuberger and Schwartz, 2011). Three of these patients had skin lesions, including lymphamotoid papulosis (Belousova et al., 2018), that resolved by 5 months post-cessation of doxycycline; one patient experienced hyperpigmentation of the feet and legs (Böhm et al., 2002) that significantly improved at 1 year; and two patients had acute granuloma triggered by sun exposure (Lim and Triscott, 2003) that had resolved completely by 14 months post-cessation. Other adverse events that occurred concurrently and persisted following the cessation of doxycycline included onycholysis of the fingernails
(Gventer and Bruneti, 1985), which improved by the 3-month follow-up, and one case of diarrhea that resolved after an empirical anti-parasitic agent was prescribed (timing not provided) (Neuberger and Schwartz, 2011). One case of intracranial pressure returned to normal within 3 weeks of discontinuing doxycycline, but a consequent optic atrophy developed, resulting in a permanent loss of an estimated 70% of color vision and visual fields (Lochhead and Elston, 2003). These reports are suggestive of some persistent adverse events associated with doxycycline although the majority resolved or improved.
In the course of its review of the literature on doxycycline, the committee identified and reviewed available studies that reported results stratified by demographic, medical, or behavioral factors to assess whether the risk for adverse events when using doxycycline for prophylaxis is associated with being part of or affiliated with a specific group. This was not done exhaustively, and the evidence included in this section is generally limited to concurrent adverse events observed with use of doxycycline. Many of these studies did not meet the inclusion criteria of following their population for at least 28 days post-doxycycline-cessation, but the committee considers these findings to be important indicators when considering the evidence as a whole. The following risk groups were specifically considered: pregnant women and those with comorbid diseases or disorders.
According to FDA, doxycycline use is classified as a pregnancy class D drug (i.e., contraindicated in pregnancy) (FDA, 2018a). Specifically, pregnancy class D indicates that there is positive evidence of human fetal risk based on adverse reaction data from investigational or postmarketing experience or studies in humans, but the potential benefits may warrant the use of the drug in pregnant women, despite the potential risks. This classification stems from a “tetracycline class effect,” whereby tetracycline has been associated with teratogenicity, permanent yellowish-brown teeth discoloration after in utero exposure and in children under 8 years of age, and, more rarely, fatal hepatotoxicity reported in pregnant women (Cross et al., 2016). With respect to guidelines, according to Centers for Disease Control and Prevention malaria recommendations, doxycycline is contraindicated for use during pregnancy. Recommendations from the United Kingdom allow doxycycline for malaria prevention if other options are unsuitable, but the course of doxycycline, including the 4 weeks after travel, must be completed before 15 weeks’ gestation (Public Health England, 2018). However, there is little scientific evidence of adverse pregnancy outcomes associated with doxycycline use during pregnancy. An expert review of published data on experiences with doxycycline use during pregnancy by the Teratogen Information
System concluded that therapeutic doses during pregnancy are unlikely to pose a substantial teratogenic risk (the quantity and quality of data were judged to be limited to fair), but the data are insufficient to state that there is no risk (Friedman and Polifka, 2000). Czeizel and Rockenbauer (1997) conducted a case–control study in mothers of infants with and without congenital anomalies (18,515 and 32,804, respectively) and found a weak, marginally significant association with total malformations and doxycycline use anytime during pregnancy; however, the association was not seen when the analysis was restricted to maternal treatment during the period of organogenesis (i.e., the second and third months of gestation). Other studies have reported mixed results on the impact of doxycycline use during pregnancy on congenital malformations; Cooper et al. (2008) reported no increased incidence of major congenital malformations in infants whose mothers had taken doxycycline, while Muanda et al. (2017) reported a two-fold increased risk of circulatory system malformation and cardiac malformations, and a three-fold increased risk of ventricular/atrial septal defect that may be the result of pro-inflammatory cytokines, matrix metalloprotease (MMP) inhibition, or placental anomalies related to doxycycline use.
While the tetracycline class of antibiotics is associated with increased blood urea levels, doxycycline specifically was shown to be safe for use in patients with renal failure or insufficiency (George and Evans, 1971). The dosage of doxycycline should be doubled in individuals taking antiepileptic drugs, such as carbamazepine, phenytoin, and phenobarbitone. These drugs cause doxycycline to be metabolized more quickly than usual and can reduce the effectiveness of this antimalarial medication; therefore, some research has suggested that twice the normal prophylactic doxycycline dose should be taken to ensure sufficient protection from malaria (Minshall, 2015).
Overall, there is a lack of systematic studies of the long-term actions of doxycycline at prophylactic doses (100 mg per day in adult humans) on brain or nervous system function. There is little evidence to support or refute a role for doxycycline in promoting somatic and brain dysfunction. The committee found no evidence of persistent or latent adverse neurologic or psychiatric consequences in human or in preclinical models at doses relevant to malarial prophylaxis. In one study comparing doxycycline with other tetracyclines, evidence for vertigo was reported following minocycline, but not doxycycline, treatment (Cunha et al., 1982).
Doxycycline inhibits matrix metalloproteases (MMPs), which are enzymes that break down extracellular matrix and are associated with enhanced tissue dam-
age and inflammation (Bench et al., 2011; Parks et al., 2004). Studies performed in multiple organisms (mice, rats, cattle, chickens) indicate beneficial effects in the prevention or treatment of connective tissue–related conditions (joint inflammation, cardiac fibrotic changes, etc.) (Bench et al., 2011; Donato et al., 2017; Haerdi-Landerer et al., 2007; Lizotte-Waniewski et al., 2016; Peters et al., 2002; Riba et al., 2017). Given their mechanism of action, interference with MMPs would not be predicted to have adverse neurologic consequences. Indeed, increased MMP activity is associated with central nervous system damage and neurodegenerative processes (Rempe et al., 2016), and MMP inhibition by doxycycline could thus be beneficial. For example, one study showed doxycycline treatment can reduce blood–brain barrier leakage following malarial infection in mice (Schmidt et al., 2018), suggesting beneficial actions on neuropathologic processes.
Long-term doxycycline is used to turn gene expression on or off in mouse models using tetracycline-responsive transgene promoters. In these types of studies, doxycycline may be either injected or administered continuously over protracted time periods (weeks) via the drinking water; the latter method is more common. The drug binds to a tetracycline-sensitive promoter in a transgene, and once bound it will promote either activation or inhibition of the expression of downstream gene products. These studies often involve a doxycycline-only control group, and doxycycline alone does not affect the reported endpoints being measured in these studies (many null effects) (Belteki et al., 2005). However, doxycycline alone has not been examined as an independent variable, and thus the neurologic and behavioral impact of doxycycline has not been tested against non-doxycycline controls.
The fact that doxycycline has antibiotic and antimicroglial activity makes it possible that the drug can alter brain inflammatory processes. Microglia are receiving a lot of attention as possible mediators of brain disorders, including depression. There is some evidence to suggest that doxycycline may be beneficial in limiting microglial-related neuroinflammatory processes, showing efficacy in reducing microglial proliferation following the intracerebral injection of toxic amyloid beta peptide and attenuating cytokine expression in a murine model of Alzheimer disease (Balducci et al., 2018). Similarly, the gut microbiome appears to modulate affective behaviors in mice and rats (Bastiaanssen et al., 2019). There are not a lot of data analyzing the impact of doxycycline on microbiota (Saarela et al., 2007). Thus, while it is possible that doxycycline could affect the brain via a rearrangement of the microbiome, there is no definitive evidence to support adverse events of long-term doxycycline exposure on brain function via this mechanism.
Doxycycline is an antibiotic that affects inflammatory processes as well as decreases gut microbiota (including beneficial probiotic bacteria) (Saarela et al., 2007). One study found that doxycycline can exacerbate colon cancer in a murine model and that it can actually enhance gut inflammation in this paradigm (Nanda et al., 2016). However, there is also evidence for gastroprotection in an ulcer model (Singh et al., 2011). The weight of the data suggest that adverse doxycy-
cline effects on gastrointestinal function may be related to inflammatory changes in the gut, which is likely to vary substantially between individuals. These occur in a minority of users and dissipate once treatment is discontinued. Doxycycline can cause gastrointestinal symptoms in other species, including rats, mice, horses, cattle, and cats (Davis et al., 2006; German et al., 2005; Nanda et al., 2016; Riond and Riviere, 1998; Trumble, 2005). The studies often involve suprapharmacologic dosing over short time periods. Some of the consequences can be severe and usually involve the foregut (esophagus, stomach). The intensity of symptoms appears to be related to the formulation (acidity) (Malmborg, 1984).
The exact mechanism by which doxycycline produces esophagitis and esophageal ulcers is not completely understood. Its acidity has been considered a major factor with respect to its ability to damage the esophageal mucosa. Doxycycline accumulates within the basal layer of squamous epithelium in rats (Giger et al., 1978), which may inhibit protein synthesis and cause cellular degeneration. Human case reports of doxycycline-induced esophageal ulcers have supported the experimental evidence, in which diffuse degeneration of the basal layer was observed while the upper layer of esophageal mucosa was unaffected (Banisaeed et al., 2003). Other factors such as the drug dissociation rate (Bailey et al., 1990), pH (Carlborg et al., 1983), osmolarity, and intrinsic chemical toxicity (Bailey et al., 1990) are also implicated in the pathogenesis of drug-induced esophageal injury.
Doxycycline is known to produce photosensitivity in response to ultraviolet (UV)-A radiation, mediated by oxidative stress and mitochondrial toxicity. Accordingly, individuals undergoing doxycycline treatment (including for malarial prophylaxis) are warned to avoid sun exposure, and encouraged to wear protective clothing and apply broad-spectrum sunscreen (protecting against both UV-A and UV-B radiation) (Tan et al., 2011).
Although some people who take doxycycline do develop concurrent adverse events, such as photosensitivity, and there have been a few case reports of severe concurrent adverse events, the available post-cessation epidemiologic evidence does not find an association between the use of doxycycline for malaria prophylaxis and persistent or latent adverse events. The committee identified seven epidemiologic studies that included some mention of adverse events that occurred ≥28 days post-cessation of doxycycline that provided the most directly relevant information for assessing persistent health effects (Andersen et al., 1998; Eick-Cost et al., 2017; Lee et al., 2013; Meier et al., 2004; Schneiderman et al., 2018; Schwartz and Regev-Yochay, 1999; Tan et al., 2017). The studies are heterogeneous in the populations that were included (active military, veterans, U.S. Peace Corps volunteers, travelers, and endemic populations), in the modes of data collection on
drug exposure, adverse events, and covariates (administrative records, researcher collected, self-report), and particularly in the nature of the health outcomes that were considered.
In most cases the focus of the studies was on neurologic or psychiatric conditions or a general assessment of adverse events of all types. Within a particular adverse event category, such as psychiatric conditions, the information elicited ranged from more minor symptoms (such as anxiety) to severe clinical disorders (e.g., psychosis, depression, PTSD), posing a challenge to the committee’s ability to make an integrated assessment. Furthermore, the relevant studies were notably inconsistent in the reporting of results, and they covered different time periods in relation to the cessation of drug exposure. Given the inherently imperfect information generated by any one study, it would be desirable to have similar studies to assess the consistency of findings, but the diversity of methods used makes it very difficult to combine information across studies with confidence. Each of the included epidemiologic studies possessed strengths and limitations related to the specific methodology used, and the findings from those studies with the highest methodologic quality were given more weight when drawing conclusions. To avoid repetition for each outcome category, a short summary of the attributes of each study that was considered to be most contributory to the evidence base or that presented evidence germane to multiple outcome categories is presented first. The evidence summaries for each outcome category refer back to these short assessment summaries.
In addition to the post-cessation epidemiologic studies, the committee also considered supplemental evidence when making its conclusions, including recognized concurrent adverse events, case reports of persistent or latent adverse events, studies of adverse events in pregnant women and people with comorbid conditions, and information from experimental animal models or cell cultures. Consistent with the chapter syntheses of other antimalarial drugs, this synthesis is organized by body system category: neurologic disorders, psychiatric disorders, gastrointestinal disorders, eye disorders, cardiovascular disorders, and other outcomes and disorders, including dermatologic and biochemical parameters. Each conclusion consists of two parts: the first sentence assigns the level of association, and the second sentence offers additional detail regarding whether further research in a particular area is merited based on consideration of all the available evidence.
Epidemiologic Studies Presenting Contributory Evidence
Eick-Cost et al. (2017) used DoD administrative databases to perform a large retrospective cohort study among active-duty service members who filled at least one prescription for mefloquine, doxycycline, or A/P between 2008 and 2013. The primary study objective was to assess and compare the risk of incident and recurrent ICD-9-CM-coded neurologic and psychiatric outcomes that were reported at medical care visits during concurrent antimalarial use plus 365 days after the end
of a prescription. This was a well-designed study and included several important features that increased its methodologic quality: a large sample size, the use of an administrative data source for both exposure and outcomes, and careful consideration of potential confounders including demographics, psychiatric history, and the military characteristics of deployment and combat exposure. Because neurologic and psychiatric diagnoses occurring during current and recent use were analyzed together without distinguishing between events that occurred within 28 days of antimalarial use and those that occurred ≥28 days post-cessation, the study provides no quantitative information regarding the persistence of most events other than the notation in the text that the results did not change when restricted to the post-cessation period. The use of medical diagnoses is likely to be more reliable for the outcomes than self-report, but there was no validation of the diagnoses recorded in the administrative databases, and symptoms or events that did not result in a medical visit or diagnosis would have been missed. For PTSD diagnoses there was no information on when the index trauma occurred.
Schneiderman et al. (2018) conducted an analysis of self-reported health outcomes associated with the use of antimalarials in a population-based cohort of deployed and nondeployed U.S. veterans, using information collected as part of the NewGen Study. Exposure and outcomes were systematically obtained, and psychiatric outcomes were measured by standardized assessment instruments. Antimalarial medication use was grouped by mefloquine, chloroquine, doxycycline, primaquine, mefloquine in combination with other drugs, other antimalarials, and not specified or no antimalarial drug exposures. Health outcomes were self-reported using standardized instruments: the SF-12 for general health status, PCL-C for PTSD, and the PHQ. The overall sample was large, and the researchers used a reasonably thorough set of covariates in models estimating drug–outcome associations, including deployment and combat exposure. Although the time period of drug use and the timing of health outcomes were not directly addressed, given that the population was all veterans who had served between 2001 and 2008 and that the survey was not administered until 2009–2011, it is reasonable to assume that antimalarial drug use had ceased some time before. The methodology and response rate (34% total; weighted 20.5% of deployed and weighted 8.8% of nondeployed individuals used doxycycline) for this study may have led to the introduction of non-response, recall, or selection biases; however, the committee believed that the investigators used appropriate data analysis techniques to mitigate the effects of any biases that were present.
Meier et al. (2004) conducted a study using data from the UK-based GPRD to assess the incidence and to compare the odds of first-time neurologic or psychiatric diagnoses in individuals aged 17–79 years using mefloquine compared with individuals using other antimalarial drugs, including doxycycline, for malaria prophylaxis. The use of data from GPRD (a well-established platform designed for both clinical practice and research) allowed for adequate power to detect differences in outcomes and for the uniform collection of exposures (although recorded drug prescriptions do not equate to use or adherence) and outcomes (based on
clinical diagnoses coded from medical care visits) that were not subject to recall bias. Events that did not result in a medical care visit or that occurred outside of the national health care system would have been missed, and there may also be some differences between the travelers who traveled to malaria-endemic areas versus areas that are not endemic for malaria, which could lead to some apparent differences in outcomes between the groups. However, it is unlikely that this would result in differential selection bias. Diagnoses were defined a priori, which excluded other outcomes, including the potential to identify rare outcomes. The antimalarial-exposed populations were large, an appropriate comparison group of travelers not using any form of malaria prophylaxis was included, and health outcomes were reported in defined time periods, including current use through 90 days after a prescription ended (termed recent use in analyses) and 91–540 days following the cessation of use (termed past use in analyses). Adjustments were made for several confounders, including age, sex, calendar time, practice, smoking status, and BMI using appropriate study design or analytic methods. Each study included a nested case–control component that allowed for the control of important covariates.
The primary aim of Tan et al. (2017) was to assess the prevalence of several health conditions experienced by returned Peace Corps volunteers associated with the use of prophylactic antimalarial drugs. The number of participants was large (8,931 participants), and 16% of those who used an antimalarial had used doxycycline. A number of important covariates, such as psychiatric history and alcohol use, were collected, but the study had several methodologic limitations. These limitations included its study design (self-report, Internet-based survey), exposure characterization based on self-report (which introduces several potential biases such as recall bias, sampling bias, and confounding), outcome assessment (based on the self-report of health-provider-diagnosed conditions up to 20 years post-service), the use of mixed comparison groups, a lack of detail regarding the analysis methods, and poor response rate (11%, which likely introduces selection bias). The evidence generated by this study was thus considered to only weakly contribute to the inferences of doxycycline use and persistent or latent adverse events or disorders.
Although some studies grouped adverse events under a more general category of “neuropsychiatric effects” for discussion, the committee separated neurologic and psychiatric symptoms and conditions to the extent possible. The FDA label and package insert state that intracranial hypertension may be associated with use of doxycycline at the dose and frequency recommended for malaria prophylaxis, with clinical manifestations that include headache, blurred vision, diplopia, vision loss, and papilledema via fundoscopy. The risk of intracranial hypertension is increased in women of childbearing age who are overweight or have a history of
intracranial hypertension and in those using isotretinoin concomitantly. Among the case reports that followed individuals for 28 days or more post-doxycycline-cessation, one case of intracranial pressure was reported that returned to normal within 3 weeks of discontinuing doxycycline and starting treatment, but consecutive optic atrophy developed resulting in a permanent loss of an estimated 70% of color vision and visual fields (Lochhead and Elston, 2003). Based on a systematic review of short-term travelers, there were no statistically significant differences for headache or dizziness associated with concurrent drug use when doxycycline users were compared with mefloquine users. Other concurrent neurologic adverse events were reported by individual cohort studies, including balance disorder, fatigue, hypoaesthesia (numbness), and palpitations and tinnitus, but for all of these outcomes there was either no difference in risk or a higher risk for mefloquine users than for doxycycline users (Tickell-Painter et al., 2017). Individuals with epilepsy who are taking antiepileptic drugs may need to double the dosage of doxycycline because these drugs cause doxycycline to be metabolized more quickly than usual and may reduce its effectiveness against malaria (Minshall, 2015).
Experimental animal and human cell culture studies that used doxycycline were also examined for evidence of mechanisms that could plausibly support adverse events. The committee found no evidence of persistent or latent adverse neurologic events in preclinical models at doses relevant to malaria prophylaxis. The committee found little evidence to support or refute a role for doxycycline in promoting somatic and brain dysfunction. Doxycycline may inhibit matrix metalloproteases, but this is not predicted to have adverse neurologic consequences (indeed, MMP activation is associated with central nervous system damage and neurodegenerative processes, suggesting doxycycline may be of benefit in this context).
Two epidemiologic studies included neurologic outcomes that occurred at least 28 days following the cessation of doxycycline (Eick-Cost et al., 2017; Tan et al., 2017). Both studies examined different neurologic outcomes with little overlap and used different methods to identify neurologic events. Eick-Cost et al. (2017) used ICD-9-CM-coded outcomes of confusion, tinnitus, vertigo, and convulsions, whereas Tan et al. (2017) examined “neuropsychologic” as a category that separately included dementia, migraines, seizures, tinnitus, vestibular disorder, and “other neuropsychologic” disorders. While both studies have limitations, Eick-Cost et al. (2017) provided the most evidence for potential persistent or latent neurologic outcomes.
In their analysis of data from DoD administrative databases, Eick-Cost et al. (2017) examined neurologic outcomes, and analyses were stratified by deployment and, separately, by psychiatric history. The results of a sensitivity analysis in which the risk period was restricted to 30 days post-prescription were not reported, although the authors stated that the results were similar to those of the primary analyses. Adjusted incident rates for confusion, vertigo, and convulsions—but not for tinnitus—were higher among the nondeployed than among
the deployed groups who used doxycycline. When stratified by deployment, no statistically significant difference for any of the neurologic outcomes was found between deployed doxycycline users and mefloquine users. Among the nondeployed, doxycycline users had a statistically significantly increased risk of vertigo compared with mefloquine users, but no difference was found for the other three neurologic outcomes. When the population was restricted to the first mefloquine or doxycycline prescription per individual and included individuals with a prior history of a neurologic or psychiatric diagnosis, individuals with a neurologic diagnosis in the year preceding the prescription had statistically significantly elevated risks for a subsequent diagnosis of the same condition for all neurologic conditions reported (tinnitus, vertigo, and convulsions) compared with individuals without a diagnosis in the prior year. There were no statistically significant differences between mefloquine and doxycycline users for tinnitus, vertigo, or convulsions for people who had a prior neurologic diagnosis or for when users of these drugs were compared in people without a prior neurologic diagnosis. Overall, the largely null results that were reported suggest that null results would also be found if the analysis were restricted to outcomes occurring 28 days post-cessation. Tan et al. (2017) also reported no association between the use of doxycycline and any of the “neuropsychologic” adverse events examined; however, effect estimates were not presented.
Based on the available evidence, the committee concludes that there is insufficient or inadequate evidence of an association between the use of doxycycline for malaria prophylaxis and persistent or latent neurologic events. Current evidence does not suggest further study of such an association is warranted, given the lack of evidence regarding biologic plausibility, adverse events associated with concurrent use, or findings from the existing epidemiologic studies.
No psychiatric adverse events are listed in the FDA label or package insert for doxycycline. In a systematic review of concurrent symptoms among short-term travelers, doxycycline users were statistically significantly less likely than mefloquine users to report psychiatric events, including abnormal dreams, insomnia, anxiety, and depressed mood, but the pooled effect estimates were very imprecise. Whereas there were 15 episodes of abnormal thoughts and perceptions among mefloquine users, no episodes were reported for doxycycline users in the cohort studies examined (Tickell-Painter et al., 2017). The committee identified 14 published case reports and case series, totaling 23 individuals, of adverse events related to the use of doxycycline for malaria prophylaxis. One case of irritable mood and suicidality was reported that resolved once doxycycline was discontinued, and no other concurrent or persistent psychiatric symptoms associated with the use of doxycycline were reported. Considering experimental animal and other biologic plausibility studies overall, systematic studies of the long-term
actions of doxycycline at prophylactic doses on brain or central nervous system function are generally lacking. There is little evidence to support or refute a role for doxycycline in promoting somatic and brain dysfunction, and the committee found no evidence of persistent or latent adverse psychiatric or behavioral events in human or in preclinical models at doses relevant to malaria prophylaxis. The gut microbiome appears to modulate affective behaviors in mice and rats, but there is no definitive evidence to support an effect of doxycycline exposure on brain function via this mechanism.
Four of the epidemiologic studies with post-cessation follow-up included information on at least one adverse psychiatric outcome (Eick-Cost et al., 2017; Meier et al., 2004; Schneiderman et al., 2018; Tan et al., 2017). While all of these studies have methodologic limitations, Eick-Cost et al. (2017), Meier et al. (2004), and Schneiderman et al. (2018) provided the strongest evidence regarding the use of doxycycline and persistent or latent psychiatric outcomes. All four studies used different methods for measuring outcomes, and the psychiatric outcomes of interest varied across studies. Eick-Cost et al. (2017) examined adjustment disorder, anxiety disorder, depressive disorder, PTSD, psychoses, suicide ideation, paranoia, hallucinations, insomnia, and death by suicide using clinical diagnoses coded in DoD administrative databases. Meier et al. (2004) also used clinical diagnoses coded in a health care administrative database to examine incident depression, psychoses, panic attacks, and death by suicide. Schneiderman et al. (2018) used standardized self-report instruments to examine the outcomes of PTSD, thoughts of death or self-harm, other anxiety disorders, and major depression. Tan et al. (2017) used unverified self-reported symptoms to derive clinical diagnoses of major depressive disorder, bipolar disorder, anxiety disorder, schizophrenia, and “other.”
In their analysis of active-duty service members, Eick-Cost et al. (2017) found that with the exception of paranoia and death by suicide, the adjusted incident rates for psychiatric outcomes were higher among the deployed than among the nondeployed groups who used doxycycline. When comparisons between mefloquine and doxycycline use were stratified by deployment, the only statistically significant difference for any of the psychiatric outcomes for the deployed was a slightly decreased risk for anxiety disorders among doxycycline users. Among the nondeployed, doxycycline users had statistically significantly increased risks of adjustment disorder, insomnia, anxiety disorder, depressive disorder, and PTSD compared with mefloquine users, but no difference was found for the other five psychiatric outcomes. When the population was restricted to the first mefloquine or doxycycline prescription per individual and included individuals with a prior history of a neurologic or psychiatric diagnosis, individuals with a psychiatric diagnosis in the year preceding the prescription had statistically significantly elevated risks for a subsequent diagnosis of the same condition for all of the psychiatric conditions that were reported (adjustment disorder, anxiety, insomnia, depressive disorder, and PTSD) compared with individuals without a diagnosis in the prior year. There were no statistically significant differences between
mefloquine and doxycycline users for any of the psychiatric outcomes when comparisons were limited to people who had a prior psychiatric diagnosis or for when users of these drugs were compared in people without a prior psychiatric diagnosis. Schneiderman et al. (2018) also found deployed doxycycline users to have increased frequencies of mental health diagnoses compared with nondeployed doxycycline users for the four psychiatric outcomes examined. However, in the adjusted logistic regression models with all covariates considered (including demographics, deployment, and combat exposure), the use of doxycycline was not associated with any of the adverse psychiatric events in comparison with nonusers of antimalarial drugs: lower composite mental health score, PTSD, thoughts of death or self-harm, other anxiety, and major depression. When combat exposure intensity was specifically considered, the weighted prevalence estimates indicated that the prevalence of disorders increased with increasing combat exposure intensity. The results of Meier et al. (2004), which analyzed travelers, corroborated the findings of Schneiderman et al. (2018) and the deployed group of Eick-Cost et al. (2017) in finding that the use of doxycycline was not associated with a difference in depression diagnoses. Both Eick-Cost et al. (2017) and Meier et al. (2004) examined psychosis and death by suicide, and neither study found a statistically significant difference between doxycycline users and nondoxycycline users. Tan et al. (2017) also reported that in the set of psychiatric outcomes examined, none, except insomnia, was elevated among doxycycline users compared with those not using doxycycline. Regarding insomnia, Eick-Cost et al. (2017) found its risk to be statistically significantly increased among nondeployed doxycycline users compared with mefloquine users. However, Tan et al. (2017) also found no difference in the prevalence of insomnia between those with prolonged or prolonged exclusively doxycycline use and those with no doxycycline use.
Eick-Cost et al. (2017) and Schneiderman et al. (2018) examined PTSD diagnoses in active-duty U.S. military and veteran populations, respectively, and included estimates that adjusted for deployment and combat. Both studies found that deployed doxycycline users reported increased frequencies of PTSD compared with nondeployed doxycycline users. In fully adjusted models, Schneiderman et al. (2018) did not find any difference in PTSD for doxycycline users compared with those who used no antimalarials. Similarly, Eick-Cost et al. (2017) found no difference in risk for PTSD among the deployed that used doxycycline compared with those who used mefloquine. However, among the nondeployed, Eick-Cost et al. (2017) found that the risk of PTSD was statistically significantly increased for doxycycline users relative to mefloquine users. When the population was restricted to the first mefloquine or doxycycline prescription per individual and included individuals with a prior history of a neurologic or psychiatric diagnosis, individuals with a PTSD diagnosis in the year preceding the prescription had statistically significantly elevated risks for a subsequent diagnosis of PTSD compared with individuals without a diagnosis in the prior year. When comparing doxycycline and mefloquine users to those with and without prior psychiatric
diagnoses, there were no statistically significant differences between mefloquine and doxycycline users for PTSD.
In sum, although there are a few findings of increased risk among specific outcomes relative to certain groups and in comparison with mefloquine, in general the results of the post-cessation epidemiologic studies provide modest evidence of no increase in risk of persistent adverse psychiatric events among individuals using doxycycline for malaria prophylaxis.
Based on the available evidence, the committee concludes that there is insufficient or inadequate evidence of an association between the use of doxycycline for malaria prophylaxis and persistent or latent psychiatric events. Current evidence does not suggest further study of such an association is warranted, given the lack of evidence regarding biologic plausibility, adverse events associated with concurrent use, or findings from the existing epidemiologic studies.
The well-established concurrent adverse events of doxycycline on gastrointestinal symptoms, including nausea, vomiting, and diarrhea, justify a closer look at potentially persistent or latent gastrointestinal disorders following the cessation of use. The FDA label and package insert warn users of Clostridium difficile‒associated diarrhea that can occur more than 2 months after drug cessation and further warn that this can cause increased morbidity and mortality because these infections can be refractory to antimicrobial therapy and may require colectomy. Although Clostridium difficile infection is listed as an adverse event of doxycycline, the current evidence suggests that doxycycline may provide protection against such infections (Tariq et al., 2018; Turner et al., 2014). The package insert also includes language warning that patients can develop watery and bloody stools (with or without stomach cramps and fever) as late as 2 or more months after antibiotic cessation, as well as pancreatitis. A systematic review in short-term travelers found that doxycycline users were statistically significantly more likely than mefloquine users to report nausea, vomiting, and diarrhea (Tickell-Painter et al., 2017).
Post-diarrheal syndromes can be associated with persistent adverse gastrointestinal events, but neither the post-cessation epidemiologic studies nor the evidence presented in the systematic reviews examining concurrent adverse gastrointestinal adverse events support such an association. There is some evidence from the biologic plausibility literature indicating that doxycycline may exert effects on the gastrointestinal tract, especially the esophagus and stomach, but the findings are inconsistent, and the studies often involve suprapharmacologic dosing over short time periods. The intensity of symptoms appears to be related to the acidity of the formulation. Human case reports of doxycycline-induced esophageal ulcers have supported the experimental evidence, in which diffuse degeneration of the basal layer was observed while the upper layer of esophageal mucosa was unaffected (Banisaeed et al., 2003). Other factors such as the drug dissociation rate, pH,
osmolarity, and intrinsic chemical toxicity are also implicated in the pathogenesis of drug-induced esophageal injury (Bailey et al., 1990; Carlborg et al., 1983). Doxycycline is an antibiotic and consequently affects inflammatory processes and also decreases gut microbiota (including beneficial probiotic bacteria), suggesting that adverse doxycycline outcomes on gastrointestinal function may be related to inflammatory changes in the gut, which are likely to vary substantially among individuals. Although there is the potential for a concurrent irritant to become a chronic problem, there is no biologic-plausibility support for acute diarrhea that could be indicative of more chronic symptoms.
Two of the post-cessation epidemiologic studies provided information on gastrointestinal disorders (Lee et al., 2013; Tan et al., 2017). Lee et al. (2013) did not provide results for associations of the gastrointestinal outcomes with doxycycline use stratified by deployed versus nondeployed status and location of deployment to a developing or developed country. The only way to draw any inferences about the impact of doxycycline is by comparing the odds ratios for “doxycycline use and deployment to developing country” and “no doxycycline use and deployment to developing country” with both those estimates relative to “no doxycycline use and no deployment to a developing country.” Using that indirect, approximate estimate, doxycycline appears to be associated with an increased risk for all three gastrointestinal illnesses examined (although the authors infer an association only for IBS and IBD, not gastroenteritis). Given the peculiar approach to the analysis, the inability to directly examine the impact of doxycycline, the small number of cases, exposures to doxycycline being based on self-report, and the other design limitations and likely biases, these results are quite limited in value. In a similarly methodologically limited study, Tan et al. (2017) found no association between doxycycline use and gastrointestinal diseases.
Based on the available evidence, the committee concludes that there is insufficient or inadequate evidence of an association between the use of doxycycline for malaria prophylaxis and persistent or latent gastrointestinal events. Current evidence suggests further study of such an association is warranted, given the evidence regarding biologic plausibility, adverse events associated with concurrent use, or findings from the existing epidemiologic studies.
The FDA package insert does not contain any information on eye disorders associated with the use of doxycycline, although secondary effects of blurred vision, diplopia, and vision loss may occur as a result of intracranial hypertension. A systematic review conducted in short-term travelers examining concurrent adverse events of malaria prophylaxis found that, based on two cohort studies, visual impairment was statistically significantly less commonly reported among doxycycline users than mefloquine users (Tickell-Painter et al., 2017). One case report of intracranial hypertension that resulted in a loss of vision was identified,
but no other case reports that presented information on eye disorders persisting beyond 28 days post-doxycycline cessation were found. No studies of experimental animal studies were identified that examined the biologic plausibility of eye disorders.
One methodologically limited post-cessation epidemiologic study (Tan et al., 2017) was identified that presented data on eye disorders, which included macular degeneration, retinopathy, and “other.” No differences in associations between eye disorders and doxycycline use compared with no doxycycline use were reported.
Based on the available evidence, the committee concludes that there is insufficient or inadequate evidence of an association between the use of doxycycline for malaria prophylaxis and persistent or latent eye disorders. Current evidence does not suggest further study of such an association is warranted, given the lack of evidence regarding biologic plausibility, adverse events associated with concurrent use, or findings from the existing epidemiologic studies.
The FDA label and package insert does not present any information regarding an association between concurrent adverse cardiovascular events and doxycycline use, and the committee did not identify any case reports reporting such an association. A systematic review conducted in short-term travelers examining concurrent adverse events found no associations between the use of doxycycline compared with mefloquine and cardiovascular disorders (Tickell-Painter et al., 2017). The committee did not identify any case reports that presented information on cardiovascular disorders persisting beyond 28 days post-doxycycline-cessation. In studies of experimental animals, doxycycline was found to inhibit matrix metalloproteases, and, as such, it may have beneficial effects in the prevention or treatment of connective-tissue-related conditions, including cardiac fibrotic changes.
One methodologically limited post-cessation epidemiologic study (Tan et al., 2017) found no association between doxycycline use and cardiovascular outcomes (arrhythmia, congestive heart failure, myocardial infarction, or “any” cardiac disorder).
Based on the available evidence, the committee concludes that there is insufficient or inadequate evidence of an association between the use of doxycycline for malaria prophylaxis and persistent or latent cardiovascular events. Current evidence does not suggest further study of such an association is warranted, given the lack of evidence regarding biologic plausibility, adverse events associated with concurrent use, or findings from the existing epidemiologic studies.
Other Outcomes and Disorders
A well-recognized concurrent adverse event of doxycycline is increased photosensitivity and skin rashes, which is thought to be mediated by oxidative
stress and mitochondrial toxicity. Individuals with lighter complexions may be more prone to photosensitivity while taking doxycycline. Other adverse dermatologic events—some of which may be severe—are also presented on the FDA label and package insert and include exfoliative dermatitis, erythema multiforme, Stevens-Johnson syndrome, toxic epidermal necrolysis, and drug reaction with eosinophilia and systemic symptoms. The committee found case reports of acute skin and nail issues and persistent adverse events of onycholysis of the fingernails which improved by 3 months; skin lesions, including lymphamotoid papulosis, which resolved by 5 months post-cessation of doxycycline; hyperpigmentation of the feet and legs, which significantly improved at 1 year; and two patients with acute granuloma triggered by sun exposure, which resolved completely by a 14-month post-discontinuation follow-up. Vaginitis and yeast infections have been associated with doxycycline use, but these conditions are often included under the category of non-specific skin reactions.
Another specific and well-documented persistent adverse event of doxycycline use that results from exposure during the second half of pregnancy is permanent tooth discoloration and enamel hypoplasia in the fetus, which the potential for can persist well beyond the period of doxycycline use in the mother. There is not an obvious extrapolation of this adverse event to other types of health problems in adults. Reversible tooth discoloration in adults’ teeth has also been reported, but these are cosmetic rather than clinical problems.
Tan et al. (2017) was the only epidemiologic study that met the committee’s inclusion criteria that systematically examined dermatologic outcomes. Although this study had many limitations and at best can only contribute weak evidence, no differences were found in the prevalence of several disease diagnoses extrapolated from adverse events derived from reported and feared adverse events with doxycycline use (i.e., recurrent yeast infections and allergic or contact dermatitis) but no statistical results were presented for these outcomes related to doxycycline exposure. Schwartz and Regev-Yochay (1999) followed 158 Israeli travelers who took part in rafting trips on the Omo River, Ethiopia, and who had visited a travel clinic to obtain malaria prophylaxis. Of the travelers prescribed doxycycline, one traveler withdrew from doxycycline use because of the development of a rash. No other adverse events or withdrawals were noted in the doxycycline users, and the persistence of the rash was not reported. Travelers were followed for an average of 16.6 months (range 8–37 months), but no drug-associated adverse events were reported (or appeared to be collected).
In a double-blinded randomized placebo-controlled trial of azithromycin and doxycycline as prophylaxis for malaria among 232 semi-immune adults, Andersen et al. (1998) collected blood tests at baseline, at weeks 5 and 10 during drug administration, and at 4 weeks post-drug-administration for hematology and serum chemistry testing. The authors reported that there were no substantial differences between the groups regarding the results of the serum chemistry and hematology tests, including at the 4-week post-dosing time point (although detailed results
were not presented). Although it met the inclusion criteria, given this limited information, this study did not provide any evidence that could be used to assess persistent or latent adverse events.
Al-Mofarreh, M. A., and I. A. Al Mofleh. 2003. Esophageal ulceration complicating doxycycline therapy. World J Gastroenterol 9(3):609-611.
Andersen, S. L., A. J. Oloo, D. M. Gordon, O. B. Ragama, G. M. Aleman, J. D. Berman, D. B. Tang, M. W. Dunne, and G. D. Shanks. 1998. Successful double-blinded, randomized, placebo-controlled field trial of azithromycin and doxycycline as prophylaxis for malaria in western Kenya. Clin Infect Dis 26(1):146-150.
Arguin, P. M., and A. J. Magill. 2017. For the record: A history of malaria chemoprophylaxis. https://wwwnc.cdc.gov/travel/yellowbook/2018/infectious-diseases-related-to-travel/emfor-the-recorda-history-of-malaria-chemoprophylaxisem (accessed December 18, 2018).
Arthur, J. D., P. Echeverria, G. D. Shanks, J. Karwacki, L. Bodhidatta, and J. E. Brown. 1990. A comparative study of gastrointestinal infections in United States soldiers receiving doxycycline or mefloquine for malaria prophylaxis. Am J Trop Med Hyg 43(6):608-613.
Atigari, O. V., C. Hogan, and D. Healy. 2013. Doxycycline and suicidality. BMJ Case Rep 2013:bcr2013200723.
Bailey, R. T., Jr., L. Bonavina, P. E. Nwakama, T. R. DeMeester, and S. C. Cheng. 1990. Influence of dissolution rate and pH of oral medications on drug-induced esophageal injury. DICP: The Annals of Pharmacotherapy 24(6):571-574.
Balducci, C., G. Santamaria, P. La Vitola, E. Brandi, F. Grandi, A. R. Viscomi, M. Beeg, M. Gobbi, M. Salmona, S. Ottonello, and G. Forloni. 2018. Doxycycline counteracts neuroinflammation restoring memory in Alzheimer’s disease mouse models. Neurobiol Aging 70:128-139.
Banisaeed, N., R. M. Truding, and C. H. Chang. 2003. Tetracycline-induced spongiotic esophagitis: A new endoscopic and histopathologic finding. Gastrointest Endosc 58(2):292-294.
Bastiaanssen, T. F. S., C. S. M. Cowan, M. J. Claesson, T. G. Dinan, and J. F. Cryan. 2019. Making sense of . . . the microbiome in psychiatry. Int J Neuropsychopharmacol 22(1):37-52.
Belousova, I. E., L. Kyrpychova, A. V. Samtsov, and D. V. Kazakov. 2018. A case of lymphomatoid papulosis type E with an unusual exacerbated clinical course. Am J Dermatopathol 40(2):145-147.
Belteki, G., J. Haigh, N. Kabacs, K. Haigh, K. Sison, F. Costantini, J. Whitsett, S. E. Quaggin, and A. Nagy. 2005. Conditional and inducible transgene expression in mice through the combinatorial use of Cre-mediated recombination and tetracycline induction. Nucleic Acids Res 33(5):e51.
Bench, T. J., A. Jeremias, and D. L. Brown. 2011. Matrix metalloproteinase inhibition with tetracyclines for the treatment of coronary artery disease. Pharmacol Res 64(6):561-566.
Berger, R. S. 1988. A double-blind, multiple-dose, placebo-controlled, cross-over study to compare the incidence of gastrointestinal complaints in healthy subjects given Doryx® and Vibramycin®. J Clin Pharmacol 28(4):367-370.
Binh, V. Q., N. T. Chinh, N. X. Thanh, B. T. Cuong, N. N. Quang, B. Dai, T. Travers, and M. D. Edstein. 2009. Sex affects the steady-state pharmacokinetics of primaquine but not doxycycline in healthy subjects. Am J Trop Med Hyg 81(5):747-753.
Bjellerup, M., and B. Ljunggren. 1994. Differences in phototoxic potency should be considered when tetracyclines are prescribed during summer-time. A study on doxycycline and lymecycline in human volunteers, using an objective method for recording erythema. Br J Dermatol 130(3):356-360.
Böhm, M., P. F. Schmidt, B. Lödding, H. Uphoff, G. Westermann, T. A. Luger, G. Bonsmann, and D. Metze. 2002. Cutaneous hyperpigmentation induced by doxycycline: Histochemical and ultrastructural examination, laser microprobe mass analysis, and cathodoluminescence. Am J Dermatopathol 24(4):345-350.
Bott, S., C. Prakash, and R. W. McCallum. 1987. Medication-induced esophageal injury: Survey of the literature. Am J Gastroenterol 82(8):758-763.
Brisson, M., and P. Brisson. 2012. Compliance with antimalaria prophylaxis in a combat zone. Am J Trop Med Hyg 86(4):587-590.
Carlborg, B., O. Densert, and C. Lindqvist. 1983. Tetracycline-induced esophageal ulcers: A clinical and experimental study. Laryngoscope 93(2):184-187.
Cavens, T. R. 1981. Onycholysis of the thumbs probably due to a phototoxic reaction from doxycycline. Cutis 27(1):53-54.
Chaabane, A., N. B. Fadhel, Z. Chadli, H. B. Romdhane, N. B. Fredj, N. A. Boughattas, and K. Aouam. 2018. Association of non-immediate drug hypersensitivity with drug exposure: A case control analysis of spontaneous reports from a Tunisian pharmacovigilance database. Eur J Int Med 53:40-44.
Cooper, W. O., S. Hernandez-Diaz, P. G. Arbogast, J. A. Dudley, S. M. Dyer, P. S. Gideon, K. S. Hall, L. A. Kaltenbach, and W. A. Ray. 2008. Antibiotics potentially used in response to bioterror-ism and the risk of major congenital malformations. Paediatr Perinat Epidemiol 23(1):18-28.
CPRD (Clinical Practice Research Datalink). 2019. Home. https://www.cprd.com (accessed December 3, 2019).
Cross, R., C. Ling, N. P. Day, R. McGready, and D. H. Paris. 2016. Revisiting doxycycline in pregnancy and early childhood—Time to rebuild its reputation? Expert Opin Drug Saf 15(3):367-382.
Cunha, B. A., C. M. Sibley, and A. M. Ristuccia. 1982. Doxycycline. Ther Drug Monit 4(2):115-135.
Cunningham, J., J. Horsley, D. Patel, A. Tunbridge, and D. G. Lalloo. 2014. Compliance with long-term malaria prophylaxis in British expatriates. Travel Med Infect Dis 12(4):341-348.
Czeizel, A. E., and M. Rockenbauer. 1997. Teratogenic study of doxycycline. Obstet Gynecol 89(4):524-528.
Davis, J. L., J. H. Salmon, and M. G. Papich. 2006. Pharmacokinetics and tissue distribution of doxycycline after oral administration of single and multiple doses in horses. Am J Vet Res 67(2):310-316.
Dickinson, B. D., R. D. Altman, N. H. Nielsen, M. L. Sterling, and the AMA Council on Scientific Affairs. 2001. Drug interactions between oral contraceptives and antibiotics. Obstet Gynecol 98(5 Pt 1):853-860.
DoD (Department of Defense). 2009. Policy memorandum on the use of mefloquine (Lariam®) in malaria prophylaxis. September 4, #HA 09-017. Provided by COL Andrew Wiesen, M.D., M.P.H., Director, Preventive Medicine, Health Readiness Policy, and Oversight, Office of the Assistant Secretary of Defense (Health Affairs), DoD, January 25, 2019.
Donato, M., B. Buchholz, C. Morales, L. Valdez, T. Zaobornyj, S. Baratta, D. T. Paez, M. Matoso, G. Vaccarino, D. Chejtman, O. Agüero, J. Telayna, J. Navia, A. Hita, A. Boveris, and R. J. Gelpi. 2017. Loss of dystrophin is associated with increased myocardial stiffness in a model of left ventricular hypertrophy. Mol Cell Biochem 432(1-2):169-178.
Eick-Cost, A. A., Z. Hu, P. Rohrbeck, and L. L. Clark. 2017. Neuropsychiatric outcomes after mefloquine exposure among U.S. military service members. Am J Trop Med Hyg 96(1):159-166.
FDA (U.S. Food and Drug Administration). 2005a. Package insert for Doryx® (doxycycline hyclate) and Doryx® MPC delayed-release tablets. https://www.accessdata.fda.gov/drugsatfda_docs/label/2005/050795lbl.pdf (accessed July 12, 2019).
FDA. 2005b. Package insert for Doryx® (doxycycline hyclate) delayed-release capsules. https://www.accessdata.fda.gov/drugsatfda_docs/label/2005/050582s024lbl.pdf (accessed July 7, 2019).
FDA. 2007. Package insert for Vibramycyin calcium (doxycycline calcium oral suspension, USP) oral suspension syrup, Vibramycin hyclate (doxycycline hyclate, USP) capsules, Vibramycin monohydrate (doxycycline monohydrate) for oral suspension, Vibra-Tabs (doxycycline hyclate tablets, USP). https://www.accessdata.fda.gov/drugsatfda_docs/label/2008/050006s79,050007s20,050480s42,050533s36lbl.pdf (accessed July 7, 2019).
FDA. 2014. Package insert for Acticlate™ (doxycycline hyclate USP) tablets. https://www.accessdata.fda.gov/drugsatfda_docs/label/2014/205931s000lbl.pdf (accessed July 7, 2019).
FDA. 2016. Package insert for Acticlate® CAP (doxycycline hyclate) capsules. https://www.accessdata.fda.gov/drugsatfda_docs/label/2014/205931s000lbl.pdf (accessed July 7, 2019).
FDA. 2017. Package insert for Acticlate® (doxycycline hyclate) tablets and Acticlate® CAP (doxycycline hyclate) capsules. https://www.accessdata.fda.gov/drugsatfda_docs/label/2017/205931s003,208253s001lbl.pdf (accessed July 7, 2019).
FDA. 2018a. Package insert for Doryx® (doxycycline hyclate) and Doryx® MPC delayed-release tablets. https://www.accessdata.fda.gov/drugsatfda_docs/label/2018/050795s026lbl.pdf (accessed July 12, 2019).
FDA. 2018b. Package insert for Doryx® (doxycycline hyclate) delayed-release capsules. https://www.accessdata.fda.gov/drugsatfda_docs/label/2018/050582s030lbl.pdf (accessed July 12, 2019).
FDA. 2018c. Package insert for Vibramycin® calcium (doxycycline calcium oral suspension, USP) oral suspension syrup, Vibramycin® hyclate (doxycycline hyclate capsules, USP) capsules, Vibramycin® monohydrate (doxycycline monohydrate) for oral suspension, Vibra-Tabs® (doxycycline hyclate tablets, USP) film coated tablets. https://www.accessdata.fda.gov/drugsatfda_docs/label/2018/050006s091lbl.pdf.
Friedman, J. M., and J. E. Polifka. 2000. Teratogenic effects of drugs: A resource for clinicians (TERIS). Baltimore, MD: The Johns Hopkins University Press. Pp. 149-195.
George, C. R., and R. A. Evans. 1971. Tetracycline toxicity in renal failure. Med J Aust 1(24):1271-1273.
German, A. J., M. J. Cannon, C. Dye, M. J. Booth, G. R. Pearson, C. A. Reay, and T. J. GruffyddJones. 2005. Oesophageal strictures in cats associated with doxycycline therapy. J Fel Med Surg 7(1):33-41.
Geschwind, A. 1984. Oesophagitis and oesophageal ulceration following ingestion of doxycycline tablets. Med J Aust 140(4):223.
Giger, M., A. Sormenberg, H. Brandli, M. Singeisen, R. Giiller, and A. L. Blum. 1978. Das tetracyclin-ulkus der speiser6hre. Klinisches Bild und in-vitro-untersuchungen [in German with English abstract]. Dtsch Med Wschr 103:1038-1040.
Golledge, C. L., and T. V. Riley. 1995. Clostridium difficile-associated diarrhoea after doxycycline malaria prophylaxis. Lancet (London, England) 345(8961):1377-1378.
Gventer, M., and V. A. Brunetti. 1985. Photo-onycholysis secondary to tetracycline. A case report. J Am Podiat Med Assoc 75(12):658-660.
Haerdi-Landerer, M. C., M. M. Suter, and A. Steiner. 2007. Intra-articular administration of doxycycline in calves. Am J Vet Res 68(12):1324-1331.
Hasan, S. A. 2007. Interaction of doxycycline and warfarin: An enhanced anticoagulant effect. Cornea 26(6):742-743.
Higgins, J. P. T., J. A. Lopez-Lopez, B. J. Becker, S. R. Davies, S. Dawson, J. M. Grimshaw, L. A. McGuinness, T. H. M. Moore, E. A. Rehfuess, J. Thomas, and D. M. Caldwell. 2019. Synthesising quantitative evidence in systematic reviews of complex health interventions. BMJ Glob Health 4(Suppl 1):e000858.
Kitchener, S. J., P. E. Nasveld, R. M. Gregory, and M. D. Edstein. 2005. Mefloquine and doxycycline malaria prophylaxis in Australian soldiers in East Timor. Med J Aust 182(4):168-171.
Klein, N. C., and B. A. Cunha. 1995. Tetracyclines. Med Clin N Am 79(4):789-801.
Korhonen, C., K. Peterson, C. Bruder, and P. Jung. 2007. Self-reported adverse events associated with antimalarial prophylaxis in Peace Corps volunteers. Am J Prev Med 33(3):194-199.
Kundu, C. N., S. Das, A. Nayak, S. R. Satapathy, D. Das, and S. Siddharth. 2015. Anti-malarials are anti-cancers and vice versa—One arrow two sparrows. Acta Tropica 149:113-127.
Landman, K. Z., K. R. Tan, P. M. Arguin, and the Centers for Disease Control and Prevention. 2014. Knowledge, attitudes, and practices regarding antimalarial chemoprophylaxis in U.S. Peace Corps volunteers—Africa, 2013. MMWR 63(23):516-517.
Lee, T. W., L. Russell, M. Deng, and P. R. Gibson. 2013. Association of doxycycline use with the development of gastroenteritis, irritable bowel syndrome and inflammatory bowel disease in Australians deployed abroad. Intern Med J 43(8):919-926.
Lim, D. S., and J. Triscott. 2003. O’Brien’s actinic granuloma in association with prolonged doxycycline phototoxicity. Australas J Dermatol 44(1):67-70.
Lizotte-Waniewski, M., K. Brew, and C. H. Hennekens. 2016. Hypothesis: Metalloproteinase inhibitors decrease risks of cardiovascular disease. J Cardiovasc Pharmacol Ther 21(4):368-371.
Lobel, H. O., M. A. Baker, F. A. Gras, G. M. Stennies, P. Meerburg, E. Hiemstra, M. Parise, M. Odero, and P. Waiyaki. 2001. Use of malaria prevention measures by North American and European travelers to East Africa. J Travel Med 8(4):167-172.
Lochhead, J., and J. S. Elston. 2003. Doxycycline induced intracranial hypertension. BMJ (Clin Res Ed) 326(7390):641-642.
Malmborg, A. S. 1984. Bioavailability of doxycycline monohydrate. A comparison with equivalent doses of doxycycline hydrochloride. Chemotherapy 30(2):76-80.
Meier, C. R., K. Wilcock, and S. S. Jick. 2004. The risk of severe depression, psychosis or panic attacks with prophylactic antimalarials. Drug Saf 27(3):203-213.
Meropol, S. B., K. A. Chan, Z. Chen, J. A. Finkelstein, S. Hennessy, E. Lautenbach, R. Platt, S. D. Schech, D. Shatin, and J. P. Metlay. 2008. Adverse events associated with prolonged antibiotic use. Pharmacoepidemiol Drug Saf 17(5):523-532.
Michel, R., S. Bardot, B. Queyriaux, J. P. Boutin and J. E. Touze. 2010. Doxycycline–chloroquine vs. doxycycline–placebo for malaria prophylaxis in nonimmune soldiers: A double-blind randomized field trial in sub-Saharan Africa. Trans R Soc Trop Med Hyg 104(4):290-297.
Minshall, I. 2015. Epilepsy and anti-malarial medication. https://www.epilepsyresearch.org.uk/wpcontent/uploads/2015/05/antimalarials4.pdf (accessed November 11, 2019).
Morris, T. J., and T. P. Davis. 2000. Doxycycline-induced esophageal ulceration in the U.S. military service. Mil Med 165(4):316-319.
Muanda, F. T., O. Sheehy, and A. Bérard. 2017. Use of antibiotics during pregnancy and the risk of major congenital malformations: A population based cohort study. Br J Clin Pharmacol 83(11):2557-2571.
Nanda, N., D. K. Dhawan, A. Bhatia, A. Mahmood, and S. Mahmood. 2016. Doxycycline promotes carcinogenesis & metastasis via chronic inflammatory pathway: An in vivo approach. PLOS ONE 11(3):e0151539.
Neuberger, A., and E. Schwartz. 2011. Clostridium difficile infection after malaria prophylaxis with doxycycline: Is there an association? Travel Med Infect Dis 9(5):243-245.
Ohrt, C., T. L. Richie, H. Widjaja, D. Shanks, J. Fitriadi, D. J. Fryauff, J. Handschin, D. Tang, B. Sandjaja, E. Tjitra, L. Hadiarso, G. Watt, and F. S. Wignall. 1997. Mefloquine compared with doxycycline for the prophylaxis of malaria in Indonesian soldiers: Randomized, double blind, placebo controlled trial. Ann Intern Med 126:963-972.
Overbosch, D., H. Schilthuis, U. Bienzle, R. H. Behrens, K. C. Kain, P. D. Clarke, S. Toovey, J. Knobloch, H. D. Nothdurft, D. Shaw, N. S. Roskell, J. D. Chulay, and the Malarone International Study Team. 2001. Atovaquone-proguanil versus mefloquine for malaria prophyalxis in nonimmune travelers: Results from a randomized, double-blind study. Clin Infect Dis 33(7):1015-1021.
Pages, F., J. P. Boutin, J. B. Meynard, A. Keundjian, S. Ryfer, L. Giurato, and D. Baudon. 2002. Tolerability of doxycycline monohydrate salt vs. chloroquine–proguanil in malaria prophylaxis. Trop Med Int Health 7(11):919-924.
Pang, L., N. Limsomwong, and P. Singharaj. 1988. Prophylactic treatment of vivax and falciparum malaria with low-dose doxycycline. J Infect Dis 158(5):1124-1127.
Parks, W. C., C. L. Wilson, and Y. S. Lopez-Boado. 2004. Matrix metalloproteinases as modulators of inflammation and innate immunity. Nat Rev Immunol 4(8):617-629.
Penning-van Beest, F. J., J. Koerselman, and R. M. Herings. 2008. Risk of major bleeding during concomitant use of antibiotic drugs and coumarin anticoagulants. J Thromb Haemost 6(2):284-290.
Peragallo, M. S. 2001. The Italian army standpoint on malaria chemoprophylaxis. Med Trop (Mars) 61(1):59-62.
Peters, T. L., R. M. Fulton, K. D. Roberson, and M. W. Orth. 2002. Effect of antibiotics on in vitro and in vivo avian cartilage degradation. Avian Dis 46(1):75-86.
Phillips, M. A., and R. B. Kass. 1996. User acceptability patterns for mefloquine and doxycycline malaria chemoprophylaxis. J Travel Med 3(1):40-45.
Public Health England. 2018. Guidelines for malaria prevention in travelers from the UK. https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/774781/ACMP_guidelines_2018.pdf (accessed December 3, 2019).
Rempe, R. G., A. M. S. Hartz, and B. Bauer. 2016. Matrix metalloproteinases in the brain and blood–brain barrier: Versatile breakers and makers. J Cereb Blood Flow Metab 36(9):1481-1507.
Riba, A., L. Deres, K. Eros, A. Szabo, K. Magyar, B. Sumegi, K. Toth, R. Halmosi, and E. Szabados. 2017. Doxycycline protects against ROS-induced mitochondrial fragmentation and ISO-induced heart failure. PLOS ONE 12(4):e0175195.
Rieckmann, K. H., A. E. Yeo, D. R. Davis, D. C. Hutton, P. F. Wheatley, and R. Simpson. 1993. Recent military experience with malaria prophylaxis. Med J Aust 158(7):446-449.
Riond, J. L., and J. E. Riviere. 1988. Pharmacology and toxicology of doxycycline. Vet Hum Toxicol 30(5):431-443.
Rudland, S., M. Little, P. Kemp, A. Miller, and J. Hodge. 1996. The enemy within: Diarrheal rates among British and Australian troops in Iraq. Mil Med 161(12):728-731.
Saarela, M., J. Maukonen, A. von Wright, T. Vilpponen-Salmela, A. J. Patterson, K. P. Scott, H. Hämynen, and J. Mättö. 2007. Tetracycline susceptibility of the ingested Lactobacillus acidophilus LaCH-5 and Bifidobacterium animalis subsp. lactis Bb-12 strains during antibiotic/probiotic intervention. Int J Antimicrob Agents 29(3):271-280.
Sack, R. B., M. Santosham, J. L. Froehlich, C. Medina, F. Orskov, and I. Orskov. 1984. Doxycycline prophylaxis of travelers’ diarrhea in Honduras, an area where resistance to doxycycline is common among enterotoxigenic Escherichia coli. Am J Trop Med Hyg 33(3):460-466.
Sánchez, J. L., R. F. DeFraites, T. W. Sharp, and R. K. Hanson. 1993. Mefloquine or doxycycline prophylaxis in U.S. troops in Somalia. Lancet 341(8851):1021-1022.
Saunders, D. L., E. Garges, J. E. Manning, K. Bennett, S. Schaffer, A. J. Kosmowski, and A. J. Magill. 2015. Safety, tolerability, and compliance with long-term antimalarial prophylaxis in American soldiers in Afghanistan. Am J Trop Med Hyg 93(3):584-590.
Schlagenhauf, P., A. Tschopp, R. Johnson, H. D. Nothdurft, B. Beck, E. Schwartz, M. Herold, B. Krebs, O. Veit, R. Allwinn, and R. Steffen. 2003. Tolerability of malaria prophylaxis in nonimmune travellers to sub-Saharan Africa: Multicentre, randomised, double blind, four arm study. BMJ 327(7423):1078.
Schlagenhauf, P., R. Johnson, E. Schwartz, H. D. Nothdurft, and R. Steffen. 2009. Evaluation of mood profiles during malaria prophylaxis: A randomized, double-blind, four-arm study. J Travel Med 16(1):42-45.
Schlagenhauf, P., M. E. Wilson, E. Petersen, A. McCarthy, L. H. Chen, J. S. Keystone, P. E. Kozarsky, B. A. Connor, H. D. Nothdurft, M. Mendelson, and K. Leder. 2019. Malaria chemoprophylaxis. J Travel Med 4(15):145-167.
Schmidt, K. E., J. M. Kuepper, B. Schumak, J. Alferink, A. Hofmann, S. W. Howland, L. Renia, A. Limmer, S. Specht, and A. Hoerauf. 2018. Doxycycline inhibits experimental cerebral malaria by reducing inflammatory immune reactions and tissue-degrading mediators. PLOS ONE 13(2):e0192717.
Schneiderman, A. I., Y. S. Cypel, E. K. Dursa, and R. M. Bossarte. 2018. Associations between use of antimalarial medications and health among U.S. veterans of the wars in Iraq and Afghanistan. Am J Trop Med Hyg 99(3):638-648.
Schwartz, E., and G. Regev-Yochay. 1999. Primaquine as prophylaxis for malaria for nonimmune travelers: A comparison with mefloquine and doxycycline. Clin Infect Dis 29(6):1502-1506.
Shanks, G. D., P. Roessler, M. D. Edstein, and K. H. Rieckmann. 1995a. Doxycycline for malaria prophylaxis in Australian soldiers deployed to United Nations missions in Somalia and Cambodia. Mil Med 160(9):443-445.
Shanks, G. D., A. Barnett, M. D. Edstein, and K. H. Rieckmann. 1995b. Effectiveness of doxycycline combined with primaquine for malaria prophylaxis. Med J Aust 162(6):306-307, 309.
Shapiro, L. E., S. R. Knowles, and N. H. Shear. 1997. Comparative safety of tetracycline, minocycline, and doxycycline. Arch Dermatol 133(10):1224-1230.
Sharafeldin, E., D. Soonawala, J. P. Vandenbroucke, E. Hack, and L. G. Visser. 2010. Health risks encountered by Dutch medical students during an elective in the tropics and the quality and comprehensiveness of pre- and post-travel care. BMC Med Educ 10:89.
Singh, L. P., A. Mishra, D. Saha, and S. Swarnakar. 2011. Doxycycline blocks gastric ulcer by regulating matrix metalloproteinase-2 activity and oxidative stress. World J Gastroenterol 17(28):3310-3321.
Smith, E. L., A. al Raddadi, F. al Ghamdi, and S. Kutbi. 1995. Tetracycline phototoxicity. Br J Dermatol 132:316-317.
Sonmez, A., A. Harlak, S. Kilic, Z. Polat, L. Hayat, O. Keskin, T. Dogru, M. I. Yilmaz, C. H. Acikel, and I. H. Kocar. 2005. The efficacy and tolerability of doxycycline and mefloquine in malaria prophylaxis of the ISAF troops in Afghanistan. J Infect 51(3):253-258.
Story, M. J., P. I. McCloud, and G. Boehm. 1991. Doxycycline tolerance study. Incidence of nausea after doxycycline administration to healthy volunteers: A comparison of 2 formulations (Doryx® vs Vibramycin®). Eur J Clin Pharmacol 40(4):419-421.
Tan, K. R., A. J. Magill, M. E. Parise, P. M. Arguin, and the Centers for Disease Control and Prevention. 2011. Doxycycline for malaria prophylaxis and treatment: Report from the CDC expert meeting on malaria prophylaxis. Am J Trop Med Hyg 84(4):517-531.
Tan, K. R., S. J. Henderson, J. Williamson, R. W. Ferguson, T. M. Wilkinson, P. Jung, and P. M. Arguin. 2017. Long-term health outcomes among returned Peace Corps volunteers after malaria prophylaxis, 1995-2014. Travel Med Infect Dis 17:50-55.
Tariq, R., J. Cho, S. Kapoor, R. Orenstein, S. Singh, D. S. Pardi, and S. Khanna. 2018. Low risk of primary Clostridium difficile infection with tetracyclines: A systematic review and metaanalysis. Clin Infect Dis 66(4):514-522.
Taylor, W. R., T. L. Richie, D. J. Fryauff, C. Ohrt, H. Picarima, D. Tang, G. S. Murphy, H. Widjaja, D. Braitman, E. Tjitra, A. Ganjar, T. R. Jones, H. Basri, and J. Berman. 2003. Tolerability of azithromycin as malaria prophylaxis in adults in northeast Papua, Indonesia. Antimicrob Agents Chemother 47(7):2199-2203.
Terrell, A. G., M. E. Forde, R. Firth, and D. A. Ross. 2015. Malaria prophylaxis and self-reported impact on ability to work: Mefloquine versus doxycycline. J Travel Med 22(6):383-388.
Thillainayagam, M., and S. Ramaiah. 2016. Mosquito, malaria and medicines—A review. Res J Pharm Technol 9(8):1268-1276.
Tickell-Painter, M., N. Maayan, R. Saunders, C. Pace, and D. Sinclair. 2017. Mefloquine for preventing malaria during travel to endemic areas. Cochrane Database Syst Rev 10:CD006491.
Trumble, C. 2005. Oesophageal stricture in cats associated with use of the hyclate (hydrochloride) salt of doxycycline. J Fel Med Surg 7(4):241-242.
Tuck, J., and J. Williams. 2016. Malaria protection in Sierra Leone during the Ebola outbreak 2014/15: The UK military experience with malaria prophylaxis, Sep 14–Feb 15. Travel Med Infect Dis 14(5):471-474.
Turner, R. B., C. B. Smith, J. L. Martello, and D. Slain. 2014. Role of doxycycline in Clostridium difficile infection acquisition. Ann Pharmacother 48(6):772-776.
Tzianetas, I., F. Habal, and J. S. Keystone. 1996. Short report: Severe hiccups secondary to doxycycline-induced esophagitis during treatment of malaria. Am J Trop Med Hyg 54(2):203-204.
Vilkman, K., S. H. Pakkanen, T. Laaveri, H. Siikamaki, and A. Kantele. 2016. Travelers’ health problems and behavior: Prospective study with post-travel follow-up. BMC Infectious Diseases 16(1):328.
Wallace, M. R. 1996. Malaria among United States troops in Somalia. Am J Med 100(1):49-55.
Waner, S., D. Durrhiem, L. E. Braack, and S. Gammon. 1999. Malaria protection measures used by in-flight travelers to South African game parks. J Travel Med 6(4):254-257.
Watanasook, C., P. Singharaj, V. Suriyamongkol, J. J. Karwacki, D. Shanks, P. Phintuyothin, S. Pilungkasa, and P. Wasuwat. 1989. Malaria prophylaxis with doxycycline in soldiers deployed to the Thai–Kampuchean border. Southeast Asian J Trop Med Public Health 20(1):61-64.
Zhao, S., C. Choksuchat, Y. Zhao, S. A. Ballagh, G. A. Kovalevsky, and D. F. Archer. 2009. Effects of doxycycline on serum and endometrial levels of MMP-2, MMP-9 and TIMP-1 in women using a levonorgestrel-releasing subcutaneous implant. Contraception 79(6):469-478.