C

Literature Review—Report Summaries by Body System

This appendix provides detailed descriptions of the reports summarized in Chapter 4, which were identified from the evidence review strategy outlined in Appendix B. Of the 60,000 papers identified in the search, the committee chose papers that most appropriately linked healthcare utilizations to impairment severity, disability, and disease severity. The Social Security Administration (SSA) defines severe impairment and disability similarly, the difference being the extent of a person’s inability to perform tasks over some period, that is, to perform any gainful activity in the case of severe impairment as opposed to performing substantial gainful activity in the case of disability (SSA, 2017).¹ Other organizations define disability differently. Disease severity refers to the presence and extent of disease in the human body, which can be evaluated through diagnostic testing and physiologic examination (Finlayson et al., 2004). Although the statement of task specifies that the committee should focus on impairment severity and SSA’s definition of disability, the literature focuses on measures of disease severity. Thus, most of the papers included in this appendix link health-care utilizations with disease severity.

The committee focused on summarizing papers that were based on nationwide data, large samples, and study populations of working-age adults in the United States, but for body systems on which there was a paucity of literature, the committee also described papers that did not fit those criteria, such as ones that looked at mortality as an end point, ones that used study populations outside the United States, and ones that had a mean population age over 65 years. Because this was not a systematic review, papers were not graded, but each included study is described in detail. The number of studies included for each body-system summary is not proportional to the number of studies that the committee found in its literature search for each body system. Rather, the committee included papers it believed would be helpful in addressing the statement of task.

Definitions of severe impairment and disability varied in the literature reviewed by the committee. The committee’s summaries used the terms as they were reported in the original articles.

___________________

¹ See CFR § 404.1505 for the full definition of disability.

MUSCULOSKELETAL SYSTEM

The committee initially identified 197 abstracts of potential interest and retrieved 77 full-text articles for review. It searched for a variety of musculoskeletal disorders and injuries, but most of the literature that the committee found was related to fractures. The six studies most relevant to the committee’s task are described below.

In a retrospective cross-sectional study, Menendez and Ring (2015) analyzed the Nationwide Emergency Department Sample (NEDS) database, developed for the Healthcare Cost and Utilization Project (HCUP), of 285,661 people admitted to emergency departments (EDs) for proximal humerus fracture in 2010 and 2011 to identify predictors of hospital admission compared with direct discharges to home. Multivariate logistic-regression modeling showed that although hospital admission was associated with such measures of clinical severity as the presence of polytrauma or open fracture and a higher Charlson comorbidity index, many other, nonclinical factors were also predictive, such as Medicare and Medicaid insurance, metropolitan facilities, and residence in the Northeast even of people who had less severe injuries.

Young et al. (2009) conducted a retrospective database review of 4,889 workers’ compensation (WC) cases with bone fracture in nine states to examine interactions between health-care utilization, work disability, rurality, and duration of work. Two measures of severity were used: multiple versus single body-part injury and closed versus open fracture. Work disability duration was measured by the total number of full days individuals were paid for time off work due to injury. After adjusting for age, sex, state, body part injured, industry, and occupation, Young et al. found that the relationship between health-care utilization and duration of disability or injury severity was influenced by whether the person lived in a rural or urban area. In particular, rural residents have shorter periods of work disability than urban residents when health-care utilization is low; urban residents have shorter disability when health-care utilization is high. Thus, health-care utilization does not predict work disability across the board; other factors affect health-care utilization to render it an insufficient proxy for disability or duration of disability in people who are receiving WC.

Young and colleagues (2015) reviewed a subset of the previously mentioned dataset to look at the relationship between the use of physical medicine and rehabilitation (PM&R) service and work disability score as well as differences between urban and rural residents. The subset included 2,216 people who had bone fracture, received physical therapy, and took at least 7 days off work. The study differentiated urban versus rural patterns of PM&R utilization based on four factors: number of services per week prior to returning to work (rural residents averaged fewer services), whether the patient used two or more passive PM&R services per week in weeks 1 to 8 of disability (less common for rural residents), the mean number of active PM&R services received in weeks 5 to 8 (higher for rural residents), and whether the patient received three weekly passive PM&R services in weeks 5 to 8 (more common for rural residents). After adjusting for such individual and injury characteristics as age, sex, state, site of injury, and occupation, they found that those who used health-care services in a rural setting returned to work faster than those who used services in an urban pattern. Longer disability duration was associated with long episodes of care, failure to transition to self-management, and passive services. Their work focused on methods of reducing health-care costs and, like their previous study, focused on WC cases.

A prospective study by Nemunaitis and colleagues (2016) considered predictors of functional outcomes after trauma. Although they focused on functional outcomes, their study of 879 mostly white male trauma survivors at a level 1 trauma center looked at injury-severity scores (ISSs) and Glasgow Coma Scale (GCS) scores of patients admitted to the ED. They found that older people and those who had government insurance were more likely to have poor discharge functional independence measure (FIM) scores than those who had commercial insurance, and that those who had lower ISS had higher discharge FIM scores. The study focused on functional outcomes, and one could surmise that those who had lower ISS and higher FIM scores used fewer health-care services; but the study did not address health-care utilizations, and any such conclusion would be a projection of the data presented.

Lipscomb et al. (2015) designed a retrospective cross-sectional study of whether declines in WC rates represent true improvement in health or a shift of care to other payment systems. The study excluded people who had multiple work-related injuries. In a sample of 18,768 mostly male carpenters in Washington State, private health-care utilization increased as WC claims declined. The report does not discuss long-term disability or return to work.

A retrospective chart review of a level 1 trauma center at Grady Memorial Hospital in Atlanta, Georgia, examined reasons for increased length of stay of patients 18–55 years old (mean age, 27 years) who were admitted for femoral-shaft fractures (Pendleton et al., 2007). Analyses were adjusted for demographics, including age and sex. The authors found that greater length of stay could be due to social reasons, such as homelessness; medical reasons, such as continued bleeding from surgical wounds, anemia that prevents progression with physical therapy, or infection; or hospital delays, such as time to surgery, time to be admitted to a rehabilitation facility, or radiology delays.

SPECIAL SENSES AND SPEECH

The committee searched for studies that relate health-care utilization with impairment severity for disorders of special senses and speech. It found three abstracts and retrieved all three for full-text review. No studies on health-care utilizations as indicators of impairment severity were identified. One study on ED visits by deaf patients is presented to provide evidence on types of utilization that are more or less probable for particular medical conditions.

A retrospective cohort study in an outpatient center in Rochester, New York, compared 200 randomly selected deaf users of American Sign Language (ASL) with 200 randomly selected hearing English speakers (McKee et al., 2015). The authors compared ED use between deaf and nondeaf patients. The random samples were not matched on demographics or comorbidities. After adjusting for age, sex, race, smoking history, and Medicaid status, McKee et al. found that the deaf ASL users were 97 percent more likely than the hearing controls to have utilized an ED in the preceding 36 months. The study did not review which factors led to increased ED utilization by deaf people. The study results suggest that deafness could be a comorbid condition that increases the likelihood of ED visits.

RESPIRATORY DISORDERS

The committee’s literature search yielded 139 abstracts; the committee read 62 full-text papers and found 8 of them to be useful. Most of the studies focused on chronic obstructive pulmonary disease (COPD), and one focused on asthma.

Alcazar et al. (2012) identified factors associated with hospital admission for COPD, including respiratory symptoms, pulmonary function tests, anxiety and depression, and quality of life. In a multicenter, cross-sectional study of 127 patients (mostly males) whose mean age was 67 years, the authors analyzed numerous factors associated with hospital admission by using a logistic regression model. They found that a significantly greater percentage of those hospitalized than of those not hospitalized received mucolytics, oxygen therapy, and home mechanical ventilation. Significantly more hospitalized patients than nonhospitalized patients had a dyspnea grade of 3 or 4. Lung function was also lower in patients who had been hospitalized when measured in milliliters of forced expiratory volume in one second (FEV1). The researchers found differences in forced vital capacity, which was significantly lower in patients who required hospitalization. As measured by the EQ-5D index, a standardized instrument for measuring generic health status, quality of life was significantly lower in the hospitalized than in the nonhospitalized patients. Anxiety and depression tended to be lower among hospitalized patients, but differences were not statistically significant. The BODE index (a capacity index for COPD that stands for body-mass index, airflow obstruction, dyspnea, and exercise) and SpO2 (partial oxygen saturation) were significantly lower in those hospitalized for COPD exacerbation and indicated greater severity of illness than those not hospitalized for COPD exacerbation. This study directly associates COPD hospital admissions with various measures of severity and quality of life.

Ekberg-Aronsson et al. (2008) sought to determine whether GOLD² stages (another measure of COPD severity) are associated with hospital admission rate. Higher GOLD stages indicate more severe COPD indicated by lower FEV1. On the basis of data from the Swedish Malmo Preventive Programme (MPP), the researchers studied 22,044 middle-aged people who participated in health screening in 1974–1992 and obtained hospital admission data on them until 2002. The researchers analyzed the association between hospital admission rate and GOLD stages, adjusting for age. The authors found that among smokers, hospital admission rates due to all causes were associated with higher GOLD stage. The results of the study provide further evidence that COPD hospitalization is associated with measures of disease severity, although the study involved a non-US population, and access to and threshold to admit in European hospitals are different from the US.

Fan et al. (2007) identified factors that predict COPD exacerbations, defined as hospitalization or ED visit for COPD. The researchers used data from the National Emphysema Treatment Trial (NETT), a randomized controlled trial of two methods of treatment for emphysema conducted in 1998–2002. They linked NETT clinical-trial data with utilization data from Medicare claims. The study population consisted of 610 patients with an average age of 66.5 years. COPD exacerbation was defined as a COPD-related ED visit or hospitalization with a primary discharge diagnosis code of ICD-9 (International Classification of Diseases, Ninth Edition) 491, 492, 493, or 496. Multivariate logistic regression was used to predict the outcome

___________________

² GOLD stands for the Global Initiative for Chronic Obstructive Lung Disease. GOLD was launched in 1997 in collaboration with the National Heart, Lung, and Blood Institute (NHLBI is part of the National Institutes of Health) and the World Health Organization.

of COPD hospitalization or ED visit during the 1-year followup period. They found that BODE score, a measure of disease severity, predicted 1-year hospitalizations and ED visits for COPD. As for functional measures, dyspnea measured with a shortness-of-breath questionnaire was the most predictive of exacerbations. The study used a large, well-characterized sample, but the average age was higher than SSA’s population of interest. Given those qualifications, the study showed that hospitalizations and ED visits for COPD exacerbations are associated with illness severity and measures of lung function.

Mullerova et al. (2015) studied outcomes and risk factors associated with hospitalizations for COPD exacerbation. The authors used data on 2,138 patients, with a mean age of 63 years, from the Evaluation of COPD Longitudinally to Identify Predictive Surrogate Endpoints (ECLIPSE) study, a 3-year longitudinal observational study conducted at 46 centers in 12 countries to characterize disease progression in COPD (Vestbo et al., 2008). They investigated factors associated with hospitalization for COPD exacerbations by using Cox proportional hazards and adjusting for a wide array of demographic and clinical variables. They also studied the association between time to first hospitalization for exacerbation during the study period and hospitalization for exacerbation before study entry. Mullerova et al. (2015) found that the factor most predictive of a hospitalization for exacerbation was a prior history of hospitalization for exacerbations (hazard ratio 2.71, p <0.001). Other factors that predicted hospitalization for exacerbations included severity of airflow limitation, poor health status, radiologic evidence of emphysema, and systemic inflammation.

Akazawa et al. (2008) conducted a retrospective case-control study design of trends in utilization for COPD patients compared with control. The authors examined medical and pharmacy claims data on more than 30 million people in the United States. The authors used data from United Healthcare, a large managed care plan in the United States. All plans in the analysis provide full insurance coverage for physician, hospital, and prescription drug services. COPD patients 40 years old and older were matched with three random controls by age, sex, region, and index date. COPD cohort members were identified on the basis of billing claims that had primary diagnoses that indicated COPD (ICD-9 code 491.xx, 492.xx, 496.xx) or pharmaceutical claims for fluticasone propionate–salmeterol combination, salmeterol, ipratropium, or tiotropium. Trends in utilization 36 months before diagnosis were compared by using multivariate regression models between COPD patients and controls. COPD patients used 1.6 times more inpatient and ED services and 1.5 times more office visits than controls, after adjustment for age, sex, region, and most common comorbid conditions. COPD patients’ health-care utilization experienced a marked increase in the month before diagnosis.

Quintana et al. (2015) identified predictors of length of stay of patients who had COPD exacerbations. They conducted a prospective cohort study of 1,453 patients who visited 16 EDs and were admitted to the hospital. The authors performed multivariate multilevel linear and logistic regression to find predictors of length of stay of COPD patients. The authors found that the best predictors of length of stay of COPD patients were baseline dyspnea, physical activity levels, and fatigue at 24 hours since admission; intensive care unit or intensive respiratory care unit admission; the need for antibiotics; and complications during hospitalization.

Sharif et al. (2014) studied readmission for COPD. The objective of their study was to determine predictors of readmission within 30 days of patients 40–64 years old who were hospitalized for COPD. They performed logistic regression of a retrospective cohort study by using a large national commercial database (N = 8,263). They found that early readmission was associated with patient factors, provider factors, and systems factors. Patient factors included

history of heart failure, lung cancer, osteoporosis, and depression. Provider factors included absence of prescription of statins within 12 months of hospitalization and absence of prescription of short-acting bronchodilators. System factors included length of stay and lack of postdischarge followup. The study described what affects intervals between hospitalizations for COPD patients and showed the complexity of predicting readmission for COPD.

Bai et al. (2007) investigated whether hospitalizations for worsening asthma or serious and reversible reduction in FEV1 are associated with a decline in lung function. A prospective European cohort of 281 adults who had predominantly moderate persistent asthma was initially evaluated in 1962–1975 and then re-evaluated in 1990 for measures of lung function. Measures of lung function included bronchial responsiveness, serum immunoglobulin E, and detailed lung-function tests. Participants were divided into two groups as having “frequent” and “infrequent” exacerbations, depending on how they compared with the median. The researchers analyzed the effect of hospitalizations and severe exacerbations on the annual decline of FEV1 by using a linear mixed-effects model. They found that the average FEV1 decline was 16.9 mL/year greater in the “frequent” group than in the “infrequent” group (p = 0.03). The study demonstrates that frequent exacerbations in asthmatics, including exacerbations that involve hospitalization, might be associated with lung-function decline. (Note that it was based on a European population.)

CARDIOVASCULAR SYSTEM

The committee initially identified 261 abstracts of potential interest and retrieved 116 full articles for review. It found more literature on cardiovascular diseases (CVDs) than any other body system in its literature search. Some articles provided an overview of CVDs, but most focused on specific cardiovascular conditions. The committee reports on 23 studies that focused on overall CVDs, heart failure (HF), ischemic heart disease (IHD), arrhythmias, and valvular disease as examples of studies that investigated disability associated with these disorders. For clarity, the 23 studies summarized in this section are grouped according to those categories.

General Cardiovascular Disease

A large prospective observational population-based study of Florida Medicaid patients compared a group of 15,775 patients who participated in a disease management intervention with 32,034 patients who received usual care (Afifi et al., 2007). All patients had at least one of the following chronic conditions: diabetes mellitus, congestive heart failure (CHF), hypertension, and asthma. The intervention consisted of telephone counseling by a trained managed-care specialist in 2001–2004. In models adjusted for demographics, severity of disease, comorbidities, and previous utilization, significantly lower rates of annual hospitalizations (ranging from 0.07 to 0.38 stays), lower lengths of stay (0.4 to 2.54 days), and fewer ED visits (0.10 to 0.91 visits) were reported in the disease management group. The study emphasizes the role that good patient counseling—even by telephone—can have in reducing expensive inpatient and emergency care. It also supports the idea that less expensive out-of-hospital followup might have a favorable effect on outcomes and that variation among patients and health-care systems might have to be considered in assessing inpatient utilization as a proxy for functionality.

Using data from the Baltimore Study of Black Aging, Thorpe et al. (2016) investigated self-reported activities of daily life in 602 black Americans who had chronic conditions, including CVD and diabetes. In models adjusted for age, education, income, and marital status,

men who had CVD had almost 3 times as high a risk of disability as men who did not report CVD. Inasmuch as people who had arthritis, depression, or diabetes were also found to have increased disability, the study documented the role of chronic conditions in functionality and suggested that multiple chronic comorbidities must be addressed when disability is evaluated.

Fan et al. (2009) studied the role of psychologic distress associated with disability in adults who had cardiovascular conditions. Data from the 2007 Behavioral Risk Factor Surveillance System, which collects data through random digit dial telephone surveys, on 177,663 respondents who were 35 years and older were used to evaluate disability on the basis of self-reports of activity limitations and use of special equipment. Overall, 12.5 percent had a history of CVD, and 9.2 percent had severe psychologic distress. People who had a history of CVD had a higher prevalence of psychologic distress than those who did not. Moreover, disability status was significantly greater in those who had psychologic distress than in those who did not, regardless of the rate of rehabilitation services received. No differences related to sex or employment status were found in levels of psychologic distress in those who had CVD after adjustment for other sociodemographic factors. The data confirm that psychologic factors must be considered to affect disability in people who have CVD.

Heart Failure and Cardiomyopathy

A 2013 systematic review evaluated the effects of social factors on hospital readmissions or mortality in which HF was an initial diagnosis (Calvillo-King et al., 2013). Some 52 studies published in 1980–2012 met criteria for inclusion. Greater age was associated with worse HF outcomes, but results by sex were inconclusive. Race influenced outcomes, with nonwhites having more readmissions but lower mortality. In studies that evaluated social factors, low socioeconomic status (SES, generally defined by income or Medicaid status), rural residence, home instability, lack of social support, being unmarried, and such medical factors as smoking, drug use, and medical visit nonadherence significantly increased the risk of readmissions. Those factors and psychiatric comorbidities, lack of home resources, and greater distance to a hospital were associated with increased mortality from HF. This systematic review emphasizes the need to consider a broad array of demographic and social factors as influences on hospital readmissions for HF.

Acknowledging that early readmission or death might indicate poor management for HF, Foraker and colleagues (2011) assessed the association of neighborhood median household income and Medicaid status with HF hospital readmissions of 1,342 participants in the Atherosclerosis Risk in Communities (ARIC) study followed over 17 years. They adjusted for covariates—race and study community, sex, age at incident HF hospitalization, and selected socioeconomic, clinical, and behavioral characteristics—and found that within 1 year 19 percent of patients had died, 59 percent had been readmitted, and 62 percent had been readmitted or died. Incident HF hospitalizations were more common among ARIC cohort participants who lived in low- and medium-neighborhood income areas than among those who lived in high-neighborhood income areas at baseline. Low-neighborhood income participants who had a high comorbidity index score at the time of the incident HF hospitalization were readmitted at a higher rate than high-neighborhood income participants in the same comorbidity category. Medicaid recipients who did not have a high burden of comorbidity tended to have a higher risk of first readmission and were readmitted more often than participants who were not receiving Medicaid. The study concluded that comorbidity burden appears to modify the association of neighborhood factors, Medicaid status, readmission, and death of HF patients.

Although quality of life is assumed to be intricately associated with functionality, few studies have attempted to investigate health-care utilization as a predictor. Acknowledging the value of information on frequency of hospitalizations, one study investigated inpatient stays as an approximation of patient functional status (Berchialla et al., 2010). A total of 235 patients participating in the Heart Muscle Disease Study Group who were treated in the Maggiore Hospital Department of Cardiology in Trieste, Italy, and had a diagnosis of dilated cardiomyopathy were followed over 15 years, from 1978–1992. Using a semi-Markov representation of the hospital process, the authors calculated that the probability of a second hospitalization within 1 year of the first is about 0.50 and within 2 years about 0.30. After the third hospitalization, the probability of not having another hospitalization within 1 year was 0.10. Use of beta blockers was a primary factor in increasing the interval between hospitalizations. The authors noted the importance of extracting as much detailed information as possible from hospital records. They emphasized that although the hospitalization process is only a rough approximation of patient status, it might be a reasonable approach with such diseases as dilated cardiomyopathies that have relatively fast worsening.

As noted previously, quality of life is often used to indicate patients’ functionality in epidemiologic studies. A study of 1,458 patients who participate in the Efficacy of Vasopressin Antagonism in HF Outcome with Tolvaptan study (EVEREST) used the Kansas City Cardiomyopathy Questionnaire (KCCQ)³ scores after discharge followed an HF admission to evaluate quality of life (Allen et al., 2011). Within 24 weeks of discharge, 478 (32.8 percent) patients had died and 192 (13.2 percent) had serial KCCQ scores under 45 (scores range from 0 to 100). After adjustment for 23 predischarge covariates, independent predictors of quality of life included a low admission KCCQ score, high brain type natriuretic peptide (BNP), hyponatremia, tachycardia, hypotension, absence of beta-blocker therapy, and history of diabetes mellitus and arrhythmia. Of interest is the role of BNP as a predictor of HF outcome, independently of other clinical factors and comorbidities. The authors noted that those predictors can be used to target aggressive treatment options for HF patients, but they also identify new measures that might be helpful in identifying patient functionality.

Only one study that specifically addressed functional limitations and readmission of HF patients was found (Yamada et al., 2012). In 215 patients who completed the Performance Measure for Activities of Daily Living-8 (PMADL-8) (higher scores indicating worse functionality) over a mean followup of 20 months, multivariate analyses resulted in only the PMADL-8 score associated with readmission for HF (hazard ratio [HR]: 2.49, 95% confidence interval [CI]: 1.27–4.90). Those in the highest functionality group had fewer events than others. Results indicate the prognostic value of self-reported physical function and its importance in predicting hospital readmissions.

In an approach similar to that above, Betihavas et al. (2015) developed a risk-prediction model for unplanned readmissions for HF. A prospective cohort of 280 patients in the Which Heart Failure Intervention Is Most Cost-Effective and Consumer Friendly in Reducing Hospital Care (WHICH?) trial, comparing home-based versus clinic-based interventions, was followed to document 18-month readmissions of participants. Factors associated with an increased risk of hospitalization for HF included age (HR: 1.07, 95% CI: 0.90–1.26) for each 10-year increase in age, living alone (HR: 1.09, 95% CI: 0.74–1.59), sedentary lifestyle (HR: 1.44, 95% CI: 0.92–-

___________________

³ The Kansas City Cardiomyopathy Questionnaire (KCCQ) is a well-established instrument used to evaluate the health status of heart failure (HF) patients (see Creber et al., 2012).

2.25), and the presence of multiple comorbid conditions (HR: 1.69, 95% CI: 0.38–7.58). This model confirms the importance of including comorbidities in assessing HF outcomes.

Because HF outcomes are usually reported in studies of older adults, the committee includes a study by Bibbins-Domingo et al. (2009), which addressed racial differences in incident HF in young adults. A cohort of 5,115 blacks and whites of both sexes who were 18–30 years old were followed for up to 20 years to identify the incidence and risk factors for HF. Only 27 participants developed HF, and all but one were black. Mean age at onset was 39 ± 6 years, which resulted in a cumulative incidence of HF before the age of 50 years of 1.1 percent (95% CI: 0.6–1.7). In blacks, risk factors for incident HF were higher diastolic blood pressure, higher body mass index, lower high-density lipoprotein cholesterol, and kidney disease. Myocardial infarction (MI), drug use, and alcohol use were not associated with the risk of HF. The study is an important contribution to the literature on working-age adults, identifying the race-related disparity in HF, but social factors were not addressed.

In contrast, a study tracked the natural history of HF matched white and nonwhite patients by using propensity scores (Gambassi et al., 2008). Of the 7,788 participants, who had chronic systolic and diastolic HF, enrolled in the Digitalis Investigation Group trial, 14.4 percent were nonwhite. Their propensity scores were used to match 1,018 pairs of white–nonwhite dyads on the basis of baseline characteristics, including 35 clinical measures of comorbidities and severity indexes. No measures of SES were included. Over a median of 38 months of followup, no racial differences were found in mortality (HR: 0.95, 95% CI: 0.80–1.14), all-causes hospitalization (HR: 1.03, 95% CI: 0.90–1.18), all-cause mortality (HR: 0.82, 95% CI: 0.6–1.11), or HF mortality (HR: 1.5, 95% CI: 0.91–1.22). The study found an increased risk of hospitalization for worsening HF among nonwhite HF patients. The authors concluded that racial difference in HF did not exist after controlling for a multitude of baseline factors, but the clinical measures reflected only baseline indicators of disease at the time of HF. Inasmuch as racial differences existed in baseline characteristics before matching, it is important to recognize that race might be a marker of other prognostic covariates.

A study of 78,801 patients from 257 hospitals participating in the AHA Get with the Guidelines-HF Program (2005–2008) evaluated HF outcomes and included 22.6 percent blacks and 6.0 percent Hispanics (Thomas et al., 2011). Clinical characteristics, adherence to guideline-based HF measures, and in-hospital mortality were addressed. Relative to white patients, Hispanic and black patients were younger (median age, 78.0, 63.0, and 64.0 years, respectively), had lower left ventricular ejection fractions, and were more likely to have diabetes mellitus and hypertension. Black and Hispanic patients had lower in-hospital mortality than white patients. In the context of a national HF quality-improvement program, HF care was equitable and improved in all racial and ethnic groups over time.

Ischemic Heart Disease

A large study evaluated national trends in MI hospitalization rates in patients under 65 years old by using the National Inpatient Sample (NIS) 1997–2006 (Towfighi et al., 2011). Temporal trends in sex-specific hospitalization rates were assessed in 2,824,615 US patients who were admitted with MI. The demographic and hospital characteristics for MI admissions in 1997 and 2006 were similar; in both years, most patients were white and male and had private insurance. Most MI admissions were to urban hospitals, and most patients were admitted through the ED. Among those who were 35–64 years old, men consistently had greater MI hospitalization rates than women. The temporal trends showed reductions in MI hospitalization

rates in both men and women, with a slightly greater reduction in men. The age-standardized rate decreased during the decade by 25 percent and 18 percent, from 168 to 126 and from 56 to 46 MI hospitalizations per 100,000 men and women, respectively. The absolute reduction in MI hospitalization rates was greater in men than in women. The most prominent decrease occurred in men 55–64 years old. The authors noted that the lower MI hospitalization rates in women than in men could be due partly to a failure to recognize and diagnose acute coronary syndromes in women. Sex should be taken into consideration inasmuch as severity of and prognosis in MI can vary between men and women in US hospitalizations.

A study of adverse events and later health-care utilization by patients who had a previous hospitalization for acute coronary syndrome (ACS) used data from a large commercial managed-care system in the United States (Korsnes et al., 2015). Patients were at least 18 years old at initial ACS hospitalization (the index episode) and had at least 12 months of continuous health-plan enrollment before and after the end of the index episode. Of 75,231 study patients identified, 3.3 percent had a serious adverse CVD event and 8.3 percent a second coronary event during the 12-month followup. Median time to first adverse and second coronary event from the end of the index episode was 4.6 and 3.7 months, respectively. Statin use and lower age were associated with lower episode-related costs. The study documented the relatively short time between initial hospitalization and recurrence for patients initially hospitalized with an ACS.

A study was undertaken to evaluate socioeconomic status (SES) on the basis of residential zip code in relation to prehospital clinical, access, and transport variables that might influence outcomes after acute MI (Agarwal et al., 2014). Using 372,984 discharges with a principal diagnosis of acute MI in the 2003–2011 NIS database, the study found significantly higher mortality in the lowest SES quartile than in the highest quartile (odds ratio [OR]: 1.11, 95% CI: 1.06–1.1). The mean adjusted cost of hospitalization was almost 4 times as high in patients in the highest SES quartile than in the lowest quartile. The authors concluded that disparities related to factors incorporated in SES must be considered when outcomes related to acute MI are evaluated.

Arrhythmia

An evaluation of health-care utilization associated with new-onset atrial fibrillation (AF) (less than 30 days after MI) and late-onset AF (at least 30 days after MI) was conducted among 1,512 patients enrolled in the Rochester Epidemiology Project (southeast Minnesota) with a mean followup of 3.9 years (Chamberlain et al., 2013). Hospitalizations and ED and outpatient visits at the Mayo Clinic in 2002–2010 were documented in the 237 patients who had prior AF, 163 who had new-onset AF, 113 who had late-onset AF, and 989 who did not have AF. Patients who had AF were older at the index MI and had a greater number of comorbidities; in particular, those who had AF before MI had worse renal function and were more likely to have prevalent HF and COPD than those who developed AF after MI or those who did not have AF. Rates of utilization differed in those who had AF from those who did not on the basis of timing of the arrhythmia. In fully adjusted models, those who had prior AF exhibited an increased risk of hospitalization (HR: 1.57, 95% CI: 1.29–1.91), those who had new-onset AF exhibited an increased risk (HR: 1.34, 95% CI: 1.11–1.63), and late-onset AF was associated with a greater increase in risk (HR: 2.18, 95% CI: 1.67–2.83) compared with people who did not have AF. A similar pattern was observed for both ED visits and outpatient visits, although with slightly smaller HRs. Absolute rates of utilization were not presented. The study provided evidence of AFs complicating MI and resulting in higher utilization associated with adverse outcomes and

poor prognosis. The combination of CVD outcomes and management of coexisting conditions should be considered in assessing disability.

The ARIC study also provided data on utilization associated with AF (Bengtson et al., 2014). Participants who had incident AF enrolled in fee-for-service Medicare for at least 12 continuous months during 1991–2009 (N = 932) were matched for age, sex, race, and field center with up to three participants who did not have AF (N = 2,729). During a mean followup of 4.1 years, there were 2,604 hospitalizations of the 932 AF participants; the median length of stay was 5 days (interquartile range: 3–9 days). There were 2,965 hospitalizations of the 2,729 AF participants during a mean followup of 4.2 years; the median length of stay was 5 days (interquartile range: 3–8 days). The unadjusted mean number of days of hospitalization per year was 13.2 (95% CI: 11.6–15.0) and 2.8 (95% CI: 2.5–3.1) for AF and non-AF participants, respectively. After adjustment for potential confounders, the rate of days in the hospital was 3.94 (95% CI: 3.29–4.73) times greater for AF participants. The unadjusted annual rate of outpatient utilization was 53.3 (95% CI: 50.5–56.3) and 22.9 (95% CI: 22.1–23.8) for AF and non-AF participants, respectively. Health-care utilization after AF diagnosis did not differ significantly by sex or race. Although the results did not provide correlations between utilization and functional status, they are useful in understanding both inpatient and outpatient utilizations by patients who had AF. The rates of utilization might have value in refinement of the SSA Listings. It should be noted that most participants in the ARIC study were over 65 years old.

Two studies reviewed rates of hospital readmission after inpatient admission for AF. Kim et al. (2009) conducted a retrospective cohort analysis by using data from the Healthcare Information Systems National Managed Care Benchmark Database (2002–2006) on 4,174 patients. Information on the first readmission within 1 year of the index hospitalization was analyzed. Overall, 12.5 percent of chronic AF patients were readmitted for AF with a mean time to readmission of 142.5 days (median, 108 days). Among newly diagnosed AF patients, 10.1 percent were readmitted for AF with a mean time to readmission of 133.8 days (median, 112 days). For chronic AF, 17.6 percent, 43.4 percent, and 65.8 percent of readmissions occurred within 1, 3, and 6 months, respectively, versus 22.7 percent, 44.5 percent, and 67.2 percent, respectively, for newly diagnosed AF. Opolski et al. (2015) investigated readmissions and repeat procedures after catheter ablation for atrial fibrillation or flutter (AF/AFL). These data were on 2,022 patients enrolled in the National Health Fund of Poland. After discharge for the index hospitalization, 123 (6.1 percent) and 540 (26.7 percent) patients were hospitalized because of AF/AFL within 30 days and 1 year, respectively. At 1-year followup, 192 (9.5 percent) patients underwent AF/AFL ablations. The patients that underwent the second ablation were younger than the ones who did not (56.6 ± 11.0 years versus 59.1 ± 10.8 years; p = 0.019) and the time of the index hospitalization was shorter (3.75 ± 2.16 days versus 4.45 ± 3.26 days; p = 0.03). Within 30 days and at the 1-year followup, 194 (9.6 percent) and 747 (36.9 percent) patients, respectively, were hospitalized. The data showed that more than one-fourth of patients who underwent AF/AFL ablation were hospitalized for arrhythmia recurrence in 1 year. Although information to associate the utilizations with disability was not sufficient, the study documents the high burden of inpatient care required for patients with AF.

Valvular Heart Disease

Using longitudinal data from the NIS, Badheka et al. (2015) describes trends in hospitalizations from 2000 to 2012 that resulted from aortic valve–related discharges in people 60 years old and older. The NIS, developed for the HCUP and sponsored by the Agency for

Healthcare Research and Quality (AHRQ), is the largest available database on hospital inpatient stays in the United States. Over the 12-year period, 113,847 hospitalizations with aortic valvular disease as the primary diagnosis (N = 561,880 for weighted national estimate) were reported in the United States in patients over 60 years old. The number of hospitalizations increased progressively from 7,213 in 2000 to 11,531 in 2012 (p <0.001). Hospitalized patients who had aortic valvular disease were predominantly non-Hispanic whites (69 percent) and over 70 years old (76 percent). Admissions of patients who had more comorbidities doubled from 25 percent in 2000 to 50 percent in 2012. The most frequent coexisting conditions in these patients were hypertension (58 percent), heart failure (35 percent), renal failure and anemia (26 percent each), diabetes (24 percent), and chronic pulmonary disease (18 percent). There were more admissions in the South (29 percent) than in any other region, followed by the Northeast (26 percent), the Midwest (22 percent), and the West (20 percent). Overall in-hospital mortality of patients hospitalized for aortic valvular (AV) disease significantly decreased from 4.5 percent in 2000 to 3.5 percent in 2012 (p <0.001). The total length of hospital stay has decreased from 8.4 to 7.8 days over this period (p <0.001); the cost of care during hospitalization increased significantly: from $1.28 billion in 2001 to $2.13 billion in 2011 (p <0.001) after adjustment for inflation. With the availability of more treatment options, an aging population, and the increasing cost of hospitalization, the burden associated with aortic valvular disease is of growing importance in the United States, but the effects on adults under 65 years old and return-to-work capacity are not known.

Only one study that evaluated outcomes in younger adults who had congenital valvular heart disease (VHD) (van der Linde et al., 2013) was found. A total of 414 patients (average 29 years, 68 percent male) were selected from prospective databases: the CONCOR database (the Dutch registry of adult patients who have chronic heart disease [CHD]) and the Leuven and Toronto database (of adults who have CHD). Patients were followed over a median duration of 4.1 (2.5–5.1) years. Chart abstraction was used to evaluate baseline and progression of aortic stenosis (AS). Increased left ventricular mass was significantly associated with faster AS progression (p <0.001). Aortic dilatation was present in 34 percent at baseline and 48 percent at followup (p <0.001). The rate of aortic dissection was 0.06 percent per patient-year. Some 70 patients required an aortic-valve intervention (4.4 percent per patient-year), with AS progression rate as the most powerful predictor (HR: 5.11, 95% CI: 3.47–7.53). The authors concluded that in young patients with mild to moderate congenital AS, the disease generally does not progress. However, people who have left ventricular hypertrophy are at risk for fast disease progression and should be monitored carefully. The results of the study might provide insights into VHD outcomes for working-age adults.

Many studies of VHD have focused on a specific (often new) procedure to compare outcomes of usual care. Severe AS, the most common acquired VHD, which has a poor prognosis once symptoms appear, is an example. In a review of 16 studies, Deutsch et al. (2013) investigated outcomes beyond survival, including functional status and health-related quality of life, and compared transcatheter aortic valve replacement (TAVR) with surgical aortic valve replacement. Reviewing metrics for assessing both physical components (such as physical function and body pain) and mental components (such as emotional health and social function) of health in 16 studies, the authors concluded that evidence is accumulating that TAVR in high-risk surgical patients who have severe symptomatic AS is associated with marked functional benefits, but most of these patients are elderly. Significant improvements in quality of life were detectable as early as 1 month after TAVR and were followed by clinical stabilization and were

detectable up to 1 year. The 1-year health status of the TAVR population has consistently been shown to become similar to age-matched general population norms. Whereas most studies involved older patients who had multiple comorbidities, disabilities, and short life expectancy, this review emphasized the value of using quality-of-life evaluation in assessing the outcome of therapeutic interventions in clinical practice.

Biomarkers also might prove valuable in objectively evaluating the health status of VHD patients after surgical procedures. Mizutani and colleagues (2017) sought to investigate the 2-year prognostic effect of BNP concentration at discharge after TAVR. In 1,094 patients followed during 2013–2016, the discriminating BNP concentration for discerning 2-year mortality was 202 pg/mL. High BNP was also found to have a statistically significant effect on other net outcomes, with an adjusted HR of 2.28 (1.36–3.82, p = 0.002). This study was conducted in older adults and needs to be duplicated in younger populations, but it suggests that incorporation of BNP stratification with other clinical variables might substantially improve predictive accuracy of outcomes of interventions for VHD.

DIGESTIVE SYSTEM

The committee initially identified 67 abstracts of potential interest and retrieved 24 full-text articles for review. The six studies most relevant to the committee’s task are discussed below.

The committee identified one study that associated severity of a digestive condition with work productivity and resource utilization (Wahlqvist et al., 2008). It used data from the 2004 National Health and Wellness Survey, a self-administered, Internet-based questionnaire that captured health and behavior data on US adults. A total of 10,028 respondents who had gastroesophageal reflux disease (GERD) were matched to an equal number of age- and sex-matched controls. Symptom severity was self-assessed as mild, moderate, or severe. Work productivity was measured by using the Work Productivity and Activity Impairment questionnaire. Resource utilizations examined included physician visits, ED visits, hospitalizations, and GERD-related prescription-drug use. The authors found that as symptom severity increased, there was a significant increase in some health-care utilizations and a decrease in productivity by some measures. Specifically, for physician visits, there was a significant increase in health-care use from mild to both moderate and severe symptoms; for ED visits, there was a significant increase from mild to severe symptoms only; for hospitalizations, there was an increase for each level of severity that was not statistically significant; and for use of prescription proton pump inhibitors and histamine-2 receptor antagonists, there was a significant increase from mild to both moderate and severe symptoms. With respect to productivity, “hours absent from work per week” differed significantly between mild and severe symptoms. Although the study shows that increased severity is associated with an increase in some health-care utilizations by US patients who have GERD, the severity measure here is self-reported symptom severity rather than SSA’s definition of impairment severity.

Another study focused on pain related to chronic pancreatitis. Mullady et al. (2011) studied whether patients who had chronic pancreatitis with various pain patterns experienced resource utilization, quality of life, and disability differently. This was a prospective cohort study in 20 US medical centers that consisted of 540 patients. The authors found that chronicity of pain, rather than severity of pain, was associated with greater rates of resource utilization and disability and lower quality of life. Patients in the cohort who had constant pain were more likely

to miss more than 5 days of work per month and more than three hospitalizations per year; however, it is unclear whether disability was associated with loss of productivity or other factors.

Dudekula et al. (2011) also focused on pain frequency, but for gastroparesis. The objective of their study was to determine predictors of hospital stays of patients who had gastroparesis. The authors conducted a retrospective chart review of 326 patients who were admitted to a hospital in Pittsburgh for gastroparesis in 2004–2008. They determined gastroparesis admission by using the ICD-9 code 536.3 and found that more clinic visits took place in year 1 than in years 2–5 (p <0.01), whereas ED visits and hospitalization rates did not differ much over time. Nausea, vomiting, and abdominal pain were listed as primary concerns for greater than 70 percent of hospitalizations and ED visits. The length of hospital stay differed significantly based on the etiology of gastroparesis and the presence of psychiatric comorbidity. Factors that did not affect length of stay included chronic pain and opioid use. The report of the study provides some insight into trends in utilization for gastroparesis, although it does not specifically address impairment severity.

In a Swiss study of 1,187 patients followed up for an average of 13 months, Siebert et al. (2013) sought to identify predictors of work disability in patients who had inflammatory bowel disease (IBD). In Crohn’s disease, such disease-related characteristics as fistulizing disease, duration of disease, number of relapses, and response to therapy were found to be associated with loss of work productivity and with disability.

Allegretti et al. (2015) studied risk factors for readmission within 90 days in patients who had IBD. They conducted a retrospective analysis of 356 patients at a hospital in Boston. IBD diagnosis was considered to include ulcerative colitis (ICD-9 code 556.X) or Crohn’s disease (555.X). They developed a Cox proportional-hazards model around two covariates of particular interest: depression and steroid use in the preceding 6 months. The final Cox model showed three variables that were risk factors for admission: depression (HR: 1.99, CI: 1.33–3.00), chronic pain (HR: 1.88, CI: 1.14–3.10), and steroid use in the last 6 months (HR: 1.33, CI: 0.92–2.04).

Myer et al. (2013) used the 2007 NEDS sample (N = 15 million) to identify leading causes of ED visits due to digestive diseases. They used logistic regression to analyze predictors of hospitalization after an ED visit. Leading primary diagnoses were abdominal pain, nausea and vomiting, and functional disorders of the digestive system. ED visits resulting in hospitalization were higher for primary digestive diagnoses versus nondigestive visits (21.6 percent versus 14.7 percent).

GENITOURINARY DISORDERS

The committee initially identified 67 abstracts of potential interest and retrieved 26 full-text articles for review. The four studies most relevant to the committee’s task are discussed below.

One study provided evidence from national data that physician visits and hospitalizations can predict severity of chronic kidney disease (CKD) (Alexander et al., 2009), but it used a medical definition of severity rather than SSA’s definition. There is also some evidence that worsening renal function (WRF) is associated with length of stay and readmission rate (Lanfear et al., 2011) and that some measures of utilization, when combined in a complex model, can predict WRF (Perkins et al., 2013).

Looking at National Health and Nutrition Examination Survey data on adult participants (N = 15,258), Alexander et al. (2009) studied the association between CKD and health-resource

utilization on the basis of self-reported physician visits and hospitalizations. They found that the mean number of annual physician visits increased significantly with CKD stage. That pattern was most evident in Mexican Americans, those under 65 years old, those who had Medicaid insurance, and those who had hemoglobin concentrations under 11 g/dL. Hospitalizations followed the same pattern. The pattern was most pronounced in Mexican Americans, younger adults, those who had body mass indexes (BMIs) less than 25, those who had hemoglobin concentrations less than 11 g/dL, diabetics, and those who did not report regular exercise.

Lanfear et al. (2011) conducted a retrospective analysis of 2,456 patients who had diagnoses of heart failure from 2000 to 2008 in a large health system in Michigan to look at long-term outcomes and readmission related to WRF. They found after adjustment that WRF, defined as a creatinine increase of at least 0.3 mg/dL, was associated with increased length of stay (7.0 days versus 3.8 days, p <0.001) and increased rates of readmission (p <0.001). Perkins et al. (2013) conducted a retrospective analysis to create a tool to predict 30-day readmissions for heart failure in people who had non-dialysis-dependent CKD. The final model included a number of variables, including resource utilization, comorbid conditions, medications, and laboratory results. Resource-utilization predictors included admission through the ED, admission as a transfer from another hospital, number of clinic visits in the 3 months before admission, discharge to a facility other than the home, and length of stay.

A study relating renal and cardiac disease looked at acute kidney injury (AKI) as a severity index for 30-day readmission after cardiac surgery (Brown et al., 2014). The authors collected data on 2,209 patients who underwent coronary artery bypass grafting (CABG) or valve surgery in seven hospitals in New England during 2008–2010. They evaluated the association between stages of AKI and 30-day readmission by using multivariate logistic regression. AKI stage 1 was defined as a 50 percent increase in serum creatinine concentration, stage 2 as a 100 percent increase, and stage 3 as a 200 percent increase if the baseline serum creatinine was at least 4.0 mg/dL or if it was a new dialysis. The results showed that those who did not develop AKI had a 9.3 percent 30-day readmission rate, those who had AKI stage 1 had a 16.1 percent readmission rate, those who had stage 2 had a 21.8 percent readmission rate, and those who had stage 3 had a 28.6 percent readmission rate (p <0.001). Thus, AKI stage could be a severity marker for 30-day readmission of patients who were undergoing CABG or valve surgery.

HEMATOLOGIC DISORDERS

The committee found 12 abstracts and retrieved all 12 for full-text review. The six studies most relevant to the committee’s task are discussed below; all are related to sickle-cell disease (SCD).

Aisiku et al. (2009) collected diary data on ED utilization and pain descriptors and analyzed laboratory severity variables of 232 patients who had sickle-cell disease. People who have SCD are often stigmatized as opioid-seeking overutilizers of EDs. The authors found that about 35 percent were high users of the ED. High users had lower hematocrit levels, more transfusions, more pain days, more pain crises, higher mean pain and distress, and worse quality of life than other SCD patients on the basis of the Medical Outcome Study 36-Item Short Form Health Survey physical-function summary scales. After controlling for severity and frequency of pain, high ED utilizers were found not to use opioids more frequently than other SCD patients. The authors correlate ED utilization with several possible measures of disease severity.

In a trend analysis of SCD patients in Delaware, Anderson et al. (2014) examined hospitalizations, readmissions, and the cost of admissions for the study population. The study included adult residents who had SCD and received acute care in a small community hospital or a large community hospital system in 2007–2009. The authors ran descriptive statistics on age, length of stay, and hospital costs. There were 518 acute-care episodes during the period. They found that hospital costs and hospital readmissions within 7 and 30 days were rising while length of stay for acute care remained constant. The conclusions were that the cost of admissions for crises was high and that readmissions were common.

Brousseau et al. (2010) studied ED visits and hospital admissions for sickle-cell crises in a retrospective cohort study of more than 20,000 subjects over a 2-year period by using HCUP data from 2005–2006. Their primary outcome measures were hospital and ED utilization and readmission rate. They found frequent use of acute care encounters, defined as hospital and ED visits, by SCD patients. More than half the study population experienced acute care encounters during the study period. In total, the study population had more than 100,000 such encounters and averaged 2.59 encounters per patient per year, including 1.52 hospitalizations and 1.08 ED visits. Utilization was highest for people 18–30 years old. For those hospitalized, the 30-day readmission rate was 33.4 percent. The high 30-day readmission rate for sickle-cell crisis is a potential indicator of impairment severity, but the authors did not address this.

Leschke et al. (2012) performed a retrospective study of Wisconsin Medicaid claims data on adults and children in 2003–2007. Their primary outcome measures were readmission 14 and 30 days after a hospitalization for sickle-cell pain crisis. Of the 408 patients included in the study, 10 percent were readmitted within 14 days, and 17 percent within 30 days. The study also found that outpatient followup after hospitalization for sickle-cell crisis was associated with reduced readmission after both 14 and 30 days. The study did not directly discuss impairment severity; the high rate of readmission after sickle-cell admissions and the reduction in readmission in conjunction with outpatient followup are notable.

In a small study of 70 participants looking at health-care utilization and pain in people who had SCD, Sanders et al. (2010) compared utilization and pain variables between older (37–62 years old) and younger (18–36 years old) adult patients. The age distinction was based on survival data presented by Powars et al. (2005) that indicated a median survival time of sicklecell patients of 36–39 years. The older group had more education; there was no significant difference in employment or marital status between the older and younger groups. The authors found that older sickle-cell patients were more likely to utilize outpatient facilities, whereas younger patients were more likely to use ED services. The two groups reported equivalent pain intensity, as measured by the Brief Symptom Inventory (Derogatis, 1993). The study indicates that health-care utilization by sickle-cell patients varies by age and cannot be generalized to all adults.

Wolfson et al. (2011) used statewide ED discharge data and found that 69 percent of sickle-cell patients in California utilized an ED each year, and they were most likely to use one facility. Over one-third were Medicaid patients. Adults were more likely than children to use multiple facilities and to be uninsured. In another study that used statewide data, the same group (Wolfson et al., 2012) found that adults had a higher frequency of utilization than children (possibly because of the coverage of children, which in California is superior to the overall coverage of adults). Distance from comprehensive SCD care and insurance status were significant predictors of ED utilization for SCD. Those who lived farther from facilities were more likely to use the ED but less likely to be hospitalized. The authors provided further

evidence that health-care utilizations by sickle-cell patients correlated with factors other than impairment severity.

SKIN DISORDERS

The committee identified four abstracts of potential interest and retrieved all four for full-text review. The two studies that are most relevant to the committee’s task are discussed below.

Farrell et al. (2008) looked at the contribution of obesity to functional decline in patients who suffered from acute burns in a group of 221 mostly male patients whose average age was 43 years in a midwestern regional acute burn center. In the sample, 76 percent were discharged to home from the hospital, and the average length of stay was 16 days. The average total body surface area (TBSA) of the burns was 15 percent. The analysis found that in patients more than 57.5 years old who had less than 30.75 percent TBSA burns, BMI played a role in whether they were discharged directly to home versus to an inpatient setting. Of patients who had a BMI greater than 27.4, only 21 percent returned home; of patients who had a BMI of 27.4 or less, 65 percent returned home. The study also looked at FIM—ability to transfer to and from bed, ambulate, dress, and feed oneself. Functional independence was measured on a scale of 1–7 in which 1 indicated a requirement of total assistance and 7 indicating complete independence. The authors found that of patients who were 54.5–72.5 years old and had less than 22.5 percent TBSA burns, those who had a BMI greater than 25.15 had an average FIM score of 3.778, whereas those who had a BMI less than 25.15 had an average FIM score of 5.400. The study made a case that obesity is a comorbidity that can impair functional ability in this population of burn patients, as evidenced by discharge disposition and functional measures.

Kimball et al. (2014) studied the presence of malignancies and hospitalizations of a group of people who had psoriasis, a chronic inflammatory skin condition. In a national sample of more than 40,000 people, the authors calculated the standardized incidence of hospitalizations for severe psoriasis infections and total malignancy rates over a 36-month period. They examined outcomes of people who had psoriasis vis-à-vis their treatment modality. The authors found that people who had psoriasis were at greater risk for lymphoma regardless of treatment: 9 cases per 10,000 person-years compared with 6 cases per 10,000 person-years. There was little difference in the rates of lymphoma in people who had different treatments. The risk of hospitalization was highest in those who were exposed to tumor necrosis factor antagonistic therapies, not phototherapy or no treatment. Thus, psoriasis might be a comorbid condition that indicates increased risk of hospitalization for patients who have lymphoma.

ENDOCRINE DISORDERS

The committee initially identified 25 abstracts of potential interest and retrieved all 25 full-text articles for review. Four studies that looked at the effect of diabetes mellitus (DM) as a comorbid condition on disease severity and hospitalization are discussed below.

Jang et al. (2016) assessed hospitalized patients who survived an EMS-assessed out-of-hospital cardiac arrest in the ED in 2009–2013 (N = 2,651) and who received a diagnosis of DM to determine whether having DM affected survival to discharge and recovery of patients who had cardiac disease. They found that DM had a significant negative association with survival outcomes in those who had cardiac disease but no significant association in those who did not

have cardiac disease. The study provided evidence that survival of patients hospitalized for DM is related to cardiac disease.

Another study (Nadjiri et al., 2016) examined use of coronary computed tomographic angiography in diabetic patients and showed that patients who had a poorer performance on their imaging had a higher event rate when the end point was a composite of cardiac events defined as all-causes death, nonfatal myocardial infarction, or unstable angina requiring hospitalization. Diabetes as a comorbid condition might predict increased hospital and ED utilization for cardiac events.

Another study looked at the association between depression and hypoglycemic episodes that required ED visits or hospitalization (Katon et al., 2013). This longitudinal cohort study of 4,117 diabetic patients used major depression as the exposure of interest and the ICD-9 code for a hypoglycemic episode that required an ED visit or hospitalization as the outcome of interest. The authors found depression to be significantly associated with time to first severe hypoglycemic episode. The study provided further evidence that comorbidities are important in associating health-care utilizations with disease severity in diabetes patients.

Regarding duration of utilization, one study looked at predictors of morbidity among patients after pancreatectomy (Jaap et al., 2016). A retrospective chart review from 2004 to 2013 (N = 180) found that the best predictor of length of stay and surgical complications was sarcopenia.

NEUROLOGIC DISORDERS

The committee initially identified 96 abstracts of potential interest and retrieved 67 full-text articles for review. The eight studies most relevant to the committee’s task are discussed below.

Using data collected in 1998–2009 through the Medical Expenditure Panel Survey, Libby et al. (2012) compared health-care utilization by people with a diagnosis of epilepsy (N = 1,026, weighted to be nationally representative, N = 864,958) with health-care utilization by those who did not have epilepsy (N = 383,090). After covariate adjustment, people who had epilepsy had higher than expected rates of health-care provider visits and medication prescriptions. The relationship between health-care utilization and work-related disability was not evaluated, but those who had epilepsy were less likely to be employed and had more missed work days because of injury or illness. Wage-based lost productivity was greater than that observed in the combination of people who had depression, diabetes, anxiety, or asthma.

A retrospective study using the Medstat Marketscan Commercial Claims and Encounters database showed that people who had multiple sclerosis (MS) (N = 1,411) were 3.5 times as likely as matched healthy controls (N = 7,055)—excluding people who had comorbid conditions—to be hospitalized, 2 times as likely to visit the ED at least once, and 2 times as likely to receive one of the rehabilitation therapies (physical, occupational, or speech therapy) during the year after the first diagnosis or MS medication treatment (Asche et al., 2010). In addition, 30 percent were prescribed MS-specific medications, and, as a group, people who had MS were more likely than healthy controls to be prescribed antipsychotics, antidepressants, anticonvulsants, urinary antibiotics, and amphetamines.

Thomas and Ellis (2013) used data on North Carolina Medicaid recipients who received Supplemental Security Income (SSI) (N = 60,190) to examine whether health-care utilization predicted gainful employment, defined by eligibility code 1619b (earnings level indicating that a

person no longer needs cash payments). People who had any kind of disability were included in this analysis as long as they lived in the community and had received only outpatient services (no inpatient claims). People who had psychiatric or cognitive disability were found to use healthcare services the most. More health-care service days were associated with less likelihood of employment, but the effect size was small. When diagnostic groups were examined relative to employment, those who had developmental disability and other neurologic disorders were among those who had the lowest odds of employment. The relationship between health-care utilization and employment was not evaluated relative to diagnosis.

Jones et al. (2016) utilized a national survey of neurologists and their MS patients (N = 715 patients) in 2013–2014 to evaluate the relationship between health-care utilization and disability. The neurologists used the Expanded Disability Status Scale (EDSS) to evaluate disability, in which a score over 5 indicates severity that can impair activities of daily living (ADLs) and impede working a full day. Some of the patients (N = 335) also answered questions about employment status, work productivity, and time lost at work because of MS. During the preceding 12 months, patients who had EDSS scores over 5 had more encounters with neurologists (incidence rate ratio [IRR] 1.4), nurses (IRR 44.4), physical therapists (IRR 9.9), and urologists (IRR 7.2). They also had more hospitalizations (IRR 3.3). The relationship between health-care utilization and work-related disability was not directly evaluated; however, when compared with patients who had no or minimal disability, patients who had EDSS scores over 5 were more likely to be unemployed (odds ratio [OR]: 12.4), more likely to have had problems in getting a job or promotion (OR: 17.0), or more likely to have had to stop working because of MS (OR: 10.4).

Health-care utilization by and outcomes of patients who have sustained traumatic brain injury (TBI) have also been examined. Collie and Prang (2013) examined trajectories of healthcare utilization over 5 years in Australia by using compensation-claims data on 316 adults who had sustained severe TBI. They found four types of trajectories, two of which showed that health-care utilization remained at a relatively high level at the end of the 5 years. The people who had those trajectories showed greater and increasing use of attendant care to assist with disability. Attendant care is generally provided to assist with self-care, supervision, and health management, so it is unlikely that a person who has attendant care for TBI would be able to work. The study shows some link between health-care utilization and severity of TBI, but the relevance of the findings to work-related disability is even less direct than that found for MS.

Forslund et al. (2013) sought to identify predictors of employment 2 years after moderate to severe TBI (N = 100) and included receipt of rehabilitation services at 1 year among the factors evaluated. Those who were receiving rehabilitation services at 1 year were less likely to be employed at 2 years; this suggests that rehabilitation services are a marker of injury severity. In the multiple-regression analysis, rehabilitation services did not remain a significant predictor.

Zhang et al. (2015) analyzed public datasets from the 1995 National Institute of Neurological Disorders and Stroke tissue plasminogen activator study (N = 605) and found that discharge disposition at the completion of the acute hospitalization for stroke—home versus skilled nursing facility (SNF), rehabilitation, or death—was a strong predictor of disability at 3 months (modified Rankin score: ≥ 3), with an OR for acute rehabilitation versus home of 13.51 and an OR for SNF versus home of 28.50.

Bolge et al. (2010) examined health-care utilization and lost work productivity among those who reported chronic sleep maintenance insomnia characterized by nighttime awakenings (CINA) in the 2006 US National Health and Wellness Survey. Those who had CINA had

significantly greater health-care utilizationh than those who did not, were less likely to be working full time, and, if working, had more absenteeism and presenteeism. The relationship between health-care utilization and work productivity was not directly examined.

MENTAL DISORDERS

The committee initially identified 125 abstracts of potential interest and retrieved 53 full-text articles for review. The two studies that linked health-care utilization with impairment severity associated with mental disorders are discussed below. The majority of the papers identified for mental disorders pertain to mental disorders as a comorbid condition—those 20 papers are discussed more generally in Chapter 4.

Naz et al. (2007) followed a cohort of 87 respondents in New York who were 15–60 years old and had a major depressive disorder with psychotic features over a period of 4 years. They found that poorer utilization of prehospital resources predicted remission but not relapse, whereas medication use was not associated with remission or relapse. In this multivariate analysis, other factors were found to be associated with longer time to remission, including longer intervals between initial episode and hospitalization, lack of insurance, and lower prehospital Global Assessment of Functioning Scale scores.

A study of first-episode schizophrenia patients who participated in a randomized trial of enhanced psychosocial interventions provides direct information about the lack of association of pre-award utilizations and the eventual award of disability compensation (Rosenheck et al., 2017). In the Recovery After an Initial Schizophrenia Episode–Early Treatment Program study, first-episode psychosis patients were randomized to receive treatment as usual (TAU) or an enhanced psychosocial intervention. Over the course of the 2-year period of the study, 34 percent of the participants were awarded Social Security disability benefits. There were no differences between the enhanced intervention and TAU in the proportion of cases awarded benefits, although the enhanced intervention led to improvements in quality of life and symptoms. Predictors of receiving a disability award during the treatment period were less education, being unemployed or not going to school, less likelihood of private health insurance, longer duration of untreated psychosis, greater age, more clinician-rated disability, and higher positive symptom ratings on the Postive and Negative Syndrome Scale. Thus, more severe illness predicted disability compensation, and the intervention, although successful in some ways, did not change the trend toward compensation.

CANCER

The focus of the literature search for cancer is not on a particular body system but rather on diagnoses that might affect many organs or systems, although many articles tend to address a single site. The committee identified 99 abstracts and reviewed 28 articles thoroughly. The 14 that provide information of potential value to the committee’s task are discussed below.

Rahman and colleagues (2015) followed 196 consecutive patients at the DanaFarber/Brigham and Women’s Cancer Center who had newly diagnosed glioblastoma to estimate their risk of hospitalization after treatment. Median overall survival of patients was 15.6 months. Some 46 percent were hospitalized during the chemoradiation phase of their treatment because of generalized weakness (17 percent), seizures (16 percent), or venous thromboembolism (13

percent). Hospitalization during chemoradiation was associated with a 47 percent increase in mortality. The data provided here demonstrate that morbidity associated with treatment for a cancer diagnosis is often the source of inpatient admissions.

To understand rates of readmission after surgery for gastric cancer, Merchant et al. (2015) followed 8,887 patients for 90 days after curative-intent surgery. Of them, 29 percent were readmitted to the hospital as inpatients at followup, most of them in the first 30 days after discharge, and almost 14 percent of the 29 percent had more than one readmission (range, 2 to 10 readmissions). After adjustment for other factors, readmission was associated with worse 5-year overall survival with an increased risk of death of 40 percent (HR: 1.4; 95% CI: 132–149). Readmission was increased by 15 percent in patients who had one comorbidity and 45 percent in those who had two or more. Although these results document the high rate of hospital readmissions in patients after gastric surgery, the statistics do not provide numbers that are useful in helping to decide the number of readmissions that constitutes disability or inability to work.

Karhade (2016) investigated 30-day readmissions after surgery for primary and secondary spinal tumors in a study that used a national registry to evaluate incidence and predictors of readmissions, adverse events, and reoperations. Data on 2,207 patients from the National Surgical Quality Improvement Program registry were analyzed to identify predictors of study outcomes on the basis of demographics, tumor characteristics, preoperative functional status, comorbidities, laboratory measures, and hospital factors related to the surgery. Readmission occurred in 10.2 percent of patients a median of 18 days after surgery. The most common causes were surgical-site infections (23.7 percent), systemic infections (17.8 percent), venous thromboembolism (VTE, 12.8 percent), and central nervous system complications (11.9 percent). Predictors of readmission included comorbidities (dyspnea, hypertension, and anemia), disseminated cancer, prior steroid use, and extended surgical hospitalization. Major complications occurred in 14.4 percent of patients: primarily VTEs, surgical-site infections, and sepsis, which were related to dependent functional status, emergency status, male sex, and comorbidities. The 30-day mortality was 3.3 percent. Those results highlight the role of comorbidities in postoperative sequelae of cancer surgery and the need to consider increased surveillance after hospital discharge.

Several papers from the same study team (Braithwaite, 2010; Izano, 2013, 2014) described functional limitations of women who had breast cancer. In a prospective cohort study of 2,202 women who had stage I, II, or III breast cancer diagnosed in 1997–2000 in the Kaiser Permanente Cancer Registry or the Utah Cancer Registry (Braithwaite, 2010), the long-term prognostic role of functional limitations was described, and complete information on body functions was collected, including endurance, strength, range of motion, and small-muscle dexterity after adjuvant treatment. During a followup of 9 years, 269 deaths (12.2 percent of the cohort) occurred, 7 percent from breast cancer and 5 percent from competing causes. Functional limitations were found in 39 percent of patients and were statistically associated with mortality, although the risk was greater for stage I than for stage III tumors. Women who had functional limitations tended to be older, less educated, and obese. The authors concluded that physical function in breast cancer patients was an important predictor of survival irrespective of clinical, lifestyle, and sociodemographic factors and should be addressed to improve longevity and quality of life of survivors. It is important to understand that disability often precedes death of cancer patients.

Two papers by Izano and colleagues described the effects of functional limitations and decline on mortality in black and white women. In a cohort of 975 women who had newly

diagnosed breast cancer in the Detroit-based Health and Functioning in Women Study (Izano et al., 2013), all-causes mortality increased significantly in women who had any functional limitation and specifically in those who had difficulties in pushing or pulling large objects, writing, handling small objects, or walking a half-mile. However, functional decline was associated with breast cancer mortality in regional or remote but not localized disease. Results did not differ by race, but the relationships were strongest in overweight and obese women. Data on 999 women taken from the same cohort who were followed for a median of 11 years (Izano et al., 2014) showed higher mortality in black women who had greater functional limitations and later-stage breast cancer. Comorbidities were associated with other-causes mortality and did not differ by race.

A study of 647 cancer survivors who were 55–65 years old and enrolled in the Penn State Cancer Survivor Study investigated the long-term effects of cancer and its treatment on employment and productivity to quantify disability that is attributable to cancer and to compare rates of disability in cancer survivors with rates in similarly aged people who have other conditions from the Health and Retirement Study (HRS) (Short et al., 2008). Work disability was measured as self-reported impairment or health problems that limited the kind or amount of paid work that subjects could do, a standardized metric strongly associated with employment status. Greater than 70 percent of cancer survivors had been employed at the time of their diagnosis, and about half reported that their disability was related to the cancer itself. The study found the rate of work disability in cancer-free survivors to be almost twice as high as in people who had no chronic conditions. Similar rates of disability, however, were found in cancer survivors and in those who had other chronic conditions, such as heart disease, stroke, diabetes, and lung disease; this highlights the chronic nature of effects even in successfully treated cancer patients. The authors concluded that cancer survivorship should be viewed as a chronic condition that requires a broad array of social services. The study is important in that it documented the effects of cancer on work disability, although it noted that, unlike other chronic conditions, disability is commonly caused by treatment that results in chronic pain and poor functional outcomes rather than by the disease itself.

Another study reported that cancer survivors under 65 years old were more than 3 times more likely to be unable to work because of a health condition than those who had no history of cancer or other chronic diseases (Hewitt et al., 2003). Using the National Health Interview Survey for 1998–2000, the authors compared 4,878 cancer survivors with 90,737 people who had no history of cancer to review general health status, psychologic disability, limitations in ADLs, physical function, and health-related inability to work. Those who had a history of cancer were more likely to have three or more chronic conditions and higher rates of psychologic problems, ADL difficulties, and functional limitations. Almost 17 percent reported being unable to work, and an additional 7 percent had health-related work limitations. Of those who had a history of cancer, 19.5 percent received SSI disability benefits. The authors also noted that cancer survivors who had comorbid chronic conditions had a much higher likelihood of disability than those without comorbid chronic conditions, although the odds of having the functional limitations described here remained higher than the odds in noncancer respondents even when they did not have other chronic conditions. Use of physician care was assessed and found to be similar to that by controls, but hospitalizations were not assessed. The authors noted that inasmuch as cancer is primarily a disease of the elderly, the aging of the US population will increase disabilities and the need for supportive services to address them. The study did not

identify health-care utilizations that would be useful in distinguishing cancer survivors’ disabilities, but it did yield evidence that disabilities are likely to increase.

Another national population-based sample was used to estimate the burden of illness of cancer survivors compared with people who did not have a history of cancer (Yabroff et al., 2004). Multiple measures of disease burden—including utility (a summary measure of health status in multiple domains of health-related quality of life), lost productivity, and functional limitations—were all higher in 1,823 cancer survivors than in 5,469 age-, sex-, and education-matched controls. Cancer survivors were less likely to have held a job in the preceding month, more likely to have been unable to work because of health, more limited in the type of work that they could do, and had lost more days from work than controls. Of the cancer survivors, 18 percent had been unable to work in the preceding month, 27 percent were more limited in the kind of work that they could do, and had lost 50 days from work in the preceding year; productivity varied among cancer types. That length of time since diagnosis did not alter results reflects a poor prognosis for complete recovery. Health-care utilization was not reported in this study, but it confirmed the employment limitations that result from a cancer diagnosis.

Bradley and Bednarek (2002) took a different approach in analyzing data on cancer survivors at earlier stages who might be younger and more likely to return to work. Their goal was to investigate employment patterns of 253 long-term cancer survivors registered in the Detroit Surveillance, Epidemiology, and End Results Program (SEER) population. Patients were interviewed 5–7 years after their diagnosis of breast, prostate, colon, or lung cancer. Telephone surveys used questions on employment taken from two other studies, the health and retirement study and the Current Population Survey. The patients had a mean age of 62 years; data on demographics, employment status, hours worked, reasons for not working, absenteeism, job changes, and issues regarding retirement were collected. More than two-thirds of patients who had been working at the time of their diagnosis were employed 5–7 years later. Of patients who were no longer actively employed, 54 percent had retired, 24 percent had left their jobs because of poor health, and 9 percent were not working because their business had closed. The authors concluded that although it was clear that cancer had imposed some degree of limitation on some patients, especially in physically demanding jobs, the employment outlook for many patients will be promising as screening becomes more routine and working-age adults’ conditions will be diagnosed earlier.

Other studies have focused on quality of life (QOL) as a patient outcome. Reduced mental and physical QOL was found to be significantly higher in patients who had several types of cancer (prostate, colorectal, and spinal), in particular demographic groups, and in people who had particular types of treatment (Sharma et al., 2007; Charlton et al., 2015; Choi et al., 2016; Farris et al., 2017). Those papers described risk factors for reduced QOL of cancer survivors, but direct associations with health-care utilization were not reported.

IMMUNE-SYSTEM DISORDERS

The committee initially identified 36 abstracts of potential interest and retrieved 33 full-text articles for review. The six studies most relevant to the committee’s task are discussed below.

Two studies looked at ED visits of and disease severity in patients who had immune disorders. Panopalis et al. (2010) conducted a retrospective cohort study (N = 807) in San Francisco of patients who had systemic lupus erythematosus (SLE) to determine predictors of

frequent ED use. One factor examined was disease severity as determined by score on the Systemic Lupus Activity Questionnaire (SLAQ). Their data show that SLE patients who had more than three ED visits per year had higher disease severity than those who had one or two ED visits per year. SLE patients who had zero visits per year had the lowest disease severity. Another study examined ED utilization by human immunodeficiency virus (HIV) patients (Josephs et al., 2010). In this study, 951 patients were asked during in-person interviews to recall their ED visits in the preceding 6 months. The researchers then identified factors associated with the ED visits and admission to the hospital from the ED. They found that the likelihood of 6-month ED visits was 1.4 times higher among acquired immunodeficiency syndrome (AIDS) patients who had low CD4 (cluster of differentiation 4) counts (< 50/mm³) than among patients who had normal CD4 counts. The likelihood of 6-month ED visits was 2.2 times higher among disabled than among nondisabled AIDS patients and 1.3 times higher in unemployed than in employed AIDS patients. Those findings suggest that frequency of ED utilization might reflect disease severity in patients who have immune diseases.

Two studies linked hospitalizations with disease severity as defined by clinical manifestations of AIDS, such as CD4 count. Kerr and colleagues (2012) conducted a retrospective analysis of 2,454 patients who were infected with AIDS. They found that HIV patients who had the clinical manifestations of AIDS for more than 1 year were more frequently hospitalized and spent more total days in the hospital than patients who had AIDS for less than 1 year, who in turn were more frequently hospitalized and spent more total days in the hospital than asymptomatic HIV patients. In a prospective cohort study of HIV-infected patients at 11 HIV care sites around the United States, Yehia et al. (2012) found that the annual rate of hospitalization was more than 5 times higher among AIDS patients who had low CD4 counts (<50/mm³) than among AIDS patients who had normal CD4 counts. Similarly, the annual risk of hospitalization was nearly 2 times higher among AIDS patients who had a high viral load (>100,000 copies) than among those who had a normal viral load. Those studies provided evidence that greater clinical severity is associated with increased hospital utilization by HIV and AIDS patients.

Given that SLE is associated with one of the highest readmission rates among chronic diseases, Yazdany et al. (2014) examined hospital discharge data from five states—31,903 SLE patients, including 9,244 readmissions within 30 days of discharge—to identify predictors of readmission. One predictor found was that patients readmitted to the hospital within 30 days of discharge had significantly greater disease severity (as determined by the Ward Index) than those who were not. Thus, hospital readmission could be associated with disease severity in SLE patients.

One study identified by the committee linked outpatient-clinic utilization with disease severity. A cluster analysis of data on 1,748 HIV-infected patients at a large medical center found that those who had fewer than two clinic visits per year were less likely to have a favorable viral load and less likely to have a favorable CD4 count than those who had three or more visits per year (Palma et al., 2015). Those who had more than six visits per year were also less likely to have a favorable CD4 count. The study suggested that frequency of outpatient-clinic use might reflect disease severity in HIV patients.

The committee’s literature search found one study that linked prescription medication utilization with disease severity in people who had immune disorders. Fielden and colleagues (2008) conducted a retrospective study of 1,605 HIV-infected patients who initiated highly active antiretroviral therapy (HAART) in 1996–2001. Their results showed that AIDS patients

who had less than 95 percent adherence to HAART were 1.88 times more likely to be hospitalized than those with higher rates of medication adherence. The study thus suggested a link between adherence to HAART and disease severity as measured by hospitalization of AIDS patients.

REFERENCES

Berchialla, P., F. Foltran, A. Di Lenarda, G. Barbati, L. Franchin, L. Salvatore, and D. Gregori. 2010. Hospitalization rate as a proxy for the quality of life evaluation: The case of dilated cardiomiopathy. Journal of Preventive Medicine and Hygiene 51(4):152–156.

Betihavas, V., S. A. Frost, P. J. Newton, P. Macdonald, S. Stewart, M. J. Carrington, Y. K. Chan, and P. M. Davidson. 2015. An absolute risk prediction model to determine unplanned cardiovascular readmissions for adults with chronic heart failure. Heart, Lung and Circulation 24(11):1068–1073.

Bibbins-Domingo, K., M. J. Pletcher, F. Lin, E. Vittinghoff, J. M. Gardin, A. Arynchyn, C. E. Lewis, O. D. Williams, and S. B. Hulley. 2009. Racial differences in incident heart failure among young adults. New England Journal of Medicine 360(12):1179–1190.

Bolge, S. C., V. N. Joish, R. Balkrishnan, H. Kannan, and C. L. Drake. 2010. Burden of chronic sleep maintenance insomnia characterized by nighttime awakenings. Population Health Management 13(1):15–20.

Bradley, C. J., and H. L. Bednarek. 2002. Employment patterns of long-term cancer survivors. Psychooncology 11(3):188–198.

Braithwaite, D., W. A. Satariano, B. Sternfeld, R. A. Hiatt, P. A. Ganz, K. Kerlikowske, D. H. Moore, M. L. Slattery, M. Tammemagi, A. Castillo, M. Melisko, L. Esserman, E. K. Weltzien, and B. J. Caan. 2010. Long-term prognostic role of functional limitations among women with breast cancer. Journal of the National Cancer Institute 102(19):1468–1477.

Brousseau, D. C., P. L. Owens, A. L. Mosso, J. A. Panepinto, and C. A. Steiner. 2010. Acute care utilization and rehospitalizations for sickle cell disease. JAMA 303(13):1288–1294.

Brown, J. R., C. R. Parikh, C. S. Ross, R. S. Kramer, P. C. Magnus, K. Chaisson, R. A. Boss, Jr., R. E. Helm, S. R. Horton, P. Hofmaster, H. Desaulniers, P. Blajda, B. M. Westbrook, D. Duquette, K. LeBlond, R. D. Quinn, C. Jones, A. W. DiScipio, and D. J. Malenka. 2014. Impact of perioperative acute kidney injury as a severity index for thirty-day readmission after cardiac surgery. The Annals of Thoracic Surgery 97(1):111–117.

Calvillo-King, L., D. Arnold, K. J. Eubank, M. Lo, P. Yunyongying, H. Stieglitz, and E. A. Halm. 2013. Impact of social factors on risk of readmission or mortality in pneumonia and heart failure: Systematic review. Journal of General Internal Medicine 28(2):269–282.

Chamberlain, A. M., S. J. Bielinski, S. A. Weston, W. Klaskala, R. M. Mills, B. J. Gersh, A. Alonso, and V. L. Roger. 2013. Atrial fibrillation in myocardial infarction patients: Impact on health care utilization. American Heart Journal 166(4):753–759.

Charlton, M. E., K. B. Stitzenberg, C. Lin, J. A. Schlichting, T. R. Halfdanarson, G. Y. Juarez, J. F. Pendergast, E. A. Chrischilles, and R. B. Wallace. 2015. Predictors of long-term quality of life for survivors of stage 2/3 rectal cancer in the cancer care outcomes research and surveillance consortium. Journal of Oncology Practice 11(4):e476–e486.

Choi, B. Y., D. M. DiNitto, C. N. Marti, and N. G. Choi. 2016. Impact of mental health and substance use disorders on emergency department visit outcomes for HIV patients. The Western Journal of Emergency Medicine 17(2):153–164.

Collie, A., and K. H. Prang. 2013. Patterns of healthcare service utilisation following severe traumatic brain injury: An idiographic analysis of injury compensation claims data. Injury 44(11):1514–1520.

Creber, R. M., R. Polomano, J. Farrar, and B. Riegel. 2012. Psychometric properties of the Kansas City Cardiomyopathy Questionnaire (KCCQ). European Journal of Cardiiovasculoar Nursing 11(2):197–206.

Derogatis, L. R. 1993. Brief symptom inventory: Administration, scoring, and procedures manual. Minnesota: National Computer Systems.

Deutsch, M. A., S. Bleiziffer, Y. Elhmidi, N. Piazza, B. Voss, R. Lange, and M. Krane. 2013. Beyond adding years to life: Health-related quality-of-life and functional outcomes in patients with severe aortic valve stenosis at high surgical risk undergoing transcatheter aortic valve replacement. Current Cardiology Reviews 9(4):281–294.

Dudekula, A., M. O’Connell, and K. Bielefeldt. 2011. Hospitalizations and testing in gastroparesis. Journal of Gastroenterology and Hepatology 26(8):1275–1282.

Ekberg-Aronsson, M., K. Lofdahl, J. A. Nilsson, C. G. Lofdahl, and P. M. Nilsson. 2008. Hospital admission rates among men and women with symptoms of chronic bronchitis and airflow limitation corresponding to the gold stages of chronic obstructive pulmonary disease—a population-based study. Respiratory Medicine 102(1):109–120.

Everhart, J. E., and C. E. Ruhl. 2009. Burden of digestive diseases in the United States part iii: Liver, biliary tract, and pancreas. Gastroenterology 136(4):1134–1144.

Fan, A. Z., T. W. Strine, R. Jiles, J. T. Berry, and A. H. Mokdad. 2009. Psychological distress, use of rehabilitation services, and disability status among noninstitutionalized US adults aged 35 years and older, who have cardiovascular conditions, 2007. International Journal of Public Health 54 (Suppl 1):100–105.

Fan, V. S., S. D. Ramsey, B. J. Make, and F. J. Martinez. 2007. Physiologic variables and functional status independently predict COPD hospitalizations and emergency department visits in patients with severe COPD. Journal of Chronic Obstructive Pulmonary Disease 4(1):29–39.

Farrell, R. T., R. L. Gamelli, R. F. Aleem, and J. M. Sinacore. 2008. The relationship of body mass index and functional outcomes in patients with acute burns. Journal of Burn Care & Research 29(1):102–108.

Farris, M. S., K. A. Kopciuk, K. S. Courneya, S. E. McGregor, Q. Wang, and C. M. Friedenreich. 2017. Identification and prediction of health-related quality of life trajectories after a prostate cancer diagnosis. International Journal of Cancer 140(7):1517–1527.

Fielden, S. J., M. L. Rusch, B. Yip, E. Wood, K. Shannon, A. R. Levy, J. S. Montaner, and R. S. Hogg. 2008. Nonadherence increases the risk of hospitalization among HIV-infected antiretroviral naive patients started on HAART. Journal of the International Association of Physicians in AIDS Care 7(5):238–244.

Finlayson, T. L., C. A. Moyer, and S. S. Sonnad. 2004. Assessing symptoms, disease severity, and quality of life in the clinical context: A theoretical framework. American Journal of Managed Care 10(5):336–344.

Foraker, R. E., K. M. Rose, C. M. Suchindran, P. P. Chang, A. M. McNeill, and W. D. Rosamond. 2011. Socioeconomic status, Medicaid coverage, clinical comorbidity, and rehospitalization or death after an incident heart failure hospitalization: Atherosclerosis risk in communities cohort (1987 to 2004). Circulation: Heart Failure 4(3):308–316.

Forslund, M. V., C. Roe, J. C. Arango-Lasprilla, S. Sigurdardottir, and N. Andelic. 2013. Impact of personal and environmental factors on employment outcome two years after moderate-to-severe traumatic brain injury. Journal of Rehabilitation Medicine 45(8):801–807.

Gambassi, G., S. A. Agha, X. Sui, C. W. Yancy, J. Butler, G. Giamouzis, T. E. Love, and A. Ahmed. 2008. Race and the natural history of chronic heart failure: A propensity-matched study. Journal of Cardiac Failure 14(5):373–378.

Hewitt, M., J. H. Rowland, and R. Yancik. 2003. Cancer survivors in the United States: Age, health, and disability. The Journals of Gerontology. Series A, Biological Sciences and Medical Sciences 58(1):82–91.

Izano, M., W. A. Satariano, R. A. Hiatt, and D. Braithwaite. 2013. The impact of functional limitations on long-term outcomes among African-American and white women with breast cancer: A cohort study. British Medical Journal Open 3(10):e003232.

Izano, M., W. A. Satariano, M. C. Tammemagi, D. Ragland, D. H. Moore, E. Allen, A. Naeim, M. E. Sehl, R. A. Hiatt, K. Kerlikowske, O. Sofrygin, and D. Braithwaite. 2014. Long-term outcomes among African-American and white women with breast cancer: What is the impact of comorbidity? Journal of Geriatric Oncology 5(3):266–275.

Jaap, K., M. Hunsinger, J. Dove, K. McGinty, E. Stefanowicz, J. Fera, J. Wild, M. Shabahang, and J. Blansfield. 2016. Morphometric predictors of morbidity after pancreatectomy. The American Surgeon 82(12):1221–1226.

Jang, D. B., S. D. Shin, Y. S. Ro, K. J. Song, K. O. Ahn, S. S. Hwang, Y. T. Kim, S. O. Hong, and J. A. Choi. 2016. Interaction of the diabetes mellitus and cardiac diseases on survival outcomes in out-of-hospital cardiac arrest. American Journal of Emergency Medicine 34(4):702–707.

Jones, E., J. Pike, T. Marshall, and X. Ye. 2016. Quantifying the relationship between increased disability and health care resource utilization, quality of life, work productivity, health care costs in patients with multiple sclerosis in the US. BMC Health Services Research 16:294.

Josephs, J. S., J. A. Fleishman, P. T. Korthuis, R. D. Moore, and K. A. Gebo. 2010. Emergency department utilization among HIV-infected patients in a multisite multistate study. HIV Medicine 11(1):74–84.

Karhade, A. V., V. S. Vasudeva, H. H. Dasenbrock, Y. Lu, W. B. Gormley, M. W. Groff, J. H. Chi, and T. R. Smith. 2016. Thirty-day readmission and reoperation after surgery for spinal tumors: A national surgical quality improvement program analysis. Neurosurgical Focus 41(2):E5.

Katon, W. J., B. A. Young, J. Russo, E. H. Lin, P. Ciechanowski, E. J. Ludman, and M. R. Von Korff. 2013. Association of depression with increased risk of severe hypoglycemic episodes in patients with diabetes. The Annals of Family Medicine 11(3):245–250.

Kerr, J. C., T. G. Stephens, J. J. Gibson, and W. A. Duffus. 2012. Risk factors associated with inpatient hospital utilization in HIV-positive individuals and relationship to HIV care engagement. Journal of Acquired Immune Deficiency Syndromes 60(2):173–182.

Kim, M. H., J. Lin, M. Hussein, and D. Battleman. 2009. Incidence and economic burden of suspected adverse events and adverse event monitoring during AF therapy. Current Medical Research and Opinion 25(12):3037–3047.

Kimball, A. B., J. Schenfeld, N. A. Accortt, M. S. Anthony, K. J. Rothman, and D. Pariser. 2014. Incidence rates of malignancies and hospitalized infectious events in patients with psoriasis with or without treatment and a general population in the U.S.A.: 2005–09. British Journal of Dermatology 170(2):366–373.

Korsnes, J. S., K. L. Davis, R. Ariely, C. F. Bell, and D. Mitra. 2015. Health care resource utilization and costs associated with nonfatal major adverse cardiovascular events. Journal of Managed Care & Specialty Pharmacy 21(6):443–450.

Lanfear, D. E., E. L. Peterson, J. Campbell, H. Phatak, D. Wu, K. Wells, J. A. Spertus, and L. K. Williams. 2011. Relation of worsened renal function during hospitalization for heart failure to long-term outcomes and rehospitalization. American Journal of Cardiology 107(1):74–78.

Leschke, J., J. A. Panepinto, M. Nimmer, R. G. Hoffmann, K. Yan, and D. C. Brousseau. 2012. Outpatient follow-up and rehospitalizations for sickle cell disease patients. Pediatric Blood Cancer 58(3):406–409.

Libby, A. M., V. Ghushchyan, R. B. McQueen, J. F. Slejko, J. L. Bainbridge, and J. D. Campbell. 2012. Economic differences in direct and indirect costs between people with epilepsy and without epilepsy. Medical Care 50(11):928–933.

Lipscomb, H. J., A. L. Schoenfisch, W. Cameron, K. L. Kucera, D. Adams, and B. A. Silverstein. 2015. Contrasting patterns of care for musculoskeletal disorders and injuries of the upper extremity and knee through workers' compensation and private health care insurance among union carpenters in Washington state, 1989 to 2008. American Journal of Industrial Medicine 58(9):955–963.

McKee, M. M., P. C. Winters, A. Sen, P. Zazove, and K. Fiscella. 2015. Emergency department utilization among deaf American sign language users. Disability and Health Journal 8(4):573–578.

Menendez, M. E., and D. Ring. 2015. Factors associated with hospital admission for proximal humerus fracture. American Journal of Emergency Medicine 33(2):155–158.

Merchant, S. J., P. H. Ituarte, A. Choi, V. Sun, J. Chao, B. Lee, and J. Kim. 2015. Hospital readmission following surgery for gastric cancer: Frequency, timing, etiologies, and survival. Journal of Gastrointestinal Surgery 19(10):1769–1781.

Mizutani, K., M. Hara, S. Iwata, T. Murakami, T. Shibata, M. Yoshiyama, T. Naganuma, F. Yamanaka, A. Higashimori, N. Tada, K. Takagi, M. Araki, H. Ueno, M. Tabata, S. Shirai, Y. Watanabe, M. Yamamoto, and K. Hayashida. 2017. Elevation of b-type natriuretic peptide at discharge is associated with 2-year mortality after transcatheter aortic valve replacement in patients with severe aortic stenosis: Insights from a multicenter prospective ocean-tavi (optimized transcatheter valvular intervention-transcatheter aortic valve implantation) registry. Journal of the American Heart Association 6(7).

Mullady, D. K., D. Yadav, S. T. Amann, M. R. O'Connell, M. M. Barmada, G. H. Elta, J. M. Scheiman, E. J. Wamsteker, W. D. Chey, M. L. Korneffel, B. M. Weinman, A. Slivka, S. Sherman, R. H. Hawes, R. E. Brand, F. R. Burton, M. D. Lewis, T. B. Gardner, A. Gelrud, J. DiSario, J. Baillie, P. A. Banks, D. C. Whitcomb, M. A. Anderson, and N. Consortium. 2011. Type of pain, pain-associated complications, quality of life, disability and resource utilisation in chronic pancreatitis: A prospective cohort study. Gut 60(1):77–84.

Mullerova, H., D. J. Maselli, N. Locantore, J. Vestbo, J. R. Hurst, J. A. Wedzicha, P. Bakke, A. Agusti, and A. Anzueto. 2015. Hospitalized exacerbations of COPD: Risk factors and outcomes in the eclipse cohort. Chest 147(4):999–1007.

Myer, P. A., A. Mannalithara, G. Singh, G. Singh, P. J. Pasricha, and U. Ladabaum. 2013. Clinical and economic burden of emergency department visits due to gastrointestinal diseases in the United States. American Journal of Gastroenterology 108(9):1496–1507.

Nadjiri, J., J. Hausleiter, S. Deseive, A. Will, E. Hendrich, S. Martinoff, and M. Hadamitzky. 2016. Prognostic value of coronary ct angiography in diabetic patients: A 5-year follow up study. International Journal of Cardiovascular Imaging 32(3):483–491.

Naz, B., T. J. Craig, E. J. Bromet, S. J. Finch, L. J. Fochtmann, and G. A. Carlson. 2007. Remission and relapse after the first hospital admission in psychotic depression: A 4-year naturalistic follow-up. Psychological Medicine 37(8):1173–1181.

Nemunaitis, G., M. J. Roach, J. Claridge, and M. Mejia. 2016. Early predictors of functional outcome after trauma. PM & R: The Journal of Injury, Function, and Rehabilitation 8(4):314–320.

Opolski, G., L. Januszkiewicz, E. Szczerba, B. Osinska, D. Rutkowski, Z. Kalarus, and J. Kazmierczak. 2015. Readmissions and repeat procedures after catheter ablation for atrial fibrillation. Cardiology Journal 22(6):630–636.

Palma, A., D. W. Lounsbury, L. Messer, and E. B. Quinlivan. 2015. Patterns of HIV service use and HIV viral suppression among patients treated in an academic infectious diseases clinic in North Carolina. AIDS and Behavior 19(4):694–703.

Panopalis, P., J. Z. Gillis, J. Yazdany, L. Trupin, A. Hersh, L. Julian, L. A. Criswell, P. Katz, and E. Yelin. 2010. Frequent use of the emergency department among persons with systemic lupus erythematosus. Arthritis Care and Research: The Official Journal of the Arthritis Health Professions Association 62(3):401–408.

Pendleton, A. M., L. K. Cannada, and M. Guerrero-Bejarano. 2007. Factors affecting length of stay after isolated femoral shaft fractures. Journal of Trauma and Acute Care Surgery 62(3):697–700.

Perkins, R. M., A. Rahman, I. D. Bucaloiu, E. Norfolk, W. DiFilippo, J. E. Hartle, and H. L. Kirchner. 2013. Readmission after hospitalization for heart failure among patients with chronic kidney disease: A prediction model. Clinical Nephrology 80(6):433–440.

Powars, D. R., L. S. Chan, A. Hiti, E. Ramicone, and C. Johnson. 2005. Outcome of sickle cell anemia: A 4-decade observational study of 1056 patients. Medicine (Baltimore) 84(6):363–376.

Quintana, J. M., A. Unzurrunzaga, S. Garcia-Gutierrez, N. Gonzalez, I. Lafuente, M. Bare, N. F. de Larrea, F. Rivas, and C. Esteban. 2015. Predictors of hospital length of stay in patients with exacerbations of COPD: A cohort study. Journal of General Internal Medicine 30(6):824–831.

Rahman, R., P. J. Catalano, D. A. Reardon, A. D. Norden, P. Y. Wen, E. Q. Lee, L. Nayak, R. Beroukhim, I. F. Dunn, A. J. Golby, M. D. Johnson, E. A. Chiocca, E. B. Claus, B. M. Alexander, and N. D. Arvold. 2015. Incidence, risk factors, and reasons for hospitalization among glioblastoma patients receiving chemoradiation. Journal of Neuro-Oncology 124(1):137–146.

Rosenheck, R. A., S. E. Estroff, K. Sint, H. Lin, K. T. Mueser, D. G. Robinson, N. R. Schooler, P. Marcy, and J. M. Kane. 2017. Incomes and outcomes: Social Security disability benefits in first-episode psychosis. American Journal of Psychiatry 174(9):886–894.

Sanders, K. A., S. M. Labott, R. Molokie, S. R. Shelby, and J. Desimone. 2010. Pain, coping and health care utilization in younger and older adults with sickle cell disease. Journal of Health Psychology 15(1):131–137.

Sharif, R., T. M. Parekh, K. S. Pierson, Y. F. Kuo, and G. Sharma. 2014. Predictors of early readmission among patients 40 to 64 years of age hospitalized for chronic obstructive pulmonary disease. Annals of the American Thoracic Society 11(5):685–694.

Sharma, A., D. M. Sharp, L. G. Walker, and J. R. Monson. 2007. Predictors of early postoperative quality of life after elective resection for colorectal cancer. Annals of Surgical Oncology 14(12):3435–3442.

Short, P. F., J. J. Vasey, and R. Belue. 2008. Work disability associated with cancer survivorship and other chronic conditions. Psychooncology 17(1):91–97.

Siebert, U., J. Wurm, R. M. Gothe, M. Arvandi, S. R. Vavricka, R. von Kanel, S. Begre, M. C. Sulz, C. Meyenberger, and M. Sagmeister. 2013. Predictors of temporary and permanent work disability in patients with inflammatory bowel disease: Results of the Swiss inflammatory bowel disease cohort study. Inflammatory Bowel Diseases 19(4):847–855.

SSA (Social Security Administration). 2017. Substantial gainful activity. https://www.ssa.gov/oact/cola/sga.html (accessed May 27, 2017).

Thomas, K. C., and A. R. Ellis. 2013. Patterns of healthcare use and employment among people with disabilities. Disability and Health Journal 6(2):133–140.

Thomas, K. L., A. F. Hernandez, D. Dai, P. Heidenreich, G. C. Fonarow, E. D. Peterson, and C. W. Yancy. 2011. Association of race/ethnicity with clinical risk factors, quality of care, and acute outcomes in patients hospitalized with heart failure. American Heart Journal 161(4):746–754.

Thorpe, R. J., Jr., A. J. Wynn, J. L. Walker, J. R. Smolen, M. P. Cary, S. L. Szanton, and K. E. Whitfield. 2016. Relationship between chronic conditions and disability in African American men and women. Journal of the National Medical Association 108(1):90–98.

Towfighi, A., D. Markovic, and B. Ovbiagele. 2011. National gender-specific trends in myocardial infarction hospitalization rates among patients aged 35 to 64 years. American Journal of Cardiology 108(8):1102–1107.

van der Linde, D., E. R. Andrinopoulou, E. N. Oechslin, W. Budts, A. P. van Dijk, P. G. Pieper, E. M. Wajon, M. C. Post, M. Witsenburg, C. K. Silversides, A. Oxenius, A. J. Bogers, J. J. Takkenberg, and J. W. Roos-Hesselink. 2013. Congenital valvular aortic stenosis in young adults: Predictors for rate of progression of stenosis and aortic dilatation. International Journal of Cardiology 168(2):863–870.

Vestbo, J., W. Anderson, H. O. Coxson, C. Crim, F. Dawber, L. Edwards, G. Hagan, K. Knobil, D. A. Lomas, W. MacNee, E. K. Silverman, and R. Tal-Singer. 2008. Evaluation of COPD longitudinally to identify predictive surrogate end-points (eclipse). The European Respiratory Journal 31(4):869–873.

Wahlqvist, P., M. Karlsson, D. Johnson, J. Carlsson, S. C. Bolge, and M. A. Wallander. 2008. Relationship between symptom load of gastro-oesophageal reflux disease and health-related quality of life, work productivity, resource utilization and concomitant diseases: Survey of a US cohort. Alimentary Pharmacology & Therapeutics 27(10):960–970.

Wolfson, J. A., S. M. Schrager, T. D. Coates, and M. D. Kipke. 2011. Sickle-cell disease in California: A population-based description of emergency department utilization. Pediatric Blood & Cancer 56(3):413–419.

Wolfson, J. A., S. M. Schrager, R. Khanna, T. D. Coates, and M. D. Kipke. 2012. Sickle cell disease in California: Sociodemographic predictors of emergency department utilization. Pediatric Blood & Cancer 58(1):66–73.

Yabroff, K. R., W. F. Lawrence, S. Clauser, W. W. Davis, and M. L. Brown. 2004. Burden of illness in cancer survivors: Findings from a population-based national sample. Journal of the National Cancer Institute 96(17):1322–1330.

Yamada, S., Y. Shimizu, M. Suzuki, and T. Izumi. 2012. Functional limitations predict the risk of rehospitalization among patients with chronic heart failure. Circulation Journal 76(7):1654–1661.

Yazdany, J., B. J. Marafino, M. L. Dean, N. S. Bardach, R. Duseja, M. M. Ward, and R. A. Dudley. 2014. Thirty-day hospital readmissions in systemic lupus erythematosus: Predictors and hospital- and state-level variation. Arthritis & Rheumatology 66(10):2828–2836.

Yehia, B. R., J. A. Fleishman, P. L. Hicks, M. Ridore, R. D. Moore, and K. A. Gebo. 2012. Inpatient health services utilization among HIV-infected adult patients in care 2002–2007. Journal of Acquired Immune Deficiency Syndromes 53(3):397–404.

Young, A., S. Muhlner, A. Kurowski, and M. Cifuentes. 2015. The association between physical medicine and rehabilitation service utilization and disability duration following work-related fracture. Work 51(2):327–336.

Young, A. E., M. Cifuentes, R. Wasiak, and B. S. Webster. 2009. Urban-rural differences in work disability following occupational injury: Are they related to differences in healthcare utilization? Journal of Occupational and Environmental Medicine 51(2):204–212.

Zhang, Q., Y. Yang, J.L. Saver. 2015. Discharge destination after acute hospitalization strongly predicts three month disability outcome in ischemic stroke. Restorative Neurology and Neuroscience 33(5):771-775.

This page intentionally left blank.