and benefits is important, perhaps more so when much of the expected benefit will accrue to the broader community, not just the individual.12

Adherence to Treatment for Latent Infection

One consideration in the choice among alternative drug regimens is the trade-off between the higher degree of efficacy of a longer duration of treatment and the lower levels of individual adherence to such regimens. A number of studies suggest that rates of initiation and completion of treatment of latent tuberculosis infection are sometimes quite low. One study of an indigent urban population found that only 55 of 466 people with tuberculosis infection were prescribed drug therapy and only 20 of those completed it (Schluger et al., 1999). Another study involving high-risk inner-city residents identified 809 people with a positive skin test result of whom 409 fit ATS/CDC criteria for therapy for latent tuberculosis infection; only 84 (20 percent) actually completed treatment (Bock et al., 1999). Although the rate of treatment adherence might be expected to be high among health care workers, studies suggest otherwise. For example, one study found that only 8 to 10 percent of physicians whose skin tests had converted from negative to positive had been treated for latent tuberculosis infection (Ramphal-Naley et al., 1996). In another study, of 40 health care workers who were identified as having been eligible for isoniazid treatment following a skin test conversion, only 15 (38 percent) had completed at least 6 months of therapy (Fraser et al., 1994).

The highest rates of completion of treatment among health care workers were reported in a study of a hospital in a city with relatively high rates of active tuberculosis (Blumberg et al., 1996). Of 125 workers with a recent positive tuberculin skin test result, all got a chest radiograph and almost all (98 percent) saw a physician. Although 84 percent started therapy for latent tuberculosis infection, only 66 percent of those who started therapy completed at least 6 months of treatment. Almost three-quarters of the 34 physicians in the group completed therapy whereas not quite half (44 of 91) of other workers completed therapy. Of all those who started but did not complete treatment, one-third stopped because of side effects.


Studies suggest that people often misunderstand numerical explanations of risks and benefits provided by health professionals (Schwartz et al., 1997). How information is provided is important (Ransohoff and Harris, 1997). For example, people may find it easier to understand frequency information (e.g., a 4 in 1,000 chance of some outcome) than probability information (e.g., a 0.4 percent chance of the outcome) (Gigerenzer, 1996). They may also better understand comparisons of absolute risk to a baseline (e.g., reduction in risk to 4 in 1,000 with treatment compared to 12 in 1,000 without treatment) than they understand presentation of relative risk information with no baseline (Schwartz et al., 1997). Treatment may appear more attractive when described in terms of gain (e.g., 99 percent of patients will not develop the disease) rather than loss (e.g., 1 percent will develop the disease) (Mazur and Hickam, 1990; Hux and Naylor, 1995).

The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement