In this study of 291 patients undergoing retreatment for TB, outcomes differed considerably by group -- 74% of patients with relapse, 48% of patients with failure, and 41% of patients with default had treatment success -- similar to previous studies[5, 7]. Default from retreatment was extremely common at 25%, higher than previous country-wide estimates. This may reflect temporal changes in treatment completion but more likely represents differences in study populations, as we focused on TB "hot spots", or urban centers with comparatively high TB incidence. Recent studies have demonstrated that, in urban settings, adherence is linked to patient knowledge about TB and provision of disease-specific education by the health care provider to the patient. In busy urban clinics, time for education may be limited. Default from retreatment was most frequent among those who had defaulted from initial treatment, while failure was most common among those with previous failure. Although retreatment guidelines are often the same for patients with failure, default from, or relapse after initial treatment, these results suggest that groups may benefit from different management strategies[9, 10]. For example, treatment failure is commonly due to drug resistance, while recurrence may be due to poor adherence, high mycobacterial burden (such as in cavitary disease), or exogenous reinfection. Default patients may require intensified case management and education, rather than more intensive treatment.
The present study shows that, even when available, drug susceptibility testing is underutilized. It was performed in only 10% of retreatment patients. All 3 failure patients who underwent DST testing had MDR-TB, while 3 of 22 of relapse patients and 0 of 5 default patients tested did. While these DST results were only available for three failure patients and, therefore, not representative, these data and those from other studies suggest that MDR risk is not uniform among retreatment subgroups, with increased prevalence of MDR among patients with initial treatment failure[2, 11–13]. According to a population-based study conducted among retreatment cases in Morocco, 12.2% had MDR-TB, but the study did not divide retreatment patients into failure, relapse, or default subgroups. Taken together, these findings support use of DST in all retreatment patients, earlier DST testing in those with clinical and microbiological indications of impending treatment failure, and use of second-line drugs for retreatment of patients with initial treatment failure until DST results are known. In Morocco, DOTS coverage is 100%, and concerted efforts to dramatically enhance DST use are underway.
Published medical risk factors for failure or relapse include HIV infection, diabetes mellitus, low body weight, cavitation on chest x-ray, high bacterial burden, short treatment duration, drug resistance, and positive culture after two months of treatment[14–16]. Sociodemographic factors include unemployment, drug abuse, alcoholism, smoking, and poor treatment adherence. Treatment default is known to be associated with substance abuse, foreign birth, male gender, previous default, low socioeconomic status, psychiatric illness, unemployment, migration, side effects, [17, 18] long distance to the clinic, social stigma, and poorly-implemented DOTS but, of course, differ by setting [16, 19–21]. In our study population, HIV infection is rare; among TB patients, less than 1% are HIV-infected (unpublished data, Morocco NTP). Further, alcohol use in Morocco is uncommon, and smoking is extremely uncommon among women. Moreover, in the urban clinics studied, the majority of patients are non-immigrants, the clinics are geographically accessible, and DOTS coverage is 100%. Thus, many traditional risk factors for poor TB treatment outcomes are less prominent in Morocco, making it harder to prospectively identify patients at risk. However, continued sputum smear positivity after 3 months of treatment is a strong predictor of subsequent poor outcomes, and should prompt DST testing in all patients. As missed treatment doses may herald impending default, enhanced communication between the local clinics that dispense TB treatment and physicians at the regional health centers that prescribe it may be one country-specific strategy to help pinpoint those individuals who are missing doses and are at high risk of defaulting altogether. Small sample sizes limited our ability to evaluate subgroups, but even so, we were able to identify male sex, substance use (tobacco, alcohol, or illicit drug use), and missed doses during the intensive phase as likely risk factors for treatment default. Higher odds of hospitalization probably reflected the need for hospitalization to ensure adherence rather than increased disease severity. Further exploring risk factors for treatment default may help control programs identify those likely to benefit from targeted interventions such as health education, substance abuse counseling, enhanced tracking, or reinforcement of DOTS supervision[22, 23].
As a retrospective chart review, our risk factor evaluation was limited by the availability of data present in clinical charts. Information about tobacco, alcohol, and illicit drug use was not routinely recorded, for example. Classifying those with missing substance use data as nonusers likely biased analyses of this risk factor toward the null; however, estimates of smoking, tobacco, and illicit drug use were similar to national substance use statistics. Also, in cases of recurrence, it was not possible to distinguish between relapse and reinfection, so we limited our risk factor analysis to those who had received initial TB treatment within two years of recurrence and were, thus, more likely to have true relapse. Our ability to identify independent risk factors in subgroup analyses was limited by small sample sizes; questions regarding risk factors in these subgroups would best be answered in larger, prospective studies. Finally, DST testing was not universally performed in retreatment patients, so selection bias is possible, as clinicians are more likely to send those at high risk of resistance for testing. In our study, 20% of retreatment patients with DST had MDR-TB, compared with 12% in a national prevalence survey.