Directly Observed Therapy for Treatment Completion of Pulmonary Tuberculosis: Title and subTitle BreakConsensus Statement of the Public Health Tuberculosis Guidelines Panel

ONE OF THE MOST disturbing public health trends in the United States in recent years has been the dramatic reemergence of tuberculosis (TB). After declining consistently for several decades, the incidence of TB surged after 1985, and by 1993 there were 62000 more cases of TB than had been predicted by pre-1985 trends.¹

A significant proportion of these cases is due to recent transmission of infection, confirming the suspicion that untreated cases serve as reservoirs to infect others.^{2 - 4} Even 1 active case can produce a miniepidemic.^{5 - 8} In San Francisco, Calif, for example, a single case of TB resulted in clinical disease in 34 patients and infection in many others.² Left untreated, approximately 50% of active cases result in death, while transmission of infection continues to occur within the community.

Current TB treatment goals of the Centers for Disease Control and Prevention (CDC)⁹ and the Advisory Council for the Elimination of Tuberculosis¹⁰ call for 90% of all cases to complete a recommended 6-month, 3-drug to 4-drug course of therapy within 12 months. When this treatment goal is achieved, at least 90% of patients with active disease are cured, and fewer than 5% experience relapse.¹¹

Compounding the resurgence of TB is its substantial increase in resistance to both single-drug and multidrug treatment.^{12 - 13} Incomplete or incorrect antibiotic therapy serves as a major cause of acquired drug resistance, which may be transmitted to others.¹⁴ Multidrug-resistant TB (MDRTB, ie, resistance to at least isoniazid and rifampin) has been associated with treatment failure and high mortality, especially for patients coinfected with human immunodeficiency virus (HIV). However, patients who are infected with strains susceptible to standard chemotherapy (even HIV-infected patients, including those with advanced disease) can expect high cure rates when they complete therapy.¹⁵

The economic consequences of TB are substantial. US health care expenditures in 1991 for TB were estimated to be $703 million.¹⁶ Moreover, treatment costs for MDRTB may exceed $200000 per case.¹⁷ In New York City (the national TB epicenter during this resurgent period), the cost of treating 1 patient with MDRTB approached $1 million.¹⁸

Explanations proposed for the increase in TB and MDRTB include substantial declines in public funding for TB control (leading to deterioration in public health infrastructure),¹⁹ HIV infection,^{20 - 21} physician mismanagement,²² immigration from endemic regions,²³ homelessness,²⁴ substance abuse,²⁵ and the failure of patients to complete therapy.²⁶ These factors are exacerbated when TB therapy is self-supervised, because the probability of treatment completion appears to be lowest in such cases.^{27 - 28}

Among all these factors the greatest unmet public health challenge in TB control is treatment completion.²⁹ At least 20% of all patients treated for pulmonary TB in 1993 did not complete therapy even within 12 months.³⁰ Moreover, TB treatment completion rates reported to the CDC by health departments in the 64 largest US cities between 1987 and 1991 averaged from 73.7% to 77.3%.³¹ For some cities, especially those with a high incidence of TB, and several states, treatment completion rates were substantially lower.

The ability to predict adherence to and completion of treatment is unreliable.^{32 - 34} No demographic variable, occupation, level of income, or level of education reliably and consistently predicts adherence to therapy. Adherence to most medical regimens is inversely proportional to the length of therapy, the number of drugs administered, the frequency of drug administration, and the overall complexity of the treatment regimen. These findings have significant implications for TB chemotherapy, which typically involves 3 to 4 antimicrobials (and a greater number of pills), potential adverse effects, and a minimum of 6 consecutive months of therapy (completed within 12 months).^{35 - 36}

To increase the probability of treatment completion, directly observed therapy (DOT) has been recommended as the standard of care for pulmonary TB.^{9 - 10 ,37} The critical aim of treatment completion is to avoid the effects of nonadherence to therapy, such as persistent infectiousness on the part of the patient and higher rates of treatment failure, relapse, and drug resistance. It is argued that DOT prevents these problems because patient consumption of medication is confirmed through observation whereby treatment is completed and cure achieved. Treatment completion, therefore, is a critical benchmark of cure.

Although DOT has been recommended by many professional organizations and expert panels on TB, to our knowledge no comprehensive review of evidence to support these recommendations has been published. The Public Health Tuberculosis Guidelines Panel was established in 1995 to review scientific evidence on the effectiveness of DOT compared with other interventions, and to make public health recommendations regarding successful practices in TB treatment completion.

The Council on Linkages Between Academia and Public Health Practice was established in 1992 with funding from the Health Resources Services Administration (HRSA) to implement the recommendations of the Faculty/Agency Forum.³⁸ In 1995, the council undertook a project to develop model public health guidelines in 4 areas: childhood immunizations, preventive cardiology, lead poisoning prevention, and TB treatment completion. The council established the Public Health Tuberculosis Guidelines Panel to develop guidelines for treatment completion of pulmonary TB.

In consultation with leading TB control organizations, the council selected individuals for the panel who had substantial experience in the public health, clinical, social and behavioral science, and epidemiological aspects of TB control. Given the public health focus of this project, panel members also were selected because they were practitioners in federal, state, and local public health, or had experience with health issues and public health issues relevant to vulnerable populations. Other individuals participated in the initial development of the guideline process (see acknowledgments). The entire panel was convened twice, once in 1995 and once in 1996. Panel members shared the findings from their examination of the published literature and arrived at their recommendations through periodic conference calls and smaller work groups.

The methods used by the panel have been described elsewhere.³⁹ In brief, the English-language medical literature was reviewed by searching MEDLINE with the following key words and phrases: directly observed therapy, supervised therapy, compliance, treatment completion, case management, and treatment adherence for TB for the period 1966 through August 1, 1996.

Articles that met these criteria underwent structured review based on the following criteria: study design (randomized or semirandomized trials, prospective nonrandomized studies, retrospective or case-control studies, and cross-sectional studies), sample size, interventions (supervised or nonsupervised therapy), target population, drug resistance, and treatment completion as an outcome. The review process also examined incentives and enablers because they often distinguish specific treatment interventions. Thus, the review process examined the intensity of DOT and quantified differences in the level of supervised therapy.

Given its focus on treatment of active TB cases, the panel did not review scientific evidence related to treatment completion for preventive therapy. The panel determined that preventive therapy (ie, single-drug therapy) is sufficiently different from treatment completion of active cases (ie, multidrug therapy). Higher completion rates with the use of supervised preventive therapy might overstate the actual effectiveness of DOT. The panel determined that evidence regarding directly observed preventive therapy was not generalizable to active TB case management, and should not be used in reaching its recommendations.

In addition, there is an extensive history of TB case management recommendations by expert panels and professional organizations (eg, the CDC, the American Thoracic Society, the American Lung Association, and the World Health Organization).⁴⁰ These organizations typically recommend the use of DOT for all active cases. The Public Health Tuberculosis Guidelines Panel saw little added value in simply repeating these recommendations, particularly since they did not include specific evidence on the effectiveness of different strategies in achieving high treatment completion rates. Instead, the panel's recommendations are derived from a review of the medical literature.

The panel based its recommendations on the evidence graded according to a published system (Table 1). These recommendations were drafted by the panel chair and reviewed twice by the panel, and comments from panel members were incorporated.

The literature search yielded 497 articles. Using the established selection criteria, the panel reduced this initial search to an unduplicated set of 27 eligible studies with treatment completion for pulmonary TB as an outcome. These studies included 5 randomized or semirandomized trials, 12 prospective trials without controls, 7 retrospective studies, 2 case-control studies, and 1 cross-sectional report. The remaining citations, which included review articles, commentaries, expert opinions, and studies with outcomes other than treatment completion, were excluded.

The 12 studies based on comprehensive, patient-centered DOT strategies, such as fully supervised DOT with multiple incentives and enablers, reported the highest treatment completion rates (Table 2). For these studies, reported treatment completion rates ranged from 86% to 96.5% for a variety of patient populations, including alcoholic patients, substance abuse patients, incarcerated patients, homeless persons, and patients infected with HIV. The rate of TB relapse reported in these studies ranged from 0% to 11.5%.^{42 - 53}

These DOT studies used a combination of multiple incentives and enablers, including intermittent regimens designed around a patient's lifestyle (therapy at home, school, work, or a clinic 2 to 3 times per week); relevant social and economic enablers and incentives (food, clothing, books, stipends, transportation, treatment contracts, bilingual staff, or reminders); and culturally appropriate outreach (community and non–health workers) and tracking. Depending on the patient population, DOT was supplemented with substance abuse therapy, housing for homeless patients (during treatment), comprehensive case management (beginning at the time of hospitalization), and treatment education for patients and their families, and referrals for other social services.

For the 4 studies of DOT without extensive enablers and incentives,^{54 - 57} treatment completion rates ranged from 85% to 87.6%, and reported rates of relapse ranged from 0.8% to 4.9% (Table 3).

With modified DOT (Table 4), supervision is used for only part of the treatment period (typically during the hospitalization phase of therapy), and thereafter, patients are self-supervised. This strategy appears to be less effective, with treatment completion rates ranging from 78.6% to 82.6%.^{58 - 59}

The 9 studies with nonsupervised strategies (Table 5) reported treatment completion rates ranging from 41.9% to 82% and relapse rates ranging from 2.1% to 4.5%.^{11 ,60 - 67} These interventions include education and motivation strategies directed at the patient, the patient's family, or the clinician; blister packs for medications; selective DOT for noncompliers; and self-administered therapy (SAT).

The median completion rate for each intervention based on all studies is shown in Figure 1.

The literature review also revealed 2 studies that concluded that DOT is more cost-effective than SAT. In one study, the cost per case cured using DOT was $13925 compared with $15003 for SAT, and use of DOT predicted fewer TB relapses (31 vs 132 per 1000 patients treated) and fewer TB-related deaths (3 vs 13).⁶⁸ In another study, the cost per case cured using DOT was $3999 compared with $12167 using SAT.⁶⁹ Both studies were conducted from the perspective of a public health department, involved community-based DOT approaches, and sustained their findings across multiple set points in their respective models.

Because of the documented efficacy in treatment completion for pulmonary TB, the Public Health Tuberculosis Guidelines Panel recommends a patient-centered strategy based on DOT. This strategy should include appropriate incentives and enablers based on the individual needs of the patient, including transportation, outreach, financial and social incentives, reminder systems, and tracking for failed appointments. These appear particularly effective for patients trying to balance TB treatment completion with HIV infection, unemployment, substance abuse, or homelessness.

Directly observed therapy has had intuitive appeal within the public health community for decades. The panel's review of the published evidence underscores the effectiveness of this strategy compared with other strategies, including SAT. These findings have significant implications for public health practice, since a single case of TB can generate clusters of new cases.

The effectiveness of DOT appears robust. The panel's findings define an enabler-based or incentive-based model that has been duplicated across time, geography, and social infrastructure. For example, DOT appears to be effective for a variety of high-risk groups, including the unemployed, substance abuse patients, alcoholic patients, the homeless, refugees, and patients with HIV infection. Moreover, this patient-centered strategy appears equally effective regardless of country or community.

There are several caveats regarding the panel's findings. First, the evidentiary basis for the panel's recommendations includes studies conducted in a variety of countries and cultures. Drawing conclusions from international literature can be fraught with pitfalls.^{70 - 71} For example, definitions of exactly what is meant by DOT can vary. However, the panel attempted to account for this by assessing different versions or levels of DOT intensity. The panel found that in some studies DOT was modified and limited only to the inpatient portion of TB care. In other studies, DOT extended beyond the practice of observing the patient ingest medications to the use of incentives and enablers.

The panel found that the more action taken to enable patients with TB to take medications, the more likely the patients are to complete therapy. Programs should consider using outreach for isolated populations, adding appropriate incentives when necessary, and making treatment patient centered, rather than clinician or clinic centered. In short, building TB treatment regimens around patients' lifestyles appears to be not only more effective in terms of treatment completion, but also more acceptable to patients.

There are several areas for future research. First, it would be useful to examine more closely the effectiveness of various incentives and enablers among different patient populations. This information could facilitate the design and implementation of more customized incentives and enablers and produce more cost-effective TB control across diverse settings. With only 2 cost-effectiveness studies identified by the panel, additional research is necessary to determine the critical components that ultimately drive this cost-effective equation.

Also, more studies regarding the implementation of public health recommendations are needed. Based on the experiences of the Preventive Services Task Force, guidelines should be accompanied by an implementation strategy, analogous perhaps to "Put Prevention Into Practice."⁷² Implementation strategies are timely because TB control will increasingly be influenced by health department outsourcing of programs and emerging private-sector initiatives such as through managed care.⁷³ These trends provide opportunities to establish an effective continuum of patient care by linking clinical services with public health practices.

Finally, the panel emphasizes that recommendations regarding DOT and treatment completion are only one component of an effective TB control program. Timely case identification, successful associate and contact investigations, and effective prevention programs for high-risk populations also are critical ingredients of successful TB control.⁷⁴ Unless these components are appropriately addressed, even DOT will be ineffective in the fight against TB.