High death rates in Peruvian patients defaulting from multi-drug-resistant tuberculosis treatment

Although only 10% of Peruvian patients on a multiple drug resistant tuberculosis (MDR TB) treatment programme defaulted from treatment, 40% of the patients who defaulted had culture-positive sputum at the time of default and one in two of the defaulters died, according to the findings of a retrospective study published in the June 15^th edition of Clinical and Infectious Diseases.

MDR TB is disease caused by a strain of Mycobacterium tuberculosis that is resistant to isoniazid and rifampicin in about 4% of TB patients. The high potential for spreading MDR TB and the high mortality rate observed in the Peruvian patients underscores the public health threat posed by defaulting of MDR TB treatment in resource-poor settings.

Defaulting, defined as prolonged treatment interruption, not only increases the risk of TB recurrence and TB-related death in the patient, but also the circulation in the population of TB bacilli that are refractory to treatment with existing drugs. This has ominous public health implications regarding TB control.

MDR TB therapy requires 18 months treatment whereas ordinary TB treatment lasts for less than a year. This is a constraint for patients in resource poor settings who have to reckon with other pressing survival issues such as malnutrition and poverty. The second-line drugs used for MDR TB treatment also cause serious adverse events. These factors contributed to decreased patient adherence to MDR TB treatment in up to 48% of patients in the study.

While it is known that default from first-line TB therapy is associated with patient- and programme-related factors, not many studies have examined risk factors related to MDR TB treatment default. The only such study to date has implicated substance use, dissatisfaction with healthcare worker attitudes, and low or unstable socioeconomic status.

A community-based MDR TB treatment programme was launched in 1996 in Peru. This provided a unique opportunity for a team of investigators from Peru and the United States to identify risk factors which are associated with treatment default on one hand, and death in defaulters on the other.

The study took place in the Peruvian capital Lima. Study subjects were patients with laboratory-confirmed MDR TB who initiated their first individualised treatment regimen from February 1999 to July 2002. The treatment programme was optimised to facilitate adherence using several approaches.

These approaches included directly observed treatment by community healthcare workers, identification and aggressive management of adverse events, provision of individualised socioeconomic support, and psychiatric consultations and group therapy in patients with mental health needs.

Patient clinical charts were retrospectively reviewed to abstract data on a number of variables which might be predictors of treatment default or death in defaulters. These included baseline clinical characteristics, radiographic findings, history of substance use, TB treatment history, socioeconomic variables, microbiologic test results from monthly sputum smear and culture examinations, and severe adverse events.

Data were also collected on the nutritional state, self-reported substance abuse, housing, and access to water. Standard definitions were developed for defaulting, adequate versus poor bacteriologic response, cure, treatment completion, treatment failure, and death.

Home visits were made for patients who defaulted from treatment to establish vital status, collect missing baseline socioeconomic information, and to estimate the rate of mortality among patients who defaulted from therapy.

Sixty-seven (10%) of 671 patients defaulted from MDR TB therapy. The median time to treatment default was 438 days, and 27 (40%) of the 67 patients who defaulted from treatment had culture-positive sputum at the time of default.

A statistical model identified substance use, substandard housing conditions, later year of enrollment, and health district but not severe adverse events as significant predictors of default from therapy.

Forty-seven (70%) of 67 patients who defaulted from therapy were traced; over 50% of the defaulters died. Poor bacteriologic response, being less than one year on treatment at the time of default, low education level, and having a psychiatric disorder were significant predictors of death after defaulting.

In summary, more than a third of Peruvian patients who defaulted from MDR TB treatment had culture-positive sputum at the time of treatment default, and over 50% of the defaulters died. Public health efforts must aim at preventing treatment default in order to reduce TB-related mortality and the risk of spreading MDR TB in the population.