INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Completion of antituberculosis treatment is the foremost priority of tuberculosis (TB) control programs (1). Interrupted or incomplete treatment increases the risk of treatment failure, relapse of disease, and acquisition of drug-resistant TB (2). In turn, treatment failure and relapse can increase transmission of TB. TB patients who move during the course of treatment may be at increased risk for interrupted and incomplete treatment. Patient movement has contributed to several multistate outbreaks of drug-resistant TB, and may have facilitated the spread of a highly lethal clone of multidrug-resistant Mycobacterium tuberculosis from New York City to other parts of the United States (6).

In 1993, over 1,300 TB patients in the United States moved to another reporting area before completing treatment (10). Tracking these patients to ensure that they complete treatment could prevent further emergence and spread of drug-resistant TB in the United States. To our knowledge, there has been no systematic evaluation of patients who have moved to another reporting area before completing treatment. We report on the treatment outcomes and characteristics of a population-based cohort of TB patients who moved between local health jurisdictions. We also determined risk factors for TB treatment default in California.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Study Cohort

Our study population included 3,222 verified TB patients reported to the California Department of Health Services (CDHS) by 60 of 61 local health jurisdictions in 1993. This population constituted 62% (3,222 of 5,160) of all reported cases. Our study cohort included all patients who were at least 15 yr of age and alive at the time of diagnosis, had started antituberculosis treatment, and did not move out of California during treatment. All cases met the Centers for Disease Control and Prevention (CDC) surveillance case definition for TB (11, 12).

Routine Data Collection, Data Sources, and Definitions

Patient characteristics, initial drug-susceptibility results, the reason for discontinuing TB treatment, the dates on which antituberculosis drugs were started and discontinued, circumstances of administration of treatment (completely self-administered by the patient versus delivered under direct observation by trained personnel), and the specimen collection dates for the initial positive and first consistently negative sputum cultures were submitted on the Report of Verified Case of Tuberculosis (RVCT) and its follow-up forms (13). Patients codiagnosed as having acquired immune deficiency syndrome (AIDS) were identified by cross-matching of our cohort against the registry of AIDS cases reported to CDHS from 1982 to 1993. All AIDS cases met reporting criteria from the CDC (14).

Multidrug-resistant TB was defined as resistance to isoniazid and rifampin of the first isolate collected for drug-susceptibility testing during the disease episode. Duration of treatment was defined as the number of days from the date on which the first regimen was prescribed to the date on which the last prescribed medications could have been ingested by the patient. The reason for discontinuing TB treatment was derived from the RVCT form, and was reported as either a treatment outcome (patient completed treatment, died before treatment was completed, defaulted from treatment, other) or an interim case-management category (moved). Patients who were lost to follow-up or who refused further treatment were classified as having defaulted from treatment. Patients who discontinued treatment because of medical complications were classified as other. Patients who transferred from the jurisdiction in which they were diagnosed as having TB to a new local health jurisdiction with a known forwarding residence or work address were classified as having moved. In California, institutions such as prisons, jails, or hospitals are not considered independent local health jurisdictions or reporting entities. When patients were diagnosed in these institutions, they were reported to the local health jurisdiction in which the diagnosing facility was located. Patients released from these institutions were only classified as having moved if they relocated from their original health jurisdiction to another local health jurisdiction.

Follow-up of Patients who Moved within California

To ascertain treatment outcomes for patients who moved within California, we instituted a tracking system. Each local health jurisdiction submitted an RVCT Follow-up 2 form when treatment was discontinued. If treatment was discontinued because the patient moved to a new local health jurisdiction, that jurisdiction was contacted and instructed to follow-up on the patient and submit another RVCT Follow-up 2 form. Additional outreach was made by staff from CDHS to ensure that follow-up information was submitted. All 61 local health jurisdictions in California participated in the follow-up of patients who moved.

Statistical Analysis

Categorical variables were assessed using the two-tailed Mantel- Haenszel chi-square or Fischer's exact test. Odds ratios (ORs), relative risks (RRs), and 95% confidence intervals (CIs) were calculated. Variables were initially included in a logistic regression analysis if their respective univariate analyses yielded p < 0.10. Using a forward, stepwise procedure, our final model was developed with entry and retention criteria of p < 0.10 and p < 0.05, respectively. Statistical interactions between appropriate variables were explored. Statistical analyses were done with SAS software (version 6.10; SAS Institute, Cary, NC) and Epi-Info (version 6.0; CDC, Atlanta, GA).

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Of the 3,222 TB patients in our study population, we excluded 646 patients: 363 were under 15 yr of age, 82 were dead at the time of diagnosis, 42 had not started taking antituberculosis drugs, 147 had moved out of California and 12 had no treatment outcome information submitted. The remaining 2,576 patients formed the study cohort.

In our cohort, 2,035 (79.0%) patients completed treatment, 236 (9.2%) died, 141 (5.5%) defaulted from treatment, and 17 (0.7%) had other treatment outcomes in the health jurisdiction in which their disease was diagnosed; 147 (5.7%) patients moved to another local health jurisdiction in California. We ascertained treatment outcomes for 131 of the 147 (89.1%) patients who moved.

To determine the risk of defaulting from TB treatment among patients who moved, we compared their treatment outcomes with those of the 2,429 patients who did not move. Patients who moved defaulted nearly six times more often (RR = 5.5; 95% CI = 4.1 to 7.4), but had the same risk of death (Table 1).

To determine the association between directly observed treatment, patient movement, and treatment default, we compared patients who moved with those who did not. Of patients for whom information was available, 96 of 124 (77.4%) who moved had directly observed treatment, as compared with 650 of 2,429 (26.8%) patients who did not move. Patients who moved were nearly three times more likely to have directly observed treatment (RR = 2.9; 95% CI = 2.6 to 3.3). However, their rate of default did not differ by whether treatment was directly observed or self-administered (32% versus 32%).

Among patients who defaulted from treatment after moving, we evaluated the median duration of treatment and proportion with documented sputum conversion. Of 42 defaulters, 29 had drug-sensitive disease, six had disease resistant to one or more first-line antituberculosis drugs, and seven had culture-negative disease. For 24 of 29 patients with drug-sensitive disease, a regimen was started that included isoniazid, rifampin, and pyrazinamide. These patients defaulted a median of 138 d after initiation of treatment (interquartile range: 62 to 287 d), which is equivalent to three-quarters of the standard 6-mo short-course regimen. The five remaining patients were given a regimen that included isoniazid and rifampin, but not pyrazinamide. These patients defaulted a median of 179 d after initiating treatment (interquartile range: 152 to 299 d), which is equivalent to two-thirds of the standard 9-mo treatment regimen. Of 25 patients with culture-proven drug-sensitive pulmonary disease, only 11 patients (45%) were documented to have undergone sputum culture conversion from positive to negative.

To identify risk factors for moving before treatment was completed in the diagnosing local health jurisdiction, we compared patients who moved with those who completed treatment. Patient residence at time of diagnosis was the most significant predictor of movement (Table 2). In particular, patients whose TB was diagnosed in a state prison had an extremely high risk for movement, and accounted for 38.7% of those who moved. Although patients with a history of homelessness had a moderate risk for movement, they accounted for 14.3% of patients who moved. Patients diagnosed in long-term-care facilities demonstrated a high risk of movement but accounted for 6.7% of patients who moved.

To determine whether the newly ascertained treatment outcomes of patients who moved altered our assessment of risk factors for defaulting, we compared two multivariate models. First, independent risk factors for defaulting from treatment (as compared with completing treatment) were identified among patients who did not move. Second, we repeated this analysis with all patients, including those who moved. In the second model, the number of defaulters increased by 30% (from 141 to 183 patients), and diagnosis while in a state prison emerged as the single most important risk factor for defaulting (Table 3). Unlike diagnosis of multidrug-resistant TB, which was strongly associated with defaulting but represented fewer than 3% of all defaulters, diagnosis of TB in a state prison was both strongly associated with defaulted treatment and represented a substantial proportion (17%) of defaulters.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

To our knowledge, no other study has traced a population-based cohort of TB patients who moved for the purpose of determining their treatment outcomes. In our study, three of every 10 patients who moved between local health jurisdictions in California defaulted from treatment. Patients who moved defaulted nearly six times more often than patients who did not move. Our findings highlight the need to ensure continuous treatment for TB patients who move.

Failure to complete treatment prolongs the morbidity of individual patients and facilitates transmission of M. tuberculosis. Nearly half of patients who defaulted from treatment after moving received less than two-thirds of the recommended treatment regimen for their disease, and one-quarter received less than one-third of the recommended regimen. Assuming that patients took their medications without interruption, we estimate that 10% to 15% of these defaulters will relapse with disease (2). Since absolute continuity of treatment is unlikely, the relapse rate may be even higher. Less than half of patients with culture-proven pulmonary TB who defaulted after moving were documented to have sputum-culture conversion. Our study suggests that many patients who defaulted after moving may not be permanently cured, and may be infectious.

Patient residence at the time of diagnosis of TB was the most significant predictor of patient movement. Patients whose disease was diagnosed while they were residing in a state prison accounted for nearly 40% of those who moved; these patients moved and defaulted more often than other patients with TB. Patients diagnosed in a local jail had a lower, but still significant, risk for moving and defaulting. Although patients with TB diagnosed while they were in long-term-care facilities had a high risk of movement, they comprised few patients and were unlikely to default. Congregate settings such as correctional institutions, long-term-care facilities, homeless shelters, single-room-occupancy hotels, hospices, and other residential care facilities can be conducive to transmission of M. tuberculosis (6, 15). If patients default from treatment after moving, become infectious, and continue to reside in congregate settings, transmission of TB may be amplified.

Although our study was not designed to determine why patients defaulted from treatment, we found that patients who moved were more likely to receive directly observed treatment. However, administration of directly observed treatment did not appear to alter the risk for defaulted treatment after moving. This suggests that other interventions may be needed to improve the treatment outcomes of these patients.

An additional objective of this study was to determine how patient movement influenced our understanding of TB treatment default in California. Historically, federal and state summaries of TB treatment outcomes excluded patients who moved from the reporting area (20, 21). In our study, excluding the treatment outcomes of patients who moved led to an underestimate of treatment defaulters in California (141 versus 183) and masked the strongest risk factor associated with treatment default (diagnosis while residing in a state prison).

Our study has several limitations. First, we do not know the exact amount of medication taken by patients who defaulted after moving. We used data reported to the state TB registry by local health authorities, and did not directly abstract data from medical records. If patients did not take their medications every day, or if there were interruptions in the continuity of treatment during their movement, we would have overestimated the duration of treatment. Second, some patients without documentation of sputum conversion may have undergone sputum conversion, but without sputa being collected or negative cultures being reported. Third, some of those patients reported to have defaulted after moving may have received and completed treatment, but local health authorities were not notified. In this case, we would have overestimated the number of patients who defaulted from treatment. Finally, our results may not be generalizable to patients who moved out of California.

In summary, our study found that TB patients who moved between local health jurisdictions in California defaulted from treatment more often than patients who did not move. These patients were also more likely to have TB diagnosed while they were in congregate settings that can be conducive to transmission of TB. If patients who were not permanently cured returned to congregate settings, TB transmission may have been amplified. Failure to ensure complete treatment among patients who move will adversely affect the health of individual patients and the control of TB in many high-risk populations and settings.