Poor Compliance Makes Treatment of Latent Tuberculosis Infection Unsatisfactory

Methods

We retrospectively analyzed 50 858 patients treated for LTBI through all public TB clinics in Texas from 1995 to 2002 using an existing database. We also analyzed a database for the years 2002 through 2008 to identify cases in the initial database who developed active disease (presumed reactivation).The Texas Department of State Health Services contracts with approximately 30 part-time physicians to provide services in TB clinics. These physicians follow treatment guidelines from the department based on Centers for Disease Control and Prevention recommendations but can individualize care according to patient circumstances. ⁵ Patients are seen monthly to review symptoms and obtain laboratory tests if toxicity is a concern. The usual treatment regimen is INH for 9 months. Treatment data on these patients were maintained in a computerized database and were accessed by one of the authors (S.B.). The Institutional Review Board for Studies on Human Subjects of the Texas Department of State Health Services and the Texas Tech University Health Sciences Center reviewed and approved the study with waiver of informed consent.

Data collected included sex, age, ethnicity, race, birth country, history of BCG vaccination, date of LTBI report, tuberculin skin test result, chest X-ray report, treatment regimen, dates of starting and stopping treatment, duration of treatment, reasons for discontinuing treatment, and development of active TB. Information was incomplete on the history of TB contact, intravenous drug use, incarceration, HIV status, and underlying medical conditions. Detailed data on adverse drug reactions and laboratory monitoring and the reasons for deviation from the treatment guidelines were not available. The LTBI treatment regimens examined in this study were daily INH therapy and INH/RMP combination therapy using standard doses of INH and RMP. ⁵

The primary outcomes included the percentage of patients in each treatment group who completed therapy, the percentage of patients with adverse reactions who discontinued therapy, and the number of patients who completed therapy but later developed TB during the 6-year follow-up until 2008. Treatment regimens, treatment duration and completion, and adverse reactions were determined and recorded by the treating clinics. Treatment completion was documented in the database for each patient without comment on specifics about the case. We used χ² tests to compare proportions. Multivariate logistic regression was used to adjust for confounding variables to identify significant demographic factors in the development of tuberculous disease after treatment for LTBI. A 2-sided P value < .05 was considered statistically significant. All analyses were completed using SPSS 19.

Results

We abstracted information on 50 858 cases of LTBI who received treatment during this 8-year period. The racial distribution included 53% Hispanics, 17% non-Hispanic Caucasians, 23% African Americans, and 6% Asians. Thirty-eight percent were foreign born; 68% of foreign-born patients were from Mexico, 7% from Vietnam, 2% from the Philippines, and 2% from India. In sum 50 578 cases (99%) were treated with INH therapy, and 280 cases were treated with the combination of INH/RMP. The duration of treatment was 6 to 12 months in the INH group and 4 to 12 months in the combination group. The reasons for deviation from the standard treatment with the 9-month INH regimen were not consistently available in the database. Sixty-one percent (n = 31 124) completed their treatment for LTBI. Of the 151 patients in the combination group who completed treatment, 10% took treatment for 4 to 6 months, 58% for 6 months, 26% for 7 to 9 months, and 6% for 10 to 12 months. Eighteen percent (n = 49) of the combination group experienced adverse reactions and did not complete therapy. Sixty-nine percent of these patients took the combination regimen for at least six months.

We focused our comparative analysis on patients who had a positive PPD and a normal chest X-ray to create more homogeneous comparison groups. This step eliminated patients with abnormal chest X-rays and patients with no recorded PPD. The initial comparisons included age, race, sex, country of origin, and drug regimen (Table 1). This analysis suggested that TB reactivation was more frequent in patients aged 19 to 59 years compared to patients younger than 19 and patients older than 59 (P < .01), in black patients compared to white patients and Asians (P < .001), and in patients from the United States compared to foreign born (P < .002). Multivariate logistic regression analysis indicated that age had a statistically significant effect and that patients in the age group of 19 to 59 had a higher likelihood reactivation than patients 18 years or younger (odds ratio, 2.09; 95% confidence interval, 1.14 to 3.85; P = .02) (Table 2). This regression analysis also suggested that patients from Mexico were less likely to reactivate than patients from the US (odds ratio, 0.59; 95% confidence interval, 0.35 to 1.01; P = .053). Drug regiments did not appear to have a significant effect on reactivation. Only 1 patient in the INH/RMP group reactivated during follow-up, and this individual was lost to follow-up during the management of his latent tuberculous infection. No patient who had a normal chest X-ray and completed the combination INH/RMP therapy developed active TB during follow-up. Approximately 65% of patients treated with INH who reactivated took the drug for 6 months or less.

OPEN IN VIEWER

Table 1.

Demographic Factors and Tuberculosis Reactivation, No. (%) ^a

Factor	Positive	Negative	P
Age, y	117	33 101	.01
< 18	12 (10.3)	7179 (21.7)
19-59	101 (86.3)	2507 (75.7)
> 60	4 (3.4)	849 (2.6)
Sex	174	44 155	.93
Male	111 (63.8)	27 757 (62.9)
Female	63 (36.2)	16 398 (37.1)
Race	173	43 615	< .001
White	99 (57.2)	31 139 (71.4)
Black	65 (37.6)	9838 (22.6)
Asian	9 (5.2)	2583 (5.9)
American Indian	0	55 (0.1)
Country of birth	171	41 163	.002
US	130 (76)	25 928 (63)
Mexico	28 (16.4)	10 611 (25.8)
Foreign (except Mexico)	13 (7.6)	4624 (11.2)
Drug regimen	169	43 736	.55
Isoniazid	168 (99.4)	43 528 (99.5)
Isoniazid + rifampin	1 (0.6)	208 (0.5)

a

Some information was not recorded in the database, and the number of patients in the various sections differs.

OPEN IN VIEWER

Table 2.

Logistic Regression Analysis and Tuberculosis Reactivation ^a

Factor	Adjusted Odds Ratio (95% Confidence Interval)	P
Age, y
< 18	—	—
19-59	2.09 (1.14-3.85)	.02
> 60	2.59 (0.83-8.07)	.10
Race
White	—	—
Black	1.12 (0.72-1.74)	.61
Asian	1.28 (0.38-4.34)	.70
American Indian	0	.99
Country of birth
US	—	—
Mexico	0.59 (0.35-1.01)	.053
Foreign (except Mexico)	0.46 (0.17-1.23)	.12
Sex: Male	1.18 (0.78-1.77)	.43
Drug: Isoniazid + rifampin	1.94 (0.27-14.03)	.51

a

Dash (—) indicates reference.

Discussion

This study has all the difficulties expected with a retrospective data review that involved multiple sites, multiple physicians, and multiple clinic staff members providing patient care and documentation. However, it does include a very large number of cases in a state with a large immigrant population at risk for drug-resistant LTBI, and it suggests that the combination therapy with INH/RMP for LTBI prevents active TB as effectively as the standard treatment with INH. Sixty-seven percent of patients who completed the monotherapy with INH and still developed TB took INH for only 6 months. The higher number of TB cases in our monotherapy group could be reasonably explained by either a less optimal duration of treatment or less protective efficacy possibly secondary to INH resistance. Linas and coworkers assumed that the percentage reduction in TB was 30% in contacts completing 3 to 5 months of INH and 60% in those completing 6 to 8 months. ⁴ More patients in our combination therapy group had adverse reactions and stopped treatment. This was probably related to the prolonged therapeutic regimen in this group, since more than 70% of patients with adverse reactions took the treatment for more than 4 months.

This study highlights the low completion rate for prophylactic therapy in fieldwork and identifies 2 important issues. Are shorter course regimens safe and effective? Can compliance and completion rates be improved? The completion rate in this study is comparable to those in similar studies and is close to the assumed rate used by Linas et al in their analysis.^4,6 There is an inverse relationship between the duration of LTBI treatment and the completion rate, and the 9-month INH regimen is a risk factor for failure to complete LTBI treatment (odds ratio, 2.08; 95% confidence interval, 1.23 to 3.57). ⁷ Completion rates of 4-month RMP monotherapy have ranged from 72% to 91% in retrospective, prospective, and randomized controlled studies.^6,8,9 Studies of 3- or 4-month INH/RMP regimens have reported completion rates of 84% to 93%.^10-15 Studies using this combination for 3 to 4 months have reported a severe adverse event rate from 2.3% to 7.3 %.^10-13,16-18

None of the patients in the INH/RMP group with normal X-rays who completed therapy in our study developed active TB during the 6-year follow-up period. Cases of active TB have also occurred at very low frequency in other INH/RMP combination studies except for 1 randomized controlled trial from Hong Kong, which included patients with silicosis.^10,11,14,18 Uncontrolled clinical series in the United States and Canada have also demonstrated that the INH/RMP regimen is at least as effective as INH monotherapy for LTBI treatment.^12,19-21 In addition, more protection may occur with INH/RMP combination in populations with a high prevalence of INH-resistant TB. A study of 6-month RMP therapy in adults exposed to INH-resistant TB showed no active cases developed during a 2-year follow-up, equivalent to a minimal protective effect of 56%. ²¹ Khan’s decision-analysis model suggests that a RMP-based regimen should be used to treat LTBI in foreign-born persons from Vietnam, Haiti, and the Philippines due to the high prevalence of INH-resistant TB in those countries. ²² Therefore, the available literature supports the conclusion that shorter RMP-based regimens have good efficacy and potentially higher completion rates.

Most studies of LTBI treatment, including ours, report that the compliance with prophylactic therapy is low. Our study also suggests that some treatment decisions seemed inconsistent with guidelines. This may reflect inadequate staffing and/or staff turnover, incomplete data collection, insufficient resources available to public health clinics, discontinuities in patient care, and other non-health-related factors affecting immigrants and poor populations. Clearly, low compliance rates reduce the individual and population benefits from LTBI treatment. More involvement of primary care practices and other community health organizations could improve compliance. Kerse and coworkers studied the effect of physician-patient relationship on medication compliance in primary care offices. ²³ Higher levels of physician-patient concordance were associated with improved compliance; the fact that the physician was the usual source of care also appeared to improve compliance. Therefore, joint management between the public health department and local primary care physicians might improve compliance and increase completion rates. Primary care physicians could reinforce the importance of prophylaxis and monitor patients for side effects. Studies in India have demonstrated that a mixed model of private practice and public health services can improve TB detection and treatment using directly observed therapy. ²⁴

Poor compliance occurs more frequently in poor and disadvantaged populations. The state health departments need to consider involving community resources in patient monitoring and drug delivery. Larson et al published a review of strategies and organizational structures needed to increase community participation in health initiatives for marginalized populations. ²⁵ This approach depends on the inclusion of neighborhood organizations, churches, public schools, and grassroots groups. This requires community stakeholders to have equal footing with medical personnel and an iterative process with constant review of the process and outcomes. Mayo and coworkers reported results from a collaborative project for the detection and treatment of TB in a homeless shelter, which involved public health nurses, nurse practitioners, and school of nursing faculty and students. This project improved screening and increased the rate of completion prophylaxis to 77%. ²⁶ The state departments of health could involve other health providers, such as school nurses, local hospitals, and special clinics for disadvantaged patients, in the comanagement of these patients. In addition, community organizations, such as churches and advocacy groups, could help with monitoring, provide lay education, and provide resources for transportation.

Our study has definite limitations, and these include incomplete data collection and inconsistent adherence to guidelines. The retrospective study design and unbalanced treatment groups prevent definite conclusions regarding the efficacy of short-course INH/RMP in our state. However, the literature does support the implementation of this regimen in TB clinics that provide LTBI treatment mainly to foreign-born persons, presumably with a higher likelihood of primary INH resistance. This regimen should also reduce administrative costs and increase completion rates.^5,27 The strengths of our study include the large number of cases and a realistic look at outcomes in routine field use. Departments of health need to consider the use of process indicators to evaluate program strengths and weaknesses in the delivery of TB care, since the long-term follow-up needed to track reactivation is not a practical method for program review. In addition, LTBI treatment has inherent difficulties related to the prolonged times used in prophylaxis, especially for poor populations and immigrants who do not have adequate access to health care or a good understanding of treatment goals. Therefore, departments of health should consider the involvement of primary care physicians, nonphysician health care workers, and community organizations in the treatment of LTBIs and active TB infections to enhance completion rates.