Treatment of latent Mycobacterium tuberculosis infection with 12 once weekly directly-observed doses of isoniazid and rifapentine among persons experiencing homelessness

Objectives To investigate treatment outcomes and associated characteristics of persons experiencing homelessness who received 12-weekly doses of directly observed isoniazid and rifapentine (3HP/DOT) treatment for latent TB infection (LTBI). Methods Among homeless persons treated with 3HP/DOT during July 2011 –June 2015 in 11 U.S. TB programs, we conducted descriptive analyses of observational data, and identified associations between sociodemographic factors and treatment outcomes. Qualitative interviews were conducted to understand programmatic experiences. Results Of 393 persons experiencing homelessness (median age: 50 years; range: 13–74 years), 301 (76.6%) completed treatment, 55 (14.0%) were lost to follow-up, 18 (4.6%) stopped because of an adverse event (AE), and 19 (4.8%) stopped after relocations or refusing treatment. Eighty-one (20.6%) had at least one AE. Persons aged ≥65 were more likely to discontinue treatment than persons aged 31–44 years. Programs reported difficulty in following up with persons experiencing homelessness because of relocations, mistrust, and alcohol or drug use. Conclusions This study demonstrates the feasibility of administering the 3HP/DOT LTBI regimen to persons experiencing homelessness, a high-risk population.


Introduction
Approximately one quarter of the world's population is infected with Mycobacterium tuberculosis without evidence of active TB disease, a condition known as latent tuberculosis infection (LTBI). Reactivation of LTBI, compared with rapid progression from recent infection, is estimated to account for greater than 80% of new TB cases in the United States. [1,2] Marginalized populations, including persons who live in or have lived in congregate settings (e.g., homeless shelters and correctional facilities), have a higher risk of exposure to M. tuberculosis. [3] In the United States, persons with a recent history of homelessness have an estimated TB disease incidence rate of 36-47/100,000 population [4] and have been documented to have more comorbid conditions than other TB patients. [5] Approximately 5% of TB cases in the United States are among persons who reported being homeless sometime within the 12 months preceding diagnosis. [2,6] Studies of persons experiencing homelessness have reported higher rates of LTBI prevalence (20%) [7,8] than that (5%) among the general US population. [9] Targeted LTBI testing and treatment for populations experiencing homelessness is included as an important US strategy for TB elimination. [10] One of the greatest challenges to the success of LTBI treatment programs is inadequate patient adherence. [9,11] As a result of inadequate adherence, the heretofore standard LTBI therapy of 9 months of isoniazid (9H) has completion rates under routine conditions of �50% and much less among persons experiencing homelessness. [12][13][14] The long treatment duration for 9H is reported to be a major reason for lack of patient adherence. [15,16] Additionally, concerns regarding hepatotoxicity exist with 9H. [15] In 2011, the U.S. Centers for Disease Control and Prevention (CDC) recommended use of a short-course regimen of 12 once-weekly directly-observed doses of rifapentine and isoniazid (3HP/DOT) for LTBI. [17] The PREVENT-TB clinical trial provided evidence of higher treatment completion with 3HP/DOT (82%), compared with 9 months of self-administered isoniazid (69%) (P < .01); the trial also determined that 3HP/DOT was non-inferior to 9H when comparing efficacy. [18] An evaluation under typical programmatic conditions documented the substantial improvement in treatment completion with 3HP/DOT over 9H, and provided some of the first data on 3HP/DOT treatment completion among persons experiencing homelessness. [19] However, additional data were needed to help describe factors related to patient adherence and the regimen's acceptability among those experiencing homelessness.
In this study, we investigated treatment discontinuation rates and associated characteristics of persons experiencing homelessness who received 3HP/DOT as LTBI treatment in programmatic settings. We also assessed programmatic barriers and challenges to implementing use of 3HP/DOT among persons experiencing homelessness.

Methods
This study used a subset of data from a previously published larger study [19] from a cohort of 3,288 eligible patients with LTBI who had been followed prospectively from initiation through completion of 3HP/DOT treatment at 16 TB program sites in the United States during July 2011-December 2013. Patients eligible for inclusion in this sub-analysis were 181 clinic patients at 10 program sites who were reported as having been homeless during �12 months before TB evaluation and as having received �1 dose of 3HP/DOT. An additional cohort of 212 persons experiencing homelessness (starting treatment between March 2013 and June 2015) was solicited from a U.S. site using similar processes for collection of data on the same variables. The additional site initiated the use of 3HP/DOT among persons experiencing homelessness after noting a worsening incidence of TB among persons experiencing homelessness in their community. Exclusion and inclusion criteria across sites were consistent with CDC guidelines [17] regardless of start date, with the exceptions of one site having an increased age-threshold in their decision to only treat patients aged �18 years (instead of �12 years of age) and another site having an added criterion that limited treating patients with liver function enzymes >3 times the upper limit of normal. For LTBI diagnostic testing, the majority of sites used QuantiFERON-TB Gold In-Tube Test (Quest Diagnostics, Secaucus, New Jersey) interferon gamma release assays (IGRA) to limit loss of patients observed in the additional clinic visit needed for tuberculin skin test (TST) results; one site used TST primarily.
Although program sites were not selected randomly, they were geographically distributed across the United States. Patients were considered ineligible if they had current active TB disease, were contacts of an index-patient with isoniazid-or rifampin-resistant TB, had a negative test for LTBI (tuberculin skin test [TST] or interferon-gamma release assay [IGRA]) at the time of TB evaluation, became pregnant during LTBI treatment, or were human immunodeficiency virus (HIV)-infected and on antiretroviral medications. [17] The definition of homelessness was consistent with that used by the U.S. Department of Housing and Urban Development developed in the original McKinney-Vento Act of 1987 and reestablished in the Hearth Act amendment to the McKinney-Vento Act of 2012; a homeless individual is defined as one "who lacks a fixed nighttime residence, whose primary residence is a temporary shelter, or whose primary nighttime residence that is a public or private place not designed for or ordinarily used as a regular sleeping accommodation for human beings, including a car, park, abandoned building, bus or train station, airport, or camping ground." [20] Program sites were designated as being in basic (two sites), standard (four sites) or comprehensive (five sites) tiers, depending on the detail of data collected. All sites collected the following patient information: demographics, weekly dose and symptom monitoring for adverse reactions to the regimen at each directly observed treatment (DOT) visit, baseline and monthly laboratory monitoring, and final disposition. Sites in the standard and comprehensive tiers collected additional information regarding medical or behavioral risk factors and concurrent medications during 3HP/DOT treatment.

Quantitative data collection
Quantitative data from 11 sites were entered into a Microsoft Access (Microsoft Corporation, Redmond, Washington) database and analyzed by using SAS version 9.3 (SAS Institute Inc., Cary, North Carolina). Premature treatment discontinuation was defined as receipt of <11 doses of 3HP/DOT over 16 weeks. For descriptive statistics, we determined frequencies and percentages by patient characteristics. We also described frequencies and percentages of patients reporting adverse events (AEs) with and without treatment discontinuation. An AE was defined as any symptom reported since last receipt of a 3HP/DOT dose. Throughout treatment, patients were monitored for symptoms including those common to rifapentine and isoniazid using an identified symptom review checklist. [19] At each weekly DOT visit or at any interim visit or telephone call, patients were asked to report any AE or symptoms experienced since administration of the prior week's medication dose.

Quantitative data analyses
Among persons experiencing homelessness, we conducted bivariate log-binomial analyses to assess independent associations between patient characteristics and treatment discontinuation. We reported relative risk (RR) values with their 95% confidence intervals (CIs). We considered P values of < .05 as statistically significant. We also examined the effect of factors associated with 3HP treatment discontinuation using two multivariable log-binomial models. Factors considered for initial inclusion in the models were chosen based on the statistically significant findings from our bivariate analyses and clinical judgement of our subject matter expert. For the first model, we looked at the effect of patient socio-demographic characteristics and treatment reason on 3HP treatment discontinuation (N = 393). The second model only included patients from sites that collected data on medical and behavioral risk factors (N = 329), and examined the effect of socio-demographic characteristics, treatment reason, medical and behavioral risk factors on 3HP treatment discontinuation (N = 329). In both multivariable models, associations between covariates and 3HP/DOT treatment discontinuation were considered significant at a level of P < .05.

Qualitative data collection
Qualitative data were collected through in-depth interviews with health care workers from a convenience sample of three program sites that treated a substantial proportion of study patients experiencing homelessness. The interviews were conducted to elicit responses about the programmatic experience of using 3HP/DOT to treat LTBI and to gain a more thorough understanding of quantitative results.
Interview questions were developed by using a thematic approach, then divided into three broad themes. The first theme focused on program-related questions (e.g., 3HP/DOT implementation start date, cost implications and staffing requirements). The second theme focused on the process of 3HP/DOT implementation, with certain questions relating to use of incentives and DOT while administering 3HP/DOT as well as description of the 3HP/DOT implementation process. Finally, questions related to the 3HP/DOT implementation process elicited provider perceptions of 3HP/DOT by those involved in the cascade of care. (See Appendix A for the specific list of questions in each broad theme). Two interviews were conducted by telephone, and one was conducted in person. The interview responses were recorded, transcribed into a Microsoft Word document, reviewed with site representatives for accuracy of information, and then collated into a spreadsheet.

Ethics approval
From a human subjects ethics approval perspective, CDC determined this project to be public health practice, not research. Each participating site obtained ethical review and approval for participation in accordance with local requirements.
In the two final multivariable models, no factor was significantly associated with 3HP/DOT treatment discontinuation,-neither when controlling for the association of socio-demographic factors with 3HP/DOT treatment discontinuation nor when controlling for medical and behavioral factors.

Qualitative data
Three sites (7 health care staff) participated in qualitative interviews. All three sites primarily targeted LTBI testing of persons experiencing homelessness, although one site tested additional populations. Staff perceived 3HP/DOT implementation cost to be high compared with standard 9H treatment at all three sites; nonetheless, staff understood that treating LTBI was a worthy investment for prevention of future, more-costly TB cases. Program staff did not receive additional funding for 3HP/DOT implementation; one program manager had convinced providers serving their homeless population outside of the health department setting to seek Medicaid reimbursement for 3HP/DOT. Only one site had hired additional staff (a nurse practitioner) to assist. All sites agreed that dedicated staff time or additional staff were required for expansion of services to contacts at homeless shelters. Sites used multiple resources and tools (e.g., informational brochures, personnel, and mental health or addiction services) to improve outreach to this population. Staff at one site emphasized using clear communication language throughout the entire spectrum of the LTBI testing and treatment cascade. As recommended, 3HP/DOT was administered by DOT at all sites; one site preferred administering medications in the clinic setting, whereas others preferred administering them at their local shelters. Two sites provided incentives to patients, including food (e.g., $5 gift cards for fast-food restaurants) or transportation vouchers; one of these sites had a treatment completion rate of 98%. Site staff expressed personal beliefs that such incentives as bus tokens and gift cards were effective in increasing adherence when given to patients. Site staff did not make substantial changes to their protocols to accommodate treating persons experiencing homelessness.
Site staff were creative in finding external resources to support their LTBI treatment programs. Site managers collaborated with local entities during care delivery; one site partnered with shelter staff in providing transportation and targeted testing, and another partnered with correctional facilities for patients who became incarcerated during treatment. One consistent perception among staff was that they had limited experience with the 3HP/DOT regimen. Hospitals and laboratories also served as partners for completing the evaluations to rule out TB before initiating LTBI treatment; chest radiographs were frequently completed and interpreted at hospitals.
A challenge that programs encountered with treating persons experiencing homelessness was that continued instability in housing led to frequent patient relocation. Some programs reported encountering patients who mistrusted the regimen because it was characterized as new. The high percentage of persons using alcohol and drugs among this population also created additional challenges (such as frequent treatment interruptions) in administering and monitoring treatment.
Programmatic challenges mentioned by clinic staff included staffing and time requirements for treatment by DOT, patients' difficulty with 3HP/DOT pill burden (three 300-mg pills/dose for isoniazid and six 150-mg pills/dose for rifapentine), higher costs of 3HP/DOT relative to Table 2   9H, and difficulty in following up with persons experiencing homelessness. Frequently reported AEs included high blood pressure while on 3HP/DOT medication. Previous and current substance users also reported concerns with triggering feelings or psychological feelings of being intoxicated ("high") during blood draws. All sites agreed that the short duration of 3HP/DOT held great promise for increased adherence and higher treatment completion rates needed for TB prevention among this population.

Discussion
Although multivariable analyses did not yield significant medical, sociodemographic, or behavioral factors associated with treatment discontinuation, results from this study demonstrated that 3HP, a regimen of 12 doses of once-weekly rifapentine and isoniazid with DOT, was successful as a LTBI treatment option among persons who are reported as experiencing homelessness. The US Preventive Services Task Force recommends screening for and treating LTBI among persons in this high-risk population. [3] Compared with historic LTBI treatment completion using 6-12 months of isoniazid among homeless persons (25%-33%), [21] the treatment completion rate (77%) in our study was much higher. [18,19] Given the frequent movement of this population, this high completion rate is difficult to achieve and is consistent with evidence in other populations that 3HP/DOT is positively associated with treatment completion. [18,19,22] As elicited from the interview responses, factors contributing to the high completion were presumed to include short duration of the regimen, [22,23] low rates of adverse events, [24] and use of incentives or enablers (food or transportation vouchers). A large-scale clinical trial (PREVENT TB) found greater rates of overall adverse reactions to 3HP compared with 9H, but lower rates of serious hepatotoxicity. [18,25]    Past studies have documented that difficulty in treating LTBI among persons experiencing homelessness is principally a result of issues of instability (e.g., unstable housing, food insecurity, and high prevalence of mental health conditions), [26][27][28][29] and not necessarily a result of adverse events. Consistent with findings from these past studies, of 92 patients in our study who discontinued treatment, 55 (60%) discontinued because of being lost to follow-up. The rate of noncompletion of treatment resulting from reasons other than adverse events was consistent with the PREVENT TB clinical trial results. In that trial, the North American region reported that the proportion of discontinuation of LTBI treatment not associated with adverse events was 57% higher among participants with a history of homelessness than among participants with no such history (P < .001). [26,30] Our qualitative data collection efforts indicated that the relatively high treatment completion rate achieved with 3HP/DOT also may be attributed to the implementation process adopted by the programs. Two of the three interviewed programs (with treatment completion rates of 98% and 69%) administered treatment with incentives, ranging from food and transportation vouchers to gift cards. Past studies have demonstrated that targeted strategies among the homeless population (e.g., monetary incentives and education programs) improved adherence. [11,31] Specifically, the timing of incentives (as elicited from site representatives) is important for adherence; an immediate monetary incentive is better than a deferred one. [31] In addition to incentives, administering treatment by DOT contributed to the level of success attained in treating this population. All three programs administered therapy by DOT, which has been documented to increase the likelihood that patients take their medications. [18,31,32] DOT delivered to patients where they were residing, such as in local shelter, also likely contributed to improved completion rates. CDC now also recommends use of 3HP by self-administration (SAT). [33] At U.S. sites, the iAdhere clinical trial found lower completion (77%) among all trial participants with self-administered 3HP compared with DOT administration (85%), but found the SAT administration to be non-inferior to that by DOT. [34] Among iAdhere trial participants who experienced homelessness, there was no significant difference in non-completion in adjusted analyses [34].
Project site staff mentioned that DOT, regardless of its effectiveness, contributed to the high cost of administering this therapy and might be an overall deterrent to wider adoption of 3HP under published guidelines that include administration by DOT. [17] Additionally, site representatives indicated that the high cost of rifapentine relative to that of isoniazid is another deterrent; however, since 2014, the price to TB programs of rifapentine was cut in half. [35] Regardless, the benefits of higher adherence to 3HP balances these upfront costs by increasing the number of TB cases prevented. [35] Studies of the cost effectiveness of 3HP (directly observed or self-administered) versus isoniazid in homeless populations are needed and are in progress. Lowering the rifapentine price and allowing 3HP self-administration might dramatically improve the cost-effectiveness of the 3HP regimen. [7,35] The use of Video-DOT can also be leveraged as an alternative to in-person DOT for monitoring treatment. Furthermore, a fixed-dose 3HP pill might also positively impact treatment completion.

Limitations
Given that this study was conducted under field conditions at programmatic sites, this paper includes certain limitations. First, the study was observational; self-selected sites implemented the project under their program's routine practice, which was not standardized across sites. Given the observational design of this project, all persons experiencing homelessness and requiring LTBI treatment may have not been captured, leading to an underestimation of this cohort of patients. Possible limitations in interpreting completion rates might also exist because certain programs differed in instructions for stopping treatment as a result of an adverse event. Other limitations include underreporting of risk factors for low completion rates (e.g., substance abuse, mental health problems, or HIV infection status) because these factors were based on self-reporting. Lastly, the convenience sampling method used in selecting patients for the project may limit the generalizability of findings to other sites.

Public health implications
This analysis demonstrated the feasibility of administering the 3HP/DOT LTBI regimen among persons experiencing homelessness, a population that is at high risk and has a high prevalence of behavioral risk factors that contribute to poor outcomes. This analysis demonstrated higher rates of treatment completion in this population than seen historically with the heretofore-standard 9H regimen. Use of 3HP as a shorter regimen appears to be conducive to completing treatment for LTBI and may play an important role in accelerating the decline of active TB disease in the United States by preventing future infectious TB cases among populations at risk.