Integrating HIV care and treatment into tuberculosis clinics in Lusaka, Zambia: results from a before-after quasi-experimental study

We conducted a quasi-experimental before-after study from June 1, 2010 through March 31, 2012 to assess whether our TB/HIV treatment and care integration intervention had an effect on two primary outcomes: linkage to HIV care and early ART initiation. We defined “linkage to HIV care” as documented evidence of having enrolled in the national HIV program. We defined “early” ART initiation as a patient receiving ART within 8 to 12 weeks of anti-TB therapy (ATT) start. Patients were eligible for study inclusion if they had a recorded ATT start date documented in the TB register, and had documented HIV infection. We excluded patients who: were already on ART at the time of TB diagnosis; transferred into a study site TB clinic from a non-study site; were receiving HIV care through a private clinic or non-governmental organization; started ATT outside the defined pre- and post-intervention time periods; or were concurrently enrolled in another study. No patients with documented multi-drug resistant TB received treatment at primary health centre level during the study period.

We observed outcomes of interest among patients meeting study eligibility criteria at two TB clinics situated within two CIDRZ-supported primary health centres in Lusaka, Zambia—“Clinic A” and “Clinic B”. Observations occurred during an 8-month pre-intervention period (June 1, 2010 through January 31, 2011) and an 8-month post-intervention period (August 1, 2011 to March 31, 2012). Patients who started ATT during the pre-intervention period were considered part of a “pre-intervention cohort”; those who initiated ATT during the post-intervention period were part of the “post-intervention cohort”. The pre- and post-intervention periods were exactly the same duration, and were specified to enable an intervening 1-month “wash-out” window between when the last patient joining the pre-intervention cohort completed their 6-month ATT course and the start of our intervention during the post-intervention period.

Patients in the pre-intervention cohort received the standard of care. Under the standard of care, patients with TB/HIV co-infection are provided TB and HIV care separately in two distinct clinics using a referral-driven model. Typically, the standard of care follows one of two clinical pathways (detailed below) depending on whether HIV or TB is diagnosed first: 1) a newly diagnosed or established HIV-positive patient undergoes evaluation and treatment for TB; or 2) a new TB patient tests positive for HIV.

Newly diagnosed HIV-positive patients may be referred to the ART clinic from any of a number of locations in the facility, including the HCT room, the maternal & child health department, the out-patient department, or other department. At the ART clinic, the patient is enrolled in the national HIV program (requiring registration in the SmartCare electronic medical record system), and undergoes baseline clinical and laboratory evaluation in preparation for ART initiation. Those newly establishing HIV care also receive co-trimoxazole preventive therapy (CPT). All newly diagnosed HIV-positive patients undergo TB screening at the time of their first clinical evaluation by sputum smear microscopy, or, less commonly, by Xpert MTB/RIF (Cepheid, Sunnyvale, CA, USA) where available. Chest x-ray is not routinely available in all health facilities, and where available often requires that a user fee be paid. Patients found to be smear-negative are assessed by a clinician and may be treated empirically based on clinical findings suggestive of TB disease. After establishing HIV care, HIV-positive patients are screened for TB-compatible symptoms at every ART clinic visit per national guidelines [20, 21]. HIV-positive patients diagnosed with TB clinically (i.e. based on symptoms, physical exam, and/or chest x-ray) or microbiologically (i.e. by smear microscopy or Xpert MTB/RIF) are referred to the TB clinic to initiate ATT.

All newly diagnosed TB patients are offered HCT at enrolment into TB care, typically within the TB Clinic itself. To that end, the TB clinic usually houses one or more HCT counselling rooms, and is staffed by one to two nurses and peer health educators. Patients testing HIV-positive are referred to the ART Clinic to establish HIV care and undergo further evaluation for ART initiation as detailed above.

All patients with drug-susceptible TB, irrespective of HIV status, receive first-line ATT with a WHO-recommended, 6-month rifampicin-based fixed-dose combination regimen. ATT is initiated in the TB clinic according to national guidelines from the National Tuberculosis & Leprosy Control Programme (NTP). In Zambia, TB treatment is supported by a lay cadre of peer health educators who provide psychosocial support and adherence counselling to patients.

We implemented the same model of TB/HIV service integration in two routine TB clinics—Clinic A and Clinic B. The integration intervention was introduced first at Clinic A in August 2011, and then at Clinic B in October 2011. The intervention had five core components: 1) health worker training and mentorship; 2) timely provider-initiated HIV testing and counselling (PITC); 3) on-site HIV care enrolment; 4) dedicated ART clinic days; and 5) synchronized TB and HIV patient follow-up. We further describe these components below.

We trained all nurses and clinical officers from both the TB clinic and ART clinic on TB/HIV co-management. Training emphasized proper ATT and ART prescribing practices, and recognition and management of drug-related side effects and toxicity. Importantly, clinicians were trained to begin ART for all co-infected patients as soon as possible, and preferably within 8 to 12 weeks of ATT initiation, regardless of CD4 count. CIDRZ clinician mentors worked intensively with facility staff for approximately 6 months at the start of the intervention to reinforce training concepts. CIDRZ peer educators were trained to provide health education talks to patients on the relationship between TB and HIV and the need for early and sustained co-treatment.

All newly diagnosed TB patients were offered PITC at enrolment into TB care (Fig. 1). TB patients newly identified as HIV-positive who had not previously enrolled in HIV care were informed about the importance of linking to HIV care and starting ART, and were offered immediate enrolment into HIV care on site. MOH nursing staff conducted initial HIV care enrolment procedures in line with national guidelines in place at the time of the study, including: completion of the MOH HIV care enrolment form and blood draws for baseline laboratory tests (including complete blood count, CD4, creatinine, and liver function tests). Following HIV care enrolment, patients were scheduled for an on-site ART initiation visit with a clinical officer to undergo a thorough clinical evaluation, including history, physical exam, and laboratory test result review, and to start ART. HIV-positive TB patients already on ART at the time of ATT initiation were given the option to continue their HIV care at the ART clinic or to transfer their HIV care to the TB clinic for the duration of ATT.

To ensure adequate patient follow-up and efficient use of limited human resources for clinician staffing, each TB clinic scheduled a weekly ‘ART clinic day’ when the same MOH clinical officer was posted to provide “one-stop shop” TB/HIV services, including ART initiation. On the scheduled day, the MOH clinical officer evaluated patients, conducted a physical exam, and reviewed lab results and ART eligibility. Once patients enrolled in HIV care, and initiated ART, they attended the TB clinic according to their directly observed therapy (DOT) schedule (daily, weekly or monthly) for TB treatment. ART and TB follow-up was synchronized to harmonize clinical care and drug collection, and to maximize patient convenience. Specifically, the same clinical officer provided follow-up clinical evaluation for both TB and HIV and attendant co-morbidities and drug-related toxicities, and patients collected both their antiretrovirals (ARVs) and TB medications from the same open-air drug dispensary at the TB clinic. Eligible patients who declined to initiate ART continued to receive TB care in the TB clinic and were encouraged to start ART at each visit. Upon ATT completion, patients were referred to the ART clinic in each facility for continuing their HIV care.

For both the pre- and post-intervention periods, data were collected prospectively during routine clinical care, and recorded onto the following primary data sources: national TB registers, patient TB treatment cards, paper-based HIV care files, and the national SmartCare HIV electronic medical record system. From these data sources, we subsequently abstracted patient data fields of interest, including: age, sex, TB clinic serial number, TB registration date, TB case identification number, TB registration type (i.e. new, relapse, transferred in, treatment after default, or failure), TB anatomical site (i.e. pulmonary TB [PTB] or extra-pulmonary TB [EPTB]), TB treatment start date, recorded HIV status, SmartCare registration number, date of HIV care enrolment, date of initial history and physical exam, date of first CPT dispensation, date of first ART dispensation, and WHO TB treatment outcome [22, 23]. Blank data fields were coded as “missing.” Per routine medical record keeping practice, any patient without a documented HIV care enrolment date is considered not to have linked to care, and any patient without a first ART dispensation date is thought not to have started ART. All data were de-identified at the time of abstraction to protect patient confidentiality. For the pre-intervention cohort, we administratively censored HIV care enrolment and ART initiation 6 months after the end of the pre-intervention period, on July 31, 2011. Similarly, for the post-intervention cohort, we censored HIV care enrolment and ART initiation 6 months after the end of the post-intervention period, on September 30, 2012.

Patient characteristics at the time of ATT initiation (i.e. at “baseline”) were compared between pre- and post-intervention cohorts using the Chi-square test for categorical variables and the Student’s t-test for continuous variables. Simple proportions were used to describe the primary outcomes: 1) the percentage of HIV-associated TB patients newly enrolled in HIV care by 4, 8, and 12 weeks of ATT initiation; and 2) the percentage of HIV-associated TB patients who newly initiated ART by 4, 8, and 12 weeks of starting ATT. We compared the proportion of patients with a primary outcome between the pre- and post-intervention cohorts, stratified by study clinic, using the Cochran-Mantel-Haenszel test. We estimated time to HIV care enrolment and ART initiation using Kaplan-Meier methods with follow-up time measured from ATT initiation. Cumulative failure functions were compared using the Log rank test. Cox proportional hazard modelling was employed to estimate the association between exposure to our TB/HIV integration intervention and ART start by 8 weeks post-ATT initiation. We selected 8 weeks post-ATT initiation as our threshold of interest as this is the WHO-recommended time frame by which ART should be initiated in patients with HIV-associated TB [24]. To estimate the effect of the intervention on successful TB treatment (i.e. TB cure or treatment completion), multi-variable logistic regression modelling was used to determine the odds ratio and associated 95% confidence interval (CI). For both our Cox proportional hazard and logistic regression models, we adjusted for potential confounders that were selected on the basis of clinical and programmatic plausibility, including age, sex, intervention clinic, TB anatomical site, and TB registration type (i.e. new case or cases of relapse, default, or treatment after failure). Patients who transferred out or had missing outcomes data were excluded from the models. We considered two-sided p-values ≤0.05 statistically significant. All statistical analyses were performed using STATA version 12.1 (College Station, TX, USA).

Ethical approval for this study was granted by the University of Zambia Biomedical Research Ethics Committee and the Institutional Review Board at the University of North Carolina at Chapel Hill, without requiring patient consent given the use of de-identified, routinely collected data.

Over the 22-month study period, Clinic A and B registered 578 and 983 total HIV-positive and HIV-negative TB patients, respectively; HIV prevalence among TB patients at Clinic A and B was 68.3% and 55.5%, respectively. A total of 473 patients met study eligibility criteria, including 248 patients in the pre-intervention cohort and 225 patients in the post-intervention cohort.

Patient age and sex distribution did not differ significantly between the pre- and post-intervention cohorts, with a majority of patients being male with a mean age of 33 years (Table 1). Eighty-five percent of patients in both the pre- and post-intervention cohorts were classified as having new TB. A similar majority of patients in both cohorts had pulmonary tuberculosis, with the remaining patients in both cohorts having extra-pulmonary tuberculosis. While 52.8% of patients in the pre-intervention cohort hailed from Clinic A, 65.8% of patients in the post-intervention cohort were from Clinic B—a statistically significant difference (Table 1). The proportion of HIV-positive TB patients not linked to care at baseline as measured by enrolment in the HIV program at the time of ATT initiation was significantly higher in the post-intervention cohort at 74.2%, compared to 64.1% in the pre-intervention cohort (p = 0.02) (Table 1).

A total of 147 patients had documented evidence of HIV program enrolment at the time of ATT initiation, and thus were excluded from the linkage to care analysis. Among patients who were not linked at baseline, a significantly higher proportion of patients in the post-intervention cohort linked to HIV care by 4, 8, and 12 weeks after ATT initiation, compared to patients in the pre-intervention cohort at the same time points (Table 2). By 12 weeks after ATT initiation, 62.9% of patients in the post-intervention cohort had linked to HIV care, compared to 54.7% in the pre-intervention cohort (p = 0.03). Stratified by clinic, we observed a significantly higher proportion of patients in the post-intervention cohort at Clinic A linking to care by 4, 8, and 12 weeks. For Clinic B, we did not observe a statistically significant difference in the proportion of patients in the pre- and post-intervention cohorts linking to HIV care by the time points of interest (Table 2).

In time-to-event analyses, a significantly higher proportion of patients in the post-intervention cohort compared to the pre-intervention cohort linked to care by 4 weeks (53.9% vs. 43.4%, p = 0.03), but not by 8 weeks (59.3% vs. 51.6%, p = 0.07) or 12 weeks (62.9% vs. 54.7%, p = 0.06) (Fig. 2). The median time to HIV care linkage was 59 days in the pre-intervention cohort versus 25 days in the post-intervention cohort.

We evaluated ART initiation in both patients who were linked and not linked to HIV care at baseline (Table 3). The proportion of all patients at Clinics A and B initiating ART in the post-intervention cohort by 4 weeks after ATT initiation was higher in the post-intervention cohort (29.3%) than the pre-intervention cohort (25.0%), however, this difference was not statistically significant (p = 0.18) (Table 3). A significantly higher proportion of patients initiated ART by 8 weeks in the post-intervention (45.3%) versus pre-intervention cohort (38.3%) (p = 0.04). Similarly, by 12 weeks after ATT, significantly more HIV-positive TB patients receiving the intervention had started ART (52.0%) than those who had not been exposed to the intervention (41.5%) (p < 0.01). The proportion of patients starting ART within 4, 8, and 12 weeks of ATT initiation was observed to be consistently higher in Clinic A than Clinic B for both the pre- and post-intervention cohorts.

In time-to-event analyses, a significantly higher proportion of patients in the post-intervention versus pre-intervention cohort had initiated ART early, by 12 weeks (52.0% vs. 41.5%, p = 0.03) after TB treatment initiation (Fig. 3), but not by 8 weeks (45.3% vs. 38.3%, p = 0.11) or 4 weeks (29.3% vs. 25.0%, p = 0.28). Overall, median time to ART was 264 days in the pre-intervention cohort and 78 days in the post-intervention cohort.

In our adjusted multi-variable Cox proportional hazard model controlling for age, sex, clinic, TB anatomical site, and TB registration type, TB patients who received the integration intervention started ART by 8 weeks post-ATT at 1.33 times the rate (HR = 1.33, 95% CI: 1.00–1.77) as patients who were not exposed to the intervention, and this observation was statistically significant (Table 4).

Among the 473 total patients with HIV-associated TB in the pre- and post-intervention cohorts, 284 (60.0%) had a documented TB treatment outcome. Of these, a higher proportion had a successful outcome (i.e. either microbiological cure or documented treatment completion) in the post-intervention cohort (83.3%) compared to the pre-intervention cohort (73.9%), although the difference was of borderline statistical significance (p = 0.051) and the effect was less pronounced in Clinic B than Clinic A (Table 5).

In adjusted multivariable logistic regression modelling, patients receiving the TB/HIV integration intervention were 2.02 times (95% CI: 1.11–3.67) as likely to have a successful TB treatment outcome (i.e. achieving cure or completing treatment) as patients not receiving the intervention (Table 6).