In 2014, tuberculosis (TB) surpassed HIV as the leading cause of death from an infectious disease worldwide and, despite advances in prevention and treatment, it remains the leading cause of morbidity and mortality among persons living with HIV (PLHIV) [1]. In Swaziland, the burden of both diseases is particularly high. In 2016 it had the highest HIV prevalence in the world (27% for adults aged 15–49), a TB incidence rate of 398 cases per 100,000 population, and the second highest prevalence of TB and HIV coinfection (with 70% of TB patients estimated to be HIV-positive) [2–4].

To reduce the burden of TB among PLHIV, the World Health Organization (WHO) has long recommended a set of collaborative TB/HIV activities including TB preventive therapy, intensified TB case finding, and TB infection control measures for PLHIV [5]. Additionally, TB patients should be screened for undiagnosed HIV infection and initiated on early antiretroviral therapy (ART) if found to be HIV-positive, which has been shown in randomized controlled trials to reduce mortality as well as loss to follow-up [5–9]. Current guidelines recommend that ART is started within two weeks of TB treatment initiation for patients with profound immunosuppression (a CD4 count less than 50 cells/μL) to reduce morbidity and mortality, and within eight weeks for all other cases (with the caution that immediate ART in the context of TB meningitis may worsen outcomes) [5,10–12].

In accordance with WHO guidelines, the Swaziland Ministry of Health and partners have been implementing collaborative TB/HIV activities over recent years, with most public facilities now providing co-located TB and HIV services. As of 2014, 97% of TB patients had a known HIV status. However, only 79% of notified HIV-positive TB patients were reported by WHO to have initiated ART, and TB-related mortality remained unacceptably high among PLHIV (135 deaths per 100,000, compared with 51 deaths per 100,000 within the general population), even in the face of declining TB notification rates [1, 13]. In addition, the overall timing of ART initiation relative to TB treatment in Swaziland—a marker of TB/HIV program quality and predictor of outcomes—remained largely unknown.

We conducted a retrospective review of routine programmatic facility data to evaluate the provision of timely ART for HIV-positive TB patients in Swaziland, in order to inform recommendations to the National TB Control and National AIDS programs for improving collaborative TB/HIV services.

Over the last decade, after declaring HIV and TB to be public health emergencies, the Government of the Kingdom of Swaziland and partners have prioritized efforts to strengthen collaborative TB/HIV activities, and to reduce the burden of both diseases in the country [13–14]. This retrospective study of known HIV-positive TB patients demonstrates the relative success of integrated and co-located TB/HIV services in Swaziland in terms of ART uptake. Approximately 40% of people diagnosed with TB were already on ART before they began TB treatment, and 99% were on ART by six months after TB treatment initiation. This latter statistic demonstrates near-perfect compliance with 2015 national guidelines to provide ART to all known HIV-positive TB patients regardless of CD4 cell count [15], and is exceptional when compared with published literature from other integrated settings in sub-Saharan Africa (which have shown ART uptake among co-infected patients ranging from 58–80%) [16–22]. ART uptake elsewhere among HIV-positive TB patients is even lower in non-integrated settings [23–25], and was estimated by WHO at the time of this study to be only 78% worldwide [1].

In addition to overall ART uptake, the timing of ART initiation relative to the start of TB treatment is important [5,10–12]. In our study population of co-infected patients not already on ART at the time of TB treatment initiation, almost 90% initiated ART within eight weeks as per Swaziland national guidelines, with a median time to initiation of 15 days. This also far surpasses achievements in similar settings, which have found 59–78% ART initiation within eight weeks [16,18–19,21–22] and median time to initiation ranging from 34 to 75 days (except in one Malawi study, which achieved a median 14 days to ART initiation in the context of newly integrated TB/HIV guideline implementation) [16–19,21]. This relative success in Swaziland perhaps validates recent initiatives such as task-shifting to nurse-led ART initiation and successful integration of HIV services into TB clinics. Indeed in Shiselweni, the region supported by Médecins Sans Frontières (MSF) to initially and most rapidly implement task-shifting and treatment decentralization within Swaziland, co-infected patients had higher odds of receiving timely ART uptake in our study than in other regions.

Despite this overall success, when stratified by baseline CD4 cell count, we found that only about three quarters (73%) of co-infected patients not already on ART actually received what is considered to be timely ART per WHO best practice guidelines. Although almost 90% of those with a CD4 cell count over 50/μL began ART within the recommended eight weeks of TB treatment initiation, only about 45% of patients with a CD4 cell count less than 50/μL (those with highest mortality risk who would benefit most from early ART) were found to have received ART within two weeks. Although all patients in this latter group were eventually documented to have started ART (indicating that they remained in care at least until six months of follow up), this finding indicates room for improvement. The fact that the proportion of patients receiving ART within two weeks was only slightly higher for patients with a CD4 cell count less than 50/μL than for those with a CD4 cell count more than 50/μL suggests that providers were not optimally differentiating the timing of ART for co-infected patients based on their CD4 count. In addition to reinforcement of national guidelines, sensitization of providers to the benefits of early ART—especially for the sickest TB patients in whom they may currently be hesitating to start ART soon after TB treatment (perhaps due to concerns about pill burden and/or immune reconstitution inflammatory syndrome)—will be critical.

CD4 cell count at TB treatment initiation was not documented in almost a quarter of patients, which is consistent with reports of reagent stock outs and/or other laboratory shortages during the study period. Interestingly, PLHIV with a missing CD4 cell count had similar timing of ART initiation to those with a documented CD4 less than 50 cells/μL. We suggest that, for those in whom a CD4 cell count is missing (e.g., not point-of-care or otherwise easily available), timely ART should mean initiation within two weeks of TB treatment—especially in settings where PLHIV still present late to care—to ensure early ART for potentially severely immunocompromised patients who may be asymptomatic at WHO staging. This may become additionally important if programs opt-out of baseline CD4 testing in the context of new WHO guidance to treat all PLHIV regardless of their immune status [9].

Overall, 85% of co-infected patients in this cohort were documented to have achieved TB cure or to have completed TB treatment, compared with a 71% WHO-estimated TB cure rate for PLHIV in 2014, and a 79% treatment success rate reported in Swaziland’s 2014 Annual TB Programme Report of national-level data [26–27]. Less than 3% of patients experienced documented treatment failure or defaulted from treatment, although 7% died during follow-up. Timely ART initiation relative to TB treatment did not result in significantly different TB treatment outcomes at six months. This has been shown in other studies as well, although several randomized controlled trials have supported the fact that early ART does improve survival specifically for those with very low CD4 cell counts [10–12,24,28–30]. Unfortunately, we were not able to assess HIV treatment outcomes for our study population in order to infer any effect of ART timing on these outcomes.

The major strength of this study is that it drew from routine programmatic data from a diverse set of public health facilities in Swaziland, three years after the national TB/HIV guidelines were revised, and can thus be used for programmatic evaluation and planning. There are, however, some limitations. Sites were selected purposefully and thus may not be nationally representative, and the number of patients included in the study who were not already on ART at the time of TB treatment limited us to detection of medium to large effect sizes (i.e. odds ratios). Using a retrospective chart review, we were unable to make temporal observations (such as potential reasons for ART delay), and data accuracy depended on how well patient-level data was entered into site registers and charts. Additionally, because of logistical constraints we only followed patients for six months after TB treatment initiation, and thus were unable to measure longer-term HIV treatment outcomes (such as viral suppression), which may have been affected by timeliness of ART. Finally, we cannot specifically draw conclusions about improvements specifically due to integration of TB/HIV services in Swaziland, since this is the first time systematic data has been collected in the country on ART timing relative to TB treatment. However, results from this study can serve as a baseline from which to measure future progress.

This assessment of TB/HIV services provided in Swaziland’s co-located clinics found an exceptionally high degree of timely ART uptake among known HIV-positive TB patients, although less so among patients with low or missing CD4 cell counts. This data is being used as an assessment of program quality, and efforts are occurring in country to address delayed ART uptake among patients with CD4 counts less than 50 cells/μL and the high degree of missing CD4 cell counts. However, unlike several studies in other settings, this provides evidence that near-perfect levels of ART uptake for HIV-positive TB patients (albeit with some delay) can be achieved in integrated but low-resource TB/HIV settings, advancing efforts to reach ambitious 90-90-90 targets and to reduce HIV-associated TB incidence in Swaziland. Indeed, as ART has been scaled up in Swaziland, TB case notification rates have concurrently declined [13], and a national TB prevalence survey currently underway will hopefully confirm a true decrease in HIV-associated TB. However, further intensification of TB case finding (currently estimated at only 61% by WHO) [2] would likely unearth more “missed” HIV-positive TB cases who could benefit from timely ART services.