Monitoring prevention or emergence of HIV drug resistance: results of a population-based foundational survey of early warning indicators in mainland Tanzania

Antiretroviral therapy (ART) extends the lives of persons living with HIV and AIDS and improves the quality of their lives [1, 2]. The rapid increase in ART coverage in low- and middle-income countries, especially in Sub-Saharan Africa, has also led to a clear decline in premature mortality [3, 4]. ART, however, is not without complications, and suboptimal adherence can lead to therapeutic failure and the development of antiretroviral resistance [2]. The development of resistance requires changes in ART regimens with limited therapeutic choices each time resistance occurs. Additionally, individuals with drug-resistant HIV strains can transmit them to infants or sexual and needle-sharing partners, leading to infections that are already resistant to first-line therapy and more difficult to treat [2].

In Tanzania, like most low-income country settings, routine individual-level testing for HIV drug resistance (HIVDR) using laboratory genotyping and phenotyping is not feasible due to resource constraints [5, 6]. To monitor the appearance of HIVDR at the population level, the World Health Organization (WHO) developed a generic protocol for each country to adapt based on the routine patient monitoring system (PMS) in order to monitor and mitigate HIVDR [5]. The protocol focuses on eight elements: (i) creating an HIVDR technical working group to develop and implement the strategy, (ii) monitoring HIVDR early warning indicators (EWIs) at ART sites, (iii) conducting sentinel monitoring of HIVDR during treatment, (iv) conducting HIVDR threshold surveys to assess the incidence of transmitted HIVDR, (v) establishing an HIVDR database, (vi) designating an in-country or regional WHO-accredited HIVDR genotyping laboratory, (vii) conducting a national-level annual HIVDR situation review and making recommendations and (viii) the production of an annual report containing evidence-based recommendations to minimize the emergence and transmission of HIVDR. The EWIs, in particular, are attractive for ongoing monitoring because salient data can be retrieved from national ART PMSs [7], at relatively low cost and quicker than the laboratory based assays for the assessment of HIVDR mutations.

A limited number of studies are available from Africa that assesses EWI performance [6, 8‐11], and none from Tanzania. Furthermore, we could not find any studies that attempt to use programmatic factors as predictors for EWI success. Using four indicators that could be captured using the existing PMS, we undertook a baseline facility-level survey to a) ascertain Tanzanian national performance against WHO targets for four HIVDR EWIs and b) determine the correlates of achievements or failures in reaching the targets.

We invited 35 facilities to participate in the survey, and all agreed. The 35 facilities included 28 public and 7 faith-based facilities; 4 referral hospitals, 17 non-referral hospitals and 14 health centers; and 20 urban, 6 semi-urban and 9 rural facilities. During 2010, a total of 13,466 adult patients began ART at these facilities; we abstracted data from 3,608 (27%) charts. All facilities adhered to the national ART management guidelines regarding eligibility for ART initiation. Six facilities had been providing ART services for longer than five years while 14 had been in operation for less than three years (Table 1).

Of the 3,608 charts we reviewed from sampled facilities, the weighted national average was 100% for EWI 1a, 26% (95% CI 25–26) for EWI 2, 54% (95% CI 53–54) for EWI 3a, and 38% (95% CI 37–38) for EWI 5b (Table 1). Of the 35 clinics, all met the WHO target for EWI 1a, 17 clinics (49%) for EWI 2, 10 clinics (29%) for EWI 3a and 1 clinic (3%) for EWI 5b (Table 2). None of the clinics met all four EWI targets.

Twenty-nine (83%) ART facilities were operating their clinics for at least eight hours per day. Patients at 6 (17%) facilities travelled an average of less than 10 kilometers, whereas those at 16 (46%) facilities travelled over 30 kilometers on average to attend the clinic. At 14 (40%) facilities the average time patients spent in the clinic (which included wait time and time spent with providers) was less than 60 minutes, and at 6 (17%) it was over 180 minutes. Thirty (86%) facilities had no supplementary ART dispensing locations; only two (6%) ART facilities provided both refill sites and outreach services for ART. Twenty-five (71%) facilities provided home-based services.

Facilities that met EWI 2 and 3a were more likely to have their patients spend more than 180 minutes on average in the clinic (p = 0.011 and 0.007, respectively), Table 3. The only facility that met EWI 5b was a faith-based urban referral hospital operating for less than 8 hours per day open.

We found that all surveyed facilities had 100% of patients initiated on appropriate first-line ART regimens (EWI 1a). Meeting EWI 1a is a critical success because initiating patients on a mixture of therapies would a) complicate initiation protocols and require routine genotyping at baseline, which is not currently viable in Tanzania; b) tap into usable therapies that should be conserved for patients who fail first-line treatment; c) preclude the system from monitoring patients who fail first-line and must be switched to second-line regimens, a crucial indicator for detecting resistance.

On the other hand, we found that none of the facilities surveyed met all four WHO EWI targets, and only one facility met EWI 5b (ART clinic appointment keeping during the first 12 months of ART). These results are very concerning, as missing even a few doses - particularly during the first six months of therapy when viral loads are high - can lead to treatment failure and drug resistance [2, 15, 16]. A WHO report found that among 27 African countries surveyed, less than half of the clinics met this target, and globally, less than three fifths of the clinics met EWI 5b [5].

Anecdotally, pharmacy record keeping in Tanzania is inconsistent, therefore missing appointments is the next best proxy for estimating treatment interruption [5]. It is possible, however, that some patients are missing ART clinic appointments but still picking up their antiretrovirals at the pharmacy, so we cannot conclusively say that treatment interruption is occurring at this magnitude. Nevertheless, the results for EWI 5b are alarming. Furthermore, WHO's recently harmonized set of early warning indicators no longer includes EWI 5 [17], which means appointment keeping will no longer be a proxy for ART adherence and possible drug failure.

We also found that facilities whose patients spent the most time in the clinic (more than 3 hours, cumulatively in all departments including wait time), were more likely (without controlling for clinics’ ART cohorts sizes) to fulfill EWI 2 (LTFU 12 months after ART initiation) and EWI 3a (retention on first-line ART at 12 months) than others. Since we are not able to distinguish wait time from time spent with providers, we have two possible explanations. If longer time spent in the clinic is, in fact, due to more patient-provider contact time, there is a sizeable body of literature on patient satisfaction relative to time spent with provider [18, 19], and we speculate that greater patient satisfaction may lead to lower rates of LTFU. We also theorize that patients who spend more time with their providers may understand and adhere to their regimen better, leading to improved retention on first-line regimens. Conversely, if longer visits were due to greater wait times, research shows that patient satisfaction may be high even when wait times are long [20, 21]. Furthermore, it is plausible that the association between longer wait times and higher EWI performance is confounded by higher quality services at some facilities.

Hypothetically, patients may be willing to wait longer at facilities they consider higher quality (with more skilled providers and better diagnostic equipment), while these facilities are also more likely to perform higher in EWIs. Nonetheless, the association between time spent in the clinic and performance in EWI 2 and 3a should be interpreted with caution, as measurement error and respondent bias may be a factor. If providers overestimated the average time they spend with patients, we would expect to see a smaller effect size after taking this into account.

We were not successful in identifying many significant predictors of high EWI performance. Referral hospitals did not perform better than lower level facilities with regard to patient retention on first-line ART (EWI 3a) as we expected. Perhaps this is because more experienced clinicians in referral hospitals may be better equipped to recognize patients who are failing their first-line regimen and move them to second-line therapy. We also expected that facilities with supplemental ART dispensing locations (such as refill sites and outreach services) and home based care would be more successful with LTFU and appointment keeping, but this was not the case. It is possible that with a few sampled facilities and only some facilities providing outreach services, that we did not have sufficient power to detect this difference. Lastly, we anticipated facilities whose patients walk longer average distances would perform worse overall, but perhaps some patients choose to travel to more distal facilities to avoid the stigma of receiving treatment near their homes.

Our study is subject to several limitations. It is the first assessment of its kind undertaken in Tanzania, and in future studies, we hope to obtain better measures for treatment interruption, as well as more robust measures of facility characteristics, such as patient visit times, to better understand how patients spending more time in the clinic results in less attrition and greater retention on first-line ART. Given the human resource shortages Tanzania faces, we are not suggesting that providers should spend more time with patients, but that understanding the factors that contribute to more efficient clinic visits is crucial for balancing human resource constraints with providing both effective and quality care. Secondly, our sample was not perfectly representative of all ART facilities in the nation, and we plan to extend subsequent surveys to areas not covered in this initial survey, including the private facilities, as well as to randomly sample lower-level facilities to avoid selection bias. Finally, there are inherent limitations in the survey methods we employed. We were unable to develop data on the WHO HIVDR EWIs that cannot be obtained from routine patient monitoring, which we plan to address with additional special surveys in our next monitoring round. Additionally, we plan in the future to triangulate EWI data, like those collected in this study, with HIVDR threshold and monitoring surveys, both of which are undergoing development. Despite these limitations, we believe that our data point out successes (initiating ART) and where our system needs immediate improvements (attrition, retention on first-line ART and clinic appointment keeping). It is with periodic surveys such as this that ART services in Tanzania can be improved to provide the best care and treatment for our patients.

In conclusion, Tanzania performed very well in EWI 1a, ART prescribing practices, but its performance in the other three EWIs fell far short of the WHO targets. This study provides a baseline for future monitoring of EWIs in Tanzania. It also conveys lessons learned for surveys that aim to identify practices contributing to the emergence or prevention of HIVDR. EWIs are a quicker assessment technique and they are crucial for tracking HIVDR in resource-limited settings where individual-level monitoring is not feasible. In addition to the national monitoring of EWIs, clinics should be empowered to use their own data to track EWIs more routinely at a facility level in order to address problems or gaps in a timely manner. With better understanding of the programmatic factors that contribute to decreased HIV drug resistance, we may continue to improve the quality of ART services provided for Tanzanians living with HIV and AIDS, and ultimately, the quality of their lives.