We updated a previous policy review conducted by Church et al., [8] in which national HIV policies published between January 2003 and June 2013 were documented in six sub-Saharan African countries. The focus of the updated policy review was to assess the uptake of the 2013 WHO recommendations related to the eligibility threshold for ART initiation, the availability of first-line ART regimes, and recommendations to improve retention. Internet searches were conducted to obtain national guidelines, addendums to the national ART guidelines, and clinical guidelines for the management of ART patients published between July 2013 and July 2015.

We surveyed health facilities located within or on the border of ten HDSS in Kenya (Nairobi and Kisumu), Malawi (Karonga), South Africa (Agincourt and uMkhanyakude), Tanzania (Ifakara and Kisesa), Uganda (Masaka and Rakai) and Zimbabwe (Manicaland) (Fig. 1). These HDSS are part of the network for Analysing Population-based HIV/AIDS data on Africa (ALPHA) [9] and conduct population-based HIV surveillance in eastern and southern Africa (http://alpha.lshtm.ac.uk/). All facilities serving the population of the HDSS sites were sampled except in Nairobi, where a convenience sample was selected and in Ifakara, where only facilities with more than 100 ART patients were included.Fig. 1

Location of the health and demographic surveillance systems

Data from Nairobi, Karonga, Agincourt, Umkhanyakude, Masaka, Rakai, Kisesa and Ifakara were entered into an MS SQL Server (Microsoft Corp) database in Nairobi. Data from Kisumu and Manicaland, were collected on tablets using software developed using the Open Data Kit (ODK) platform (https://opendatakit.org/), and merged with data from the other institutions. Cleaning, merging and analysis of the datasets was carried out using Stata 14 [11].

Descriptive statistics were used to compare selected indicators in each country across the two rounds. The results are presented as the percentage of facilities that reported adherence to each of the indicators in round 1 and 2. Median numbers (with interquartile range (IQR)) were obtained for continuous variables.

Seven policy documents were published between July 2013 and July 2015. We identified one policy document from Kenya [12], Malawi [13], Tanzania [14], Uganda [15], Zimbabwe [16] and two from South Africa [17, 18]. All countries had some explicit policies that aimed to increase ART access in line with the WHO 2013 guidelines on ART access and retention in care (see Additional file 1). Kenya, Malawi, South Africa, Tanzania, Uganda and Zimbabwe had adopted policies on decentralisation and nurse-led ART initiation prior to the WHO recommendations. However, no national policy mentioned that lay cadres could distribute ART in the community. Only Malawi’s policies indicated that a three-month supply of ART should be given to clinically stable patients.

In total, 145 facilities were included in the analysis, with data collected in both round 1 (2013 to 2015) and round 2 (2015 to 2016). The time between round 1 and round 2 averaged 20.9 months (range 16–25 months). Nineteen facilities were excluded (four from Kenya, five from South Africa, two from Tanzania, and eight from Uganda) because they were not surveyed in both rounds.

The majority of facilities were either small health centres or dispensaries (75% in round 1 and 82% round 2) and 81% were government owned in both rounds (Table 2). All facilities had a median of three full-time nurses or midwives (IQR: 1.0–5.0 in round 1 and IQR: 2.0–5.5 in round 2) providing HIV services. The proportion of facilities having doctors or clinical officers providing services was 36% in round 1 and 46% in round 2. In round 1, Kenya had the highest median number of weekly ART patients seen per health care worker (11 (IQR: 5–17)) and Malawi had the lowest (1 (IQR: 0–1)). In round 2, South Africa had the highest median number of weekly ART patients seen per health care worker (21 (IQR: 11–44) and again Malawi had the lowest (0 (IQR: 0–2)). In most facilities the caseloads were <50 ART patients/nurse or clinician per week, with only one facility in South Africa in round 2 reporting caseloads of >100 ART patients/nurse or clinician per week.Table 2

Health facility characteristics

Across all countries 97% of the facilities in round 1 and 96% in round 2 were providing ART free of charge to all patients. In Malawi, all facilities in both rounds had nurses initiating ART. In other sites, the proportion of facilities with nurses initiating ART increased between round 1 and 2 from 76% to 93% in Kenya, and 67 to 100% in Zimbabwe, with smaller changes in Tanzania (60 to 80%), Uganda (79 to 83%) and South Africa (96 and 96%).

All facilities in South Africa and Tanzania (in both rounds) reported that they had an HIV treatment and prevention guideline (for any year) (Table 3). The proportion of facilities that had HIV treatment and prevention guidelines increased in Kenya (87 to 100%), and Zimbabwe (40 to 80%), with smaller changes in Malawi (60 to 80%) and Uganda (84 to 100%).Table 3

Quality of care indicators

In both rounds, 75% of facilities indicated that at least one staff member had received training on ART. Almost all facilities received supervisory visits at least once a year in both rounds: Malawi (80 to 100%), South Africa (91 to 100%), Uganda (95 to 100%) Kenya (95 to 98%), Zimbabwe (100 to 94%) and Tanzania (93 and 93%).

No facilities in Malawi and Zimbabwe reported stock-outs of first-line ART in round 2 (from 20 and 7% respectively in round 1). In other countries, stock-outs were still apparent with little change seen in the proportion of facilites that experienced first-line ART stock-out in the past year between round 1 and 2 (Kenya: 18 to 8%, Uganda: 16 to 17%, South Africa: (13 and 13%) and Tanzania: (33 and 33%). Similarly, the overall proportion of facilities reporting stock-outs of regimens containing tenofovir (TDF) with efavirenz (EFV) was similar between round 1 and 2 (18% in round 1 and 11% in round 2).

The proportion of facilites that were stocked with isoniazid preventive therapy (IPT) increased from 58 to 93% in Kenya and from 87 to 100% in South Africa, but showed little change in other countries: Uganda (0 to 11%), Malawi (100 to 80%), Zimbabwe (53 to 43%) and Tanzania (13 and 13%). All facilities in Kenya, Malawi, and Uganda had stocked co-trimoxazole (CTX) in round 1 and 2, with similarly high proportions in South Africa (91 to 96%) and Zimbabwe (92 to 97%). Tanzania had the lowest proprotion of facilities stocking CTX (60 to 40%).

All facilities providing ART services reported a median increase in the patient load from 32 (IQR: 7–60) in round 1 to 36 (IQR: 10–84) in round 2. Taking the denominator as the number of HIV positive patients who enrolled in the clinic, we estimated the proportion who initiated ART within three months. In Malawi, the proportion increased from 53% in round 1 to 90% in round 2, while in all other countries the proportion was unchanged with 70 to 83% of all enrolled HIV positive patients initiating ART within three months (see Fig. 2).Fig. 2

Overall proportion of patients initiating ART within three months of enrolment at clinic, by country for each round

The proportion of facilities that prescribed the first-line ART regimen (TDF with EFV) recommended in the 2013 WHO guidelines rose between round 1 and 2 in all countries (Fig. 3). This was universally implemented in Malawi in both rounds. In round 1, six facilities in Zimbabwe were still prescribing d4T-based regimens. However, this dropped to one facility in round 2.Fig. 3

Proportion of health facilities starting patients on different ART regimens during the last three months, by country. *KEN 1-Kenya round 1 *KEN 2-Kenya round 2 *MWI 1-Malawi round 1 *MWI 2-Malawi round 2 *ZAF 1-South Africa round 1 *ZAF 2-South Africa round 2 *TZA 1-Tanzania round 1 *TZA 2-Tanzania round 2 *UGA 1-Uganda round 1 *UGA 2-Uganda round 2 *ZIM 1-Zimbabwe round 1 *ZIM 2-Zimbabwe round 2

Table 4 shows that implementation of additional strategies to support PLHIV and improve retention were high in both rounds with few changes noted across all countries (conducting home visits and/or making phone calls to patients who missed their clinic visit within two weeks (89 to 96%), establishment of peer support groups (84 to 88%), routine pill counts (74 and 74%). Similarly, high levels of implementation were noted for the provision of home-based care (94 to 87%), and nutritional supplements and food packages to malnourished patients (78 to 77%) in both round 1 and 2.Table 4

Selected retention indicators

To our knowledge this is the first study to use two sequential cross-sectional surveys to compare implementation of policies on ART access and retention across six African countries with a generalised HIV epidemic. This study also had limitations that should be considered when interpreting the findings. First, the survey was conducted in health facilities serving HDSS sites in each country, and thus the facilities were not selected to be nationally representative. In three large facilities in Rakai, Masaka and Kisumu, additional support was provided for service delivery which could have positively affected the quality of care in comparison to other facilities which operated under typical national supervision. Second, populations within HDSS sites may be different in terms of their treatment-seeking behaviours than those in the general population [31] which may affect the patient volumes witnessed in this study. Third, the completion of the survey by facility managers may present reporting bias and the results may not be a true reflection of the quality of care offered in that facility. Fourth, although the same facilities were used in both surveys, different personnel within the facility may have responded to the survey questionnaire, and any changes in the service may have local reasons (such as personnel changes or transport disruption). Fifth, the analysis on these data were primarily to describe changes, and not to test the significance of any changes.

The conceptual framework that underpinned this study was developed prior to the first facility survey round and was based on a review of literature and policy in circulation up to 2015. Whilst comprehensive at the time, the provision of HIV care and treatment is a rapidly evolving field and it is possible that additional indicators would now be included. Nevertheless this analysis provides a detailed overview of the implementation status of HIV access and retention policies across 145 facilities in six sub-Saharan African countries between 2013 and 2016, a time when considerable changes were taking place at the policy level.