Following the WHO revision of the guidelines for antiretroviral therapy (ART) in 2013 , and subsequently in 2015/2016 [2,3], many countries began the process of transitioning towards providing immediate treatment for HIV, which has also been called Early Access to ART for All (EAAA), ‘Universal Test and Treat’ (UTT), and ‘Treatment as Prevention’ (TasP). The evidence from the HPTN052 trial [4,5] and several HIV epidemiologic studies [6–9] promise a decline in HIV incidence by adopting the EAAA strategy. These studies have demonstrated reductions in HIV incidence and suggest economic benefits of early access to testing and treatment.
Missing from the current evidence on EAAA is how it will affect HIV patients’ satisfaction with care in routine treatment programs . In theory, EAAA could affect patient satisfaction through many pathways. It could have a positive impact on patient satisfaction if patients value the early onset of treatment, which frees them from the psychological burden of delaying treatment while waiting for the HIV disease to get worse. Conversely, it could have a negative impact on patient satisfaction – for example, through the psychological and emotional stress patients might feel if required to commit to life-long treatment immediately after receiving a life-changing diagnosis. EAAA might also cause significant increases in patient volume, which in turn, could lead to less health worker time spent per patient, longer queues, and increased health worker stress – all of which could reduce patient satisfaction. Lastly, and maybe most importantly, EAAA might negatively affect average patient satisfaction through compositional changes to the patient population receiving ART . Patients with late-stage HIV disease are more likely to experience the powerful recovery and health improvement that follows commencement of treatment compared to those with early-stage disease. Individuals in early stages of HIV disease are more likely to still feel healthy when treatment starts and the main physical change following HIV treatment in these patients may be the experience of HIV side effects. This difference in the experience of recovery could affect patient satisfaction.
Patient satisfaction is important for several reasons. For one, it has intrinsic value, because the treatment process should be responsive to patients’ legitimate demands and respectful of their underlying wants [12,13]. Moreover, patient satisfaction also has instrumental value through its effect on patients’ behaviors. Patients who are dissatisfied with their care are more likely to disengage from care [14,15] and fail to reengage after disengagement [15,16]. Dissatisfied patients are also more likely to refuse or delay treatment  and to miss medications if already on treatment [14,18]. Therefore, if EAAA reduces patient satisfaction, it could negatively impact HIV treatment programs and the course of the HIV epidemic.
The Maximizing ART for Better Health and Zero New HIV Infections (MaxART) trial, a pragmatic stepped-wedge cluster-randomized controlled trial (SW-cRCT) of EAAA in Eswatini, was designed to ‘quantify the causal impact of early access to ART for all on patient satisfaction’  as a prespecified secondary endpoint. Viral with viral suppression and retention in care were the primary endpoints. This study, therefore, aims to determine the causal impact of EAAA on overall patient satisfaction, as well as four specific domains of the patient experience (wait time, consultation time, involvement in treatment decision-making, and respectful treatment). To our knowledge, the results below are the first rigorous causal test of the hypothesis that EAAA changes patient satisfaction.
This SW-cRCT was carried out in public-sector health facilities in Eswatini's Hhohho region (Fig. 1). Thirteen government clinics and one regional hospital (14 clusters in total) were randomized into pairs, and then randomized to transition from the standard of care to EAAA at different time points such that two facilities transitioned from the control to the intervention arm every 4 months . Every health facility started in the control phase (C), and at the beginning of each new step, two health facilities transitioned to a 4-month transition phase (T). During the transition phase, the health facility switched from using the existing HIV eligibility guidelines to the EAAA strategy. This process continued until all health facilities had successfully transitioned to the intervention stage (I) where EAAA was fully implemented (Fig. 2). Health workers and study participants were blinded to the transition timing until it happened. Patients were free to attend any health facility they chose throughout the duration of the study. Repeated cross-sectional data were collected from patients visiting a facility on randomly selected clinic days during each step of the study. All health facilities (clusters) had open enrollment for HIV treatment for individuals aged 18 years and older. We enrolled all eligible patients at the beginning of the study (month 1) (Fig. 2).
Ethical approval for the study was obtained from the Eswatini National Health Research Review Board in July 2014 (Reference Number: MH/599C/FWA 000 15267), and a nonhuman subject research exemption was obtained from the Harvard Institutional Review Board as the research team at Harvard, which was responsible for this analysis, only had access to de-identified data.
All patients living with HIV and aged 18 years or older who visited any of the 14 health facilities during the study period were eligible for inclusion in the study. Pregnant and breastfeeding women were excluded, as well as patients who were unable to provide written informed consent.
The primary intervention was implementation of the EAAA strategy. This strategy involved providing immediate access to HIV treatment for all patients living with HIV and attending public-sector clinics. Health facilities in the control arm used the standard of care, which determined treatment eligibility based on prevailing guidelines. From September 2014 to November 2015, these were ART initiation if the CD4+ cell count was less than 350 cells/μl or WHO HIV clinical stage 3 or 4. On the 1st of December 2015, Eswatini adopted new guidelines recommending ART initiation if the CD4+ cell count was less than 500 cells/μl (or WHO HIV clinical stage 3 or 4).
Patients who visited clinics in the control (C) phase were enrolled in ART if they were eligible under existing guidelines, or on pre-ART if they were HIV-positive but not eligible for ART. During the transition (T) phase, all patients living with HIV and visiting the clinic for the first time were automatically enrolled on ART after a session of counselling, while patients who were enrolled in the pre-ART program were transitioned to ART on their next visit. In the intervention phase (I), all patients living with HIV were immediately initiated on ART (Fig. 2).
Our main outcome of interest was patient satisfaction, which was prespecified as a secondary outcome in the MaxART trial published protocol paper . During each step of the study, we selected a random set of data collection days for each health facility. During these days, a random sample of HIV patients were selected by one or two interviewers, using the method of selecting the next patient entering the consultation room . Sampling was not stratified or restricted by HIV service accessed (ART, pre-ART, or HIV testing). The interviewer administered a questionnaire to assess overall patient satisfaction, as well as satisfaction with four different dimensions of the patient experience: wait time, time spent with the health worker, involvement in treatment decision-making, and respectful treatment. Patients responded to each question by selecting the appropriate response on a Likert scale that ranged from ‘very good’ to ‘very bad’. The responses were coded on a 5-point scale from 1 to 5, with 1 = ‘very good’, 2 = ‘good’, 3 = ‘indifferent’, 4 = ‘bad’, and 5 = ‘very bad’. The questionnaire is provided in Supplementary File 1, https://links.lww.com/QAD/B547.
A total of 2629 patient interviews were carried out over the study period from September 2014 through August 2017. All patients were asked questions about overall patient satisfaction, whereas a random sub-sample of 701 patient interviews also included questions on satisfaction with the four specific domains of the patient experience.
To test the causal impact of EAAA on patient satisfaction, we fitted multilevel ordered logistic models [with individuals at the first-level and health facility (cluster) at the second-level] with a random intercept for each facility. In the base model, we regressed patient satisfaction on exposure to EAAA and controlled for potential time trends with a fixed effect for each study step, as recommended for repeated cross-sectional samples [21–23]. As robustness checks, we also fitted models that included the following additional covariates: time since ART diagnosis, time since start of ART treatment, and a measure of overall health status. We fitted each model separately for overall patient satisfaction and for each of the four specific domains of the patient experience.
All models follow the basic specification in Eq. (1) below.
where i indexes the individual, j indexes the cluster, and t indexes the step (time). m is a category with (), τ is the cut point for that category, θ is the treatment impact, T is a binary variable, which takes the value 0 if facility is in the control phase and 1 if facility is in the EAAA phase. X is a vector of other independent variables (which we added in sensitivity analyses), β is a vector of logit coefficients, and δ represents the time-steps. represents the cluster-level random effects, and represents the individual error terms.
We assessed the robustness of the results for overall patient satisfaction by fitting additional regression models, which controlled for important variables that could potentially influence patient satisfaction. These included overall health status, number of months on ART, number of months since HIV diagnosis, age, sex, and educational status. In addition, we conducted sub-group analyses to test for differences in impact of EAAA before and after the switch in treatment threshold during the study from CD4+ cell count less than 350 cells/μl to CD4+ cell count less than 500 cells/μl. To account for time-constant health facility covariates, such as type of facility or catchment-area population, we also fitted a model with facility fixed effects. Furthermore, we tested the effect of duration of implementation of EAAA on patient satisfaction, because duration of implementation might influence results. Finally, we examined background trends in facility patient volume over the duration of the study.
We conducted 2692 patient surveys. Table 1 shows patient characteristics by intervention and control arm. Baseline covariates were balanced across the intervention and the control arms of this randomized controlled trial. The mean age of the entire study population was 38 years (SD 12) while the mean age for intervention and control arm were 38 (SD 12) and 38 (SD 12), respectively. Seventy-two percent of the study population were women, and the percentage of women in the intervention and the control arm were 70 and 75%, respectively. Fifty-five percent of the overall population were married with no significant difference between the intervention (54%) and the control (56%) arm (Table 1).
There were also no significant differences in health status variables. Mean quality of life was 2.1 (SD 0.94) for the entire study population and 2.1 (SD 0.94) and 2.1 (SD 0.95) for intervention and control arms, respectively. The average number of months since HIV diagnosis was 59 (SD 45) for overall population, 61 (SD 46) for the intervention arm, and 56 (SD 43) for the control arms. The average number of months since commencement of HIV treatment was 45 (SD 39), with no significant difference between the intervention (mean 47, SD 41) and the control (mean 42, SD 37) arm.
Impact of early access to antiretroviral therapy for all
In general, patients were satisfied with the quality of care received throughout the study. On a scale of 1–5 (with 1 = ‘very good’, and 5 = ‘very bad’), the average overall patient satisfaction was 1.7 (SD 0.7), whereas satisfaction with the other domains of the patient experience were: 2.3 (SD 1.0) for wait time, 1.7 (SD 0.6) for involvement in treatment decisions, 1.7 (SD 0.6) for consultation time, and 1.6 (SD 0.6) for respectful treatment (Table 1).
We did not find any significant impact of EAAA on either overall patient satisfaction or on any of the four domains of the patient experience that we measured. Our estimations assume proportional odds and the resulting odds ratios (ORs) are interpreted as follows: If we select a certain level m of satisfaction with the patient experience (e.g. ‘good’), the odds ratio compares all the observations in groups greater than m to all the observations in groups less than or equal to m. We did not find any statistically significant impact of EAAA on patient satisfaction or on satisfaction in any of the four domains of the patient experience that we measured. The OR of comparing EAAA to control on overall patient satisfaction was 0.91 (95% CI 0.66–1.25, P = 0.559). The ORs describing EAAA impact were also close to one and insignificant for wait time (OR 1.04; 95% CI 0.61–1.78, P = 0.880), involvement in treatment decisions (OR 0.90; 95% CI 0.62–1.31; P = 0.595), consultation time (OR 0.86; 95% CI 0.61–1.20; P = 0.375), and respectful treatment (OR 1.35; 95% CI 0.93–1.96; P = 0.114). Tables 2 and 3 show the main results.
Although we found no significant impact of EAAA on patient satisfaction, we observed a worsening general trend over time in both the intervention and the control arm, following the first step of this trial (Figs. 3 and 4).
For overall patient satisfaction, the odds of dissatisfaction increased consistently from step 2 to step 6, but this increase only became statistically significant in the last two steps [step 5 (OR 3.07; 95% CI 1.49–6.35) and step 6 (OR 3.89; 95% CI 2.06–7.34)]. A similar pattern was seen for respectful treatment from health workers – the OR of reporting worse satisfaction increased over time but only the last two steps were significantly different from the baseline [step 5 (OR 3.98; 95% CI 1.23–12.84) and step 6 (OR 3.78; 95% CI 1.17–12.24)] (Table 2). Patient satisfaction in all four domains of the patient experience followed a similar trend over time: patient satisfaction worsened steadily across steps 2–6, and the differences from null impact became statistically significant in steps 5 and 6 (Figs. 4 and 5).
We carried out a range of sensitivity analyses (Appendix Tables A1–A3, https://links.lww.com/QAD/B547). In all sensitivity analyses, our findings remained essentially the same. First, when we adjusted for individual covariates – age, sex, and educational attainment – the difference in the EAAA impact estimates were less than +/−2% compared with our main model (Appendix Table A1, https://links.lww.com/QAD/B547). Second, there was a tendency for the EAAA impact sizes to be slightly higher in the period before the background policy change from CD4+ cell count threshold less than 350 cells/μl to CD4+ cell count threshold less than 500 cells/μl. However, these differences were insignificant and in neither of the two sub-groups are the impact sizes significantly different from the null (Appendix Table A2, https://links.lww.com/QAD/B547). Third, we ran a facility fixed effects model to control explicitly for time-constant facility level differences in our estimation of EAAA impact. In this model, the impact of EAAA on patient satisfaction remained close to the null and insignificant (Appendix Table A3, https://links.lww.com/QAD/B547). Fourth, we tested whether the duration of EAAA implementation affected EAAA impact. We found no evidence for a significant effect of implementation duration on EAAA impact (Appendix Table A4, https://links.lww.com/QAD/B547).
Finally, additional analyses of patient volumes over time show that there was an increase in patient volume for all facilities (regardless of EAAA versus standard-of-care status) between baseline and end line with an increase in average monthly visits of 68% (range 15–583%), and an increase in average number of patient-years spent seeking care of 68% (range 19–401%) (Appendix Table A5, https://links.lww.com/QAD/B547). In comparison, the change in patient volume between EAAA and standard of care was comparatively low, with an average increase in monthly visits of 33% across all health facilities (range 8–107%) and an average increase in the number of patient-years spent seeking care of 32% across all facilities (range 9–105%) (Appendix Table A6, https://links.lww.com/QAD/B547).
In this trial, immediate ART (or EAAA) did not change patient satisfaction compared with delayed ART in Eswatini, one of the countries with the highest prevalence of HIV in the world. This lack of impact was not only true for overall patient satisfaction but also for four domains of the patient experience that we measured – satisfaction with wait time, consultation time, involvement in treatment decisions, and respectful treatment. These results remained robust in a wide range of sensitivity analyses.
Patient satisfaction is important both intrinsically  – the treatment process should be responsive to patients’ legitimate demands and respectful of their underlying wants – and instrumentally – satisfied patients are more likely to take up needed treatment, continue care, and adhere well to their medication. For instance, a study of ART patients in Uganda, Tanzania, and Botswana, identified dissatisfaction with long wait times and queues as reasons for poor ART retention and adherence . Other studies have identified one or more of the domains of the patient experience that we have measured as important for ART uptake  and retention [14,15,26].
Our results are thus important for the future of national implementations of immediate ART in sub-Saharan Africa. If immediate ART had reduced patient satisfaction, immediate ART policies would have needed to proceed with caution and further implementation research would have been required to discover and test approaches to ensure that patient satisfaction does not decline with the introduction of immediate ART. Our result that immediate ART did not negatively affect patient satisfaction is indirectly reinforced by the main findings from this trial, showing that immediate ART increased both primary endpoints of retention and viral suppression. According to theories of change based on the extant health systems literature [14–18], we would have expected immediate ART to decrease retention and viral suppression if it also decreased patient satisfaction.
An incidental finding of our trial – patient dissatisfaction increased over the study observation period – may also be important for policy. This change was not causally related to the switch from delayed to immediate ART. Instead, our results suggest that it might have been related to patient volume, which increased substantially over time in both the treatment and the control arm of the trial. Indeed, additional quantitative analyses of our data suggest that the increase in patient volume because of the switch from delayed to immediate ART was small in comparison to the increase due to the general growth of the Eswatini ART program during the period of this trial. The reduction in patient satisfaction with calendar time suggests that it will be important to routinely monitor patient satisfaction, as ART programs in sub-Saharan Africa continue to grow and mature. If patient satisfaction does indeed tend to decline in the future – as our results would suggest – implementation research is needed to determine and address the particular underlying causes for this trend.
Our study had several strengths. It is the first to quantify the impact of immediate ART on patient satisfaction. In addition, our use of a rigorous implementation research design (a SW-cRCT) ensures that our evidence is causally strong. A range of sensitivity analyses that support our main results ensure that the evidence derived from this study is robust and can serve as a foundation for policy decisions regarding immediate ART. Lastly, the location of our study in Eswatini, a country with high HIV prevalence and a large number of patients receiving ART, ensured that our test of the impact of a policy change from delayed to immediate ART had both statistical power and contextual relevance. Similar power and relevance would have been difficult to obtain in settings with substantially lower HIV prevalence.
Our study also had several limitations. The confidence intervals around our point estimates were relatively wide, indicating limited statistical certainty of our findings. However, all of our point estimates of the impact of immediate ART (or EAAA) on patient satisfaction – including those in the sensitivity analyses – were very close to null impact, suggesting that our null findings are indeed because of lack of impact rather than because of power limitations. Another limitation is that we measured the impact of the switch from delayed to immediate ART on satisfaction in some domains of the patient experience but not in others, in particular travel time or the physical conditions of the health facilities. Although the domains that we did not measure are generally important for subjective and objective outcomes of ART programs, it is unlikely that the switch from delayed to immediate ART would have affected domains such as transport and infrastructure. In addition, our findings are likely limited in their generalizability to contexts similar to the study context: rural public-sector – and largely primary care – facilities in sub-Saharan African communities. Our findings may not apply to urban settings or communities with substantially lower HIV prevalence. Finally, our study was designed to establish the impact of immediate (compared with delayed) ART on patient satisfaction; it was not designed to identify the mechanisms through which such impact might occur. For our main finding – immediate ART does not affect patient satisfaction – this limitation is not important; however, for our incidental finding – patient satisfaction decreases with calendar time – mechanistic understanding would have provided valuable insight – both to bolster our confidence that this finding is real and to suggest possible intervention to address this negative trend.
In this randomized controlled trial in Eswatini, we did not find evidence for any causal impact of the switch from delayed to early ART on overall patient satisfaction. We also found no evidence for any impact of immediate ART on satisfaction with four important domains of the patient experience: wait time, involvement in treatment decisions, consultation time, and respectful treatment. At the same time, we observed a strong secular trend of decreasing patient satisfaction in both the intervention and the control arm of the trial, which may have been caused by the rapidly increasing numbers of patients on ART over our observation period. Further implementation research should identify approaches to ensure high patient satisfaction with ART in the long run.
We would like to thank all members of the Ministry of Health involved in the trial, staff and members of the Clinton Health Access Initiative in Eswatini, and all respondents who participated in the study.
Contributions: All authors designed the study. O.O. conducted the analysis and drafted the manuscript. All authors provided comments and approved the final version of the manuscript with T.B.
Other trial information
Registration: This trial is registered with ClinicalTrials.gov number NCT02909218. The study protocol was published in the Trials journal volume 18, issue 1 of 2017  and is available online.
Funding: Support for the study was provided by the Embassy of the Kingdom of the Netherlands in South Africa/Mozambique, Médecins Sans Frontières, Mylan, British Columbia Centre of Excellence in Canada, and the Dutch Postcode Lottery in the Netherlands. The funders had no role in study.
Conflicts of interest
There are no conflicts of interest.
1. World Health Organization (WHO). Consolidated guidelines on the use of antiretroviral drugs for treating and preventing HIV infection: recommendations for a public health approach
. Geneva: WHO; 2013.
2. World Health Organization (WHO). Consolidated guidelines on the use of antiretroviral drugs for treating and preventing HIV infection: what's new
. Geneva: WHO; 2015.
3. World Health Organization (WHO). Consolidated guidelines on the use of antiretroviral drugs for treating and preventing HIV infection: recommendations for a public health approach
. Geneva: WHO; 2016.
4. Cohen MS, Chen YQ, McCauley M, Gamble T, Hosseinipour MC, Kumarasamy N, et al. Antiretroviral therapy for the prevention of HIV-1 transmission
. New Engl J Med
5. Cohen MS, Chen YQ, McCauley M, Gamble T, Hosseinipour MC, Kumarasamy N, et al. HPTN 052 Study Team. Prevention of HIV-1 infection with early antiretroviral therapy
. N Engl J Med
6. Eaton JW, Johnson LF, Salomon JA, Bärnighausen T, Bendavid E, Bershteyn A, et al. HIV treatment as prevention: systematic comparison of mathematical models of the potential impact of antiretroviral therapy on HIV incidence in South Africa
. PLoS Med
7. Granich R, Crowley S, Vitoria M, Lo Y-R, Souteyrand Y, Dye C, et al. Highly active antiretroviral treatment for the prevention of HIV transmission
. J Int AIDS Soc
8. Hontelez JAC, Lurie MN, Bärnighausen T, Bakker R, Baltussen R, Tanser F, et al. Elimination of HIV in South Africa through expanded access to antiretroviral therapy: a model comparison study
. PLoS Med
9. Tanser F, Bärnighausen T, Grapsa E, Zaidi J, Newell M-L. High coverage of ART associated with decline in risk of HIV acquisition in Rural KwaZulu-Natal, South Africa
10. The HIV Modelling Consortium Prevention Editorial Writing Group. HIV treatment as prevention: models, data, and questions—towards evidence-based decision-making
. PLoS Med
11. Bärnighausen T, Salomon JA, Sangrujee N. HIV treatment as prevention: Issues in economic evaluation
. PLoS Med
12. World Health Organization (WHO). World health report 2000 – health systems: improving performance
. Geneva: WHO; 2000.
13. Roberts M, Hsiao W, Berman P, Reich M. Getting health reform right
. Oxford: Oxford University Press; 2008.
14. Mills EJ, Nachega JB, Bangsberg DR, Singh S, Rachlis B, Wu P, et al. Adherence to HAART: a systematic review of developed and developing nation patient-reported barriers and facilitators
. PLoS Med
15. Layer EH, Brahmbhatt H, Beckham SW, Ntogwisangu J, Mwampashi A, Davis WW, et al. I pray that they accept me without scolding:’ experiences with disengagement and re-engagement in HIV care and treatment services in Tanzania
. AIDS Patient Care STDs
16. Ondenge K, Renju J, Bonnington O, Moshabela M, Wamoyi J, Nyamukapa C, et al. ‘I am treated well if I adhere to my HIV medication’: putting patient-provider interactions in context through insights from qualitative research in five sub-Saharan African countries
. Sex Transm Infect
2017; 93: (Suppl 3): pii: e052973.
17. Gourlay A, Wringe A, Birdthistle I, Mshana G, Michael D, Urassa M. It is like that, we didn’t understand each other’: exploring the influence of patient-provider interactions on prevention of mother-to-child transmission of HIV service use in rural Tanzania
. PloS One
18. Roberts KJ. Physician-patient relationships, patient satisfaction, and antiretroviral medication adherence among HIV-infected adults attending a public health clinic
. AIDS Patient Care STDs
19. Walsh FJ, Bärnighausen T, Delva W, Fleming Y, Khumalo G, Lejeune CL, et al. Impact of early initiation versus national standard of care of antiretroviral therapy in Swaziland's public sector health system: study protocol for a stepped-wedge randomized trial
20. Geldsetzer P, Fink G, Vaikath M, Bärnighausen T. Sampling for patient exit interviews: assessment of methods using mathematical derivation and computer simulations
. Health Serv Res
21. Hussey MA, Hughes JP. Design and analysis of stepped wedge cluster randomized trials
. Contemp Clin Trials
22. Hemming K, Haines TP, Chilton PJ, Girling AJ, Lilford RJ. The stepped wedge cluster randomised trial: rationale, design, analysis, and reporting
23. Hemming K, Taljaard M, Forbes A. Analysis of cluster randomised stepped wedge trials with repeated cross-sectional samples
24. Chimbindi N, Bärnighausen T, Newell M-L. Patient satisfaction with HIV and TB treatment in a public programme in rural KwaZulu-Natal: evidence from patient-exit interviews
. BMC Health Serv Res
25. Hardon AP, Akurut D, Comoro C, Ekezie C, Irunde HF, Gerrits T, et al. Hunger, waiting time and transport costs: time to confront challenges to ART adherence in Africa
. AIDS Care
26. Croome N, Ahluwalia M, Hughes LD, Abas M. Patient-reported barriers and facilitators to antiretroviral adherence in sub-Saharan Africa