Date Published: August 13, 2018
Publisher: Public Library of Science
Author(s): Arianna Zanolini, Kombatende Sikombe, Izukanji Sikazwe, Ingrid Eshun-Wilson, Paul Somwe, Carolyn Bolton Moore, Stephanie M. Topp, Nancy Czaicki, Laura K. Beres, Chanda P. Mwamba, Nancy Padian, Charles B. Holmes, Elvin H. Geng, Alexander C. Tsai
Abstract: BackgroundIn public health HIV treatment programs in Africa, long-term retention remains a challenge. A number of improvement strategies exist (e.g., bring services closer to home, reduce visit frequency, expand hours of clinic operation, improve provider attitude), but implementers lack data about which to prioritize when resource constraints preclude implementing all. We used a discrete choice experiment (DCE) to quantify preferences for a number of potential clinic improvements to enhance retention.Methods and findingsWe sought a random sample of HIV patients who were lost to follow-up (defined as >90 days late for their last scheduled appointment) from treatment facilities in Lusaka Province, Zambia. Among those contacted, we asked patients to choose between 2 hypothetical clinics in which the following 5 attributes of those facilities were varied: waiting time at the clinic (1, 3, or 5 hours), distance from residence to clinic (5, 10, or 20 km), ART supply given at each refill (1, 3, or 5 months), hours of operation (morning only, morning and afternoon, or morning and Saturday), and staff attitude (“rude” or “nice”). We used mixed-effects logistic regression to estimate relative utility (i.e., preference) for each attribute level. We calculated how much additional waiting time or travel distance patients were willing to accept in order to obtain other desired features of care. Between December 9, 2015 and May 31, 2016, we offered the survey to 385 patients, and 280 participated (average age 35; 60% female). Patients exhibited a strong preference for nice as opposed to rude providers (relative utility of 2.66; 95% CI 1.9–3.42; p < 0.001). In a standard willingness to wait or willingness to travel analysis, patients were willing to wait 19 hours more or travel 45 km farther to see nice rather than rude providers. An alternative analysis, in which trade-offs were constrained to values actually posed to patients in the experiment, suggested that patients were willing to accept a facility located 10 km from home (as opposed to 5) that required 5 hours of waiting per visit (as opposed to 1 hour) and that dispensed 3 months of medications (instead of 5) in order to access nice (as opposed to rude) providers. This study was limited by the fact that attributes included in the experiment may not have captured additional important determinants of preference.ConclusionsIn this study, patients were willing to expend considerable time and effort as well as accept substantial inconvenience in order to access providers with a nice attitude. In addition to service delivery redesign (e.g., differentiated service delivery models), current improvement strategies should also prioritize improving provider attitude and promoting patient centeredness—an area of limited policy attention to date.
Partial Text: Although public health programs in Africa continue to rapidly start new patients living with HIV on life-saving treatment programs, sustained engagement of these patients—necessary for long-term success—remains a widespread challenge. Existing delivery practices are often part of the problem: programs to date sometimes expect frequent (e.g., monthly) visits to a healthcare facility for medication refills and clinical review, require standing in long queues once at those facilities, and offer only impersonal and brief interactions with healthcare workers . Although programs recognize the need for improvement, few data exist on the comparative effectiveness of many potential innovations, and therefore the way forward remains uncertain. For example, in order to make treatment more accessible, a clinic could choose to extend open hours, increase the quantity of medications dispensed (and therefore reduce visit frequency), build satellite clinics deeper in the community, or numerous other strategies . In most resource-limited environments, programs cannot implement all strategies simultaneously, and so which one to prioritize remains an important unknown.
The choice experiment was offered to a random sample of 385 of 530 persons lost to follow-up identified by a parent study. Of the 385, 105 did not consent, yielding 280 people who were lost, traced, and contacted in the field and who agreed to the experiment between December 9, 2015 and May 31, 2016 (Fig 2). The 280 who received the DCE were very similar to the 105 who refused the DCE on clinical, sociodemographic, and other measures (S2 Table). Of the 280 respondents, the average age was 35 years, 60% were females, and 60% were married. A total of 55% were on ART at the time of loss to follow-up; most (68%) reported having initiated ART. The average time since the last visit at the original clinic was 1.7 years. Overall, 170 (61%) of these lost patients were not in care, and 110 (39%) had reconnected to care either at a new facility or back at their original facility by the end of the experiment.
Among HIV-infected persons who were lost to follow-up in Zambia, a choice experiment provided clear insights about what characteristics of care patients prefer—thus offering immediately actionable information for programs seeking to address the widespread predicament of poor retention. Nice providers emerged as a characteristic of unexpected importance. In a standard willingness to wait analysis, patients were willing to trade 19 hours of waiting time to obtain nice providers. Given the fact that 19 hours exceeded the actual waiting time categories offered in the choice experiment and was a product of a linearity assumption standard in willingness-to-wait analyses, we explored alternative ways of quantifying trade-offs. We attempted a willingness to travel analysis, which suggested that patients would travel 45 km farther to access a facility with nice as opposed to rude providers. An even more conservative additional approach, in which trade-offs were not allowed to exceed values actually asked, found that patients would be willing to accept a facility that required a combination of 5 hours of waiting time (as opposed to 1), was located 10 km from home (as opposed to 5 km), and that gave a 3 (as opposed to 5)-month supply of medications, all in order to access a facility staffed with nice as opposed to rude providers. Although each method offers slightly different ways of quantifying trade-offs, the message is clear and consistent: healthcare worker attitude is critically important to patients. Other salient findings included quantifying willingness to wait or travel in order to access facilities that gave larger quantities of medication, were located nearer to their residence, and had operating hours on Saturdays.