The authors have declared that no competing interests exist.
About 85% of Zimbabwe’s >1.4 million people living with HIV are on antiretroviral treatment (ART). Further expansion of its treatment program will require more efficient use of existing resources. Two promising strategies for reducing resource utilization per patient are multi-month medication dispensing and community-based service delivery. We evaluated the costs to providers and patients of community-based, multi-month ART delivery models in Zimbabwe. We used resource and outcome data from a cluster-randomized non-inferiority trial of three differentiated service delivery (DSD) models targeted to patients stable on ART: 3-month facility-based care (3MF), community ART refill groups (CAGs) with 3-month dispensing (3MC), and CAGs with 6-month dispensing (6MC). Using local unit costs, we estimated the annual cost in 2020 USD of providing HIV treatment per patient from the provider and patient perspectives. In the trial, retention at 12 months was 93.0% in the 3MF, 94.8% in the 3MC, and 95.5% in the 6MC arms. The total average annual cost of HIV treatment per patient was $187 (standard deviation $39), $178 ($30), and $167 ($39) in each of the three arms, respectively. The annual cost/patient was dominated by ART medications (79% in 3MF, 87% in 3MC; 92% in 6MC), followed by facility visits (12%, 5%, 5%, respectively) and viral load (8%, 8%, 2%, respectively). When costs were stratified by district, DSD models cost slightly less, with 6MC the least expensive in all districts. Savings were driven by differences in the number of facility visits made/year, as expected, and low uptake of annual viral load tests in the 6-month arm. The total annual cost to patients to obtain HIV care was $10.03 ($2) in the 3MF arm, $5.12 ($0.41) in the 3MC arm, and $4.40 ($0.39) in the 6MF arm. For stable ART patients in Zimbabwe, 3- and 6-month community-based multi-month dispensing models cost less for both providers and patients than 3-month facility-based care and had non-inferior outcomes.
Zimbabwe, a country in southern Africa of just under 15 million people, has a high burden of HIV infection, with a prevalence of more than 10% among adults [
One promising strategy for expanding access to HIV treatment, improving outcomes, and increasing efficiency is the implementation and scale-up of differentiated service delivery (DSD) models. DSD models aim to move away from a “one-size-fits-all” approach to HIV service delivery and tailor care delivery to patient needs [
In addition to reducing the burden of care on patients and achieving (at least) non-inferior clinical outcomes DSD models are expected to reduce costs for both patients and providers [
We used resource and outcome data from a cluster randomized, non-inferiority trial of DSD models for patients in Zimbabwe who met criteria as being stable on ART (
The primary study outcome was retention in care at 12 months, defined as 1-attrition. Attrition included loss to follow-up and all-cause mortality; loss to follow-up was defined as not collecting ART for ≥90 days during the follow-up period [
For the analysis from the provider perspective, we followed the same methodology as reported in a sister study conducted in Lesotho [
Costs are reported in 2020 USD (
For the costs incurred by patients, we considered the transport costs to and from facility and CAG interactions, wages lost due to time seeking care and the total time patients spent accessing care. Data on patient costs were collected from a subset of 1,082 participants in all 3 study arms through semi-structured interviews with an 11-item questionnaire at study enrolment and the end of the 12-month follow-up period. Patients were asked about the number of visits to DSD interactions, the time spent on travel to access care, travel cost and distance travelled. We then used these data to estimate the average amount transport costs incurred for a patient to access a facility visit or a CAG meeting. These costs were then multiplied, at the patient level, by the respective number of facility visits and CAG meetings an individual attended.
We assigned a value of lost wages to the opportunity costs of accessing care either at a facility or a CAG visit. To do this, we assumed an average daily wage of daily GDP per capita, which amounted to $1.50 per day, because Zimbabwe does not have a stated minimum wage [
To account for diffences in viral load availability at the time of the study, we calculated the total provider cost if all participants who met the primary outcome (retained in care) were to receive a viral load. In this analysis we therefore included the cost of one viral load and associated clinic visit for all patients who were retained in care but for whom no viral load was recorded in the resource utilization dataset.
The trial enrolled 4,800 participants (1,919 in 3MF, 1,335 in 3MC, and 1,546 in 6MC) (
The trial demonstrated non-inferiority of outcomes in the two community CAG arms, compared to the facility-based arm. Retention in care at 12 months was highest in the 6MC arm (95.5%), followed by 3MC (94.8%) and 3MF (93.0%).
Outcome | 3MF | 3MC | 6MC |
---|---|---|---|
n = 1,919 | n = 1,335 | n = 1,546 | |
Expected interactions per year as per model design | 4 | 5 to 6 | 3 to 4 |
Observed interactions per year | 5.01 (1.00) | 5.02 (0.84) | 3.13 (0.71) |
Facility visits per year | 5.01 (1.01) | 1.99 (0.12) | 1.98 (0.15) |
CAG meetings per year | NA | 3.05 (0.74) | 1.18 (0.63) |
Months of ART dispensed | 11.39 (2.38) | 11.51 (1.96) | 11.28 (2.77) |
Viral load tests conducted | 0.49 (0.49) | 0.45 (0.49) | 0.08 (0.27) |
Total annual cost per patient | $187.04 (185.31; 188.78) | $177.83 (176.19; 179.46) | $167.40 (165.44; 169.36) |
Percent reduction | ref | -4.9% | -10.5% |
Total annual cost per patient retained | $195.06 (194.11; 195.99) | $182.81 (181.80; 183.83) | $172.81 (171.30; 174.31) |
Percent reduction | ref | -6.3% | -11.4% |
Annual cost of first line ART | $164.17 (163.39; 164.95) | $163.45 (162.60; 164.30) | $161.08 (159.60; 162.56) |
Laboratory costs | $6.90 (6.57; 7.24) | $6.43 (6.03; 6.82) | $1.08 (0.89; 1.28) |
Annual facility visit costs | $23.98 (23.82; 24.13) | $9.23 (9.22; 9.24) | $9.22 (9.21; 9.24) |
CAG meeting costs | 0 | $3.70 (3.67; 3.75) | $1.42 (1.39; 1.46) |
Travel cost to facility | $2.51 (2.48; 2.53) | $0.99 (0.99;0.99) | $0.99 (0.98; 0.99) |
Wages lost due to facility visits | $7.52 (7.45; 7.59) | $2.98 (2.97; 2.99) | $2.97 (2.95; 2.98) |
Wages lost due to CAG meetings | NA | $1.14 (1.13; 1.16) | $0.44 (0.43; 0.45) |
Total per patient per year in care | $10.03 (9.94; 10.12) | $5.12 (5.09; 5.14) | $4.40 (4.37; 4.42) |
Percent reduction | ref | -48.9% | -56.1% |
Time spent travelling to clinic per year (hours) | 13.35 (13.23;13.47) | 5.29 (2.27;5.30) | 5.26 (5.24;5.28) |
Time spent travelling to CAG meetings per year (hours) | 0.00 | 1.03 (1.02; 1.05) | 24 (23; 25) |
Total travel time per year (hours) | 13.35 (13.23;13.47) | 6.32 (6.30; 6.35) | 5.66 (5.64; 5.68) |
Months of ART dispensed did not differ significantly among any of the arms, with participants receiving 11.39 months in 3MF, 11.51 in 3MC, and 11.28 in 6MC. The number of viral load tests conducted—intended to average 1 per year—was very low in 6MC, with only 0.08 tests done per patient over the 12-month study period. This number rose to 0.49 and 0.45 in the 3MC and 3MF arms, respectively.
We estimated the all-inclusive cost of a facility visit as $4.62 and of a CAG interaction as $1.88. From existing literature, we assumed a cost of $14.42 per viral load test and $13.81 for a month of first-line ART [
As shown in
When stratified by setting (
Outcome | 3MF | 3MC | 6MC | |||
---|---|---|---|---|---|---|
Urban | Rural | Urban | Rural | Urban | Rural | |
387 | n = 1,532 | 387 | n = 948 | 337 | n = 1,209 | |
Expected interactions per year as per model design | 4 | 4 | 5 to 6 | 5 to 6 | 3 to 4 | 3 to 4 |
Observed interactions per year | 4.89 (0.61) | 5.05 (1.08) | 4.93 (0.46) | 5.06 (0.95) | 2.96 (0.43) | 3.18 (0.76) |
Facility visits per year | 4.89 (0.61) | 5.05 (1.08) | 1.99 (0.10) | 1.98 (0.12) | 1.99 (0.11) | 1.97 (0.16) |
CAG meetings per year | NA | NA | 2.95 (0.3) | 3.09 (0.85) | 0.98 (0.38) | 1.23 (0.67) |
Months of ART dispensed | 11.63 (1.85) | 11.34 (2.50) | 11.82 (1.43) | 11.38 (2.13) | 11.59 (3.57) | 11.19 (2.50) |
Viral load tests conducted | 0.96 (0.19) | 0.36 (0.48) | 0.96 (0.19) | 0.22 (0.42) | 0.03 (0.18) | 0.09 (0.28) |
Total annual cost per patient | $196.35 (193.42;199.47) | $184.69 (182.68;186.71) | $189.39 (187.13;191.64) | $173.11 (171.07;175.14) | $171.00 (165.68;176.34) | $166.40 (164.38;168.41) |
Percent reduction | ref | ref | -3.5% | -6.3% | -12.9% | -9.9% |
Total annual cost per patient retained | $200.59 (198.46;202.72) | $193.59 (192.55;194.63) | $192.03 (168.73;178.72) | $178.88 (177.63;180.14) | $173.72 (168.73;178.72) | $172.54 (171.24;173.85) |
Percent reduction | ref | ref | -4.3% | -7.6% | -13.4% | -10.9% |
Annual cost of first line ART | $164.02 (162.24;165.79) | $164.21 (163.34;165.08) | $165.46 (164.36;166.57) | $162.59 (161.47;163.71) | $162.83 (157.86;167.80) | $160.58 (159.31;161.84) |
Laboratory costs | $13.61 (13.26;13.95) | $5.13 (4.77;5.49) | $13.81 (13.52;14.10) | $3.28 (2.89;3.68) | $0.48 (0.19;0.76) | $1.26 (1.02;1.49) |
Annual facility visit costs | $22.96 (22.77;23.16) | $24.25 (24.06;24.43) | $9.23 (9.20;9.25) | $9.23 (9.22;9.24) | $9.24 (9.21;9.27) | $9.22 (9.20; 9.24) |
CAG meeting costs | 0 | 0 | $3.53 (3.51;3.55) | $3.78 (3.72;3.84) | $1.17 (1.12;1.22) | $1.49 (1.45;1.54) |
Travel cost to facility | $2.44 (2.41;2.47) | $2.52 (2.50;2.55) | $0.99 (0.98;1.00) | $0.99 (0.99;1.00) | $0.99 (0.98;1.00) | $0.99 (0.98;0.99) |
Wages lost due to facility visits | $7.33 (7.24;7.42) | $7.56 (7.49;7.65) | $2.98 (2.97;2.99) | $2.98 (2.97;2.99) | $2.97 (2.96; 3.00) | $2.97 (2.95; 2.98) |
Wages lost due to CAG meetings | NA | NA | $1.11 (1.10;1.12) | $1.16 (1.14;1.18) | $0.37 (0.35;0.38) | $0.46 (0.45;0.48) |
Total per patient per year in care | $9.78 (9.66;9.90) | $10.09 (9.98;10.20) | $5.09 (5.06;5.12) | $5.13 (5.10;5.16) | $4.34 (4.32;4.37) | $4.41 (4.39;4.43) |
Percent reduction | ref | ref | -47.9% | -57.19% | -55.62% | -56.29% |
Time spent travelling to clinic per year (hours) | 13.01 (12.85;13.18) | 13.43 (13.29;13.58) | 5.30 |
5.27 |
5.29 |
5.25 |
Time spent travelling to CAG meetings per year (hours) | 0.00 | 0.00 | 1.00 |
1.05 |
0.33 |
0.48 |
Total travel time per year (hours) | 13.01 (12.85;13.18) | 13.43 (13.29;13.58) | 6.30 |
6.33 |
5.62 |
5.67 |
The average annual cost to patients to receive ART care was $10.03 (95% CI $9.94-$10.12) in the 3MF arm, $5.12 ($5.09-$5.14) in the 3MC arm and $4.40 ($4.37–4.42) in the 6MC arm (
To understand variability in provider costs by location, we also estimated average provider cost/patient retained in care by geographic district. Costs varied among districts by roughly 10% in the 3MC arm, 7% in the 3MF arm; and 3% in the 6MC arm (
District | Gutu (n = 535) | Zaka (n = 1,017) | Mberengwa (n = 1,661) | Beitbridge (n = 813) | Chitingwiza (n = 774) |
---|---|---|---|---|---|
3MF | 64 (20) | 134 (25) | 151 (43) | 164 (88) | 352 (96) |
3MC | 13 (17) | 131 (55) | 26 (7) | 34 (18) | 362 (96) |
6MC | 11 (17) | 43 (24) | 41 (5) | 18 (4) | -- |
3MF | $188.97 (1.6; 330) | $192.31 (0.7; 546) | $194.92 (0.8; 347) | $202.98 (1.0; 188) | $200.59 (1.1; 373) |
3MC | $182.71 (0.8; 75) | $186.40 (0.5; 238) | $175.79 (0.9; 381) | $174.45 (2.1; 193) | $192.03(0.6; 378) |
6MC | $177.43 (2.4; 66) | $ 175.98 (1.3; 179) | $171.75 (0.7; 816) | $172.80 (2.2; 416) | -- |
When assuming that all individuals retained in care receive a viral load (
3MF | 3MC | 6MC | |
---|---|---|---|
n = 1,784 | n = 1,265 | n = 1,477 | |
Annual cost of first line ART | $164.17 (163.39; 164.95) | $163.45 (162.60; 164.30) | $161.08 (159.60; 162.56) |
Laboratory costs | 14.42 | 14.42 | 14.42 |
Annual facility visit costs | $ 26.39 (26.21; 26.57) | $11.79 (11.66; 11.92) | $13.49 (13.43; 13.56) |
CAG meeting costs | 0 | $3.70 (3.67; 3.75) | $1.42 (1.39; 1.46) |
Total provider costs | $204.98 (204.13; 205.83) | 193.36 (192.49; 194.24) | $190.42 (188.94; 191.90) |
Percent reduction | ref | -5.7% | -7.1% |
In this economic evaluation of three differentiated models of ART delivery in Zimbabwe, we found that both 3- and 6-monthly dispensing through community ART groups reduced costs for both providers and patients, primarily by reducing the number of visits required to a healthcare facility. For patients who meet eligibility criteria—those who are regarded as “stable” on ART—the two community models also achieved non-inferior retention in care, as previously reported [
In interpreting this finding of cost-effectiveness, a number of provisos should be kept in mind. First, cost savings per patient in the community models were relatively modest (5–10% less than the facility model), and, importantly, these savings will only be realized if resource allocation at facilities is also adjusted. Fewer patient visits to a clinic, for example, will only reduce healthcare budgets if the change allows the facility to pay for fewer human and physical resources. In most cases, clinics do not have the ability to match their physical space, staff complement, or other resources to relatively modest changes in demand for services, at least in the short term. Instead, the “savings” from enrolling patients in less intensive models of care—here defined to mean those requiring fewer clinic visits—are likely to be seen in shorter waiting times for patients, higher quality of care from clinicians, more time available for activities including record-keeping and outreach, and/or more time for staff breaks or shorter working hours. These changes (with the possible exception of the last one) may be of great value to the health system and its patients, but they will not result in monetary savings.
A second consideration is that one reason for the lower cost of the 6MC model, in particular, was the poor uptake of viral load testing in this arm. We believe the reasons to be a combination of geographic access—some districts did not yet have viral load testing capacity, and 6MC was overrepresented in these districts—and the small number of clinic visits made by 6MC patients. While we cannot be certain of the latter reason, the fact that two districts, Mberengwa and Beitridge, tested far more 3MF patients than 3MC or 6MC patients may indicate a need for more investment in compliance with annual viral load testing guidelines if patients receive services in the community, rather than at a facility. A sensitivity analysis in which we assumed that all patients who were retained in care received a viral load still showed cost-savings in the CAG arms, albeit more modest.
And third, while the trial and this analysis used the 3MF arm to represent standard of care in Zimbabwe, the 3MF model in fact provided more care than currently is recommended by national guidelines in Zimbabwe, which require only one clinical consultation a year. We thus cannot be sure that true standard of care costs as much as the 3MF model analyzed here, though our study design accounts for WHO guidelines that recommend clinical visits 3-6monthly for patients stable on ART [
Our findings that DSD models are potentially cost saving to both the provider and patient while achieving similar results in retaining patients in care add to a small but growing evidence base that at least some DSD models cost less than conventional, facility-based care. A parallel study conducted in Lesotho reported a similar saving (6%) to providers and substantial savings to patients [
In addition to the considerations above, our study had two other limitations. First, the total number of interactions per participant was captured in the study, but not the type of visit. The breakdown of visits was therefore based on the model as recommended. Second, we did not collect patient cost data on all patients and therefore had to assign costs on a per-visit basis by visit type, based on average costs reported by the subset of participants who completed the patient cost questionnaire.
Despite these limitations, this study demonstrates that in Zimbabwe, community ART groups can save money for both providers and patients, particularly when combined with 6-month dispensing, without compromising patient outcomes. The consistency of this finding with those of other studies suggests that for stable, adult ART patients, offering community-based models of care makes sense for the healthcare system. The next challenge will be to identify cost-effective models of care for patients who do not meet current criteria for stability, such as a suppressed viral load or experience on ART, and for children, adolescents, and other groups.
(DOCX)
(DOCX)
(PDF)