A commitment contract to achieve virologic suppression in poorly adherent patients with HIV/AIDS : AIDS

Secondary Logo

Journal Logo


A commitment contract to achieve virologic suppression in poorly adherent patients with HIV/AIDS

Alsan, Marcellaa,b; Beshears, Johnb,c; Armstrong, Wendy S.d; Choi, James J.b,e; Madrian, Brigitte C.b,f; Nguyen, Minh Ly T.d; Del Rio, Carlosd,g; Laibson, Davidb,h; Marconi, Vincent C.d,g

Author Information
AIDS 31(12):p 1765-1769, July 31, 2017. | DOI: 10.1097/QAD.0000000000001543



Antiretroviral therapy (ART) adherence is critical for HIV treatment to be successful but remains difficult for many individuals to maintain [1,2]. Barriers to adherence include socioeconomic status, mental health, and substance abuse [3–6]. Interventions that can improve adherence and demonstrate sustained virologic suppression for people living with HIV (PLWH) are needed. Conditional cash transfers (CCTs) – monetary rewards tied to adherence – have produced mixed results in improving ART adherence [7–15]. Even when financial incentives have shown positive impacts on adherence and viral suppression, the effect does not persist once incentives are withdrawn [15].

The current study leverages behavioral economics to improve the design of financial incentives for ART adherence [16–19]. Individuals often intend to engage in healthy behaviors in the future but, when the moment to engage in such a behavior arrives, they frequently fail to follow through on their intentions, instead making choices that are expedient at the time. Commitment contracts allow individuals to tie their own hands – by choosing to make future incentive payments contingent on following through on good intentions, individuals can increase their own engagement in healthy behaviors [20,21]. Commitment contracts have proved effective in promoting healthy behaviors, but to our knowledge, they have never been used in HIV care [22–26].

We hypothesized that participants offered a commitment contract for ART adherence would be more likely to be virologically suppressed at the end of the period during which the incentives were in effect, as well as at an unanticipated study visit after incentives had ended. To test our hypothesis, we used a randomized trial design combined with a comparison with a nonrandomized control group, studying patients on appropriate ART having virologic failure within a publicly funded HIV clinic serving Atlanta, Georgia, USA.



The study used a randomized trial design for two treatment arms: first, participants in the provider visit incentive (PVI) arm were told that they would receive $30 after attending each scheduled provider visit (a CCT). Second, participants in the incentive choice arm were given a choice between the above CCT and a commitment contract, which made the $30 payment conditional on the patient attending the provider visit and meeting an ART adherence threshold. A block randomization scheme, stratified on whether or not the majority of the participant's three previous viral load measurements were suppressed, assigned 21 individuals to the PVI arm and 19 to the incentive choice arm.

The study also included 70 individuals in a passive control arm, who did not receive financial incentives. Individuals in the passive control arm were not enrolled in the randomized trial but met basic study eligibility criteria during the same time period.

Inclusion criteria

Participants in the PVI and incentive choice arms were PLWH who attended the Grady Health System Infectious Disease Program (IDP). They were enrolled during November 2011–April 2012 and were followed for a median of 15 months. To be eligible, an individual's most recent HIV-1 plasma RNA viral load (pVL) must have been more than 200 copies/ml and must have been measured within the prior 18 months and at least 6 months after starting the current ART regimen. The pool was further restricted to English-speaking adults who filled prescriptions through IDP, were not using pillboxes, were not planning to relocate, and were not enrolled in another trial.

To create a matched passive control arm based on observational data, we identified individuals via IDP electronic health records (EHR). As recruitment for the PVI and incentive choice arms involved asking clinical staff to refer individuals who had difficulty with adherence, we restricted our search to adults who registered pVL more than 200 copies/ml at some point in 2011 after having been on ART for at least 6 months. To parallel the enrollment process of the PVI and incentive choice arms, we then narrowed the sample to individuals who visited IDP during 2012 and whose most recent pVL was more than 200 copies/ml and measured within the prior 6 months. We further narrowed the sample to individuals who filled prescriptions through IDP and were not in the PVI or incentive choice arms. The 2012 visit was considered the ‘enrollment visit’, and we tracked individuals forward in time from that point.

Description of the intervention

All participants received the standard of care (SOC) at IDP, which included not only medical care, but also a wide range of social services. In addition, participants in the PVI and incentive choice arms received financial incentives designed to motivate health-improving behaviors. After the initial study enrollment visit, participants in the PVI arm received a $30 payment each time they showed up as scheduled for one of their next four HIV primary care visits. At the initial study enrollment visit, participants in the incentive choice arm chose between either the incentive scheme assigned to the PVI arm [Attend Clinic Get Paid (ACGP)] or an incentive scheme that tied payments to clinic attendance and ART medication adherence [Take Medications and ACGP (TMACGP)]. More precisely, participants who selected TMACGP received a $30 payment at each of their next four HIV primary care visits if they showed up as scheduled and presented a dose-recording pill bottle cap indicating that they correctly took at least 90% of doses of a sentinel medication since the previous study visit (refer to Supplementary materials for the algorithm for assigning a sentinel medication, https://links.lww.com/QAD/B107).

Participants in the PVI and incentive choice arms were also asked to return for a sixth, unanticipated study visit approximately 3 months after the last of the four study visits to which the incentive scheme applied. To reduce attrition, participants were offered $100 for showing up to the fifth and sixth study visits.

Data collection

In both the PVI and incentive choice arms, questionnaires were administered at each study visit, and adherence was measured using a dose-recording cap (Aardex Group, Sion, Switzerland).

EHR data were collected for all three arms, with ‘study visits’ for the passive control arm selected to match the study visit schedule for the other arms as closely as possible. HIV-1 pVL was assessed using Abbott Real Time HIV-1 assay (Abbott RT; Abbott Diagnostics, Wiesbaden, Germany). The primary outcome of interest was virologic suppression (pVL ≤ 200 copies/ml, in accordance with Department of Health and Human Services guidelines at the time of the study) at the fifth study visit. A second outcome of interest was virologic suppression at the sixth visit. Missing values for pVL were coded as failures, but in Supplementary materials, https://links.lww.com/QAD/B107, we find similar results using inverse probability weighting to correct for missing values.

Statistical power

The study had 51% statistical power to detect an absolute difference of 30 percentage points between the PVI and incentive choice groups’ rates of virologic suppression at a 5% significance level for a two-sided test. Comparing the incentive choice and passive control arms, the study had 67% power to detect the same difference.

Statistical analysis

We used logistic regression to estimate the unadjusted and adjusted impact of the incentive choice treatment relative to the PVI arm and relative to the passive control arm. The predictor variables included treatment arm indicators and the stratifying variable. Our hypothesis tests were constructed relying on the asymptotic normality of the maximum likelihood estimator, but we obtained similar results when we conducted permutation tests. All statistical procedures were implemented using Stata 13 (StataCorp, College Station, Texas, USA).


Baseline characteristics and data summary

Supplementary materials, https://links.lww.com/QAD/B107 indicate that the three arms had similar demographic and clinical characteristics at baseline, although individuals in the passive control arm were older (median age 48.93 years) than individuals in the incentive choice arm (median 40.10) and PVI arm (median 42.88). Individuals in the incentive choice arm had higher pVL values leading up to the enrollment visit relative to individuals in the other arms. The passive control arm had a higher rate of missing pVL measurements at the fifth study visit compared with the other arms.

In the incentive choice arm, 48% of participants had at least one suppressed viral load measurement across the three visits prior to the enrollment visit. The percentage was 43% in the PVI arm and 36% in the passive control arm. Thus, many individuals in the study had experienced some previous success in achieving viral suppression, whereas some individuals in the study had faced much more difficulty achieving success in the past.

Plasma HIV-1 viral load suppression

Figure 1 shows that, for the three visits prior to enrollment, the percentage of individuals with suppressed viral load measurements was similar across arms. At the fifth study visit, the percentages suppressed were: incentive choice arm 42%, PVI arm 38%, and passive control arm 34%. At the sixth study visit, the percentages were: incentive choice arm 68%, PVI arm 43%, and passive control arm 41%.

Fig. 1:
Percentage virologically suppressed by study arm and visit number.

Table 1 shows logistic regression results. The adjusted odds ratio (OR) of suppression in the incentive choice arm relative to the PVI arm at the fifth visit was 1.57 [95% confidence interval (CI) 0.25–9.92; P value 0.630], and in the incentive choice arm relative to the passive control arm, it was 1.44 (95% CI 0.46–4.49; P value 0.52). At the sixth visit, the adjusted OR of virologic suppression in the incentive choice arm relative to the PVI arm was 3.38 (95% CI 0.77–14.84; P value 0.107), and in the incentive choice arm relative to the passive control arm, it was 3.93 (95% CI 1.19–13.04; P value 0.025).

Table 1:
Virologic suppression in the incentive choice arm compared with the provider visit incentive arm and the passive control arm.


The current study demonstrated the feasibility of using commitment contracts in HIV care. Many previous interventions have produced statistically significant effects on ART adherence that do not persist after the intervention ends. A notable feature of our study is that, after the incentives for ART adherence and provider visits were removed, participants who had been offered a commitment contract for ART adherence were more likely to achieve virologic suppression relative to individuals who had been assigned a CCT for provider visits and relative to individuals who had been assigned the SOC, although the difference was only statistically significant in the latter comparison. There were differences in the prevalence of missing outcomes across groups, but these differences were not statistically significant for the unanticipated postincentive visit and, therefore, were unlikely to be the explanation for the results. Thus, financial rewards coupled with individual choice can increase engagement in healthy behaviors after incentives are removed.

In the face of mixed evidence regarding the efficacy of CCTs for promoting ART adherence [7–15], our results offer a new perspective on the use of financial incentives. When individuals can choose whether or not to make financial rewards dependent on adherence, they may become more adherent both because of the direct incentive effect and because the ability to choose may increase feelings of personal engagement and empowerment in disease management [27].

Replication is needed to address the limitations of our study, including its small sample size, as well as to determine whether similar findings are obtained in other settings.


Financial support for this research was provided by the National Institutes of Health (P30AG034532), the Emory University Center for AIDS Research (P30AI050409), and the Pershing Square Fund for the Foundations of Human Behavior.

Presented in part at the Conference on Retroviruses and Opportunistic Infections (CROI) 2016. Abst. #1039.

Conflicts of interest

There are no conflicts of interest.


1. Paterson DL, Swindells S, Mohr J, Brester M, Vergis EN, Squier C, et al. Adherence to protease inhibitor therapy and outcomes in patients with HIV infection. Ann Intern Med 2000; 133:21–30.
2. Bangsberg DR, Perry S, Charlebois ED, Clark RA, Roberston M, Zolopa AR, Moss A. Nonadherence to highly active antiretroviral therapy predicts progression to AIDS. AIDS 2001; 15:1181–1183.
3. Golin CE, Liu H, Hays RD, Miller LG, Beck CK, Ickovics J, et al. A prospective study of predictors of adherence to combination antiretroviral medication. J Gen Intern Med 2002; 17:756–765.
4. Ickovics J, Meade C. Adherence to antiretroviral therapy among patients with HIV: a critical link between behavioral and biomedical sciences. J Acquir Immune Defic Syndr 2002; 31 (suppl 3):98–102.
5. Kalichman SC, Grebler T. Stress and poverty predictors of treatment adherence among people with low-literacy living with HIV/AIDS. Psychosom Med 2010; 72:810–816.
6. Kalichman SC, Ramachandran B, Catz S. Adherence to combination antiretroviral therapies in HIV patients of low health literacy. J Gen Intern Med 1999; 14:267–273.
7. Barnett PG, Sorensen JL, Wong W, Haug NA, Hall SM. Effect of incentives for medication adherence on healthcare use and costs in methadone patients with HIV. Drug Alcohol Depend 2009; 100:115–121.
8. Operario D, Kuo C, Sosa-Rubí SG, Gálarraga O. Conditional economic incentives for reducing HIV risk behaviors: integration of psychology and behavioral economics. AIDS Behav 2013; 17:2283–2292.
9. Haug NA, Sorensen JL. Contingency management interventions for HIV-related behaviors. Curr HIV/AIDS Rep 2006; 3:154–159.
10. Rigsby MO, Rosen MI, Beauvais JE, Cramer JA, Rainey PM, O’Malley SS, et al. Cue-dose training with monetary reinforcement: pilot study of an antiretroviral adherence intervention. J Gen Intern Med 2000; 15:841–847.
11. Rosen MI, Dieckhaus K, McMahon TJ, Valdes B, Petry NM, Cramer J, Rounsaville B. Improved adherence with contingency management. AIDS Patient Care STDS 2007; 21:30–40.
12. Sorensen JL, Haug NA, Delucchi KL, Gruber V, Kletter E, Batki SL, et al. Voucher reinforcement improves medication adherence in HIV-positive methadone patients: a randomized trial. Drug Alcohol Depend 2007; 88:54–63.
13. Simoni J, Amico K, Pearson C, Malow R. Strategies for promoting adherence to antiretroviral therapy: a review of the literature. Curr Infect Dis Rep 2008; 10:515–521.
14. Metsch LR, Feaster DJ, Gooden L, Matheson T, Stitzer M, Das M, et al. Effect of patient navigation with or without financial incentives on viral suppression among hospitalized patients with HIV infection and substance use: a randomized clinical trial. JAMA 2016; 316:156–170.
15. Gardner LI, Metsch LR, Anderson-Mahoney P, Loughlin AM, del Rio C, Strathdee S, et al. Efficacy of a brief case management intervention to link recently diagnosed HIV-infected persons to care. AIDS 2005; 19:423–431.
16. Loewenstein G, Brennan T, Volpp KG. Asymmetric paternalism to improve health behaviors. JAMA 2007; 298:2415–2417.
17. Loewenstein G, Asch DA, Friedman JY, Melichar LA, Volpp KG. Can behavioural economics make us healthier?. BMJ 2012; 344:e3482.
18. Loewenstein G, Asch DA, Volpp KG. Behavioral economics holds potential to deliver better results for patients, insurers, and employers. Health Aff 2013; 7:1244–1250.
19. Volpp KG, Pauly MV, Loewenstein G, Bangsberg D. P4P4P: an agenda for research on pay-for-performance for patients. Health Aff 2009; 28:206–214.
20. Angeletos G, Laibson D, Repetto A. The hyperbolic consumption model: calibration, simulation, and empirical evaluation. J Econ Perspect 2001; 15:47–68.
21. Ashraf N, Karlan D, Yin W. Tying Odysseus to the mast: evidence from a commitment savings product in the Philippines. Q J Econ 2006; 121:635–672.
22. Bryan G, Karlan D, Nelson S. Commitment devices. Annu Rev Econ 2010; 2:671–698.
23. Giné X, Karlan D, Zinman J. Put your money where your butt is: a commitment contract for smoking cessation. Am Econ J Appl Econ 2010; 2:213–235.
24. Rogers T, Milkman KL, Volpp KG. Commitment devices: using initiatives to change behavior. JAMA 2014; 311:2065–2066.
25. Royer H, Stehr M, Sydnor J. Incentives, commitments, and habit formation in exercise: evidence from a field experiment with workers at a Fortune-500 company. Am Econ J Appl Econ 2015; 7:51–84.
26. Volpp KG, Galvin R. Reward-based incentives for smoking cessation: how a carrot became a stick. JAMA 2014; 311:909–910.
27. Nokes K, Johnson MO, Webel A, Rose CD, Phillips JC, Sullivan K, et al. Focus on increasing treatment self-efficacy to improve human immunodeficiency virus treatment adherence. J Nurs Scholarsh 2012; 44:403–410.

adherence; antiretroviral therapy; behavioral economics; commitment contract; financial incentives; HIV-1 virologic suppression

Supplemental Digital Content

Copyright © 2017 Wolters Kluwer Health, Inc.