Information, Mental Models, and Provider Choice: Protocol for a Randomized Controlled Trial Among Private Clinic Users in Urban Uganda

How to cite

Grabowska, Marta; Lagarde, Mylene. 2025. "Information, Mental Models, and Provider Choice: Protocol for a Randomized Controlled Trial Among Private Clinic Users in Urban Uganda." AEA RCT Registry. AEARCTR-0017156. https://doi.org/10.1257/rct.17156-3.0

Download: BibTeX · RIS — for reference managers without a browser translator. Zotero and most others pick up this page's citation metadata directly.

Marta Grabowska* , Mylene Lagarde

* Primary investigator

Registry ID
AEARCTR-0017156
DOI
https://doi.org/10.1257/rct.17156-3.0
First registered
Last updated
Status
On going
Trial dates
2025-11-27 – 2026-04-30
Countries
Uganda (Gulu)
Registry record
http://www.socialscienceregistry.org/trials/17156

Abstract

Healthcare quality in low- and middle-income countries (LMICs) remains low and highly variable across providers, with little oversight from regulators. As healthcare is a credence good market, there are information asymmetries between patients and providers: many (good) bad quality providers are (un)able to signal high quality. In Uganda, where 40% of facilities are private for-profit and out-of-pocket spending reaches 35% of health expenditure, private providers have a large incentive to exploit this asymmetry and to respond to patient beliefs about signals of quality. The result is market failure in healthcare, where patients cannot identify quality providers and overpay for inadequate care.

There are two possible causes of the persistence of this information gap between patients and providers. First, patients have limited access to information, absent of objective signals of quality such as provider ratings. Unable to directly evaluate medical expertise or qualifications of a provider, patients infer from a subset of possible characteristics which they observe during their interactions. Second, patients' mental models about healthcare (or how observable signals are weighted in assessments of provider quality) are often incorrect. For example, patients believe overtreatment and extensive testing signal quality, even in self-limiting cases.

In this study, we seek to cross-evaluate two interventions that target these sources of information asymmetry: 1) providing objective quality signals to individuals and 2) addressing errors in conceptual frameworks of healthcare. Earlier this year, we conducted an independent audit of government, private, and non-profit health centres using standard patients, generating objective evidence on healthcare quality, including provider effort, medical appropriateness, patient experience and cost. In Gulu, the audit revealed private providers performing substantially worse on case management, with only 40% SPs managed correctly (allowing for overtreatment). By contrast, non-profit and government providers managed patients correctly 60% of the time – a 50% improvement on private clinics. This evidence will form the basis of both information treatments. Intervention 1 will be delivered to participants through a short video – a voiced over animated presentation comparing provider quality – and report cards with comparative ratings of types of providers (government, public, non-profit). Intervention 2 will include all the information components, but the video will include an additional section addressing misconceptions about healthcare providers, focusing on private clinics’ tendency to overtreatment, lower medical accuracy, and higher costs.

To evaluate the relative effectiveness of information with / without addressing the misconceptions about provider quality signals, we will conduct a randomised controlled trial in Gulu, the largest city in the Northern region of Uganda. Between November 2025 and January 2026, we will recruit 1,500 current users of private providers in central locations; administer a baseline survey and elicit beliefs about the relative quality of government, private, and non-profit providers; then randomise the participants to either receive intervention 1, intervention 2, or the control condition (no information given). Our primary outcomes of interest are: 1) beliefs about providers’ relative quality, captured in a post-intervention module at the time of the survey; and 2) choice of provider when the participant next seeks care. The choice will be captured for 3 months following the survey using WhatsApp – pictures of patient report cards will be used to verify the provider visited. Jointly, these findings will inform if simple information or more extensive public education campaigns can redirect patients away from low quality providers.

Intervention

Using data from an independent audit of government, private, and non-profit health centres using standard patients, we designed three information provision interventions:

Intervention 1) providing objective quality signals to individuals. Delivered to participants through a short video – a voiced over animated presentation comparing provider quality – and report cards with comparative ratings of types of providers (government, public, non-profit).

Intervention 2) objective quality signals + addressing errors in conceptual frameworks of healthcare. Intervention 2 will include all the information components, but the video will include an additional section addressing misconceptions about healthcare providers, focusing on private clinics’ tendency to overtreatment, lower medical accuracy, and higher costs.

Intervention 3) A final intervention is a simple screen where we reveal which providers performed best within each category according to SP study results. We do not compare how they did *across* provider types.

Experimental design

Sample and Recruitment:

We will conduct a randomized controlled trial with 1,600 participants in Gulu, Uganda. Participating adults will be eligible if 1) they had sought care within the past 12 months; 2) they had sought care from private healthcare providers within the past 12 months; 3) they are able to give a mobile phone contact number (essential for a part of the primary outcome, and enabling future follow-up data collection if desired).

We will recruit individuals who have recently used private health facilities in Gulu. In a recent pilot, conducted in a busy market location and a well-off neighbourhood of Gulu, 92% of respondents sought care in past 12 months and 2/3 of those had used private providers. We will recruit in central locations in the city. Participants will be screened for eligibility and invited to participate in a 35 minute survey. To avoid interviewing the same individuals, enumerators will work in teams under the guidance of a supervisor and recruitment sites will be split between the teams.

Randomization:
After completing the baseline portion of the survey and the belief elicitation module, participants will be randomly assigned to one of four groups:
- Control group (T0): Receives no additional information; views a placebo video of unrelated results from the SP study. n=300
- Best performer group (T1): Received the placebo video; in addition, learns only information on which providers did best in the SP study, by category. This group tests if preferred provider can be shifted *within* provider, if not between types. n=300
- Objective Quality Information (T2): Receives objective information on provider quality based on standardized patient audits conducted earlier this year. n=500
- Conceptual Framework Correction (T3): Receives treatment T2 and additional information addressing common misconceptions about healthcare quality signals. n=500

Design details

Randomization method
SurveyCTO calculations, performed per survey
Randomization unit
Individual
Clusters
N/A
Observations
1,600
Arms
300 T0, 300 T1, 500 T2, 500 T3.
Primary outcomes
1) Beliefs about providers' medical quality - by provider 2) Did not choose private provider when seeking care in 2 months subsequent to intervention
Secondary outcomes
Marginal preferences for providers characterised by: 1) overtreatment, 2) low cost, 3) higher diagnostic effort, 4) better patient manner, 5) nicer facility conditions, 6) less time at facility
Sponsors
Name: 100x Impact Accelerator Location: London Url: https://www.100ximpact.org/

Cited by

No citation data available for this registration yet.

Keywords