Changes in Medical and Cost Uncertainty and the Health Seeking Decision: Vignette Experiment in Pakistan

How to cite

Rogge, Lisa; Imping, Alina; Landmann, Andreas. 2023. "Changes in Medical and Cost Uncertainty and the Health Seeking Decision: Vignette Experiment in Pakistan." AEA RCT Registry. AEARCTR-0012326. https://doi.org/10.1257/rct.12326-1.2

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

Lisa Rogge* , Alina Imping , Andreas Landmann

* Primary investigator

Registry ID
AEARCTR-0012326
DOI
https://doi.org/10.1257/rct.12326-1.2
First registered
Last updated
Status
Completed
Trial dates
2023-11-21 – 2024-01-31
Countries
Pakistan (Khyber Pakhtunkhwa)
Registry record
http://www.socialscienceregistry.org/trials/12326

Abstract

Health care decisions are sometimes a matter of life and death, but also their financial consequences can be disastrous for many households around the world. While one can imagine how the fear of making an expensive mistake may deter households from making otherwise sensible health care choices, there is almost no scientific evidence on the importance and joint role of medical and financial uncertainty. In this project, we first provide theoretical insights based on simulation exercises and develop a novel measurement instrument for medical and cost uncertainty using hypothetical health scenarios (vignettes). Initial analyses from data collected using this instrument with low-income households in Pakistan reveal that on top of many biases, both medical as well as cost uncertainty in health care decisions exist and may deter sensible health investments, in line with the theoretical predictions. As these results indicate potentially large dividends to providing accurate information, we are now conducting an extended vignette survey that includes three within-survey information interventions targeted at changing medical or cost uncertainty separately or jointly. With this RCT, we aim to test in two intervention stages whether: i) the interventions reduce biases and uncertainties; ii) the interventions change the hypothetical health seeking decision; and iii) the joint intervention induces larger changes.

Experimental design

The experiment will be fielded within a household survey with low-income households in four districts in Khyber Pakhtunkhwa, Pakistan. It is part of a game-like survey tool that asks respondents to imagine being exposed to a health shock that is defined by a health vignette script, assess medical and cost uncertainties for six care-seeking options and finally take a hypothetical decision for one of these six options. Each respondent is presented with two health scenarios subsequently: one light scenario that does not require formal medical care (light fever) and one severe scenario (randomly selected from heart disease and appendicitis).
Within each scenario, there are two stages of interventions: first, respondents are randomly allocated to one of four equally sized groups: one control group that does not receive any additional information, and three treatment groups that receive additional information within the vignette: a) targeted at medical uncertainty (nature of health shock, recommended treatment and level of care); b) targeted at cost uncertainty (health insurance coverage); and c) information content of a) and b) combined. This information content is designed in a way that could also be communicated via a medical helpline. After presenting the vignette and information, facility-specific expected probability to get better, expected costs, medical and cost uncertainty are elicited followed by the first hypothetical healthcare decision. In the second intervention stage, all respondents are presented with an overview of their reported facility-specific expectations. Then, within treatment groups a)-c), half of the respondents will see an overview of all their answers like the control group, and the other (randomly selected) half will see the overview without medical uncertainty (a), cost uncertainty at secondary facilities (b) or both (c).
We will conduct a results-blind analysis before receiving the full data including treatment allocation.

Design details

Randomization method
Randomization on tablet within survey form at the time of the interview
Randomization unit
Individual
Clusters
4,000 households with one respondent each
Observations
4,000 households with one respondent each
Arms
Control group: 1,000 respondents Medical information treatment stage 1: 1,000 respondents; stage 2: 500 control overview and 500 overview without medical uncertainty Cost information treatment stage 1: 1,000 respondents; stage 2: 500 control overview and 500 overview without cost uncertainty Medical and cost information treatment stage 1: 1,000 respondents; stage 2: 500 control overview and 500 overview without medical and cost uncertainty
Primary outcomes
Hypothetical healthcare decision (whether and where to seek care, ranking) before and after overview Scenario- and facility-specific expected probability to get better, cost, medical and cost uncertainty

Cited by

No citation data available for this registration yet.

Keywords