NEW UPDATE: Completed trials may now upload and register supplementary documents (e.g. null results reports, populated pre-analysis plans, or post-trial results reports) in the Post Trial section under Reports, Papers, & Other Materials.
Following Up For Better Health: Improving Non-Communicable Disease Compliance in Urban India
Initial registration date
November 09, 2015
April 10, 2018 3:50 PM EDT
University of Southern California
Other Primary Investigator(s)
University of Chicago
Additional Trial Information
The majority of individuals with non-communicable diseases (NCDs) live in the developing world, where prevalence rates are growing rapidly as nations become richer and more urban. In cities, where most individuals have access to a variety of medical providers, improving adherence to treatment is key to reducing morbidity from NCDs. We will partner with Swasth India, which operates private health clinics in the slums of Mumbai, to evaluate experimentally several interventions designed to improve NCD adherence. These interventions include (a) SMS reminders to adhere to treatment, (b) patient education on NCDs and the consequences of non-adherence, (c) a discount for NCD follow-up visits and medicine, and (d) a “regular patient” lottery, where the number of follow-up visits increases the value of the lottery prize.
Bennett, Daniel and Simone Schaner. 2018. "Following Up For Better Health: Improving Non-Communicable Disease Compliance in Urban India." AEA RCT Registry. April 10.
All of the interventions in this study aim to improve compliance with treatment regimens for diabetes and hypertension. The first set of interventions will provide patients with information about the implications and proper treatment of these diseases. We will frame the information in either positive or negative terms by emphasizing either the gains from compliance or the losses from non-compliance. The second intervention is an SMS-based reminder service. We will elicit the willingness to pay for reminders and then compare the willingness to pay with the compliance benefit of reminders. Finally, we will offer price discounts for medical care, including both certain and lottery-based discounts.
Intervention Start Date
Intervention End Date
Primary Outcomes (end points)
Using a baseline survey, an intermediate follow-up survey, and a final follow-up survey, we will measure/obtain:
1. Biometric outcomes, including blood sugar, blood pressure, weight, and height. 2. Self-reported doctors’ visits, diet, and use of medication
3. Self-reported morbidities related to diabetes and hypertension, such as pain, dizziness, and lethargy, as well as secondary outcomes such as labor supply and self-reported well-being. 4. The consumption of foods that may be problematic for at-risk people. 4. Knowledge and beliefs about the NCDs
5. Willingness-to-pay for reminders for patients
6. The utilization of clinic consultations.
Primary Outcomes (explanation)
Secondary Outcomes (end points)
Secondary Outcomes (explanation)
All study participants will be randomly assigned to either the control group or a treatment group, which will receive a combination of the interventions described below.
(1) Gain and Loss-Framed Health Information. We will train a team of community health workers (CHWs) to counsel patients for diabetes and hypertension. The information session, will explain (a) what the condition is, (b) the long-term consequences of failure to manage the condition, including severe morbidity and mortality and (c) the impact of disease management on the prognosis of the disease. Individuals will be randomly assigned either to information framed in gain/positive terms (e.g. “if you comply with recommended treatment you will support your long-run health and avoid negative health complications”) or information framed in loss/negative terms (e.g. “if you do not comply with recommended treatment you will damage your long-run health and may experience negative health complications”) or will receive no information. (2) First and Second Stage Reminders: Individuals randomly selected for this treatment will all receive SMS reminders on a weekly or bi-weekly basis and will remind patients to take their medication and to follow-up with the doctor at the recommended interval. After three months, conditional on their willingness to pay for reminders for the rest of the year (measured using a BDM procedure), some of the individuals in this group will continue getting reminders for the rest of the year. These form the group that receives second stage reminders. The individuals who received reminders only for the first three months, will form the group that receives first stage reminders. There will also be a group that receives no reminders
(3) Pricing: Individuals will be randomly assigned to receive a 25% discount on visits to a Swasth clinic, or entry into a lottery that produces the same expected value of discount as does the sure discount, or will not receive any price incentives
Experimental Design Details
Using a computer randomization tool, randomization will be done on unique IDs assigned to each patient.
The randomization will be conducted at the individual level, and will be stratified by disease type and whether the participant is a pre-existing Swasth patient.
Was the treatment clustered?
Sample size: planned number of clusters
Sample size: planned number of observations
Sample size (or number of clusters) by treatment arms
There will be 1166 people enrolled in each of the pricing arms (no discount, fixed discount, lottery discount). Same goes for the three information arms (no info, positively framed info, negatively framed info). 80% of the full sample (2798) will be enrolled in stage 1 reminders and 1399 people will be enrolled in stage 2 reminders.
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
We have powered our experiment to detect a 0.22 standardized effect size between price × information arms. Since many of the effects we are interested in estimating require receipt of stage 1 reminders, we will allocate only 20 percent of the sample to "no reminders". Since the randomization will be conducted at the individual level, we have used the minimum detectable effect formula without clustering in Glennerster and Takavarasha (2013). We have calculated the sample size needed to achieve target power for a single survey round (i.e. assuming one observation per individual) since we wish to estimate short- and long-run impacts separately. We have also scaled up our sample sizes to account for an assumed 10 percent attrition rate. This calculation suggests that we need a total sample size of 3,243, which we have rounded up to 3,500 for budgeting purposes and to account for the fact that some individuals in the first-stage reminder group will need to be used to pilot the BDM mechanism.
This design implies that we have sufficient power to detect a 0.12 standardized effect size between price and information arms, a 0.15 standardized effect size between "no reminders" and "first-stage reminders", and a 0.14 standardized effect size in terms of the impact of information treatments on willingness-to-pay for reminders. We are also interested in estimating reminder treatment effects conditional on a zero-to-negative willingness-to-pay. We have reasonable power to detect treatment effects when just 20-30 percent of individuals do not positively value reminders. To put these numbers in context, assuming a 50 percent adherence rate in the control group, a 0.2 standardized effect size amounts to a minimum detectable effect of 10 percentage points in terms of adherence
INSTITUTIONAL REVIEW BOARDS (IRBs)
IRB Approval Date
IRB Approval Number
IRB Approval Date
Post Trial Information
Is the intervention completed?
Is data collection complete?