Why patients abandon treatment?

How to cite

Seira, Enrique; Gruber, Jonathan; Bronsoler, Ari. 2021. "Why patients abandon treatment?." AEA RCT Registry. AEARCTR-0007533. https://doi.org/10.1257/rct.7533-1.0

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

Enrique Seira* , Jonathan Gruber , Ari Bronsoler

* Primary investigator

Registry ID
AEARCTR-0007533
DOI
https://doi.org/10.1257/rct.7533-1.0
First registered
Last updated
Status
In development
Trial dates
2021-04-30 – 2022-06-30
Countries
Private
JEL codes
I12, I18, H44
Registry record
http://www.socialscienceregistry.org/trials/7533

Abstract

A large share of patients with chronic diseases abandon treatment. Although there is a strong presumption that this hurts their health, we were unable to find a well powered study that estimates causal effects of abandonment on health. Our study has several main objectives.
First, to estimate the causal effect of staying in treatment longer, on the health of diabetic patients. In particular on their HbA1c levels, but also on their knowledge, their expectations of treatment effectiveness and of their cost of complying with treatment. Second, to find whether it is more cost effective to incentivize clients to stay ex-ante for staying, or to bring the abandoners back through a subsidy ex-post after they abandon. The later one may be more well targeted, but the former may create dynamic incentives to exert effort form the beginning of treatment and thus may generate less dropout through dynamic complementarities. Third, to investigate whether giving clients personalized information about diabetes and their health increases their knowledge about their health and causes them to stay longer in treatment. Furthermore, to test if giving them an incentivized exam on the information we provide increases their attention ---as reflected in their knowledge—and further increases staying. Fourth, to investigate correlationally to what extent the probability of abandoning treatment increases after experiencing a negative income shock. To investigate correlationally if they leave after learning how to take care of their health by themselves. Finally, to assess whether or not they substitute to other health providers after they leave.

Intervention

Methodology and population
• We will use a randomized controlled trial in 17 clinics of our partner CdA. These clinics are located in the states of Nuevo Leon, Coahuila, Chihuahua, Jalisco and Guanajuato.
• Our population of interest consists of diabetics that have just bought a plan to get treated for their diabetes at CdA. We will obtain a sample of 3,000 of them and randomly allocate them to 3 baseline arms and 2 cross randomized arms.
• The study is a double blind RCT as neither the clients not the ones operating in the Clinics, nor the surveyors , will be aware of who belongs on which arm. Randomization will be performed at the person level. We don’t expect spillovers as clients don’t know each other and there are between 2 to 6 clients per branch per day in the experiment so almost never are together at the same time.

Baseline arms
1. (C) Control group (1000 clients). This group will get the business-as-usual treatment. That is, they pay at full price and get no extra incentives for staying or coming back. They get the usual information given by the doctors at CdA which is not personalized.
2. (EA) Ex-ante conditional on staying prize. Just after the client enrolls at the going price, we will announce to them that they are entitled prize of $250 pesos if they go to the first quarterly hemoglobin appointment, $250 pesos if they go to their second, $250 if they go to their third and $250 if they go to their fourth plus $1,000 bonus for year accomplishment.
3. (EP) Ex-post bring-back the dropouts with a randomized subsidy. Clients allocated to this arm will receive nothing (i.e. they will be equivalent to the control group) until they drop out. When they drop out however, a random subset of them will be asked to come back at a subsidized price. By dropout we mean missing appointments. There will be two times to bring back clients. A first batch will be those that miss their first appointment at the 4th month of treatment. The second batch is for those that miss their second appointment –at month 7th of treatment—(excluding the former group). The subsidy involves forgiving the months they had no pay, after they themselves pay for one extra month. A $250 prize will be paid for each quarterly checkup they attend as in point 2 above, plus $1,000 bonus for year accomplishment.

Cross-randomized arms

A. No extra information group. This group will not receive any extra information from researchers above and beyond what CdA clinics give them.

B. Information treatment: We will provide personalized information on progress they have made in their self-monitoring of blood glucose and HbA1c, and in their health in general. We will explain what the possible complications are if they stop treatment and provide statistical information about the likelihood of complications. This information will be provided bi-monthly by phone (and if possible also by email).

Experimental design

Baseline arms
1. (C) Control group (1000 clients). This group will get the business-as-usual treatment. That is, they pay at full price and get no extra incentives for staying or coming back. They get the usual information given by the doctors at CdA which is not personalized.
2. (EA) Ex-ante conditional on staying prize. Just after the client enrolls at the going price, we will announce to them that they are entitled prize of $250 pesos if they go to the first quarterly hemoglobin appointment, $250 pesos if they go to their second, $250 if they go to their third and $250 if they go to their fourth plus $1,000 bonus for year accomplishment.
3. (EP) Ex-post bring-back the dropouts with a randomized subsidy. Clients allocated to this arm will receive nothing (i.e. they will be equivalent to the control group) until they drop out. When they drop out however, a random subset of them will be asked to come back at a subsidized price. By dropout we mean missing appointments. There will be two times to bring back clients. A first batch will be those that miss their first appointment at the 4th month of treatment. The second batch is for those that miss their second appointment –at month 7th of treatment—(excluding the former group). The subsidy involves forgiving the months they had no pay, after they themselves pay for one extra month. A $250 prize will be paid for each quarterly checkup they attend as in point 2 above, plus $1,000 bonus for year accomplishment.

Cross-randomized arms

A. No extra information group. This group will not receive any extra information from researchers above and beyond what CdA clinics give them.

B. Information treatment: We will provide personalized information on progress they have made in their self-monitoring of blood glucose and HbA1c, and in their health in general. We will explain what the possible complications are if they stop treatment and provide statistical information about the likelihood of complications. This information will be provided bi-monthly by phone (and if possible also by email).

For more detail see pre-analysis plan

Design details

Randomization method
Randomized in an office computer
Randomization unit
Randomization at the level of the individual
Clusters
3000 patients
Observations
3000 patients
Arms
See pre analysis plan.
Primary outcomes
Health: HbA1c, BMI Subjective health: How do you feel today physically. Last month how many days could you not work for health reasons? Health complications; isits to hospital due to diabetes in the last 12 months. Nights been hospitalized due to diabetes in the last 12 months. In the past 12 months, among others Costs of complying with treatment: Monthly expenditure out of pocket on diabetes. How hard subjectively was to follow the diet and recommendations. Factual knowledge: Knowledge of their own health status. knowledge of risk of abandoning treatment, among others. Expectations: HbA1c in 6 months if stays in CdA and if not; Expected months in treatment. Willingness-to-pay.
Secondary outcomes
See pre-analysis plan

Cited by

No citation data available for this registration yet.

Keywords