Increasing Cancer Screening Rates and Vaccine Uptake using Digital Messages based on Insights from Behavioral Economics

We have five treatments which may be considered as our interventions. Individuals insured by the HMO (Maccabi Healthcare Services) will be randomly assigned to one of 5 treatments and control. The treatments differ in the content included in the message. Each message emphasizes a different theme that is expected to increase compliance rates, based on previous behavioral and psychological research. The different treatments are: (1) Positive framing: stating the fact that early detection of a disease increases the treatment options and the chances of full recovery. (2) Negative framing: stating the fact that late detection of a disease reduces the treatment options and the chances of full recovery. (3) Empowerment: emphasizing the personal responsibility of the individual for his/her health. (4) Physician Recommendation: mentioning that doctors within the HMO recommend performing the suggested screenings. (5) Implementation Intention: presenting the screening and potential outcomes as an "if-then plan".

In all treatments and control, each reach-out to an insured individual consists of three messages - all framed along the lines of the treatment she/he has been randomly assigned to. The second and third message are sent out 2 and 4 weeks after the first, respectively, only in the case that the individual did not yet schedule an appointment for one of her/his recommended screenings/vaccination. There are minor changes between the 3 messages as they are designed to be reminders that are not too repetitive. Messages are sent out by email but in cases in which this is not possible they will be sent out as text messages. Those who did not open the first message received by mail, receive it again as a reminder two days later.

The customer journey started on July 29th, 2020. The research has been approved by the research committee of Maccabi Healthcare Services and is scheduled to go through Helsinki committee and obtain approval for sharing the data for research purposes. The intervention in itself does not require informed consent, as it is part of the routine digital activities of the HMO. The data is currently in collection and is expected to be ready roughly by November 2020. The HMO plans to continue these digital customer journey campaigns continuously. Our plan in this research is to focus on the first campaign but, with agreement of the HMO, we may continue to study the following campaigns as well.

Data analysis: We plan to examine compliance rates (actual performance of screening/vaccinations) and intentions (scheduling an appointment) for each screening/vaccination and overall for each treatment and compare them to the control group. We will examine these variables in different time frames measured from the date of the last message. Our control variables will include demographics provided by Maccabi Healthcare Services. We will make use of different descriptive statistics as well as logit regression models. Other appropriate tools to analyze the data may be used as the data becomes available and the analysis begins.

2020-07-29

2021-12-06

A dummy variable recording whether the individual performed the screening/got the vaccination within 2-3 months since receiving the message.

A dummy variable recording whether the individual scheduled an appointment to her/his recommended screening.

The design is straightforward. 5 treatments plus a control. Treatments differ in the framing and content of the message and we will examine which of the above approaches has the highest success in increasing compliance rates.

Randomization by a computer.

Individual.

No

Roughly 150,000 individuals

Roughly 150,000

25,000 individuals in each treatment/control

Assuming n=25000 in each group, significance level 0.05, power of 95%, in a two sided test, the minimum detectable effect size is h=0.032.