Peer Effects in Workplace Health

The interventions are happening in the context of a yearly flu vaccine drive. There will be two set of interventions.
In interventions (A) we split all employees who did not opt out of communications into three groups. The first group is receives the standard message, comparable to how it was used in prior years at the company. The second group receives a message that tells them that taking vaccinations will inspire others to do the same, a passive peer effect. The third group receives a message that encourages them to get other people to get vaccinated, an active peer effect.
Concurrently we randomly select 200 people at the firm who are told that if they get vaccinated before a deadline $100 will be donated to a charity.

Study A. Messages will be sent weekly for four weeks
Control group: No message
First group message: "Flu vaccines reduce the risk of flu illness and hospitalization. Staying active improves energy, mood, and sleep."
Second group message "Flu vaccines reduce the risk of flu illness and hospitalization. Taking healthy actions may inspire your coworkers too"
Third group message "Flu vaccines reduce the risk of flu illness and hospitalization. You can play a role in encouraging friends, family and coworkers by talking to them about getting vaccinated too. Share the well-being feeling. Encourage friends, family, and coworkers to be more active too."
Study B
200 employees will be randomly selected. They receive this message "Help out a worthy cause when you upload proof of a flu shot. You recently received a message fro {the company wellness program name} encouraging you to get a flu shot. Go365 is partnering with Associate Well-being to offer you an exclusive opportunity to help out a local cause. Get your flu shot prior to October 31, 2021, and, on your behalf, $100 will be donated to Feeding America, a networkho of food banks, pantries and meal programs that serve over 40 million people - of 1 in 7 Americans - facing hunger. Every $1 donated can provide up to 10 meals per family in need - that's up to 1,000 meals a family could receive thanks to your healthy action! [Upload today button] Notes For 40 years, Feeding America has responded to the hunger crisis in America by providing food to people in need through a nationwide network of food banks. This limited, special partnership with {the company wellness program name} will help Feeding America support people in your community live more secure and stable lives. As a part of this research and to quality for the Feeding America donation, you must get a flu shot by October 31, 2021. If your shot is billed through your {company name} medical plan, verified proof will automatically be uploaded to {company wellness program name]. It it's not billed through your {company name} medical plan you must submit {Form ID} Form to {company wellness program name} together with proof within 90 days of activity completion.

2021-09-15

2021-10-15

Primary outcome is individual level flu vaccination status (binary).

This variable is drawn from the company wellness program that is linked to employees health records.

Company wellness program points, LDL, Blood pressure, glucose level, ASCVD risk, medical expenses without Rx, Rx medical expenses, attrition. Sub-population analysis by race, ethinicity, job tenure, and social preferences

company wellness program points, LDL, blood pressure, glucose level, ASCVD risk, medical expenses with and without Rx are derived from the companies wellness program that is linked with health records. Attrition, race, ethnicity and job tenure is based on HR data. Social preferences is based on measure from online experiments conducted during the second half of 2020.

Part A. Employees at the company are randomized into four groups and receive different messages encouraging them to get a flu vaccine.
This tests whether being informed of peer effects (group 2) or being encouraged to influence others (group 3) affects the flu vaccination rate compared to the control group.
Part B. 200 randomly selected employees receive a message telling them that when vaccinated an amount will be donated to a charity. The outcome of interest is the flu vaccination rate of the peers of the recepient of the message compared to those employees where the peers did not receive a message.

Part A: Ranomization with Stata's runiform routine that allocates a number of 0, 1, or 2 to each employee at the company thus allocating them to the control message, the second group or the third group.
Part B: independently from the randomization in part A 200 employees are drawn using SAS's survey select routine stratified by age, company wellness program status level, and hierarchy tier in the company.

Randomization is at the individual level.

No

28793

28793

Part A: Control (7,198), First group (7,198), Second Group (7,198), Third group (7,199)
Part B: Treatment (200)

Measurable vaccination rate is 0.56 in the control arm. This is based on flu shot vaccination rates from the past year. With a power of 0.8 the minimum detectable effect (delta) will be an increase of 2.39 percentage points (0.5839 - 0.56) With a power of 0.9 the minimum detectable effect will be an increase of 2.71 percentage points (0.5871 - 0.56)