Mitigating the Economic and Mental Health Impacts of COVID-19

We evaluate two distinct interventions:

Psychosocial support treatment. The psychosocial support treatment consists of a two-layered psychological first aid (PFA) intervention. We have adapted a well-established PFA manual developed by the WHO to the specific mental health challenges associated with COVID-19 in India, in partnership with mental health experts from a highly reputed mental health organization in Chennai (SCARF India). PFA is designed to be administered by non-specialists in order to help others in distress. It consists of asking questions about a person’s situation, providing basic support in brainstorming solutions to problems, reinforcing coping abilities, linking them to available social support, and identifying severe cases. These severe cases could include facing violence or abuse, thoughts about harming themselves or others, acute distress, or the inability to do day-to-day tasks due to stress, worry or low mood. In these severe cases, we will refer participants to the second layer of treatment, staffed by mental health professionals at SCARF.

Cash transfer. We will offer unconditional cash transfers of approximately 5,000 Rs, which is equivalent to about two weeks of the average salary in this context. The cash transfers will be delivered electronically, providing rapid support that can be administered remotely.

2020-11-17

2021-06-30

Our primary outcomes belong to the following categories: (i) mental well-being, (ii) economic conditions, (iii) demand for information, (iv) memory and attention, and (v) prosociality.

We will measure mental well-being using three standard questionnaires from psychology: the nine-item Patient Health Questionnaire (PHQ-9), the Generalized Anxiety Disorder two-item assessment (GAD-2), and the four-item Perceived Stress Scale (PSS-4). Our measures of economic conditions include survey questions on food security, ability to pay rent, worries about finances, outstanding debt and healthcare access. To measure demand for information, we elicit choices between information that varies in usefulness and affect (positive or negative). To measure memory and attention, we seed information at the midline and measure recall at the endline. This includes information related to COVID, as well as the set of news articles used in the demand for information module. Our measures of prosociality include choices of how to allocate money and effort, with variation in who benefits (eg. self vs. others; ingroup vs. outgroup). We will evaluate heterogeneity of treatment effects using baseline measures of mental health and financial strain.

We will measure secondary outcomes related to satisfaction with government response, intimate partner violence, and beliefs and attitudes about mental health treatment.

Our government satisfaction survey questions will include an assessment of response to the COVID-19 crisis and trust in government when dealing with future crises. The intimate partner violence module asks participants about previous instances of violence in their household. The mental health treatment questions measure beliefs regarding the usefulness of mental health care, including a choice between providing someone else with cash or psychological first aid.

The study consists of a randomized controlled trial with three arms: (i) a control group, (ii) a treatment group which receives a one-time cash transfer of approximately 5,000 Rs, and (iii) a treatment group which receives repeated psychosocial support over the phone from trained lay people. Our target size for the entire study is 4,200 participants, with a control group : PFA group : cash transfer group ratio of 5:5:4. However, we will enroll people in the study and assign them to treatment groups using this ratio on a rolling basis until we run out of budget or reach this target, whichever comes first.

We will conduct four surveys with each participant: (i) a demographic survey, (ii) a baseline survey, (iii) a midline survey, and (iv) an endline survey. We will use measurements from our demographic survey to stratify our randomization. In the two weeks following the baseline survey, participants will receive their treatment (a cash transfer, or an initial PFA conversation), if they are not in the control group. In the one to two weeks after this treatment, we will schedule the midline survey with each participant. Following the midline survey, each participant in the PFA group will receive two more PFA conversations in the two to three weeks afterwards. The endline surveys will be scheduled approximately three to four weeks after the midline survey.

We pre-randomize by computer using Stata’s pseudo-random number generator.

Individual

No

N/A

We will enroll participants in the study and assign them to treatment groups on a rolling basis, following a ratio of 5:5:4 of Control:PFA:Cash. Due to budget uncertainty, we will enroll participants in the study until we run out of funding or reach a total sample of 4,200, whichever occurs first.

1,500 individuals in the control group, 1,500 individuals in the PFA group, and 1,135 individuals in the cash transfer group.

Because our cash treatment arm is more expensive than the PFA treatment arm, our cash treatment group is smaller than the PFA group and the control group, at a ratio of 5:5:4. Our power calculations are calculated to detect effects at the 5% significance level with 80% power, assuming equal variance in treatment and control groups. As described above, we will enroll participants until we exhaust our funding or reach 4,200 participants. The minimum detectable effects corresponding to several possible sample sizes (at our treatment assignment ratio) are as follows: – 2,240 participants (800 control; 800 PFA; 640 cash): 0.149σ MDE – 2,800 participants (1,000 control; 1,000 PFA; 800 cash): 0.133σ MDE – 3,360 participants (1,200 control; 1,200 PFA; 960 cash): 0.121σ MDE – 3,640 participants (1,300 control; 1,300 PFA; 1040 cash): 0.116σ MDE – 3,920 participants (1,400 control; 1,400 PFA; 1120 cash): 0.112σ MDE – 4,200 participants (1,500 control; 1,500 PFA; 120 cash): 0.109σ MDE