Using Group Interpersonal Psychotherapy to Improve the Well-Being of Adolescent Girls

There are two main interventions:

1. 14 weeks of group-based interpersonal therapy (IPT-G): StrongMinds is focused on treating depression in Uganda by training community members (in this case ELA club mentors) to act as mentors in IPT-G techniques. This intervention will be offered to 13-19 year old young women who score a 10 or higher on the PHQ-8. These adolescents who take up the offer will then be enrolled in the 14 weeks of therapy. Group therapy sessions build bonds between young women and encourage them to actively engage in the healing process and to support each other in the exploration of their depression triggers. With new healthier patterns and skills, women can learn to manage their current depression and ensure future depressive episodes can be quickly identified and resolved before the onset of any long-term consequences.

2. Unconditional Cash Transfer: A one time lump sum of 200,000 UGX (~$54) be provided to all study participants in a random sub-set of intervention clusters near or at the conclusion of the 14-week therapy. This treatment variation will allow for determination of whether complimentary income support enhances the effects of IPT-G on psychological wellbeing and other outcomes of interest.

2019-09-01

2019-12-31

psychological wellbeing measured by (i) binary indicators on the GHQ-12 (>=3), (ii) a binary indicator of minimal depression (PHQ-8 <=10) (iii) score on Rosenberg self esteem (0-30) and score on Child & Youth Resilience Measure-Revised (0-34)

(i) indicator for being enrolled in school; (ii) the score on a set of competencies (skills) using a module developed by the study team; (iii) risky sex (condom use during last sexual encounter); and (iv) the incidences of marriage and childbearing since baseline.

Control Group/Standard ELA Program: 500 female adolescents (aged 13-19) in 50 clusters. The target population will be listed from the ELA BRAC clubs (ELA members) and its immediate vicinity (non-ELA members), screened using the PHQ-8, and invited to participate in the study if their PHQ- scores are above the screening threshold. The ELA clubs will operate as usual in these areas with no other intervention present.

T1. ELA + IPT-G: 500 female adolescents (aged 13-19) in 50 clusters. The target population will be listed from the ELA BRAC clubs (ELA members) and its immediate vicinity (non-ELA members), screened using the PHQ-8, and invited to participate in the study if their PHQ-8 scores are above the screening threshold. The ELA clubs will operate as usual in these areas. The study participants will also receive an
invitation to participate in a 14-week IPT-G.

T2. ELA + IPT-G + Income Support: 500 female adolescents (aged 13-19) in 50 clusters. The target population will be listed from the ELA BRAC clubs (ELA members) and its immediate vicinity (non-ELA members), screened using the PHQ-8, and invited to participate in the study if their PHQ-8 scores are above the screening threshold. The ELA clubs will operate as usual in these areas. The study participants will receive an invitation to participate in a 14-week IPT-G. They will also receive a one time unconditional and unframed cash transfer. This treatment variation will allow for determination of whether complimentary income support enhances the effects of IPT-G on psychological wellbeing and other outcomes of interest. The amount of the cash transfer will be pinned to the old age pension which is 25,000UGX per month which is about $7. for 8 months, 8*25,000=200,000UGX ($54).

Cluster randomization in the office using Stata 14.

The randomization unit is the ELA club.

Yes

150 ELA Clubs (and surrounding area)

1500

50 clubs control, 50 clubs IPT-G, 50 clubs IPT-G + cash

The primary indicator for psychological wellbeing used in our power calculations is a binary indicator for minimal depression, which indicates a PHQ‐9 score less than 10. At baseline, this measure will be equal to zero. For the power calculations (which were conducted using the clustersampsi command in Stata 14), we assumed that the control group will have a mean of 0.33 at follow‐up. With 50 clusters per arm and 10 individuals per cluster, and assuming an intra‐cluster correlation of 0.1, the minimum detectable increase in minimal depression is 12 percentage points (pp), indicating that we’d be able to detect statically significant differences at the 95% level of confidence with 80% power if the mean in either treatment group is greater than 0.45. This is well within the expected range of impacts based on earlier studies by StrongMinds, and other studies of IPT‐G. The detectable difference between IPT‐G and IPT‐G plus income support is also 12 pp (in either direction), assuming the same parameters and a mean of 0.45 in the IPT‐G group at follow‐up.