Beyond Bias: Reducing Provider Bias Towards Adolescents to Increase Contraception Take-Up in Tanzania, Burkina Faso, and Pakistan

The intervention has three components: 1) a summit that includes impactful stories told to and by family planning providers that highlight the consequences of provider bias, 2) a forum for continued communication between providers, and 3) a rewards program where clinics in which providers exhibit less biased client interactions or who have improved the most towards this end will be rewarded with social recognition and a ceremony.

2019-09-01

2020-10-01

We will have different primary outcomes for each domain of interest.

Technical quality outcome(s): Essential questions index and counseling on both short and long acting methods
Patient centered care outcome(s): Patient centered care index
Client outcome(s): Client received modern method and client received long acting method

Technical quality: Essential questions index and counseling on both short and long acting methods

For the essential questions index we will aggregate indicators for whether the provider asked about birth spacing preferences, contraceptive method preferences, and if the client had any questions for the provider. This information will be recorded in the exit surveys, provider observations, and mystery client visits. We will use the method outlined by Anderson (2008) to construct the index separately in each data source.

counseling on short and long acting methods will be based on the set of short acting methods, which includes condoms and pills and the set of long acting methods, which includes IUDs, injectables, and implants. If the visit included counseling on at least one method from each set, the visit will be coded as a one (and zero otherwise). This information will be collected in exits surveys, mystery client visits, and provider observations.

Patient centered care: Patient centered care index
The indexed will based on a scale created and validated by Sudhinaraset et al. (2018). We will use questions 37- 44 in the exit survey instrument and identical questions in the mystery client form to construct the index. Each of these questions has a response option of “No, never”, “Yes, a few times”, “Yes, most of the time”, and “Yes, all of the time”. In accordance with the index, we will set each of these 8 questions to 1 if the respondent/mystery client answered “Yes, all of the time” and to zero for all other response. We will then use the Anderson (2008) method to create the index.

Client outcomes: Client received modern method and client received long acting method
These outcomes will be based on two indicators for whether the client received a long acting method or a modern method.

Among the eligible facilities in each country, half of the facilities will be randomly assigned to receive implementation of the “Beyond Bias” intervention, while the other facilities will serve as a control group—standard adolescent and youth sexual and reproductive health or family planning program interventions—to establish a counterfactual. Provider-level assignment is infeasible due to the nature of the interventions: all providers within a given facility will have the same assignment as the facility.

Randomization done in office using Stata

Clinic

Yes

235 clinics (Tanzania: 75; Burkina Faso: 78; Pakistan: 80)

91,000 exit surveys (Tanzania: 39,000; Burkina Faso: 39,000; Pakistan: 13,000) 692 provider surveys (Tanzania: 300; Burkina Faso: 312; Pakistan: 80) 366 mystery client visits (Tanzania: 150; Burkina Faso: 156; Pakistan: 160) 612 direct observation (Tanzania: 300; Burkina Faso: 312) 210 interviews with clients (70 per country) 90 interviews with providers (30 per country) 65 interviews with clinic administrators (Tanzania: 25; Burkina Faso: 20; Pakistan: 20)

117 treatment clinics, 116 control clinics (Tanzania: 38 intervention, 37 control; Burkina Faso: 39 intervention, 39 control; Pakistan: 40 intervention, 40 control)