Table 2

Sampling strategy and sample size calculations for each Integra research component
Research Component Sampling strategy Key indicators used in sample size calculation Adjustments to sample calculation Model 1. Kenya FP Model 2. Kenya PNC Model 2. Swaziland PNC
Sampling of facilities Theoretical sampling to point of saturation within province 4 criteria: high client load; minimum of 2 FP providers; range of SRH services available; no current provision of SRH-HIV integration n/a 12 facilities: 6 Intervention and 6 Comparison 12 facilities: 6 Intervention and 6 Comparison 8 facilities: 4 Intervention and 4 Comparison*
HFAs: Facility inventory 1 per facility 1 per facility 1 per facility
HFAs: Client exit interviews & linked observations of consultations Consecutive sample and saturation Quality of Care analysis 6 CPIs No 18 per facility: 6 new clients, 6 repeat clients and 6 clients switching FP method 6 within 48 hours postpartum 6 at one week and 6 at 6 weeks per facility 6 within 48 hours postpartum 6 at one week and 6 at 6 weeks
HFAs: Structured interviews with providers Convenience All providers working in MCH-FP units No 3-6 providers per facility 3-6 providers per facility 3-6 providers per facility
HFAs: Client flow assessment Census over 5 days based on existing client load No 50 – 200 clients per day 50 – 200 clients per day 50 – 200 clients per day
Cohort Consecutive sample of female clients (recruited if within 12 weeks post-partum in PNC facilities) Kenya: 80% power to detect a 5% increase in condom use, among all women in PNC and those using other contraceptive methods in FP facilities. 1952 clients: 976 from Intervention and 976 from Comparison facilities
Swaziland: To detect a 7% increase in condom use among PNC clients. 30% loss to follow up; over-sampling of HIV-positive clients until min of 400 in Intervention & Comparison facilities 1978 clients: 989 from Intervention and 989 from Comparison facilities
Household survey 3-stage cluster survey design, for random selection of EAs, households and individuals 80% power to detect 10% increase in % of women who have ever used study facility. Design effect of 2.0 1632 (816 male; 816 female) within 10km catchment area of FP facilities n/a 816 (408 male; 408 female) within 10km catchment area of PNC facilities
Qualitative – in-depth interviews
Providers Convenience sampling 1-3 providers per facility 1-3 providers per facility 1-3 providers per facility
Cohort clients Purposive sampling to reflect issues emerging from quantitative data 25 HIV-positive clients; and 25 other clients 25 HIV-positive clients; and 25 other clients 25 HIV-positive clients; and 25 other clients
Community members Purposive sampling to reflect issues emerging from quantitative data 20 men n/a 20 men

* 2 facilities subsequently added due to lower than expected client load at comparison sites.

Warren et al.

Warren et al. BMC Public Health 2012 12:973   doi:10.1186/1471-2458-12-973

Open Data