Data collection
A cross-sectional study was conducted in a random sample of 60 (45%) FP clinics out of 132 FP clinics operating in Mombasa County in 2016. Two of the selected clinics were found to have no ongoing FP clinic activities, and were excluded from further analyses. The sample size was chosen to sample approximately half of FP clinics in Mombasa County. Study staff visited FP clinics and scheduled meetings with clinic managers. These meetings occurred on the same day as the visit if clinic managers were available. If clinic managers were unavailable, then the meeting was scheduled for a subsequent day. Brief structured interviews with FP clinics managers collected information about each clinic including size, location (urban, peri-urban, rural), size of population served, public versus private facility, presence of academic or non-governmental organization (NGO) support, what other clinics are co-located with the FP clinic, if the FP clinic holds regular management meetings, and clinic staff characteristics (number and type of staff, experience and education level of clinic manager, number of providers trained in HIV testing and counseling, and whether the clinic manager is aware of the current Kenyan National HIV guidelines). These interview questions were modelled after questions used successfully in the original SAIA trial (19). Following the interview, study staff performed an evaluation of HIV testing commodities (HIV testing kits, alcohol, cotton, lancets), to determine if they were in stock and not expired. Finally, staff abstracted data from the FP clinic register for the complete 3-month period preceding the date of data abstraction. The most recent complete and consecutive months were used for this analysis. Register pages were also examined for any duplicate entries. Digital photographs were then taken of the corresponding register pages with names masked. Study staff used these images to abstract data including number of new FP clients, number of new clients counselled for HIV testing, number of new clients tested for HIV, number newly positive for HIV, and referral processes (referred to another health facility, and reasons for referral). Over an 8-month period, multiple stakeholder meetings and results dissemination forums occurred with different members of the Mombasa County Health Management Team, sub-County Ministers of Health, reproductive health coordinators, family planning coordinators, the Chief Officer of Public Health, Chief Nursing Officer, and Director of Reproductive Health, to discuss this study and define HTC best practices. Data quality and interpretation was discussed at these meetings, including the handling of missing data in the registers. There was a broad consensus agreement that missing information on HIV counseling and testing in the registers indicated that these services were not performed. Data from the register photographs were initially recorded on paper, then entered manually into a REDCap database (20). All data collection forms were reviewed for any missing responses, and addended as needed. In addition, FP register data abstraction forms were reviewed by a second study team member who independently viewed the digital images of the FP record, verifying that the same responses were generated by both study team members. Any discrepancies identified between data abstraction by different study team members were resolved with consensus agreement of the staff. Following data entry, study staff performed a question by question assessment (line-listing) comparing hard copy data collection forms to the digital database to identify and correct any key-in errors.
Data Analysis
Because HTC was performed infrequently in the vast majority of FP clinics, any HIV counseling or testing was compared to no HIV counseling or testing. Generalized linear models with a log link and binomial family were used to calculate prevalence ratios (PR) and identify correlates of HTC. Factors associated with HTC with a p-value <0.10 in univariate analyses were included in a multivariate analysis (21). Additional analyses examined correlates using a multiplicative performance score (proportion counseled x proportion tested) and an additive performance score (proportion counseled + proportion tested). All analyses were conducted using Stata version 15 (College Station, Texas) (22).
This study was conducted in close collaboration with the Mombasa County Department of Health. Prior to data collection, several meetings were held with sub-County Medical Officers of Health to incorporate their input into development of the study procedures. This study was approved by the Human Subjects Research Committee of the University of Washington (STUDY00001851), the Kenyatta National Hospital-University of Nairobi Ethics Research Committee, and Mombasa County.