Study design and patient selection
We conducted a retrospective cross-sectional study based on the Botswana Ministry of Health and Wellness (MOHW) National Cervical Cancer Prevention Programme “see-and-treat” pilot programmatic database.[17] The database included women screened with visual inspection after acetic acid (VIA) at Bontleng clinic and those referred to Princess Marina Hospital (PMH) for colposcopy in Gaborone, Botswana, from March 2009 through August 2015. Cervical cancer screening services were initially provided for WLWH as part of comprehensive HIV care and were later extended to HIV-negative women at these sites. Screening services were offered free of charge to all Botswana citizens.
Screening services were linked to a physician-led referral colposcopy and loop electrosurgical excision procedure (LEEP) clinic at PMH. Through various channels, women came to screening services, including provider referral, self-referral following sensitization by written materials, and health education talks. Women were excluded from screening if they had previously had a hysterectomy, pelvic radiation for lower genital tract cancer, or a cervical cancer diagnosis. Screening for women who were menstruating heavily, pregnant, or had a persistent vaginal discharge was re-scheduled for after resolution of the condition.
Cervical cancer screening procedures
All patients underwent a speculum examination of the cervix by a nurse who had participated in the Botswana MOHW VIA training program. Visual assessment was performed after applying 5% acetic acid to the cervix using a cotton swab, and findings were categorized as normal, abnormal with a recommendation for cryotherapy, or abnormal with a recommendation for LEEP. Those with abnormal lesions eligible for cryotherapy were offered same-day treatment and had no histopathology specimen collected. Women with abnormal lesions ineligible for cryotherapy based on appearance, size, or extension into the cervical os, were referred to the colposcopy/LEEP clinic and evaluated by a specialist gynecologist or trained medical officers. The colposcopic appearance of lesions determined diagnostic and treatment decisions. Low-grade appearing lesions were treated with cautery after taking a biopsy; high-grade appearing lesions or those extending into the cervical were treated by LEEP. Histopathology specimens were read by pathologists blinded to VIA findings.
HIV procedures
Women with unknown HIV status at the time of screening or with documented HIV negative status more than six months prior were referred to an HIV testing center and requested to share their results. Throughout the study period, the Botswana National HIV program initiated anti-retroviral treatment (ART) at a CD4 count of ≤ 350 cells/µL.
Data collection
All women undergoing VIA screening completed a questionnaire capturing a limited set of patient-level cervical cancer risk factors, including smoking, age of sexual debut, and parity. HIV status was recorded, and for WLWH, CD4 count at the time of HIV diagnosis and whether on ART at the time of screening was documented. VIA screening outcomes were recorded in the programmatic database. Histology results of women referred for colposcopy/LEEP were extracted from the National Health Laboratory (NHL) electronic medical record when available and entered into the programmatic database.
Outcomes
The primary outcome was the association of VIA positivity and age adjusting for cervical cancer risk factors. The secondary outcomes were the association of histopathologically confirmed high-grade pre-cancer and age adjusting for cervical cancer risk factors; HIV-status association with VIA positivity and high-grade pre-cancer; and the proportions of VIA positivity and high-grade pre-cancer by both age and HIV status.
Data analysis
The analyzed dataset included only women between the ages of 25 and 49. Patient records with missing data for VIA or histopathology that could not be corrected by cross-reference with primary records were excluded from the primary and secondary analysis, respectively. The sample size for the primary outcome was calculated using a 1-sided alpha of 0.05. To attain a 99% power, we assumed VIA positivity to be 30% in women aged 25 to 29 years and 20% in women aged 30 to 49 years based on previous findings.[17] The sample size required to detect a statistically significant difference in VIA-positivity between the two age groups was 2,076 women (374 women aged 25 to 29 years and 1,702 women aged 30 to 49 years).
The cervical cancer risk factors adjusted for included: HIV status, parity, smoking, and age of sexual debut. CD4 count and ART were included in the analysis of WLWH. Descriptive statistics for these variables are presented as median [interquartile range (IQR)] and proportions. Continuous variables were categorized into binary variables and compared using the chi-square test. Categorical variables included age groups of younger and older women (25 to 29 years; 30 to 49 years), age of sexual debut (≤18; >18 years), parity (≤2; >2), CD4 count (≤350 cells/µL; >350 cells/µL), and histopathology results (benign or CIN 1 [≤CIN1] for low-grade pre-cancer; CIN2+ for high-grade pre-cancer). Patterns of missing data were described for the study cohort using percentages.
Logistic regression models computed unadjusted and adjusted odds ratios (ORs) with 95% confidence intervals (CI). Only exposure variables with a p-value of less than 0.1 for unadjusted ORs were included in the adjusted regression models.[18] A p-value of less than 0.05 was considered to be statistically significant. We used Stata 14.0 (StataCorp LLC, College Station, Texas).