Support for lowering the age of cervical cancer screening to 25 years for women living with HIV: retrospective cross-sectional program data from Botswana | BMC Women’s Health
Study design and patient selection
We conducted a retrospective cross-sectional study based on the pilot programmatic database of the National Cervical Cancer Prevention Program of the Ministry of Health and Welfare (MOHW), Botswana. . The assessment included women screened between March 2009 and August 2015 with visual inspection after acetic acid (VIA) at Bontleng Clinic, a primary care clinic in the capital Gaborone providing HIV testing and antiretroviral treatment (ART) for the district. Women with low-grade lesions were offered same-day treatment with cryotherapy for lesions that met the eligibility criteria. Women with lesions ineligible for cryotherapy were referred to Princess Marina Hospital (PMH), a regional tertiary hospital five kilometers away, for colposcopy and excision. Cervical cancer screening services were provided to WLWHs as part of comprehensive HIV care. Services were only extended to HIV-negative women towards the end of the evaluation period. Testing services were offered free of charge to all citizens of Botswana.
Screening services were linked to a physician-run colposcopy and electrosurgical excision clinic at PMH. Through various channels, women came to testing services, including referral to a provider, self-referral following outreach through written materials, and health education interviews. Women were excluded from screening if they had previously had a hysterectomy, pelvic radiation therapy for lower genital tract cancer, or diagnosed with cervical cancer. Screening of women who had heavy periods, were pregnant, or had persistent vaginal discharge was rescheduled after the disease resolved.
Cervical Cancer Screening Procedures
All patients underwent a cervical speculum examination by a nurse who had participated in the MOHW VIA training program in Botswana. Women were first assessed for lesions suspicious for cervical cancer (raised ulcerative lesions with contact bleeding) and, if present, women were referred to PMH for further assessment. Visual assessment was performed after applying 5% acetic acid to the cervix using a cotton swab. Results were categorized as normal, abnormal with cryotherapy recommendation, or abnormal with LEEP recommendation. Abnormal lesions were described as: (1) low grade if well defined and opaque, or (2) high grade if dense or had abnormal vessels. Women with low-grade lesions covering less than three-quarters of the cervix and not extending into endocervical annulment were offered same-day treatment with cryotherapy; these women did not have a histopathological sample taken. Women with abnormal lesions ineligible for cryotherapy due to appearance, size, or extension into the endocervical canal were referred to PMH’s colposcopy/RAD clinic and evaluated by a specialist gynecologist or licensed physicians. The colposcopic appearance of the lesions determined the diagnostic and therapeutic decisions. Lesions appearing to be low grade were treated by cauterization after performing a biopsy; high-grade appearing lesions and those extending into the endocervical canal were treated with LEEP. Board-certified pathologists examined histopathology specimens at the National Health Laboratory (NHL), a government reference laboratory, providing services to the MOHW and private health facilities in Botswana. Pathologists were blinded to VIA results. Histopathology findings were graded according to American Society of Colposcopy and Cervical Pathology (ASCCP) and College of American Pathologists criteria for cervical intraepithelial neoplasia (CIN). Briefly, specimens were recorded as no CIN, CIN graded one to three based on severity, or invasive cervical cancer (ICC).
Women whose HIV status was unknown at the time of testing or whose HIV-negative status was documented more than six months previously were referred to an HIV testing center and asked to share their results. Throughout the study period, the Botswana National HIV Program initiated antiretroviral therapy (ART) at a CD4 count ≤ 350 cells/µL.
All women undergoing VIA screening completed a questionnaire capturing a limited set of patient-level risk factors for cervical cancer, including smoking, age at first intercourse, 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 testing were documented. The results of the VIA screening were recorded in the programmatic database. Histological results of women referred for colposcopy/RAD were extracted from the NHL electronic medical record when available and entered into the programmatic database.
The primary endpoint was the association of VIA positivity and age, taking into account cervical cancer risk factors. Secondary outcomes were association of histopathologically confirmed high-grade abnormality and age, adjusting for cervical cancer risk factors; association of HIV status with VIA positivity and high-grade abnormality; and the proportions of VIA positivity and high-grade abnormalities by age and HIV status.
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-checking with primary records were excluded from primary and secondary analysis, respectively. The sample size for the primary outcome was calculated using a one-tailed alpha of 0.05. To achieve 99% power, we assumed VIA positivity to be 30% in women aged 25-29 and 20% in women aged 30-49 based on previous results. . The sample size needed to detect a statistically significant difference in VIA positivity between the two age groups was 2076 women (374 women aged 25-29 and 1702 women aged 30-49).
Adjusted cervical cancer risk factors included: HIV status, parity, smoking, and age at first sexual intercourse. CD4 count and ART were included in the WLWH analysis. Descriptive statistics for these variables are presented as median [interquartile range (IQR)] and proportions. Continuous variables were classified into binary variables and compared using the chi-square test. Categorical variables included age ranges of younger and older women (25–29; 30–49), age at first intercourse (≤18; >18), parity (≤2; > 2), CD4 count (≤ 350 cells/µL; > 350 cells/µL), and histopathology results (benign or CIN 1 [≤ CIN1] for low-grade abnormalities; CIN2+ for high grade anomalies being CIN2/3 and ICC). Patterns of missing data were described for the study cohort using percentages.
Logistic regression models calculated unadjusted and adjusted odds ratios (OR) with 95% confidence intervals (CI). Only exposure variables with a p– values less than 0.1 for unadjusted ORs were included in the adjusted regression models . A p-a value less than 0.05 was considered statistically significant. We used Stata 14.0 (StataCorp LLC, College Station, TX).