Abstract
BACKGROUND: Cervical cancer remains a leading cause of cancer-related deaths in low- and middle-income countries. In Ghana, only a few women of reproductive age have ever been screened for cervical cancer. Most existing evidence is derived from population-level studies focusing on all women of reproductive age, with little attention to pregnant women as a distinct subgroup. This is striking given that pregnant women frequently come into contact with antenatal services, which could serve as a critical platform for offering screening. This study examined predictors of non-screening among pregnant women in the Pru East Municipality, Bono East Region of Ghana. METHODS: A facility-based cross-sectional study was conducted among 456 pregnant women. Data were collected using closed-ended questionnaires and analysed with STATA 17 and IBM SPSS AMOS 29. Guided by Andersen's Behavioural Model of Health Services Use, descriptive and inferential statistics, including chi-square tests, logistic regression, and confirmatory factor analysis (CFA), were performed as part of data analysis. RESULTS: Overall, 87.3% (95% CI: 84.2-90.4) of pregnant women reported never having undergone cervical cancer screening before or during the current pregnancy. Most of the pregnant women (74.3%) and (78.1%) demonstrated a moderate level of knowledge regarding cervical cancer and its screening, respectively. Not having formal education, low knowledge levels, unemployment, rural residency, limited media exposure, restrictive cultural norms, community unawareness, social stigma, perceived high cost of service charges for screening, and unfriendly health worker attitudes were significantly associated with non-screening. The structural model of latent construct demonstrated excellent fitness (AGFI = 0.932; CFI = 0.914; GFI = 0.921; IFI = 0.947; NFI = 0.931; RMSEA = 0.066). Moreover, strong correlations were observed between enabling and systemic factors (r = 0.929), followed by systemic and predisposing factors (r = 0.613), and enabling and predisposing factors (r = 0.585). Regarding the structural path to non-screening, the predisposing construct had the strongest direct effect on non-testing behaviour (β = 0.39), followed by the systemic construct (β = 0.25). CONCLUSION: Beyond the observed rate, the odds of non-screening were shaped by a constellation of determinants across the three domains of Andersen's Behavioural Model of Health Services Use. Therefore, efforts to improve cervical cancer screening should move beyond fragmented approaches to integrated and multisectoral strategies that address predisposing, enabling, and systemic barriers simultaneously.