Socioeconomic and clinical factors associated with odontogenic infection types among hospitalised patients in Periurban Ghana

Background: Odontogenic infections may progress to life-threatening complications if not managed promptly, and socioeconomic circumstances have been linked to their presentation, particularly where access to dental care is limited. Because every patient studied was already hospitalised, this analysis addresses the clinical type of infection at presentation, not the risk of developing one. Methodology: A retrospective cross-sectional review of hospital records was undertaken at the Allied Ward of Sunyani Teaching Hospital (January 2021 to January 2025). Of 811 consecutive admissions, 525 were excluded: 509 for non-odontogenic conditions and 16 for records missing more than 30% of required clinical fields, leaving 286 complete records. Chi-square tests were reported with Cramer’s V and Monte Carlo exact p-values where cells were sparse. Because the unpenalised model showed quasi-complete separation, the primary analysis was a Firth-penalised multinomial logistic regression, with abscess as the reference outcome. Results: Adjusting for age, sex, employment, residence, recorded oral hygiene, systemic comorbidity and alcohol use, unemployment was associated with higher adjusted odds of Ludwig’s angina (LA) and not abscess (AOR = 2.78, 95% CI 1.27–6.09), and rural residence with necrotizing fasciitis (NF; AOR = 3.62, 95% CI 1.33–9.80). Systemic comorbidity was associated with LA (AOR = 4.14, 95% CI 1.83–9.36), NF (AOR = 8.72, 95% CI 2.95–25.75) and cellulitis (AOR = 2.07, 95% CI 1.03–4.18). Absence of a recorded poor-hygiene entry was associated with cellulitis (AOR = 8.05, 95% CI 2.42–26.83), interpreted as a documentation artefact not a protective effect. Sex and alcohol use showed no association. Number of children, the sole source of separation, was not retained; its penalised intervals included 1. Conclusion: Unemployment and systemic comorbidity were associated with more severe presentations, and rural residence with NF specifically. Findings are associational and describe variation in infection type among admitted patients rather than infection risk, within a retrospective single-centre design.