Syphilis clustering among young pregnant women in Kampala, Uganda
Listed in
This article is not in any list yet, why not save it to one of your lists.Abstract
Introduction
In Uganda, the spatial distribution of syphilis varies by age, gender, and region. Identifying clusters (subsets of administrative subdivisions) with high syphilis prevalence could boost efforts to eliminate mother-to-child transmission of syphilis. We examined spatial variations and clustering of syphilis prevalence among pregnant young women in Central Uganda.
Methods
We analysed secondary data from a randomised trial that evaluated the effectiveness of three antenatal syphilis partner notification approaches ( NCT02262390 ). This study analysed clustering of syphilis prevalence by administrative division in Kampala and Wakiso districts, using Moran’s I tests and Local Indicator of Spatial Association (LISA). We used the Kulldorff Spatial-Scan Poisson model to classify divisions with high or low syphilis prevalence (HP/LP) based on 95% statistical significance. We estimated prevalence ratios for sociodemographic and bio-behavioural HIV risk factors associated with clustering, stratified by HIV status, using modified Poisson regression.
Results
Of 422 young women diagnosed with syphilis, 26 (6%) had HIV and syphilis. The median age was 26 years (IQR 24-29). Most (314, 74%) were in monogamous marriages, and half (50%) had ≤13 years of schooling. Syphilis prevalence clustering was negatively correlated with being in a polygamous marriage (adjusted prevalence ratio [APR]=0.64; 95%: 0.47-0.88), having an unplanned pregnancy (APR=0.78; 95% CI: 0.64-0.93) and HIV testing >3 months prior (APR=0.83, 95% CI: 0.72-0.95). Syphilis prevalence was significantly higher in 3 of 12 clusters–Kasangati Town Council (Relative Risk [RR]=2.79, p<0.0001), Kawempe (RR=2.52, p<0.0001), and Nabweru (RR=1.95, p=0.0002), and lower in one cluster–Kyengera Town Council (RR=0.12, p<0.0001). Notably, no significant clustering was detected among women with HIV (p>0.05). Random patterns of syphilis prevalence were detected across all divisions (Moran’s I=0.08, p=0.19). However, some neighbouring divisions had similar prevalence: Kawempe (1.06, p=0.02) and Nabweru (0.54, p=0.045). LISA analysis confirmed high syphilis prevalence in northern divisions (Kawempe and Nabweru; p=0.01). By contrast, Central Region had neighbouring low and high prevalence divisions (Kawempe and Central; p=0.001).
Conclusion
Syphilis prevalence was similar within neighbouring divisions, but highest in Kasangati Town Council and Kawempe. Scaling up spatial analysis application tools enables the detection of clusters where interventions can be targeted to eliminate congenital syphilis.
