Prehospital delay and the care cascade in intracranial suppurative infections: an exploratory mediation analysis of in-hospital prognosis in a resource-limited ivorian cohort

Background: In our cohort of intracranial suppurative infections (ICS), the median interval from symptom onset to admission reached 19 days, a salient feature. Hypothesis: We hypothesised that a longer prehospital delay is associated with higher in-hospital mortality, partly through deterioration of consciousness at admission (Glasgow Coma Scale, GCS). Methods: Single-centre secondary analysis of 150 stays (148 patients; 16 deaths; mortality 10.7%). The care cascade separated the prehospital phase (onset → admission; n = 150) from the in-hospital phase (admission → surgery; n = 115). Associations between delay, GCS and lesion characteristics used Spearman’s coefficient and age- and sex-adjusted models. A counterfactual mediation analysis (natural direct and indirect effects as odds ratios; percentile bootstrap, 2,000 resamples) followed an a priori directed acyclic graph. Firth regression, sensitivity analyses and an E-value assessed robustness. Given 16 deaths, analyses are exploratory. Results: The median prehospital delay (19 days; IQR 15–25) was about nine to ten times the admission-to-surgery delay (2 days; IQR 1–3). A prolonged delay (> 19 days) was associated with rural residence (aOR 2.27; 95% CI 1.16–4.43) and, borderline, with referred status (aOR 1.88; 0.96–3.67). A longer delay was associated with a lower admission GCS (ρ = −0.25; p = 0.002; −0.67 points per 10 days after adjustment), with no association with lesion size. The total effect on death was imprecise (OR 1.79 per +10 days; 0.85–4.52). The share transiting through GCS was about 19% but very wide (−48 to 105%); the indirect effect (OR 1.11; 0.99–1.39) was non-significant. Conclusion: The delay in ICS care was mainly prehospital and associated with structural access factors (rurality, referral) and with neurological deterioration at admission. The delay–rurality and delay–GCS links were most robust; the effect on mortality remained imprecise. Prehospital delay is a candidate target for organisational interventions, whose effectiveness must be assessed prospectively.