Care Trajectories Rather than Outcomes: Sequence Analysis Identifies Four Hospital Pathways in Intracranial Suppurative Infections

Background. Studies of intracranial suppurative infections (ISI) in sub-Saharan Africa look at isolated variables — a delay, a Glasgow Coma Scale (GCS) score, a mortality rate — and rarely describe how a hospital stay unfolds. We asked whether a day-by-day reconstruction of the pathway tells us something that conventional prognostic analyses do not.Methods. Single-centre cohort of 150 admissions (148 patients) for ISI at Bouaké University Hospital, Côte d’Ivoire, 2016–2026. Each stay became a 31-day sequence on two channels, care (seven states) and neurological severity (three GCS bands), using rules fixed before clustering. Sequences were compared by optimal matching and grouped by partitioning around medoids. Stability was tested by bootstrap and eleven sensitivity analyses, admission factors by discrepancy analysis and Firth regressions. The study is exploratory.Results. The 150 admissions produced 100 distinct sequences, grouped into four profiles: rapid recovery (T1, 38.7%), prolonged convalescence (T2, 35.3%), complicated course (T3, 15.3%) and death (T4, 10.7%). Length of stay, GCS, complication and death build the sequences, so their differences between groups are expected and validate nothing. Among other variables, only lesion diameter clearly separated the groups (ε² = 0.185). Symptom-to-admission delay had only a weak effect (ε² = 0.049; unadjusted p = 0.017) that disappeared after adjustment, and referral status was linked to nothing (p = 0.55). Entropy formed a plateau from day 2 to day 12, peaking at 0.72 on day 10. On bootstrap, only T4 was stable (Jaccard 0.985; T1 0.741, T3 0.672, T2 0.599). Four of eleven sensitivity analyses departed clearly from the main solution.Conclusion. Stays fall into three survivor profiles of uneven clarity and one profile of death. The registry dates neither complications nor critical care, so the sequences mainly describe length of stay, complications and outcome; they say nothing about the order of events or about when to intervene. Multicentre validation and a registry that dates complications and critical care are needed.