Intravenous Ketamine for Refractory Cancer Pain in 433 Adults and Children: A Real-World Retrospective Study from a Comprehensive Cancer Centre in Jordan

Background: Ketamine is used as an adjunct analgesic for cancer pain refractory to opioids, but evidence remains limited to small series, predominantly from high-income settings. Few studies include both adults and children, and data from the Middle East are sparse. Methods: We conducted a retrospective study of all patients who received intravenous ketamine for pain at King Hussein Cancer Center (KHCC), Jordan, between February 2018 and April 2026. Pain scores, medications, and vital signs were extracted from the structured EHR. Unstructured variables (pain mechanism, adverse events) were extracted using a schema-constrained large language model pipeline (GPT-4.1-mini) and validated in a 10% random sample by two investigators (AA, EK). The primary outcome was the paired within-patient change in peak pain (maximum NRS in 72 hours pre-ketamine vs maximum NRS in 24 hours post), tested by the Wilcoxon signed-rank test. Results: 433 patients received IV ketamine (183 [42%] paediatric; 72% deceased at follow-up). Among 141 patients with paired peak data, the median paired reduction in peak pain at 24 hours was 1 point (IQR 0 to 5; mean 2.5 ± 3.8; p < 0.001; matched-pairs rank-biserial r = 0.58). Among 107 patients with baseline peak pain >=3, 67 (63%) achieved a >=2-point reduction. Planned sensitivity analyses using peak-to-nadir and mean-to-mean metrics confirmed the direction of the effect. Adverse events were documented in 25 patients (5.8%); retrospective ascertainment likely underestimates true incidence. Conclusions: In the largest single-centre series to date, adjunctive IV ketamine was associated with reduced peak cancer pain in adults and children. The observational design precludes causal inference. These data support ketamine’s feasibility as a palliative analgesic adjunct across age groups in a middle-income setting. The bimodal response pattern (63% responders alongside non-responders) is a hypothesis-generating finding for future trial enrichment. A placebo-controlled randomised trial stratified by pain mechanism is a logical next step where equipoise can be established.