| Summary: | Traumatic lumbar punctures (TLPs) in children with acute lymphoblastic leukemia are associated with a poorer prognosis. The objective of this study was to determine risk factors for TLPs using a retrospective cohort. We compared and contrasted three different regression methods for the analysis of repeated-measures data. In the multivariable model using generalized estimating equations, variables significantly associated with TLPs were age < l year or ≥ 10 years; body mass index percentile ≥ 95; platelet counts < 100 x 103/µL; fewer days since previous LP, and a preceding TLP. The same variables, with similar estimates and confidence-intervals, were identified by the random-effects model. In a fixed-effects model where each patient was used as their own control, days since prior LP and the effect of using image-guidance were significant. Random-effects and GEE lead to similar conclusions, whereas fixed-effects discards between-subject comparisons and leads to different estimates and interpretation of results.
|
|---|
