Objective To determine the value of the Immediate Post-Concussion Assessment and Cognitive Testing (ImPACT) tool in measuring neurocognitive recovery following concussion.

Design Nested case-control.

Setting Sport Medicine Centre, Alberta, Canada.

Participants Ice hockey players (n=96; 88.5% male, 11-17 years old) following concussion who completed ImPACT at baseline and medical clearance to return to play (RTP) were compared to controls (n=31) matched for age, sex, and testing time interval.

Assessment of risk factors Players who sustained a concussion were referred to a study physician for assessment and RTP clearance based on clinical assessment and SCAT2/3. Physicians were blinded to ImPACT scores.

Outcome measures ImPACT composite scores were measured (i.?e., verbal memory, visual memory, reaction time, visual motor) at each time interval [i.e., baseline and RTP (cases)/second measurement (controls)]. Between-interval ImPACT reliable change scores (80%CI) were compared.

Main results At RTP (Median days since injury: 22, range: 2–84), 43/96 (44.8%) concussed players and 11/31 (35.5%) controls had ≥1 composite scores reliably worse than at baseline. Conditional logistic regression analysis (n=31 matched pairs) did not demonstrate a difference in the odds of having ≥1 reliably worse composite score between concussed and controls [OR=2.00 (95% CI: 0.5–8.0)] or ≥1 reliably better composite score [OR=1.2 (95% CI: 0.37–3.93)].

Conclusions The utility of baseline ImPACT scores in medical decisions to RTP may be limited by the reliability of composite scores in youth populations. Caution is warranted when interpreting ImPACT scores post-concussion on an individual level using baseline performance. Future research including multivariable analyses is warranted in a larger sample.

Competing interests Brian Brooks is a co-author of a paediatric memory test for which he receives royalties. No financial interests in ImPACT are declared.

None for all other authors.