King-Devick Sensitivity and Specificity to Concussion in Collegiate Athletes

Abstract

Context: The King-Devick (K-D) test is used to identify oculomotor impairment after concussion. However, the diagnostic accuracy of the K-D test over time has not been evaluated.

Objectives: To (1) examine the sensitivity and specificity of the K-D test at 0 to 6 hours postinjury, 24 to 48 hours postinjury, the beginning of a return-to-play (RTP) protocol (asymptomatic), unrestricted RTP, and 6 months postconcussion and (2) compare outcomes between athletes with and those without concussion across confounding factors (sex, age, sport contact level, academic year, learning disorder, attention-deficit/hyperactivity disorder, migraine history, concussion history, and test administration mode).

Design: Retrospective, cross-sectional design.

Setting: Multiple institutions in the Concussion Assessment, Research and Education Consortium.

Patients or other participants: A total of 320 athletes with a concussion (162 men, 158 women; age = 19.80 ± 1.41 years) were compared with 1239 total collegiate athletes without a concussion (646 men, 593 women; age = 20.31 ± 1.18 years).

Main outcome measure(s): We calculated the K-D test time difference (in seconds) by subtracting the baseline from the most recent time. Receiver operator characteristic (ROC) curve and area under the curve (AUC) analyses were used to determine the diagnostic accuracy across time points. We identified cutoff scores and corresponding specificity at both the 80% and 70% sensitivity levels. We repeated ROC with AUC analyses using confounding factors.

Results: The K-D test predicted positive results at the 0- to 6-hour (AUC = 0.724, P < .001), 24- to 48-hour (AUC = 0.701, P < .001), RTP (AUC = 0.640, P < .001), and 6-month postconcussion (AUC = 0.615, P < .001) tim points but not at the asymptomatic time point (AUC = 0.513, P = .497). The 0- to 6-hour and 24- to 48-hour time points yielded 80% sensitivity cutoff scores of -2.6 and -3.2 seconds (ie, faster), respectively, but 46% and 41% specificity, respectively. The K-D test had a better AUC when administered using an iPad (AUC = 0.800, 95% CI = 0.747, 0.854) compared with the spiral-bound card system (AUC = 0.646, 95% CI = 0.600, 0.692; P < .001).

Conclusions: The diagnostic accuracy of the K-D test was greatest at 0 to 6 hours and 24 to 48 hours postconcussion but declined across subsequent postconcussion time points. The AUCs did not differentiate between groups across confounding factors. Our negative cutoff scores indicated that practice effects contributed to improved performance, requiring athletes to outperform their baseline scores.

Keywords: baseline testing; diagnostic accuracy; mild traumatic brain injury; oculomotor performance; postconcussion assessment.

Figure 1

Receiver operating characteristic curve and area under the curve (AUC; ± SE) across postconcussion time points: A, 0–6 h postconcussion (concussion group = 1239, control group = 159), B, 24–48 h postconcussion (concussion group = 1239, control group = 229), C, beginning of a return-to-play protocol (asymptomatic; concussion group = 1239, control group = 283), D, unrestricted return to play (concussion group = 1239, control group = 279), and E, 6 months postconcussion (concussion group = 1239, control group = 171). The horizontal dotted line at y = 80 indicates 100% specificity at a sensitivity of 80%. ^a Inverted receiver operating characteristic curve.

Figure 2

Receiver operating characteristic curves at the within–48-h time point segmented by, A, sex, B, sport contact level, C, academic year, D, learning disorder, E, attention-deficit/hyperactivity disorder, F, migraine history, G, concussion history, and H, administration mode (P < .05). The horizontal dotted line at y = 80 indicates 100% specificity at a sensitivity of 80%. ^a Indicates difference in area under the receiver operating characteristic curve (P < .001).