Within-subject haemoglobin variation in elite athletes: a longitudinal investigation of 13 887 haemoglobin concentration readings

The Athlete Biological Passport (ABP) estimates individualized reference ranges for key blood markers, such as haemoglobin concentration ([Hb]), using predetermined population mean, between- and within-subject variances. Here, we aim to reassess previously published estimates for within-subject [Hb] variance and determine whether sex-, analyzer-, sport-, or season-specific values are required. Our reference population contains 7723 male (mean ± SD, 22.3 ± 4.6 years of age) and 6164 female (21.6 ± 4.3) athlete observations from 49 sports. [Hb] was calculated using one of three cytometers; Bayer-H3 (1997-1999, n = 4554), ADVIA-120 (1999-2010, n = 8636) and Sysmex XT-2000i (2010-2012, n = 697). The final model was a linear mixed model for [Hb] with analyzer (H3, ADVIA, Sysmex), sex (male, female), sport (power-endurance, endurance, skill, team, disabled and non-athletes), season (summer, winter), and the interaction between sex and sport as fixed effects and athlete as a random effect. The model included an exponential correlation structure to allow for within-subject autocorrelation, and allowed different within-subject variances for each sport. Within-subject [Hb] variance (g(2) /L(2) ) was significantly less for power endurance (35.09, 95% CI 33.50 to 36.76), disabled (25.82, 95% CI 21.71 to 35.28) and non-athletes (34.30, 95% CI 28.53 to 35.87) than for endurance (40.35, 95% CI 39.62 to 47.22) and team sports (38.70, 95% CI 37.68 to 39.76) athletes. No new evidence was found to justify adjusting the current within-subject [Hb] variance estimate.