Cardio-Respiratory Poligraphy (CRP) is useful for evaluating patients with suspected Obstructive Sleep Apnea (OSA) in specific situations. Outcome: To describe interobserver variability between an expert and an in-training scorers in the analysis of CRP carried out in patients with suspected OSA. To evaluate the effects of an specific training program in CRP scoring.
20 patients older than 18 years, with suspected OSA, studied with CRP level 3 (Stardust II, Respironics). Scoring was performed by an expert scorer -physician specialized in Sleep Medicine- (A) and an in-training technician with a short experience in scoring PSG (B). Standard criteria established by AASM were applied. Scorers were blinded of patients data and another scorer‘s results. After first analysis, in-training observer was submitted to an specific CRP scoring training under the supervision of two specialized physicians. 30 days after training program was completed, 20 CRP studies were scored again. Both observers were blinded of patients data and another scorer‘s results. OSA severity was established according to Respiratory Disturbance Index (RDI), normal < 5ev/h; mild 5–14.9; moderate 15–29.9, severe 30–49.9 and very severe > 50. Interobserver variability in the diagnostic result was valued as the percentage of agreement and kappa coefficient, a kappa value > 0.7 was considered acceptable. Pre and post-training RDI mean differences between scorers were compared.
13/20 males; age 48.1 ± 17.3 ys; BMI 28.6 ± 5. RDI mean difference between scorers was 4.82 ± 5.55 ev/h. When we compared diagnostic classification for every patient between observers the % of concordance was 55% and kappa 0.42. In the post-training evaluation the RDI mean difference fell down to 2.03 ± 2.40 ev/h (p = 0.02) whereas the % of concordance raised to 90% and kappa coeff to 0.87.
When we use CRP for diagnosing OSA the level of agreement between expert scorer and in-training technician is low, but substantially improve after an specific training.
An specific training in CRP scoring and interpretation is simple and very necessary.
Facundo Nogueira, No Financial Disclosure Information; No Product/Research Disclosure Information