Assessing the effectiveness of the Friedman classification and its components in predicting obstructive sleep apnea: a Brazilian cross-sectional study.

purposeThis study assessed the effectiveness of the Friedman classification, including the modified Mallampati index (MMI) and tonsil size grading in predicting obstructive sleep apnea (OSA), comparing them to the validated STOP-BANG and GOAL questionnaires.

methodsA cross-sectional study of outpatients undergoing home sleep apnea testing (HSAT) was conducted from November 2022 to May 2025. Standardized examinations of the oral cavity and oropharynx were recorded blind to HSAT results. OSA severity was defined as any, moderate-to-severe, or severe OSA. Discriminatory accuracy was assessed by the area under the curve (AUC).

resultsOverall, 558 adults were analyzed, with 78.7% diagnosed with OSA, including 36.7% with moderate-to-severe OSA, and 19.9% with severe OSA. For all severity levels of OSA, the MMI showed a sensitivity of 65.4% to 73.0% and a specificity of 38.5% to 42.0%. Tonsil size showed a sensitivity of 19.6% to 22.5% and a specificity of 81.9% to 83.2%. The Friedman classification had a sensitivity of 61.3% to 66.7% and a specificity of 42.5% to 47.9%. The Friedman instrument and its components performed poorly as screening instruments at all OSA severity levels, with AUC values of less than 0.6. Conversely, STOP-BANG and GOAL exhibited adequate discrimination with AUC values from 0.677 to 0.768 and 0.669 to 0.721, respectively. No significant differences in AUC values were found between the STOP-BANG and GOAL tools for all OSA severity levels.

conclusionThe Friedman classification and its components are insufficient as screening tools for OSA in adults, unlike the STOP-BANG and GOAL questionnaires.