ABSTRACT
BACKGROUND: There is scarce information about whether the diagnosis of OSA supported only by medical record data can be a useful and reliable tool to initiate a CPAP treatment. OBJECTIVES: The aim of this study is to develop and assess the accuracy of clinical parameters for the diagnosis and prescription of CPAP in patients with suspected OSA. METHODS: Adult patients who underwent polysomnography and completed the Berlin questionnaire, a clinical record, and the Epworth sleepiness scale were included in the study. A situation was simulated in which two blinded and independent observers would be able to indicate CPAP treatment if the patients were snorers with frequent apnea reports (≥3-4 times a week) and overweight (BMI > 25 kg/m(2)) plus one of the following: diurnal symptoms (tiredness after sleeping or at waking time ≥3-4 times a week or Epworth sleepiness scale >11), arterial hypertension, cerebrovascular accident, coronary event, type II diabetes or cardiac arrhythmias (observer 1, clinical criteria) or on the basis of the respiratory disturbance index, significant tiredness (≥3-4 times a week) or sleepiness (Epworth >11) and associated comorbidities (observer 2, reference method). The area under the ROC curve (ABC-ROC), sensitivity, specificity, and likelihood ratios were calculated. RESULTS: Among 516 subjects (72 % men), the median age was 52 years, BMI 28.3 kg/m(2), and RDI 19.7 events/h. The ABC-ROC, sensitivity, specificity, and positive likelihood ratio of the clinical parameters were of 0.64 to 0.65, 31 to 33 %, 97 to 98 %, and 11 to15 respectively. No differences in the diagnostic performance of the clinical criteria were observed between men and women. CONCLUSIONS: These clinical parameters made it possible to indicate CPAP in approximately one third of the population with OSA which would have required it on the basis of their PSG and clinical history. This approach showed high specificity; hence, few patients who did not meet the criteria for CPAP use would have received this treatment.
