0537 Understanding Factors That Influence the Short-Term Reliability of Obstructive Sleep Apnea Symptom Subtypes

Abstract Introduction Multiple studies from clinical and population samples have established the existence of symptom-based subtypes of OSA. An important step towards clinical translation is understanding factors influencing the short-term reliability of these subtypes. We describe new results from an ongoing study in the Sleep Apnea Global Interdisciplinary Consortium (SAGIC) to answer this question. Methods Subjects with newly diagnosed OSA completed a questionnaire on symptoms twice (at least two weeks apart) prior to therapy. Symptom subtype at each timepoint was determined based on questionnaire responses using a prediction model derived in existing SAGIC patients. To understand short-term reliability, we calculated simple kappa coefficients indicating slight (0.00-0.20), fair (0.21-0.40), moderate (0.41-0.60), substantial (0.61-0.80), and almost perfect (0.81-1.00) reliability and percent agreement. Logistic regression was performed to evaluate factors associated with the likelihood of falling into different subtypes at the two timepoints. Results Data included 394 SAGIC participants from Ohio State (n=101), Kansas (n=20), Sydney (n=82), Perth (n=58), Germany (n=15), Chile (n=23), and Beijing (n=95). At the first visit, 123 (31.2%) participants had the disturbed sleep subtype, 180 (45.6%) were minimally symptomatic, and 91 (23.1%) were excessively sleepy. At the second visit 128 (32.5%) participants had the disturbed sleep subtype, 191 (48.5%) were minimally symptomatic, and 75 (19%) were excessively sleepy. Overall, 76.9% of participants had the same subtype at both timepoints and the kappa was 0.64, indicating substantial reliability. Patients whose initial subtype determination was more certain (e.g., a higher predicted probability) and those initially reporting the minimally symptomatic subtype were more likely to have the same subtype at the next assessment, whereas those with higher BMI were more likely to change subtypes; age, gender, and site were not significantly associated with reliability. Conclusion While recruitment is ongoing across all SAGIC sites, this analysis suggests substantial reliability of symptom subtypes over a short timeframe (without treatment). Patients whose initial subtype was predicted more confidently and those with the minimally symptomatic subtype were more likely to report the same subtype at follow-up. This further supports the potential clinical utility of symptom subtypes in advancing more personalized patient care. Support (if any)