Table 4 Screening models for OSA risk
Predictors | Logistic regression | Receiver operating characteristic curve | ||||||
|---|---|---|---|---|---|---|---|---|
OR (95% CI) | p value | Cutoff value | sensitivity | specificity | PPV | NPV | AUC | |
Original models | ||||||||
 Male | 4.2 (1.02–17.05) | 0.047 | Male | 87% | 39% | 83% | 46% | 0.63 |
 BMI | 8.1 (2.06–31.88) |  < 0.01 |  > 24.8 | 78% | 69% | 90% | 47% | 0.75 |
 CAP | 2.4 (0.69–8.44) | 0.17 |  > 243.5 | 67% | 54% | 84% | 32% | 0.54 |
 HSI | 10.1 (1.21–84.03) | 0.03 |  > 38.3 | 46% | 92% | 96% | 32% | 0.72 |
Combined models | Adjusted OR (95% CI) | p value | Predictors | Â | Â | Â | Â | Â |
 Model A |  |  |  ≥ 3 | 65% | 85% | 94% | 41% | 0.75 |
  Male | 2.5 (0.5–12.4) | 0.28 |  |  |  |  |  |  |
  BMI > 24.8 | 5.3 (1.1–25.6) | 0.04 |  |  |  |  |  |  |
  CAP > 243.5 | 0.6 (0.1–3.0) | 0.54 |  |  |  |  |  |  |
  HSI > 38.3 | 6.4 (0.7–61.2) | 0.11 |  |  |  |  |  |  |
 Model B |  |  |  ≥ 2 | 80% | 62% | 88% | 47% | 0.71 |
  Male | 2.4 (0.5–11.7) | 0.28 |  |  |  |  |  |  |
  BMI > 24.8 | 7.0 (1.5–33.6) | 0.02 |  |  |  |  |  |  |
  CAP > 243.5 | 0.9 (0.2–4.2) | 0.87 |  |  |  |  |  |  |
 Model C |  |  |  ≥ 2 | 78% | 85% | 95% | 52% | 0.81 |
  Male | 2.3 (0.5–11.1) | 0.30 |  |  |  |  |  |  |
  BMI > 24.8 | 4.5 (1.0–19.4) | 0.045 |  |  |  |  |  |  |
  HSI > 38.3 | 5.6 (0.6–51.9) | 0.13 |  |  |  |  |  |  |