Prediction model for prolonged duration of fever, with and without CRP
| Variables | OR | 95% CI | P-value | OR | 95% CI | P-value |
|---|---|---|---|---|---|---|
| Sore throat (PH) | 2.80 | 1.30 to 6.01 | 0.01 | 2.81 | 1.30 to 6.04 | 0.01 |
| Palpable lymph nodes (PE) | 1.87 | 1.01 to 3.49 | 0.05 | 1.87 | 1.00 to 3.49 | 0.05 |
| Duration of fever before consultation (PH) | 0.93 | 0.79 to 1.10 | 0.39 | 0.93 | 0.79 to 1.10 | 0.40 |
| CRP | 1.00 | 0.99 to 1.01 | 0.89 | |||
| Area under the curve (mean, SD) | 0.64 (0.02) | 0.64 (0.03) |
CRP = C-reactive protein. OR = odds ratio. PE = physical examination. PH = patient history. Multicollinearity: there was multicollinearity between ‘sore throat’ and ‘signs of throat infection’. Therefore, the variable concerning physical examination (‘signs of throat infection’) was dropped. The patient history forms included categorical variables with possible answers: ‘no, little, very, very much, and do not know’. These variables were dichotomised using a cut-off point between ‘little’ (including ‘do not know’) and ‘very’. The physical examination forms included categorical variables with possible answers: ‘no, little, very, very much’. These variables were dichotomised using a cut-off point between ‘little’ and ‘very’.