Skip to main content

Table 5 Multivariable prediction models derived from backward stepwise logistic regression on the four prediction subsets among 334 complete questionnaire respondents (model 1 through 4)

From: Signs and symptoms do not predict, but may help rule out acute Q fever in favour of other respiratory tract infections, and reduce antibiotics overuse in primary care

  Model 1 Model 2 Model 3 Model 4
B P SE OR B P SE OR B P SE OR B P SE OR
Variable
Sex 1.13 <0.001 0.33 3.09 0.66 0.04 0.32 1.94 0.56 0.08 0.32 1.75 0.91 0.01 0.32 2.48
Age 40–59 yrs      0.83 0.01 0.31 2.29         
Age ≥60 yrs −0.91 0.01 0.37 0.40      0.82 0.01 0.33 2.26 −0.57 0.10 0.34 0.57
Active smoking      0.77 0.03 0.35 2.16 −0.62 0.08 0.35 0.54 0.87 0.01 0.34 2.39
Fever 0.83 0.07 0.46 2.30 1.24 <0.001 0.38 3.45 1.56 <0.001 0.41 4.77 1.25 <0.001 0.38 3.48
Flu-like illness 0.98 0.02 0.41 2.66             
Pneumonia 1.24 0.00 0.38 3.45 1.02 0.01 0.40 2.78 1.25 0.00 0.41 3.47 1.43 0.00 0.42 4.16
Confusion 0.99 0.06 0.53 2.68             
Severe fatigue 0.59 0.10 0.36 1.81             
Abdominal pain −1.15 0.01 0.42 0.32 −0.76 0.05 0.38 0.47 −0.75 0.05 0.38 0.47 −0.88 0.02 0.37 0.42
Earache −1.02 0.04 0.49 0.36             
Chest pain      0.61 0.08 0.35 1.84         
Cough      −0.63 0.05 0.32 0.53 −0.99 0.04 0.47 0.37 −0.86 0.01 0.33 0.42
Headache          1.14 0.04 0.56 3.12 0.64 0.06 0.35 1.90
Constant −2.42 <0.001 0.56 n.a. −2.15 <0.001 0.47 n.a. −1.80 <0.001 0.46 n.a. −1.55 <0.001 0.50 n.a.
χ2    71.00     52.00     65.00     67.00  
df    9.00     8.00     8.00     8.00  
P    <0.001     <0.001     <0.001     <0.001  
Nagelkerke R2    0.37     0.28     0.34     0.33  
Hosmer & Lemeshow    0.69     0.46     0.81     0.71  
Classification accuracy    75.1%     71.8%     73.1%     73.1%