Generalized Linear Mixed-Effects Model to Predict Hospitalization

PredictorAdjusted Odds Ratio (95% CI)P
  29–56 d0.11 (0.06–0.22)<.001b
  57–90 d0.07 (0.03–0.14)<.001b
 Gestational agec.02b
  <37 weeks' gestation1.64 (0.80–3.38).18
  Unknown0.38 (0.17–0.87).02b
 History of chronic illness2.27 (0.85–6.07).10
History of presenting illness
 Decreased urine output4.13 (1.88–9.07)<.001b
 Ear discharge1.97 (1.17–3.31).01b
 New difficulty feeding1.36 (0.81–2.30).25
 Diarrhea0.58 (0.21–1.59).29
Physical examination
 Respiratory distress40.54 (19.17–85.70)<.001b
 Dehydration2.89 (0.87–9.61).08
 Ill appearance1.64 (0.52–5.18).40
Diagnostic test
 Complete blood countd<.001b
  Completed and within normal limits10.95 (5.95–20.15)<.001b
  Completed and abnormal9.25 (4.04–21.14)<.001b
  Completed and within normal limits2.11 (1.15–3.87).02
  Completed and abnormal2.14 (0.49–9.26).31
  • a Reference group: 0 to 28 d old.

  • b µStatistical significance (P < .05). In this model, age, gestational age, history of decreased urine output, history of ear discharge, respiratory distress on examination, and completion of complete blood count were independently associated with hospitalization. In the subanalysis including only the 1459 infants meeting simplified AAP diagnostic criteria (data available on request), the following differences were observed: chronic illness was not included in the model (univariable P > .1) and dehydrated appearance was statistically significant in the regression model (P = .02).

  • c Reference group: ≥37 wk gestation.

  • d Reference group: complete blood count not obtained. Test was considered abnormal if the WBC count was <5 or >15 × 103 cells/µL.

  • e Reference group: urinalysis not obtained. Test was considered abnormal if any of the following was present: leukocyte esterase, nitrite, or >5 WBCs per high-power field.