Evaluation of sequential serum amyloid A measurements for the prediction of sepsis and survival in critically ill neonatal foals
Eva de Bruijn, Mathilde Laetitia Pas, Donatienne Castelain, Alexander Dufourni, Ellen Paulussen, Bart PardonAbstract
Background
Single measurements of serum amyloid A (SAA) concentration at hospital admission has limited sensitivity for predicting sepsis and death.
Hypothesis/Objectives
To determine whether differences in SAA during the first 2 days of hospitalization could predict sepsis and death in critically ill foals.
Animals
One hundred twenty-seven critically ill neonatal foals, <14 days of age.
Methods
Prospective cohort study. SAA was measured at hospital admission and on day 2 of hospitalization using a validated point-of-care test. Logistic regression was conducted to evaluate the predictive value of individual SAA measurements for diagnosing sepsis and assess whether change in SAA during 48 h could predict sepsis or death. Cox survival analysis assessed the association between SAA concentration and death.
Results
Serum amyloid A (SAA) concentrations were measured on day 0 (n = 127) and day 2 (n = 97) of hospitalization. On admission and day 2, SAA was not significantly associated with death (P = .20 and P = .08), but was significantly associated with neutropenia (OR 1.001; 95% CI, 1.001-1.1002; P ≤ .001), blood culture positivity (OR 1.001; 95% CI, 1.0-1.1001; P = .002), or both (OR 1.002; 95% CI, 1.001-1.1003; P ≤ .001), with optimal cut-offs of 419 μg/mL on day 0, with moderate sensitivity and specificity (80% and 81%, respectively). Changes between day 0 and day 2 were not predictive for death (P = .49), neutropenia (P = .36), blood culture positivity (P = .41) or both (P = .17).
Conclusions and clinical importance
In this cohort, SAA only had moderate ability to rule in or rule out sepsis or predict death. Repeated measurements after 48 h did not improve accuracy, suggesting that SAA should not be used as a standalone test.