Abstract
Purpose:
Develop novel screening criteria through evaluation of potential risk factors for treatment-requiring retinopathy of prematurity (ROP), including threshold or type 1, in infants who were screened in 11 Neonatal Intensive Care Units in Colombia, South America between January 2008 and June 2012.
Methods:
Retrospective study of 380 infants with a birth weight (BW) of ≤ 1750 g and/or a gestational age (GA) of ≤ 32 weeks that were screened by one physician. Univariate and multivariate logistic regression models were fit to determine the odds ratio (OR) and 95% confidence intervals (CI) for treatment-requiring ROP, in association with putative risk factors (year of exam, gestational age, birth weight, sepsis, oxygen, hyaline membrane disease (HMD), transfusion and intraventricular hemorrhage).
Results:
The mean GA and BW for the 92 cases (24%) of treatment-requiring ROP were 28.2±2.00 weeks and 1049±281 grams respectively, compared to 30.6±2.01 weeks and 1394±320 grams in controls. Statistically significant risk factors in multivariate analysis included birth weight (OR=0.998, CI=0.996-0.999, p<0.001), gestational age (OR=0.69, CI=0.56-0.83, p<0.001), sepsis (OR=2.2, CI=1.11-4.51, p=0.0228), and year of exam (OR=5.10, CI=2.34-11.41, p<0.0001). HMD was not statistically significant, but was perhaps clinically significant (OR=2.05, CI=0.97-4.54, p=0.0609). History of oxygen administration, transfusion, perinatal surgeries and intraventricular hemorrhage were not independent predictors of high risk ROP.
Conclusions:
Our results indicate that sepsis and possibly HMD are of particular importance in ROP screening in developing nations such as Colombia. It also emphasizes the individual predictive value of both gestational age and birth weight. This information will help to ensure identification of all patients at risk of developing ROP that could benefit from treatment, while decreasing the time and resources required of overextended physicians and the exposure to exams in premature infants. Specifically, utilizing this data to generate a predictive score would decrease the number of patients requiring screening by 26% while still retaining 99% sensitivity (area under ROC curve = 0.86).
Keywords: 706 retinopathy of prematurity •
464 clinical (human) or epidemiologic studies: risk factor assessment •
460 clinical (human) or epidemiologic studies: health care delivery/economics/manpower