March 2012
Volume 53, Issue 14
Free
ARVO Annual Meeting Abstract  |   March 2012
Improving The Fit In Logistic Regression Models Of Retinopathy Of Prematurity: The Square Of Birth Weight As A New Covariate Of Risk
Author Affiliations & Notes
  • Simon Dulku
    Birmingham & Midland Eye Centre, City Hospital, Birmingham, United Kingdom
  • Chinedu N. Igwe
    Birmingham & Midland Eye Centre, City Hospital, Birmingham, United Kingdom
  • Roger L. Holder
    Department of Primary Care, University of Birmingham, Birmingham, United Kingdom
  • Lucilla Butler
    Birmingham & Midland Eye Centre, City Hospital, Birmingham, United Kingdom
  • Footnotes
    Commercial Relationships  Simon Dulku, None; Chinedu N. Igwe, None; Roger L. Holder, None; Lucilla Butler, None
  • Footnotes
    Support  None
Investigative Ophthalmology & Visual Science March 2012, Vol.53, 5872. doi:
  • Views
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Simon Dulku, Chinedu N. Igwe, Roger L. Holder, Lucilla Butler; Improving The Fit In Logistic Regression Models Of Retinopathy Of Prematurity: The Square Of Birth Weight As A New Covariate Of Risk. Invest. Ophthalmol. Vis. Sci. 2012;53(14):5872.

      Download citation file:


      © ARVO (1962-2015); The Authors (2016-present)

      ×
  • Supplements
Abstract

Purpose: : To determine whether birth weight (BW), Z-score, or centile gives a better prediction of the risk of requiring treatment in a logistic regression model of retinopathy of prematurity (ROP) and to test for higher order relationships in these covariates.

Methods: : A retrospective study of 299 infants of 32 weeks Gestational Age (GA) or less who were screened for ROP at City Hospital, Birmingham, United Kingdom between 1 January 2001 and 31 October 2009. A stepwise logistic regression model was used to examine the relative merits of GA, BW, Z-score and centile as predictors of the risk of requiring treatment for ROP (termed severe ROP). Then, three logistic regression models were compared: Model 1: GA and BW; Model 2: GA and Z-score; Model 3: GA and centile. Higher order relationships were explored using general linear model (GLM) analysis.

Results: : Stepwise logistic regression chose GA and BW as the best predictors of risk of severe ROP. All models were statistically significant predictors of risk and achieved similar area under receiver operating characteristic curve (AUC) (R2 for Models 1-3 = 0.211, 0.206 and 0.205; AUC=0.805, 0.812 and 0.812 respectively). BW was better modeled as a full quadratic incorporating GA, BW and BW2 with R2 = 0.249 and AUC 0.819.

Conclusions: : The best model for our population was a full quadratic comprising a combination of GA, BW and BW2. The relationship of birth weight to the risk of severe ROP is therefore non-linear. BW2 is an important new covariate of risk in ROP.

Keywords: retinopathy of prematurity 
×
×

This PDF is available to Subscribers Only

Sign in or purchase a subscription to access this content. ×

You must be signed into an individual account to use this feature.

×