Risk factors for blindness
| Risk factor | No cerebral injury (n=15 589) | Cerebral injury* (n=4274) | ||||
| n with blindness/n total | Adjusted OR† (95% CI) | P value | n with blindness/n total | Adjusted OR† (95% CI) | P value | |
| No severe ROP (n=15 295) | 27/12 459 (0.2%) | 1.0 (reference group) | 47/2836 (1.7%) | 8.38 (5.3 to 13.3) | <0.0001 | |
| Severe ROP‡ (n=4568) | 53/3130 (1.7%) | 8.14 (4.5 to 14.6) | <0.0001 | 86/1438 (6.0%) | 28.7 (16.0 to 51.7) | <0.0001§ |
In each cell, data are displayed as (infants with primary outcome/total infants in category).
*The marginal adjusted OR for any cerebral injury in the presence of severe ROP was 3.53 (95% CI 2.26 to 5.50), p<0.0001.
†The adjusted ORs and p values are from a logistic regression model using GEE; the model adjusted for sex, birth weight, multiple birth, maternal race, cerebral injury, severe ROP, the interaction between cerebral injury and severe ROP, and centre as a cluster effect. Additionally, the following variables were removed during backward stepwise selection: maternal public insurance, gestational age, birth year, antenatal steroids, Hispanic ethnicity, maternal hypertension and maternal insulin-dependent diabetes. Multiplicative interaction between cerebral injury and ROP was found to be statistically significant (p=0.01).
‡The marginal adjusted OR for any severe ROP in the presence of cerebral injury was 3.43 (95% CI 2.31 to 5.09), p<0.0001.
§Compared with children with neither risk factor, the presence of both cerebral injury and severe ROP was associated with an adjusted OR of 28.73 (95% CI 15.96 to 51.71), p<0.0001. This comes from a logistic regression model assessing cerebral injury and severe ROP as a four-level variable (cerebral injury only, severe ROP only, both, neither) with no interaction term.
GEE, generalised estimating equation; ROP, retinopathy of prematurity.