Objective To determine whether elevation of a biological marker of inflammation would be a better predictor of bronchopulmonary dysplasia (BPD) development than lung function measurement results.

Design Prospective study.

Setting Tertiary neonatal intensive care unit.

Patients 78 prematurely born infants (median gestational age 29 (range 24–32) weeks) were studied; 39 developed BPD.

Interventions BPD was diagnosed as oxygen dependence at 28 days.

Main outcome measures Levels of a biological marker of inflammation (carbon monoxide) were assessed by measurement of end-tidal carbon monoxide (ETCO) and lung function by measurement of functional residual capacity (FRC) and compliance (Crs) and resistance (Rrs) of the respiratory system on days 3 and 14 after birth. Possible predictive factors were modelled for BPD and for BPD severity.

Results Gestational age, birth weight, ETCO, FRC and Crs results on days 3 and 14 differed significantly between infants who did and did not develop BPD. In multifactorial logistic regression, only birth weight and ETCO results (on day 14) remained significant predictors of BPD with an area under the curve of 0.97. The final multifactorial model for the severity of BPD included those two factors, plus septic episodes.

Conclusion These results emphasise the importance of ongoing inflammation in the development of BPD; ETCO levels, rather than lung function test results, were the more accurate predictor of BPD development.