Phototherapy (PT) is widely used and assumed to be safe as well as effective in reducing total bilirubin (TB) levels. Our recent NICHD Network Trial showed that aggressive use of phototherapy reduces neurodevelopmental impairment (NDI), but may increase deaths among ELBW infants. Among ventilator treated infants \<750 g birth weight (BW) (n =696), conservative Bayesian analyses (using a neutral prior probability) identified a 99% (posterior) probability that aggressive phototherapy reduced profound NDI but a 99% probability that it increased deaths relative to conservative phototherapy. The possibility that PT increases deaths among high risk infants is also suggested by the Collaborative Phototherapy trial (performed in the 1970s), the only large RCT in which LBW infants were randomly assigned to receive PT or no PT. The relative risk for death among those randomized to PT relative to those randomized to no PT was 1.32 (0.9-1.82) among all LBW infants and 1.49 (0.93-2.40) among ELBW infants. These findings are consistent with a major increase in mortality but have been ignored because the p was \>0.05, an error often made in ignoring important potential treatment hazards when power is limited.
Multiple studies, most performed decades ago in larger infants, found that short on/off cycles of PT (e.g. 15 min on/60 min off, 1 h on/3 h off, or 1 h on/1 h off ) are as effective as uninterrupted PT to reduce TSB. (Cycles with \>6 h off PT do not appear to be as effective as uninterrupted PT). The clinical use of uninterrupted rather than cycled PT appears to be based largely on the assumption that PT is safe for all infants.
