In this study we present the largest, most inclusive study to date evaluating the impact and incidence of rAKI in critically ill neonates. The current study shows that rAKI is common in our NICU, affecting approximately 7% of all infants and 39% of those with any AKI. Many of the risk factors known to be associated with AKI also predict rAKI in our cohort. Lastly, we show that rAKI is independently associated with increased LMV and increased LOS. The current study highlights the importance of rAKI in neonates and the need for further study of these high-risk patients.
As the impact of AKI in critically ill neonates has become clear over the last decade, research is transitioning to better understanding high risk populations. To our knowledge, to date, there has been only one published single center study of rAKI in a NICU. Adegboyega et al. examined incidence, risk factors, and outcomes of rAKI in infants less than 28 weeks GA. In the current study we show that rAKI occurred in 39% of the infants that had a single episode of AKI in a broad neonatal population, including all GA categories. Furthermore, we describe the incidence of rAKI by GA cohort for the first time (50% in < 28 weeks GA, 30% in 28–35 weeks GA, 37% in > 36 weeks GA). To our knowledge this is the first study describing the incidence of rAKI in critically ill neonates > 28 weeks GA. In the current study we show that AKI episodes did not become more severe the more they recurred.
There are known risk factors for AKI, but it has not been well studied if these same risk factors are associated with rAKI episodes. We found infants with rAKI had lower BW and younger GA than those with sAKI or no AKI. Notably, we found that many, but not all the risk factors known to contribute to AKI, also contribute to rAKI development. Our study showed that PDA, sepsis, NTX exposure, hypotension with pressor requirement, CHD, and known renal disease were associated with rAKI suggesting that perhaps repeated occurrences or continued insults from these risk factors may lead to rAKI episodes. Interestingly, while HIE (or asphyxia) is known to be associated with AKI, we did not find an association between HIE and rAKI. This may reflect the relatively short duration of hospitalizations in this population as compared to premature neonates. In future studies it will be important to independently study the risk factors associated with rAKI in more detail.
AKI has been shown to be associated with increased morbidity and mortality across a variety of neonatal populations. In the previously mentioned study by Adegboyega et al., they showed that preterm infants with rAKI had higher mortality than those with sAKI. In the current study we showed that rAKI was associated with increased LMV and increased LOS. Those with rAKI had longer LOS and LMV than those without AKI and those with sAKI episodes. An interesting finding in the current study is that the mortality was highest among those with sAKI, and sAKI was independently associated with > 5x higher odds of mortality compared to no AKI and almost 4x higher odds of mortality than those with rAKI. One explanation for this is that an infant must survive the first episode of AKI to go on to develop a recurrent episode of AKI. Regardless, this underscores the importance of prevention of even a single AKI occurrence in the NICU. The current study highlights that rAKI is a separate clinically distinct entity from sAKI that warrants close attention and further study. Our study has several strengths including a large, broad cohort of infants of all gestational ages cared for in our NICU where AKI detection and care is standardized. Another strength of the study is that we utilized up-to-date methodologies such as modified, neonatal KDIGO criteria for diagnosis of AKI and only classified rAKI once there was complete resolution of any previous AKI episodes. There are some limitations to our study. We relied upon clinical documentation in electronic medical records which can be incomplete. There is also potential for missed AKI episodes as urine output criteria was not assessed due to the difficulty of measuring urine output in infants thus AKI and rAKI incidences may be underestimated here. While the modified, neonatal KDIGO criteria is considered the gold standard for diagnosing and staging AKI in the NICU, our study highlights its limitations. It does not capture all AKI episodes in infants whose SCr does not downtrend as expected after birth. Also, per modified neonatal KDIGO criteria, AKI is defined as 1.5 times baseline which means that a small increase in SCr from 0.2 mg/dL to 0.3 mg/dL is classified as AKI, which may be overestimating AKI episodes in infants. Continued revision and refinement of this definition is warranted.