In this study, we compared the different outcomes according to AKI duration in children. We identified the temporal evolution of AKI based on kidney injury duration. The discrimination between the biological parameters, AKI causes and AKI severity was significant among the three groups of transient AKI,  persistent AKI, and acute kidney disease. Patients with acute kidney disease presented the highest mortality rates, followed by those in the transient AKI and persistent AKI groups, respectively. The underlying disease severity, reflected by the biological parameters, as well as disease evolution and management are associated with these outcomes. Besides younger age, the biological parameters from day one of AKI and the intrinsic cause were associated with a prolonged AKI episode. Regardless of AKI severity, progression to CKD was higher in patients with acute kidney disease and in children with intrinsic AKI. Subsequent AKI episodes did not increase the risk of CKD.

A recent meta–analysis of AKI incidence in hospitalised children reported a high incidence of AKI (26%) [16]. However, we report a higher incidence than the 0.39% AKI incidence reported by Sutherland in an American population [17], probably because we incorporated a wider spectrum of ages (including neonates). Indeed, we divided the cohort based on the duration of AKI and we found two studies that followed this division [10, 11].

In the first one, Nagata et al., compared the effects of transient AKI, persistent AKI and acute kidney disease on the long–term renal prognosis in adult settings [11]. In their study, persistent AKI and acute kidney disease had higher incidences (40%) compared to transient AKI (19%) [11]. Conversely, LoBasso et al., found transient AKI in 87% of the children undergoing cardio-pulmonary by-pass who developed AKI, followed by 6.7% in persistent AKI, and 6.2% in acute kidney disease, respectively [10]. Our results are closer to those reported in adults, despite the larger number of transient AKI cases in our cohort. The effect of kidney injury duration depends on the initiating AKI event, renal functional reserve and proper management. This is why, in the latter study, the initiating event, represented by cardio-pulmonary by-pass, induced a pre-renal AKI episode that resolved within 48 h [10]. Unlike Lo Basso’s cohort, our study included patients from both intensive care units and hospitalisations, from all medical specialities, including neonates. The reported incidence of acute kidney disease in children is heterogeneous, ranging between 6.3% [10] and 56.3% [19], and information is scarce [7, 9, 10, 18, 19]. The discrepancy regarding AKD incidence comes from smaller, retrospective observational studies that analysed specific groups at risk for AKI. For instance, Patel included 528 children with AKI from all paediatric units, yet he reported the highest incidence of acute kidney disease (56.3%). However, he found a much lower acute kidney disease incidence in patients undergoing non-kidney organ transplantation (13%). Similarly, the 35.3% acute kidney disease incidence reported by Daraskevicius in children who underwent allogeneic haematopoietic stem cell transplantation seems to be overestimated due to the reduced number of cases (51) [18]. Yet, in some instances, AKI evolution is strongly influenced by the initial AKI episode management, thus resulting in a reduced number of patients progressing to acute kidney disease [10]. Our results (29.1%) are comparable to a previous study that included patients aged from 1 month to 18 years old, with an overall incidence of acute kidney disease of 42.3% [9].

Pre-renal causes were the leading ones in all three groups. Decreased renal perfusion was the main risk factor for developing transient AKI. Rapid reversal of AKI was dependent on its cause. Unlike adults, children have a higher renal reserve in the absence of chronic diseases (i.e.: diabetes mellitus, arterial hypertension, cardiovascular disease). In the absence of early recovery, moderate to severe AKI incidence increases, mirroring the severity of the underlying disease. Prolonged pre-renal injury can lead to intrinsic damage because of decreased renal perfusion, hypoxia and high levels of pro–inflammatory cytokines [20]. In our persistent AKI group, the presence of an intrinsic cause of AKI increased the risk of progressing to acute kidney disease. In this group the incidence of inflammatory conditions (sepsis, septic shock, systemic inflammatory response syndrome, and hypoxia) is higher and reversal takes longer as compared to hypovolemic states. Our results are consistent with previous studies showing that, regardless of AKI duration or AKI severity, children with AKI have poor outcomes [7, 8].

Similar to published data, our study also identified biochemical risk factors associated with progression of AKI to acute kidney disease, these include lower haemoglobin, lower platelet counts, lower serum proteins and higher inflammatory markers as well as higher urea and higher baseline and maximum SCr levels [18, 19]. While transient AKI is considered a “functional” AKI with rapid reversal, persistent kidney injury over time may generate renal remodelling (i.e. tubulointerstitial inflammation and fibrosis) [20]. Children with prolonged AKI are often exposed to nephrotoxic medication that further contributes to the continuum of renal injury [20]. Nevertheless, progression to acute kidney disease is multifactorial. In acute kidney disease, there is an obvious shift from pre-renal towards intrinsic damage and intrinsic causes of AKI modulate AKI duration [20].

In this retrospective cohort study, we observed that patients who progress from AKI to acute kidney disease are at high risk of developing CKD. Several studies involving adults underline the high risk of progression to CKD after acute kidney disease, ranging from 14.7% to 37.4% [21,22,23,24]. Our results showed a lower incidence of new-onset CKD following an AKI episode (4.05%) when compared to Patel (8%-19%) [7, 19]. Nevertheless, patients in the acute kidney disease group had a fivefold increased risk of progressing to CKD, similar to results from the most extensive meta–analysis in adults [24] and to Patel’s results [19]. Interestingly, we found a high incidence of acute kidney disease in patients with pre-existing CKD (40.9%). Thus, CKD increases the susceptibility to acute kidney disease. We also highlighted that subsequent AKI episodes did not influence the risk of new-onset CKD. We did not evaluate the impact of other risk factors such as number of days of nephrotoxic medications, post-AKI proteinuria or chronic underlying diseases (even though in paediatric populations the incidence of chronic diseases is very low).

Currently, data regarding CKD development in paediatric AKI survivors suggest that certain conditions confer a greater risk for new-onset CKD. For instance, the Translational Research Investigating Biomarker Endpoints in AKI (TRIBE-AKI) study showed that perioperative AKI did not increase the risk of CKD 5 years after paediatric cardiac surgery [25]. In addition, the Follow-Up Renal Assessment of Injury Long-Term After Acute Kidney Injury (FRAIL-AKI) study did not identify cardiopulmonary bypass AKI as a risk factor of progression to CKD even though several novel urinary biomarkers remained elevated up to 7 years after an AKI episode [26]. On the other hand, Madsen linked perioperative AKI with a 3.8-fold higher risk of CKD in patients developing AKI after cardiac surgery over a follow-up of almost 5 years [27]. It is not possible to affirm that AKI causes CKD in children, yet certain factors seem to increase the risk. Unlike adults, where various AKI-associated factors can increase the risk of CKD (such as AKI severity, number of AKI episodes) [28], in our paediatric cohort only acute kidney disease and intrinsic AKI cause increased the risk of CKD. Similar results were published by Patel [7, 19].

Both AKI and acute kidney disease were associated with mortality among hospitalised patients [9, 15, 17]. AKI mortality was previously reported by Meena to be around 11%, similar to our results (13.1%) [16]. The reported acute kidney disease mortality in paediatric patients ranges from 10% to 31.8% [9, 10, 18]. Mortality increases with severity of kidney injury. Similar to previous studies, acute kidney disease shows the highest mortality rates [9,10,11, 18, 19]. Deng reported a 10.2% mortality rate in patients with acute kidney disease [8], similar to our results, yet much lower than the 37.7% and the 31.8% mortality rates reported by Patel and LoBasso, respectively [10, 19]. However, when compared to the literature, our transient AKI group showed unexpectedly high mortality rates, 14.66% compared to the 8.9% rate reported mortality by Patel [10]. Although transient AKI can be rapidly reversed, the associated hemodynamic changes observed in the acute state of pre-renal AKI seem to increase the mortality risk along with other associated comorbidities. In addition, our cohort comprised a mixed paediatric population, including neonates. However, it is fair to assume that some patients may have been included in the transient AKI group at the time, but they may have actually had a prolonged AKI evolution before admission.

This study aimed to assess if AKI–acute kidney disease–CKD is a continuum of time-dependent renal injury associated with high morbidity and mortality. In summary, we found that new onset CKD was not influenced by AKI severity or subsequent AKI episodes but by AKI cause and duration. Acute kidney disease is an independent risk factor for CKD. While rapidly resolving AKI has worse survival rates compared to patients with prolonged AKI, progression to CKD is lower. The high mortality rates prove that these patients do not die of AKI but rather, with AKI complicating the course of the underlying disease.

The main limitation of our study is the single centre and retrospective nature. In addition, the lack of urine output information is a major limitation in AKI diagnosing and staging. Another limitation is represented by the lack of data on exposure to nephrotoxins, post-AKI proteinuria, chronic diseases etc., to be considered in the analysis of the risk of developing new-onset CKD. In addition, the lack of information on paediatric intensive care unit (PICU) admission and/or length of stay may limit the strength of our results. The absence of baseline SCr in patients with contrast-associated-AKI increases the difficulty of correctly staging AKI. Due to the retrospective nature of the study, it is possible that some of the patients in the transient AKI group should have been assigned to the persistent AKI or acute kidney disease groups at inclusion. Future prospective studies need to address this issue and shed more light on the impact of AKI duration on mortality in the paediatric population. The strengths of our study include the large number of patients from all paediatric age groups, allowing to have new insights on AKI duration, and acute kidney disease development and progression to CKD following an AKI episode. Prospective studies with standardised follow-up protocols are needed in the paediatric field in order to validate our findings and overcome their limitations.

This study underlines that the continuum of kidney injury should be addressed by including the duration of AKI along with the function-based stages and quasi-anatomic nomenclature. Physicians should reconsider the cause and disease management plans when kidney injury lasts longer than 48 h.