AKI cases constituted over a third (126, 37.7%) of all patients followed up by the nephrology service. This is comparable to findings from a Ghanaian study in which AKI made up 31% of renal admissions over two years [29]. In contrast Obiagwu et al. in Kano Nigeria reported a prevalence of only 13.9% using KDIGO 2012 criteria [30]. The cohort from Kano consisted of only clinic attendees. As highlighted by Olowu et al many African AKI cohorts report a high loss to follow-up rate so this may explain the lower prevalence reported in Kano as not all the AKI patients may have returned for a clinic review [9]. A study from northern Iran reported an even lower prevalence of only 7.3% but this study was conducted in 2004 before consensus AKI criteria were widely adopted [31].
The male-to-female ratio of enrolled patients was 1.4. Similar findings of male predominance were noted in other AKI data derived from Africa [7, 9, 15]. This pattern may be due to congenital anomalies of the kidney and urinary tract (CAKUT) which tend to occur more frequently in boys.
AKI patients had a median age of 7.5 (0.3-16) years and only three infants were included in this cohort. In contrast, studies conducted in Malawian, Ugandan and Nigerian hospitalised children with AKI, recruited significantly younger subjects with mean ages of 4 years, 1.7 years and 4.8 years respectively [15, 19, 32]. Additionally, the present study showed no significant age difference between patients with MAKI and those without MAKI (8.2 versus 8.0, p=0.82). However, MAKI patients in the cohort were more likely to be older than 5 years of age. In contrast, Ugandan children with MAKI were much younger (1.7+/- 1.1 years) [17, 19]. Historical reports of renal dysfunction in malaria seem to suggest that MAKI occurs more commonly in older children and non-immune adults from areas classified as zones of low malaria transmission [21, 33].
Most patients (107, 85%) had severe AKI at admission. This is consistent with findings from earlier studies [7, 9]. One possible explanation could be the time it takes for patients in many low-resource regions to get access to healthcare facilities [9]. Another reason could be that most of the children in this cohort had community-acquired acute kidney injury (CA-AKI) which is associated with a greater risk of severe AKI in some studies [34]. Also, this finding may be a mere reflection of selection bias resulting from the fact that the nephrology service is more likely to be consulted on children on the more severe end of the AKI spectrum. Data on the epidemiology of AKI among all hospitalized children is needed to better characterise this entity.
In contrast, a meta-analysis of worldwide pooled AKI data showed that only 11% of AKI patients had severe AKI [5]. The meta-analysis only included one study from Africa and the majority of patients in the meta-analysis had hospital-acquired acute kidney injury (HA- AKI). Studies have linked severe AKI to worse outcomes and increased risk of persistently abnormal creatinine [18, 35, 36]. It is thus important to identify strategies and interventions that can improve earlier diagnosis of AKI so that children are not put at increased risk of developing ESRD, which has a dire prognosis for many in low-resource regions.
Most (88, 70%) of the referring facilities were able to recognise AKI before sending the children to our hospital. It is known that AKI is often missed in hospitalised patients until it becomes severe or clinically apparent [7, 10, 37]. This may result in the late referral of patients and consequently may account for the high prevalence of severe AKI seen in many African cohorts [9]. Almost all (72, 93.5 %) of the MAKI patients had received an anti-malarial drug before referral. This suggests that factors other than delayed access to anti-malarial medications may be responsible for the development of MAKI. There is need to improve the quality of epidemiological data on AKI from primary healthcare facilities to better characterise the predictors of MAKI development in children [38]. In addition, there is a need to train frontline workers on renal monitoring and initial non-dialytic management of children with suspected AKI especially the optimization of perfusion and avoidance of nephrotoxins [7, 10, 39-42]
There was a marked seasonal distribution in the number of AKI cases with most of the cases concentrated in the first half of the year which is the rainy season. Cases begin to rise in the first quarter of the year and then peak in the second quarter of the year. This seasonal variation in AKI may be attributable to seasonal changes in malaria cases which was the commonest aetiology of AKI in this cohort. It is known that there are variations in both sporozoite rates in the mosquito vector and in human infection rates that are related to seasonal changes in climatic conditions such as temperature and rainfall [43-46]. The second quarter peak of AKI cases was not observed in the years 2020 and 2021. This coincides with COVID-19 waves during the cool dry season in Zambia which were characterised by travel restrictions and which could have influenced referral patterns and consequently the number of AKI cases seen in that period.
MAKI was the predominant aetiology identified. This contrasts with findings by Olowu et al. which reported the leading aetiology of AKI in African children as septicaemia (22.5%) and showed that malaria accounted for only 12% of cases [9]. The findings from the current study suggest that in some regions of Africa, malaria is still the predominant aetiology of AKI contrary to what has been reported elsewhere [6,11,17,19]. Although MAKI was identified as the leading cause of AKI in this cohort, no single test can confirm that AKI was solely due to malaria. No biopsies were performed and access to autoimmune screen tests was limited. In malaria-endemic areas some of these cases may merely represent asymptomatic malaria infection in children with AKI caused by different aetiologies.
In recent times there has been a noted increase in reports of AKI in children with malaria [11, 19, 21]. This may be the result of improved detection due to the adoption of newer AKI consensus guidelines [47]. In contrast, studies from West Africa in Ghanaian and Nigerian children report a very low prevalence of MAKI among children with malaria [27, 28]. This discrepancy in the importance of MAKI in different populations may be due to differences in biological or socio-economic factors [20, 21, 24]. In particular, differences in malaria transmission intensity may influence the pattern of disease manifestation including the incidence of MAKI [20, 21, 48]. It is known that change in malaria infection intensity as a result of malaria control programmes may lead to the occurrence of more severe disease in older people because the population is left with a higher proportion of non-immune individuals [20, 48,49]
The MAKI patients in the present cohort were largely referred from medium transmission zones. Anecdotal information from the second paediatric dialysis centre in the northern region (Figure 3:Arthur Davison Children’s Hospital) shows that most MAKI cases are from moderate transmission areas on the Copperbelt province with very few children referred for MAKI from high malaria transmission zones like Luapula, Muchinga and Northern provinces in the far north. Paediatricians in Luapula, Northern, Muchinga and Western provinces corroborate this observation. There is a need for improved epidemiological mapping of AKI. Additionally, this data acts as a reminder to health workers that as the war on malaria is in the process of being completely won, malaria control programmes may leave sections of the population at increased risk of severe disease and altered manifestations of malaria. This calls for adjustments to control programmes so as to expand their target populations to these newer vulnerable sub-populations [48].
Predictors of MAKI were presence of fever, absence of hypertension at admission, and a higher admission serum creatinine (Table 5). Malaria often causes a catabolic type of AKI and many of the patients present with lower blood pressure as a result of dehydration caused by poor fluid intake and vomiting or due to systemic inflammatory response [21].
Patients were more likely to have MAKI if they had a low Glasgow coma scale at presentation (OR 2.2, p 0.5). Children with severe forms of malaria such as cerebral malaria are known to be at higher risk of AKI [50]. Some of the postulated pathophysiology of MAKI may be similar to those described in cerebral malaria [26]. Plasmodium falciparum produces a molecule, Plasmodium falciparum erythrocyte membrane protein 1(PfEMP1), which is expressed on the surface of infected red cells. This molecule acts as a ligand for the adhesion of the infected red cells to endothelial cells in a process known as sequestration. The resulting endothelial activation leads to expression of more endothelial adhesion molecules and consequently results into compromised renal perfusion [21, 24, 25]. The identification of immunological biomarkers may help to better predict patients at increased risk of MAKI and facilitate earlier interventions [3].
Two- thirds of the AKI patients in the cohort required dialysis. This is similar to findings by Olowu et al. where 66% of the children with AKI needed dialysis [9]. In contrast Susantitaphong et al in their meta-analysis of worldwide data found a dialysis need of merely 11%. The two reviews differ in that the former had a population dominated by CA-AKI whereas the latter had children that chiefly had HA-AKI who were predominantly from HIC.
The dialysis access rate for this cohort was 100% even though close to half of the children utilized improvised PD catheters. Olowu et al cited reports of similar dialysis access rates from Sudan and South Africa [9]. Similar to Zambia, both these countries have state-funded dialysis programmes. In contrast, the pooled data showed an overall dialysis access rate of only 45 % with adults having an access rate of only 49.1 %, and children having access of 64 % [9]. Given that only patients with overt renal dysfunction are likely to have been identified at referring health facilities it is plausible that the need for dialysis in our population may be higher than reported.
The predictors of the need for dialysis in this cohort were the presence of anuria or oliguria at admission. Due to the small number of patients that had a record of admission serum sodium, potassium, and bicarbonate, these variables were not included in the derivation of the logistic regression model. A multicentre study in Spanish PICU patients found that thrombocytopenia and serum creatinine could predict need for dialysis at discharge from PICU [51]. All the patients in the present cohort had severe AKI thus predictive rules such as the renal angina score would have been unsuitable. More prospective studies that will recruit patients at risk of AKI are needed so as to better characterise the predictors of severe AKI development and the need for renal replacement therapy [3].
In children particularly in low-income countries, the main stay of RRT is PD and this was reflected in our cohort where 88% (82) of all children receiving dialysis underwent this modality. Pooled data from Africa showed a similar PD rate of 80% [9, 16]. Close to half (43.5%) of the catheters used were improvised PD catheters. Others have described the use of improvised PD catheters and fluids [16, 52]. PD is assumed to be cheaper than HD but it is still beyond the reach of many in LICs. Much still needs to be done to secure supplies of child-appropriate dialysis consumables in our population [10, 53].
The PD catheter use rate was 1.8 (147 catheters/ 82 patients) catheters per AKI episode. The median duration of dialysis need was 11 days (0-408) while the mean lifespan of PD catheters was 6.6+/- 6.4 days for improvised catheters and 23.7+/-46.6 days for Tenchoff catheters (p=0.004). This discrepancy in the life span of the two catheter types, in the context of the prolonged period of dialysis need, explains the high catheter use rate. Because of the mean duration of dialysis need, use of Tenchoff catheters for AKI patients is advocated.
Even though the overall complication rate was not significantly different between Tenchoff catheters and improvised catheters, there was a difference in the pattern of complications (Table 6). The overall peritonitis rate was 34.7%. This is lower than rates reported among patients undergoing PD for AKI in South Africa (47.5% and 41%) but higher than rates reported in DRC [16, 54, 55]. Surprisingly the peritonitis rate was higher in association to Tenchoff catheters as compared to improvised catheters (41% versus 26.7%, p=0.07). There is a need to improve infection control measures.
At discharge from the renal ward, a third (27%) of children had a poor outcome (absconded, CKD, death, ESRD or were discharged for palliation). There was no difference in mortality between patients with MAKI and those without MAKI (5.2% versus 6.1%, p 0.93). This is similar to findings from a study conducted among Ugandan children, which found that patients with malarial febrile illness had a mortality of (no AKI, 0.8% versus MAKI, 4.1%, p = 0.005) compared to children who had AKI caused by non-malaria febrile illness (no AKI, 4.0% vs. AKI, 4.6%, p = 0.81) [19]. In contrast, Namazzi et al, in another Ugandan study, reported a higher mortality rate of 26.5% among children with stage 3 AKI [18].
The presence of fever, being less than 5 years old and having a higher white cell count were protective from a poor outcome (Table 8). The odds of patients having a poor outcome were increased for children with MAKI but this was not statistically significant (OR 6.4, p 0.51). This poor sensitivity of various clinical variables to predict MAKI may be because all the patients in the cohort had already developed severe AKI by the time they presented and predictive models would be more sensitive if applied in populations with early kidney dysfunction.
One year post-presentation with AKI the mortality rate had risen from 5.6% at the time of discharge to 9.5%. This is comparable to a mortality rate of 13.8% that was observed in a meta-analysis of pooled worldwide childhood AKI data [5]. In contrast Olowu et al reported a pooled mortality among 1942 children from Africa of 34% [9]. Mortality rates from the present study may be inaccurate due to the very high rate of loss to follow-up of 77% (97/126 patients), an observation that was also made by Olowu et al [9]. One major factor accounting for this high attrition rate is limited economic resources to permit families to stay out the full course of the treatment. Most patients come from agrarian areas and their families can ill afford to be away from work for extended periods of time. There is need to expand PD facilities to district hospitals which are located in closer proximity to patients especially since recent government policy has been to ensure that at least two medical doctors are available to man such facilities. There is also an urgent requirement to strengthen post-AKI follow-up because children who have suffered severe AKI are at risk of developing CKD [3]. One way this could be done is to empower health personnel at the district level with the necessary skills to monitor and follow up patients with renal dysfunction.
A fifth of children (19.8%) children were diagnosed with CKD or ESRD at discharge from the renal ward. This is higher than the pooled rate of 10% reported by Olowu et al [9]. There was no difference in outcome between patients with MAKI and those without MAKI (15.6% versus 20.4%, p=0.35). Namazzi et al reported that 15.6 % of Ugandan children still had persistently elevated creatinine a month after suffering from MAKI (Acute kidney disease (AKD)) [18]. They did not report CKD rates. The CKD rate could have been distorted by the fact that there was no access to the pre-morbid and baseline serum creatinine results of the patients. Thus pre-existing renal dysfunction could not be determined [9, 11]. In this cohort 5 children were reported to have reduced cortico-medullary differentiation at admission but this was not predictive of a poor patient outcome on logistic regression (OR 2.23, p 0.65). One year post diagnosis, only four out of the 20 CKD patients were still being followed up. Given this very high loss to follow-up rate, it is possible that the actual number of children with poor outcomes may, unfortunately, be higher.
This study has several limitations. Firstly, this is a retrospective study therefore not all patient data was available on the folders. Secondly, most of the data on sepsis-associated AKI was not collated because sepsis patients are not followed up by the nephrology service post PICU and so their folders were unavailable. Thirdly, long-term outcomes for a large proportion of the cohort are unknown because they were lost to follow-up. Lastly, there is no single test that can specifically diagnose MAKI thus it is possible that MAKI could have been over-estimated. In malaria endemic areas a positive malaria test in the context of AKI may merely represent asymptomatic malaria in a patient with AKI from a different aetiology.