A reappraisal of risk factors for hypertension after pediatric acute kidney injury

Abstract

Background:

Acute kidney injury (AKI) is common in hospitalized children and increases the risk of chronic kidney disease (CKD) and hypertension, but little is known about the patient level risk factors for pediatric hypertension after AKI. The aims of this study are to evaluate the prevalence and risk factors for new onset hypertension in hospitalized children with AKI and to better understand the role of acute kidney disease (AKD) in the development of hypertension.

Methods:

This study was an observational cohort of all children ≤ 18-year-old admitted to a single tertiary care children’s hospital from 2015-2019 with a diagnosis of AKI. Hypertension was defined as blood pressure >95^th percentile for sex, age, height, diagnosis of hypertension on the problem list, or prescription of antihypertensive medication for > 90 days after AKI.

Results:

410 children were included in the cohort. 78 (19%) developed hypertension > 90 days after AKI. A multivariable logistic regression model identified AKD, need for kidney replacement therapy, congenital heart disease, and non-kidney solid organ transplantation as risk factors for hypertension after AKI.

Conclusions:

Incident hypertension after 3 months is common among hospitalized children with AKI, and AKD, need for dialysis, congenital heart disease, and non-kidney solid organ transplant are significant risk factors for hypertension after AKI. Monitoring for hypertension development in these high-risk children is critical to mitigate long term adverse kidney and cardiovascular outcomes.

Graphical Abstract

Introduction

Pediatric hypertension is quite common, and the incidence and prevalence have been increasing. The global prevalence of hypertension is estimated to be approximately 4% among people aged 19 years or younger and age specific prevalence of childhood onset hypertension has increased about 1.8 fold in both young children and adolescents between the year 2000 and 2015 [1]. Hypertension increases risk of cardiovascular disease in adulthood which is a leading cause of mortality [2]. There are numerous risk factors for pediatric hypertension and acute kidney injury (AKI) has been identified as one [3].

AKI is common in hospitalized children, affecting 5-9% of all hospitalized children, 10-35% of critically ill children, and 27-30% of critically ill neonates [4-7]. Children with underlying conditions have been found to be at high risk for AKI with rates ranging from 10-53% in children who have undergone cardiac surgery, children with sepsis and post infection complications, and those with a history of malignancy or hematopoietic stem cell transplant (HSCT) who have been exposed to nephrotoxins [8]. Previously, most AKI episodes were thought to be a self-limited, but we now know that AKI can result in longer term kidney dysfunction including acute kidney disease (AKD), chronic kidney disease (CKD), kidney failure, proteinuria, hypertension, and death [3, 9]. Studies in children have shown variable prevalence and risk factors for these adverse kidney outcomes largely due to substantial differences in sample sizes, outcome definitions, and previous lack of consensus AKI and AKD definitions [3, 8]. Few large studies have concurrently evaluated multiple patient level risk factors for the development of hypertension in children following an episode of AKI. No studies have evaluated the effect of persistent kidney injury beyond 7 days, now termed AKD, as a risk factor for hypertension. We have previously shown in a cohort of non-kidney solid organ transplant recipients that AKD, rather than peak stage of AKI, may be a better predictor of chronic kidney disease. This finding that prolonged kidney injury may also increase risk of other long term kidney outcomes was the impetus for assessing AKD as a risk factor for hypertension in the present study [10].

The aims of the present study are to describe the prevalence and patient level risk factors for new onset hypertension in an unselected observation cohort of pediatric AKI survivors at a single, tertiary care pediatric center. We hypothesized that hypertension would occur commonly after AKI in hospitalized children.

Methods

Study Design and Population:

This study was an observational cohort study of all children ≤ 18 who were admitted to all units of our tertiary care children’s hospital between 2015-2019 with a diagnosis of AKI. The cohort was previously described by Patel et al. and is briefly described here [11]. The cohort was identified using ICD-9 and ICD-10 codes for AKI. AKI diagnosis and stage was confirmed by a single pediatric nephrologist. For children with more than one AKI event in the study period we selected the first AKI event as their date of study enrollment. We excluded children who were only treated in the outpatient setting, those with less than two creatinine values between day 8 and 90 after AKI as this is the minimum number of values needed to evaluate for AKD development, those with a pre-existing diagnosis of hypertension, those who were dependent on chronic dialysis at the time of study enrollment and previous kidney transplant recipients. Pre-existing hypertension was defined as two outpatient blood pressure readings > 95^th percentile for age, sex, and height, diagnosis of hypertension on the child’s problem list, or an active prescription for an antihypertensive medication at the time of study enrollment. A flowchart of patient inclusion and exclusion is presented in Figure 1. Patients were followed for a minimum of 2 years and up to 7 years after AKI or until the patient’s death. The study was approved by the Institutional Review Board at Duke University and followed the tenets of the Declaration of Helsinki.

Fig. 1.

Flowchart of children included in the study

Outcomes:

The primary outcome was prevalence of new onset hypertension > 90 days after AKI. Hypertension was defined as two blood pressure readings > 95^th percentile for age, sex, and height, new diagnosis of hypertension on problem list, or new prescription for antihypertensive medication. Blood pressure data was gathered from the child’s outpatient medical record in our hospital electronic medical record. For this reason, blood pressure measurements were taken in a variety of general pediatric and subspecialty clinics in our hospital system. Data regarding measurement modality (oscillometric vs manual) was not recorded in the medical record but conventionally most outpatient clinics in our hospital system obtain blood pressure by oscillometric measurement. Information regarding blood pressure cuff size was not available. If multiple measurements were taken on the same day the values were averaged. Children with only one blood pressure measurement >90 days after AKI were categorized as normotensive if they also did not have a diagnosis of hypertension on their problem list and/or a prescription for an antihypertensive medication.

Covariate Definitions:

Participants were assessed for the covariates outlined in Supplementary Table 1. Covariates included demographic information, comorbid conditions, peak AKI stage, AKI duration, need for kidney replacement therapy in the first 7 days of AKI, and AKD development. AKI duration was defined as number of days creatinine was elevated between days 1-7. Baseline creatinine was defined as lowest serum creatinine in the 6 months prior to AKI and if one was not available then creatine was calculated using the Schwartz equation and an assumed eGFR of 120 ml/min/1.73m^2 for children outside of the neonatal period [12]. For neonates, baseline creatinine was defined as the creatinine nadir after birth. AKI, AKD, and CKD were defined based on KDIGO criteria [13-15]. History of CKD was defined based on KDIGO guidelines and included children with stages 1 – 5 CKD who were not dialysis dependent [15]. Prematurity was classified as gestation age ≤ 36 weeks at birth as this is the gestational age at which nephrogenesis is completed [16]. Gestational age was gathered from the neonatal intensive care unit (NICU) history and physical note for children admitted to the NICU at the time of the study or from the child’s problem list for those children admitted another hospital unit.

Statistical Analysis:

Covariates of interest were utilized to build a logistic regression model to determine risk factors associated with hypertension after AKI. We then used a stepwise variable selection (forward inclusion p < 0.05; backward elimination p > 0.1) of all covariates in the full logistic regression model to build a parsimonious logistic regression model predicting probability of hypertension in children with AKI. We conducted standard assumption diagnostics and report model parameters in odds ratios (OR) with 95% confidence intervals (CI). Statistical significance was set at p-value ≤ 0.05. All statistical analyses were done using Stata 17.0 (Stata Corporation, College Station, TX, USA).

Results

Baseline Characteristics:

528 children were identified for initial inclusion in the cohort as described in the previous study by Patel et al [11]. Additionally, 84 children were excluded due to mortality at < 3 months post AKI and 34 were excluded due to a pre-existing diagnosis of hypertension. No children included in the final analysis were found to have zero blood pressure measurements >90 days after AKI. A total of 410 children were included in the final analysis as outlined in Figure 1. Baseline demographics of the cohort of children who developed hypertension and those who did not are summarized in Table 1. Children who developed hypertension tended to be older and have higher BMI at the time of their AKI diagnosis. A higher proportion of children who developed hypertension were admitted to the pediatric intensive care unit (PICU) or pediatric cardiac intensive care unit (PCICU) at the time of AKI and a lower proportion admitted to the NICU. There was a higher proportion of children of self-reported African American race who developed hypertension after AKI compared to children of other self-reported races.

Table 1.

Baseline demographics of children who developed new diagnosis of hypertension after AKI compared to those who did not develop hypertension

	Hypertension (N = 78)	No hypertension (N = 332)
Sex at birth
Male	41 (52.6%)	176 (53.0%)
Female	37 (47.4%)	156 (47.0%)
Age at AKI diagnosis (months)	23 (1.25–101)	2 (0.23–75)
BMI (kg/m2)	16.5 (14.6–20.2)	14.4 (9.1–17.2)
Hospital location at AKI diagnosis
General Floor	19 (24.4%)	102 (30.7%)
PICU	24 (30.8%)	63 (19.0%)
PCICU	29 (37.2%)	62 (18.7%)
NICU	6 (7.7%)	105 (31.6%)
Self-reported race/ethnicity
Asian	5 (6.4%)	11 (3.3%)
Black	35 (44.9%)	121 (36.4%)
Hispanic	6 (7.7%)	30 (9.0%)
Mixed	1 (1.3%)	15 (4.5%)
Other	6 (7.7%)	23 (6.9%)
White	25 (32.1%)	132 (39.8%)

Continuous variables reported as median (Q1–Q3) and categorical variables reported as N (%)

Prevalence of Hypertension after AKI:

Figure 2 shows the prevalence of hypertension in our cohort. 78 (19%) of children developed hypertension in the follow up period. Seven of 88 (8%) children with stage 1 AKI, 24 of 150 (16%) children with stage 2 AKI, and 47 of 172 (27.3%) children with stage 3 AKI developed hypertension (p-value < 0.001). There was notably a linear trend with increasing prevalence of new hypertension as AKI stage worsened. Of the 201 children without AKD, 24 (11.9%) developed hypertension and of the 209 children with AKD, 54 (25.8%) developed hypertension after AKI (OR 2.6, 95% CI 1.51-4.35, p-value <0.001). Between day 8-90, 111 children were found to be hypertensive. Of this subgroup 53 (47.7%) went on to have hypertension at >90 days compared to only 20 of 282 children (7.1%) of children who were not hypertensive between day 8-90.

Fig. 2.

A Incidence of hypertension by stage of AKI. b Incidence of hypertension by AKD status

Risk factors for hypertension after AKI:

The multivariable logistic regression model is shown in Table 2 and univariate analysis is shown in Supplemental Tables 2a and 2b. AKD (OR 2.63, 95% CI 1.41-5.03, p-value 0.003), congenital heart disease (OR 2.27, 95% CI 1.17-4.45, p-value 0.015), and non-kidney solid organ transplant (OR 5.48, 95% CI 2.41-12.79, p-value <0.001) were associated with development of hypertension after AKI. A multivariable logistic regression model using forward and backward variable selection is shown in Table 3. This model confirmed the findings of the full multivariable logistic regression model and additionally found that need for kidney replacement therapy in the first 7 days of AKI was associated with new hypertension diagnosis (OR 2.27, 95% CI 1.01-4.95, p-value 0.042. This model had an AUC of 0.778. Interestingly those with a history of prematurity had lower odds of developing hypertension in the follow up period.

Table 2.

Multivariable logistic regression model for new hypertension after AKI

Covariate	Odds ratio	95% CI	p-value
Sex	0.96	0.55–1.68	0.898
BMI	1.01	0.95–1.07	0.776
Age at AKI diagnosis	1.00	0.99–1.01	0.491
Family history of kidney disease	1.21	0.66–2.21	0.543
Prematurity	0.31	0.13–0.72	0.008
Congenital heart disease	2.27	1.17–4.45	0.015
Malignancy	0.74	0.29–1.76	0.510
Previous AKI	1.32	0.53–3.12	0.539
CKD	1.91	0.58–5.67	0.257
Solid organ transplant	5.48	2.41–12.79	<0.001
AKI stage 2 vs stage 1	1.49	0.59–4.15	0.413
AKI stage 3 vs stage 1	2.50	0.98–7.11	0.066
Duration of AKI	1.04	0.89–1.21	0.628
AKD	2.63	1.41–5.03	0.003
Need for kidney replacement therapy	1.84	0.78–4.29	0.159

Table 3.

Forward and backward variable selection multivariable logistic regression model for hypertension after AKI

Covariate	Odds ratio	95% CI	p-value
Prematurity	0.25	0.11–0.49	<0.001
Congenital heart disease	1.89	1.07–3.33	0.027
Solid organ transplant	5.52	2.51–12.32	<0.001
AKD	3.34	1.88–6.14	<0.001
Need for kidney replacement therapy	2.27	1.01–4.95	0.042

Discussion

To the best of our knowledge, this study is the first to assess the patient level risk factors for hypertension as well as AKD as a risk factor for hypertension in an unselected cohort of hospitalized children with AKI. We showed that hypertension is quite common after AKI and that AKD is an independent risk factor for hypertension after AKI irrespective of sociodemographic factors and comorbidities. We also showed that patient level risk factors such as congenital heart disease and solid organ transplant are associated with hypertension after AKI.

We have previously shown that AKD may be a better predictor of CKD development than AKI stage in pediatric non-kidney solid organ transplant recipients [10]. The present study demonstrates that AKD is also a strong predictor of hypertension after AKI. These findings suggests that the duration of kidney injury may be just as important in predicting future risk of adverse kidney outcomes as the peak stage of initial kidney injury. A higher proportion of children in our study who had high blood pressure during day 8-90 also went on to have sustained hypertension at >90 days after AKI suggesting that blood pressure monitoring and management during this period warrants further study. We additionally identified children with a congenital heart disease and non-kidney solid organ transplants as other high risk patient populations for hypertension post AKI which confirms previously reported findings [8]. These children have numerous reasons for persistent kidney injury including underlying physiology, medication exposures, and severity of illness that worsen their risk. By identifying these groups as high risk we can better study these additional risk factors and identify possible modifiable risks and interventions to reduce the risk of hypertension and future cardiovascular disease including nutrition and exercise counseling.

Hypertension was quite common in our study occurring in 19% of children. Other studies have shown variable prevalence of hypertension after AKI. In a large meta-analysis of 10 studies including a total of 346 children, hypertension was reported in 6.6% of children after six years of follow up. This is likely in some part due to significant variability in AKI definition ranging from absolute eGFR or serum creatinine value to percent change in serum creatinine to need for kidney replacement therapy. These 10 studies were also rather small with sample sizes ranging from 17 to 100 [3]. There is variability in reported prevalence and risk factors for hypertension and CKD after AKI in studies of neonates, children undergoing cardiac surgery, critically ill children, and in children with malignancy or HSCT [3, 17]. A study of 200 children with AKD in India found that the prevalence of persistent hypertension (defined as requiring one more or antihypertensive medication) at 90 days was 12%. The prevalence of hypertension at day 8-90 was higher in our study (27%) likely due to our more inclusive definition of hypertension [18]. Interestingly, in our study prematurity was not identified as an independent risk factor for future hypertension development which we anticipate was in part due to collider stratification bias. By conditioning on AKI, we distorted the known association between prematurity and hypertension. There was a total of 136 premature infants included in our study. Within this group 11 (8%) developed hypertension, which is higher than rates reported by other large studies of premature infants and infants admitted to the neonatal intensive care unit [19, 20]. These risk factors remain important when considering a child’s future risk for adverse kidney outcomes in combination with other risk factors. Lastly, the children who did not develop hypertension were quite young compared to those who did develop hypertension. This is likely due to the high proportion of infants in this group compared to the group that did develop hypertension. Measurement and treatment of hypertension in infants is fraught with issues and possibly contributed to the lower absolute number of young children in the hypertension group. This can better be addressed in prospective studies with protocolized follow up and study personnel who are trained to obtain blood pressure measurements on even young infants. Overall variations in outcomes suggest that risk of hypertension after AKI is not predicted by AKI event alone but rather is multifactorial due to characteristics of the kidney injury, patient risk factors, and exposures [19, 21-26]. Modifiable risk factors should be studied in conjunction with non-modifiable, patient level risk factors in at risk populations to better assess how they might compound risk of future hypertension development.

Our study has numerous strengths that set it apart from previous studies. Strengths of our study include strict application of the most current KDIGO definitions of AKI, AKD, and CKD, broad and age-appropriate definition of hypertension, large sample size, and multivariable analysis assessing numerous risk factors simultaneously. Limitations of our study include incomplete data due to retrospective nature of the study. By using ICD-9 and −10 codes to identify initial participants there was likely some selection bias introduced. Studies of using administrative health data have found that administrative codes have poor sensitivity but good specificity for identifying AKI in children but that sensitivity improves 1.5-2 fold with each increase in AKI stage [27]. This suggests that children with AKI stage 1 are least likely to be included in studies utilizing ICD-9 and −10 codes for identifying children for study inclusion which introduces selection bias. Many studies in children on adverse kidney outcomes after AKI do use large databases; use of clinical diagnosis of AKI using creatinine and urine output based definitions would strengthen and improve generalizability of future studies as epidemiologic studies have shown that stage 1 AKI is significantly more common than severe AKI in even critically ill children and young adults [4]. Lastly, our study used non-standardized office blood pressures rather than the gold standard of ambulatory blood pressure monitoring (ABPM) for diagnosis of hypertension. Due to the retrospective nature of the study, we were unable to standardize method by which blood pressure was checked (oscillometric vs manual), technique including having the patient sit in a chair in a quiet environment, using appropriate size blood pressure cuff, or standardized training for the blood pressure checker. Future, prospective studies would be strengthened with the use of a standardized protocol for checking blood pressure or use of ABPM.

In this study, we have shown that hypertension is quite common in pediatric AKI survivors. We have shown that AKD is a significant risk factor for post AKI hypertension in children and have identified children with a history of congenital heart disease and non-kidney solid organ transplantation as high risk for future hypertension. Lastly, we found that children who required dialysis in the first seven days of AKI were at higher risk for future hypertension suggesting that these children would also benefit from intensive of their blood pressure after hospital discharge. By identifying these high-risk groups of children, as well as by identifying AKD as a risk factor for hypertension after AKI, we have set the stage for future studies to explore the mechanistic reasons for hypertension after AKI as well as to better explore the role of repeated injuries, altered physiology, and continued nephrotoxin exposure on risk of hypertension after AKI. Understanding the risk for future kidney disease can also allow for more kidney focused care including more aggressive blood pressure control, prompt detection and management of proteinuria, as well as focused lifestyle counseling to prevent secondary risk factors for hypertension and cardiovascular disease in at risk children.

Data availability:

All data are stored according to the Duke University Institutional Review Board approval and are available upon request.