Blood Pressure Control in Adolescents With CKD and Risk of Kidney Failure in Young Adulthood Authors: Dorey A. Glenn, Morgan Botdorf, Andrea Allen, Kimberley Dickinson, Kaleigh Wieand, Amy Goodwin Davies, Victoria Besser, Bradley P. Dixon, Caroline Gluck, John Hammond, Lyndsay Harshman, Mahmoud Kallash, Chryso Katsoufis, Sara Kibrom, Alexander Kula, Mark Mitsnefes, Yosuke Miyashita, Zubin Modi, Larry Patterson, Michelle R. Denburg, Christopher B. Forrest, Jarcy Zee Published: 2026 ABSTRACT Rationale & ObjectiveUnderstanding risk factors for chronic kidney disease (CKD) progression during adolescence is essential to improve outcomes for these individuals as adults. The objective of the present study was to retrospectively analyze electronic health records of children with CKD to understand the effect of cumulative systolic blood pressure (SBP) load during adolescence on time to kidney replacement therapy (KRT) or death during young adulthood.Study DesignRetrospective cohort study.Setting & ParticipantsAdolescents with CKD were enrolled from 14 academic medical centers in the Preserving Kidney Function in Children with Chronic Kidney Disease (PRESERVE) study. Individuals aged between 1 and less than 18 years with 2 or more estimated glomerular filtration rate measurements between 30 and <90 mL/min/1.73 m2 separated by ≥90 days without an intervening estimated glomerular filtration rate ≥90 mL/min/1.73 m2 were included.ExposureCumulative SBP load was defined using area under and above the SBP curve (ie, time and magnitude) and time-only approaches using the 50th, 75th, and 90th SBP percentiles.OutcomesTime to KRT (long-term dialysis initiation or kidney transplantation) or death were ascertained via linkage with the US Renal Data System.Analytical ApproachCox proportional hazards models were used to investigate the relationship between blood pressure (BP) control and the composite of KRT or death.ResultsThe cohort included 2,585 individuals with a median follow-up of 7.45 years (IQR, 6.05-9.20), among whom 4.6% (n = 118) met the KRT or death outcome between ages 18 and 30 years. In an adjusted Cox model, each unit increase (percentile point × time) in cumulative SBP load >90th percentile was associated with 1.36 (95% CI, 1.17-1.58) times higher hazard of KRT or death at age 18-19 years. Control of SBP to <90th percentile for 25%, 50%, and 100% of time between ages 14 and 18 years was associated with a 47%, 72%, and 92% risk reduction of KRT or death at age 18-19 years compared with no SBP control.LimitationsMisclassification of BP control related to white coat or masked hypertension. Adherence to prescribed antihypertensive medications was not assessed.ConclusionsWorse SBP control during adolescence was associated with a markedly increased risk of kidney failure in young adulthood. Cumulative SBP load derived from electronic health record data can inform risk of adverse long-term kidney outcomes. Chronic kidney disease (CKD) is characterized by irreversible loss of kidney function resulting in excess morbidity, mortality, and health care expenditure.1 Advanced CKD (stages 3-5) occurs in 0.36% of individuals aged 12-17 years in the United States and is most commonly the result of congenital anomalies of the kidney and urinary tract and glomerular disease.2 Alterations in growth, health development, and cognition make CKD management during adolescence challenging for patients, caregivers, and health care providers.3 Understanding risk factors for CKD progression in the adolescent age group is essential to improve outcomes for these individuals as adults.4 CKD is commonly accompanied by elevated blood pressure (BP).5,6 However, the relationship between cumulative BP load during adolescence and subsequent progression to kidney failure has not been well studied.7,8 Although treatment of hypertension can slow loss of kidney function and reduce cardiovascular risk in children with CKD,9, 10, 11 there is a lack of real-world data to inform target BP thresholds and characterize short- and long-term risk. The objective of the current study was to retrospectively analyze electronic health records (EHRs) of children with CKD from 14 academic medical centers in the Preserving Kidney Function in Children with Chronic Kidney Disease (PRESERVE) study to understand the effect of cumulative systolic blood pressure (SBP) load during adolescence on time to kidney replacement therapy or death during young adulthood. Methods Study Design and Data Source Using National Patient-Centered Clinical Research Network (PCORnet) infrastructure, we conducted a multicenter EHR-based retrospective cohort study to examine the association between SBP control in adolescents with CKD (aged 14 to less than 18 years) and risk of KRT or death in young adulthood (ages 18-30 years) (Fig 1). EHR data were analyzed from 14 academic medical institutions within 5 participating clinical research networks that were part of the PRESERVE study12 (Item S1).Figure 1Study design. Abbreviations: CKD, chronic kidney disease; KRT, kidney replacement therapy. Study design. Abbreviations: CKD, chronic kidney disease; KRT, kidney replacement therapy. EHR data from January 1, 2009 to December 31, 2020 were harmonized to the PCORnet Common Data Model (v6.0), extracted in February 2024, and linked with the United States Renal Data System (USRDS) registry2 to obtain information on outcome events (kidney transplant, dialysis initiation, or death). Data from the EHRs included patient diagnoses, clinical measurements, procedures, and vital signs from health care encounters. Cohort Formation A detailed description of participant inclusion for the PRESERVE study has been previously reported by Denburg et al,12 with selection criteria outlined in Table S1. Cohort formation for the present analysis is presented in Fig S1. Individuals between the ages of 1 and less than 18 years with 2 or more estimated glomerular filtration rate (eGFR) measurements between 30 and <90 mL/min/1.73 m2 separated by ≥90 days without an intervening eGFR value ≥90 mL/min/1.73 m2 were included. They were also required to have 3 or more valid SBP measurements over 2 years at different ages during the baseline BP assessment window (ie, ages 14, 15, and 16; 15, 16, and 17; 14, 15, and 17; or 14, 16, and 17 years). BP measurements were obtained exclusively from outpatient visits, and participants were excluded if they met any of the following criteria: (1) SBP ≥300 mm Hg; (2) diastolic BP ≥160 mm Hg; (3) SBP less than diastolic BP within a paired measurement; or (4) any BP measurement obtained on the day before, the day of, or the day after anesthesia exposure. If an individual had multiple measurements recorded on the same day, the mean value was used. Individuals with history of long-term dialysis or kidney transplantation before age 18 were excluded. Variable Definitions Exposure: BP Control As established methods for assessing BP control using observational cohort data are lacking, our primary analysis used a measure representing cumulative BP load13,14 to estimate each participant’s level of SBP control between ages 14 and less than 18 years (Fig 2). To calculate age-, sex-, and height-specific SBP percentile values, we compared our observed SBP values to normative BP data from the 2017 Clinical Practice Guideline for Screening and Management of High Blood Pressure in Children and Adolescents.15 Next, a cumulative SBP load >90th SBP percentile was calculated using an area under the curve (AUC) approach accounting for both the duration and magnitude of SBP elevation. Linear interpolation between observed SBP measurements was used to create a continuous SBP curve, and the area under the SBP curve >90th percentile for age, sex, and height was calculated. The AUC was normalized for the duration of each participant’s baseline BP assessment window to account for varying observation intervals. A higher area value, with units of percentile point × time (pp×time), indicates poorer SBP control, reflecting a longer duration and/or greater degree of elevated BP than the SBP percentile threshold.Figure 2Illustration of the systolic blood pressure (SBP) assessment methodology for a sample participant. The red areas represent cumulative SBP load >90th percentile, and the green areas represent cumulative SBP load <90th percentile. The percentage of time (red/green bar) above the target SBP is defined as the total number of days with SBP above the threshold (days with SBP in the red shaded area) divided by the total number of days included in the SBP observation period. Illustration of the systolic blood pressure (SBP) assessment methodology for a sample participant. The red areas represent cumulative SBP load >90th percentile, and the green areas represent cumulative SBP load <90th percentile. The percentage of time (red/green bar) above the target SBP is defined as the total number of days with SBP above the threshold (days with SBP in the red shaded area) divided by the total number of days included in the SBP observation period. For descriptive analyses, participants with an AUC of 0 were defined as having well-controlled SBP. Moderate and poorly controlled BP groups were defined among the remaining participants using the median AUC value as a cut point (ie, less than median AUC considered moderately controlled SBP and greater than or equal to median AUC considered poorly controlled SBP). Secondary measures of cumulative SBP load above the 75th and 50th percentiles were similarly calculated and dichotomized into BP control groups at the median time. As an additional secondary exposure, we also calculated a duration-based measure of BP control. This was defined as the proportion of time between ages 14 and less than 18 years SBP was ≥90th percentile.15 Duration-based measures were also calculated using the 75th and 50th SBP percentiles and dichotomized into BP control groups at the median time. To quantify the potential benefit of SBP control below the target SBP percentile threshold, we defined a third exposure metric, the cumulative SBP load <90th percentile. This was calculated as the area above the SBP curve (AAC) and below the 90th percentile, normalized to the duration of the baseline BP assessment window. A higher AAC indicates better BP control. Similar AAC measures were calculated using the 75th and 50th SBP percentiles. Model Covariates Model covariates were assessed before or at the time of a participant’s first BP measurement during the baseline BP assessment window. Demographic variables extracted from the EHR included sex assigned at birth and race/ethnicity. If a participant identified as Hispanic, their race/ethnicity was classified as Hispanic. For non-Hispanic participants, race was categorized as Black/African American, White, Asian/Pacific Islander, or Other. Insurance status at the last visit before the beginning of the baseline BP assessment window was classified using payer type codes from the PCORnet Common Data Model (commercial/self-pay, Medicaid/ State Children’s Health Insurance Program, and other). Antihypertensive medication prescription, proteinuria, CKD etiology, and eGFR were assessed using EHR data from the 12 months before the first BP measurement in the baseline BP assessment window. When multiple eGFR measurements were available, the most recent measurement to the start of the BP assessment window was used. Antihypertensive medication exposure was determined by a prescription with start date within the 12-month lookback window for an angiotensin-converting enzyme inhibitor, angiotensin receptor blocker, β-blocker, or calcium channel blocker. Proteinuria was captured using urine dipstick measurement, because urinary protein-creatinine ratios were not available because of data quality limitations. Urine dipstick values of 2+, 3+, or 4+ were categorized as positive while values of 1+, trace, or missing were treated as negative, with 1+ considered negative to minimize false positives from concentrated urine specimens. CKD etiology was categorized hierarchically as malignancy-related, glomerular, or nonglomerular disease/non–malignancy-related. Malignancy-related CKD was defined by the presence of chemotherapeutic medications or a cancer diagnosis code, regardless of timing. Glomerular disease was defined, in the absence of any malignancy codes, as the presence of at least 2 glomerular disease diagnosis codes occurring on different dates or a glomerular disease diagnosis code and a nontransplant-related kidney biopsy procedure code. All remaining cases of CKD were classified as nonglomerular/nonmalignancy. eGFR was calculated using the Chronic Kidney Disease in Children Study “under 25” (U25) equation based on serum creatinine, age, sex, and height.16 Health care utilization was quantified using the total number of BP measurements in the baseline BP assessment window and the number of hospitalizations lasting >1 day duration in the 12 months before the first BP measurement in the baseline BP assessment window. Nonkidney-related medical complexity was assessed over a 3-year period before the start of the BP assessment period using the Pediatric Medical Complexity Algorithm.17 This approach classifies International Classification of Diseases, Tenth Revision, Clinical Modification codes into 18 different body systems and categorizes individuals into either complex chronic disease, noncomplex chronic disease, or no chronic disease. Outcome: KRT or Death Outcome follow-up began at each participant’s 18th birthday and ended at KRT event, death, time of the last health care encounter, or on December 31, 2020, whichever came first. KRT events (long-term dialysis initiation or kidney transplantation) were ascertained via linkage between PRESERVE study participants and the USRDS. PRESERVE participant information was linked to the USRDS database by an honest broker using name, date of birth, and social security number, with the latter provided only when available and permitted by the participating institution. The linked USRDS data included information on kidney failure etiology, date of dialysis initiation, dialysis modality, transplant information, causes of death, and medical comorbid conditions. Statistical Analysis Descriptive statistics for demographic and clinical variables were generated for the full sample and separately for each BP control group (well-controlled, moderately controlled, poorly controlled). Kaplan–Meier curves were estimated for time to the composite outcome of KRT or death across the BP control groups. Cox proportional hazards models were used to investigate the relationship between BP control and the composite of KRT or death. Each SBP control measure was modeled as a continuous predictor in separate models. Primary analyses evaluated cumulative SBP load >90th percentile. Secondary analyses evaluated the alternative SBP metrics (percent time and cumulative SBP load <90th percentile) and alternative percentile thresholds (75th and 50th percentiles). All models were adjusted for sex, race/ethnicity, CKD etiology, renin-angiotensin-aldosterone system (RAAS) blockade use, hospitalizations, medical complexity, eGFR, and the number of BP measurements. RAAS blockade use was included as a covariate because its antiproteinuric effects, independent of its BP-lowering effects, may influence CKD progression and thus confound the relationship between BP control and outcomes. Other antihypertensive medications were not adjusted for. We also accounted for the potential correlations between patients within institutions using robust sandwich-type standard errors. For each model, we examined Martingale residuals to assess the linearity of the association between BP control and the KRT or death outcome. The proportional hazards assumption was assessed using Schoenfeld residual tests (using P value <0.05) and plots against time for each model. Given the presence of these violations in multiple models, we included an interaction between time intervals and BP control to estimate piecewise hazard ratios. The follow-up intervals were specified as follows: ages 18 to less than 19 years (0-1 years of follow-up), 19 to less than 20 years (1-2 years of follow-up), 20 to less than 23 years (2-5 years of follow-up), and 23-30 years (5-12 years of follow-up). Results The final analytic cohort included 2,585 individuals with median follow-up of 7.45 years (IQR: 6.05-9.20), among whom 4.6% (n = 118) experienced a KRT or death endpoint between ages 18-30 years. Among these individuals, 42 underwent dialysis only, 29 received a kidney transplant only, and 47 began long-term dialysis before kidney transplantation. Compared with those who did not progress to KRT, those with a KRT event were more likely to have an underlying glomerular etiology for their kidney disease (51% vs 15%), proteinuria (85% vs 39%), and antihypertensive medication prescriptions (Table S3). Based on cumulative SBP load (ie, AUC) >90th SBP percentile, 655 participants had well-controlled BP, 964 patients had moderately controlled BP (less than median of 1.37 pp×time), and the remaining 966 patients had poorly controlled BP (≥1.37 pp×time) (Table 1). There was a greater proportion of female participants across all BP control groups. There were a higher proportion of African American participants (29%) and lower proportion of White participants (52%) with poor BP control than well and moderately controlled BP. There was also a higher proportion of participants with Medicaid/State Children’s Health Insurance Program participation among those with poor BP control. BP control did not differ by eGFR or non–kidney-related medical complexity. Antihypertensive medications were prescribed in 57% of the full cohort and were most commonly RAAS blockers (47%) and calcium channel blockers (25%). All antihypertensive medication types were prescribed more frequently in those individuals with moderately and poorly controlled BP than those with well-controlled BP.Table 1Cohort Characteristics by BP Control Category at Age 14 Using Cumulative SBP Load Above the 90th Percentile SBP ThresholdBP ControlaOverallN = 2,585Well ControlledN = 655Moderately ControlledN = 964Poorly ControlledN = 966Women1,398 (54.1%)358 (54.7%)521 (54.0%)519 (53.7%)Race/ethnicity White1,435 (55.5%)388 (59.2%)549 (57.0%)498 (51.6%) African American625 (24.2%)127 (19.4%)215 (22.3%)283 (29.3%) Hispanic318 (12.3%)81 (12.4%)122 (12.6%)115 (11.9%) Other207 (8.0%)59 (9.0%)78 (8.1%)70 (7.2%)Institution A149 (5.8%)28 (4.3%)52 (5.4%)69 (7.1%) B534 (20.6%)166 (25.3%)178 (18.4%)190 (19.7%) C156 (6.0%)48 (7.3%)69 (7.2%)39 (4.0%) D98 (3.8%)27 (4.1%)30 (3.1%)41 (4.2%) E399 (15.4%)90 (13.7%)158 (16.4%)151 (15.6%) F46 (1.8%)<11 (<1.7%)<11 (<1.1%)<11 (<1.1%) G505 (19.5%)104 (15.9%)193 (20.0%)208 (21.5%) H116 (4.5%)28 (4.3%)51 (5.3%)37 (3.8%) I116 (4.5%)31 (4.7%)54 (5.6%)31 (3.2%) J188 (7.3%)42 (6.4%)68 (7.0%)78 (8.1%) K184 (7.1%)47 (7.2%)62 (6.4%)75 (7.8%) L63 (2.4%)17 (2.6%)20 (2.1%)26 (2.7%) M31 (1.2%)<11 (<1.7%)<11 (<1.1%)<11 (<1.1%)Payer Commercial/self-pay700 (27.1%)229 (35.0%)261 (27.1%)210 (21.7%) Medicaid/SCHIP721 (27.9%)151 (23.1%)254 (26.3%)316 (32.7%) Other/unknown1,164 (45.0%)275 (42.0%)449 (46.5%)440 (45.5%)eGFR, mL/min/1.73 m279.5 [66.9, 88.2]77.8 [66.2, 87.6]79.1 [65.0, 88.3]80.4 [69.7, 88.3]CKD etiology Glomerular432 (16.7%)118 (18.0%)174 (18.0%)140 (14.5%) Malignancy399 (15.4%)93 (14.2%)157 (16.3%)149 (15.4%) Nonglomerular/nonmalignancy1,754 (67.8%)444 (67.8%)633 (65.6%)677 (70.1%)Proteinuriab1,061 (41.3%)240 (37.2%)399 (41.6%)422 (43.8%)Medical complexityc Complex chronic1,122 (43.4%)267 (40.8%)433 (44.9%)422 (43.7%) Noncomplex chronic166 (6.4%)39 (6.0%)53 (5.5%)74 (7.7%) Neither1,297 (50.2%)349 (53.3%)478 (49.5%)470 (48.7%)No. of hospitalizations 02,124 (82.2%)566 (86.4%)792 (82.2%)766 (79.3%) 1238 (9.2%)54 (8.2%)93 (9.6%)91 (9.4%) ≥2223 (8.6%)35 (5.3%)79 (8.2%)109 (11.3%)No. of BP measurements13.0 [8.0, 22.0]9.0 [6.0, 15.0]13.0 [8.0, 21.0]15.0 [10.0, 27.0]Any BP med use, n (%)1,465 (56.7%)253 (38.6%)509 (52.7%)703 (72.8%)RAAS blockade use1,206 (46.6%)222 (33.9%)403 (41.8%)581 (60.1%)ACEi1,093 (42.3%)208 (31.8%)362 (37.5%)523 (54.1%)ARB338 (13.1%)51 (7.8%)118 (12.2%)169 (17.5%)Note: Values are n (%) or median [interquartile range]. Cells with counts fewer than 11 are suppressed to protect confidentiality.Abbreviations: ACEi, angiotensin-converting enzyme inhibitor; ARB, angiotensin receptor blocker; BP, blood pressure; CKD, chronic kidney disease; eGFR, estimated glomerular filtration rate; RAAS, renin-angiotensin-aldosterone system; SBP, systolic blood pressure; SCHIP, State Children’s Health Insurance Program.aWell-controlled (no area >90th percentile), moderately controlled (<1.37 percentage point time >90th percentile), and poorly controlled (≥1.37 percentage point time >90th percentile).bDipstick proteinuria was defined based on urine protein test results from the 12 months before the first BP observation. Results showing 2+, 3+, and 4+ g protein on a dipstick urinalysis were considered positive for proteinuria.cDefined using Pediatric Medical Complexity Algorithm assessed over the 3 years before the BP control assessment period. Complex chronic disease is defined by a condition affecting at least 2 body systems lasting ≥1 year or a progressive or malignant condition. Noncomplex chronic disease is defined as a chronic condition that lasts ≥1 year, involves 1 body system, and does not meet criteria for a complex chronic condition. Patients are flagged as not having a chronic disease if they have an acute condition that is not expected to last >1 year or if they do not have any acute or chronic conditions. Table 1Cohort Characteristics by BP Control Category at Age 14 Using Cumulative SBP Load Above the 90th Percentile SBP ThresholdBP ControlaOverallN = 2,585Well ControlledN = 655Moderately ControlledN = 964Poorly ControlledN = 966Women1,398 (54.1%)358 (54.7%)521 (54.0%)519 (53.7%)Race/ethnicity White1,435 (55.5%)388 (59.2%)549 (57.0%)498 (51.6%) African American625 (24.2%)127 (19.4%)215 (22.3%)283 (29.3%) Hispanic318 (12.3%)81 (12.4%)122 (12.6%)115 (11.9%) Other207 (8.0%)59 (9.0%)78 (8.1%)70 (7.2%)Institution A149 (5.8%)28 (4.3%)52 (5.4%)69 (7.1%) B534 (20.6%)166 (25.3%)178 (18.4%)190 (19.7%) C156 (6.0%)48 (7.3%)69 (7.2%)39 (4.0%) D98 (3.8%)27 (4.1%)30 (3.1%)41 (4.2%) E399 (15.4%)90 (13.7%)158 (16.4%)151 (15.6%) F46 (1.8%)<11 (<1.7%)<11 (<1.1%)<11 (<1.1%) G505 (19.5%)104 (15.9%)193 (20.0%)208 (21.5%) H116 (4.5%)28 (4.3%)51 (5.3%)37 (3.8%) I116 (4.5%)31 (4.7%)54 (5.6%)31 (3.2%) J188 (7.3%)42 (6.4%)68 (7.0%)78 (8.1%) K184 (7.1%)47 (7.2%)62 (6.4%)75 (7.8%) L63 (2.4%)17 (2.6%)20 (2.1%)26 (2.7%) M31 (1.2%)<11 (<1.7%)<11 (<1.1%)<11 (<1.1%)Payer Commercial/self-pay700 (27.1%)229 (35.0%)261 (27.1%)210 (21.7%) Medicaid/SCHIP721 (27.9%)151 (23.1%)254 (26.3%)316 (32.7%) Other/unknown1,164 (45.0%)275 (42.0%)449 (46.5%)440 (45.5%)eGFR, mL/min/1.73 m279.5 [66.9, 88.2]77.8 [66.2, 87.6]79.1 [65.0, 88.3]80.4 [69.7, 88.3]CKD etiology Glomerular432 (16.7%)118 (18.0%)174 (18.0%)140 (14.5%) Malignancy399 (15.4%)93 (14.2%)157 (16.3%)149 (15.4%) Nonglomerular/nonmalignancy1,754 (67.8%)444 (67.8%)633 (65.6%)677 (70.1%)Proteinuriab1,061 (41.3%)240 (37.2%)399 (41.6%)422 (43.8%)Medical complexityc Complex chronic1,122 (43.4%)267 (40.8%)433 (44.9%)422 (43.7%) Noncomplex chronic166 (6.4%)39 (6.0%)53 (5.5%)74 (7.7%) Neither1,297 (50.2%)349 (53.3%)478 (49.5%)470 (48.7%)No. of hospitalizations 02,124 (82.2%)566 (86.4%)792 (82.2%)766 (79.3%) 1238 (9.2%)54 (8.2%)93 (9.6%)91 (9.4%) ≥2223 (8.6%)35 (5.3%)79 (8.2%)109 (11.3%)No. of BP measurements13.0 [8.0, 22.0]9.0 [6.0, 15.0]13.0 [8.0, 21.0]15.0 [10.0, 27.0]Any BP med use, n (%)1,465 (56.7%)253 (38.6%)509 (52.7%)703 (72.8%)RAAS blockade use1,206 (46.6%)222 (33.9%)403 (41.8%)581 (60.1%)ACEi1,093 (42.3%)208 (31.8%)362 (37.5%)523 (54.1%)ARB338 (13.1%)51 (7.8%)118 (12.2%)169 (17.5%)Note: Values are n (%) or median [interquartile range]. Cells with counts fewer than 11 are suppressed to protect confidentiality.Abbreviations: ACEi, angiotensin-converting enzyme inhibitor; ARB, angiotensin receptor blocker; BP, blood pressure; CKD, chronic kidney disease; eGFR, estimated glomerular filtration rate; RAAS, renin-angiotensin-aldosterone system; SBP, systolic blood pressure; SCHIP, State Children’s Health Insurance Program.aWell-controlled (no area >90th percentile), moderately controlled (<1.37 percentage point time >90th percentile), and poorly controlled (≥1.37 percentage point time >90th percentile).bDipstick proteinuria was defined based on urine protein test results from the 12 months before the first BP observation. Results showing 2+, 3+, and 4+ g protein on a dipstick urinalysis were considered positive for proteinuria.cDefined using Pediatric Medical Complexity Algorithm assessed over the 3 years before the BP control assessment period. Complex chronic disease is defined by a condition affecting at least 2 body systems lasting ≥1 year or a progressive or malignant condition. Noncomplex chronic disease is defined as a chronic condition that lasts ≥1 year, involves 1 body system, and does not meet criteria for a complex chronic condition. Patients are flagged as not having a chronic disease if they have an acute condition that is not expected to last >1 year or if they do not have any acute or chronic conditions. Cohort Characteristics by BP Control Category at Age 14 Using Cumulative SBP Load Above the 90th Percentile SBP Threshold Note: Values are n (%) or median [interquartile range]. Cells with counts fewer than 11 are suppressed to protect confidentiality. Abbreviations: ACEi, angiotensin-converting enzyme inhibitor; ARB, angiotensin receptor blocker; BP, blood pressure; CKD, chronic kidney disease; eGFR, estimated glomerular filtration rate; RAAS, renin-angiotensin-aldosterone system; SBP, systolic blood pressure; SCHIP, State Children’s Health Insurance Program. Well-controlled (no area >90th percentile), moderately controlled (<1.37 percentage point time >90th percentile), and poorly controlled (≥1.37 percentage point time >90th percentile). Dipstick proteinuria was defined based on urine protein test results from the 12 months before the first BP observation. Results showing 2+, 3+, and 4+ g protein on a dipstick urinalysis were considered positive for proteinuria. Defined using Pediatric Medical Complexity Algorithm assessed over the 3 years before the BP control assessment period. Complex chronic disease is defined by a condition affecting at least 2 body systems lasting ≥1 year or a progressive or malignant condition. Noncomplex chronic disease is defined as a chronic condition that lasts ≥1 year, involves 1 body system, and does not meet criteria for a complex chronic condition. Patients are flagged as not having a chronic disease if they have an acute condition that is not expected to last >1 year or if they do not have any acute or chronic conditions. The distribution of percent time and cumulative SBP load by KRT outcomes >90th, 75th, and 50th percentile SBP thresholds were skewed toward poorer BP control (ie, rightward shift) in those with KRT (Fig S2a-f, Table S2). A steeper decline in Kaplan–Meier survival probabilities in the poorly controlled BP group was observed across all SBP thresholds and for both SBP measures, particularly between ages 18 and 23 years (Figs 3a, b and S4a-d).Figure 3Unadjusted Kaplan–Meier survival curves for the outcome of kidney replacement therapy or death stratified by SBP control group defined by (a) cumulative SBP load >90th percentile and (b) percentage of time <90th percentile SBP threshold. Abbreviation: SBP, systolic blood pressure. Unadjusted Kaplan–Meier survival curves for the outcome of kidney replacement therapy or death stratified by SBP control group defined by (a) cumulative SBP load >90th percentile and (b) percentage of time <90th percentile SBP threshold. Abbreviation: SBP, systolic blood pressure. In a fully adjusted Cox model, cumulative SBP load >90th percentile between ages 14 and 18 was significantly associated with an increased hazard of KRT or death. Based on evaluation of Martingale residuals, the association was linear across AUC levels. Each pp×time increase in cumulative SBP load >90th percentile was associated with a 1.36 (95% confidence interval [CI], 1.17-1.58) times higher hazard of KRT or death at age 18-19. Hazard ratios for ages 19-20, 20-23, and 23-30 years were 1.18 (95% CI, 0.96-1.46), 1.13 (95% CI, 0.99-1.29), and 1.03 (95% CI, 0.81-1.30), respectively (Fig 4a, Table S4). A similar decreasing pattern over time was observed for the 75th and 50th percentile SBP thresholds. Hazard ratio estimates using percent time with SBP >90th percentile were similar to analyses assessing BP control using the cumulative SBP load >90th percentile method (Fig 4b, Table S5). Using the coefficients from these percent time models, percent risk reduction of KRT or death at age 18-19 for 0%, 25%, 50%, 75%, and 100% of time during adolescence (ages 14 to less than 18 years) with SBP ≤50th, 75th, and 90th SBP percentiles were calculated and summarized in Fig 5.Figure 4Forest plots showing hazard of kidney replacement therapy or death between ages 18 and 30 years at different follow-up ages, based on (a) 1 percentile point × time increment in cumulative SBP load above each SBP percentile threshold, (b) a 10% increment in the percentage of time above each SBP percentile threshold, and (c) 1 percentile point × time increment in cumulative SBP load below each SBP percentile threshold. Abbreviation: SBP, systolic blood pressure.Figure 5Percent risk reduction of kidney replacement therapy or death for young adults with chronic kidney disease at age 18 to <19 years by increasing time with SBP at or below the 50th, 75th, and 90th SBP percentiles during adolescence (ages 14 to < 18 years). Relative risk reduction is based on confounder-adjusted data shown in Table S5. Abbreviation: SBP, systolic blood pressure. Forest plots showing hazard of kidney replacement therapy or death between ages 18 and 30 years at different follow-up ages, based on (a) 1 percentile point × time increment in cumulative SBP load above each SBP percentile threshold, (b) a 10% increment in the percentage of time above each SBP percentile threshold, and (c) 1 percentile point × time increment in cumulative SBP load below each SBP percentile threshold. Abbreviation: SBP, systolic blood pressure. Percent risk reduction of kidney replacement therapy or death for young adults with chronic kidney disease at age 18 to <19 years by increasing time with SBP at or below the 50th, 75th, and 90th SBP percentiles during adolescence (ages 14 to < 18 years). Relative risk reduction is based on confounder-adjusted data shown in Table S5. Abbreviation: SBP, systolic blood pressure. The distribution of cumulative SBP load below the 90th, 75th, and 50th percentile by KRT outcome are shown in Fig S3a-c. Kaplan–Meier curves for KRT or death by SBP control group using this measure are shown in Fig S5a-c. In a fully adjusted Cox model, each pp×time increase in AAC cumulative SBP load below the 90th percentile was associated with a 0.94 (95% CI, 0.91-0.97) times lower hazard of KRT or death at age 18-19 years. The effect of improved BP control on the hazard of KRT or death between ages 19 and 30 years was not statistically significant (Fig 4c, Table S6). C-statistics ranged from 0.889 to 0.895 across all primary and secondary models (Table S7). Discussion Hypertension is a modifiable risk factor for progression of CKD in childhood and the development of target end-organ damage, such as left ventricular hypertrophy.9,11 Yet, hypertension may go under- or untreated.5 We found that real-world BP measurements can inform risk of adverse long-term kidney outcomes and define clinically meaningful target thresholds. We used multicenter EHR data to characterize the risk of KRT among children with CKD. We focused on SBP control during the ages of 14.0-17.9 years when adolescents experience significant physiologic18 and socioemotional changes that place them at risk for CKD progression. We found that worse BP control was associated with a markedly increased risk of kidney failure necessitating KRT in young adulthood and that this risk tapers over time. Risk increased linearly with increasing uncontrolled SBP. This implies that large increases in the amount of time below the SBP threshold could substantially reduce risk of kidney failure. Using both area (ie, time and magnitude) and time-only based methods, we observed that improved BP control was associated with a lower risk of kidney failure. Such data may be used to encourage health care providers to address uncontrolled BP and counsel patients and caregivers regarding the benefits of improved medication adherence, closer BP monitoring, or dietary interventions, etc. Future work on other factors influencing BP control, including social determinants of health, are also warranted to understand high-risk groups for potential policy interventions. Clinical trials in both pediatric and adult CKD populations support intensified BP lowering to delay progression of CKD and KRT (eg, SPRINT19). The ESCAPE trial (n = 385) demonstrated a renoprotective benefit of 24-hour mean arterial pressure reduction to <50th percentile in children with both glomerular disorders and hypoplasia-dysplasia.11 We similarly found a protective benefit for BP control below the SBP thresholds we examined (50th, 75th, and 90th). Moreover, we found greater benefits as the magnitude and/or duration increase of values below these SBP percentile thresholds. As opposed to the protocolized and controlled setting of a clinical trial, these findings support the dynamic nature of real-world BP management, especially in adolescents. For example, even for an individual with fluctuations above and below a given BP threshold, a significant reduction in risk can be achieved by sustained efforts over time. Clinical practice guidelines recommend that 24-hour mean arterial pressure in children with CKD be controlled to <50th percentile for age, sex, and height or SBP to <90th percentile using manual auscultatory office BP measurement.9 Although ambulatory BP monitoring (ABPM) may better characterize BP control and predict left ventricular hypertrophy in children and adolescents than in-office BP measurements, ABPM is not always readily available or feasible for repeated assessments.20 For example, cumulative BP loads have been associated with incident CKD in adults with normal kidney function and no history of hypertension or antihypertensive medication use, highlighting the prognostic value of methods to assess BP load, such as ABPM, over time.13 The adolescent years are characterized by poor health outcomes among those with CKD, such as high rates of graft rejection among those having undergone kidney transplantation.21 There is also a scarcity of evidence to inform an optimal approach to supporting adolescent patients with CKD with the social, developmental, psychological, and health-related obstacles they experience during the transition from pediatric to adult care.22 Although overall mortality during follow-up in our analysis was low (<1%), it is well known that adults with childhood-onset kidney failure have high rates of mortality, predominantly related to cardiovascular disease.23,24 Consequently, our findings have important implications for health care providers and families navigating the transition from pediatric to adult care. First, poor BP control during late adolescence should be recognized as a strong risk factor for disease progression, with steps taken to ensure that BP is well controlled over time using assessment by ABPM, coordination and communication with primary care providers regarding BP goals, or rigorous home monitoring, when feasible. Providers transitioning care of an adolescent with CKD should communicate a thorough assessment of BP control as part of a comprehensive transition plan. Similarly, those individuals with a history of poor BP control as adolescents should be considered high risk for cardiovascular mortality.10 The principal strength of the study is its size, geographic and racial diversity, duration of follow-up, use of real-world clinical data, and rigorous ascertainment of the outcome event using linked USRDS data. Much of the hypertension literature separates adults from children. Our data helps to inform the gap in large-scale clinical research that spans the pediatric and adult divide. Study limitations include potential misclassification of BP related to white coat or masked hypertension. Future studies incorporating ABPM data could inform the direction of bias from these phenomena. Adherence to prescribed antihypertensive medications was not assessed. Medical management that may have occurred outside of the study sites was also not captured. In the United States, most children with CKD are cared for by pediatric nephrologists at tertiary care referral centers. External validity to settings where primary care providers may be managing CKD or to populations with limited access to care may be limited. Last, the cumulative SBP load (AUC) and AAC methods used to define BP control did not capture potential BP fluctuations above and below a given BP percentile within an individual. The clinical implications of such fluctuations should be explored in future analyses. In summary, using multicenter EHR data, we demonstrated that among adolescents with CKD, worse BP control is associated with a markedly increased risk of kidney failure in young adulthood. Substantial risk reduction can be achieved with even moderate improvements in BP control. Clinicians and patients should enhance efforts to address this modifiable risk factor for kidney failure.