Abstract
Introduction
Few population-based data exist in the U.S. about adults with IgA nephropathy (IgAN) and risks of kidney, cardiovascular, and mortality outcomes versus those with nonglomerular chronic kidney disease (CKD) or without CKD.
Methods
We applied natural language processing (NLP) algorithms to electronic health record (EHR) data within a large integrated healthcare delivery system to identify adults with biopsy-proven IgAN between 2010 and 2020. We next identified 2 separate comparison cohorts of adults with nonglomerular CKD and adults without CKD matched on age and sex, and compared the rates of end-stage kidney disease (ESKD), worsening CKD, acute kidney injury (AKI), cardiovascular outcomes, and death through 2021 using multivariable Cox proportional hazards models.
Results
We identified 1651 adults with biopsy-confirmed IgAN who had a mean age of 43 years and 48% women, 41% Asian or Pacific Islander, and 3.3% Black. Compared with matched adults with nonglomerular CKD (n = 9863), those with IgAN had higher adjusted rates of ESKD (adjusted hazard ratio [aHR]: 2.79, 95% CI: 2.16–3.62), worsening CKD (aHR: 3.05, 95% CI: 2.68–3.46), and AKI (aHR: 1.45, 95% CI: 1.22–1.73) but no significant adjusted differences in cardiovascular events and death. Compared with matched adults without CKD (n = 16,510), patients with IgAN had substantially higher adjusted rates of adverse kidney outcomes as well as higher adjusted rates of hospitalization for heart failure (aHR: 8.06, 95% CI: 2.90–22.37) and death (aHR: 2.90, 95% CI: 2.08–4.02) but not acute myocardial infarction or stroke/transient ischemic attack.
Conclusion
Adults with IgAN are at greater risk of adverse kidney outcomes versus nonglomerular CKD and substantially higher risks of heart failure and death compared with no CKD.
Keywords: cardiovascular events, IgA nephropathy, kidney failure
Graphical abstract
Glomerulonephritis (GN) is a broad term used to describe a large, diverse group of inflammatory kidney diseases that cause injury to glomeruli and ultimately lead to the development of CKD and an increased risk of progressing to ESKD requiring kidney replacement therapy. IgAN is the most common GN worldwide with an estimated incidence of 2.5 cases per 100,000 and is reported to be associated with higher rates of CKD progression and ESKD compared with other causes of kidney disease.1, 2, 3
Although adults with CKD and ESKD are known to have higher risks of cardiovascular disease (CVD) events and death, few studies have specifically examined the risks of CVD events among contemporary patients with IgAN. Using a Finnish registry in 2006, Myllymaki et al.4 reported a nearly 3-fold higher incidence of CVD events in 221 adults with IgAN compared to the general population. However, that study did not compare vascular events in IgAN to patients with other forms of glomerular disease or other kidney diseases and had other important methodological limitations. In a study among 633 adults with IgAN in Norway, Knoop et al.5 observed that 45% of all deaths in the cohort were attributed to CVD, but did include a comparable group of CKD patients without IgAN. Collectively, existing published studies of both kidney and CVD outcomes in patients with IgAN are limited by small sample sizes, restricted sociodemographic diversity, primarily include data from earlier eras and from selected practice settings, and did not examine outcomes with similar CKD patients without IgAN, especially within the U.S.
To address these knowledge gaps, we conducted a population-based study within a US-based, large, integrated healthcare delivery system to characterize recent clinical outcomes among adults with biopsy-proven IgAN as compared with other patients with CKD attributed to nonglomerular etiologies and patients without CKD.
Methods
Source Population
Kaiser Permanente Northern California (KPNC) is an integrated health care delivery system with 21 hospitals and > 240 freestanding clinics where currently > 4.6 million members receive comprehensive care (i.e., inpatient, emergency department, and ambulatory encounters) within Northern and Central California. The KPNC membership is highly representative of the local and statewide population with respect to age, sex, race/ethnicity, and socioeconomic status.6
This study was approved by the KPNC institutional review board, and a waiver of informed consent was obtained as this is a retrospective, EHR data-only study.
Identification of Adults With IgAN Using Natural Language Processing
We first aimed to identify a cohort of adults with biopsy-proven IgAN who were alive with valid KPNC health plan membership for any period between January 1, 2010 and December 31, 2020. Given the limitations of administrative codes for identifying and characterizing causes of CKD, we used a combination of physician-assigned diagnosis codes and an NLP algorithm applied to unstructured and semistructured EHR data. First, we identified all KPNC members who had an International Classification of Diseases, Ninth Edition (ICD-9) or Tenth Edition (ICD-10) diagnosis code for any nephrotic or nephritic syndrome (NS), GN, or IgAN (ICD-9: 580.X, 581.X, 582.X, and 583.X; ICD-10: N00.X, N01.X, N02.X, N03.X, N04.X, N05.X, N06.X, and N08) at any time using available data. Although no diagnosis codes are mapped specifically to IgAN, we considered ICD-9 code 583.9 and ICD-10 code N02.8 to be potentially indicative of IgAN operationally. We then excluded patients who had died, lost KPNC membership, or developed ESKD before January 1, 2010. ESKD was defined as receipt of chronic dialysis or kidney transplant and was ascertained through a comprehensive health system ESKD treatment registry.7
To identify patients with biopsy-proven IgAN, we developed and validated a rule-based NLP algorithm to analyze unstructured biopsy reports and clinical progress notes. We used a random sample of 100 patients with qualifying diagnosis codes described above to derive the NLP algorithms and a separate random sample of 100 patients to validate the algorithms and report performance. Derivation and validation samples were stratified by qualifying diagnosis code (i.e., code indicative of IgAN vs. code for other NS/GN) and presence of an available kidney biopsy report at KPNC to enrich the samples for positive IgAN cases. All EHR data for patients included in the derivation and validation samples were manually reviewed by board-certified physicians (ASG and AM) for the presence or absence of biopsy-proven IgAN as the “gold standard” for classification. The NLP algorithm was developed using Linguamatics I2E software (version 6.2.0, IQVIA, Cambridge, UK), an ontology-based interactive information extraction system, and accounted for clinical negations and common spelling errors.8,9 To meet the definition of biopsy-proven IgAN, patients had to have a positive mention of IgAN along with documented evidence of a kidney biopsy in a biopsy report or clinical notes. We note that no specific histological findings were examined using NLP; rather, NLP was used to ascertain the treating provider’s assessment of the biopsy. Biopsy dates were also extracted using NLP. The validation sample had 32 adjudicated biopsy-proven IgAN cases, and the final NLP algorithm classified them with a sensitivity of 93.9%, specificity of 98.5%, positive predictive value of 96.9% and negative predictive value of 97.0%. All NLP queries are available in YAML Ain't Markup Language format in the online supplement.
Finally, we applied the validated NLP algorithms to all biopsy reports and progress notes in the full sample of patients with qualifying diagnosis codes to identify the cohort of patients with biopsy-proven IgAN. The index date for follow-up was assigned using the following priority: (i) first confirmed biopsy date between 2010 and 2020 identified through biopsy records or clinical notes; (ii) first IgAN diagnosis date between 2010 and 2020 if evidence of a biopsy was present but the specific date was unavailable; or (iii) January 1, 2010 if the confirmed biopsy date was before 2010 given the study inception period.
Identification of a Comparison Cohort of Adults with Nonglomerular CKD
After identifying adults with confirmed IgAN, we identified a comparison cohort of adults with CKD not attributable to IgAN or other glomerular diseases. Eligible patients with nonglomerular CKD were identified as having at least 2 outpatient department estimated glomerular filtration rate (eGFR) measurements < 60 ml/min per 1.73 m2 at least 90 days apart within 4 years before or during the study period. Patients with ESKD before January 1, 2010 or having any diagnosis codes for GN or NS were excluded. Presumed nonglomerular CKD patients were then individually matched without replacement to patients with IgAN at a variable ratio up to 10:1 based on the following characteristics and eligibility on the corresponding IgAN index date: age (± 2 years), sex (male or female per EHR data), at least 6 months of previous KPNC membership to ensure full capture of baseline comorbidities, no previous ESKD or outpatient eGFR < 15 ml/min per 1.73 m2, and no previous documented nephrotic-range proteinuria (defined as a urine albumin-to-creatinine ratio > 3000 mg/g, urine protein-to-creatinine ratio > 2 g/g, or at least 2 measurements of 3+ proteinuria from urine dipstick tests). Matching was primarily performed to identify an eligible comparison cohort at the time of the IgAN diagnosis rather than control for covariates; all qualifying criteria for patients with nonglomerular CKD were required to have been met at any time before the IgAN index date to be eligible for matching. Although matching at a 1:k ratio provides minimal to no improvements in precision when k > 5, we used a variable ratio of up to 1:10 ratio to include a more representative sample of patients with nonglomerular CKD, especially considering the heterogeneity in populations and minimal adjustment for confounding at this stage. In variable ratio matching, each patient with IgAN is matched to between 1 and 10 comparison patients with nonglomerular CKD, depending on the population available.
Identification of a Comparison Cohort of Adults With No CKD
We also identified a comparison cohort of adults with no documented glomerular disease or documented CKD using similar methods. Eligible patients with no CKD were identified as having no eGFR measurements < 60 ml/min per 1.73 m2 or ESKD before January 1, 2010 or having any diagnosis codes for GN or NS. Patients without CKD were individually matched without replacement to patients with IgAN at a fixed ratio of 10:1 on the following at the corresponding IgAN index date: age (± 1 year), sex (male or female per EHR data), at least 6 months of previous KPNC membership, no previous ESKD or eGFR < 60 ml/min per 1.73 m2, and no previous documented nephrotic-range proteinuria.
Covariates
We obtained demographic characteristics (age, sex, self-reported race, and Hispanic ethnicity) from health plan databases to describe the study cohorts at baseline. We defined relevant comorbidities within 5 years before the index date by diagnosis or procedure codes supplemented with laboratory test results, outpatient vital signs, and/or prescribed medications using EHR-based data that was cleaned using standardized procedures and linked at the individual-patient level in the Kaiser Permanente Virtual Data Warehouse, as previously described and validated.10,11 Baseline medication use was ascertained using filled outpatient prescriptions within 120 days before the index date. Laboratory values and vital signs were ascertained as the most recent outpatient value within 1 year before the index date. All eGFR values were calculated using the 2021 CKD Epidemiology Collaboration serum creatinine-based equation without the Black race coefficient.12 Proteinuria was measured using the most recent available urine albumin-to-creatinine ratio measurements within 4 years before the index date; where unavailable, urine albumin-to-creatinine ratio was estimated from available urine protein-to-creatinine ratio and urine dipstick results, respectively.13
Follow-up and Outcomes
We censored follow-up time at death, loss of health plan membership, or December 31, 2021, whichever was earliest. We ascertained ESKD through a validated health system registry.7 Incident or worsening CKD was a composite outcome defined as a 50% reduction in eGFR from baseline, a reduction in eGFR to < 15 ml/min per 1.73 m2, or ESKD. For this outcome only, patients without a baseline outpatient eGFR measurement within 1 year before their index date were assumed to have an eGFR of 120 ml/min per 1.73 m2, in which case a 50% reduction would necessitate a follow-up outpatient eGFR of ≤ 60 ml/min per 1.73 m2. Hospitalized AKI was defined according to the Kidney Disease: Improving Global Outcomes serum creatinine based criteria as a relative increase in inpatient serum creatinine of ≥ 50% from baseline, or an absolute increase of 0.3 mg/dl within 48 hours during the hospitalization. For this outcome, the baseline serum creatinine value for each hospitalization was defined as the most recent outpatient, nonemergency department value within 7 to 365 days before admission. Cardiovascular outcomes included hospitalization for acute myocardial infarction (MI), hospitalization or emergency department visit for ischemic stroke or transient ischemic attack (TIA), and hospitalization for heart failure (HF), ascertained using previously validated ICD-9 and ICD-10 primary discharge diagnosis codes for hospitalizations and emergency department visits occurring in KPNC and non-network facilities that were comprehensively captured in our EHR. Death was ascertained based on comprehensive information from health plan administrative and clinical databases, member proxy reporting, Social Security Administration vital status files, and state death certificate information.14 As a sensitivity analysis, we additionally censored follow-up for cardiovascular outcomes at the time of ESKD.
Statistical Analyses
Analyses were conducted using SAS (version 9.4, SAS Institute Inc, Cary, NC). We first characterized each of the cohorts and compared demographics, comorbidities, medication use, and laboratory values at baseline using standardized differences derived from Cohen’s d for continuous and binary variables and Cramer’s V for categorical variables. These values reflect a standardized magnitude of the difference between groups comparable across characteristics with varying distributions; for example, a value of 0.10 would indicate a 10% difference between groups.
We then calculated crude incidence rates per 100 person-years (PY) with associated Poisson 95% CI for all outcomes in each cohort. For nonfatal outcomes that were potentially recurrent (hospitalized AKI, MI, stroke/TIA, or HF), we calculated rates incorporating the first event only and including recurrent events. Next, we developed both unadjusted and multivariable stratified Cox proportional hazards regression models to calculate hazard ratios (HR) and associated 95% CI for all outcomes adjusting for demographics, comorbidities, and laboratory values that were substantially different at baseline in each cohort and stratifying on the matched groups. We conducted sensitivity analyses additionally adjusting for prednisone and cardioprotective medication use at baseline. To account for competing risks of death, we also modeled the cumulative incidence of each outcome using Fine and Gray subdistribution hazard models and calculate subdistribution hazard ratios. For the IgAN versus nonglomerular CKD matched cohort, we additionally calculated HR stratified by baseline proteinuria status. For nonfatal recurrent outcomes, we assume that events follow a Poisson distribution, and therefore developed multivariable adjusted Poisson regression models using a generalized estimating equations approach to calculate rate ratios and associated 95% CI accounting for matched groups. To account for missing data in continuous covariates included in all models, we used a multiple imputation approach across 50 imputed datasets using all available covariate and outcome information. Variables with greater than 50% missing values were not considered for adjustment. We conducted a sensitivity analysis restricting the matched cohort to patients with IgAN with index dates during the study period, which are presumed to be reflective of incident IgAN.
To facilitate comparison with the nonglomerular CKD cohort, we created a 1:1 propensity score-matched cohort of IgAN and nonglomerular CKD patients. The propensity score model included age, sex, race/ethnicity, tobacco use, CVD history (previous atrial fibrillation, stroke/TIA, myocardial infarction, heart failure, coronary artery bypass surgery, and percutaneous coronary intervention as separate variables), hypertension, diabetes mellitus, dyslipidemia, eGFR, proteinuria, and anti-hypertensive and statin medication use. We used a nearest neighbor matching approach and excluded individuals with propensity scores outside the region of common support.
Results
Cohort Assembly and Characteristics
Among 30,175 eligible KPNC patients with diagnosis codes for IgAN, GN, or NS between 2010 and 2020, we identified 1651 patients with biopsy-confirmed IgAN using NLP algorithms (Figure 1). Of these 1651 patients with IgAN, 1267 patients were matched to 9863 eligible patients with nonglomerular CKD and all 1651 patients with IgAN were matched to 16,510 eligible patients with no CKD on age and sex at the time of the IgAN diagnosis. Compared with matched patients who had nonglomerular CKD, patients with IgAN were more likely to be of Asian American or Pacific Islander descent, had a lower overall comorbidity burden (with a particularly lower prevalence of heart failure, diabetes, hypertension, and chronic lung disease), less overweight or obese, and were receiving fewer cardioprotective medications, but had higher cholesterol and proteinuria levels (Table 1). Compared with matched patients who did not have CKD, patients with IgAN were more likely to be of Asian American or Pacific Islander descent, had a higher prevalence of hypertension and dyslipidemia, were receiving more cardioprotective medications, and had higher cholesterol and proteinuria levels (Table 1). The median (interquartile range) uncensored follow-up time for all outcomes was 5.6 years (2.8–10.2) in the nonglomerular CKD comparison cohort and 5.9 years (2.9–11.2) in the non-CKD comparison cohort. In the propensity score-matched cohort of IgAN and nonglomerular CKD, there were 883 matched pairs; characteristics of this cohort are shown in Supplementary Table S1.
Figure 1.
Assembly of cohorts of IgAN, non-glomerular chronic kidney disease, and no chronic kidney disease.
Table 1.
Baseline characteristics
| Characteristics | Comparison cohort with nonglomerular CKD |
Comparison cohort with no CKD |
||||
|---|---|---|---|---|---|---|
| IgA Nephropathy (n = 1267) | Nonglomerular CKD (n = 9863) | Std. diff. | IgA Nephropathy (n = 1651) | No CKD (n = 16,510) | Std diff | |
| Demographics | ||||||
| Age (yrs) | ||||||
| Mean (SD) | 46.0 (13.7) | 50.9 (11.9) | 0.08 | 42.5 (14.7) | 42.5 (14.7) | 0.00 |
| Median (interquartile range) | 45.0 (36.2–55.7) | 49.6 (42.8–58.9) | 40.8 (31.6–52.6) | 40.9 (31.5–52.8) | ||
| Women, n (%) | 612 (48.3) | 4589 (46.5) | 0.00 | 790 (47.8) | 7928 (48.0) | 0.00 |
| Self-reported race, n (%) | 0.37 | 0.15 | ||||
| American Indian or Alaska Native | 5 (0.4) | 61 (0.6) | 5 (0.3) | 83 (0.5) | ||
| Asian American or Pacific Islander | 526 (41.5) | 1518 (15.4) | 677 (41.0) | 3260 (19.7) | ||
| Black or African American | 36 (2.8) | 2335 (23.7) | 55 (3.3) | 1049 (6.4) | ||
| Multiracial | 74 (5.8) | 567 (5.7) | 94 (5.7) | 697 (4.2) | ||
| White | 407 (32.1) | 4207 (42.7) | 533 (32.3) | 7780 (47.1) | ||
| Unknown | 219 (17.3) | 1176 (11.9) | 287 (17.4) | 3642 (22.1) | ||
| Hispanic ethnicity, n (%) | 235 (18.5) | 1422 (14.4) | 0.03 | 302 (18.3) | 3238 (19.6) | 0.01 |
| Tobacco use, n (%) | 0.13 | 0.04 | ||||
| Current smoker | 67 (5.3) | 883 (9.0) | 90 (5.5) | 1480 (9.0) | ||
| Former smoker | 216 (17.0) | 2697 (27.3) | 249 (15.1) | 2501 (15.1) | ||
| Never smoker | 984 (77.7) | 6284 (63.7) | 1312 (79.5) | 12,530 (75.9) | ||
| Medical history | ||||||
| Diabetic chronic kidney disease | 93 (7.4) | 2305 (23.4) | 0.74 | 93 (5.9) | 0 (0.0) | 4.21 |
| Hypertensive chronic kidney disease | 245 (19.3) | 2716 (27.5) | 0.25 | 266 (16.9) | 0 (0.0) | 4.85 |
| Atrial fibrillation or flutter | 40 (3.2) | 400 (4.1) | 0.01 | 45 (2.7) | 149 (0.9) | 0.05 |
| Ventricular fibrillation or tachycardia | 0 (0.0) | 63 (0.6) | 0.06 | 0 (0.0) | 10 (0.1) | 0.01 |
| Ischemic stroke or transient ischemic attack | 8 (0.6) | 174 (1.8) | 0.05 | 8 (0.5) | 59 (0.4) | 0.01 |
| Acute myocardial infarction | 1 (0.1) | 153 (1.6) | 0.07 | 2 (0.1) | 43 (0.3) | 0.01 |
| Coronary artery bypass surgery | 3 (0.2) | 80 (0.8) | 0.03 | 5 (0.3) | 22 (0.1) | 0.01 |
| Percutaneous coronary intervention | 3 (0.2) | 110 (1.1) | 0.05 | 3 (0.2) | 33 (0.2) | 0.00 |
| Heart failure | 23 (1.8) | 692 (7.0) | 0.12 | 25 (1.5) | 52 (0.3) | 0.05 |
| Mitral or aortic valvular disease | 24 (1.9) | 282 (2.9) | 0.03 | 25 (1.5) | 107 (0.6) | 0.03 |
| Venous thromboembolism | 19 (1.5) | 288 (2.9) | 0.06 | 20 (1.2) | 51 (0.3) | 0.04 |
| Other thromboembolic events | 2 (0.2) | 24 (0.2) | 0.00 | 2 (0.1) | 4 (0.0) | 0.02 |
| Hospitalized bleed | 8 (0.6) | 126 (1.3) | 0.04 | 10 (0.6) | 36 (0.2) | 0.02 |
| Diabetes mellitus | 135 (10.7) | 3647 (37.0) | 0.28 | 158 (9.6) | 1053 (6.4) | 0.04 |
| Hypertension | 577 (45.5) | 6944 (70.4) | 0.22 | 702 (42.5) | 2578 (15.6) | 0.20 |
| Dyslipidemia | 578 (45.6) | 5912 (59.9) | 0.09 | 688 (41.7) | 3536 (21.4) | 0.14 |
| Hyperthyroidism | 27 (2.1) | 392 (4.0) | 0.05 | 40 (2.4) | 241 (1.5) | 0.02 |
| Hypothyroidism | 106 (8.4) | 1241 (12.6) | 0.06 | 123 (7.5) | 728 (4.4) | 0.04 |
| Chronic liver disease | 84 (6.6) | 610 (6.2) | 0.02 | 96 (5.8) | 352 (2.1) | 0.07 |
| Chronic lung disease | 215 (17.0) | 2590 (26.3) | 0.11 | 277 (16.8) | 2457 (14.9) | 0.02 |
| Diagnosed depression | 99 (7.8) | 1970 (20.0) | 0.05 | 125 (7.6) | 1441 (8.7) | 0.01 |
| Diagnosed dementia | 5 (0.4) | 117 (1.2) | 0.05 | 5 (0.3) | 94 (0.6) | 0.01 |
| Vital signs | ||||||
| Body mass index, kg/m2 | ||||||
| Mean (SD) | 28.9 (6.6) | 32.1 (8.3) | 0.56 | 28.6 (6.9) | 28.3 (7.0) | 0.04 |
| Median (interquartile range) | 27.8 (24.3–32.0) | 30.8 (26.5–36.3) | 27.4 (24.0–31.7) | 27.1 (23.7–31.3) | ||
| Missing, n (%) | 273 (21.5) | 862 (8.7) | 395 (23.9) | 5143 (31.1) | ||
| Systolic blood pressure, mmHg | ||||||
| Mean (SD) | 127.0 (17.1) | 127.4 (17.0) | 0.01 | 126.1 (17.1) | 121.6 (14.0) | 0.29 |
| Median (interquartile range) | 126.0 (116.0–137.0) | 127.0 (117.0–136.0) | 125.0 (115.0–136.0) | 121.0 (112.0–131.0) | ||
| Missing, n (%) | 244 (19.3) | 639 (6.5) | 356 (21.6) | 4831 (29.3) | ||
| Medications | ||||||
| Prednisone | 121 (9.6) | 0 (0.0) | 0.22 | 151 (9.1) | 0 (0.0) | 0.29 |
| Angiotensin-converting enzyme inhibitor | 324 (25.6) | 3597 (36.5) | 0.11 | 412 (25.0) | 1214 (7.4) | 0.18 |
| Angiotensin II receptor blocker | 206 (16.3) | 1456 (14.8) | 0.04 | 242 (14.7) | 445 (2.7) | 0.18 |
| Aldosterone receptor antagonist | 11 (0.9) | 379 (3.8) | 0.10 | 13 (0.8) | 32 (0.2) | 0.03 |
| Diuretic | 263 (20.8) | 3968 (40.2) | 0.19 | 301 (18.2) | 1217 (7.4) | 0.11 |
| Alpha blocker | 37 (2.9) | 362 (3.7) | 0.01 | 39 (2.4) | 152 (0.9) | 0.04 |
| Central alpha receptor antagonist | 27 (2.1) | 330 (3.3) | 0.04 | 29 (1.8) | 45 (0.3) | 0.07 |
| Beta blocker | 238 (18.8) | 3801 (38.5) | 0.19 | 270 (16.4) | 1077 (6.5) | 0.11 |
| Calcium channel blocker | 272 (21.5) | 2585 (26.2) | 0.05 | 314 (19.0) | 492 (3.0) | 0.22 |
| Antiarrhythmic drug | 5 (0.4) | 137 (1.4) | 0.05 | 5 (0.3) | 59 (0.4) | 0.00 |
| Oral anticoagulant | 42 (3.3) | 513 (5.2) | 0.04 | 46 (2.8) | 129 (0.8) | 0.06 |
| Antiplatelet drug | 6 (0.5) | 242 (2.5) | 0.07 | 7 (0.4) | 71 (0.4) | 0.00 |
| Statins | 304 (24.0) | 4287 (43.5) | 0.15 | 356 (21.6) | 1700 (10.3) | 0.10 |
| Other lipid-lowering drugs | 32 (2.5) | 602 (6.1) | 0.08 | 36 (2.2) | 137 (0.8) | 0.04 |
| Nitrates | 8 (0.6) | 324 (3.3) | 0.09 | 10 (0.6) | 65 (0.4) | 0.01 |
| Vasodilators | 43 (3.4) | 721 (7.3) | 0.08 | 47 (2.8) | 82 (0.5) | 0.08 |
| Non-steroidal anti-inflammatory drugs | 100 (7.9) | 1060 (10.7) | 0.04 | 118 (7.1) | 1406 (8.5) | 0.01 |
| Diabetic therapy | 94 (7.4) | 3108 (31.5) | 0.28 | 110 (6.7) | 804 (4.9) | 0.02 |
| Laboratory values | ||||||
| Hemoglobin, g/dl | ||||||
| Mean (SD) | 13.0 (1.9) | 13.2 (1.9) | 0.08 | 13.0 (1.9) | 13.8 (1.5) | 0.47 |
| Median (interquartile range) | 13.0 (11.7–14.4) | 13.3 (12.0–14.5) | 13.1 (11.7–14.4) | 13.8 (12.9–14.9) | ||
| Missing, n (%) | 400 (31.6) | 2726 (27.6) | 563 (34.1) | 10,956 (66.4) | ||
| Hemoglobin A1C, % | ||||||
| Mean (SD) | 5.9 (1.0) | 7.0 (1.8) | 1.28 | 5.9 (1.0) | 6.3 (1.4) | 0.29 |
| Median (interquartile range) | 5.7 (5.4–6.2) | 6.5 (5.7–7.8) | 5.7 (5.4–6.1) | 5.8 (5.4–6.6) | ||
| Missing, n (%) | 829 (65.4) | 4681 (47.5) | 1146 (69.4) | 13,961 (84.6) | ||
| High density lipoprotein, mg/dl | 0.12 | 0.11 | ||||
| < 35 | 66 (5.2) | 972 (9.9) | 78 (4.7) | 349 (2.1) | ||
| 35–39 | 93 (7.3) | 950 (9.6) | 110 (6.7) | 585 (3.5) | ||
| 40–49 | 212 (16.7) | 1960 (19.9) | 247 (15.0) | 1507 (9.1) | ||
| 50–59 | 133 (10.5) | 1232 (12.5) | 157 (9.5) | 1225 (7.4) | ||
| ≥ 60 | 153 (12.1) | 1046 (10.6) | 186 (11.3) | 1273 (7.7) | ||
| Unknown | 610 (48.1) | 3704 (37.6) | 873 (52.9) | 11,572 (70.1) | ||
| Low density lipoprotein, mg/dl | 0.14 | 0.15 | ||||
| < 70 | 101 (8.0) | 1312 (13.3) | 115 (7.0) | 524 (3.2) | ||
| 70–99 | 210 (16.6) | 2270 (23.0) | 252 (15.3) | 1485 (9.0) | ||
| 100–129 | 179 (14.1) | 1658 (16.8) | 223 (13.5) | 1760 (10.7) | ||
| 130–159 | 116 (9.2) | 914 (9.3) | 132 (8.0) | 970 (5.9) | ||
| 160–199 | 55 (4.3) | 345 (3.5) | 61 (3.7) | 338 (2.0) | ||
| ≥ 200 | 35 (2.8) | 100 (1.0) | 42 (2.5) | 54 (0.3) | ||
| Unknown | 571 (45.1) | 3265 (33.1) | 826 (50.0) | 11,380 (68.9) | ||
| Total cholesterol, mg/dl | 0.11 | 0.13 | ||||
| < 200 | 402 (31.7) | 4333 (43.9) | 486 (29.4) | 3154 (19.1) | ||
| 200–240 | 159 (12.5) | 1316 (13.3) | 189 (11.4) | 1409 (8.5) | ||
| > 240 | 114 (9.0) | 594 (6.0) | 134 (8.1) | 450 (2.7) | ||
| Unknown | 592 (46.7) | 3621 (36.7) | 842 (51.0) | 11,498 (69.6) | ||
| Serum creatinine, mg/dl | ||||||
| Mean (SD) | 1.7 (1.2) | 1.5 (0.5) | 0.18 | 1.6 (1.3) | 0.8 (0.2) | 0.88 |
| Median (interquartile range) | 1.4 (0.9–2.0) | 1.4 (1.2–1.6) | 1.3 (0.9–2.0) | 0.8 (0.7–1.0) | ||
| Missing, n (%) | 277 (21.9) | 185 (1.9) | 387 (23.4) | 8551 (51.8) | ||
| Estimated glomerular filtration rate, ml/min per 1.73m2 | ||||||
| Mean (SD) | 63.0 (33.9) | 48.8 (11.5) | 0.33 | 67.3 (35.8) | 98.6 (15.7) | 1.13 |
| Median (interquartile range) | 57.1 (34.3–91.5) | 52.0 (41.8–59.2) | 62.2 (36.0–98.6) | 98.7 (87.4–110.5) | ||
| Missing, n (%) | 277 (21.9) | 0 (0.0) | 387 (23.4) | 8551 (51.8) | ||
| Estimated urine albumin-to-creatinine ratio, mg/g | ||||||
| Mean (SD) | 907.2 (1393.2) | 194.1 (656.2) | 0.67 | 871.9 (1366.8) | 20.1 (79.6) | 0.88 |
| Median (interquartile range) | 337.4 (142.4–1225.7) | 25.2 (3.6–205.9) | 305.4 (109.0–1153.9) | 0.0 (0.0–14.6) | ||
| Missing, n (%) | 269 (21.2) | 1542 (15.6) | 361 (21.9) | 10,640 (64.4) | ||
| Proteinuria (uACR) categories | 0.42 | 0.67 | ||||
| Mild (< 30 mg/g) | 128 (10.1) | 4676 (47.4) | 186 (11.3) | 5244 (31.8) | ||
| Moderate (30–299 mg/g) | 203 (16.0) | 1739 (17.6) | 284 (17.2) | 494 (3.0) | ||
| Severe (≥ 300 mg/g) | 667 (52.6) | 1907 (19.3) | 820 (49.7) | 133 (0.8) | ||
| Unknown | 269 (21.2) | 1542 (15.6) | 361 (21.9) | 10,640 (64.4) | ||
CKD, chronic kidney failure; SD, standard deviation; Std. diff., standardized difference; uACR, urine albumin-to-creatinine ratio.
Kidney Outcomes
Patients with IgAN had higher crude rates of ESKD (3.71 vs. 1.70 per 100 PY) and worsening CKD (9.34 vs. 5.14 per 100 PY), but not hospitalized AKI (2.81 vs. 3.11 per 100 PY), than patients with nonglomerular CKD (Table 2). After adjusting for potential confounders, patients with IgAN had higher adjusted rates of ESKD (aHR: 2.79, 95% CI: 2.16–3.62), worsening CKD (aHR: 3.05, 95% CI: 2.68–3.46), and hospitalized AKI (aHR: 1.45, 95% CI: 1.22–1.73) (Figure 2). These associations were driven by patients with urine albumin-to-creatinine ratio < 30 mg/g; when stratified by baseline proteinuria status, the association between IgAN and all kidney outcomes was attenuated among those with proteinuria (Supplementary Figure S1). In the propensity score-matched cohort, patients with IgAN had significantly higher adjusted rates of ESKD, worsening CKD, and hospitalized AKI, but point estimates were attenuated compared to the primary analysis (Supplementary Figure S2). Restricting the analysis to incident patients with IgAN (766 patients with IgAN, 5839 nonglomerular CKD controls) displayed similar results as the primary analysis (Supplementary Figure S3).
Table 2.
Kidney, cardiovascular, and mortality outcomes for IgAN and matched patients with nonglomerular CKD
| Outcome | N Events | Person-yrs | Rate per 100 person-yes (95% CI) | Unadjusted Hazard Ratio (95%CI) |
|---|---|---|---|---|
| Kidney outcomes | ||||
| End-stage kidney disease | ||||
| IgA Nephropathy | 259 | 6975.4 | 3.71 (3.04–4.54) | 1.92 (1.65–2.23) |
| Nonglomerular CKD | 967 | 56983.67 | 1.70 (1.59–1.81) | Ref |
| Incident or worsening chronic kidney disease | ||||
| IgA Nephropathy | 525 | 5622.7 | 9.34 (8.18–10.66) | 1.76 (1.58–1.95) |
| Nonglomerular CKD | 2554 | 49671.27 | 5.14 (4.95–5.35) | Ref |
| Hospitalized acute kidney injury | ||||
| IgA Nephropathy | 223 | 7365.2 | 2.81 (2.32–3.40) | 0.98 (0.85–1.14) |
| Nonglomerular CKD | 1734 | 55779.14 | 3.11 (2.96–3.26) | Ref |
| Cardiovascular outcomes | ||||
| Hospitalization for myocardial infarction | ||||
| IgA Nephropathy | 12 | 8183.3 | 0.15 (0.07–0.30) | 0.43 (0.24–0.76) |
| Nonglomerular CKD | 254 | 60631.93 | 0.42 (0.37–0.47) | Ref |
| Hospitalization or emergency department visit for stroke or transient ischemic attack | ||||
| IgA Nephropathy | 19 | 8157.9 | 0.23 (0.13–0.42) | 0.57 (0.36–0.92) |
| Nonglomerular CKD | 287 | 60291.23 | 0.48 (0.42–0.53) | Ref |
| Hospitalization for heart failure | ||||
| IgA Nephropathy | 13 | 8189.6 | 0.16 (0.08–0.31) | 0.43 (0.25–0.77) |
| Nonglomerular CKD | 242 | 60674.25 | 0.40 (0.35–0.45) | Ref |
| All-cause death | ||||
| IgA Nephropathy | 92 | 8224.5 | 1.12 (0.85–1.47) | 0.69 (0.56–0.87) |
| Nonglomerular CKD | 1133 | 61379.17 | 1.85 (1.74–1.96) | Ref |
CKD, chronic kidney disease.
Figure 2.
Adjusted hazard ratios for IgAN and clinical outcomes compared with matched nonglomerular CKD and non-CKD patients. ∗Adjusted for: age, self-reported race, tobacco use, previous myocardial infarction, previous stroke or transient ischemic attack, previous coronary revascularization, heart failure, diabetes mellitus, hypertension, dyslipidemia, chronic lung disease, depression, estimated glomerular filtration rate, and proteinuria (estimated urine albumin-to-creatinine ratio). Missing values for estimated glomerular filtration rate and estimated proteinuria imputed using multiple imputation across 50 datasets. ∗∗Adjusted for: age, self-reported race, tobacco use, previous atrial fibrillation or flutter, heart failure, diabetes mellitus, hypertension, dyslipidemia, chronic liver disease. CKD, chronic kidney disease; CI, confidence interval; ED, emergency department; IgAN, IgA nephropathy; TIA, transient ischemic attack.
Compared with patients who did not have CKD, patients with IgAN had significantly higher crude rates of all kidney outcomes (ESKD: 3.44 vs. 0.01 per 100 PY; worsening CKD: 8.21 vs. 0.38 per 100 PY; hospitalized AKI: 2.54 vs. 0.20 per 100 PY) (Table 3), which persisted after multivariable adjustment indicating a more than 500-fold higher adjusted hazard of ESKD (aHR: 540.16, 95% CI: 141.94–2055.61), 50-fold higher worsening CKD (aHR: 52.42, 95% CI: 40.24–68.28), and 25-fold higher hospitalized AKI (aHR: 24.79, 95% CI: 18.22–33.72) (Figure 2).
Table 3.
Kidney, cardiovascular, and mortality outcomes for IgAN and matched patients without CKD
| Outcome | N Events | Person-yrs | Rate per 100 person-yrs (95% CI) | Unadjusted hazard ratio (95% CI) |
|---|---|---|---|---|
| Kidney outcomes | ||||
| End-stage kidney disease | ||||
| IgA Nephropathy | 326 | 9487.2 | 3.44 (0.99–11.98) | 352.51 (166.52–746.23) |
| No CKD | 10 | 107,302.0 | 0.01 (0.01–0.02) | Ref |
| Incident or worsening chronic kidney disease | ||||
| IgA Nephropathy | 641 | 7808.5 | 8.21 (6.58–10.25) | 26.86 (22.74–31.71) |
| No CKD | 406 | 105,783.0 | 0.38 (0.35–0.42) | Ref |
| Hospitalized acute kidney injury | ||||
| IgA Nephropathy | 275 | 9977.3 | 2.54 (1.85–3.47) | 19.42 (15.52–24.30) |
| No CKD | 218 | 106,780.8 | 0.20 (0.18–0.23) | Ref |
| Cardiovascular outcomes | ||||
| Hospitalization for myocardial infarction | ||||
| IgA Nephropathy | 14 | 11,024.0 | 0.13 (0.06–0.28) | 2.25 (1.21–4.21) |
| No CKD | 71 | 107,043.4 | 0.07 (0.05–0.08) | Ref |
| Hospitalization/emergency department visit for stroke or transient ischemic attack | ||||
| IgA nephropathy | 26 | 10,988.2 | 0.24 (0.13–0.42) | 1.53 (0.98–2.38) |
| No CKD | 170 | 106,672.5 | 0.16 (0.14–0.19) | Ref |
| Hospitalization for heart failure | ||||
| IgA nephropathy | 18 | 11,026.7 | 0.16 (0.07–0.39) | 5.80 (3.00–11.20) |
| No CKD | 40 | 107,229.1 | 0.04 (0.03–0.05) | Ref |
| All-cause death | ||||
| IgA nephropathy | 105 | 11,078.9 | 0.95 (0.69–1.30) | 3.10 (2.43–3.96) |
| No CKD | 382 | 107,330.9 | 0.36 (0.32–0.39) | Ref |
CKD, chronic kidney disease.
Further adjustment for baseline medication use yielded similar results (Supplementary Table S2). Inclusion of recurrent hospitalized AKI events showed a higher but nonstatistically significant rate of AKI among patients who had IgAN compared with nonglomerular CKD (adjusted rate ratio: 1.23, 95% CI: 0.99–1.52; Supplementary Table S3) but did not meaningfully alter the results in the non-CKD cohort (Supplementary Table S4). Subdistribution hazard ratios accounting for competing risks of death displayed similar results as the primary analysis (Supplementary Figure S4).
Cardiovascular Outcomes and Death
Patients with IgAN had lower crude rates of hospitalization for MI (0.15 vs. 0.42 per 100 PY), hospitalization/ED visit for stroke/TIA (0.23 vs. 0.48 per 100 PY), HF hospitalization (0.16 vs. 0.40 per 100 PY), and death (1.12 vs. 1.85 per 100 PY) compared with patients who had nonglomerular CKD (Table 2). These crude differences were all attenuated and not statistically significant after multivariable adjustment (Figure 2). There were no differences in associations between IgAN and cardiovascular outcomes when stratified by proteinuria status (Supplementary Figure S1). Results were similar in the propensity score-matched cohort (Supplementary Figure S2), the incident IgAN cohort (Supplementary Figure S3), and after accounting for competing risks of death (Supplementary Figure S4). When excluding events occurring after the onset of ESKD, patients with IgAN had lower rates of HF hospitalization before ESKD; point estimates for MI and stroke were also lower but imprecise (Supplementary Figure S5).
In contrast, patients with IgAN had significantly higher crude rates of hospitalization for MI (0.13 vs. 0.07 per 100 PY), HF hospitalization (0.16 vs. 0.04 per 100 PY) and death (0.95 vs. 0.36 per 100 PY), but not hospitalization/ED visit for stroke/TIA (0.24 vs. 0.16 per 100 PY) compared with patients who did not have CKD (Table 3). After multivariable adjustment, patients with IgAN had significantly higher adjusted rates of HF hospitalization (aHR: 8.06, 95% CI: 2.90–-22.37) and death (aHR: 2.90, 95% CI: 2.08–4.02) but not for hospitalization for MI (aHR: 1.28, 95% CI: 0.55–3.00) or hospitalization/ED visit for stroke/TIA (aHR: 1.49, 95% CI: 0.89–2.51) (Figure 2). After restricting events to those before ESKD, associations with all cardiovascular outcomes were attenuated (Supplementary Figure S5).
After additional adjustment for baseline medication use, patients with IgAN displayed a significantly lower rate of hospitalization for MI compared with patients who had nonglomerular CKD (aHR: 0.41, 95% CI: 0.20–0.84) (Supplementary Table S1). All other associations in both comparison cohorts remained similar to the main results. Inclusion of recurrent events did not change results in the nonglomerular CKD cohort (Supplementary Table S3). In the non-CKD comparison cohort, patients with IgAN had a significantly higher adjusted rate of hospitalization for MI (aHR: 1.54, 95% CI: 1.13–2.10) but other associations remained similar to the main results (Supplementary Table S2).
Discussion
We found that compared with nonglomerular CKD, IgAN was associated with significantly higher rates of ESKD, worsening CKD, and episodes of AKI, with or without accounting for potential confounders, but did not experience differential rates of cardiovascular events or all-cause death. Compared with those who did not have CKD, adults with IgAN had substantially higher risk for adverse kidney outcomes as expected, but also were at higher risk for HF hospitalization and death.
Previous studies have primarily focused on kidney outcomes in persons with IgAN. For example, among a multicenter cohort of 901 adults with IgAN (Oxford derivation, North American validation studies, validation of the Oxford Classification of IgAN in a European cohort study) followed for a median of 5.6 years, Barbour et al.15 observed crude 5- and 10-year risks of a composite kidney outcome (i.e., 50% reduction in eGFR or ESKD) of 11.2% and 26.8%, respectively. Barbour et al.16 subsequently analyzed an expanded international multicenter cohort of 3927 adults with IgAN (Europe, China, Japan, and North and South America) and observed a 5-year risk of 14.7% (95% CI: 13.1%–16.3%) for the composite kidney outcome of 50% reduction in eGFR or ESKD. Pitcher et al.17 observed a kidney failure or death rate of 50% in a cohort of 2299 adults with biopsy-proven IgAN in the UK; however, this study only included patients with reduced kidney function (i.e., eGFR < 60 ml/min per 1.73 m2) resulting from IgAN. Our high observed rates of ESKD and worsening CKD are consistent with these studies, and provide additional new insights as we demonstrate the significantly higher adjusted rates of adverse kidney outcomes compared with nonglomerular CKD. We found that IgAN was associated with a higher risk of episodes of AKI compared with nonglomerular CKD even after adjusting for baseline eGFR and other AKI risk factors, consistent with more progressive CKD as a contributing factor as well as IgAN-specific pathways (e.g., rapidly progressive glomerulonephritis, acute tubular necrosis related to red blood cell cast obstruction or heme toxicity, or postinfectious proliferative glomerulonephritis in the setting of diabetes or other comorbidities).18, 19, 20, 21 Reduced eGFR and proteinuria are known risk factors for AKI, and our study provides new insights about the excess risk of AKI with IgAN compared with nonglomerular CKD.
In contrast, evaluation of cardiovascular risks in those with IgAN compared with other forms of CKD or no CKD has been more limited. Among 658,168 US patients with ESKD initiating chronic dialysis between 1997 and 2014, O’Shaughnessy et al.22 compared the rate of a composite cardiovascular event (MI, ischemic stroke, or cardiovascular or cerebrovascular death). They found that compared with ESKD attributed to IgAN, adjusted rates of the composite cardiovascular event were significantly higher than for ESKD attributed to diabetes, lupus nephritis, focal segmental glomerulosclerosis, membranous nephropathy, membranoproliferative glomerulonephritis, vasculitis, or autosomal dominant polycystic kidney disease.22 However, these results do not necessarily generalize to less severe CKD, and we found that after adjustment for a wide range of confounders, there was no significantly higher rate of MI, HF, or stroke/TIA for IgAN compared with nonglomerular CKD. In our study, not surprisingly, IgAN was linked to higher adjusted rates of hospitalization for HF compared with no CKD given the known excess risk of HF associated with reduced eGFR and/or proteinuria23,24; however, IgAN was not associated with significantly different rates of MI or stroke/TIA. The latter finding is in contrast to the report from Jarrick et al.25 who compared 3945 patients with IgAN with 19,272 age-sex-matched adults in the general population in Sweden and observed a higher rate of ischemic heart disease (aHR 1.86, 95% CI:1.63–2.13), but only adjusted for education, country of birth, cancer, diabetes, and systemic inflammatory diseases. Canney et al.26 examined a cohort of adults with primary glomerular diseases from British Columbia, Canada (2000–2012) and reported a higher age-sex-adjusted rate (standardized incidence ratio 1.38, 95% CI:1.01–1.85) of a composite outcome of coronary artery, cerebrovascular and peripheral vascular events, and death attributed to MI or stroke for IgAN compared with the general population, but the study did not account for other explanatory variables nor was it powered to evaluate individual types of cardiovascular events.
With regards to mortality, the vast majority of studies only examined cohorts of patients with IgAN with no comparison group. Jarrick et al.27 reported that during median follow-up of 13.6 years, there was a 1.74-fold higher adjusted risk of all-cause death in 3622 adults with IgAN (mean age 34.9 years) diagnosed between 1974–2011 compared with 18,041 general population controls in Sweden who were matched on age, sex, calendar year, and county of residence, although cardiovascular morbidity was not adjusted for in the model. In contrast, we found a nearly 3-fold higher adjusted risk of death for IgAN compared with no CKD but no significant difference compared with nonglomerular CKD after accounting for sociodemographic characteristics, cardiovascular and noncardiovascular comorbidity, and other potential confounders.
IgAN involves excess inflammation (with mediating factors such as transforming growth factor β and interleukin-6) and cellular proliferation that contribute to glomerular and interstitial fibrosis,28,29 which promotes the development and more rapid progression of CKD and its associated kidney and selected cardiovascular complications. IgAN may also lead to arterial stiffness as well as lower responsiveness to angiotensin II, suggesting excess renin-angiotensin system activity,30 that collectively may contribute to enhanced cardiovascular and mortality risk independent of reductions in eGFR or proteinuria.
Our study was strengthened by inclusion of a population-based sample of biopsy-proven IgAN based on a validated NLP algorithm applied to semistructured and unstructured EHR data that did not rely on administrative billing codes. Our study cohorts were demographically diverse and represented recent kidney, cardiovascular, and survival outcomes among IgAN and matched nonglomerular CKD and non-CKD adults derived from the same source population. Our study also had limitations. Despite the high accuracy of our NLP algorithm of identifying biopsy-proven IgAN, there may be some misclassification of patients with CKD and IgAN because of incomplete EHR data as some patients may receive biopsies outside of KPNC and their results may not have been available. The assigned index date of IgAN may not necessarily reflect the true biological start date of IgAN, potentially introducing bias because of left truncation of follow-up time, although our sensitivity analyses of presumed incident patients with IgAN were consistent with the main study findings. However, our findings are most applicable to follow-up after clinically-recognized IgAN within our healthcare system rather than biological initiation of IgAN. We were also unable to distinguish between primary and secondary IgAN, and detailed histopathological findings (e.g., consensus-based Oxford MEST-C classification31,32) were not available which precluded evaluation of these findings with subsequent risks of kidney, cardiovascular, and mortality outcomes. Information on albuminuria was missing in a significant proportion of patients. As this is a retrospective cohort study, observed associations could be because of residual confounding effects, rather than the exposure of IgAN itself. However, in addition to hard matching on age and sex, we accounted for potential differences in a broad spectrum of potential confounders as well as receipt of cardiovascular and renoprotective therapies. Our findings were also robust in a variety of sensitivity analyses. Kidney function and proteinuria may also be considered mediators of the relationship between IgAN and certain clinical outcomes and may not necessarily be appropriate adjustment factors in models. Although some urgent or emergent care may be provided by out-of-network providers, it is systematically captured through administrative/billing claims databases with our health care delivery system, so clinical outcomes should be comprehensively identified.33, 34, 35 Finally, despite the diverse, contemporary study population, results may not fully generalize to uninsured populations or all practice settings.
In sum, IgAN was independently associated with worse kidney outcomes compared with nonglomerular CKD, and also linked to excess rates of HF hospitalizations and all-cause death compared with no CKD. Beyond maximizing use of currently available renoprotective and cardioprotective therapies, additional research is needed to identify potential novel modifiable factors as well as more accurate risk prediction to tailor more personalized preventive strategies.
Disclosure
ASG declares having received research funding from the National Institute of Diabetes, Digestive and Kidney Diseases; National Heart, Lung, and Blood Institute; Novartis; Bristol-Myers Squibb; and Edward Life Sciences. All other authors have nothing to disclose.
Acknowledgments
Funding for this styudy was provided by the Novartis Pharmaceutical Corporation. The funders had no role in the design of the study, data collection, data analysis, interpretation of data, the decision to submit results for publication, or in the preparation, review or approval of the manuscript.
Data Sharing Statement
The data that support the findings of this study are available from the corresponding author upon reasonable request. Access to deidentified individual participant data and statistical analysis plan will be provided to qualified researchers for purposes of replicating the results or conducting further analyses. Data sharing is subject to approval by the institutional review board and compliance with applicable data protection regulations. Requests should be directed to ASG at alan.s.go@kp.org.
Footnotes
Figure S1. Adjusted hazard ratios for IgAN and clinical outcomes compared with matched nonglomerular CKD patients, stratified by baseline albuminuria status.
Figure S2. Association of IgAN with clinical outcomes compared with propensity score-matched nonglomerular CKD patients (n=883 matched pairs).
Figure S3. Association of presumed incident IgAN (n=766) with clinical outcomes compared with matched patients who had nonglomerular CKD (n=5836).
Figure S4. Adjusted subdistribution hazard ratios for IgAN and clinical outcomes compared with matched patients with nonglomerular CKD, accounting for the competing risk of death.
Figure S5. Adjusted hazard ratios for IgAN and clinical outcomes compared to matched patients with nonglomerular CKD and matched patients without CKD, with additional censoring at ESKD.
Table S1. Characteristics of propensity score-matched patients with IgAN and nonglomerular CKD.
Table S2. Multivariable association of IgAN with clinical outcomes compared with matched patients with nonglomerular CKD and with matched patients without CKD, after additional adjustment for medication use at baseline.
Table S3. Crude outcomes and adjusted associations from Poisson models for nonfatal recurrent events for patients with IgAN and matched patients with nonglomerular CKD.
Table S4. Crude outcomes and adjusted associations from Poisson models for nonfatal recurrent events for patients with IgAN with matched patients without CKD.
STROBE Checklist.
Supplementary Material
Figure S1. Adjusted hazard ratios for IgAN and clinical outcomes compared with matched nonglomerular CKD patients, stratified by baseline albuminuria status. Figure S2. Association of IgAN with clinical outcomes compared with propensity score-matched nonglomerular CKD patients (n = 883 matched pairs). Figure S3. Association of presumed incident IgAN (n = 766) with clinical outcomes compared with matched patients who had nonglomerular CKD (n = 5836). Figure S4. Adjusted subdistribution hazard ratios for IgAN and clinical outcomes compared with matched patients with nonglomerular CKD, accounting for the competing risk of death. Figure S5. Adjusted hazard ratios for IgAN and clinical outcomes compared to matched patients with nonglomerular CKD and matched patients without CKD, with additional censoring at ESKD. Table S1. Characteristics of propensity score-matched patients with IgAN and nonglomerular CKD. Table S2. Multivariable association of IgAN with clinical outcomes compared with matched patients with nonglomerular CKD and with matched patients without CKD, after additional adjustment for medication use at baseline. Table S3. Crude outcomes and adjusted associations from Poisson models for nonfatal recurrent events for patients with IgAN and matched patients with nonglomerular CKD. Table S4. Crude outcomes and adjusted associations from Poisson models for nonfatal recurrent events for patients with IgAN with matched patients without CKD. STROBE Checklist.
References
- 1.McGrogan A., Franssen C.F., de Vries C.S. The incidence of primary glomerulonephritis worldwide: a systematic review of the literature. Nephrol Dial Transplant. 2011;26:414–430. doi: 10.1093/ndt/gfq665. [DOI] [PubMed] [Google Scholar]
- 2.Gutierrez E., Zamora I., Ballarin J.A., et al. Long-term outcomes of IgA nephropathy presenting with minimal or no proteinuria. J Am Soc Nephrol. 2012;23:1753–1760. doi: 10.1681/ASN.2012010063. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Lv J., Yang Y., Zhang H., et al. Prediction of outcomes in crescentic IgA nephropathy in a multicenter cohort study. J Am Soc Nephrol. 2013;24:2118–2125. doi: 10.1681/ASN.2012101017. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Myllymaki J., Syrjanen J., Helin H., Pasternack A., Kattainen A., Mustonen J. Vascular diseases and their risk factors in IgA nephropathy. Nephrol Dial Transplant. 2006;21:1876–1882. doi: 10.1093/ndt/gfl062. [DOI] [PubMed] [Google Scholar]
- 5.Knoop T., Vikse B.E., Svarstad E., Leh S., Reisæter A.V., Bjørneklett R. Mortality in patients with IgA nephropathy. Am J Kidney Dis. 2013;62:883–890. doi: 10.1053/j.ajkd.2013.04.019. [DOI] [PubMed] [Google Scholar]
- 6.Koebnick C., Langer-Gould A.M., Gould M.K., et al. Sociodemographic characteristics of members of a large, integrated health care system: comparison with US Census Bureau data. Permanente J. 2012;16:37–41. doi: 10.7812/TPP/12-031. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Pravoverov L.V., Zheng S., Parikh R., et al. Trends associated with large-scale expansion of peritoneal dialysis within an integrated care delivery model. JAMA Intern Med. 2019;179:1537–1542. doi: 10.1001/jamainternmed.2019.3155. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Milward D., Bjareland M., Hayes W., et al. Ontology-based interactive information extraction from scientific abstracts. Comp Funct Genomics. 2005;6:67–71. doi: 10.1002/cfg.456. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Cormack J., Nath C., Milward D., Raja K., Jonnalagadda S.R. Agile text mining for the 2014 i2b2/UTHealth Cardiac risk factors challenge. J Biomed Inform. 2015;58(suppl):S120–S127. doi: 10.1016/j.jbi.2015.06.030. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Go A.S., Magid D.J., Wells B., et al. The cardiovascular Research Network: a new paradigm for cardiovascular quality and outcomes research. Circ Cardiovasc Qual Outcomes. 2008;1:138–147. doi: 10.1161/CIRCOUTCOMES.108.801654. [DOI] [PubMed] [Google Scholar]
- 11.Ross T.R., Ng D., Brown J.S., et al. The HMO research network virtual data warehouse: A public data model to support collaboration. EGEMs (Wash DC) 2014;2:1049. doi: 10.13063/2327-9214.1049. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Inker L.A., Eneanya N.D., Coresh J., et al. New creatinine- and cystatin C-based equations to estimate GFR without race. N Engl J Med. 2021;385:1737–1749. doi: 10.1056/NEJMoa2102953. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Sumida K., Nadkarni G.N., Grams M.E., et al. Conversion of urine protein-creatinine ratio or urine dipstick protein to urine albumin-creatinine ratio for use in chronic kidney disease screening and prognosis : an individual participant-based meta-analysis. Ann Intern Med. 2020;173:426–435. doi: 10.7326/M20-0529. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Curb J.D., Ford C.E., Pressel S., Palmer M., Babcock C., Hawkins C.M. Ascertainment of vital status through the National Death Index and the Social Security Administration. Am J Epidemiol. 1985;121:754–766. doi: 10.1093/aje/121.5.754. [DOI] [PubMed] [Google Scholar]
- 15.Barbour S.J., Espino-Hernandez G., Reich H.N., et al. The MEST score provides earlier risk prediction in lgA nephropathy. Kidney Int. 2016;89:167–175. doi: 10.1038/ki.2015.322. [DOI] [PubMed] [Google Scholar]
- 16.Barbour S.J., Coppo R., Zhang H., et al. Evaluating a new international risk-prediction tool in IgA nephropathy. JAMA Intern Med. 2019;179:942–952. doi: 10.1001/jamainternmed.2019.0600. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Pitcher D., Braddon F., Hendry B., et al. Long-term outcomes in IgA nephropathy. Clin J Am Soc Nephrol. 2023;18:727–738. doi: 10.2215/CJN.0000000000000135. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Nasr S.H., Markowitz G.S., Whelan J.D., et al. IgA-dominant acute poststaphylococcal glomerulonephritis complicating diabetic nephropathy. Hum Pathol. 2003;34:1235–1241. doi: 10.1016/S0046-8177(03)00424-6. [DOI] [PubMed] [Google Scholar]
- 19.Szeto C.C., Lai F.M., To K.F., et al. The natural history of immunoglobulin A nephropathy among patients with hematuria and minimal proteinuria. Am J Med. 2001;110:434–437. doi: 10.1016/S0002-9343(01)00659-3. [DOI] [PubMed] [Google Scholar]
- 20.Sevillano A.M., Diaz M., Caravaca-Fontán F., et al. IgA nephropathy in elderly patients. Clin J Am Soc Nephrol. 2019;14:1183–1192. doi: 10.2215/CJN.13251118. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Zhang L., Li J., Yang S., et al. Clinicopathological features and risk factors analysis of IgA nephropathy associated with acute kidney injury. Ren Fail. 2016;38:799–805. doi: 10.3109/0886022X.2016.1163153. [DOI] [PubMed] [Google Scholar]
- 22.O’Shaughnessy M.M., Liu S., Montez-Rath M.E., Lafayette R.A., Winkelmayer W.C. Cause of kidney disease and cardiovascular events in a national cohort of US patients with end-stage renal disease on dialysis: a retrospective analysis. Eur Heart J. 2019;40:887–898. doi: 10.1093/eurheartj/ehy422. [DOI] [PubMed] [Google Scholar]
- 23.Go A.S., Chertow G.M., Fan D., McCulloch C.E., Hsu C.Y. Chronic kidney disease and the risks of death, cardiovascular events, and hospitalization. N Engl J Med. 2004;351:1296–1305. doi: 10.1056/NEJMoa041031. [DOI] [PubMed] [Google Scholar]
- 24.Bansal N., Zelnick L., Bhat Z., et al. Burden and outcomes of heart failure hospitalizations in adults with chronic kidney disease. J Am Coll Cardiol. 2019;73:2691–2700. doi: 10.1016/j.jacc.2019.02.071. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Jarrick S., Lundberg S., Sundström J., Symreng A., Warnqvist A., Ludvigsson J.F. Immunoglobulin A nephropathy and ischemic heart disease: a nationwide population-based cohort study. BMC Nephrol. 2021;22:165. doi: 10.1186/s12882-021-02353-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Canney M., Gunning H.M., Zheng Y., et al. The risk of cardiovascular events in individuals with primary glomerular diseases. Am J Kidney Dis. 2022;80:740–750. doi: 10.1053/j.ajkd.2022.04.005. [DOI] [PubMed] [Google Scholar]
- 27.Jarrick S., Lundberg S., Welander A., et al. Mortality in IgA nephropathy: A nationwide population-based cohort study. J Am Soc Nephrol. 2019;30:866–876. doi: 10.1681/ASN.2018101017. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Rauen T., Floege J. Inflammation in IgA nephropathy. Pediatr Nephrol. 2017;32:2215–2224. doi: 10.1007/s00467-017-3628-1. [DOI] [PubMed] [Google Scholar]
- 29.Zheng N., Xie K., Ye H., et al. TLR7 in B cells promotes renal inflammation and Gd-IgA1 synthesis in IgA nephropathy. JCI Insight. 2020;5 doi: 10.1172/jci.insight.136965. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Abdi-Ali A., Mann M.C., Hemmelgarn B.R., et al. IgA nephropathy with early kidney disease is associated with increased arterial stiffness and renin-angiotensin system activity. J Renin Angiotensin Aldosterone Syst. 2015;16:521–528. doi: 10.1177/1470320313510586. [DOI] [PubMed] [Google Scholar]
- 31.Trimarchi H., Barratt J., Cattran D.C., et al. Oxford Classification of IgA nephropathy 2016: an update from the IgA Nephropathy Classification Working Group. Kidney Int. 2017;91:1014–1021. doi: 10.1016/j.kint.2017.02.003. [DOI] [PubMed] [Google Scholar]
- 32.Working Group of the International Ig ANN. Cattran D.C., Coppo R., et al. The Oxford classification of IgA nephropathy: rationale, clinicopathological correlations, and classification. Kidney Int. 2009;76:534–545. doi: 10.1038/ki.2009.243. [DOI] [PubMed] [Google Scholar]
- 33.Goldberg R.J., Gurwitz J.H., Saczynski J.S., et al. Comparison of medication practices in patients with heart failure and preserved versus those with reduced ejection fraction (from the Cardiovascular Research Network [CVRN]) Am J Cardiol. 2013;111:1324–1329. doi: 10.1016/j.amjcard.2013.01.276. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Go A.S., Yang J., Ackerson L.M., et al. Hemoglobin level, chronic kidney disease, and the risks of death and hospitalization in adults with chronic heart failure: the anemia in Chronic Heart Failure: outcomes and Resource Utilization (Anchor) Study. Circulation. 2006;113:2713–2723. doi: 10.1161/CIRCULATIONAHA.105.577577. [DOI] [PubMed] [Google Scholar]
- 35.Go A.S., Lee W.Y., Yang J., Lo J.C., Gurwitz J.H. Statin therapy and risks for death and hospitalization in chronic heart failure. JAMA. 2006;296:2105–2111. doi: 10.1001/jama.296.17.2105. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Figure S1. Adjusted hazard ratios for IgAN and clinical outcomes compared with matched nonglomerular CKD patients, stratified by baseline albuminuria status. Figure S2. Association of IgAN with clinical outcomes compared with propensity score-matched nonglomerular CKD patients (n = 883 matched pairs). Figure S3. Association of presumed incident IgAN (n = 766) with clinical outcomes compared with matched patients who had nonglomerular CKD (n = 5836). Figure S4. Adjusted subdistribution hazard ratios for IgAN and clinical outcomes compared with matched patients with nonglomerular CKD, accounting for the competing risk of death. Figure S5. Adjusted hazard ratios for IgAN and clinical outcomes compared to matched patients with nonglomerular CKD and matched patients without CKD, with additional censoring at ESKD. Table S1. Characteristics of propensity score-matched patients with IgAN and nonglomerular CKD. Table S2. Multivariable association of IgAN with clinical outcomes compared with matched patients with nonglomerular CKD and with matched patients without CKD, after additional adjustment for medication use at baseline. Table S3. Crude outcomes and adjusted associations from Poisson models for nonfatal recurrent events for patients with IgAN and matched patients with nonglomerular CKD. Table S4. Crude outcomes and adjusted associations from Poisson models for nonfatal recurrent events for patients with IgAN with matched patients without CKD. STROBE Checklist.