Data from: External validation of an electronic health record-based diagnostic model for histological acute tubulointerstitial nephritis
Data files
Dec 12, 2024 version files 64.67 KB
-
ain-model-dryad.xlsx
59.86 KB
-
README.md
4.80 KB
Abstract
Background
Accurate diagnosis of acute tubulointerstitial nephritis (AIN) often requires a kidney biopsy. We previously developed a diagnostic statistical model for predicting biopsy-confirmed AIN by combining four laboratory tests after evaluating over 150 potential predictors from the electronic health record. Here, we validate this diagnostic model in two biopsy-based cohorts at Johns Hopkins Hospital (JHH) and Yale, which were geographically and temporally distinct from the development cohort, respectively.
Methods
We analyzed patients who underwent kidney biopsy at JHH and Yale University (2019-2023). We assessed discrimination (AUC) and calibration using previously derived model coefficients and recalibrated the model using an intercept correction factor that accounted for differences in baseline prevalence of AIN between development and validation cohorts.
Results
We included 1982 participants: 1454 at JHH and 528 at Yale. JHH (5%) and Yale (17%) had lower proportions of biopsies with AIN than the development set (23%). The AUC was 0.73 (0.66-0.79) at JHH and 0.73 (0.67-0.78) at Yale, similar to the development set (0.73 (0.64-0.81)). Calibration was imperfect in validation cohorts, particularly at JHH, but improved with application of intercept correction factor. The model increased AUC of clinicians' prebiopsy suspicion for AIN by 0.10 to 0.77 (0.71-0.82).
Conclusion
AIN diagnostic model retained discrimination in two validation cohorts but needed recalibration to account for local AIN prevalence. The model improved clinicians’ ability to predict AIN.
README: External validation of an electronic health record-based diagnostic model for histological acute tubulointerstitial nephritis
Please see accompanying manuscript for detailed methods: https://pubmed.ncbi.nlm.nih.gov/39500309/
https://doi.org/10.5061/dryad.3bk3j9kvf
Description of the data and file structure
Study design and participants: IRB-approved prospective study (HIC2000027890) enrolled participants who were scheduled to undergo a clinical kidney biopsy between July 2020 and June 2023 and consented to participate in the Yale Kidney Biobank (NCT04343417).
Data sources: We manually reviewed the electronic health record for clinician notes before the time of biopsy for pre-biopsy diagnosis. We classified the prebiopsy diagnosis as AIN if this was noted as a prebiopsy diagnosis in any position. In a sensitivity analysis at Yale, we classified the prebiopsy diagnosis as AIN only when this was the most likely suspected diagnosis. When clinicians’ prebiopsy suspicion for AIN was unavailable through chart review (n=21, 4%), we assumed that the clinicians did not suspect AIN. We also reviewed clinician notes after biopsy to determine the attributed causes of AIN.
AIN diagnostic model: The AIN diagnostic model provides a probability of AIN using four clinically available variables obtained before kidney biopsy: serum creatinine, blood urea nitrogen to creatinine ratio, urine protein, and urine specific gravity. When multiple values were available, we used the value available closest to the time of kidney biopsy.
Diagnosis of Acute Interstitial Nephritis (1=AIN, 0=Not AIN): We classified as an AIN case those whose diagnosis included either “interstitial nephritis” or “tubulointerstitial nephritis” and classified the rest as controls. We classified as controls those with AIN accompanying a primary glomerulonephritis diagnosis since the interstitial infiltrate in these cases would be expected to be secondary to the glomerulonephritis. We evaluated diagnosis listed on any position in the final diagnosis section of the biopsy report.
Variables
- Urine specific gravity: This is a measure of the density of urine compared to the density of water. For each patient, this value was determined by performing an dipstick urinalysis test on a urine sample most commonly acquired before the kidney biopsy.
- Protein present in urine reported as 0 if >=2+ or 1 if <=1+ on using dipstick. This is a measure of the amount of protein in the urine. For each patient, this value was determined by performing an dipstick urinalysis test on a urine sample most commonly acquired before the kidney biopsy.
- Diagnosis of Acute Interstitial Nephritis (1=AIN, 0=Not AIN): We classified as an AIN case those whose diagnosis included either “interstitial nephritis” or “tubulointerstitial nephritis” and classified the rest as controls. We classified as controls those with AIN accompanying a primary glomerulonephritis diagnosis since the interstitial infiltrate in these cases would be expected to be secondary to the glomerulonephritis. We evaluated diagnosis listed on any position in the final diagnosis section of the biopsy report.
- Creatinine closest to biopsy: This is a measure of the level of creatinine in the blood of each participant prior to the biopsy procedure. These measurements are taken from the electronic health record (EHR) and values are taken from as close to the kidney biopsy procedure as possible.
- Blood urea nitrogen (BUN) to creatinine ratio closest to biopsy: This calculation is derived from dividing the BUN level in a laboratory result by the creatinine level. The BUN and creatinine for each patient is extracted from the electronic health record (EHR), as closer to the kidney biopsy procedure as possible.
- Coefficient given by model 1: output of logistic regression model of AIN diagnsotic index
- Probability of acute interstitial nephritis (AIN) diagnosis given by model 1 : before recalibration
- Coefficient given by model 2: output of logistic regression model of AIN diagnsotic index after recalibration
- Probability of acute interstitial nephritis (AIN) diagnosis given by model 2 : after recalibration
- Medication causing AIN, if present: Noted as n/a if could not be determined.
- Biopsy proven diagnosis: Diagnosis on biopsy labelled as below:
- ATN Acute tubular necrosis/injury
- AIN Acute interstitial nephritis
- DKD Diabetic Kidney Disease
- Fibrosis Fibrosis
- OTHER GN Other glomerulonephritis
- IGA Immunoglobulin A Nephropathy
- FSGS Focal and Segmental Glomerulosclerosis
- LUPUS Lupus Nephritis
- Prebiopsy AIN - clinical impression: obtained by chart review. 1 if AIN, 0 if not AIN.