Researchers developed and validated a prediction tool for estimating the risk of incident hypothyroidism in a representative, contemporary cohort of US adults with moderate-to-advanced chronic kidney disease (CKD), defined as an estimated glomerular filtration rate (eGFR) of less than 30 mL/min/1.73m2 (stage 4-5 CKD).
Overall study cohort consisted of 15,642 adults with stage 4-5 CKD taken from the Optum Labs Data Warehouse of more than 200 million diverse US adults with commercial or Medicare Advantage health insurance.
Included patients had a baseline thyroid stimulating hormone (TSH) level in the normal range (0.5-5.0 mIU/L); incident hypothyroid cases were those with a subsequent TSH level > 5.0 mIU/L.
During median 3.4 years follow-up, 1650 (10.5%) of the patients developed incident hypothyroidism.
Researchers used two thirds of the cohort (10,428) to derive the prediction model, and the remaining third (5214) to perform internal validation.
White race and ethnicity, higher baseline TSH, hypertension, congestive heart failure, receipt of an angiogram or computed tomography (CT) scan with iodinated contrast, and amiodarone use were each associated with higher risk for incident hypothyroidism, while non-Hispanic Black race and ethnicity, higher body mass index (> 30 kg/m2), and higher serum albumin (≥ 4.0 g/dL) were each associated with lower risk for incident hypothyroidism.
Among these factors, higher baseline TSH level and amiodarone use linked with the largest increases in the risk for incident hypothyroidism.
The model’s discrimination was good, producing similar C-statistics in both the development dataset 0.77, and validation dataset 0.76, indicating that the model accounts for about 76%-77% of the cause of incident hypothyroidism in these patients.
This "convenient risk-prediction tool can inform clinical management of these patients by identifying those who warrant prioritized screening, serial monitoring, and long-term treatment" of hypothyroidism and could potentially be embedded for automated analysis in an electronic health record (EHR), say the report’s authors. "The findings have the potential to improve the quality of care of CKD patients."
Researchers primarily affiliated with the University of California, Irvine, ran the study. They published their report in the Journal of Clinical Endocrinology and Metabolism.
Study cohort included patients with moderate-to-advanced CKD, hence the model should be used cautiously for patients with milder kidney dysfunction.
Model’s performance may depend on the accuracy of data on comorbidities, procedures, and laboratory findings.
Data used for model development and validation ascertainment of incident hypothyroidism relied entirely on serum levels of TSH.
Ascertainment of incident hypothyroidism did not take into account subsequent measures of TSH, and it is possible that patients with modestly aberrant TSH levels at baseline may have later reverted to normal levels.
Study data came from a retrospective review of claims and EHR data that did not include information on the indications for TSH testing.
Thyroid autoimmunity is a known risk factor for hypothyroidism in the general population but limited measurement of anti-thyroid peroxidase antibodies in the study cohort precluded using this covariate in the prediction model.
The authors cannot not rule out that some relevant clinical data, including thyroid-function test results, were not measured or recorded because they were in claims not included in the Optum dataset.
The study received no commercial funding. None of the authors report relevant financial relationships.
Mitchel L. Zoler is a reporter with Medscape and MDedge based in the Philadelphia region. @mitchelzoler
Lead image: One Photo/Dreamstime
Medscape Medical News © 2023
Cite this: Predicting Hypothyroid Onset in Adults With Kidney Disease - Medscape - Jun 14, 2023.