Identifying drug combinations associated with acute kidney injury using association rules method

Prasad Nishtala, Te-yuan Chyou

Research output: Contribution to journalArticlepeer-review

17 Citations (SciVal)
56 Downloads (Pure)


Background: Older adults are at an increased risk of acute kidney injury (AKI) because of aging, multiple comorbidities, and polypharmacy. Objectives: The aim of this case-crossover study was to apply association rule (AR) analysis to ascertain drug combinations contributing to the risk of AKI in adults aged 65 years and older. Methods: We sourced a nationwide representative sample of New Zealanders aged ≥65 years from the pharmaceutical collections and hospital discharge information. Prescription records (2005-2015) of drugs of interest were sourced from New Zealand pharmaceutical collections (Pharms). We classified medication exposure, as a binary variable, at individual drug level belonging to medication classes including antimicrobials, antihistamines, diuretics, opioids, nonsteroidal anti-inflammatory medications. Several studies have associated the drugs of interest from these medication classes with AKI in older adults. We extracted the first-time coded diagnosis of AKI from the National Minimal Data Set. A unique patient identifier linked the prescription data set to the event data set, to set up a case-crossover cohort, indexed at the first AKI event. ARs were then applied to identify frequent drug combinations in the case and the control periods (l-day observation with a 35-day washout period), and the association of AKI with each frequent drug combination was tested by computing a matched odds ratio (MOR) and its 95% confidence interval (CI). Results: We identified 55 747 individuals (mean age 82.14) from 2005 to 2014 with incident AKI and exposed to at least one of the drugs of interest. ARs identified several medication classes including antimicrobials, nonsteroidal anti-inflammatory drugs, and opioids are associated with AKI. The frequently used medicines associated with AKI are trimethoprim (MOR = 1.68; 95% CI = [1.54-1.80]), ondansetron (MOR = 1.43; 95% CI = [1.25-1.64]), codeine phosphate plus metoclopramide (MOR = 1.37; 95% CI = [1.11-1.63]), and norfloxacin (MOR = 1.24; 95% CI [1.05-1.42]). Conclusions: We applied ARs, a novel methodology, to big data to ascertain drug combinations associated with AKI. ARs uncovered previously implicated medication classes that increase the risk of AKI in older adults. The finding that ondansetron increases the risk of AKI requires further investigation.

Original languageEnglish
Pages (from-to)467-473
Number of pages7
JournalPharmacoepidemiology and Drug Safety
Issue number4
Early online date20 Feb 2020
Publication statusPublished - 1 Apr 2020

Bibliographical note

Funding Information:
The authors would like to thank the Analytical Services, Ministry of Health of New Zealand for providing the data sets.

Publisher Copyright:
© 2020 John Wiley & Sons Ltd


  • acute kidney injury
  • association rules
  • case-crossover design
  • older people
  • pharmacoepidemiology
  • pharmacovigilance

ASJC Scopus subject areas

  • Epidemiology
  • Pharmacology (medical)


Dive into the research topics of 'Identifying drug combinations associated with acute kidney injury using association rules method'. Together they form a unique fingerprint.

Cite this