Identification of Helicobacter pylori infected patients, using administrative data

S. Thirumurthi, R. Desilva, D. L. Castillo, P. Richardson, N. S. Abraham

Research output: Contribution to journalArticlepeer-review

11 Scopus citations


Background: Helicobacter pylori is a prevalent organism implicated in peptic ulcer disease. Aim: To validate administrative data for diagnosis of H. pylori-infected patients. Methods: Administrative data identified patients with ICD-9 code for H. pylori (041.86) or prescription of eradication therapy; diagnosis was confirmed by chart abstraction. Multivariable regression assessed predictors of infection considering drug therapy, ICD-9 code 041.86, procedure code, in-patient or out-patient diagnostic code, age, gender and race to generate an algorithm for validation. Results: The test cohort of 531 patients (361 potential cases; 170 random controls) was primarily male (94%), Caucasian (59%) and elderly [67 years (s.d. 10)]. The positive predictive value (PPV) of ICD-9 code 041.86 was 100% and 97.4% if from an in-patient or out-patient encounter, respectively. Eradication drug therapy had a PPV of 73.7% (triple therapy) and 97.7% (quadruple therapy). The strongest predictors were out-patient ICD-9 code 041.86 (OR 8.1; 95% CI: 7.0-9.1); eradication drug therapy (OR 7.4; 95% CI: 6.6-8.3); oesophagogastroduodenoscopy (OR 3.5; 95% CI: 3.3-3.6); and age ≥70 (OR 1.2; 95% CI: 1.1-1.4). An algorithm including these data elements yielded a c-statistic of 0.93 and PPV of 97.9%. Conclusions: Administrative data can diagnose H. pylori-infected patients. The diagnostic algorithm includes presence of eradication drug therapy overlapping with an out-patient ICD-9 code 041.86 among elderly adults.

Original languageEnglish (US)
Pages (from-to)1309-1316
Number of pages8
JournalAlimentary Pharmacology and Therapeutics
Issue number11-12
StatePublished - Dec 2008

ASJC Scopus subject areas

  • Hepatology
  • Gastroenterology
  • Pharmacology (medical)


Dive into the research topics of 'Identification of Helicobacter pylori infected patients, using administrative data'. Together they form a unique fingerprint.

Cite this