CoChat
Sign Up
Cited by 602Medical Coding and Health InformationElectronic Health Records SystemsStatistical Methods in Epidemiology

Validation studies of the health improvement network (THIN) database for pharmacoepidemiology research

Lewis, James D., Schinnar, Rita, Bilker, Warren B., Wang, Xingmei, Strom, Brian L.·2007·Pharmacoepidemiology and Drug Safety
DOI 10.1002/pds.1335
Open this paper in CoChat — free account

Abstract

Background: The Health Improvement Network (THIN) is a new medical records database that contains records from general practices some of which have or continue to participate in the General Practice Research Database (GPRD) and others that never participated in GPRD. We sought to replicate in THIN well‐established associations from the medical literature and to compare results from the GPRD practices to the non‐GPRD practices within THIN. Methods: Using THIN data from 1986–2003, we conducted case‐control studies of associations between diseases (e.g., hypertension and stroke) and between diseases and drugs (e.g., aspirin and colon cancer). Conditional logistic regression was used to calculate odds ratios adjusted for potential confounders. Differences between GPRD and non‐GPRD practices were assessed by testing for a statistical interaction by practice type in each outcome‐exposure association. Results: We observed the expected positive associations (p < 0.05) of stroke with hypertension and diabetes mellitus; of myocardial infarction with hypertension, hypercholesterolemia, obesity, and smoking; and of peptic ulcer disease with aspirin, NSAIDs, and potassium. We observed the expected negative associations (p < 0.05) of colorectal cancer with aspirin, NSAIDs, and cox‐2 inhibitors. The expected protective effect of aspirin use for myocardial infarction was not observed. In all cases, the results obtained from the GPRD practices were similar to the results obtained from the non‐GPRD practices, only being statistically different for the associations of myocardial infarction with diabetes and aspirin use. Conclusions: THIN data that are collected outside of the GPRD appear as valid as the data collected as part of the GPRD. Copyright © 2006 John Wiley & Sons, Ltd.

Integrity signals

Registry

Crossref · 10.1002/pds.1335

Cites

12 of 12 references verified0 citation errors · 0 not checkable

Statistics

Recomputing p-values needs the paper’s full text.

Claims the paper makes

This paper’s full text isn’t openly available, so its claims and self-check weren’t run. The registry and citation checks still apply.

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

  • The exposure groups (bipolar disorder and schizophrenia) and the outcomes (mortality, CVD deaths, CVD, suicide and self-harm) have been well validated.

    “The exposure groups (bipolar disorder and schizophrenia)44,45 and the outcomes (mortality, CVD deaths, CVD, suicide and self-harm)46,47,50,51 have been well validated.”

    From Mortality gap for people with bipolar disorder and schizophrenia: UK-based cohort study 2000–2014 (2017) · cites this paper as [51]

  • Validation of identification methods has been increasingly emphasized as essential for studies using routinely collected health data, particularly for disease codes in studies using administrative data collected for billing purposes.

    “Validation of identification methods has been increasingly emphasized as essential for studies using routinely collected health data, particularly for disease codes in studies using administrative data collected for billing purposes [33].”

    From The Reporting of Studies Conducted Using Observational Routinely-Collected Health Data (RECORD) statement. (2015) · cites this paper as [33]

Cites

Flagged first, then by citations

Cited by

Sorted by citations

Related

Authors
Lewis, J. D., Schinnar, R., Bilker, W. B., Wang, X., & Strom, B. L.
Affiliations
University of Pennsylvania (US)
Corresponding author
Lewis, James D.
ORCID
Venue
Pharmacoepidemiology and Drug Safety
Publisher
Wiley
Published
2006-10-25
Volume
16
Issue
4
Pages
393-401
ISSN
  • 1053-8569 (ISSN-L)
  • 1099-1557
DOI
10.1002/pds.1335
OpenAlex
W1979698089
Cited by
602
References
12
Citations by year
2026: 9 · 2025: 26 · 2024: 16 · 2023: 24 · 2022: 25 · 2021: 19 · 2020: 52 · 2019: 24 · 2018: 35 · 2017: 49 · 2016: 48 · 2015: 45 · 2014: 36 · 2013: 32 · 2012: 54
FWCI
34.29
Type
journal article
Language
en
Open access
closed
Licence
onlinelibrary.wiley.com/termsAndConditions#vor (vor)
Locations
  • Pharmacoepidemiology and Drug Safety (publishedVersion)
  • PubMed (publishedVersion)
Funders
Agency for Healthcare Research and Quality (HS10399)
Topics
  • Health Sciences › Health Professions › Health Information Management › Medical Coding and Health Information
  • Health Sciences › Health Professions › Health Information Management › Electronic Health Records Systems
  • Physical Sciences › Mathematics › Statistics and Probability › Statistical Methods in Epidemiology
Keywords
Health Improvement Network, Clinical Practice Research Datalink, pharmacoepidemiology, myocardial infarction, nonsteroidal anti-inflammatory drugs, peptic ulcer, diabetes mellitus, hypercholesterolemia, obesity, smoking, COX-2 inhibitors
MeSH terms
Anti-Inflammatory Agents, Non-Steroidal, Aspirin, Colonic Neoplasms, Family Practice, United Kingdom, Humans, Hypercholesterolemia, Hypertension, Myocardial Infarction, Obesity, Peptic Ulcer, Risk Factors, Smoking, Reproducibility of Results, Logistic Models, Odds Ratio, Case-Control Studies, Medical Records Systems, Computerized, Potassium Compounds, Pharmacoepidemiology, Risk Assessment, Databases as Topic, Stroke, Diabetes Complications

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review