Synapse
⌘+K
Synapse
PulseExploreClubsResearchersJournals
Instagram
HomeClubsExplore
July 17, 2019BMC MedicineOpen Access

The uncertainty with using risk prediction models for individual decision making: an exemplar cohort study examining the prediction of cardiovascular disease in English primary care

View Full Paper
Ask AI
Bookmark
Share

Why the study?

Risk prediction models commonly inform treatment decisions, but uncertainty around risk scores beyond confidence intervals is rarely explored.

Does the choice of modelling decisions affect individual cardiovascular risk predictions in patients eligible for primary prevention?

Population

3,792,474 patients eligible for cardiovascular risk prediction in the CPRD

Comparison

Different QRISK-based prediction models (model A through model F) varying modelling decisions

Design

Cohort study

Authors

APAlexander PateManchester Academic Health Science CentreRERichard EmsleyGreater Manchester Mental Health NHS Foundation TrustDADarren M. AshcroftQueen's University Belfast

Discussion

Loading...

Member takes

Implication

Wide variation in individual risk estimates across models; leaves open whether standardized approaches are needed for primary prevention decisions.

Structured PICO

Does the choice of modelling decisions affect individual cardiovascular risk predictions in patients eligible for primary prevention?

P
Population
3,855,660 patients aged 25-84 eligible for cardiovascular risk prediction from the Clinical Practice Research Datalink (CPRD) with linked hospitalisation and mortality records, without history of CVD or statin treatment at baseline (1,965,078 females, 1,890,582 males).
I
Intervention
Risk prediction models (Models B-F) incorporating additional risk factors, secular trends, geographical variation, and imputation methods compared to standard QRISK models.
C
Comparator
Model A (QRISK2 standard model).
O
Outcome
Time until the first CVD event (transient ischaemic attack, ischaemic stroke or coronary heart disease) identified either through CPRD, HES or ONS records.composite

Risk prediction models using routinely collected data show large variability in individual patient risk depending on modelling decisions, which could reclassify millions of patients for statin eligibility despite performing similarly on standard metrics.

Limitations

  • Could not assess the variation in risk that may be caused by risk factors missing from the database (such as diet or physical activity).

Cite This Study

Pate et al. (2019) studied this question.

synapsesocial.com/papers/6a81f0fbe02763da79e04481https://doi.org/10.1186/s12916-019-1368-8

Topics

Statin therapyWomen and heart diseaseCardiovascular prevention
View Full Paper
Ask AI
Bookmark
Share

Also Consider

Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context:

  1. 1An independent and external validation of QRISK2 cardiovascular disease risk score: a prospective open cohort study2010 · 254 citations
  2. 2Predicting cardiovascular risk in England and Wales: prospective derivation and validation of QRISK22008 · 1,497 citations
  3. 3Primary prevention of cardiovascular disease: A review of contemporary guidance and literature2017 · 479 citations
  4. 4Dietary fat intake and prevention of cardiovascular disease: systematic review2001 · 391 citations
  5. 5ReseArch with Patient and Public invOlvement: a RealisT evaluation – the RAPPORT study2015 · 317 citations