Article Text

Download PDFPDF
Original research article
Developing and validating a cardiovascular risk score for patients in the community with prior cardiovascular disease
  1. Katrina K Poppe1,2,
  2. Rob N Doughty2,3,
  3. Sue Wells1,
  4. Dudley Gentles1,
  5. Harry Hemingway4,
  6. Rod Jackson1,
  7. Andrew J Kerr1,5
  1. 1 School of Population Health, University of Auckland, Auckland, New Zealand
  2. 2 Department of Medicine, University of Auckland, Auckland, New Zealand
  3. 3 Greenlane Cardiovascular Service, Auckland City Hospital, Auckland, New Zealand
  4. 4 Farr Institute of Health Informatics Research and Institute of Health Informatics, University College London, London, UK
  5. 5 Counties Manukau District Health Board, Middlemore Hospital, Auckland, New Zealand
  1. Correspondence to Dr Katrina K Poppe, Epidemiology and Biostatistics, School of Population Health, University of Auckland Private Bag 92019, Auckland 1142, New Zealand ; k.poppe{at}


Objective Patients with atherosclerotic cardiovascular disease (CVD) vary significantly in their risk of future CVD events; yet few clinical scores are available to aid assessment of risk. We sought to develop a score for use in primary care that estimates short-term CVD risk in these patients.

Methods Adults aged <80 years with prior CVD were identified from a New Zealand primary care cohort study (PREDICT), and linked to national mortality, hospitalisation and dispensing databases. A Cox model with an outcome of myocardial infarction, stroke or CVD death within 2 years was developed. External validation was performed in a cohort from the UK.

Results 24 927 patients, 63% men, 63% European, median age 65 years (IQR 58–72 years), experienced 1480 CVD events within 2 years after a CVD risk assessment. A risk score including ethnicity, comorbidities, body mass index, creatine creatinine and treatment, in addition to established risk factors used in primary prevention, predicted a median 2-year CVD risk of 5.0% (IQR 3.5%–8.3%). A plot of actual against predicted event rates showed very good calibration throughout the risk range. The score performed well in the UK cohort but overestimated risk for those at highest risk, who were predominantly patients defined as having heart failure.

Conclusions The PREDICT-CVD secondary prevention score uses routine measurements from clinical practice that enable it to be implemented in a primary care setting. The score will facilitate risk communication between primary care practitioners and patients with prior CVD, particularly as a resource to show the benefit of risk factor modification.

  • cardiovascular disease
  • risk score
  • secondary prevention
  • electronic health record

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


  • Contributors KKP and DG: analyses and model construction. KKP and AJK: manuscript drafting. RJ, SW and AJK: PREDICT study concept, design and implementation. All authors: critical revision of the manuscript for important intellectual content and approval of the final version.

  • Funding The PREDICT CVD study has been supported by the Health Research Council of NZ via project grants (03/183, 08/121) and a programme grant (11/800). KKP has received a Heart Foundation of NZ Research Fellowship, RND is the recipient of the Heart Foundation of NZ Chair of Heart Health and SW was the recipient of a Stevenson Fellowship in Health Innovation and Quality Improvement.Competing interests Outside of this study SW has received a research grant from Roche Diagnostics Ltd. Exeter, in the PDF the competing interest statement has bunched up and joined the funding statement.

  • Competing interests Outside of this study, SW has received a research grant from Roche Diagnostics.

  • Provenance and peer review Not commisioned; externally peer reviewed.

Linked Articles