Elsevier

The Lancet

Volume 397, Issue 10291, 12–18 June 2021, Pages 2264-2274
The Lancet

Articles
Cardiovascular risk prediction in type 2 diabetes before and after widespread screening: a derivation and validation study

https://doi.org/10.1016/S0140-6736(21)00572-9Get rights and content

Summary

Background

Until recently, most patients with diabetes worldwide have been diagnosed when symptomatic and have high cardiovascular risk, meaning most should be prescribed cardiovascular preventive medications. However, in New Zealand, a world-first national programme led to approximately 90% of eligible adults being screened for diabetes by 2016, up from 50% in 2012, identifying many asymptomatic patients with recent-onset diabetes. We hypothesised that cardiovascular risk prediction equations derived before widespread screening would now significantly overestimate risk in screen-detected patients.

Methods

New Zealanders aged 30–74 years with type 2 diabetes and without known cardiovascular disease, heart failure, or substantial renal impairment were identified from the 400 000-person PREDICT primary care cohort study between Oct 27, 2004, and Dec 30, 2016, covering the period before and after widespread screening. Sex-specific equations estimating 5-year risk of cardiovascular disease were developed using Cox regression models, with 18 prespecified predictors, including diabetes-related and renal function measures. Equation performance was compared with an equivalent equation derived in the New Zealand Diabetes Cohort Study (NZDCS), which recruited between 2000 and 2006, before widespread screening.

Findings

46 652 participants were included in the PREDICT-1° Diabetes subcohort, of whom 4114 experienced first cardiovascular events during follow-up (median 5·2 years, IQR 3·3–7·4). 14 829 (31·8%) were not taking oral hypoglycaemic medications or insulin at baseline. Median 5-year cardiovascular risk estimated by the new equations was 4·0% (IQR 2·3–6·8) in women and 7·1% (4·5–11·2) in men. The older NZDCS equation overestimated median cardiovascular risk by three times in women (median 14·2% [9·7–20·0]) and two times in men (17·1% [4·5–20·0]). Model and discrimination performance measures for PREDICT-1° Diabetse equations were also significantly better than for the NZDCS equation (eg, for women: R2=32% [95% CI 29–34], Harrell's C=0·73 [0·72–0·74], Royston's D=1·410 [1·330–1·490] vs R2=24% [21–26], C=0·69 [0·67–0·70], and D=1·147 [1·107–1·187]).

Interpretation

International treatment guidelines still consider most people with diabetes to be at high cardiovascular risk; however, we show that recent widespread diabetes screening has radically changed the cardiovascular risk profile of people with diabetes in New Zealand. Many of these patients have normal renal function, are not dispensed glucose-lowering medications, and have low cardiovascular risk. These findings have clear international implications as increased diabetes screening is inevitable due to increasing obesity, simpler screening tests, and the introduction of new-generation glucose-lowering medications that prevent cardiovascular events. Cardiovascular risk prediction equations derived from contemporary diabetes populations, with multiple diabetes-related and renal function predictors, will be required to better differentiate between low-risk and high-risk patients in this increasingly heterogeneous population and to inform appropriate non-pharmacological management and cost-effective targeting of expensive new medications.

Funding

Health Research Council of New Zealand, Heart Foundation of New Zealand, and Healthier Lives National Science Challenge.

Introduction

In 2003, the New Zealand Ministry of Health guidelines recommended that cardiovascular preventive treatment decisions should be informed by calculated cardiovascular risk, using a Framingham Heart Study-based cardiovascular risk prediction calculator that required clinicians to assess patients' diabetes status.1, 2 The majority of middle-aged New Zealanders met guideline eligibility criteria for 5-yearly cardiovascular risk assessments, effectively introducing almost universal diabetes screening.

Subsequently, a national More Heart and Diabetes Checks health target was introduced to increase cardiovascular risk assessments in the eligible population from 50% in 2012 to 90% by 2016.3 The target included an assessment of diabetes status that was facilitated by the replacement of fasting blood glucose with the simpler non-fasting glycated haemoglobin (HbA1C) as the recommended screening test.4 The 90% target was achieved in September, 2016.3

Research in context

Evidence before this study

A 2019 systematic review identified and compared the performance of 15 cardiovascular disease risk prediction models developed in diabetes populations and 11 models developed in general populations but later validated in diabetes populations. The authors found that the discriminative performance of the prediction models was only modest and only half the studies had been externally validated. They concluded that improvements in performance through the identification of additional predictors, and further validation studies, were required before the models should be implemented in clinical practice. To our knowledge, none of these studies were either conducted or validated in populations with widespread diabetes screening.

Added value of this study

By September, 2016, approximately 90% of middle-aged New Zealanders had been screened for diabetes, up from about 15% in 2001 and 50% in 2012. This followed the establishment of a national funded More Heart and Diabetes Checks health target in 2012. We are currently unaware of any other country that has diabetes screening levels as high as New Zealand. In this unique study, we were able to validate the New Zealand Diabetes Cohort Study (NZDCS) cardiovascular risk prediction equation, which was derived from a representative New Zealand diabetes population between 2000 and 2006, before the introduction of widespread diabetes screening. The NZDCS equation overestimated median cardiovascular risk by three times in woman and two times in men in a more contemporary New Zealand diabetes population recruited between 2004 and 2016, with many participants diagnosed through screening following the establishment of the 2012 health target. We then developed the world's first cardiovascular risk prediction equations in a contemporary diabetes population with almost universal diabetes screening. The new equations were well calibrated, and had a significantly improved ability to differentiate between high-risk and low-risk patients compared with the NZDCS equation.

Implications of all the available evidence

Recent widespread diabetes screening has radically changed the cardiovascular risk profile of patients with diabetes in New Zealand. The combined effect of increasing obesity, increased use of cardiovascular risk prediction equations requiring diabetes assessments, the introduction of a simple non-fasting glycated haemoglobin as the international diabetes diagnostic standard, and the development of new-generation glucose-lowering medications will inevitably lead to increased diabetes screening worldwide. We have shown that cardiovascular risk prediction equations developed before widespread diabetes screening will significantly overestimate cardiovascular risk in many screen-detected patients. Therefore, new equations, derived from diabetes populations including screen-detected patients, and with additional predictors to help to differentiate between high-risk and low-risk patients, will be required to more accurately predict cardiovascular risk in people with diabetes. Without new equations, low-risk patients might be overtreated with new-generation glucose-lowering medications that have only been shown to reduce cardiovascular events in patients at high cardiovascular risk.

In parallel with this timeline, between 2002 and 2016, approximately 400 000 primary care patients were recruited into the PREDICT cohort study when they completed cardiovascular risk assessments, with half of participants recruited after 2010.5 New cardiovascular risk prediction equations were derived in this cohort after we showed that the previously recommended Framingham equation overestimated cardiovascular risk by approximately 50% in the PREDICT cohort.6 These new general population equations, derived in people with and without diabetes, were incorporated into 2018 national risk management guidelines7 and are now widely used.

However, while we showed that the recommended Framingham equation1 significantly overestimated risk in the general PREDICT cohort,6 this same Framingham equation had previously been shown to significantly underestimate cardiovascular risk in people with type 2 diabetes in the New Zealand Diabetes Cohort Study (NZDCS), recruited between 2000 and 2006, before widespread screening began.8 The NZDCS investigators derived a cardiovascular risk prediction equation in this diabetes cohort and recommended that it replace the Framingham equation for assessing risk in New Zealanders with diabetes.8

We hypothesised that the NZDCS equation would now overestimate cardiovascular risk in a more contemporary New Zealand diabetes population, largely because many screen-detected patients would be identified much earlier in the course of their diabetes. Also, a recent systematic review examined the performance of both general population and diabetes-specific cardiovascular risk prediction equations in people with diabetes.9 The authors concluded that improvements in performance through the identification of additional predictors and further validation studies were required before the models should be implemented in clinical practice. Here, we describe the external validation of the diabetes-specific NZDCS equation and the derivation of equivalent new equations, with additional diabetes-related and renal function-related predictors, in the subset of people with type 2 diabetes in the PREDICT study.

Section snippets

Study design and participants

PREDICT is an ongoing open cohort study that automatically recruits participants when New Zealand primary health-care practitioners complete standardised cardiovascular risk assessments using PREDICT decision support software, as reported in detail elsewhere.5, 6 The software auto-populates PREDICT risk factor templates from patient records. Clinicians then complete missing fields before cardiovascular risk is calculated and recruitment finalised. Participant risk factor profiles captured by

Results

63 362 people aged 30–74 years with type 2 diabetes were recruited between Oct 27, 2004, and Dec 30, 2016 (figure 1), with 33 582 (53·0%) recruited after 2010. We excluded 12 289 people with previous cardiovascular disease, 2539 with heart failure, and 1882 with significant renal impairment. The remaining 46 652 people constituted the PREDICT-1° Diabetes subcohort used in all analyses presented here (table 1). They experienced 4114 first cardiovascular disease events during 244 840 person-years

Discussion

This unique study has documented how recent widespread diabetes screening has radically changed the cardiovascular risk profile of people diagnosed with type 2 diabetes in New Zealand. The cardiovascular risk distribution of a contemporary representative population of New Zealanders with diabetes bears little resemblance to the much higher risk distribution predicted by an equation developed in New Zealand just a few years before introduction of widespread screening. The main implication of

Data sharing

All individual participant (deidentified) data, including a data dictionary defining each field, will be made available to university-based academic researchers if their proposed analyses are approved by the investigators' Data Access Proposal Committee. A proposal must be considered relevant to the original aims of the research, must meet the study's ethics approval criteria, and will require one or more of the study investigators as formal collaborators. A signed data access agreement will be

Declaration of interests

We declare no competing interests.

References (36)

  • Enrolment in a primary health organisation

  • National Health Index

  • G Collins et al.

    Transparent reporting of a multivariable prediction model for individual prognosis or diagnosis (TRIPOD): the TRIPOD statement

    BMJ

    (2015)
  • I White et al.

    Multiple imputation using chained equations: issues and guidance for practice

    Stat Med

    (2011)
  • R Little et al.

    Statistical analysis with missing data

    (2002)
  • D Cox

    Regression models and life-tables (with discussion)

    J R Stat Soc Ser B

    (1972)
  • D Schoenfeld

    Partial residuals for the proportional hazards regression model

    Biometrika

    (1982)
  • D Belsley et al.

    Regression diagnostics: identifying influential data and sources of collinearity

    (1980)
  • Cited by (0)

    View full text