A simple risk score using routine data for predicting cardiovascular disease in primary care

Predicting poorer health outcomes in older community-dwelling patients with multimorbidity: prospective cohort study assessing the accuracy of different multimorbidity definitions

PURPOSE

Multimorbidity is commonly defined and measured using condition counts. The UK National Institute for Health Care Excellence Guidelines for Multimorbidity suggest that a medication-orientated approach could be used to identify those in need of a multimorbidity approach to management.

OBJECTIVES

To compare the accuracy of medication-based and diagnosis-based multimorbidity measures at higher cut-points to identify older community-dwelling patients who are at risk of poorer health outcomes.

DESIGN

A secondary analysis of a prospective cohort study with a 2-year follow-up (2010-2012).

SETTING

15 general practices in Ireland.

PARTICIPANTS

904 older community-dwelling patients.

EXPOSURE

Baseline multimorbidity measurements based on both medication classes count (MCC) and chronic disease count (CDC).

OUTCOMES

Mortality, self-reported health related quality of life, mental health and physical functioning at follow-up.

ANALYSIS

Sensitivity, specificity, positive predictive values (PPV) and negative predictive values (NPV) adjusting for clustering by practice for each outcome using both definitions.

RESULTS

Of the 904 baseline participants, 53 died during follow-up and 673 patients completed the follow-up questionnaire. At baseline, 223 patients had 3 or more chronic conditions and 89 patients were prescribed 10 or more medication classes. Sensitivity was low for both MCC and CDC measures for all outcomes. For specificity, MCC was better for all outcomes with estimates varying from 88.8% (95% CI 85.2% to 91.6%) for physical functioning to 90.9% (95% CI 86.2% to 94.1%) for self-reported health-related quality of life. There were no differences between MCC and CDC in terms of PPV and NPV for any outcomes.

CONCLUSIONS

Neither measure demonstrated high sensitivity. However, MCC using a definition of 10 or more regular medication classes to define multimorbidity had higher specificity for predicting poorer health outcomes. While having limitations, this definition could be used for proactive identification of patients who may benefit from targeted clinical care.

Validation of the cardiovascular risk model developed for Omanis with type 2 diabetes

AIM

The first cardiovascular risk prediction model in the Arab world was recently developed for Omanis with type 2 diabetes mellitus. This study aims to validate the newly developed model.

MATERIALS AND METHODS

A retrospective cohort study design was applied in this study. The model was validated in two samples; the model derivation sample and a separate validation sample, consisting of 1314 and 405 diabetics respectively. All patients were free of cardiovascular disease at the baseline (2009-2010) and were followed up until: the first cardiovascular event occurred; the patient died; or up to December 2015. All data were retrieved from the patients' medical records in a primary care setting.

RESULTS

In both the derivation and validation samples, the model showed good discrimination, with an area under the receiver operating curve of 0.73 (95% CI; 0.69-0.77) and 0.70 (95% CI: 0.59-0.75) respectively. Calibration of the model was satisfactory and the actual difference between the mean predicted and observed risk in different risk groups ranged from 0.7%-3.1% and 0.1%-4.2% in the derivation and validation samples respectively.

CONCLUSION

The recently developed cardiovascular disease risk assessment model for Omanis with type 2 diabetes achieved adequate overall validity. The model showed good discrimination and acceptable calibration; it therefore has the potential to be used in local clinical settings. However, further validation and comparison studies are needed to judge the generalizability and superiority of the model over other tools currently used in Oman.

Metformin in non-diabetic hyperglycaemia: the GLINT feasibility RCT

BACKGROUND

The treatment of people with diabetes with metformin can reduce cardiovascular disease (CVD) and may reduce the risk of cancer. However, it is unknown whether or not metformin can reduce the risk of these outcomes in people with elevated blood glucose levels below the threshold for diabetes [i.e. non-diabetic hyperglycaemia (NDH)].

OBJECTIVE

To assess the feasibility of the Glucose Lowering In Non-diabetic hyperglycaemia Trial (GLINT) and to estimate the key parameters to inform the design of the full trial. These parameters include the recruitment strategy, randomisation, electronic data capture, postal drug distribution, retention, study medication adherence, safety monitoring and remote collection of outcome data.

DESIGN

A multicentre, individually randomised, double-blind, parallel-group, pragmatic, primary prevention trial. Participants were individually randomised on a 1 : 1 basis, blocked within each site.

SETTING

General practices and clinical research facilities in Cambridgeshire, Norfolk and Leicestershire.

PARTICIPANTS

Males and females aged ≥ 40 years with NDH who had a high risk of CVD.

INTERVENTIONS

Prolonged-release metformin (500 mg) (Glucophage® SR, Merck KGaA, Bedfont Cross, Middlesex, UK) or the matched placebo, up to three tablets per day, distributed by post.

MAIN OUTCOME MEASURES

Recruitment rates; adherence to study medication; laboratory results at baseline and 3 and 6 months; reliability and acceptability of study drug delivery; questionnaire return rates; and quality of life.

RESULTS

We sent 5251 invitations, with 511 individuals consenting to participate. Of these, 249 were eligible and were randomised between March and November 2015 (125 to the metformin group and 124 to the placebo group). Participants were followed up for 0.99 years [standard deviation (SD) 0.30 years]. The use of electronic medical records to identify potentially eligible individuals in individual practices was resource intensive. Participants were generally elderly [mean age 70 years (SD 6.7 years)], overweight [mean body mass index 30.1 kg/m2 (SD 4.5 kg/m2)] and male (88%), and the mean modelled 10-year CVD risk was 28.8% (SD 8.5%). Randomisation, postal delivery of the study drug and outcome assessment using registers/medical records were feasible and acceptable to participants. Most participants were able to take three tablets per day, but premature discontinuation of the study drug was common (≈30% of participants by 6 months), although there were no differences between the groups. All randomised participants returned questionnaires at baseline and 67% of participants returned questionnaires by the end of the study. There was no between-group difference in Short Form questionnaire-8 items or EuroQol-5 Dimensions scores. Compared with placebo, metformin was associated with small improvements in the mean glycated haemoglobin level [-0.82 mmol/mol, 95% confidence interval (CI) -1.39 to -0.24 mmol/mol], mean estimated glomerular filtration rate (2.31 ml/minute/1.73 m2, 95% CI -0.2 to 4.81 ml/minute/1.73 m2) and mean low-density lipoprotein cholesterol level (-0.11 mmol/l, 95% CI -0.25 to 0.02 mmol/l) and a reduction in mean plasma vitamin B12 level (-16.4 ng/l, 95% CI -32.9 to -0.01 ng/l). There were 35 serious adverse events (13 in the placebo group, 22 in the metformin group), with none deemed to be treatment related.

LIMITATIONS

Changes to sponsorship reduced the study duration, the limited availability of information in medical records reduced recruitment efficiency and discontinuation of study medication exceeded forecasts.

CONCLUSIONS

A large, pragmatic trial comparing the effects of prolonged-release metformin and placebo on the risk of CVD events is potentially feasible. However, changes to the study design and conduct are recommended to enable an efficient scaling up of the trial. Recommendations include changing the inclusion criteria to recruit people with pre-existing CVD to increase the recruitment and event rates, using large primary/secondary care databases to increase recruitment rates, conducting follow-up remotely to improve efficiency and including a run-in period prior to randomisation to optimise trial adherence.

TRIAL REGISTRATION

Current Controlled Trials ISRCTN34875079.

FUNDING

The project was funded by the National Institute for Health Research (NIHR) Health Technology Assessment programme and will be published in full in Health Technology Assessment; Vol. 22, No. 18. See the NIHR Journals Library website for further project information. Merck KGaA provided metformin and matching placebo.

Improved risk stratification in prevention by use of a panel of selected circulating microRNAs

Risk stratification is crucial in prevention. Circulating microRNAs have been proposed as biomarkers in cardiovascular disease. Here a miR panel consisting of miRs related to different cardiovascular pathophysiologies, was evaluated to predict outcome in the context of prevention. MiR-34a, miR-223, miR-378, miR-499 and miR-133 were determined from peripheral blood by qPCR and combined to a risk panel. As derivation cohort, 178 individuals of the DETECT study, and as validation cohort, 129 individuals of the SHIP study were used in a case-control approach. Overall mortality and cardiovascular events were outcome measures. The Framingham Risk Score(FRS) and the SCORE system were applied as risk classification systems. The identified miR panel was significantly associated with mortality given by a hazard ratio(HR) of 3.0 (95% (CI): 1.09–8.43; p = 0.034) and of 2.9 (95% CI: 1.32–6.33; p = 0.008) after adjusting for the FRS in the derivation cohort. In a validation cohort the miR-panel had a HR of 1.31 (95% CI: 1.03–1.66; p = 0.03) and of 1.29 (95% CI: 1.02–1.64; p = 0.03) in a FRS/SCORE adjusted-model. A FRS/SCORE risk model was significantly improved to predict mortality by the miR panel with continuous net reclassification index of 0.42/0.49 (p = 0.014/0.005). The present miR panel of 5 circulating miRs is able to improve risk stratification in prevention with respect to mortality beyond the FRS or SCORE.

Cost effectiveness of case-finding strategies for primary prevention of cardiovascular disease: a modelling study

BACKGROUND

Policies of active case finding for cardiovascular disease (CVD) prevention in healthy adults are common, but economic evaluation has not investigated targeting such strategies at those who are most likely to benefit.

AIM

To assess the cost effectiveness of targeted case finding for CVD prevention.

DESIGN AND SETTING

Cost-effectiveness modelling in an English primary care population.

METHOD

A cohort of 10 000 individuals aged 30-74 years and without existing CVD or diabetes was sampled from The Health Improvement Network database, a large primary care database. A discrete-event simulation was used to model the process of inviting people for assessment, assessing cardiovascular risk, and initiation and persistence with drug treatment. Risk factors and drug cessation rates were obtained from primary care data. Published sources provided estimates of uptake of assessment, treatment initiation, and treatment effects. The researchers determined the lifetime costs and quality-adjusted life years (QALYs) with opportunistic case finding, and strategies prioritising and targeting patients by age or prior estimate of cardiovascular risk. This study reports on the optimum strategy if a QALY is valued at £20 000.

RESULTS

Compared with no case finding, inviting all adults aged 30-74 years in a population of 10 000 yields 30.32 QALYs at a total cost of £705 732. The optimum strategy is to rank patients by prior risk estimate and invite 8% of those who are assessed as being at highest risk (those at ≥12.76% predicted 10-year CVD risk), yielding 17.53 QALYs at a cost of £162 280. There is an 89.4% probability that the optimum strategy is to invite <35% of patients for assessment.

CONCLUSION

Across all age ranges, targeted case finding using a prior estimate of CVD risk is more efficient than universal case finding in healthy adults.

Cardiovascular screening in general practice in a low SES area

Background

Lower social economic status (SES) is related to an elevated cardiovascular (CV) risk. A pro-active primary prevention CV screening approach in general practice (GP) might be effective in a region with a low mean SES. This approach, supported by a regional GP laboratory, was investigated on feasibility, attendance rate and proportion of persons identified with an elevated risk.

Methods

In a region with a low mean SES, men and women aged ≥50/55 years, respectively, were invited for cardiovascular risk profiling, based on SCORE 10-year risk of fatal cardiovascular disease and additional risk factors (family history, weight and end organ damage). Screening was performed by laboratory personnel, at the GP practice. Treatment advice was based on Dutch GP guidelines for cardiovascular risk management. Response rates were compared to those in five other practices, using the same screening method.

Results

521 persons received invitations, 354 (68%) were interested, 33 did not attend and 43 were not further analysed because of already known diabetes/cardiovascular disease. Eventually 278 risk profiles were analysed, of which 60% had a low cardiovascular risk (SCORE-risk <5%). From the 40% participants with a SCORE-risk ≥5%, 60% did not receive medication yet for hypertension/hypercholesterolemia. In the other five GPs response rates were comparable to the currently described GP.

Conclusion

Screening in GP in a low SES area, performed by a laboratory service, was feasible, resulted in high attendance, and identification and treatment advice of many new persons at risk for cardiovascular disease.

Predictive risk algorithms in a population setting: an overview

BACKGROUND

The widespread use of risk algorithms in clinical medicine is testimony to how they have helped transform clinical decision-making. Risk algorithms have a similar but underdeveloped potential to support decision-making for population health.

OBJECTIVE

To describe the role of predictive risk algorithms in a population setting.

METHODS

First, predictive risk algorithms and how clinicians use them are described. Second, the population uses of risk algorithms are described, highlighting the strengths of risk algorithms for health planning. Lastly, the way in which predictive risk algorithms are developed is discussed briefly and a guide for algorithm assessment in population health presented.

CONCLUSION

For the past 20 years, absolute and baseline risk has been a cornerstone of population health planning. The most accurate and discriminating method to generate such estimates is the use of multivariable risk algorithms. Routinely collected data can be used to develop algorithms with characteristics that are well suited to health planning and such data are increasingly available. The widespread use of risk algorithms in clinical medicine is testimony to how they have helped transform clinical decision-making. Risk algorithms have a similar but underdeveloped potential to support decision-making for population health.

Original and REGICOR Framingham functions in a nondiabetic population of a Spanish health care center: a validation study

PURPOSE

Risk functions can help general practitioners identify patients at high cardiovascular risk, but overprediction inevitably leads to a disproportionate number of patients being targeted for treatment. To assess predicted cardiovascular risk, we analyzed the 10-year performance of the original and REGICOR Framingham coronary risk functions in nondiabetic patients.

METHODS

Ours was a longitudinal, observational study of a retrospective cohort of patients observed for 10 years in primary care practices in Badajoz, Spain. Our cohort comprised 447 nondiabetic patients aged 35 to 74 years who had no evidence of cardiovascular disease and were not on lipid-lowering or antihypertensive therapy. We assessed the patients' 10-year coronary risk measurement from the time of their recruitment. We also estimated the percentage of patients who were candidates for antihypertensive and lipid-lowering therapy.

RESULTS

The actual incidence rate of coronary events was 6.7%. The original Framingham equation overpredicted risk by 73%, whereas the REGICOR Framingham function underpredicted risk by 64%. The Brier scores were 0.06364 and 0.06093 (P = .365) for the original Framingham and REGICOR Framingham functions, respectively, and the remaining discrimination and calibration parameters were also highly similar for both functions. The original Framingham function classified 14.8% of the population as high risk and the REGICOR Framingham function classified 6.9%. The proportions of patients who, according to the original Framingham and REGICOR functions, would be candidates for lipid-lowering therapy were 14.3% and 6.7%, and for antihypertensive therapy they were 12.5% and 7.8%, respectively.

CONCLUSION

The original Framingham equation overestimated coronary risk whereas the REGICOR Framingham function underestimated it. The original Framingham function selected a greater percentage of candidates for antihypertensive and lipid-lowering therapy.

The Evaluation of Screening and Early Detection Strategies for Type 2 Diabetes and Impaired Glucose Tolerance (DETECT-2) update of the Finnish diabetes risk score for prediction of incident type 2 diabetes

AIMS/HYPOTHESIS

The Finnish diabetes risk questionnaire is a widely used, simple tool for identification of those at risk for drug-treated type 2 diabetes. We updated the risk questionnaire by using clinically diagnosed and screen-detected type 2 diabetes instead of drug-treated diabetes as an endpoint and by considering additional predictors.

METHODS

Data from 18,301 participants in studies of the Evaluation of Screening and Early Detection Strategies for Type 2 Diabetes and Impaired Glucose Tolerance (DETECT-2) project with baseline and follow-up information on oral glucose tolerance status were included. Incidence of type 2 diabetes within 5 years was used as the outcome variable. Improvement in discrimination and classification of the logistic regression model was assessed by the area under the receiver-operating characteristic (ROC) curve and by the net reclassification improvement. Internal validation was by bootstrapping techniques.

RESULTS

Of the 18,301 participants, 844 developed type 2 diabetes in a period of 5 years (4.6%). The Finnish risk score had an area under the ROC curve of 0.742 (95% CI 0.726-0.758). Re-estimation of the regression coefficients improved the area under the ROC curve to 0.766 (95% CI 0.750-0.783). Additional items such as male sex, smoking and family history of diabetes (parent, sibling or both) improved the area under the ROC curve and net reclassification. Bootstrapping showed good internal validity.

CONCLUSIONS/INTERPRETATION

The predictive value of the original Finnish risk questionnaire could be improved by adding information on sex, smoking and family history of diabetes. The DETECT-2 update of the Finnish diabetes risk questionnaire is an adequate and robust predictor for future screen-detected and clinically diagnosed type 2 diabetes in Europid populations.