Article Text
PDF
Abstract
Aims Lipid point-of-care testing (POCT) analysers are being used to screen target populations to identify individuals at high risk of developing cardiovascular disease (CVD) as part of the National Health Service (NHS) Health Checks programme. We evaluated the performance of the Cholestech LDX and CardioChek PA POCT analysers against laboratory methods in CVD risk assessment.
Methods Ten-year QRISK2, Joint British Societies' II (JBSII), and Framingham CVD risk scores were calculated for subjects recruited from Wolverhampton City PCT community NHS Health Check clinics. CVD risk scores derived using POCT capillary whole blood total cholesterol and HDL-cholesterol measurements were compared with those derived from the laboratory analysis of paired venous serum samples. Data from subjects with diabetes, overt CVD, and those who did not meet the risk algorithm age criteria were excluded.
Results All subjects classified as high risk (risk score >20%) by the three risk algorithms on the basis of the laboratory results were correctly identified by the LDX. One (2.2%) and four (7.0%) moderate-risk subjects were misclassified by the LDX as high risk, using the JBSII and Framingham risk algorithms, respectively. The CardioChek identified all subjects classed as high risk by QRISK2, but failed to identify 6/31 (19.4%) and 3/19 (15.8%) of subjects classed as high risk by the Framingham and JBSII algorithms, respectively. The CardioChek, however, did not misclassify any moderate-risk subjects as high risk.
Conclusions Identification of subjects at risk of CVD depends on the cardiovascular risk algorithm and also on the performance of the POCT device.
Statistics from Altmetric.com
Introduction
Cardiovascular disease (CVD) is a major global public health problem.1 In the UK, CVD is the leading cause of morbidity and premature mortality.2 The National Health Service (NHS) Health Checks programme was launched in 2009 to screen those in the UK population aged 40–75 years for individuals at high risk of developing CVD and hitherto undiagnosed diabetics.3 ,4 CVD mortality is known to vary according to ethnicity and socioeconomic background; rates are higher among South Asians and those from deprived backgrounds. NHS Health Checks were introduced, in part, to try and address this inequalities gap in CVD risk. Each health check combines measurements of blood pressure, Body Mass Index (BMI), blood tests for glucose, total cholesterol (TC) and high-density-lipoprotein cholesterol (HDL-C) and creatinine, and a life-style assessment including diet, exercise and smoking status.3 These data are then entered into various algorithms, which have been developed using data collected from prospective population cohort studies, to estimate CVD risk and so help healthcare professionals identify individuals requiring preventative intervention. The JBSII,5 Framingham6–8 and QRISK29 are among the more commonly used risk assessment algorithms in the UK.
Point-of-care testing (POCT) could facilitate widespread population screening in the primary care setting as part of this NHS Health Check initiative. The LDX (Cholestech Corporation, Hayward, USA) and CardioChek PA (Polymer Systems Technology, Indianapolis, USA) were the major POCT lipid and glucose analysers currently in use in the UK at the time of the study. We and others have evaluated the analytical performance of these two POCT analysers against established laboratory methods.10–26 Only one study, however, has evaluated the impact of LDX POCT lipid measurements on CVD risk assessment.15 There are no data comparing the LDX and CardioChek POCT analysers in estimating cardiovascular risk. We, therefore, studied the impact of different POCT analysers against central laboratory methods in the assessment of cardiovascular risk.
Methods
Study design and subject recruitment
Subjects were recruited from community health check clinics set up by the Wolverhampton City PCT to provide a CVD and diabetes screening service for the local population. Individuals attending the clinics and receiving a POCT capillary whole blood test for TC, HDL-C and glucose were offered the option of having a paired venous sample collected for confirmatory laboratory analysis between March and September 2010. Venous sampling was employed as a means of verifying the POCT results during the evaluation period. Absolute 10-year CVD risk scores were calculated for each individual aged 30–75 years excluding those with a history of diabetes or overt CVD. A random POCT glucose measurement was used to screen for diabetes as part of the health check. Individuals with a result of ≥11.1 mmol/L were considered to be diabetic and thus excluded from this study. The results of each health check were forwarded to the appropriate general practitioner with flags to identify subjects with a calculated 10-year CVD risk of ≥20%, isolated dyslipidaemia, isolated hypertension or a random glucose ≥11.1 mmol/L.
POCT and laboratory TC and HDL-cholesterol analysis
Paired capillary and venous samples were collected from consenting subjects. Finger-prick capillary whole blood specimens were collected from each subject and analysed using either the LDX or the CardioChek POCT analyser for TC, HDL-C and glucose according to the manufacturer's instructions, with the exception that the LDX sample collection device was used for both analysers. POCT device maintenance and internal quality control (IQC) testing were performed as recommended by the manufacturer; all devices were enrolled in an appropriate external quality assurance (EQA) scheme for the duration of the study. All three analytes were analysed simultaneously by the POCT device using a single test strip (CardioChek) or cassette (LDX). POCT measurements were performed by non-laboratory personnel who had received appropriate training and competency assessment in accordance with the Royal Wolverhampton NHS Trust POCT policy. Paired venous samples were collected into Sarstedt serum gel Z/4.7 mL (TC and HDL-C) and fluoride-EDTA/2.7 mL (glucose) tubes using the Sarstedt Safety Monovette system (Sarstedt Aktiengesellschaft & Co, Germany). Samples were centrifuged within 5 h of collection and immediately analysed for TC, HDL-C and glucose using Roche reagents and methodology on the laboratory Roche Modular P analyser (Roche Diagnostics GmbH, Mannheim, Germany). POCT device operators were not aware of laboratory measurements at time of capillary sampling. POCT and laboratory TC and HDL-C measurements were used to calculate the TC/HDL-C ratios required by the CVD risk algorithms. Interassay imprecision percentage coefficients of variation (%CV) for the laboratory, LDX and CardioChek TC and HDL-C methods during the course of the study were <3% (TC) and <5% (HDL-C), <4.1% (TC) and <5.2% (HDL-C)10 and <6.7% (TC) and <8.7% (HDL-C),10 respectively.
Cardiovascular risk assessment
Absolute 10 years QRISK2, Framingham and JBSII CVD risk scores were calculated using laboratory, clinical and demographical data for each individual who met the required age criteria; 30–75 years for the QRISK2 and Framingham algorithms, and 35–74 years for JBSII. Clinical measurements and demographical data required for CVD risk calculation were collected by the POCT device operator at time of capillary sampling. In some cases, CVD risk scores could not be calculated for an individual using each of the three algorithms due to the failure to meet the required criteria, for example, age. QRISK2 (http://qrisk.org/) and Framingham (http://cvrisk.mvm.ed.ac.uk/) risk scores were calculated using online calculators. JBSII risk scores were generated using the Joint British Societies CVD risk assessor software V.1.06 (Professor PN Durrington, Department of Medicine, University of Manchester). In accordance with the JBSII CVD risk assessment guidelines,5 the values of 160 mm Hg (systolic) and 90 mm Hg (diastolic) were assumed as an estimate of pretreatment blood pressure readings for individuals already on antihypertensives (Framingham and JBSII algorithms). Antihypertensive therapy is included in the QRISK2 algorithm. The risk factor information required for CVD risk calculation differed for each algorithm; the individual's age, gender, TC/HDL-C ratio and smoking status were considered by all three algorithms. Additionally, the Framingham calculator also took into account systolic blood pressure, but no corrections were made for ethnicity or family history. The JBSII risk calculation also included data on systolic and diastolic blood pressure and made corrections for individuals with a family history of CVD (first-degree relatives aged <55 years for a male or <65 years for a female) and those of South Asian origin. QRISK2 CVD risk scores also accounted for systolic blood pressure, ethnicity, postcode, BMI, CVD family history in first-degree relatives (aged <60 years) as well as a known personal history of diabetes mellitus, chronic kidney disease, atrial fibrillation, and rheumatoid arthritis. Each individual was subsequently classified for each algorithm according to their score as being either low (≤9.9%), moderate (10.0–19.9%) or high risk (≥20.0%).
Data plotting and statistical analysis
Data analysis and the generation of the TC/HDL-C ratio Bland-Altman27 plots were performed using GraphPad Prism V.4.00 (GraphPad Software, San Diego, California, USA) and Analyse-it V.2.22 (Analyse-it Software, Leeds, UK).
Results
Paired POCT and laboratory measurements were available for 276 subjects, of whom 249 (90.2%) met the criteria for cardiovascular risk assessment with the Framingham and QRISK2 algorithms. Of these, 162 and 87 subjects were assessed using the LDX and CardioChek, respectively. Seventeen subjects were excluded from assessment using the JBSII algorithm; eight and nine for the LDX, and CardioChek groups, respectively.
Demographic, laboratory and clinical data are summarised in table 1 and figure 1. Previously published POCT and laboratory TC and HDL-C measurements10 were used to calculate the TC/HDL-C ratios required by the CVD risk algorithms. When compared to the laboratory method, the LDX and CardioChek displayed mean (±SD) percentage biases of 9.7 (10)% (95% Limits of agreement (LOAs) −9.9% to 29.3%) and −3.7 (22.6)% (95% LOAs −48.8% to 41.4%) respectively (figure 1 and table 1). The CardioChek demonstrated a higher spread of bias as shown by the wider 95% LOA.
Summary of the population demographics and POCT and laboratory results used to calculate the CVD risk scores
Bland-Altman plot comparison of the point-of-care testing (POCT) and laboratory (total cholesterol) TC : HDL-C ratio measurements. TC : HDL-C ratio data for the LDX (n=162) and CardioChek (n=87) versus the laboratory method are shown in panels A and B, respectively. The solid horizontal black lines shown in each plot represent identity. Mean percentage biases (LDX=9.7%; CardioChek=−3.7%) and 95% limits of agreement (LDX=−9.9% to 29.3%; CardioChek=−48.8% to 41.4%) are by the solid grey and dashed black horizontal lines, respectively. POCT and laboratory TC and HDL-C data used to calculate the ratios were taken from ref. 10 HDL-C, high-density-lipoprotein cholesterol; TC, total cholesterol.
Ten-year QRISK2, JBSII and Framingham cardiovascular risk scores were calculated for each subject using the POCT and laboratory TC/HDL-C ratio values. The agreement between each POCT analyser and the laboratory method in identifying subjects with low (≤9.9%), moderate (10–19.9%) and high (≥20%) 10-year CVD risk is shown in table 2. In summary, all subjects classified as being high risk using the laboratory methods were identified by the LDX, but one (2.2%) and four (7.0%) moderate-risk subjects were misclassified as high risk using the JBSII and Framingham risk algorithms, respectively. The CardioChek correctly identified all high-risk subjects using QRISK2 algorithm but failed to identify six (19.4%) and three (15.8%) high-risk subjects using the Framingham and JBSII algorithms, respectively. The CardioChek, however, had no false positive high-risk subjects in any of the risk algorithms.
CVD risk classification comparison using the JBSII, Framingham and QRISK2 algorithms: POCT (LDX and CardioChek) versus the laboratory methods
Discussion
CVD is one of the greatest public health challenges facing the UK today, but early identification of individuals at risk of developing CVD can facilitate management to delay, or even prevent, the onset of disease. Community-based screening using POCT can form part of this approach to identify high-risk individuals who may otherwise be missed by traditional primary care routes. Numerous studies have assessed the analytical performance of the LDX and CardioChek.10–26 However, given that the lipid measurements from these devices may then be used to estimate an individual's 10-year CVD risk, only the study by Jain et al15 evaluated the impact of LDX measurements on CVD risk scoring. We, present the first study comparing the effect of different POCT analysers on the assessment of cardiovascular risk.
Our results show that the choice of POCT device may influence an individual's absolute cardiovascular risk score, but this depends on the risk algorithm used. Compared to central laboratory results, the LDX and CardioChek correctly identified all high-risk individuals when using QRISK2 without any false positive results. Using the JBSII and Framingham risk algorithms, compared to laboratory methods, the LDX correctly identified all high-risk subjects (no false negatives), but misclassified some moderate-risk subjects as high risk (false positives); whereas the CardioChek failed to identify some high-risk individuals risks (false negatives) but did not misclassify moderate-risk subjects as high risk (no false positives). These results indicate that, compared to Framingham and JBSII, QRISK2 is less influenced by the analytical performance of POCT instruments because of the greater number of risk factors incorporated into the algorithm and the lesser importance given to TC/HDL-C ratio. Whereas, the higher overall weighting of TC/HDL-C in the Framingham and JBSII risk algorithms make them more susceptible to analytical performance of POCT instruments.
By comparison with laboratory methods, the overall positive TC/HDL-C bias of LDX makes it prone to false positive results, and the negative TC/HDL-C bias of CardioChek makes it prone to false negative results (figure 1); that is, the LDX tends to overestimate cardiovascular risk, whereas the CardioChek tends to underestimate risk. The spread of bias for the TC\HDL-C ratio (demonstrated by the 95% LOA) was wider for the CardioChek than the LDX which reflects our previously reported observations for the individual TC and HDL-C measurements used to calculate this ratio.10 Two possible causes of the CardioChek's wider 95% LOA relative to the LDX may be: (1) the higher observed assay imprecision and (2) a larger number of user-dependent steps which have the potential to influence the final result.10 Furthermore, a comparison of EQA data showed that the LDX method group displayed lower mean between-analyser %CVs for TC and HDL-C compared to the CardioChek.10
By comparison with central laboratory analysis, the use of POCT in the generation of CVD risk has a number of advantages; considered less invasive than venipuncture;28 ,29 reduced number of visits and on-the-spot counselling that may be important when targeting individuals who have little contact with healthcare professionals and who may not have been identified by traditional primary care routes. With all POCT devices, good clinical governance, including thorough operator training and competency assessment and adequate IQC and EQA measures are essential. Thus, the local clinical laboratory has an important role to play in the provision of the health checks in order to ensure the delivery of a quality service as recommended in the 2010 MHRA guidance concerning the use and management of POCT in vitro diagnostic devices.30 Users may wish to consider the analytical performance of their device on whether to repeat results before initiating treatment for high (or near-high)-risk patients.
In summary, our study supports the notion that POCT can be appropriate for use as part of the NHS Health Checks program,15 providing robust governance and quality assurance procedures are in place.10 It is well recognised that different cardiovascular risk algorithms give different cardiovascular risk scores in the same individual because of the differing risk factors used and their weighting.31 However, healthcare professionals, managers and commissioners should also be aware that the analytical performance of different POCT analysers varies and this may affect the individual's CVD risk score depending on the cardiovascular risk algorithm used.
Take home messages
-
Point-of-care testing (POCT) can be appropriate for use as part of the National Health Service Health Checks program providing robust governance and quality assurance procedures are in place.
-
The local clinical laboratory has an important role to play in the provision of the health checks in order to ensure the delivery of quality service.
-
Analytical performance of different POCT analysers varies and this may affect the individual's cardiovascular disease risk score depending on the risk algorithm used.
Acknowledgments
The authors thank Dr L Heath and the staff in the Public Health Department of the Wolverhampton Primary Care Trust for their help in setting up and managing the NHS Health Check clinics. We also thank BHR Pharmaceuticals (CardioChek) and Alere Ltd (LDX) for the loan of the POCT analysers. We gratefully acknowledge the help of staff from the New Cross Hospital Heart and Lung Department for helping to organise health check days (Pauline Caines and Paula Bourke) and for providing help with the phlebotomy (Annette Russell and Kim Pincher).
References
Footnotes
-
Contributors SJW and CF conceived the study in collaboration with the Wolverhampton City Primary Care Trust's Department of Public Health. SJW researched the literature, designed the studies, processed and analysed the data, and wrote the first draft. All authors reviewed and edited the manuscript, and approved the final version of the manuscript. RG is Guarantor.
-
Funding This study was funded by the Wolverhampton City Primary Care Trust's Department of Public Health (Coniston House, Chapel Ash, Wolverhampton).
-
Competing interests None.
-
Provenance and peer review Not commissioned; externally peer reviewed.