Skip to main content
Download PDF

Development and validation of a clinical prediction model for detecting coronary heart disease in middle-aged and elderly people: a diagnostic study

European Journal of Medical Research volume 28, Article number: 375 (2023) Cite this article

Abstract

Objective

To develop and validate a multivariate prediction model to estimate the risk of coronary heart disease (CHD) in middle-aged and elderly people and to provide a feasible method for early screening and diagnosis in middle-aged and elderly CHD patients.

Methods

This study was a single-center, retrospective, case–control study. Admission data of 932 consecutive patients with suspected CHD were retrospectively assessed from September 1, 2020 to December 31, 2021 in the Department of Integrative Cardiology at China-Japan Friendship Hospital. A total of 839 eligible patients were included in this study, and 588 patients were assigned to the derivation set and 251 as the validation set at a 7:3 ratio. Clinical characteristics of included patients were compared between derivation set and validation set by univariate analysis. The least absolute shrinkage and selection operator (Lasso) regression analysis method was performed to avoid collinearity and identify key potential predictors. Multivariate logistic regression analysis was used to construct a clinical prediction model with identified predictors for clinical practice. Bootstrap validation was used to test performance and eventually we obtained the actual model. And the Hosmer–Lemeshow test was carried out to evaluate the goodness-fit of the constructed model. The area under curve (AUC) of receiver operating characteristic (ROC), calibration curve, decision curve analysis (DCA), and clinical impact curve (CIC) were plotted and utilized with validation set to comprehensively evaluate the predictive accuracy and clinical value of the model.

Results

A total of eight indicators were identified as risk factors for the development of CHD in middle-aged and elderly people by univariate analysis. Of these candidate predictors, four key parameters were defined to be significantly related to CHD by Lasso regression analysis, including age (OR 1.034, 95% CI 1.002 ~ 1.067, P = 0.040), hemoglobin A1c (OR 1.380, 95% CI 1.078 ~ 1.768, P = 0.011), ankle-brachial index (OR 0.078, 95% CI 0.012 ~ 0.522, P = 0.009), and brachial artery flow-mediated vasodilatation (OR 0.848, 95% CI 0.726 ~ 0.990, P = 0.037). The Hosmer–Lemeshow test showed a good calibration performance of the clinical prediction model (derivation set, χ2 = 7.865, P = 0.447; validation set, χ2 = 11.132, P = 0.194). The ROCs of the nomogram in the derivation set and validation set were 0.722 and 0.783, respectively, suggesting excellent predictive power and suitable performance. The clinical prediction model presented a greater net benefit and clinical impact based on DCA and CIC analysis.

Conclusion

Overall, the development and validation of the multivariate model combined the laboratory and clinical parameters of patients with CHD, which could be beneficial to the individualized prediction of middle-aged and elderly people, and helped to facilitate clinical assessments and decisions during treatment and management of CHD.

Introduction

Coronary heart disease (CHD) is a type of ischemic heart diseases characterized by atherosclerotic plaque accumulation in the coronary arteries [1] As one of the leading causes of hospitalization and death [2] CHD affects over 110 million individuals worldwide [3] and thus gives rise to a heavy burden on health expenditures [4] Advancing age is a major risk factor for cardiovascular events,[5, 6] previous evidence showed a higher incidence of CHD in males over 40 years old and the prevalence can be as high as 27.8% in patients over 60 years old [7] Currently, coronary angiography (CAG) is the reference standard for diagnosing CHD. Whereas, the more complications and worse physical conditions in elder people, including renal insufficiency, coagulation abnormalities, or intolerance to CAG [1] all of which require special attention, highlight the importance of alternative diagnostic methods that are more appropriate. Therefore, we try to establish a cardiovascular disease prediction model for this special group to provide a feasible method for early screening and diagnosis of high-risk patients.

Clinical prediction models are mathematical equations that relate multiple predictors to evaluate he probability of an outcome [8, 9] which can be used to gain insights into causality of the outcome of interest and have been recognized as reliable tools for quantifying risk in diagnostic and prognostic analyses [10, 11] Besides, nomogram is a prediction tool with the advantages of being graph-based and easy-to-understand, which can predict individualized specific risks for each patient in complex clinical settings [12, 13] And they could be valuable decision support tools to assist clinicians in the complicated choices they make regarding patient management.

We therefore developed and validated a diagnostic model combining the clinical and laboratory parameters of CHD in middle-aged and elderly people based on the clinical data of 839 eligible patients, to determine whether these factors could be incorporated into the model to provide a potential auxiliary solution for patient identification of CHD.

Materials and methods

Sampling design

This study was a single-center, retrospective, case–control study. Admission data of consecutive patients with suspected CHD between September 1, 2020 and December 31, 2021 were retrospectively assessed in the Department of Integrative Cardiology at China-Japan Friendship Hospital, which were obtained from the electronic medical records system and analyzed anonymously. All eligible patients were classified as the derivation set and validation set at a 7:3 ratio, respectively. Based on the results of CAG, patients with at least one coronary artery stenosis ≥ 50% were designated in the CHD group, and the rest were assigned to the non-CHD group in the two sets.

Inclusion criteria were as follows: (1) all patients with CHD met the diagnostic criteria from the guideline; [1, 14] (2) the age of the patient ≥ 45 years old; (3) every patient voluntarily signed informed consent for admission. Patients were excluded if they met one or more of the following criteria: (1) patients who suffered from severe cerebrovascular diseases, severe liver and kidney dysfunction, acute infection, malignant tumor, severe diseases of the endocrine and hematopoietic systems, mental diseases, pregnancy status, and patients in lactation, etc; (2) patients with contraindications to CAG or who cannot cooperate with the arterial vascular examination; (3) those without complete clinical data. This study was performed in accordance with the TRIPOD statement [15] and the Declaration of Helsinki of 1975, as revised in 2013.

Sample size

The sample size for logistic regression is calculated by the following suggested Eq. (1).

(1)

In Eq. (1), parameter represents the number of independent variables and parameter represents the smallest proportion of positive or negative cases in the population. As far as we know, the incidence of CHD in middle-aged and elderly people is about 27.8% [7] and the incidence is estimated to be 30% for the sample size calculation. Firstly, 100 samples were preliminarily included and analyzed and 9 variables were found to be significantly related to CHD after univariate analysis. Therefore, the minimum number of eligible patients required of the derivation set is: N = 10*9/0.3 = 300.

Diagnostic criteria and of CHD

According to the “2019 ESC Guidelines for the diagnosis and management of chronic coronary syndromes” [1] and “Nomenclature and criteria for diagnosis of ischemic heart disease. Report of the Joint International Society and Federation of Cardiology/World Health Organization task force on standardization of clinical nomenclature” [14] CHD was diagnosed when CAG showed ≥ 50% stenosis in at least one coronary artery. CAG was conducted by a team of professional cardiologists in this study.

Data collection

Basic information

The basic clinical information of all included patients was recorded, including demographic information (sex, age, height, weight, smoking history, etc.), clinical characteristics (coronary lesions, main symptoms, comorbidities, family histories, prior medication use, etc.). Peripheral venous blood samples were drawn in the fasting state in the morning of the 2nd day after admission and processed within 2 h. Serum indicators, including alanine aminotransferase (ALT), aspartate aminotransferase (AST), hypersensitive C-reactive protein (Hs-CRP), Homocysteine (HCY), total cholesterol (TC), triglyceride (TG), low-density lipoprotein cholesterol (LDL-C), high-density lipoprotein cholesterol (HDL-C), serum creatinine (Scr), N-terminal pro-B-type natriuretic peptide (NT-proBNP), and hemoglobin A1c (HbA1c) were tested in the Department of Clinical Laboratory of China-Japan Friendship Hospital.

Transthoracic echocardiography

Echocardiography was then performed and echocardiographic indexes consisting of left atrium diameter (LAD), left ventricular ejection fraction (LVEF), left ventricular end-diastolic diameter (LVDd), left ventricular posterior wall thickness (PWT), and interventricular septal thickness (IVST) were analyzed and recorded by two independent echocardiographers. Left ventricular mass (LVM) was determined using the anatomically validated Devereux equation [16] and normalized by body surface area (BSA), according to the formulas as follows: (1) LVM(g) = 0.8 × {1.04 × [(LVDd + PWT + IVST)3-LVDd3]} + 0.6; (2) BSA(m2) = 0.0061 ×  height (cm) + 0.0128 × weight (kg)-0.1529; (3) LVMI = LVM/BSA.

Measurements of baPWV and ABI

The limb arterial elasticity was examined before CAG in the morning, and patients abstained from alcohol-, nicotine-, caffeine-containing products, and vasodilator drugs for at least 8 h. All eligible patients were placed in a supine position after having 5 or more minutes of rest, with the limbs fully exposed. On both the left and right sides, the pressure cuff was attached to the brachial artery and the posterior tibial arteries, and then limb artery pulse waveform and blood pressure were measured and recorded automatically by a BP-203 RPEII arteriosclerosis detector (OMRON, Japan). Finally, the values of ankle-brachial index (ABI) and brachial-ankle pulse wave velocity (baPWV) were calculated. Meanwhile, heart rate, blood pressure, pulse volume waveform, and electrocardiogram were noted simultaneously during the test.

Measurement of FMD

All patients were positioned in a supine position after having 5 or more minutes of rest, with the right upper limb fully exposed, and monitored by limb lead electrocardiogram continuously. The examination was carried out before CAG in the morning and patients were instructed to avoid alcohol-, nicotine-, caffeine-containing products, and vasodilator drugs at least 8 h before the test. After the position of the brachial artery was located by ultrasound, the ultrasound probe was fixed and then a blood pressure cuff was placed around the upper arm distal to the brachial artery segment that was explored. The probe was angulated at 90° for optimal morphologic B-mode imaging and < 60° for optimal velocity acquisition. Brachial artery diameter and flow velocity were recorded using a UNEXEF38G vascular endothelial function detector (UNEX, Japan) before cuff inflation, at deflation, and after deflation at 1 min intervals for 5 min, and the difference between these measures represented brachial artery flow-mediated vasodilatation (FMD) value.

Statistical analysis

IBM SPSS Statistics software (version 26.0, Chicago, USA) and R statistical analysis software (version 4.1.2, Vienna, Austria) were used for statistical analysis. Continuous variables were shown as the mean ± standard deviation for normal distribution and median with interquartile range (P25, P75) for non-normal distribution, which were compared by the Mann–Whitney U test, respectively. Categorical variables were presented as frequency (percentage) using the Chi-Square test. Clinical characteristics of included patients were compared between derivation set and validation set by univariate analysis. The least absolute shrinkage and selection operator (Lasso) regression analysis method was performed to avoid collinearity and identify key potential predictors, meanwhile, indicators with statistically significant differences were considered for the establishment of clinical prediction model. Multivariate logistic regression analysis was used to construct a clinical prediction model with identified predictors for clinical practice. Bootstrap validation was used to test performance and eventually we obtained the actual model (seed = 120, nfolds = 3). And the Hosmer–Lemeshow test was carried out to evaluate the goodness-fit of the constructed model, with lower χ2 and higher P values indicating better calibration. The area under curve (AUC) of receiver operating characteristic (ROC), calibration curve, decision curve analysis (DCA), and clinical impact curve (CIC) were plotted and utilized with validation set to comprehensively evaluate the predictive accuracy and clinical value of the model. Values for AUC range from 0.5 to 1 and the closer to unity, the more accurate a model. A difference at P < 0.05 was considered statistically significant.

Results

Selection process

Admission data of 932 consecutive patients between September 1, 2020 and December 31, 2021 were retrospectively assessed in the Department of Integrative Cardiology at China-Japan Friendship Hospital. Of these, 93 were excluded according to the exclusion criteria: 16 with a history of malignant tumor, 6 with hepatic insufficiency, 34 with a history of chronic kidney disease, 8 with cerebral hemorrhage, 5 with cerebral infarction, and 24 without complete information. The rest of the patients were randomly classified into the derivation (n = 588) and validation sets (n = 251) at a 7:3 ratio and then assigned to the CHD group and non-CHD group following their CAG findings. The details of the selection process are shown in Fig. 1.

Fig. 1
figure 1

Flowchart of the detailed selection process

Full size image

Clinical characteristics of patients in derivation and validation sets

The baseline demographic and clinicopathologic features of patients in derivation and validation sets were listed in Table 1. Overall, there was no significant difference in gender distribution, age, BMI, proportions of smoking history, diabetes, hypertension, heart failure (NYHA I-II) and atrial fibrillation, peripheral blood parameters, Scr, Hs-CRP, blood lipid, HCY, HbA1c, NT-proBNP, FMD, as well as arterial stiffness indices and echocardiographic parameters between the two sets (P < 0.05). Besides, the frequency of ACEI/ARB was significantly higher than those in the validation set (P = 0.023).

Table 1 Basic characteristics of patients in the derivation set and validation set
Full size table

Clinical characteristics of patients in the derivation set

Patients were classified into the CHD (N = 433) and non-CHD (N = 155) groups according to the CAG results. The baseline characteristics of patients in the two groups were compared by univariate analysis, as listed in Table 2. A total of eight indicators were identified as risk factors for the development of CHD in middle-aged and elderly people. Compared with patients in the non-CHD group, patients in the CHD group showed a higher proportion of diabetes history, as well as higher age, HbA1c, baPWV, and lower TC, LDL-C, ABI, and FMD levels (P < 0.001). Similar results were obtained in the validation set. (Additional file 1: Table S1).

Table 2 Basic characteristics of patients in the derivation set
Full size table

Lasso regression analysis to identify key potential predictors

Lasso regression analysis method was performed to avoid collinearity and identify key potential predictors. On this basis, variables were again selected using a regression model, and those with P < 0.05 were integrated into the analysis and considered for the establishment of clinical prediction model. Furthermore, we then used Lasso regression simulations to assess eight candidate variables, with non-zero coefficients taking a penalty paramete (λ)., as shown in Fig. 2. Meanwhile, indicators with statistically significant differences were considered for the establishment of clinical prediction model. Finally, four key potential predictors were identified combined with the results of univariate analysis, including age, HbA1c, ABI, and FMD (λ = 0.005). These four key potential predictors would be further incorporated into the development of the model in the next step.

Fig. 2
figure 2

The process of selecting key potential predictors by Lasso regression analysis. A Coefficients profile of selected predictors using Lasso regression analysis; B Using all the sample and candidate predictors, we employ Lasso to select the primitive predictors by cross-validation method

Full size image

Construction of the diagnostic nomogram by multivariate logistic regression analysis

By the previous univariate analysis, the presence of diabetes history, age, HbA1c, baPWV, and lower TC, LDL-C, ABI, and FMD levels were identified as risk factors for the development of CHD in middle-aged and elderly people. Of these parameters, age (OR 1.034, 95% CI 1.002 ~ 1.067, P = 0.040), HbA1c (OR 1.380, 95% CI 1.078 ~ 1.768, P = 0.011), ABI (OR 0.078, 95% CI 0.012 ~ 0.522, P = 0.009), and FMD (OR 0.848, 95% CI 0.726 ~ 0.990, P = 0.037) levels were identified as independent predictors for CHD development on multivariate logistic regression analysis, as shown in Table 3. Based on this result, a new predictive equation was established: Risk score = 1.451 + 0.033*Age + 0.322*HbA1c-2.548*ABI-0.165*FMD. The nomogram was incorporated with the above independent four predictors, as shown in Fig. 3. The weights of predictors are represented by the length of the line segment, which was positively correlated with the degree to which the predictors affected clinical outcomes. The diagnostic possibility corresponds to the total points by adding the scores for each predictor, indicating the probability of CHD.

Table 3 Multivariate logistic regression analysis of independent risk factors
Full size table
Fig. 3
figure 3

Nomogram predicting CHD in middle-aged and elderly people

Full size image

Meanwhile, the Hosmer–Lemeshow test showed a good calibration performance of the clinical prediction model (χ2 = 7.865, P = 0.447), and the calibration curve indicated an excellent consistency of the model in the derivation set (Fig. 4A1). The ROC curve demonstrated the predictive power of the nomogram with an AUC value of 0.722 (Fig. 4B1), indicating that the nomogram had a high predictive value. In addition, DCA and CIC analysis was ultilized to assess the clinical utility of the prediction model (Fig. 4C1, D1), suggesting that the model had a good overall net benefit and clinical impact within most reasonable threshold probability.

Fig. 4
figure 4

Apparent performance of the prediction model in the derivation set and validation set. A1 Calibration curve of the multivariate prediction model in the derivation set. B1 ROC curve of the multivariate prediction model in the derivation set. C1 DCA of the model in the derivation set: Y-axis represents the net benefit. The red solid line represents the CHD prediction model, the thin solid line is the hypothesis that all patients get achievement of CHD and receive treatment, and the thick one is the assumption that no patients have CHD and none receive treatment. D1 CIC of the model in the derivation set. The yellow solid line represents the number of high-risk patients and the blue dotted line is the number of high-risk patients with events in the 1000 patients.A1Calibration curve of the multivariate prediction model in the validation set. (B2) ROC curve of the multivariate prediction model in the validation set. C1 DCA of the multivariate prediction model in the validation set. C1 CIC of the multivariate prediction model in the validation set

Full size image

Performance of the clinical prediction model

Validation and evaluation of performance of the clinical prediction model was conducted using Bootstrap method. Admission data of 251 eligible patients was collected and analyzed for the internal validation of the prediction model. The results of calibration plot, AUC value, DCA analysis, and CIC analysis stemmed from the validation set were similar to those from the derivation set. In the validation set, the goodness-of-fit χ2 of CHD in middle-aged and elderly people was 11.132 (P = 0.194), which indicated no evidence of poor fit between observation and prediction (Fig. 4A2). Besides, the ROC curve revealed an AUC value of 0.783 (Fig. 4B2). Moreover, the nomogram demonstrated a high net benefit in predicting the CHD probability among middle-aged and elderly people by DCA and CIC analysis (Fig. 4C2, D2). Overall, these results showed that the novel nomogram had a good predictive power and clinical utility for the prediction of CHD probability in middle-aged and elderly people.

Discussions

Recently, with the growing morbidity and mortality related to CHD in middle-aged and elderly people, the early diagnosis and treatment of CHD have received extensive attention worldwide [3] Therefore, a new prediction model for CHD patients could be the key to early screening and diagnosis and thus improve their prognosis. In this study, we attempted to construct and verify a diagnostic model based on easily available parameters such as data on demographics, complications, clinical and laboratory indicators at baseline. A total of 932 consecutive patients with suspected CHD were retrospectively evaluated, and 839 eligible patients were enrolled in the analysis. Eight indicators were recognized as risk factors for the progression of CHD in the derivation set, of these predictors, age, HbA1c, ABI, and FMD defined to be significantly related to CHD in middle-aged and elderly people by Lasso regression analysis. Moreover, the performance of the clinical prediction model was validated in the validation set, showing high net benefit, good ability, and great clinical utility of the model according to the results of calibration plot, AUC value, DCA analysis, and CIC analysis.

This study showed that predictors such as age, HbA1c, ABI, and FMD were combined as independent risk factors in the prediction model. The main reason is that CHD is a disease caused by a variety of risk factors, which is thought to be mainly associated with age, glucose metabolism, and vascular health [1] Additionally, the result is similar to some previous reports on the risk factors of CHD [17,18,19] However, fewer concerned ABI and FMD, which are associated with peripheral artery disease (PAD), in previous models focusing on CHD. These indicators are easy to acquire in the electronic medical systems of inpatients, strengthening the ease of use and comprehensiveness of the prediction model.

Consistent with previous findings, our results demonstrated that the older the patients, the higher was the risk score in the nomogram. In detail, prevalent cases of CHD began to account for a large proportion of epidemic cases of cardiovascular disease in patients over 40 years old, and the prevalence rose steeply with elder age categories [4] According to the report [3] there were approximately 10.88 million prevalent cases of CHD in patients aged 50 to 54 years, more than three times the number of cases in patients aged 40 to 44 years.

At present, the relationship between risk factors associated with diabetes and PAD, and the occurrence and development of CHD in middle-aged and elderly people are gradually obtaining more attention [1, 20, 21] In the derivation set, patients in the CHD group had a higher HbA1c level than those in the non-CHD group, indicating that HbA1c is positively associated with the risk of CHD. In addition, a high HbA1c level was shown to be an important risk factor for glucose metabolism progression in cardiovascular disease [22, 23] Knowledge of glucose metabolism is significant on account of a well-established link between adverse cardiovascular outcomes and diabetes [24,25,26] Meanwhile, ordinary measurement of HbA1c in every patient with suspected CHD was recommended [1] As outlined in the 2019 ESC Guidelines [21] targeting near-normal HbA1c for glycemic control will reduce cardiovascular complications in diabetic patients, and less-rigorous HbA1c goals may be more appropriate for senile patients with severe comorbidities on a personalized basis. Besides, it has been shown that a reduction of approximately 1% in HbA1c was associated with a 15% reduction in the relative risk of non-fatal myocardial infarction [27] and proper glycemic control at an early stage is strongly related to long-term cardiovascular benefits [28].

It confirmed that the risk of death derived from cardiovascular causes is at a higher level in patients with large-vessel PAD [29] The ABI is a sensitive marker for arterial stiffness [30] and the FMD can be tested noninvasively to evaluate vascular endothelial function associated with atherosclerosis [31] In the derivation set, patients in the CHD group had lower ABI and FMD levels than those in the non-CHD group, indicating that the ABI and FMD levels were negatively correlated with the risk of CHD. Meanwhile, this study showed that the lower the levels, the higher was the risk score in the nomogram. Indeed, impaired ABI and FMD have been reported as early biomarkers of the development of atherosclerotic disease in previous studies, and higher values generally predict better coronary vascular outcomes [32, 33] A previous report found that patients with an FMD ≥ 10% were less strongly associated with fewer cardiovascular risk factors than those with an FMD < 10% [31] Moreover, ABI may be defined as a risk modifier in the assessment of cardiovascular risk [34] According to the American Heart Association, [35] ABI is an independent predictor of the cardiovascular event risk, even in the absence of PAD symptoms. The degree of the increased risk related to a low ABI is approximately two to three times greater in patients with diagnosed cardiovascular disease than in healthy individuals, and a decline in ABI of > 0.15 over time is related to a twofold increase in mortality [30, 35]

Among the present studies, cardiovascular risk prediction models involving traditional risk factors such as sex, age, smoking history, hypertension, diabetes mellitus, and hypercholesterolemia have been utilized to evaluate risk of future cardiovascular events [36] These prediction models, however, have limited comprehensiveness and accuracy [37] resulting in the assessment of other risk predictors such as electrocardiogram [38] or parameters of obesity such as waist circumference, [17] used in combination with other traditional risk factors or alone. More accurate recognization of high-risk individuals could facilitate the development of appropriate targeted aggressive risk reduction therapies, but more proper assessments for this strategy are still required in the future.

Furthermore, both the Hosmer–Lemeshow test and the calibration curve showed a good consistency between the actual and predicted risk of CHD, which ensures the reliability and repeatability of the CHD prediction model. Meanwhile, the discrimination of the model was assessed by the AUC value. The AUC value was 0.722 in the derivation set for the model to predict functional outcome and 0.783 in the validation set, showing that the clinical prediction model had a good predictive ability [39] Besides, the results of DCA and CIC analysis illustrated that the clinical prediction model had remarkable predictive power. DCA analysis was utilized to assess the clinical utility of the prediction models, in which the net benefit is defined as the difference between the expected benefit and the expected harm of each prediction model [40] The plots indicated that the clinical prediction model showed a greater net benefit with a wider range of threshold probabilities for predicting CHD in the two sets. The DCA for the derivation set indicated that the net benefit was maximized with threshold probabilities of 0% to 40% by the “predict all” method. Moreover, the CIC was further mapped on the basis of the DCA to assess the clinical impact, presenting the approximate number of patients with predicted CHD and the number of those who were in actual situation of illness at each risk threshold. When the risk threshold is greater than 60%, the estimated value is closer to the true number. Meanwhile, a similar trend was seen in the validation set.

The study is the first to construct and validate a clinical prediction model of CHD and investigate the value of easily accessible clinical and laboratory predictors to predict the development of CHD in middle-aged and elderly people. In addition, the nomogram we developed is beneficial to the early diagnosis of CHD, especially those who are not suitable for CAG or constrained by patients' characteristics such as those with severe comorbidities or the absence of medical conditions. Its purpose is to help clinicians make appropriate clinical treatment decisions based on individual conditions.

Limitations

Nevertheless, certain limitations of the study need to be mentioned. Firstly, this study is a retrospective single-center study with a single population source leads to a certain degree of selection bias. The study only included patients with suspected CHD, which may restrict the generalizability of the findings to patients with confirmed CHD or those without suspected CHD. Secondly, this study adopted a single time-node data modeling, which could not avoid the impact of dynamic changes in the physiological and pathological states of patients over time on the model. Subsequently, the design could be further improved, and multi-time-node data could be collected and analyzed to improve and update the model. Thirdly, the clinical performance of the constructed model was only evaluated by internal validation, therefore, the clinical value in external application needs to be further verified. The study did not assess the impact of the prediction model on patient outcomes, such as mortality or morbidity, which could help providing a more comprehensive evaluation of the model's clinical utility. We look forward to more large-scale, multi-center and prospective studies with rigorous and standardized design to verify and improve the results of this study.

Conclusion

This study established and validated a clinical prediction model for several clinical predictors related to the diagnosis probability of CHD in middle-aged and elderly people, including age, HbA1c, ABI, and FMD. The performance of the model was determined to be of high net benefit, strong ability, and great clinical utility in the validation set. The results could potentially contribute to the early diagnosis and treatment of CHD in middle-aged and elderly people, which may ultimately improve the prognosis of patients with CHD.

Availability of data and materials

All data generated or analyzed during this study are included in this published article and its additional information files.

References

  1. Knuuti J, Wijns W, Saraste A, Capodanno D, Barbato E, et al. esc scientific document group. 2019 esc guidelines for the diagnosis and management of chronic coronary syndromes. Eur Heart J. 2020;41:407–77.

    Article  PubMed  Google Scholar 

  2. GBD. Mortality and causes of death collaborators. global, regional, and national life expectancy, all-cause mortality, and cause-specific mortality for 249 causes of death 1980–2015: a systematic analysis for the Global burden of disease study 2015. Lancet. 2015;388:1459–544.

    Google Scholar 

  3. Roth GA, Johnson C, Abajobir A, Abd-Allah F, Abera SF, Abyu G, et al. Global, regional, and national burden of cardiovascular diseases for 10 causes, 1990 to 2015. J Am Coll Cardiol. 2017;70:1–25. https://doi.org/10.1016/j.jacc.2017.04.052.

    Article  PubMed  PubMed Central  Google Scholar 

  4. Roth GA, Mensah GA, Johnson CO, Addolorato G, Ammirati E, Baddour LM, et al. GBD-NHLBI-JACC global burden of cardiovascular diseases writing group. global burden of cardiovascular diseases and risk factors, 1990–2019 update from the GBD 2019 study. J Am Coll Cardiol. 2020;76:2982–3021.

    Article  PubMed  PubMed Central  Google Scholar 

  5. Nowbar AN, Howard JP, Finegold JA, Asaria P, Francis DP. 2014 global geographic analysis of mortality from ischaemic heart disease by country, age and income: statistics from World Health Organisation and United Nations. Int J Cardiol. 2014;174:293–8. https://doi.org/10.1016/j.ijcard.2014.04.096.

    Article  PubMed  PubMed Central  Google Scholar 

  6. Kattainen A, Salomaa V, Härkänen T, Jula A, Kaaja R, Kesäniemi YA, et al. Coronary heart disease: from a disease of middle-aged men in the late 1970s to a disease of elderly women in the 2000s. Eur Heart J. 2006;27:296–301. https://doi.org/10.1093/eurheartj/ehi630.

    Article  PubMed  Google Scholar 

  7. Annual Report on Cardiovascular Health and Diseases in China. 2020. J Cardiovascular Pulmonary Dis. https://doi.org/10.3969/j.issn.1007-5062.2021.09.001

  8. Ranstam J, Cook JA, Collins GS. Clinical prediction models. Br J Surg. 2016;103:1886. https://doi.org/10.1002/bjs.10242.

    Article  PubMed  CAS  Google Scholar 

  9. Adams ST, Leveson SH. Clinical prediction rules. BMJ. 2012;344:d8312. https://doi.org/10.1136/bmj.d8312.

    Article  PubMed  Google Scholar 

  10. Nijman RG, Vergouwe Y, Thompson M, van Veen M, van Meurs AH, van der Lei J, et al. Clinical prediction model to aid emergency doctors managing febrile children at risk of serious bacterial infections: diagnostic study. BMJ. 2013;346:f1706. https://doi.org/10.1136/bmj.f1706.

    Article  PubMed  PubMed Central  Google Scholar 

  11. Jaja BNR, Saposnik G, Lingsma HF, Macdonald E, Thorpe KE, Mamdani M, et al. SAHIT collaboration development and validation of outcome prediction models for aneurysmal subarachnoid haemorrhage: the SAHIT multinational cohort study. BMJ. 2018;360:j5745. https://doi.org/10.1136/bmj.j5745.

    Article  PubMed  Google Scholar 

  12. Kattan MW, Marasco J. What is a real nomogram? Semin Oncol. 2010;37:23–6. https://doi.org/10.1053/j.seminoncol.2009.12.003.

    Article  PubMed  Google Scholar 

  13. Florkowski CM. Sensitivity, specificity, receiver-operating characteristic (ROC) curves and likelihood ratios: communicating the performance of diagnostic tests. Clin Biochem Rev. 2008;29:S83–7.

    PubMed  PubMed Central  Google Scholar 

  14. Nomenclature and Criteria for diagnosis of ischemic heart disease. Report of the joint international society and federation of cardiology/world Health Organization task force on standardization of clinical nomenclature. Circulation. 2015;59(607):609. https://doi.org/10.1161/01.cir.59.3.607.

    Article  Google Scholar 

  15. Collins GS, Reitsma JB, Altman DG, Moons KG. Transparent reporting of a multivariable prediction model for individual prognosis or diagnosis (TRIPOD): the TRIPOD statement. BMJ. 2015;350:g7594. https://doi.org/10.1136/bmj.g7594.

    Article  PubMed  Google Scholar 

  16. Devereux RB, Alonso DR, Lutas EM, Gottlieb GJ, Campo E, Sachs I, et al. Echocardiographic assessment of left ventricular hypertrophy: comparison to necropsy findings. Am J Cardiol. 1986;57:450–8. https://doi.org/10.1016/0002-9149(86)90771-x.

    Article  PubMed  CAS  Google Scholar 

  17. Yang X, Li J, Hu D, Chen J, Li Y, Huang J, et al. Predicting the 10-year risks of atherosclerotic cardiovascular disease in chinese population: the china-par project (prediction for ASCVD Risk in China). Circulation. 2016;134:1430–40. https://doi.org/10.1161/CIRCULATIONAHA.116.022367.

    Article  PubMed  Google Scholar 

  18. Piniés JA, González-Carril F, Arteagoitia JM, Irigoien I, Altzibar JM, Rodriguez-Murua JL, et al. Sentinel practice network of the basque country. development of a prediction model for fatal and non-fatal coronary heart disease and cardiovascular disease in patients with newly diagnosed type 2 diabetes mellitus: the basque country prospective complications and mortality study risk engine (BASCORE). Diabetologia. 2014;57:2324–33. https://doi.org/10.1007/s00125-014-3370-1.

    Article  PubMed  CAS  Google Scholar 

  19. Bendermacher BL, Teijink JA, Willigendael EM, Bartelink ML, Peters RJ, de Bie RA, et al. A clinical prediction model for the presence of peripheral arterial disease–the benefit of screening individuals before initiation of measurement of the ankle-brachial index: an observational study. Vasc Med. 2007;12:5–11. https://doi.org/10.1177/1358863X07076827.

    Article  PubMed  Google Scholar 

  20. Shishehbor MH, Castro-Dominguez Y. Peripheral artery disease: the new and old silent killer. J Am Coll Cardiol. 2022;79:1236–8. https://doi.org/10.1016/j.jacc.2022.02.006.

    Article  PubMed  Google Scholar 

  21. Cosentino F, Grant PJ, Aboyans V, Bailey CJ, Ceriello A, Delgado V, et al. ESC scientific document group. 2019 ESC guidelines on diabetes, pre-diabetes, and cardiovascular diseases developed in collaboration with the EASD. Eur Heart J. 2020;41:255–323. https://doi.org/10.1093/eurheartj/ehz486.

    Article  PubMed  Google Scholar 

  22. Hirakawa Y, Arima H, Zoungas S, Ninomiya T, Cooper M, Hamet P, et al. Impact of visit-to-visit glycemic variability on the risks of macrovascular and microvascular events and all-cause mortality in type 2 diabetes: the advance trial. Diabetes Care. 2014;37:2359–65. https://doi.org/10.2337/dc14-0199.

    Article  PubMed  CAS  Google Scholar 

  23. Borg R, Kuenen JC, Carstensen B, Zheng H, Nathan DM, Heine RJ, et al. ADAG study group. HbA(c) and mean blood glucose show stronger associations with cardiovascular disease risk factors than do postprandial glycaemia or glucose variability in persons with diabetes: the A1C-derived average glucose (ADAG) study. Diabetologia. 2011;54:69–72. https://doi.org/10.1007/s00125-010-1918-2.

    Article  PubMed  CAS  Google Scholar 

  24. Sarwar N, Gao P, Seshasai SR, Gobin R, Kaptoge S, Di Angelantonio E, et al. Emerging risk factors collaboration. diabetes mellitus, fasting blood glucose concentration, and risk of vascular disease: a collaborative meta-analysis of 102 prospective studies. Lancet. 2010;375:2215–22. https://doi.org/10.1016/S0140-6736(10)60484-9.

    Article  PubMed  CAS  Google Scholar 

  25. Rawshani A, Sattar N, Franzén S, Rawshani A, Hattersley AT, Svensson AM, et al. Excess mortality and cardiovascular disease in young adults with type 1 diabetes in relation to age at onset: a nationwide, register-based cohort study. Lancet. 2018;392:477–86. https://doi.org/10.1016/S0140-6736(18)31506-X.

    Article  PubMed  PubMed Central  Google Scholar 

  26. Tancredi M, Rosengren A, Svensson AM, Kosiborod M, Pivodic A, Gudbjörnsdottir S, et al. Excess mortality among persons with type 2 diabetes. N Engl J Med. 2015;373:1720–32. https://doi.org/10.1056/NEJMoa1504347.

    Article  PubMed  CAS  Google Scholar 

  27. Ray KK, Seshasai SR, Wijesuriya S, Sivakumaran R, Nethercott S, Preiss D, et al. Effect of intensive control of glucose on cardiovascular outcomes and death in patients with diabetes mellitus: a meta-analysis of randomised controlled trials. Lancet. 2009;373:1765–72. https://doi.org/10.1016/S0140-6736(09)60697-8.

    Article  PubMed  CAS  Google Scholar 

  28. Laiteerapong N, Ham SA, Gao Y, Moffet HH, Liu JY, Huang ES, et al. The legacy effect in type 2 diabetes: impact of early glycemic control on future complications (the diabetes & aging study). Diabetes Care. 2019;42:416–26. https://doi.org/10.2337/dc17-1144.

    Article  PubMed  CAS  Google Scholar 

  29. Criqui MH, Langer RD, Fronek A, Feigelson HS, Klauber MR, McCann TJ, et al. Mortality over a period of 10 years in patients with peripheral arterial disease. N Engl J Med. 1992;326:381–6. https://doi.org/10.1056/NEJM199202063260605.

    Article  PubMed  CAS  Google Scholar 

  30. Wang H, Wu X, Gu Y, Zhou J, Wu J. Relationship of noninvasive assessment of arterial stiffness with 10-year atherosclerotic cardiovascular disease (ASCVD) risk in a general middle-age and elderly population. Int J Gen Med. 2021;14:6379–87. https://doi.org/10.2147/IJGM.S330142.

    Article  PubMed  PubMed Central  Google Scholar 

  31. Kuvin JT, Patel AR, Sliney KA, Pandian NG, Rand WM, Udelson JE, et al. Peripheral vascular endothelial function testing as a noninvasive indicator of coronary artery disease. J Am Coll Cardiol. 2001;38:1843–9. https://doi.org/10.1016/s0735-1097(01)01657-6.

    Article  PubMed  CAS  Google Scholar 

  32. Stone K, Fryer S, Faulkner J, Meyer ML, Heffernan K, Kucharska-Newton A, et al. Associations of lower-limb atherosclerosis and arteriosclerosis with cardiovascular risk factors and disease in older adults: the atherosclerosis risk in communities (ARIC) study. Atherosclerosis. 2022;340:53–60. https://doi.org/10.1016/j.atherosclerosis.2021.10.014.

    Article  PubMed  CAS  Google Scholar 

  33. Thijssen DHJ, Bruno RM, van Mil ACCM, Holder SM, Faita F, Greyling A, et al. Expert consensus and evidence-based recommendations for the assessment of flow-mediated dilation in humans. Eur Heart J. 2019;40:2534–47. https://doi.org/10.1093/eurheartj/ehz350.

    Article  PubMed  Google Scholar 

  34. Hanssen NM, Huijberts MS, Schalkwijk CG, Nijpels G, Dekker JM, Stehouwer CD. Associations between the ankle-brachial index and cardiovascular and all-cause mortality are similar in individuals without and with type 2 diabetes: nineteen-year follow-up of a population-based cohort study. Diabetes Care. 2012;35:1731–5. https://doi.org/10.2337/dc12-0178.

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  35. Hiatt WR, Goldstone J, Smith SC Jr, McDermott M, Moneta G, Oka R, et al. American heart association writing group 1. atherosclerotic peripheral vascular disease symposium ii: nomenclature for vascular diseases. Circulation. 2008;118:2826–9. https://doi.org/10.1161/CIRCULATIONAHA.108.191171.

    Article  PubMed  Google Scholar 

  36. Greenland P, Smith SC Jr, Grundy SM. Improving coronary heart disease risk assessment in asymptomatic people: role of traditional risk factors and noninvasive cardiovascular tests. Circulation. 2001;104:1863–7. https://doi.org/10.1161/hc4201.097189.

    Article  PubMed  CAS  Google Scholar 

  37. Brindle P, Beswick A, Fahey T, Ebrahim S. Accuracy and impact of risk assessment in the primary prevention of cardiovascular disease: a systematic review. Heart. 2006;92:1752–9. https://doi.org/10.1136/hrt.2006.087932.

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  38. Kannel WB, McGee D, Gordon T. A general cardiovascular risk profile: the Framingham study. Am J Cardiol. 1976;38:46–51. https://doi.org/10.1016/0002-9149(76)90061-8.

    Article  PubMed  CAS  Google Scholar 

  39. Swets JA. Measuring the accuracy of diagnostic systems. Science. 1988;240:1285–93. https://doi.org/10.1126/science.3287615.

    Article  PubMed  CAS  Google Scholar 

  40. Fitzgerald M, Saville BR, Lewis RJ. Decision curve analysis. JAMA. 2015;313:409–10. https://doi.org/10.1001/jama.2015.37.

    Article  PubMed  CAS  Google Scholar 

Download references

Acknowledgements

We thank the patients whose anonymized data were used for this research, as well as numerous doctors and nurses working together to fight against this coronary artery disease.

Funding

This study was supported by grants from the National Natural Science Foundation of China (Grant No. 81703894) and the Elite Medical Professionals Project of China-Japan Friendship Hospital (Grant NO. ZRJY2021-GG10).

Author information

Authors and Affiliations

  1. Graduate School, Beijing University of Chinese Medicine, Beijing, China

    Shiyi Tao & Lintong Yu

  2. Department of Integrative Cardiology, China-Japan Friendship Hospital, Beijing, China

    Deshuang Yang, Li Huang & Mingjing Shao

  3. Department of Internal Medicine, Shenzhen Nanshan Chinese Medicine Hospital, Guangdong, China

    Ruiqi Yao

  4. Department of Oncology, Guang’anmenHospital, China Academy of Chinese Medical Sciences, Beijing, China

    Lanxin Zhang

Authors
  1. Shiyi Tao
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Lintong Yu
    View author publications

    You can also search for this author in PubMed Google Scholar

  3. Deshuang Yang
    View author publications

    You can also search for this author in PubMed Google Scholar

  4. Ruiqi Yao
    View author publications

    You can also search for this author in PubMed Google Scholar

  5. Lanxin Zhang
    View author publications

    You can also search for this author in PubMed Google Scholar

  6. Li Huang
    View author publications

    You can also search for this author in PubMed Google Scholar

  7. Mingjing Shao
    View author publications

    You can also search for this author in PubMed Google Scholar

Contributions

ST and MS conceived and designed the study. ST, LY and DY conducted the statistical analyses. ST, LY, RY and LZ drafted the manuscript. RY, LZ and LH consulted and supplemented the relevant information. All authors approved the final manuscript.

Corresponding author

Correspondence to Mingjing Shao.

Ethics declarations

Ethics approval and consent to participate

The study was conducted in accordance with the Declaration of Helsinki. We analyzed a dataset which has been collected for the previous studies and information related to patient identity was concealed. Thus, ethics approval and consent for participation is not applicable for this study.

Competing interests

The authors declare that they have no competing interests.

Additional information

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary Information

Additional file 1:

Table S1. Basic characteristics of patients in the validation set.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Tao, S., Yu, L., Yang, D. et al. Development and validation of a clinical prediction model for detecting coronary heart disease in middle-aged and elderly people: a diagnostic study. Eur J Med Res 28, 375 (2023). https://doi.org/10.1186/s40001-023-01233-0

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doi.org/10.1186/s40001-023-01233-0

Keywords

European Journal of Medical Research

ISSN: 2047-783X

Contact us