Summary
Background
Routine management of hypertensive adults is based on assessment of risk factors for coronary artery disease; risk factors for heart failure (HF) remain poorly investigated despite the key role of hypertension in HF development.
Aim
To assess the components of HF risk in hypertensive adults in primary care, compare physicians’ estimations of HF and global cardiovascular risks with established calculation algorithms, and assess the concordance of these algorithms.
Methods
O-PREDICT was a transverse, observational, multicentre French survey conducted in 2006 among general practitioners who included the first hypertensive, non-HF patient seen in each of three age classes (< 60, 60–70, > 70 years). Estimations of HF and global cardiovascular risks (at 4 and 10 years, respectively) were performed subjectively during the consultation and calculated a posteriori according to algorithms from the Framingham cohort and the European SCORE database, respectively. For each of these methods, patients were stratified into four risk categories (i.e., no, low, moderate, high).
Results
One thousand five hundred and thirty seven physicians recruited 4523 patients (61% men; 64.5 ± 10.9 years; systolic blood pressure 149.9 ± 15.4 mmHg); most (67.2%) patients had one or two cardiovascular/HF risk factors (dyslipidaemia 48.8%, left ventricular hypertrophy 25.3%, diabetes 18.8%, coronary artery disease 8.8%, valvulopathy 6.1%); the number increased with advancing age and in men versus women. According to the Framingham algorithm, the risk of HF (mean 5.4 ± 8.5%; 13.4% of patients at high risk) increased with advancing age ( p < 0.001), nearly doubling for each decade increase. According to the European SCORE system, global cardiovascular risk (mean 5.4 ± 4.3%) was moderate or elevated in 48.1% of patients. Concordance between physicians’ estimations and theoretical calculations for HF and global risks was poor, as was concordance between algorithms ( κ w = 0.28, 0.12, 0.11, respectively).
Conclusion
More than one in 10 hypertensive patients seen in primary care is at high risk of HF at 4 years according to the Framingham model; this algorithm appears to offer additional information to that provided by the SCORE system. Physicians’ estimations of risks correlated poorly with algorithm calculations, suggesting that the use of these tools in general practice should be encouraged.
Résumé
Justification
La prise en charge des adultes hypertendus est généralement basée sur l’évaluation des facteurs de risque de coronaropathie, alors que les facteurs de risque d’insuffisance cardiaque (IC) restent peu évalués malgré le rôle crucial de l’hypertension artérielle dans le développement de l’IC.
Objectifs
Évaluer les composantes du risque d’IC chez des patients hypertendus vus en médecine générale, et comparer les estimations par le médecin du risque d’IC et du risque cardiovasculaire global par rapport au risque calculé par des échelles validées, et enfin évaluer la concordance entre ces deux estimations.
Méthode
O-PREDICT est une étude observationnelle, transversale, multicentrique conduite en France en 2006 chez des médecins généralistes qui devaient inclure le premier patient hypertendu sans IC vu en consultation au sein de trois tranches d’âge (< 60, 60–70, > 70 ans). Les estimations du risque d’IC et du risque cardiovasculaire global (à 4–10 ans, respectivement) étaient réalisées de façon subjective pendant la consultation et calculées a posteriori sur des algorithmes publiés basés sur la cohorte de Framingham et le SCORE Européen, respectivement. Pour chacune de ces trois méthodes, les patients étaient classés en quatre catégories de risque: (nul, faible, modéré ou élevé).
Résultats
Mille cinq cent trente-sept médecins ont recruté 4523 patients (61 % d’hommes ; âge 64,5 ± 10,9 ans ; pression artérielle systolique: 149,9 ± 15,4 mmHg), la majorité (67,2 %) avec un ou deux facteurs de risque cardiovasculaire/IC [dyslipidémie (48,8 %), hypertrophie ventriculaire gauche (25,3 %), diabète (18,8 %), coronaropathie (8,8 %), valvulopathie (6,1 %)], facteurs plus fréquents avec l’âge et chez les hommes. Selon l’algorithme de Framingham, le risque d’IC (moyenne 5,4 ± 8,5 % ; risque élevé chez 13,4 % des patients) doublant presque pour chaque décade d’âge ( p < 0,001). Selon l’algorithme SCORE, le risque cardiovasculaire global (moyenne 5,4 ± 4,3 %), était modéré ou élevé pour 48,2 % des patients. Les concordances entre les estimations du médecin et les calculs des risques d’IC et cardiovasculaire global étaient mauvaises, de même que la concordance entre les deux algorithmes ( κ w = 0,28, 0,12 et 0,11, respectivement).
Conclusion
Plus d’un patient hypertendu sur dix vus en médecine générale est à haut risque de développer une IC à quatre ans selon l’algorithme de Framingham, qui semble complémentaire de l’échelle SCORE. La corrélation entre les risques estimés par le médecin et les risques calculés restent faibles, ce qui devrait encourager à utiliser ces algorithmes en médecine générale.
Background
The routine management of hypertensive adults is based on the assessment of risk factors for coronary artery disease and on the concept of global cardiovascular risk. Concomitant intensive management of multiple risk factors has emerged as a useful approach for risk reduction . Thus, the European Systematic Coronary Risk Evaluation (SCORE) system has been proposed to provide a numerical score related to different risk factors such as sex, age, smoking status, systolic blood pressure (BP), total cholesterol and diabetes, in order to stratify individual patients according to their 10-year risk for coronary artery disease mortality . Recalibration of this algorithm is necessary in low-risk European populations, such as in France .
Recent American and European recommendations concerning the management of chronic heart failure (HF) have highlighted high BP as one of the main precipitating factors . These guidelines have also identified a group of at-risk patients without clinically evident disease or structural cardiac abnormalities, but with classic risk factors for HF, who should form an important focus for modern healthcare policies. It remains that HF risk factors, particularly in hypertensive patients, are poorly investigated in current clinical practice despite the key role of hypertension in HF development. An analysis from the Framingham cohort has shown that identification of patients at high risk of developing HF within 4 years is possible using a calculation algorithm, which takes into account risk factors for HF including arterial pressure . However, these validated tools are not currently used in clinical practice, and both HF and cardiovascular risks appear to be underestimated, particularly in patients at highest risk .
Thus, the aims of the Observatoire de la prise en charge et de l’évaluation du risque d’insuffisance cardiaque chez les patients hypertendus (O-PREDICT) survey, conducted by French general practitioners (GPs) during routine practice, were to assess the components of HF risk in hypertensive adults in primary care and to compare physicians’ estimations of HF and global cardiovascular risks with those from established algorithms.
Methods
The O-PREDICT survey was a transverse, observational, multicentre survey conducted by GPs in France from March 2006 to July 2006. Each GP included the first adult (age ≥ 18 years) hypertensive, non-HF patient seen in each of three age classes (< 60, 60–70, > 70 years). Hypertension was defined as systolic BP ≥ 140 mmHg and/or diastolic BP ≥ 90 mmHg or treated hypertension.
The primary study objective was to assess the incidence of each component of HF risk in hypertensive patients free of HF in primary care. Secondary objectives were to calculate the 4-year risk of HF according to the Framingham cohort and the 10-year global cardiovascular risk using the European SCORE system , and then to compare the physicians’ estimations done during the GP visit with these calculations performed a posteriori . The concordance between both risk calculations was analysed.
Data collection and risk estimations
A questionnaire was completed by the physician for each patient, which collected clinical information, medical and cardiovascular history, cardiovascular risk factors defined by the French Haute Autorité de santé , clinical symptoms, outpatient and ambulatory BP measurements when available, complementary investigations, and BP history. Based on these data, GPs estimated the 4-year HF risk and the 10-year global cardiovascular risk level for each patient as no risk, low, moderate or high risk.
Risk calculations
Both risks were then calculated a posteriori from the final database, using algorithms for men and women, and for each age category, to be crossed with the physicians’ estimations. The 4-year HF risk was calculated based on the predisposing conditions from the Framingham cohort (i.e., age, systolic BP, heart rate, electrocardiographic and/or echocardiographic left ventricular hypertrophy, body mass index [BMI], coronary artery disease, valvulopathy and diabetes). The 10-year global cardiovascular risk was calculated according to the European SCORE system, based on sex, age, smoking status, systolic BP and cholesterol concentration . Patients were stratified consecutively into four levels of risk: no risk (≤ 1%), low (2–4%), moderate (5–9%) and high (≥ 10%) risk for HF at 4 years, and no risk (≤ 1%), low (2–4%), moderate (5–14%) and high (≥ 15%) risk for global cardiovascular risk at 10 years.
Concordance between HF and cardiovascular risk calculations
To analyse concordance between the Framingham HF risk model and the SCORE system, the following risk categories were used: ≤ 1%, 2–4%, 5–9%, 10–14% and ≥ 15%.
Statistical analysis
The assessment was done for the whole population as well for each sex and within age categories. Based on an expected minimal percentage of 4% of patients presenting with a cardiovascular factor included in the HF risk assessment and a survey accuracy fixed to 1.5%, the number of patients needed in each group was estimated to be 656 (i.e., 1640 patients in each age group for an expected maximal sex ratio of 1.5), with a total of 4920 patients for the survey . Descriptive statistics were mean, standard deviation, minimum, maximum and median values for quantitative parameters, and frequency and percentage for qualitative parameters. Between-group comparisons were made using the t-test for continuous variables and the Chi-square test for categorical variables. Multivariable regression analyses were used to assess the independent prognostic value of variables in Cox models and results are expressed as odds ratios with 95% confidence intervals; hence, concordance between the GPs’ estimates of risk and the calculated levels according to the Framingham or the SCORE system was evaluated using weighted kappa ( κ w ). To interpret the kappa ratings, benchmarks were used as suggested by Landis and Koch . All tests were bilateral with a α risk equal to 0.05. Statistical analysis was performed using SAS ® software, version 8.2 (SAS Instituted Inc, Cary, NC, USA).
Results
Study-group characteristics
A total of 1537 GPs from all metropolitan areas in France participated in the O-PREDICT survey. Of the 4585 recruited patients, 62 (1.4%) were excluded from the analysis due to major deviations such as missing age ( n = 42) and sex ( n = 19), or normal BP and no antihypertensive treatment ( n = 2). Consequently, 4523 hypertensive adult patients with a mean age of 65 ± 11 years formed the study group. The characteristics of these patients are shown in Table 1 . Sex distribution was similar across the three age groups.
| Total | Men | Women | p a | |
|---|---|---|---|---|
| Patients, n (%) | 4523 | 2765 (61.1) | 1758 (38.9) | |
| Age in years b , n (%) | 64.5 ± 10.9 | – | – | |
| < 60 | 1527 (33.8) | 1059 (38.3) | 468 (26.6) | ND |
| 60–70 | 1531 (33.9) | 900 (32.5) | 631 (35.9) | ND |
| > 70 | 1465 (32.4) | 806 (29.2) | 659 (37.5) | ND |
| Systolic BP b (mmHg) | 149.9 ± 15.4 | 150.3 ± 15.0 | 149.4 ± 15.8 | 0.032 |
| Diastolic BP (mmHg) | 86.8 ± 10.2 | 87.3 ± 10.1 | 85.9 ± 10.2 | ND |
| Heart rate b (beats/minute) | 74.9 ± 9.0 | 74.8 ± 9.1 | 75.0 ± 8.7 | 0.39 |
| BMI b (kg/m 2 ) | 27.7 ± 4.6 | 27.9 ± 4.1 | 27.5 ± 5.3 | 0.006 |
| BMI > 30 kg/m 2 , n (%) | 1104 (24.8) | 662 (24.4) | 442 (25.6) | ND |
| Total cholesterol (g/L) | 2.2 ± 0.4 | 2.2 ± 0.4 | 2.2 ± 0.4 | ND |
| High-density lipoprotein cholesterol (g/L) | 0.6 ± 0.2 | 0.6 ± 0.2 | 0.6 ± 0.2 | ND |
| Dyslipidaemia, n (%) | 2063 (48.8) | 1312 (50.8) | 751 (45.8) | ND |
| Sedentary lifestyle, n (%) | 2909 (64.0) | 1695 (60.1) | 1214 (68.7) | ND |
| Alcohol misuse, n (%) | 956 (21.3) | 852 (31.1) | 104 (6.0) | ND |
| Diabetes b , n (%) | 842 (18.8) | 542 (19.7) | 300 (17.2) | 0.0026 |
| Left ventricular hypertrophy b , c , n (%) | 816 (25.3) | 535 (26.6) | 281 (23.3) | 0.0004 |
| Coronary artery disease b , n (%) | 392 (8.8) | 301 (11.1) | 91 (5.25) | < 0.0001 |
| Heart valve disease b , n (%) | 270 (6.1) | 162 (6.0) | 108 (6.3) | 0.41 |
| Symptoms/abnormal signs, n (%) | 1710 (38.2) | 997 (36.4) | 713 (41.0) | ND |
| Exertion dyspnoea | 1118 (65.4) | 635 (63.7) | 483 (67.7) | ND |
| Lower limb oedema | 447 (26.1) | 234 (23.5) | 213 (29.9) | ND |
| Cardiac murmur | 271 (15.9) | 171 (17.2) | 100 (14.0) | ND |
| Palpitations | 223 (13.0) | 148 (14.8) | 75 (10.5) | ND |
| Sleep apnoea syndrome | 214 (12.5) | 181 (18.2) | 33 (4.6) | ND |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
