The assessment of total cardiovascular risk in hypertensive subjects in primary care

Abstract

Background. Recently published guidelines emphasize that detection of any subclinical target organ damage in hypertensive subjects should be regarded as a sign of high cardiovascular risk.

Aim. To assess the ability of conventional multivariable cardiovascular disease risk prediction tools and high-sensitivity C-reactive protein (hs-CRP) to identify hypertensive subjects with target organ damage.

Methods. Ankle-brachial index (ABI), estimated glomerular filtration rate (eGFR), electrocardiographically determined left ventricular hypertrophy (ECG-LVH), and base-line variables were measured in hypertensive subjects aged 45–70 years without established cardiovascular or renal disease or known diabetes.

Results. Of the 495 subjects, 123 (24.8% (95% CI 21.1–28.9)) had ABI <1.00, 81 (16.4% (95% CI 13.2–19.9)) had ECG-LVH, and 41 (8.3% (95% CI 6.0–11.1)) had eGFR <60 mL/min/1.73 m². In patients with SCORE <5% or Framingham risk <20%, any sign of target organ damage was found in 46% and 49% of patients, respectively.

Conclusion. Assessment of ECG-LVH, ABI, and eGFR reclassifies a significant number of hypertensive patients to the high-risk category as compared to SCORE and Framingham risk prediction tools only.

Key words::

• Ankle-brachial index
• cardiovascular risk factors
• estimated glomerular filtration rate
• Framingham model
• high-sensitivity C-reactive protein
• left ventricular hypertrophy
• SCORE
• target organ damage

Key messages

• Subclinical hypertensive target organ damage, as assessed by ABI <1.0, presence of ECG-LVH, or eGFR <60 mL/min/1.73 m², are common in hypertensive subjects even without previously diagnosed cardiovascular or renal disease or diabetes.

• Assessment of ECG-LVH, ABI, and eGFR reclassifies a significant number of hypertensive patients to the high-risk category as compared to SCORE and Framingham risk prediction tools only.

• In attempting to prevent first-time cardiovascular events a strategy which might identify the real high-risk patients is targeting and treating individuals with asymptomatic atherosclerosis.

Abbreviations
ABI	=	ankle-brachial index
ADP	=	dorsalis pedis artery
ATP	=	posterior tibial artery
AUC	=	area under the curve
CVD	=	cardiovascular disease
DBP	=	diastolic blood pressure
ECG-LVH	=	electrocardiographically determined left ventricular hypertrophy
eGFR	=	estimated glomerular filtration rate
HDL	=	high-density lipoprotein
hs-CRP	=	high-sensitivity C-reactive protein
LDL	=	low-density lipoprotein
MDRD	=	Modification of Diet in Renal Disease
ROC	=	receiver-operating characteristic
SBP	=	systolic blood pressure
SCORE	=	Systematic COronary Risk Evaluation system

Introduction

Hypertension is the leading global risk factor for mortality and the third most important cause for global burden of disease (1). It has been estimated that every other death from coronary heart disease (47%) and from stroke (54%) worldwide are attributable to high blood pressure (2). The occurrence of these major cardiovascular events is usually the result of long-term exposure to cardiovascular risk factors and, in most hypertensive subjects, is preceded by the development of asymptomatic structural and functional abnormalities at the vascular and cardiac level (3). Recently published guidelines emphasize that detection of any subclinical target organ damage in hypertensive subjects should lead to shifting these patients to the high-risk category (4–6).

The significance of target organ damage in determining the total cardiovascular risk is dependent on how carefully the damage is assessed, based on available facilities (4). The easily obtained markers of atherosclerotic disease and increased total cardiovascular risk in the primary care setting are electrocardiographically determined left ventricular hypertrophy (ECG-LVH), low ankle-brachial index (ABI), and low estimated glomerular filtration rate (eGFR). However, searching for these signs of target organ damage and determining other risk factors in order to assess the total cardiovascular risk of a hypertensive patient is a time- and effort-consuming task for a clinical practitioner.

We investigated whether multivariable risk prediction tools commonly used to predict general cardiovascular disease (CVD) risk, the Systematic COronary Risk Evaluation (SCORE) system (7), and the recently introduced Framingham general CVD risk prediction model (8) could reliably identify hypertensive subjects with high cardiovascular risk. We also compared the ability of plasma high-sensitivity C-reactive protein (hs-CRP) to detect subclinical organ damage in a cohort of hypertensive subjects in primary prevention.

Materials and methods

Subjects

The Harmonica Project is a population survey designed to evaluate cardiovascular risk factors in people aged 45–70 years living in two communities, Harjavalta and Kokemäki, in south-western Finland. A two-stage screening strategy was used: a risk factor survey was mailed to 6013 inhabitants, and, out of the 4450 (74%) subjects willing to participate in the project, those having at least one cardiovascular risk factor (n = 2752) were invited to an enrolment examination performed by a trained nurse.

The enrolment examination included medical history, physical examination, ECG, laboratory tests, and an oral glucose tolerance test to those subjects without previously diagnosed diabetes mellitus. High-risk subjects with hypertension, metabolic syndrome, newly detected glucose disorders, body mass index ≥30 kg/m², or a 10-year risk of cardiovascular disease death of 5% or more according to the SCORE system (7) (n = 1928) were further examined by an internist in order to evaluate secondary causes and target organ damage of hypertension or metabolic disorders. In Harjavalta, ABI was measured in 972/1076 (90%) asymptomatic high-risk subjects who did not have known cardiovascular or renal disease or diabetes mellitus. A detailed description of the enrolment and examination methods has been published earlier (9).

In this paper, we analyse the test results of 495 hypertensive subjects who had valid measurements of ABI, hs-CRP, and renal function. Patients with established CVD, hs-CRP values >10 mg/L, previously diagnosed diabetes, or renal disease were excluded from the study (Figure 1).

Figure 1. Design of the study. (SCORE = Systematic COronary Risk Evaluation; BMI = body mass index; ABI = ankle-brachial index.)

Definitions

Hypertension was defined as the use of antihypertensive medication, or as the mean of home blood pressure monitoring ≥135 mmHg for systolic blood pressure (SBP) or ≥85 mmHg for diastolic blood pressure (DBP) (10). If study subjects had no antihypertensive medication at enrolment, and the study nurse measured SBP ≥140 mmHg or DBP ≥90 mmHg, subjects were taught to use an automatic validated blood pressure monitor (Omron® M4-1, Japan) which was lent to them for home blood pressure monitoring. The subjects whose arm circumference was >32 cm used a larger cuff. The subjects were instructed to take duplicate blood pressure measurements in the seated position after 5 minutes of rest in the morning and evening for 1 week. The recorded measurements except those from the first day were used to calculate the mean home blood pressure, as recommended by the recent guidelines of the European Society of Hypertension (10). Pulse pressure was calculated by subtracting the mean DBP from the mean SBP.

Standard resting 12-lead ECGs were digitally recorded and stored as digital data with a Welch Allyn CardioPerfect™ system (Welch Allyn Inc., NY, US). LVH was diagnosed if the Sokolow-Lyon voltage (S_V1 + R_V5 − 6) was >38 mm or the Cornell product (Cornell voltage) (R_aVL + S_V3 plus 6 mm for women × QRS duration) was >2440 mm × ms (11,12).

Renal function was estimated by the plasma creatinine level (enzymatic method; Olympus® AU640, Japan) and eGFR. Because our test method of plasma creatinine has been calibrated to be traceable to isotope dilution mass spectrometry (IDMS), we calculated eGFR using the recently developed modified four-variable Modification of Diet in Renal Disease (MDRD) Study equation 175 × (P_Cr/88.4)^−1.154 × (Age)^−0.203 × (0.742 if female) × (1.21 if black), where P_Cr = plasma creatinine in μmol/L, and age is expressed in years (13). Race was not applicable in our study because all patients were white. Because the MDRD formula is based on data from patients with advanced renal failure, the results may not be valid in subjects with normal or near normal glomerular filtration rates. Therefore we report only eGFR levels <60 mL/min/1.73 m² as renal dysfunction.

In regard to the multivariable risk prediction tools, a high-risk subject is defined as having ≥5% risk of dying from CVD over 10 years according to the SCORE system (7) or a 10-year estimate of >20% risk of CVD event according to the Framingham general CVD risk prediction model (8). When calculating the SCORE risk estimates, we multiplied the risks of the diabetic subjects by four in women and by two in men (7).

ABI measurement

ABI was determined from blood pressure measurements in the arms and ankles with the patient supine. SBP in the brachial artery was measured in both arms using a blood pressure cuff and Doppler instrument (UltraTec® PD1v with a vascular probe of 5 MHz, United Kingdom) in the antecubital fossa. SBP at the left and right dorsalis pedis arteries (ADP) and if not found at the left and right posterior tibial arteries (ATP) was then measured with Doppler detection with a blood pressure cuff applied to the ankle just proximal to the malleoli. ABI was the lower ankle SBP divided by the higher brachial SBP. Normal ABI was defined as 1.00–1.40.

Laboratory tests

The laboratory tests were determined in blood samples which were obtained after at least 12 hours of fasting.

An oral glucose tolerance test was performed by measuring a fasting plasma glucose and a 2-hour plasma glucose from capillary whole blood after ingestion of a glucose load of 75 g with the HemoCue® Glucose 201+ system (Ängelholm, Sweden), which converts the result from capillary whole blood to plasma glucose values. According to the World Health Organization criteria, diabetes was diagnosed if 2-hour plasma glucose concentration was ≥12.2 mmol/L (14).

Total cholesterol, high-density lipoprotein (HDL) cholesterol, and triglycerides were measured enzymatically (Olympus® AU640, Japan). Low-density lipoprotein (LDL) cholesterol was calculated according to Friedewald’s formula.

High-sensitivity CRP was assayed using a microparticle-enhanced turbidometric method on a Konelab 60i analyser (Thermo Electron, Vantaa, Finland). The cut-off point of high risk was defined hs-CRP >3 mg/L (15). Patients with hs-CRP values >10 mg/L were excluded from the study.

Informed consent

All of the participants provided written informed consent for the project and subsequent medical research. The study protocol and consent forms were reviewed and approved by the ethics committee of Satakunta Hospital District.

Statistics

The data are given as mean with standard deviation, median with interquartile range, or as counts with percentages. The statistical comparison between genders was made by bootstrap type t test. The diagnostic characteristics of the test, i.e. accuracy, sensitivity, specificity, and positive likelihood ratios, were calculated for each test (Framingham CVD risk, SCORE risk, and hs-CRP) in predicting target organ damage as assessed by ABI, eGFR, and ECG-LVH. ROC (receiver-operating characteristic) curves were used for determination of optimal cut-off point, and the respective areas under the curve were calculated with a bias-corrected bootstrap confidence interval. The optimal cut-off point maximizes the sum of sensitivity and specificity. Differences between the areas under the curves (AUC) were evaluated using an algorithm by DeLong (16).

Results

Base-line characteristics and medication use of the subjects are presented in Table I. There were no differences between genders in the prevalence of ECG-LVH (P = 0.11) or mean ABI values (P = 0.19), but the mean hs-CRP (P = 0.011) level was higher in female subjects and mean eGFR higher in men (P < 0.001). According to oral glucose tolerance test, previously unknown diabetes was diagnosed in 34/495 (6.9%) without significant difference between the genders (P = 0.91).

Table I. Base-line characteristics of the study population.

	Femalen = 272	Malen = 223	Alln = 495
Demographic
Age, years, mean (SD)	60 (7)	59 (7)	60 (7)
Weight, kg, mean (SD)	78.5 (15.7)	90.4 (14.3)	83.9 (16.2)
Body mass index, kg/m^2, mean (SD)	29.9 (5.7)	29.2 (4.2)	29.6 (5.1)
Waist circumference, cm	94.1 (13.0)	102.5 (10.7)	97.9 (12.7)
Current smokers, n (%)	30 (11)	37 (17)	67 (14)
Clinical, mean (SD)
Duration of HA, years, median (IQR)	5 (0–11)	4 (0–11)	4 (0–11)
LVH, n (%)	51 (19)	30 (13)	81 (16)
Blood pressure, mmHg, mean (SD)
Systolic	153.2 (17.3)	154.0 (15.9)	153.6 (16.7)
Diastolic	89.8 (8.0)	92.2 (8.0)	90.9 (8.1)
Pulse pressure	63.5 (14.6)	61.8 (13.4)	62.7 (14.1)
ABI, mean (SD)	1.06 (0.12)	1.07 (0.12)	1.07 (0.12)
eGFR, mL/min/1,73 m^2, mean (SD)	77.4 (15.3)	83.5 (16.6)	80.2 (16.2)
Framingham CVD score, %, mean (SD)	14 (9)	24 (13)	19 (12)
SCORE, %, mean (SD)	5 (7)	10 (7)	7 (7)
Biochemical, mean (SD)
hs-CRP, mg/L	2.75 (2.26)	2.24 (1.92)	2.52 (2.13)
Total cholesterol, mmol/L	5.31 (0.92)	5.14 (0.85)	5.23 (0.90)
LDL-cholesterol, mmol/L	3.14 (0.86)	3.06 (0.75)	3.11 (0.81)
HDL-cholesterol, mmol/L	1.57 (0.38)	1.41 (0.42)	1.50 (0.41)
Triglycerides, mmol/L	1.35 (0.58)	1.47 (0.73)	1.40 (0.65)
Fasting glucose, mmol/L	5.59 (0.85)	5.90 (1.15)	5.73 (1.01)
2-h glucose, mmol/L	8.00 (2.44)	7.80 (2.55)	7.91 (2.49)
Current medication, n (%)
Vasodilatators	135 (50)	127 (57)	262 (53)
Beta-blockers	87 (32)	65 (29)	152 (31)
Diuretics	48 (18)	38 (17)	86 (17)
Statins	47 (17)	34 (15)	81 (16)

HA = hypertension; IQR = interquartile range; LVH = left ventricular hypertrophy; ABI = ankle-brachial index; eGFR = estimated glomerular filtration rate; CVD = cardiovascular disease; SCORE = Systematic Coronary Risk Evaluation; hs-CRP = high-sensitivity C-reactive protein; LDL = low-density lipoprotein; HDL = high-density lipoprotein; Vasodilatators = angiotensin-converting enzyme inhibitor, angiotensin receptor antagonist, or calcium antagonist.

Of the 495 study subjects, 123 (24.8% (95% CI 21.1–28.9)) had ABI <1.00, 81 (16.4% (95% CI 13.2–19.9)) had ECG-LVH, and 41 (8.3% (95% CI 6.0–11.1)) had eGFR <60 mL/min/1.73 m².

Framingham CVD risk >20%, SCORE risk >5%, and hs-CRP > 3 mg/L all showed poor accuracy to detect subclinical target organ damage as assessed by ABI <1.00, eGFR <60 mL/min/1.73 m², or presence of ECG-LVH, with a sensitivity of 0.31 to 0.57, specificity of 0.51 to 0.70 and a nonsignificant positive likelihood ratio varying from 0.90 to 1.19 (Table II).

Table II. The diagnostic characteristics of the test (Framingham CVD risk, SCORE risk and hs-CRP) with a cut-off value of high risk in predicting target organ damage. Parentheses show 95% confidence intervals.

	Cut-off point	Accuracy	Sensitivity	Specificity	LR+
ABI <1.0
Framingham	≥20%	57 (52–61)	0.31 (0.23–0.40)	0.65 (0.60–0.70)	0.90 (0.66–1.20)
SCORE	≥5%	51 (47–56)	0.50 (0.41–0.60)	0.51 (0.46–0.57)	1.04 (0.84–1.26)
hs-CRP	>3	60 (56–65)	0.32 (0.24–0.41)	0.70 (0.65–0.75)	1.05 (0.77–1.41)
eGFR <60
Framingham	≥20%	64 (60–68)	0.37 (0.22–0.53)	0.67 (0.62–0.71)	1.09 (0.69–1.59)
SCORE	≥5%	52 (47–56)	0.54 (0.37–0.69)	0.51 (0.47–0.56)	1.10 (0.79–0.43)
hs-CRP	>3	67 (63–71)	0.34 (0.20–0.51)	0.70 (0.65–0.74)	1.13 (0.70–1.70)
LVH
Framingham	≥20%	62 (58–66)	0.37 (0.27–0.49)	0.67 (0.62–0.72)	1.14 (0.82–1.53)
SCORE	≥5%	53 (49–58)	0.57 (0.45–0.68)	0.52 (0.47–0.57)	1.19 (0.95–1.46)
hs-CRP	>3	64 (59–68)	0.32 (0.22–0.43)	0.70 (0.65–0.74)	1.06 (0.74–1.48)

LR+ = positive likelihood ratio; ABI = ankle-brachial index; SCORE = Systematic Coronary Risk Evaluation; hs-CRP = high-sensitivity C-reactive protein; eGFR = estimated glomerular filtration rate; LVH = left ventricular hypertrophy; CVD = cardiovascular disease.

Receiver-operating characteristic (ROC) curves were calculated to define the optimal cut-off points for Framingham CVD risk, SCORE risk, and hs-CRP in predicting the presence of target organ damage (Table III). No statistically significant difference was found in area under the curve (AUC) between these three tests. The optimal cut-off points balancing the sensitivity and specificity of the test did not give markedly better, clinically meaningful diagnostic information.

Table III. The diagnostic characteristics of the test (Framingham CVD risk, SCORE risk and hs-CRP) with an optimal cutoff value by receiver-operating characteristic (ROC) curves in predicting target organ damage. Parentheses show 95% confidence intervals.

	AUC (95% CI)	Cut-off point	Accuracy	Sensitivity	Specificity	LR+
ABI <1.0
Framingham, %	0.49 (0.43–0.55)	17	55 (50–59)	0.44 (0.35–0.54)	0.58 (0.53–1.63)	1.06 (0.83–1.32)
SCORE, %	0.51 (0.45–0.57)	19	74 (70–78)	0.11 (0.06–0.18)	0.94 (0.92–0.96)	2.02 (1.06–3.78)
hs-CRP	0.52 (0.46–0.59)	6.4	75 (70–78)	0.14 (0.08–0.21)	0.95 (0.92–0.97)	2.57 (1.40–4.68)
P-value	0.30
eGFR <60
Framingham, %	0.54 (0.45–0.61)	14	48 (43–52)	0.66 (0.49–0.80)	0.46 (0.41–0.51)	1.22 (0.93–1.49)
SCORE, %	0.54 (0.46–0.63)	3	37 (33–41)	0.80 (0.65–0.91)	0.33 (0.29–0.37)	1.20 (0.97–1.37)
hs-CRP	0.59 (0.51–0.68)	1.5	45 (41–50)	0.78 (0.62–0.89)	0.42 (0.38–0.47)	1.35 (1.08–1.57)
P-value	0.68
LVH
Framingham, %	0.56 (0.50–0.62)	10	35 (30–39)	0.90 (0.81–0.96)	0.24 (0.20–0.28)	1.18 (1.06–1.28)
SCORE, %	0.58 (0.51–0.64)	5	53 (49–58)	0.57 (0.45–0.68)	0.52 (0.47–0.57)	1.19 (0.95–1.46)
hs-CRP	0.53 (0.47–0.59)	1.5	47 (42–51)	0.70 (0.59–0.80)	0.42 (0.37–0.47)	1.22 (1.02–1.42)
P-value	0.41

AUC = area under the curve; LR+= positive likelihood ratio; ABI = ankle-brachial index; SCORE = Systematic Coronary Risk Evaluation; hs-CRP = high-sensitivity C-reactive protein; eGFR = estimated glomerular filtration rate; LVH = left ventricular hypertrophy; CVD = cardiovascular disease.

The number of the study subjects classified as high-risk (test positive) or non-high-risk (test negative) subjects according to the risk estimation tests, and the number of patients with target organ damage within each test category are shown in Figure 2.

Figure 2. Number of the study subjects classified as high-risk (test positive) or non-high-risk (test negative) subjects according to the risk estimation tests, and the number of patients with target organ damage within each test category. Cut-off points were 5% in SCORE, 20% in Framingham CVD score, and >3 mg/L in hs-CRP. (SCORE = Systematic COronary Risk Evaluation system; CVD = cardiovascular disease; hs-CRP = high-sensitivity C-reactive protein; ABI = ankle-brachial index; eGFR = estimated glomerular filtration rate; LVH = left ventricular hypertrophy).

The SCORE system estimated that 252/495 (50.9%) patients have low CVD risk, i.e. <5% risk of dying from CVD over 10 years. In this group of hypertensive patients 24.2% (95% CI 19.1–30.0) (61/252) had ABI <1.00, 7.5% (95% CI 4.6–11.5) (19/252) had eGFR <60 mL/min/1.73 m², and 13.9% (95% CI 9.9–18.8) (35/252) had ECG-LVH.

According to the Framingham CVD risk prediction model, 326/495 (65.9%) of the study subjects have a 10-year estimate of CVD risk of <20%. Among them, 25.8% (95% CI 21.1–30.9) had ABI <1.00, 8.0% (95% CI 5.3–11.5) had eGFR <60 mL/min/1.73 m², and 15.3% (95% CI 11.6–19.7) had ECG-LVH.

High-sensitivity CRP <3 mg/L was measured in 345/495 (69.7%). Based on this criterion, 24.4% (95% CI 20.0–29.3) of the patients with low ABI, 7.8% (95% CI 5.2–11.1) with low eGFR, and 16.0% (95% CI 12.2–20.3) with ECG-LVH were classified as low-risk subjects.

The positive predictive values were at highest 25.8% (95% CI 19.1–33.6) for ABI <1.00, 9.2% (95% CI 5.2–15.1) for eGFR <60 mL/min/1.73 m², and 18.1% (95% CI 12.5–24.8) for ECG-LVH.

Discussion

We investigated the usefulness of commonly used cardiovascular risk stratification methods for identifying high-risk subjects in 495 hypertensive patients without CVD, renal disease, or previously known diabetes, i.e. in primary prevention setting. We observed that SCORE and Framingham risk estimation tools or hs-CRP value >3 mg/L cannot reliably identify hypertensive patients with subclinical markers of target organ damage.

We chose ABI value <1.0 to indicate subclinical peripheral arterial disease. A recent meta-analysis of 16 cohort studies, in which participants aged 47–78 years were derived from a general population, showed that in men with ABI 0.91–1.00, compared to reference ABI 1.11–1.20, the hazard ratios for total mortality, cardiovascular mortality, and major coronary events were 1.61, 1.68, and 1.43, respectively (17). The corresponding figures in women were 1.52, 1.84, and 1.53.

The predictive value of border-line ABI 0.9–1.1 on adverse CVD outcomes was also confirmed in two German follow-up studies conducted in unselected patients from primary care (18,19).

We used a simplified method to measure ankle systolic pressures from ADP only, and ATP was used if ADP pulsation was not reliably found. The lower ankle pressure was used for ABI calculation. Leg perfusion may be better addressed using the higher of the ATP or ADP pressures to calculate ABI (20). However, using the lower of the two ankle pressures has been shown to identify a higher number of patients with increased risk for future cardiovascular events (21). The method used in our study somewhat over-estimates the prevalence of low ABI compared to using the higher of the ankle pressures, but on the other hand under-estimates the prevalence if the lower of the two ankle pressures is used. Nevertheless, the simplified method we used for ABI measurement and the easy-to-remember cut-off value <1.0 for identifying high-risk subjects would aid clinical decision-making in the hectic office of the primary care physician.

Glomerular filtration rate (GFR) <60 mL/min/ 1.73 m² is selected as the cut-off value for definition of chronic kidney disease because it represents a reduction by more than half of the normal value of 125 mL/min/1.73 m² in young adults, and this level of GFR is associated with the onset of laboratory abnormalities characteristic of kidney failure, including increased prevalence of several CVD risk factors (22). The introduction of estimated glomerular filtration rate (eGFR) calculated using simple formulas has renewed interest in reliable measurement of renal excretory function also in primary care. We calculated eGFR using the MDRD formula (13) which is not biased by body-weight, unlike the Cockcroft-Gault formula (23). Most of our study subjects were overweight or obese; 417/495 (84.2%) had body mass index ≥25, and 207/495 (41.8%) had body mass index ≥30.

We used Sokolow-Lyon voltage and the Cornell product as the criteria for ECG-LVH as is recommended in the 2007 guidelines of the European Society of Hypertension and the European Society of Cardiology (4). These criteria for LVH are independent predictors of cardiovascular events (24). Even though the sensitivity of ECG-LVH is lower in obese than in lean subjects, the Sokolow-Lyon voltage criterion is a good predictor of mortality also in overweight subjects (25). Although ECG-LVH is not the gold standard for detecting LVH, the ECG remains a quickly and easily available clinical tool to most practitioners at a relatively small cost (26).

The addition of the inflammatory biomarker hs-CRP to traditional risk factors has been shown to reclassify up to 30% of women at intermediate risk according to Framingham risk score into clinically relevant high- or low-risk categories (27). For the time being, hs-CRP has no established role in the European guidelines on CVD prevention (28).

In our study, hs-CRP >3 mg/L predicted the presence of subclinical target organ damage with a sensitivity of only ≈30%, although the specificity of 70% raised the accuracy of the high hs-CRP to ≈60%.

The SCORE and Framingham risk prediction models are not designed to detect subclinical arteriosclerosis but clinical events or death. However, subclinical organ damage can be seen as an intermediate stage in the continuum of cardiovascular disease and should be sought carefully (4). The Framingham risk score is not validated in Finland, and ideally cardiovascular risk prediction should be based on a prospective population cohort study undertaken in the population to which the risk score is to be applied (29). However, most of the subjects in the Framingham Heart Study are Caucasians as are Finnish people. On the contrary, the biggest contributor to the high-risk SCORE charts is the Finnish population-based survey FINRISK with 37,296 24–64-year-old Finns (30), so the high-risk SCORE charts are very well validated for the Finnish population.

In conclusion, subclinical hypertensive target organ damage, as assessed by ABI <1.0, presence of ECG-LVH, or eGFR <60 mL/min/1.73 m², is common in the primary care setting. However, conventional multivariable risk prediction tools, such as SCORE or the Framingham general CVD risk prediction model, or the novel risk factor hs-CRP, showed poor accuracy to detect the presence of subclinical vascular, renal, or cardiac organ damage as assessed by these simple clinical measures. Assessment of ECG-LVH, ABI, and eGFR reclassifies a significant number of hypertensive patients to the high-risk category as compared to the SCORE and Framingham risk prediction tools only.

This is an important message, since primary care physicians are inclined to determine the cardiovascular risk of individuals solely with risk estimation tools and without searching for subclinical organ damage in their hectic offices. A striking discrepancy in risk estimation between SCORE and the guidelines of the European Society of Hypertension which takes into account target organ damage was recently observed also in a Belgian study (31).

Although low ABI, ECG-LVH, and reduced eGFR have been shown to predict future adverse cardiovascular events, we cannot determine the additional prognostic significance of these measures based on this cross-sectional study. Therefore, large prospective follow-up studies are needed to test and validate the precise incremental prognostic value of ABI, ECG-LVH, and eGFR over the SCORE and Framingham risk prediction models in the future. However, given the simplicity and suitability of these measurements to daily clinical practice, and their potential ability to recognize high-risk individuals who would be otherwise misclassified as having low cardiovascular risk based on conventional risk prediction tools alone, more systematic use of ABI, ECG-LVH, and eGFR measurements in the general hypertensive population seems justified.

Acknowledgements

This work was supported by the State Provincial Office of Western Finland, the Central Satakunta Health Federation of Municipalities, the Finnish Cultural Foundation, and the Ida Montin Foundation.

Declaration of interest: The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the paper.