Pulse pressure strongly predicts cardiovascular disease risk in patients with type 2 diabetes from the Swedish National Diabetes Register (NDR)

Diabetes & Metabolism 35 (2009) 439–446

Original article

Pulse pressure strongly predicts cardiovascular disease risk in patients with type 2 diabetes from the Swedish National Diabetes Register (NDR) P.M. Nilsson a,∗ , J. Cederholm b , K. Eeg-Olofsson c , B. Eliasson c , B. Zethelius d , S. Gudbjörnsdóttir c , for the Swedish National Diabetes Register (NDR) b

a Department of Clinical Sciences, Lund University, University Hospital, 205 02 Malmö, Sweden Family Medicine and Clinical Epidemiology Section, Department of Public Health and Caring Sciences, Uppsala University, Uppsala, Sweden c Department of Medicine, Sahlgrenska University Hospital, University of Gothenburg, Gothenburg, Sweden d Geriatrics Section, Department of Public Health and Caring Sciences, Uppsala University, Uppsala, Sweden

Received 16 December 2008; received in revised form 6 April 2009; accepted 27 April 2009 Available online 12 October 2009

Abstract Objectives. – To analyze pulse pressure (PP) as a risk predictor for coronary heart disease (CHD), stroke and cardiovascular disease (CVD; CHD and/or stroke) in type 2 diabetic patients. Methods. – A total of 11,128 female and male type 2 diabetic patients with known baseline PP values and no CVD, aged 50–74 years, were followed for a mean duration of 5.6 years (1998–2003). A subgroup of 5521 patients with known mean PP values (mean values at baseline and at the end of the study) was also included. Results. – Hazard ratios (HRs) with 95% CI for fatal/nonfatal CHD with baseline or mean PP ≥ 75 mmHg, compared to < 75 mmHg, were 1.23 (1.07–1.40; P = 0.003) and 1.32 (1.07–1.62; P = 0.009), respectively, after adjusting for mean blood pressure (MBP), age, gender, diabetes duration, HbA1c , body mass index (BMI), lipid-reducing drugs, microalbuminuria > 20 ␮g/min, antihypertensive drugs and hypoglycaemic treatment, using Cox regression analyses. Fully-adjusted respective HRs for stroke were 1.17 (0.98–1.39) and 1.21 (0.90–1.61) and, for CVD, 1.23 (1.10–1.37; P < 0.001) and 1.28 (1.07–1.52; P = 0.007). Fully-adjusted HRs for baseline PP increased per quartile and, CHD, stroke or CVD, were 1.09 (1.03–1.16; P = 0.004), 1.14 (1.05–1.23; P = 0.002) and 1.11 (1.05–1.17; P < 0.001), respectively. The data suggest that, if a mean PP ≥ 75 mmHg were to be avoided, then 15% and 17% of CHD and or CVD, cases, respectively, in such a cohort might be prevented after multivariable adjustments, with a further 10% of cases avoided if also adjusted for MBP and age. Increasing baseline MBP, age and microalbuminuria were independently and significantly associated (P < 0.001) with increasing baseline or mean PP. Conclusion. – Increased PP is a powerful independent risk predictor of CVD in type 2 diabetic patients, and lowering PP can lead to a marked reduction in risk. © 2009 Elsevier Masson SAS. All rights reserved. Keywords: Pulse pressure; Diabetes; Cardiovascular disease; Myocardial infarction; Epidemiology

Résumé La pression pulsée est un indicateur puissant du risque d’événements cardiovasculaires chez des patients diabétiques de type 2 issus du Registre national suédois du diabète. Objectifs. – Évaluer la valeur de la pression pulsée (PP) comme un facteur de risque d’événements coronaires (EC), d’accidents vasculaires cérébraux ischémiques (AVC) et d’événements cardiovasculaires (ECV : EC et/ou AVC) mortels et non mortels chez des diabétiques de type 2 (DT2). Méthodes. – Onze mille cent vingt-huit patients des deux sexes atteints de DT2, dont la PP basale (PPB) avait été déterminée (âge 50–74 ans, absence d’antécédents CV à l’inclusion) ont été suivis en moyenne 5,6 ans (1998–2003). Chez 5521 d’entre eux, la pression pulsée moyenne (PPM) (moyenne de la PP à l’inclusion et en fin d’étude) était enregistrée.

∗

Corresponding author. E-mail address: [email protected] (P.M. Nilsson).

1262-3636/$ – see front matter © 2009 Elsevier Masson SAS. All rights reserved. doi:10.1016/j.diabet.2009.04.010

440

P.M. Nilsson et al. / Diabetes & Metabolism 35 (2009) 439–446

Résultats. – En analyse multivariée par régression de Cox, le risque relatif (RR) (intervalle de confiance à 95 %) d’événements coronaires chez les patients dont la PPB ou la PPM était supérieure ou égale à 75 mmHg, comparés à ceux dont les valeurs étaient inférieure à 75 mmHg, était respectivement de 1,23 (1,07–1,40 ; P = 0,003) et 1,32 (1,07–1,62 ; P = 0,009), après ajustement pour la pression artérielle moyenne (PAM), l’âge, le sexe, la durée de diabète, l’HbA1c , l’IMC, la prise d’hypocholestérolémiants, d’antihypertenseurs, l’existence d’une micro-albuminurie supérieure à 20 ␮g/min et le traitement hypoglycémiant. Le RR multi-ajusté d’accident vasculaire cérébral était respectivement de 1,17 (0,98–1,39) et de 1,21 (0,90–1,61), et celui d’événement cardiovasculaire de 1,23 (1,10–1,37 ; P < 0,001) et 1,28 (1,07–1,52 ; P = 0,007). L’augmentation du RR multi-ajusté d’événements coronaires, d’accidents vasculaires cérébraux et d’événements cardiovasculaires totaux par quartile de la PPB était respectivement de 1,09 (1,03–1,16 ; P = 0,004), 1,14 (1,05–1,23 ; P = 0,002) et 1,11 (1,05–1,17 ; P < 0,001). S’il avait été possible d’obtenir à l’inclusion une PPB inférieure à 75 mmHg, après ajustement multivarié, 15 % des événements coronaires et 17 % des accidents vasculaires cérébraux de la cohorte auraient pu être évités (10 % des événements après ajustement supplémentaire pour la PAM et l’âge). L’augmentation de la PAM et de l’âge, ainsi que la micro-albuminurie basale étaient associées de manière indépendante (P < 0,001) à la PPB et à la PPM. Conclusion. – L’augmentation de la PP est un facteur de risque d’événements cardiovasculaires puissant et indépendant chez des diabétiques de type 2. Une réduction de la PP permettrait une réduction de ce risque. © 2009 Elsevier Masson SAS. Tous droits réservés. Mots clés : Pression pulsée ; Diabète de type 2 ; Registre ; Étude longitudinale ; Maladies cardiovasculaires ; Infarctus du myocarde ; Accidents vasculaires cérébraux ; Épidémiologie

1. Introduction

2.2. Patients

Patients with type 2 diabetes have a two- to four-fold higher risk of cardiovascular disease (CVD) than do nondiabetic, healthy subjects [1,2]. The importance of systolic blood pressure (BP) as a risk factor for cardiovascular events has been verified by the United Kingdom Prospective Diabetes Study (UKPDS), and strict BP control has proved effective in reducing the risk of stroke and diabetes-related mortality [3]. Such patients are also thought to have increased arterial rigidity [4–8], and pulse pressure may be regarded as an indirect measure of arterial stiffness in middle-aged and older subjects [4,9–11]. However, only a few studies have investigated the predictive value of pulse pressure as a risk factor for CVD in patients with type 2 diabetes [12,13]. The aim of this observational prospective study was to analyze the association between pulse pressure, coronary heart disease (CHD), stroke and CVD (CHD and/or stroke) in a large sample of type 2 diabetic patients from the Swedish diabetes population registered at primary-care centers and hospital diabetes clinics nationwide.

This observational prospective study involved 11,128 female and male type 2 diabetic patients from the NDR who were free of CVD at baseline and aged 50–74 years, with data available for all of the analyzed variables. The patients were followed prospectively from 1998 to 2003 for analysis of baseline pulse pressure as a risk factor for first-incident fatal or nonfatal CHD, stroke and CVD. A subgroup of 5521 patients was also included with data available for estimation of mean pulse pressure during the study period, based on the mean values of pulse pressure at baseline and at the end of the study. Pulse pressure at the end of the study was defined as the value during the year prior to an event, whether fatal or nonfatal, or the value during 2003. The definition of type 2 diabetes was treatment with diet only, treatment with oral hypoglycaemic agents only, or onset age of diabetes ≥ 40 years and treatment with insulin only or combined with oral agents. Only 1% had a diabetes-onset age < 30 years, and 3% had an onset age < 40 years.

2. Patients and methods 2.1. The Swedish National Diabetes Register (NDR) The Swedish National Diabetes Register (NDR) was initiated in 1996 as a tool for local quality assurance in diabetes care. Annual reporting to the NDR is carried out by trained physicians and nurses via the Internet and through databases of clinical records, comprising information collected during patient visits at hospital outpatients clinics and primary healthcare centers across Sweden, from approximately 200 units participating in the present study. All included patients gave their informed consent to register before participating in the study. The study was approved by the regional ethics committee at the University of Gothenburg. In addition, reports on trends in risk factor control according to the NDR have been previously published [14–18].

2.3. Examinations at baseline Clinical characteristics at baseline in 1997 and 1998 included type of hypoglycaemic treatment, age, diabetes duration, gender, systolic and diastolic BP, weight, height, smoking status, use of lipid-lowering drugs, cumulative microalbuminuria and use of antihypertensive drugs. Body mass index (BMI) was calculated as weight/height2 (kg/m2 ). The Swedish standard for recording BP, used in the NDR, takes the mean value of two supine readings (Korotkoff: 1–5) with a cuff of appropriate size after at least five minutes of rest. Pulse pressure was defined as systolic BP minus diastolic BP. Mean BP (MBP) was defined as two-thirds of the diastolic BP plus one-third of the systolic BP. A smoker was defined as a patient who smoked one or more cigarettes a day, or smoked tobacco using a pipe, or who had stopped smoking within the past 3 months. Laboratory analyses of HbA1c were carried out at local laboratories and were quality-assured across the country by regular calibration, using the high-performance liquid chromatography (HPLC) Mono-S method. In the present study, all HbA1c values

P.M. Nilsson et al. / Diabetes & Metabolism 35 (2009) 439–446

were converted to the Diabetes Control and Complications Trial (DCCT) standard values using the formula HbA1c (DCCT) = 0.923 × HbA1c (Mono-S) + 1.345; R2 = 0.998 [19]. Microalbuminuria was defined as cumulative, with urine albumin excretion > 20 ␮g/min in two out of three consecutive tests. 2.4. Follow-up and definition of endpoints All patients were followed from the baseline examination until the first incidence of a cardiovascular event or death, or until censor date 31 December 2003. The mean follow-up duration was 5.6 years. All patients were free of CHD, stroke and heart failure at baseline. Three major endpoints were used in this study: • first-incident fatal or nonfatal CHD; • first-incident fatal or nonfatal stroke; • first-incident fatal or nonfatal CVD (defined as CHD or stroke, whichever came first). A fatal CHD event was defined as fatal ischaemic heart disease (ICD-10 codes I20–I25; www.who.int/classifications/ icd/en/) or sudden cardiac death (ICD-10 codes R96.0–1). A nonfatal CHD event was defined as nonfatal myocardial infarction (ICD-10 code I21), unstable angina (ICD-10 code I20.0), percutaneous coronary intervention and/or coronary artery bypass grafting (CABG). In patients with multiple CHD events, only the first event was considered in our study. A stroke event was defined as fatal or nonfatal intracerebral haemorrhage, cerebral infarction or unspecified stroke (ICD-10 codes I61, I63, I64, I67.9). Heart failure was defined as ICD-10 code I50.

441

All endpoint events were retrieved by data linkage with the Swedish Cause of Death and Hospital Discharge Registers (National Board of Health and Welfare, Sweden), a reliable, validated alternative to revised hospital discharge and death certificates [20,21]. In total, 1175 first-incident fatal/nonfatal CHD events, 699 stroke events and 1728 CVD events occurred, based on 55,016 person/years, while 460 CHD events, 237 stroke events and 635 CVD events were seen in the subgroup of 5521 patients, based on 27,356 person/years. 2.5. Statistical methods Absolute risk was expressed as events per 1000 person/years. Cox regression analysis was used to estimate hazard ratios (HRs) with 95% confidence intervals (95% CI) for pulse pressure and CHD, stroke or CVD, adjusted for several covariables in three models (Tables 1 and 2). Patients were subdivided into two groups with pulse pressure either ≥ 75 mmHg or < 75 mmHg, as 75 mmHg was found to be the 75th percentile in the distribution of both baseline and mean pulse pressures. Pulse pressure was also applied as a continuous variable that increased in quartiles. A test of linear trend using the difference in the Wald x2 statistic of global fit from two models, with pulse pressure applied as either a continuous or a nominal variable, showed that the former model produced a better fit. The proportionalhazards assumption was confirmed for all covariates with the Kolmogorov-type supremum test, using resampling, and with a test of all time-dependent covariates simultaneously. Maximum likelihood estimation was used to evaluate interactions between all analyzed variables, and significant interactions were added as covariates:

Table 1 Hazard ratios (95% CI) for higher vs lower values of baseline pulse pressure (PP) in relation to first-incident cardiovascular events, using Cox regression analysis in 11,128 type 2 diabetic patients, aged 50–74 years, followed for 6 years.

Fatal/Nonfatal coronary heart disease PP ≥ 75 vs < 75 mmHg (Model 1b ) PP ≥ 75 vs < 75 mmHg (Model 2c ) PP ≥ 75 vs < 75 mmHg (Model 3d ) PP per quartile (Model 3d ) Fatal/Nonfatal stroke PP ≥ 75 vs < 75 mmHg (Model 1b ) PP ≥ 75 vs < 75 mmHg (Model 2c ) PP ≥ 75 vs < 75 mmHg (Model 3d ) PP per quartile (Model 3d ) Fatal/Nonfatal cardiovascular disease PP ≥ 75 vs < 75 mmHg (Model 1b ) PP ≥ 75 vs < 75 mmHg (Model 2c ) PP ≥ 75 vs < 75 mmHg (Model 3d ) PP per quartile (Model 3d )

Patients (n)

Events (n)

Absolute riska

Hazard ratio (95% CI)

P value

3255/7873

447/728

27.6/18.2

11,128

1175

20.9

1.51 (1.35–1.70) 1.48 (1.30–1.68) 1.23 (1.07–1.40) 1.09 (1.03–1.16)

< 0.001 < 0.001 0.003 0.004

3255/7873

289/410

17.7/10.2

11,128

699

12.3

1.75 (1.50–2.03) 1.51 (1.28–1.78) 1.17 (0.98–1.39) 1.14 (1.05–1.23)

< 0.001 < 0.001 0.08 0 002

3255/7873

681/1047

43.3/26.6

11,128

1728

31.4

1.63 (1.48–1.79) 1.51 (1.36–1.68) 1.23 (1.10–1.37) 1.11 (1.05–1.17)

< 0.001 < 0.001 < 0.001 < 0.001

NB: 75 mmHg corresponded to the 75th percentile in the distribution of PP, and the PP increase per quartile corresponded to an increase per 10 mmHg; CI: confidence interval. a Events per 1000 patient-years. b Unadjusted crude hazard ratios. c Adjusted for gender, diabetes duration, HbA , body mass index, smoking status, lipid-lowering drugs, microalbuminuria, antihypertensive drugs, type of 1c hypoglycaemic treatment and significant interactions. d Adjustment as in Model 2, and also for age and mean blood pressure.

442

P.M. Nilsson et al. / Diabetes & Metabolism 35 (2009) 439–446

Table 2 Hazard ratios (95% CI) for higher vs lower values of mean pulse pressure (PP) and first-incident cardiovascular events, using Cox regression analysis in 5521 type 2 diabetic patients, aged 50–74 years, followed for 6 years.

Fatal/Nonfatal coronary heart disease PP ≥ 75 vs < 75 mmHg (Model 1b ) PP ≥ 75 vs < 75 mmHg (Model 2c ) PP ≥ 75 vs < 75 mmHg (Model 3d ) Fatal/Nonfatal stroke PP ≥ 75 vs < 75 mmHg (Model 1b ) PP ≥ 75 vs < 75 mmHg (Model 2c ) PP ≥ 75 vs < 75 mmHg, (Model 3d ) Fatal/Nonfatal cardiovascular disease PP ≥ 75 vs < 75 mmHg (Model 1b ) PP ≥ 75 vs < 75 mmHg, (Model 2c ) PP ≥ 75 vs < 75 mmHg, (Model 3d )

Patients (n)

Events (n)

Absolute riska

Hazard ratio (95% CI)

P value

1501/4020

169/291

22.8/14.3

1.59 (1.31–1.92) 1.58 (1.29–1.93) 1.32 (1.07–1.62)

< 0.001 < 0.001 0.009

1505/4016

93/144

12.4/7.0

1.76 (1.36–2.29) 1.48 (1.12–1.95) 1.21 (0.90–1.61)

< 0.001 0.006 0.2

1515/4006

239/396

32.6/19.8

1.64 (1.40–1.93) 1.53 (1.29–1.82) 1.28 (1.07–1.52)

< 0.001 < 0.001 0.007

NB: 75 mmHg corresponded to the 75th percentile in the distribution of PP. CI: confidence interval. a Events per 1000 patient-years. b Unadjusted crude hazard ratios. c Adjusted for gender, diabetes duration, HbA , body mass index, smoking status, lipid-lowering drugs, microalbuminuria, antihypertensive drugs, type of 1c hypoglycaemic treatment and significant interactions. d Adjustment as in Model 2, and also for age and mean blood pressure.

• antihypertensives × microalbuminuria; • age × smoking; • MBP × antihypertensives. The Hosmer-Lemeshow test demonstrated a nonsignificant x2 statistic, indicating excellent goodness-of-fit of the models. The homogeneity of the effect of pulse pressure on CHD, stroke and CVD outcomes across the subgroups of each covariable was tested by adding interaction terms to the relevant Cox models. In addition, the percent partial population-attributable risk with 95% CI was calculated [22], estimating the percent cases of fatal/nonfatal CHD, stroke and CVD in the cohort that might have been prevented if a pulse pressure ≥ 75 mmHg had been avoided. The association between age vs pulse pressure and baseline MBP vs pulse pressure were analyzed by univariate regression. Multiple regression was used to analyze baseline MBP, age and microalbuminuria as independent variables vs pulse pressure as a dependent variable, with MBP and age increased per standard deviation (SD) to allow for comparisons of the power of the estimated regression coefficients (± standard error [SE]) of the continuous variables, while microalbuminuria was included as a dichotomous variable. All statistical analyses were performed using SAS version 9.1.3 software (SAS Institute, Cary, NC, USA). A P value < 0.05 was considered statistically significant. 3. Results Table 3 shows the baseline characteristics of the study sample of 11,128 men and women with type 2 diabetes, aged 50–74 years and free of CVD at baseline, divided into subgroups according to baseline pulse pressure ≥ 75 mmHg or < 75 mmHg, as 75 mmHg corresponded to the 75th percentile in the

distribution of baseline pulse pressure. Those with a baseline pulse pressure ≥ 75 mmHg were older, had higher MBP and more microalbuminuria, and 60% were treated with antihypertensive drugs. Mean values of baseline pulse pressure in the ≥ 75 mmHg and < 75 mmHg subgroups were 85.1 mmHg and 58.4 mmHg, respectively, a difference of 27 mmHg. Table 3 also shows the corresponding baseline characteristics in the subgroup of 5521 patients with mean pulse pressure ≥ 75 mmHg or < 75 mmHg, as 75 mmHg corresponded to the 75th percentile in the distribution of mean PP. Mean values of pulse pressure in these two groups were 83.7 mmHg and 60.1 mmHg, respectively, a difference of 24 mmHg. 3.1. Pulse pressure and cardiovascular events Cox regression analysis was performed to determine HRs (95% CI) for pulse pressure ≥ 75 mmHg compared with < 75 mmHg, and first-incident fatal/nonfatal CHD, stroke and CVD, during the mean 5.6 years of follow-up. With baseline (Table 1) and mean (Table 2) pulse pressures, crude HRs for CHD were 1.51 (1.35–1.70) and 1.59 (1.31–1.92), respectively (P < 0.001, Model 1). The corresponding HRs for CHD were 1.48 (1.30–1.68) and 1.58 (1.29–1.93), respectively (P < 0.001), after adjusting for gender, diabetes duration, HbA1c , BMI, smoking status, lipidlowering drugs, cumulative microalbuminuria, antihypertensive drugs and type of hypoglycaemic treatment (Model 2). HRs for CHD were further attenuated after adjusting for age and MBP (Model 3): 1.23 (1.07–1.40; P = 0.003) and 1.32 (1.07–1.62; P = 0.009), respectively. Crude HRs for stroke with baseline pulse pressure ≥ 75 mmHg vs < 75 mmHg and mean pulse pressure ≥ 75 mmHg vs < 75 mmHg were 1.75 (1.50–2.03) and 1.76 (1.36–2.29) (P < 0001; Model 1), respectively, but were nonsignificant after

P.M. Nilsson et al. / Diabetes & Metabolism 35 (2009) 439–446

443

Table 3 Clinical characteristics at baseline of 11,128 patients with type 2 diabetes, aged 50–74 years, stratified by pulse pressure (PP), and a subgroup of 5521 patients with higher or lower mean PP, during the study period. All patients

Patients (n) Pulse pressure (mmHg) Mean BP (mmHg) Systolic BP (mmHg) Diastolic BP (mmHg) Age (years) Diabetes duration (years) HbA1c (%) BMI (kg/m2 ) Male gender (%) Smokers (%) Antihypertensives (%) Hypertension (%) Lipid-lowering drugs (%) Microalbuminuria (%)

Subgroup

Baseline PP ≥ 75 mmHg

Baseline PP < 75 mmHg

Mean PP ≥ 75 mmHg

Mean PP < 75 mmHg

3255 85.1 (9.9) 109.9 (9.9) 166.7 (13.3) 81.6 (9.5) 65.5 (6.1) 9.1 (7.4) 7.6 (1.3) 29.1 (4.7) 47.6 14.2 63.5 100 13.2 27.1

7873 58.4 (9.6)*** 101.0 (9.0)*** 139.9 (13.0)*** 81.6 (8.3) 61.3 (6.8)*** 7.5 (6.5)*** 7.7 (1.3) 28.9 (4.9)* 58.7*** 16.1*** 43.2*** 82.7*** 12.7 18.7***

1501 83.7 (8.6) 108.7 (9.9) 163.8 (14.9) 81.1 (9.3) 65.2 (6.2) 9.4 (7.3) 7.7 (1.2) 28.9 (4.7) 47.3 13.7 62.2 99.9 13.9 26.7

4020 60.1 (8.7)*** 101.2 (9.5)*** 140.9 (14.5)*** 81.2 (8.5) 61.2 (6.8)*** 7.9 (6.6)*** 7.7 (1.3) 28.7 (4.8) 60.1*** 15.2 45.2*** 81.9*** 14.8 20.4***

Data are presented as means with standard deviation (SD) or frequencies (%). BP: blood pressure; BMI: body mass index. NB: 75 mmHg corresponded to the 75th percentile in the distribution of PP. * P < 0.05 (patients with higher vs lower PP). *** P < 0.001.

adjustment according to Model 3: 1.17 (0.98–1.39) and 1.21 (0.90–1.61), respectively. Crude HRs for CVD with baseline pulse pressure ≥ 75 mmHg vs < 75 mmHg and mean pulse pressure ≥ 75 mmHg vs < 75 mmHg were 1.63 (1.48–1.79) and 1.64 (1.40–1.93) (P < 0.001; Model 1), and 1.23 (1.10–1.37; P < 0.001) and 1.28 (1.07–1.52; P = 0.007) after adjustment according to Model 3. HRs for baseline pulse pressure increased per quartile in relation to risk of CHD, stroke and CVD, adjusted according to Model 3, and were 1.09 (1.03–1.16; P = 0.004), 1.14 (1.05–1.23; P = 0.002) and 1.11 (1.05–1.17; P < 0.001), respectively. The effect of pulse pressure on CHD, stroke and CVD was consistent across subgroups, as defined by all covariables in the study (P value for heterogeneity > 0.05 in all comparisons due to no interaction with the subgroup covariables).

8.0 mmHg and 6.0 mmHg, respectively, per 10 mmHg increase in baseline MBP (P < 0.001), while also increasing by 7.7 mmHg and 6.8 mmHg, respectively, per 10-year increase in age (P < 0.001). Multiple regression analyses, all including baseline MBP (per SD), age (per SD) and microalbuminuria as independent variables vs either baseline or mean pulse pressure as dependent variables, and adjusted for gender, revealed that MBP was independently associated with both baseline and mean pulse pressures – regression coefficient ± SE: 7.5 ± 0.1 and 5.6 ± 0.2, respectively (P < 0.001), and that age was somewhat less strongly associated with baseline and mean pulse pressures – 4.2 ± 0.1 and 3.8 ± 0.2, respectively (P < 0.001). In addition, microalbuminuria was also independently associated with baseline and mean pulse pressures: 2.0 ± 0.2 and 1.4 ± 0.4, respectively (P < 0.001).

3.2. Population-attributable risk proportions 4. Discussion The percent partial population-attributable risk with mean pulse pressure ≥ 75 mmHg in the subgroup of 5521 patients was 15% (95% CI: 9–21%) for fatal/nonfatal CHD, 17% (8–26%) for fatal/nonfatal stroke and 17% (11–22%) for fatal/nonfatal CVD, after adjustment according to Model 2. In contrast, after adjustment according to Model 3, the percent partial populationattributable risk with mean pulse pressure ≥ 75 mmHg was 10% (3–17%) for CHD, 8% (−3–18%) for stroke and 10% (3–16%) for CVD. 3.3. Pulse pressure and baseline variables Univariate regression analyses showed that baseline pulse pressure in all 11,128 patients, and mean pulse pressure during the study in the subgroup of 5521 patients increased by

The main finding of this large-scale, observational, prospective study of type 2 diabetic patients was that elevated pulse pressure is a strong risk predictor of fatal/nonfatal CHD, stroke and CVD, independent of other cardiovascular risk factors and clinical characteristics. Furthermore, if elevated pulse pressure could be avoided in such a cohort, then 15–17% of all cases of CVD might be prevented independent of other risk factors, and 10% might be prevented after also adjusting for age and MBP. Blood pressure comprises two components: a stable component (MBP); and a pulsatile component (pulse pressure). The MBP is the pressure that distributes a steady flow of blood and oxygen to the tissues and organs, and reflects the total vascular resistance and the wall-to-lumen ratio in resistance vessels

444

P.M. Nilsson et al. / Diabetes & Metabolism 35 (2009) 439–446

(mainly small arteries). Pulse pressure is the consequence of intermittent ventricular ejection from the heart, and of the exchange capacities of the central (aorta) and larger peripheral arteries. In younger subjects, an increased pulse pressure generally reflects an increased stroke volume ejected from the left ventricle, which decreases with advancing age. In middle-aged and older patients (age 50 years or more), an elevated brachial pulse pressure more closely reflects arterial rigidity due to changes in the structure and functioning of the larger peripheral and central arteries. The increase in pulse pressure observed in the presence of arterial stiffness is dependent on an increase in systolic BP that could promote cardiac hypertrophy and increase the risk of stroke. It is also influenced by a concomitant decrease in diastolic BP, thereby compromising coronary perfusion that, in turn, might increase the risk of myocardial ischaemia [4,9,10]. Previously, several studies have shown that elevated brachial pulse pressure is a significant predictor of left ventricular hypertrophy, CHD, stroke, heart failure, and cardiovascular and total mortality in hypertensive patients as well as in the general population [10,23–28]. Increased peripheral arterial rigidity, estimated directly by increased pulse-wave velocity, has also been associated with left ventricular enlargement [29], and an increased risk of cardiovascular and total mortality, in patients with hypertension [9,10,30,31]. Arterial stiffness is increased in type 2 diabetic patients [4–8], and stiffer arteries and the consequently steeper increases in pulse pressure are seen with increasing age in such patients [12]. Furthermore, according to two earlier studies of type 2 diabetic patients [12,13], pulse pressure and peripheral arterial rigidity, estimated directly by increased pulse-wave velocity [32], were predictive of future CHD and CVD mortality. Schram et al. [12] examined 208 type 2 diabetic patients, aged 50–74 years, during 8.6 years of follow-up, and found HRs of 1.27 (95% CI 1.00–1.61) for CVD mortality per 10 mmHg increase in pulse pressure on Cox regression analysis, after adjusting for gender, age and MBP. Further adjustment for other risk factors gave similar HRs. Cockcroft et al. [13] examined 2911 type 2 diabetic patients (mean age 66 years) during 4 years of follow-up, and found a significant odds-ratio of 1.69 (P = 0.002) for fatal/nonfatal CHD per 10 mmHg increase in pulse pressure on logistic-regression analysis, after adjusting for gender, age, smoking status and total-to-HDL cholesterol ratio. The present observational study has demonstrated, to our knowledge for the first time in a large sample of more than 11,000 patients with type 2 diabetes, a strong, statistically significant and independently increased risk of CVD with elevated pulse pressure. Adjusted HRs were 1.5 (P < 0.001) for fatal/nonfatal CHD, stroke and CVD with baseline or mean pulse pressure ≥ 75 mmHg compared with < 75 mmHg (a difference of 24–27 mmHg), after adjustment for clinical characteristics and several important cardiovascular risk factors, including cumulative microalbuminuria as a marker of diabetic nephropathy as well as a strong risk factor per se for CHD (Model 2). Furthermore, as age and MBP are important confounding variables of outcome, and also have a relatively strong association with pulse pressure, it should be noted that the HRs were

somewhat attenuated, but were nevertheless in the range of 1.2–1.3 for CHD and CVD risk (P = 0.009 to < 0.001) with elevated baseline or mean pulse pressure, after also adjusting for age and MBP (Model 3). The significance of the association between baseline pulse pressure increases per quartile and CVD was weaker in the study by Schram et al. [12] than in the present study, probably because of the lower absolute risk of 7.2 compared with 31.4 events per 1000 person/years. The increasing baseline and mean pulse pressures were related to baseline MBP (8.0 mmHg and 6.0 mmHg, respectively, per 10 mmHg increase in MBP) and age (7.7 mmHg and 6.8 mmHg, respectively, per 10-year increase in age) on univariate analyses. These associations were significant (P < 0.001), independent of each other on multivariate analyses. Similar findings have also been reported in type 2 diabetic patients by Schram et al. [12], which suggests that age and MBP may interact to bring about accelerated arterial ageing, as reflected by elevated pulse pressures in middle-aged and older diabetic patients. Static wall stress increases with higher MBP, resulting in tissue fatigue characterized by alterations in collagen and elastin content that could also lead to increased arterial stiffness [33,34]. Cumulative microalbuminuria per se is a strong risk factor for CHD. Indeed, an association between increased pulse pressure and microalbuminuria was reported in a cross-sectional survey of nondiabetic hypertensive and atherosclerotic men [35], suggesting that increased pulse pressure is a predictor of microalbuminuria, although the opposite – that microalbuminuria is a predictor of increased pulse pressure – cannot be excluded. We found baseline cumulative microalbuminuria to be significantly related to pulse pressure, independent of both age and baseline MBP on multivariate analyses. However, as this association included mean and baseline pulse pressures, this analysis could not determine with certainty which of them – cumulative microalbuminuria or pulse pressure – was the predictor. Thus, a markedly increased risk for CVD in the presence of elevated pulse pressure was found in the present study, independent of the effects of age, MBP and cumulative microalbuminuria, among other covariables on these outcomes, suggesting that pulse pressure needs to be reduced in type 2 diabetic patients. Lifestyle changes, such as physical exercise [36], reduced salt intake [37] and increased intake of omega-3 fatty acids [38] can reduce arterial stiffness. However, more evidence is needed to establish the effect of antihypertensive drugs on pulse pressure. Nevertheless, angiotensin-converting enzyme (ACE) inhibitors are purported to have beneficial effects on large artery wall function and morphology, thereby reducing arterial stiffness [39], as has also been shown by aldosterone blockade using spironolactone [40]. The data on hypoglycaemic treatment, diabetes duration, HbA1c , BMI, BP antihypertensive, both antihypertensive and lipid-lowering drugs were found to be reliable in the present study, although smoking data may be somewhat biased due to underreporting by either the patients or examiners. Plasma creatinine was not available in this study, but the use of cumulative microalbuminuria should be a reasonable reflection of the

P.M. Nilsson et al. / Diabetes & Metabolism 35 (2009) 439–446

presence of diabetic nephropathy. Reduced creatinine clearance has been associated with increased vascular stiffness, indicating that kidney alterations may predict vascular rigidity [41]; other established, long-term, predictors of elevated pulse pressure in healthy subjects are fasting glucose, smoking status, BMI and triglycerides [11]. However, creatinine clearance was not a necessary covariate in our study, as the aim was to analyze the effect of pulse pressure on cardiovascular outcomes irrespective of causative mechanisms for increased pulse pressure. One limitation of the present study was that blood lipid values were not recorded at baseline in the NDR. However, the evidence from previous studies supports the use of lipid-lowering drugs rather than blood lipids as a marker of the presence of hyperlipidaemia. When lipid-lowering drugs were used instead of blood lipids among other risk-factor predictors in a multivariate analysis to elaborate a model for risk prediction of CVD, it was possible to demonstrate an excellent correlation between the predicted risk and observed risk during the follow-up [17]. In addition, two adjusted models (with and without lipid-lowering drugs as covariables) showed the same relative changes in HRs for obesity and cardiovascular events as found in two adjusted models (with and without blood lipids as covariables) in a previous large study [18]. CVD events retrieved from the National Cause of Death and Hospital Discharge Registers were also reliable, according to previous validations of reporting to these registers [19,20]. The age of the included patients was limited to 50–74 years to avoid younger patients, in whom elevated pulse pressure most likely reflects increased stroke volume, and to avoid the risk of lessprecise endpoint diagnoses in older patients. Also, the applied definition of type 2 diabetes should have excluded most of the younger patients who might have had latent autoimmune diabetes in adults (LADA). One strength of the present study was the large numbers of patients, person-years and CVD events, and the high absolute risk of 31.4 CVD events per 1000 person-years during the mean 5.6 years of follow-up. Another strength was the inclusion of a subgroup of patients for mean pulse pressure throughout the study period, and the fact that the effect of pulse pressure on the outcome was consistent across subgroups, as defined by the covariates in this observational study. Furthermore, patients were collected from the general diabetes population at primary-care centers and hospital diabetes clinics throughout Sweden, with no exclusions because of the presence or absence of risk factors or comorbidities, as is otherwise often the case in randomized controlled trials; the limitations as a result of such strict inclusion and exclusion criteria can limit the applicability of any study findings to the general patient population. In conclusion, increased pulse pressure is a powerful independent predictor of incident CVD in patients with type 2 diabetes, and a marked risk reduction appears to be possible if pulse pressure is lowered, underscoring the need for therapeutic measures to decrease pulse pressure in such patients. In addition to older age, higher baseline MBP and cumulative microalbuminuria were also independently related to higher baseline and mean pulse pressures.

445

Conflicts of interest The authors have no conflicts of interest to declare. Acknowledgements We would like to thank the regional NDR coordinators, and all participating nurses, physicians and other staff who have contributed to the NDR. Most of all, we wish to thank the patients who support the NDR, both individually and collectively, through their patient organization, the Swedish Diabetes Federation. The Swedish Association of Local Authorities and Regions and the Swedish Society of Diabetology jointly fund the NDR. References [1] Haffner S, Lehto S, Ronnemaa T, Pyorala K, Laakso M. Mortality from coronary heart disease in subjects with type 2 diabetes and in nondiabetic subjects with and without prior myocardial infarction. N Engl J Med 1998;339:229–34. [2] Stamler J, Vaccaro O, Neaton JD, Wentworth D. Diabetes, other risk factors, and 12 year cardiovascular mortality for men screened in the Multiple Risk Factor Intervention Trial (MRFIT). Diabetes Care 1993;16:434–44. [3] United Kingdom Prospective Diabetes Study (UKPDS) Group. Tight blood pressure control and risk of macrovascular and microvascular complications in type II diabetes: UKPDS 38. BMJ 1998;317:703–12. [4] Stehouwer CDA, Henry RMA, Ferreira I. Arterial stiffness in diabetes and the metabolic syndrome: a pathway to cardiovascular disease. Diabetologia 2008;51:527–39. [5] Aoun S, Blacher J, Safar ME, Mourad JJ. Diabetes mellitus and renal failure: effects on large artery stiffness. J Hum Hypertens 2001;15:693–700. [6] Ravikumar R, Deepa R, Shanthirani C, Mohan V. Comparison of carotid intima-media thickness, arterial stiffness,and brachial artery flow mediated dilatation in diabetic and nondiabetic subjects (The Chennai Urban Population Study [CUPS-9]). Am J Cardiol 2002;90:702–7. [7] Schram MT, Henry RM, van Dijk RA, Kostense PJ, Dekker JM, Nijpels G, et al. Increased central artery stiffness in impaired glucose metabolism and type 2 diabetes: the Hoorn Study. Hypertension 2004;43:176–81. [8] Bella JN, Devereux RB, Roman MJ, Palmieri V, Liu JE, Paranicas M, et al. Separate and joint effects of systemic hypertension and diabetes mellitus on left ventricular structure and function in American Indians (the Strong Heart Study). Am J Cardiol 2001;87:1260–5. [9] Safar ME, London GM, for the Clinical Committee of Arterial Structure and Function, on behalf of the Working Group on Vascular Structure Function of the European Society of Hypertension. Therapeutic studies and arterial stiffness in hypertension: recommendations of the European Society of Hypertension. J Hypertens 2000;18:1527–35. [10] Schiffrin EL. Vascular stiffness and arterial compliance. Implications for Systolic Blood Pressure. Am J Hypertens 2004;17:39S–48S. [11] Mokhtari A, Bellinetto-Ford L, Melander O, Nilsson PM. Determinants of increasing pulse pressure during 23 years’ follow-up as a marker of arterial stiffness and vascular ageing. Blood Press 2008;17:291–7. [12] Schram MT, Kostense PJ, Van Dijk RA, Dekker JM, Nijpels G, Bouter LM, et al. Diabetes, pulse pressure and cardiovascular mortality: the Hoorn Study. J Hypertens 2002;20:1743–51. [13] Cockcroft JR, Wilkinson IB, Evans M, McEwan P, Peters JR, Davies S, et al. Pulse pressure predicts cardiovascular risk in patients with type 2 diabetes mellitus. Am J Hypertens 2005;18:1463–7. [14] Gudbjörnsdottir S, Cederholm J, Nilsson PM, Eliasson B. The National Diabetes Register in Sweden: An implementation of the St. Vincent Declaration for Quality Improvement in Diabetes Care. Diabetes Care 2003;26:1270–6. [15] Nilsson PM, Gudbjörnsdottir S, Eliasson B, Cederholm J, for the Steering Committee of the National Diabetes Register, Sweden. Hypertension in

446

[16]

[17]

[18]

[19]

[20]

[21]

[22]

[23]

[24]

[25]

[26]

P.M. Nilsson et al. / Diabetes & Metabolism 35 (2009) 439–446 diabetes – trends in control and relation to macrovascular morbidity in repeated national surveys from Sweden. J Hum Hypertens 2003;17:37–44. Eliasson B, Cederholm J, Nilsson P, Gudbjörnsdottir S, for the steering committee of the Swedish National Diabetes Register. The gap between guidelines and reality: type 2 diabetes in a national diabetes register 1996–2003. Diabet Med 2005;22:1420–6. Eeg-Olofsson K, Cederholm J, Nilsson PM, Zethelius B, Gudbjörnsdóttir S, Eliasson B, et al. Overweight, obesity and cardiovascular diseases and mortality in type 2 diabetes: an observational study in 13 087 patients. Diabetologia 2009;52:65–73. Cederholm J, Eeg-Olofsson K, Eliasson B, Zethelius B, Nilsson PM, Gudbjörnsdottir S, on behalf of the Swedish National Diabetes Register. Risk prediction of cardiovascular disease in type 2 diabetes: a risk equation from the Swedish National Diabetes Register (NDR). Diabetes Care 2008;31:2038–43. Hoelzel W, Weykamp C, Jeppsson JO, Miedema K, Barr JR, Goodall I, et al. IFCC reference system for measurement of hemoglobin A1c in human blood and the national standardization schemes in the United States, Japan, and Sweden: a method-comparison study. Clin Chem 2004;1: 166–74. Merlo J, Lindblad U, Pessah-Rasmussen H, Hedblad B, Rastam J, Isacsson SO. Comparison of different procedures to identify probable cases of myocardial infarction and stroke in two Swedish prospective cohort studies using local and national routine registers. Eur J Epidemiol 2000;16:235–43. Tunstall-Pedoe H, Kuulasmaa K, Amouyel P, Arveiler D, Rajakangas AM, Pajak A, Myocardial infarction and coronary deaths in the World Health Organization MONICA project. Registration procedures, events rates, and case-fatality rates in 38 populations from 21 countries in four continents. Circulation 1994;90:583–612. Spiegelman D, Hertzmark E, Wand HC. Point and interval estimates of partial population-attributable risks in cohort studies: examples and software. Cancer Causes Control 2007;18:571–9. Darne B, Girerd X, Safar M, Cambien F, Guize L. Pulsatile versus steady component of blood pressure: a cross-sectional analysis and a prospective analysis on cardiovascular mortality. Hypertension 1989;13: 392–400. Vaccarino V, Holford TR, Krumholz HM. Pulse pressure and risk for myocardial infarction and heart failure in the elderly. J Am Coll Cardiol 2000;36:130–8. Madhavan S, Ooi WL, Cohen H, Alderman MH. Relation of pulse pressure and blood pressure reduction to the incidence of myocardial infarction. Hypertension 1994;23:395–401. Benetos A, Safar M, Rudnichi A, Smulyan H, Richard JL, Ducimetiere P. Pulse pressure: a predictor of long-term cardiovascular mortality in a French male population. Hypertension 1997;30:1410–5.

[27] Alderman MH, Cohen H, Madhavan S. Distribution and determinants of cardiovascular events during 20 years of successful antihypertensive treatment. J Hypertens 1998;16:761–9. [28] Mitchell GF, Moye LM, Braunwald E, Rouleau J-L, Bernstein V, Geltman EM, et al. Sphygmomanometrically determined pulse pressure is a powerful independent predictor of recurrent events after myocardial infarction in patients with impaired left ventricular function. Circulation 1997;96:4254–60. [29] Roman MJ, Ganau A, Saba PS, Pini R, Pickering TG, Devereux RB. Impact of arterial stiffening on left ventricular structure. Hypertension 2000;36:489–94. [30] Boutouyrie P, Tropeano AI, Asmar R, Gautier I, Benetos A, Lacolley P. Aortic stiffness is an independent predictor of primary coronary events in hypertensive patients: a longitudinal study. Hypertension 2002;39:10–5. [31] Laurent S, Boutouyrie P, Asmar R, Gautier I, Laloux B, Ducimetiere P, et al. Aortic stiffness is an independent predictor of all-cause and cardiovascular mortality in hypertensive patients. Hypertension 2001;37:1236–41. [32] Cruickshank K, Riste L, Anderson SG, Wright JS, Dunn G, Gosling RG. Aortic pulse-wave velocity and its relationship to mortality in diabetes and glucose intolerance: an integrated index of vascular function? Circulation 2002;106:2085–90. [33] Nichols WW, O’Rourke MF, editors. McDonald’s blood flow in arteries. Theoretical, experimental and clinical principles. 4th ed. London: Arnold; 1998. [34] O’Rourke M. Arterial stiffness, systolic blood pressure, and logical treatment of arterial hypertension. Hypertension 1990;15:339–47. [35] Pedrinelli R, Dell’Omo G, Penno G, Bandinelli S, Bertini A, Di Bello V, et al. Microalbuminuria and pulse pressure in hypertensive and atherosclerotic men. Hypertension 2000;35:48–54. [36] Cameron JD, Dart AM. Exercise training increases total systemic arterial compliance in humans. Am J Physiol 1994;266:H693–701. [37] Safar M, Laurent S, Safavian A, Pannier B, Asmar R. Sodium and large arteries in hypertension. Effects of indapamide. Am J Med 1988;84:15–9. [38] Nestel PJ, Pomeroy SE, Sasahara T, Yamashita T, Liang YL, Dart AM, et al. Arterial compliance in obese subjects is improved with dietary plant n-3 fatty acid from flaxseed oil despite increased LDL oxidizability. Arterioscler Thromb Vasc Biol 1997;17:1163–70. [39] Dart AM, Kingwell BA. Pulse pressure - a review of mechanisms and clinical relevance. J Am Coll Cardiol 2001;37:975–84. [40] Davis J, Gavin A, Band M, Morris A, Struthers A. Spironolactone reduces brachial pulse wave velocity and PIIINP levels in hypertensive diabetic patients. Br J Clin Pharmacol 2005;59:520–3. [41] Mourad J-J, Pannier B, Blacher J, Rudnichi A, Benetos A, London GM, et al. Creatinine clearance, pulse wave velocity, carotid compliance and essential hypertension. Kidney Int 2001;59:1834–41.