Joc 60078

7/27/2019 Joc 60078

1/8

ORIGINAL CONTRIBUTION

Diurnal Blood Pressure Pattern andRisk of Congestive Heart FailureErik Ingelsson, MD, PhD

Kristina Bjorklund-Bodegard, MD, PhD

Lars Lind, MD, PhD

Johan Arnlov, MD, PhD

Johan Sundstrom, MD, PhD

CONGESTIVE HEART FAILURE

(CHF) isone ofthe mostcom-

mon, costly, disabling, anddeadly diseases.1 It consti-

tutes a huge burden on health servicesand accounts for 1% to 2% of the totalhealth carecosts in industrialized coun-tries.2 Once diagnosed as having CHF,patients have a 1 in 3 chance of dyingwithin1 year and a 2 in3 chanceof dy-ing within 5 years.1 The mortality asso-ciated with CHF exceeds that of mostcancers, although recent reports sug-gest an improving prognosis.3

The predominant causes of CHF are

hypertension and coronary heart dis-ease,andhighblood pressure(BP)issug-gestedto bethemostimportantrisk fac-tor for CHF at a population level.4

Ambulatory BP monitoring providesinformation that is not obtained fromconventional office-based BP measure-ment, such as mean BP over a 24-hourperiod and night-daypatterns.Previousstudieshave establishedthat 24-hour BPmeasurements are powerful predictorsof cardiovascular morbidity andmortal-ity independent of office-measured BP

andotherestablished cardiovascular riskfactors.5-8 However, no previousstudieshave examined 24-hour ambulatory BPas a predictor of incident CHF in per-sons free of CHF at baseline.

Thus, the primary aim of the cur-rent study was to analyze 24-hour am-bulatory BP characteristics as predic-tors of CHF incidence in a community-

based sample of elderly men, adjustingfor antihypertensive treatment and tra-ditional risk factors for CHF. The sec-ondary aim was to investigate whether

24-hour ambulatory BP patterns con-fer any value regarding the risk of fu-ture CHF beyond that conveyed by con-ventional office-measured BP.

METHODS

Study Sample

This study is based on the Uppsala Lon-gitudinal Study of Adult Men cohort

(http://www.pubcare.uu.se/ULSAM/), ahealth investigation focused on iden-tifying metabolic risk factors for car-diovascular disease. All 50-year-old men

Author Affiliations: Departmentsof Public Health andCaring Sciences (Drs Ingelsson, Bjorklund-Bodegard,Arnlov, and Sundstrom) and Medical Sciences (Drs Lindand Sundstrom), Uppsala University, Uppsala, Swe-den; and AstraZeneca Research and Development,Molndal, Sweden (Dr Lind).Corresponding Author: Erik Ingelsson, MD, PhD,Department of Public Health and Caring Sciences,Geriatrics, Uppsala University, Uppsala Science Park,SE-751 85 Uppsala, Sweden ([email protected]).

Context High blood pressure is the most important risk factor for congestive heartfailure (CHF) at a population level, but the relationship of an altered diurnal blood pres-sure pattern to risk of subsequent CHF is unknown.

Objectives To explore 24-hour ambulatory blood pressure characteristics as predic-tors of CHF incidence and to investigate whether altered diurnal blood pressure pat-terns confer any additional risk information beyond that provided by conventional of-fice blood pressure measurements.

Design, Setting, and Participants Prospective, community-based, observationalcohort in Uppsala, Sweden, including 951 elderly men free of CHF, valvular disease,and left ventricular hypertrophy at baseline between 1990 and 1995, followed up un-til the end of 2002. Twenty-four-hour ambulatory blood pressure monitoring was per-formed at baseline, and the blood pressure variables were analyzed as predictors ofsubsequent CHF.

Main Outcome Measure First hospitalization for CHF.

Results Seventy men developed heart failure during follow-up, with an incidencerate of 8.6 per 1000 person-years at risk. In multivariable Cox proportional hazardsmodels adjusted for antihypertensive treatment and established risk factors for CHF(myocardial infarction, diabetes, smoking, body mass index, and serum cholesterol level),a 1-SD (9mm Hg) increase in nighttime ambulatory diastolic blood pressure (hazardratio [HR], 1.26; 95% confidence interval [CI], 1.02-1.55) and the presence of non-dipping blood pressure (night-day ambulatory blood pressure ratio 1; HR, 2.29;95% CI, 1.16-4.52) were associated with an increased risk of CHF. After adjusting foroffice-measured systolic and diastolic blood pressures, nondipping blood pressure re-

mained a significant predictor of CHF (HR, 2.21; 95% CI, 1.12-4.36 vs normal night-day pattern). Nighttime ambulatory diastolic blood pressure and nondipping blood pres-sure were also significant predictors of CHF after exclusion of all participants who hadan acute myocardial infarction before baseline or during follow-up.

Conclusions Nighttime blood pressure appears to convey additional risk informationabout CHF beyond office-measured blood pressure and other established risk factorsfor CHF. The clinical value of this association remains to be established in future studies.

JAMA. 2006;295:2859-2866 www.jama.com

2006 American Medical Association. All rights reserved. (Reprinted) JAMA, June 28, 2006Vol 295, No. 24 2859

wnloaded From: http://jama.jamanetwork.com/ on 12/16/2012

7/27/2019 Joc 60078

2/8

living in Uppsala in 1970-1973 were in-vited to join the study, of which 82%(2322 men) participated in the inves-tigation.9 The cohort was reinvesti-gated 20 years later (baseline of thepresent study: 1991-1995).

Of the 1681 available 70-year-oldmen invited to the follow-up investi-gation, 73% (1221 men) participated.For the present study, 1036 of thesemen had valid 24-hour ambulatory BPrecordings and data on all covariates.Based on the hospital register, 12 fur-ther participants were excluded be-cause of a previous diagnosis of CHFand 9 were excluded because of a pre-vious diagnosis of valvular disease; 64were excluded because of electrocar-diographic left ventricular hypertro-phy (ECG-LVH) at the baseline exami-

nation. Participants with ECG-LVHwere excluded because there was evi-dence of a significant interaction be-tween this variable and 24-hour am-bulatory diastolic BP (DBP; P=.03) andnighttime ambulatory DBP (P=.03). Be-cause the number of participants withECG-LVH was low, the analyses wererestricted to participants without it.Thus, 951 men were eligible for thepresent investigation. A subsample(n= 819) excluding all participants withmyocardial infarctionat baseline or dur-

ing follow-up was also examined. Allparticipants provided written in-formed consent and the ethics commit-tee of Uppsala University approved thestudy.

Baseline Examinations

Examinations performedwhenthe par-ticipants were 70 years old included amedical examination, a questionnaire,blood sampling (after an overnightfast), supine office BP measurement,24-hour ambulatory BP monitoring,

anthropometric measurements, andlipid determinations as previouslydescribed.8,10

Office-based BP wasmeasured in theright arm witha sphygmomanometer us-ing the appropriate cuff size. Two re-cordings were made to the nearest 2mm Hg after a 10-minute supine rest,andthe mean ofthe2 measurements was

used for the analyses. Systolic and dia-stolic BP were defined as Korotkoffphases I and V, respectively.

Twenty-four-hour ambulatory sys-tolic BP (SBP) and DBP were recordedusing Accutracker II equipment

(SunTech Medical Instruments Inc, Ra-leigh, NC). Blood pressure recordingswere made every 20 or 30 minutes be-tween 6 AM and 11 PM and every 20 or60 minutes between 11 PM and 6 AM.Data were edited by omitting all read-ings presumed to be erroneous, includ-ing readings of 0, DBP readings of morethan 170 mm Hg, SBP readings of morethan 270 or less than 80 mm Hg, andall readings in which the difference be-tween SBP and DBP was less than 10mm Hg. The measurements were in-terpreted and the data editedby skilled

laboratory technicians blinded to thestudy outcome. The Accutracker II de-vice showed satisfactory accuracy andprecision according to available docu-mentation at the time of investigationin the early 1990s.11 The coefficients ofvariation for 24-hour SBP and DBPwere6.8% and 5.5%, respectively, deter-mined in a reproducibility study per-formed in 22 participants approxi-mately 1 month after the baselineexamination. Short, fixed clock timein-tervals were used, defining daytime as

10 AM to 8 PM and nighttime as mid-night to 6 AM. This method is used toeliminate the retiring and rising peri-ods, during which BP is subject to con-siderable variation; it also helps to di-minish variationsbetween different agegroups and cultures.12

Pulse pressure was estimated as thedifference between SBP and DBP. Sus-tained hypertension was defined as of-fice BP of 140/90mm Hg or higher anddaytime ambulatory BP of 135/85mm Hg or higher; isolated ambulatory

hypertension (or masked hyperten-sion) was defined as office BP of lessthan 140/90 mm Hg and daytime am-bulatory BP of 135/85 mm Hg or higher;and isolated office hypertension (orwhite-coat hypertension) was definedas office BP of 140/90 or higher anddaytime ambulatory BP of less than135/85 mm Hg.13

NocturnalBP declinewasassessedbythe ratio between mean nighttime anddaytime SBP, and nondipping wasdefined as the ratio between nighttimeand daytime SBP of 1 or higher; ie, anabsence of nocturnal BP reduction,also

called inverted dipping. There are sev-eral reasons for supporting this defini-tion and the use of night-dayBP ratios;first, the ratio depends less on the spe-cific BP level than on the absolute noc-turnal BP decrease; second, the ratiosarenormalizedfordaytimeBP level;andthird,a ratio of 1 approximatesthe 95thpercentile of the distribution of night-day ratios in normotensive persons.14

Furthermore, this most extreme formof circadian rhythm disturbance hasbeen previously demonstrated to beassociated with the highest risk of car-

diovascular mortality.15

The presence of diabetes at baselinewas defined as fasting plasmaglucose of7.0 mmol/L (126 mg/dL) or higher oruse of oral hypoglycemic agents or in-sulin.16 The ECG-LVH was defined ashigh-amplitude R waves according to therevised Minnesota code,17 together witha left ventricular strain pattern.4 Cod-ing of smoking was based on interviewreports andmedication data were basedon questionnaire responses. At base-line, 310 participants were taking anti-

hypertensive medications: 186 withmonotherapy,101 with2 antihyperten-sivedrugs, and 23with 3 ormore drugs.The presence of valvular disease (Inter-national Classification of Diseases, NinthRevision [ICD-9] codes 394-397 and424or International Statistical Classifica-tion of Diseases, 10th Revision [ICD-10]codes I05-I08 and I34-I37) and priormyocardial infarction (ICD-9 code 410or ICD-10 code I21) was assessed fromthe hospital discharge register. The pre-cision of the myocardial infarction di-

agnosis in the Swedish hospital dis-charge register is high.18

Follow-up and Outcomes

Ninety-four men had a hospital dis-charge register diagnosis of CHF be-tween the 70-year-old baseline and De-cember 31, 2002. The ICD heart failurecodes 428 (ICD-9) and I50 (ICD-10)

DIURNAL BLOOD PRESSURE AND CONGESTIVE HEART FAILURE

2860 JAMA, June 28, 2006Vol 295, No. 24 (Reprinted) 2006 American Medical Association. All rights reserved.


7/27/2019 Joc 60078

3/8

and hypertensive heart disease withcongestive heart failure code I11.0(ICD-10) were considered to be pos-sible CHF. The medical records fromthe relevant hospitalizations were re-viewed by 2 physicians (E.I. and L.L.)

who, blinded to the baseline data, clas-sified the cases as definite, question-able, or miscoded.Cases included in thepresent study were considered defi-nite. The classification relied on thedefinition proposed by the EuropeanSociety of Cardiology19 and the reviewprocess hasbeen described in detailpre-viously.20 No participants were lost tofollow-up.

Statistical Analyses

Analyses were defined a priori. Data arereportedasmean(standarddeviation) or

number(percentage). Logarithmic trans-formationwasperformedto achieve nor-mal distribution for the 24-hour pulsepressure, nighttime SBP, and nighttimepulse pressure variables. The prognos-tic values of a 1-SD increase for the con-tinuous variables,or presence vs absencefor the dichotomous variables, for CHFincidence were investigated with Coxproportional hazards analyses. The dif-ferent hypertension definitions (sus-tained hypertension, isolated ambula-tory hypertension, and isolated office

hypertension) were compared with thereferencelevel with lowestCHF risk (ie,no hypertension, office BP 140/90mm Hg and daytime ambulatory BP135/85 mm Hg) using Cox propor-tional hazards analyses. Nonlinear rela-tionships were assessed by examiningincidence rates in quintiles of the inde-pendent variables, and night-day ratioof SBP was nonlinearly related to CHFincidence, supporting the use of adichotomous nondipping variable. Pro-portionalhazardsassumptionswere con-

firmed by Schoenfeld tests. We investi-gated theindependent variables in 5 setsof models in a hierarchical fashion:

1. Unadjusted;2. Models adjusted for antihyper-

tensivemedication (diuretics,-block-ers, angiotensin-converting enzyme in-hibitors, calcium antagonists, and-blockers as separate covariates);

3. Models adjusted for antihyper-tensive medication and established riskfactors for CHF (prior acute myocar-dial infarction, diabetes, smoking, bodymass index, and serum cholesterol)de-termined at baseline;

4. Models adjusted for antihyper-tensive medication, established riskfac-

tors for CHF, and office-measured SBPand DBP; and

5. Models adjusted for antihyper-tensive medication, established riskfac-tors for CHF, and 24-hour ambulatorySBP and DBP.

Models 2, 3, and 4 were consideredto be primary, whereas models 1 and5 were considered secondary. All mod-els were repeated in a subsample(n=819) without myocardial infarc-tion before baseline or during follow-up. To rule out an effect modification

by established risk factors on the rela-tion of ambulatory BP variablesto CHF,we investigated interaction termsbetween each of the established riskfactors (those listed for model 3,ECG-LVH, and interim myocardial in-farction) and the different BP vari-ables. The statistical power was 91% todetect a hazard ratio (HR) of 1.5 for a

1-SD increase in BP level. Two-tailed95% confidence intervals (CIs) and Pvalues are reported, with P.05 re-garded as statistically significant. PASS2002 statistical software (NCSS, Kays-ville, Utah) was used for power calcu-lations and STATA version 8.2 (StataCorp, College Station, Tex) forthe other

analyses.

RESULTS

Participants had a median follow-uptime of 9.1 years (range, 0.1-11.4 years),contributing to 8129 person-years atrisk. Seventy participants developedCHF during follow-up, with an inci-dence rate of 8.6 per 1000 person-years at risk. TABLE 1 shows the clini-cal characteristics at baseline andTABLE 2 shows the baseline BP charac-teristics. The frequency distribution of

the ratio between mean nighttime anddaytime SBP in thetotal cohort is showninFIGURE 1. The mean (SD) nighttime-daytime SBP ratio was 0.85 (0.10).

In unadjusted Coxproportional haz-ards analyses (model 1), all office BPmeasurements, 24-hour ambulatory BPmeasurements, nighttime ambulatoryBP measurements, sustained hyperten-

Table 1. Baseline Characteristics of the Cohort by Development of Heart Failure DuringFollow-up*

Characteristics

ParticipantsWho Developed CHF

(n = 70)

Participants WhoDid Not Develop CHF

(n = 881)

Prior myocardial infarction 10 (14) 53 (6)

Diabetes prevalence 12 (17) 83 (9)

Current cigarette smoking 25 (36) 169 (19)

Body mass index, mean (SD) 27.0 (3.2) 26.1 (3.3)

Serum cholesterol level, mean (SD), mmol/L 5.8 (0.8) 5.8 (1.0)

MedicationsDiuretics 18 (26) 85 (10)

-Blockers 24 (34) 159 (18)

ACE inhibitors 8 (11) 38 (4)

Calcium antagonists 19 (27) 95 (11)

-Blockers 1 (1) 12 (1)

Hypertension classificationsSustained 39 (56) 429 (49)

Isolated ambulatory 9 (13) 92 (10)

Isolated office 14 (20) 151 (17)

Nondipping 11 (16) 45 (5)

Abbreviations: ACE, angiotensin-converting enzyme; CHF, congestive heart failure.SI conversion: To convert cholesterol to mg/dL, divide by 0.0259.*Data are expressed as No. (%) of participants unless otherwise noted.Defined as weight in kilograms divided by height in meters squared.Defined as night-day ambulatory blood pressure ratio of at least 1.




7/27/2019 Joc 60078

4/8

sion, and nondipping were significantpredictors of CHF (TABLE 3). In analy-ses adjusted for antihypertensive treat-ment (model 2), nighttime ambula-tory SBP andDBPandnondipping weresignificant predictors of CHF inci-dence (Table 3). When we also ad-justed for established baseline risk fac-tors for CHF (prior acute myocardialinfarction, diabetes, smoking, bodymass index, and serum cholesterol;model 3), nighttime ambulatory DBP

and nondipping remained indepen-dent predictors of CHF in separatemodels (Table 3). A Kaplan-Meierplot for probability of survival free ofheart failure for nondipping vs nor-mal night-day BP patterns is pre-sented in FIGURE 2. When office SBPand DBP were added to the models(model 4), nondipping was a signifi-

cant predictor of CHF, whereas ambu-latory nighttime DBP became non-significant (TABLE 4). In analysesincluding 24-hour ambulatory SBP andDBP in addition to antihypertensivetreatment and establishedbaseline riskfactors (model 5), nighttime ambula-tory DBP and nondipping were signifi-cant predictors of CHF (Table 4).

To facilitate interpretation of the re-sults, we also calculated how much therisk for CHF increased for each

5mm Hg increment in nighttime am-bulatory DBP. In unadjusted models(model 1), each 5mm Hg increment ofnighttime ambulatory DBP was associ-ated with a 21% increased risk of CHF(HR, 1.21; 95% CI, 1.07-1.36). Adjust-ing for potential confounders in 4 dif-ferent models, a 5mm Hg increase innighttime ambulatory DBP was associ-

ated with a 13% to 25% increased riskof CHF (HR, 1.16; 95% CI, 1.02-1.31 inmodel2; HR, 1.14; 95% CI,1.01-1.29 inmodel 3; HR, 1.13; 95% CI, 0.98-1.30in model 4; and HR, 1.25; 95% CI, 1.01-1.54 in model 5). An increased night-time ambulatory DBP, as well as a non-

dipping BP, were also associated with anincreased absolute risk of CHF. The in-cidence of CHF was 2.7cases higher per1000 person-years at risk for each5mm Hg increment of nighttime am-bulatory DBP (ranging from approxi-mately 6 cases per 1000 person-years atrisk in the lowest BP groups [50-65mm Hg] to approximately 30 cases per1000 person-years at risk in the highestgroups[90-100 mm Hg]) and15.1 caseshigher per 1000 person-years at risk forthose with nondipping vs normal night-

day BP pattern (22.8 vs 7.7 cases).In the subsample without myocar-

dial infarction before baseline or dur-ing follow-up, several BP variablesweresignificant predictors of CHF in unad-justed analyses (model 1; Table 3). Ad-justing for antihypertensive treatment(model 2) in this subsample, office-measured SBP, 24-hour ambulatoryDBP, nighttime ambulatory SBP andDBP, and nondipping were significantpredictors of CHF incidence (Table 3).After further adjustment for the estab-

lished risk factors for CHF (model 3),office SBP, nighttime ambulatory DBP,and nondipping remained significantpredictors of CHF(Table3).When add-ing office SBP and DBP to the models(model 4), nighttime ambulatory DBPand nondipping were significant pre-dictors of CHF (Table 4). In analysesincluding 24-hour ambulatory SBP and

Figure 1. Frequency Distribution of the Ratio Between Mean Nighttime and Daytime SystolicBlood Pressure in the Overall Cohort

150

50

100

125

75

25

0

Frequency,

No.

Ratio Between Nighttime and Daytime Systolic Blood Pressure

0.6 0.8 1.0 1.2 1.4

Fifty-six participants exhibited nondipping (a nighttime-daytime systolic blood pressure ratio 1). The bluecurve is the kernel-density estimate using a bandwidth of 0.03.

Table 2. Baseline Blood Pressure Measurements of the Total Cohort and by Development of CHF During Follow-Up

Measurement

Mean (SD), mm Hg

ParticipantsWho Developed CHF

(n = 70)

ParticipantsWho Did Not Develop CHF

(n = 881)Total Cohort

(N = 951)

SBP DBP PP SBP DBP PP SBP DBP PP

Office 152 (19) 86 (10) 66 (15) 146 (18) 83 (9) 62 (14) 146 (18) 84 (9) 63 (14)

24-hour ambulatory 136 (12) 76 (7) 60 (10) 132 (16) 75 (8) 58 (12) 133 (16) 75 (8) 58 (12)

Daytime ambulatory 143 (14) 80 (9) 63 (11) 140 (16) 79 (8) 60 (12) 140 (16) 79 (8) 61 (12)

Nighttime ambulatory 124 (19) 69 (10) 55 (12) 118 (18) 66 (8) 52 (14) 119 (18) 66 (9) 52 (14)

Abbreviations: CHF, congestive heart failure; DBP, diastolic blood pressure; PP, pulse pressure; SBP, systolic blood pressure.




7/27/2019 Joc 60078

5/8

Table 3. Risk of Congestive Heart Failure by Type of Blood Pressure Measurement in the Total Sample and in a Subsample WithoutMyocardial Infarction

Hazard Ratio (95% Confidence Interval)*

Unadjusted(Model 1)

Adjusted forAntihypertensive Treatment

(Model 2)

Adjusted forAntihypertensive Treatment and

Established Risk Factors(Model 3)

Total sample (N = 951)Office measurements

SBP 1.40 (1.11-1.75) 1.21 (0.95-1.53) 1.25 (0.98-1.59)

DBP 1.31 (1.04-1.66) 1.13 (0.89-1.44) 1.16 (0.91-1.49)

PP 1.30 (1.03-1.63) 1.17 (0.93-1.48) 1.20 (0.95-1.52)

24-hour ambulatory measurementsSBP 1.26 (1.03-1.56) 1.17 (0.95-1.45) 1.13 (0.91-1.40)

DBP 1.26 (1.01-1.58) 1.17 (0.93-1.47) 1.13 (0.90-1.42)

PP 1.25 (1.00-1.55) 1.18 (0.95-1.47) 1.14 (0.91-1.42)

Daytime ambulatory measurementsSBP 1.20 (0.96-1.49) 1.13 (0.90-1.41) 1.08 (0.85-1.36)

DBP 1.11 (0.88-1.40) 1.03 (0.81-1.31) 0.99 (0.78-1.26)

PP 1.18 (0.95-1.46) 1.14 (0.92-1.41) 1.10 (0.88-1.37)

Nighttime ambulatory measurementsSBP 1.34 (1.09-1.64) 1.25 (1.01-1.54) 1.21 (0.98-1.49)

DBP 1.38 (1.12-1.71) 1.29 (1.04-1.59) 1.26 (1.02-1.55)

PP 1.25 (1.02-1.53) 1.17 (0.95-1.45) 1.14 (0.93-1.41)

Hypertension classificationsSustained 2.27 (1.06-4.86) 1.74 (0.80-3.77) 1.75 (0.80-3.85)

Isolated ambulatory 2.57 (0.99-6.67) 2.43 (0.94-6.31) 2.18 (0.83-5.76)

Isolated office 2.25 (0.94-5.37) 1.81 (0.75-4.35) 2.01 (0.82-4.91)

Nondipping 2.88 (1.51-5.48) 2.68 (1.38-5.21) 2.29 (1.16-4.52)

Subsample without myocardialinfarction (n = 819)

Office measurementsSBP 1.51 (1.13-2.03) 1.40 (1.03-1.91) 1.39 (1.02-1.89)

DBP 1.36 (1.00-1.84) 1.21 (0.88-1.67) 1.21 (0.87-1.67)

PP 1.42 (1.06-1.89) 1.35 (1.00-1.83) 1.35 (1.00-1.84)

24-ambulatory measurements

SBP 1.37 (1.07-1.76) 1.27 (0.98-1.65) 1.19 (0.92-1.54)DBP 1.52 (1.16-2.00) 1.42 (1.07-1.88) 1.32 (0.99-1.75)

PP 1.25 (0.95-1.66) 1.18 (0.89-1.56) 1.13 (0.86-1.49)

Daytime ambulatory measurementsSBP 1.28 (0.97-1.69) 1.20 (0.92-1.58) 1.13 (0.86-1.50)

DBP 1.27 (0.96-1.69) 1.20 (0.90-1.61) 1.12 (0.83-1.50)

PP 1.18 (0.90-1.56) 1.14 (0.87-1.49) 1.10 (0.83-1.44)

Nighttime ambulatory measurementsSBP 1.45 (1.13-1.86) 1.35 (1.04-1.75) 1.29 (0.99-1.66)

DBP 1.66 (1.29-2.14) 1.53 (1.19-1.97) 1.48 (1.14-1.92)

PP 1.24 (0.96-1.61) 1.16 (0.88-1.52) 1.13 (0.87-1.47)

Hypertension classificationsSustained 3.27 (1.15-9.34) 2.69 (0.93-7.77) 2.52 (0.86-7.39)

Isolated ambulatory 1.88 (0.42-8.40) 1.76 (0.39-7.88) 1.61 (0.35-7.31)

Isolated office 2.28 (0.67-7.78) 1.93 (0.56-6.70) 2.15 (0.61-7.51)Nondipping 3.17 (1.41-7.15) 2.99 (1.28-7.00) 2.82 (1.20-6.62)

Abbreviations: DBP, diastolic blood pressure; PP, pulse pressure; SBP, systolic blood pressure.*Cox proportional hazards ratios for a 1-SD increase in continuous variables or presence vs absence of dichotomous variables. The hazard ratios presented for sustained hyper-

tension, isolated ambulatoryhypertension, and isolated officehypertension were compared witha referent of no hypertension. Hazardratios and 95% confidenceintervalsshownare unadjusted (model 1), adjusted for antihypertensive treatment (model 2), or adjusted for antihypertensive treatment and established risk factors for CHF (prior acute myocar-dial infarction, diabetes, smoking, body mass index, and serum cholesterol) determined at baseline (model 3). P.05 was considered statistically significant.

P.01.P.05.P.001.Without myocardial infarction before baseline or during follow-up.Defined as night-day ambulatory blood pressure ratio of at least 1.




7/27/2019 Joc 60078

6/8

DBP in addition to antihypertensivetreatment and established baseline riskfactors (model 5), nighttime ambula-tory DBP and nondipping were signifi-cant predictors of CHF in this sub-sample (Table 4).

None of the investigated interac-tion terms was significant in the mainanalysis sample (excluding men withprevalent CHF, valvular disease, orECG-LVH) or in thesample without in-terim myocardial infarction.

COMMENT

In this community-based sample of el-derly men free of CHF, valvular diseaseand ECG-LVH at baseline, a nondip-ping night-day BP pattern, and in-creased nighttime diastolic BP pre-dicted CHF incidence independent ofantihypertensive treatment and estab-lished risk factors for CHF, includingmyocardialinfarction duringthe follow-up. Furthermore, a nondipping night-day BP pattern increased the risk ofCHF

evenafter adjusting for conventionalof-fice BP measurement.This indicatesthatnighttime BP patterns may be impor-tant in development of CHF and that atraditional office BP measurement doesnot capture all of the increased risk thatan increased nighttime BP conveys.

Twenty-four-hour ambulatory BP hasrepeatedly been demonstrated to be a

powerful predictor of future cardiovas-cular morbidity and mortality, even af-ter adjustment for conventional officeBP.5-8 Staessen et al5 demonstrated thata nondipping BP pattern was associatedwith an increased cardiovascularriskandthat nighttime BP more accurately pre-dicted cardiovascular events than day-time BP.5 Other studies also have iden-tified a nondippingpatternas a risk factorfor cardiovascular disease.21,22 Previousstudies have included CHF as a part of

a combined end point but have not ex-amined whether a 24-hour ambulatoryBP measurement predicts CHF per se.The pathophysiology of atheroscleroticdisease and CHF are not the same, andmost new-onset CHF is not preceded bymyocardial infarction.4,23,24 Therefore,studies of predictors of CHF, account-ing properly for myocardial infarctionand risk factors for atherosclerotic dis-ease, are warranted.

A reduced circadian BP variation isa common finding in CHF patients,

as reviewed by Goyal et al.

25

An in-creased nighttime ambulatory BP hasbeen relatedto left ventricular filling im-pairment in cross-sectional studies withlimited samples.26,27 In a recent cross-sectional study of patients with hyper-tension and type 2 diabetes mellitus,diastolic dysfunction was closely re-lated to increased diastolic BP and non-

dipping.28 However, to our knowl-edge, there are no previous studies of24-hour ambulatory BP patterns as pre-dictors of incident CHF.

Several studies have established thatnondipping is associated with endo-

thelial dysfunction and hemosta-sis.29-31 Recent studies indicate that en-dothelial dysfunction is an importantcomponent of the pathophysiologicalmechanisms of CHF, and it has beenassociated with the progression andprognosis of CHF.32,33 Thus, it is pos-sible that endothelial dysfunction couldbe a link between increased nocturnalBP and CHF.

Another possible common patho-physiological mechanism is increasedsympathetic activity, which is associ-ated with a nondipping BP pattern34,35

and also is a presumed causal factor inCHF.36

One point of controversy is whetherthe absence of a nighttime BP decreaseper se or an increased 24-hour BP loadcauses organ damage. To address thispossibility, 24-hour ambulatorySBP andDBP were included as covariates in ad-dition to antihypertensivetreatment andestablishedrisk factors (model 5)in bothsamples. In these analyses, nondippingand nighttime DBP remained signifi-cant predictors of CHF, indicating that

the nondipping BP pattern per se is im-portantor is an indicator of an impor-tant trait. An example of such a traitcould be sleep apnea, a condition thathasbeen suggestedto be associated bothwith CHF37,38 and a nondipping BP pat-tern.39 However, since this study wasnotdesignedto address the possibleinvolve-ment ofsleep apnea, this needs to be ex-amined in future studies.

A possible explanation for observedimportance of nighttime ambulatory BPmight be that the intraindividual BP

variation is lower than that for day-time BP. This may be due to high BPconsistencyduringsleepcompared withdaytime BP, which is influenced moreby physical and psychological activ-ity. However, it should be noted thatthis study did not primarily address thepathophysiological mechanisms be-hind the association between night-

Figure 2. Kaplan-Meier Plot for Probability of Heart FailureFree Survival in the Total Cohortfor Nondipping vs Normal Night-Day Blood Pressure Patterns

1.0

0.3

0.6

0.5

0.4

0.7

0.8

0.9

0.2

0.1

0

No. at Risk

Normal Night-Day Pattern

Normal Night-Day Pattern

Log RankP

7/27/2019 Joc 60078

7/8

time BP pattern and CHF. That has tobe examined in other settings.

Thestrengthsofthis study include thelarge,community-based population andthe long follow-up period. Further-more, allCHFcases werevalidated,lim-

itingthe inclusionof false-positive cases.However, there are some limitations tothis study. Because we only examinedmen of the same age with a similar eth-nicbackground, thisstudyhas unknowngeneralizability to women or other ageandethnic groups. However, we did cir-cumvent the powerful effects of age onCHF incidence. Moreover, due to limi-tations in sample size, men using anti-hypertensivemedications wereincludedinthestudy population,whichmayhaveaffectedtheresults through residualcon-founding,despitetheadjustmentfor spe-

cificantihypertensivetreatment.Anotherpossible limitation is that multiple sta-tistical testing is present to some degree.However, all analyses were specified apriori and the findings were consistentin all models and in the subsample.

Since evidence of effect modificationbetween some of the BP variables andECG-LVH was found, the study samplehadto be stratifiedinto 2 parts: one stra-tum with participants with ECG-LVHatbaseline andtheother without these par-ticipants. The stratum consisting of par-

ticipants with ECG-LVH was unfortu-nately too small (64 participants, ofwhom 11 developedCHF duringthe fol-low-up) to be analyzed. Thus, the analy-ses were restricted to the stratum withparticipants without ECG-LVH, whichalso might be considered to be a limita-tion of the study.

Milder, nonhospitalized cases of CHFwere not included in our end point,which maybe considereda limitationbutwould tend to bias the results toward thenull hypothesis. Since the CHF diagno-

sis was based on a review of medical re-cords, it was not possible to differenti-ate between systolic and diastolic heartfailure because echocardiography wasnot available at the time of diagnosisformany of the cases. Thus, we could notexamine whether the impact of BP pat-tern is different for systolic vs diastolicheart failure. Finally, the diagnosis of

Table 4. Risk of Congestive Heart Failure by Type of Blood Pressure Measurement in theTotal Sample and in a Subsample Without Myocardial Infarction, Adjusted for Previous FactorsPlus Office or Ambulatory Blood Pressure

Hazard Ratio (95% Confidence Interval)*

Adjusted forAntihypertensive Treatment,

Established Risk Factors,and Office Blood Pressure

(Model 4)

Adjusted forAntihypertensive Treatment,

Established Risk Factors,

and 24-hour AmbulatoryBlood Pressure

(Model 5)

Total sample (N = 951)Office measurements

SBP NA 1.23 (0.92-1.65)

DBP NA 1.11 (0.82-1.51)

PP NA 1.19 (0.91-1.56)

24-hour ambulatory measurementsSBP 1.01 (0.77-1.32) NA

DBP 1.05 (0.79-1.39) NA

PP 1.05 (0.80-1.37) NA

Daytime ambulatory measurementsSBP 0.94 (0.70-1.25) 0.78 (0.45-1.36)

DBP 0.87 (0.66-1.16) 0.53 (0.30-0.92)

PP 1.01 (0.78-1.30) 1.14 (0.65-2.02)

Nighttime ambulatory measurementsSBP 1.14 (0.89-1.44) 1.41 (0.92-2.14)

DBP 1.23 (0.97-1.58) 1.47 (1.03-2.11)

PP 1.07 (0.84-1.35) 1.24 (0.80-1.90)

Hypertension classificationsSustained 0.98 (0.35-2.76) 1.41 (0.55-3.59)

Isolated ambulatory 2.01 (0.76-5.33) 1.91 (0.69-5.29)

Isolated office 1.32 (0.48-3.62) 1.95 (0.80-4.76)

Nondipping 2.21 (1.12-4.36) 2.19 (1.06-4.53)

Subsample without myocardialinfarction (n = 819)

Office measurementsSBP NA 1.32 (0.91-1.92)

DBP NA 1.01 (0.67-1.51)

PP NA 1.37 (0.97-1.94)

24-hour ambulatory measurementsSBP 1.04 (0.76-1.44) NA

DBP 1.29 (0.90-1.84) NA

PP 0.97 (0.70-1.36) NA

Daytime measurementsSBP 0.95 (0.67-1.35) 0.66 (0.31-1.38)

DBP 1.00 (0.71-1.43) 0.45 (0.23-0.87)

PP 0.94 (0.68-1.31) 1.12 (0.52-2.41)

Nighttime measurementsSBP 1.18 (0.88-1.58) 1.53 (0.88-2.65)

DBP 1.47 (1.10-1.97) 1.62 (1.07-2.46)

PP 1.00 (0.73-1.37) 1.21 (0.65-2.24)

Hypertension classificationsSustained 1.65 (0.42-6.50) 1.97 (0.57-6.83)

Isolated ambulatory 1.52 (0.33-6.97) 1.37 (0.29-6.55)

Isolated office 1.55 (0.38-6.32) 2.11 (0.60-7.41)Nondipping 2.58 (1.09-6.12) 2.73 (1.10-6.76)

Abbreviations: DBP, diastolic blood pressure; NA, not applicable; PP, pulse pressure; SBP, systolic blood pressure.*Cox proportional hazards ratios for a 1-SD increase in continuous variables or presence vs absence of dichotomous vari-

ables. The hazard ratios presented for sustained hypertension, isolated ambulatory hypertension, and isolated office hy-pertension were compared with a referent of no hypertension. Hazard ratios and 95% confidence intervals shown areadjusted for antihypertensive treatment, established risk factors (as defined in Table 3), and office blood pressure (model4) or adjusted for antihypertensive treatment, established risk factors, and 24-hour ambulatory bloodpressure (model 5).P.05 was considered statistically significant.

P.05.Without myocardial infarction before baseline or during follow-up.Defined as night-day ambulatory blood pressure ratio of at least 1.




7/27/2019 Joc 60078

8/8

prevalent CHF at baseline was definedas previous hospitalization for CHF.Since nondipping is associated withprevalent CHF, it is plausible that partof the association between nondippingandincident CHFidentifiedin this study

could be a result of undiagnosed preva-lent CHF at baseline that only later wassevere enough to require hospitaliza-tion. However, in a substudy of 343 par-ticipants without prevalent CHF, only 5(1.5%) hadan ejection fraction less than0.40 on echocardiography; only 1 ofthese 5 had nondipping BP, and 3 de-veloped CHF. Therefore, even if CHFwere not diagnosed, it would have beenuncommon.

CONCLUSION

In conclusion, a nondipping night-

day BP pattern and an increased night-time BP predicted CHF incidence in-dependent of established risk factorsinour large, community-based sample ofelderly men. Nondipping was also a riskfactor for CHF even when taking con-ventional office BP measurement intoaccount. Nighttime BP appears to con-vey additive risk information aboutCHF, but its clinical value remains tobe established in future studies.

Author Contributions: Dr Ingelssonhad full accesstoall of thedata in thestudyand takes responsibility forthe integrity of the data and the accuracy of the data

analysis.Study conce pt and desi gn: Ingelsson, Bjorklund-Bodegard, Lind , Arnlov, Sundstrom.Acquisition of data: Ingelsson, Bjorklund-Bodegard.Analysis and interpretation of data: Ingelsson,Bjorklund-Bodegard, Lind , Arnlov, Sundstrom.Drafting of the manuscript: Ingelsson, Bjorklund-Bodegard.Critical revision of the manuscript for important in-tellectual content: Bjorklund-Bodegard, Lind, Arnlov,Sundstrom.Statistical analysis: Ingelsson, Bjorklund-Bodegard,Arnlov, Sundstrom.Obtained funding: Ingelsson.Administrative, technical, or material support:Bjorklund-Bodegard.Study supervision: Lind, Sundstrom.Financial Disclosures: Dr Lind is a part-time em-ployee at AstraZeneca Research and Development,

Molndal, Sweden, anda part-timeemployee at UppsalaUniversity(AstraZenecahas no interestsin thisprojectandhas notprovided anyfinancial support).No otherdisclosures were reported.Funding/Support: Funding was provided by PrimaryHealth Care in Uppsala County, the Swedish HeartLung Foundation (Hja rt-Lungfonden), and theThureus Foundation .Roleof theSponsors: Thefundingsources hadno rolein the design and conduct of the study, in the collec-tion, analysis, andinterpretationof thedata,or in thepreparation, review, or approval of the manuscript.

REFERENCES

1. McMurray JJ, Pfeffer MA. Heart failure. Lancet.2005;365:1877-1889.2. Berry C, Murdoch DR, McMurray JJ.Economics ofchronicheart failure.Eur J HeartFail. 2001;3:283-291.3. Levy D, Kenchaiah S, Larson MG, et al. Long-term trendsin theincidenceof andsurvival with heartfailure. N Engl J Med. 2002;347:1397-1402.4. Levy D, LarsonMG, VasanRS, KannelWB, Ho KK.The progression from hypertensionto congestive heartfailure. JAMA. 1996;275:1557-1562.5. Staessen JA, Thijs L, Fagard R, et al; Systolic Hy-pertension in Europe TrialInvestigator. Predicting car-diovascular riskusing conventional vs ambulatorybloodpressure in older patients with systolic hypertension.JAMA. 1999;282:539-546 s.6. Verdecchia P, Reboldi G, Porcellati C, et al.Risk ofcardiovascular diseasein relationto achieved officeandambulatory blood pressure control in treated hyper-tensivesubjects.J Am Coll Cardiol. 2002;39:878-885.7. ClementDL, DeBuyzereML,De BacquerDA, etal.Prognostic value of ambulatory blood-pressure re-cordingsin patients with treated hypertension. N EnglJ Med. 2003;348:2407-2415.8. BjorklundK, Lind L, Zethelius B, Berglund L, LithellH. Prognostic significance of 24-h ambulatory bloodpressure characteristics for cardiovascular morbidityin a population of elderly men.J Hypertens. 2004;22:

1691-1697.9. HedstrandH. A studyof middle-agedmen withpar-ticular reference to risk factors for cardiovasculardisease. Ups J Med Sci Suppl. 1975;19:1-61.10. Byberg L, Siegbahn A, Berglund L, McKeigue P,Reneland R, Lithell H. Plasminogen activator inhibi-tor-1 activity is independently related to both insulinsensitivity andserum triglyceridesin 70-year-old men.Arterioscler Thromb Vasc Biol. 1998;18:258-264.11. Appel LJ, Whelton PK, Seidler AJ, Patel AR, KlagMJ.The accuracy andprecisionof theAccutracker am-bulatory blood pressure monitor. Am J Epidemiol.1990;132:343-354.12. OBrien E, Asmar R, Beilin L, et al. European So-ciety of Hypertension recommendations for conven-tional, ambulatory and home blood pressuremeasurement. J Hypertens. 2003;21:821-848.13. Pickering TG, Hall JE, Appel LJ, et al. Recommen-dations for bloodpressure measurement in humans and

experimental animals, I: blood pressure measurementin humans: a statement forprofessionalsfrom theSub-committee of Professionaland Public Education of theAmericanHeart Association Council on HighBloodPres-sure Research. Circulation. 2005;111:697-716.14. StaessenJA, Bieniaszewski L,OBrienE, et al;AdHoc Working Group. Nocturnal blood pressure fallon ambulatory monitoring in a large internationaldatabase. Hypertension. 1997;29:30-39.15. Ohkubo T, Hozawa A, Yamaguchi J, et al. Prog-nostic significance of the nocturnal decline in bloodpressure in individuals with and without high 24-hbloodpressure: the Ohasama study.J Hypertens. 2002;20:2183-2189.16. Expert Committee on the Diagnosis and Classi-ficationof DiabetesMellitus.Report of theExpert Com-mittee on the Diagnosis and Classification of Diabe-tes Mellitus. Diabetes Care. 2003;26(suppl 1):S5-S20.

17. Prineas RJ, Crow RS, Blackburn H. The Minne-sota Code Manual of Electrocardiographic Findings:Stand ards and Proce dures for Measu reme nt andClassification. Bristol, England: John Wright; 1982.18. LindbladU, RastamL, RanstamJ, Peterson M. Va-lidity of register data on acute myocardial infarctionandacute stroke: theSkaraborg HypertensionProject.Scand J Soc Med. 1993;21:3-9.19. Task Force on Heart Failure of the European So-cietyof Cardiology.Guidelines for thediagnosisof heartfailure. Eur Heart J. 1995;16:741-751.20. Ingelsson E, Arnlov J, Sundstrom J, Lind L. The

validity of a diagnosisof heart failure in a hospital dis-charge register. Eur J Heart Fail. 2005;7:787-791.21. Verdecchia P, Porcellati C, Schillaci G, et al. Am-bulatory bloodpressure: an independent predictor ofprognosis in essential hypertension. Hypertension.1994;24:793-801.22. Ohkubo T, Imai Y, Tsuji I, et al. Relation be-tween nocturnal decline in blood pressure and mor-tality: the Ohasama Study. Am J Hypertens. 1997;10:

1201-1207.23. ChenYT, Vaccarino V, Williams CS, Butler J, Berk-man LF, Krumholz HM. Risk factors for heart failurein the elderly: a prospectivecommunity-based study.Am J Med. 1999;106:605-612.24. Gottdiener JS,ArnoldAM, AurigemmaGP, et al.Predictors of congestive heart failurein theelderly: theCardiovascular Health Study.J Am Coll Cardiol. 2000;35:1628-1637.25. GoyalD, Macfadyen RJ,WatsonRD, LipGY. Am-bulatory blood pressure monitoring in heart failure: asystematic review. Eur J Heart Fail. 2005;7:149-156.26. Verdecchia P, Schillaci G, GuerrieriM, et al. Preva-lenceand determinants of left ventriculardiastolicfill-ing abnormalities in an unselected hypertensivepopulation. Eur Heart J. 1990;11:679-691.27. Galderisi M, Petrocelli A, Alfieri A, Garofalo M,deDivitiis O. Impactof ambulatory bloodpressureonleft ventricular diastolic dysfunction in uncompli-

cated arterial systemic hypertension. Am J Cardiol.1996;77:597-601.28. Andersen NH, Poulsen SH, Poulsen PL, et al. Leftventricular dysfunction in hypertensivepatients with type2 diabetes mellitus. Diabet Med. 2005;22:1218-1225.29. Higashi Y, Nakagawa K, Kimura M, et al. Circa-dianvariationof blood pressure and endothelial func-tion in patients with essential hypertension: a com-parisonof dippersand non-dippers.J Am Coll Cardiol.2002;40:2039-2043.30. vonKanel R, Jain S,MillsPJ, et al.Relation of noc-turnal blood pressure dipping to cellular adhesion,inflammation and hemostasis. J Hypertens. 2004;22:2087-2093.31. Lee KW, Blann AD, Lip GY. High pulse pressureand nondipping circadian blood pressure in patientswith coronary artery disease: relationship to throm-bogenesis and endothelial damage/dysfunction. AmJ Hypertens. 2005;18:104-115.

32. Chong AY, Blann AD, Patel J, Freestone B, HughesE, Lip GY. Endothelial dysfunctionand damage in con-gestive heart failure: relation of flow-mediated dila-tion to circulating endothelial cells,plasmaindexes ofendothelial damage, and brain natriuretic peptide.Circulation. 2004;110:1794-1798.33. Tousoulis D, Charakida M, Stefanadis C. Inflam-mation andendothelial dysfunctionas therapeutictar-gets in patients with heart failure. IntJ Cardiol. 2005;100:347-353.34. SicaDA.Whatare theinfluencesof salt, potassium,the sympathetic nervous system, and the renin-angiotensinsystemon the circadian variationin bloodpressure?Blood Press Monit. 1999;4(suppl 2):S9-S16.35. Sherwood A, Steffen PR,Blumenthal JA,Kuhn C,Hinderliter AL. Nighttime blood pressure dipping:the role of the sympathetic nervous system. Am JHypertens. 2002;15:111-118.36. BellDS. Heart failure:the frequent, forgotten, and

often fatal complication of diabetes. Diabetes Care.2003;26:2433-2441.37. Javaheri S, Parker TJ, Liming JD, et al. Sleep ap-nea in 81 ambulatory male patients with stable heartfailure:types and their prevalences,consequences, andpresentations. Circulation. 1998;97:2154-2159.38. Bradley TD,LoganAG, Kimoff RJ,et al.Continu-ous positive airway pressure for central sleep apneaandheartfailure. NEngl J Med. 2005;353:2025-2033.39. Ziegler MG. Sleep disorders and the failure tolower nocturnal blood pressure. Curr Opin NephrolHypertens. 2003;12:97-102.



Documents

Joc 60078