Orthostatic hypotension and the risk of myocardial infarction in the home-dwelling elderly


  • H. Luukinen,

    1. From the Department of Public Health Science and General Practice, University of Oulu, and Unit of General Practice, Oulu University Hospital, University of Oulu, Oulu
    Search for more papers by this author
  • K. Koski,

    1. From the Department of Public Health Science and General Practice, University of Oulu, and Unit of General Practice, Oulu University Hospital, University of Oulu, Oulu
    Search for more papers by this author
  • P. Laippala,

    1. Tampere School of Public Health, Biometry Unit, University of Tampere, and Tampere University Hospital, Tampere
    Search for more papers by this author
  • K. E. J. Airaksinen

    1. Cardiology Unit, Department of Medicine, Turku University Hospital, University of Turku, Turku; Finland
    Search for more papers by this author

Heikki Luukinen MD, PhD, Department of Public Health Science and General Practice, University of Oulu, and Unit of General Practice, Oulu University Hospital, University of Oulu, PB 5000, FIN-90014, Finland.
(fax: +358 8 5375661; e-mail: heikki.luukinen@oulu.fi).


Objectives.  We investigated the prognostic significance of orthostatic hypotension on the risk of myocardial infarction (MI) amongst the elderly.

Design.  Prospective population-based study. 

Setting.  Home-dwelling population.

Subjects.  Orthostatic testing was performed between 8 a.m. and 2 p.m., irrespective of having had meals, on 792 persons, representing 82% of all home-dwelling persons aged ≥70 years living in five municipalities around the city of Oulu.

Main outcome measures.  Occurrence of cases of MI were recorded during mean 3.58 (SD 1.09) years follow-up period, from national mortality statistics and local hospital discharge registers.

Results.  Ninety cases of MI, of which 40 were fatal after initial hospitalization, occurred during the follow-up period. Orthostatic diastolic blood pressure (BP) drop 1 min after standing up was associated with subsequent MI, but systolic BP reactions had no predictive value. According to the Cox regression model, the strongest predictor of the occurrence of subsequent MI was found in regard to ≥8 mmHg drop in diastolic BP 1 min after standing up; adjusted for history of MI, diabetes mellitus, chest pain, use of calcium antagonist, β-blocker, nitrate and diuretic medication, hazard ratio of MI being 2.00 (1.11–3.59).

Conclusions.  Orthostatic testing offers a novel means to assess the risk of MI amongst elderly persons. Diastolic BP drop immediately after standing up identifies elderly subjects at a high risk of subsequent MI.


Orthostatic testing has been traditionally used for the assessment of syncope [1] and liability to falls [2] in elderly subjects. Orthostatic hypotension (OH) is a common phenomenon in the elderly, and if defined as a 20 mmHg drop in systolic blood pressure (BP) OH affects nearly one-third of the elderly [3–5]. Since orthostatic testing is a simple means to assess baroreflex-mediated rapid BP changes [6], it may offer significant prognostic information. The risk of death is increased amongst men [7] and hypertensive diabetic patients with systolic OH (OH-S) [8]. Highly variable orthostatic BP measures increase the risk of stroke in elderly nursing home residents [9]. According to the results of a recent population-based study, both diastolic OH (OH-D) and OH-S increase the risk of cardiovascular mortality [5]. However, there are no data on the potential value of OH in the prediction of MI. We examined the association between OH and fatal and nonfatal cases of MI in a large population of home-dwelling elderly people.



This study is part of a large population-based project with the primary aim of examining risk factors for falls and the effects of prevention of falls in older persons. The population of the study has been presented in our previous report [5]. The population originally consisted of all home-dwelling persons aged 70 years or over who lived in five rural municipalities around the city of Oulu in Northern Finland on 1 September 1991 (n = 969). The institutional ethics committee approved the study protocol, and informed consent was obtained before interviewing the subjects. A total of 833 persons (86%) participated in the examinations, during which orthostatic testing was performed. This was successfully carried out in 792 (82%) subjects.


Baseline data were collected by means of mailed questionnaires, interviews, clinical examinations and clinical tests performed by two teams including two nurses and a physician between 1 September 1991 and 29 February 1992. Because of time-consuming and strenuous examinations, each participant was examined on two separate days at an average interval of 3 months, without repetition of any measurement. Cases of MI were recorded from the day of the orthostatic test up to the end of the follow-up period, 31 December 1995.

The occurrence of chest pain was asked by means of a standardized postal questionnaire according to a question ‘how often do you experience chest pain?’ (no, once a week, twice a week, daily), there being no gender differences in the distribution of frequency of chest pain; χ2 = 2.24, d.f. = 3, P = 0.5240. Diabetes mellitus (no, diet-treated, tablet- or insulin-treated), and current smoking habits (yes/no) were also recorded via questionnaires. The questionnaires were checked by one of the nurses, when the subject came to the clinical examinations. Chronic diseases other than diabetes were noted by the examining physician from the medical records of the health centres, and national health insurance documents were also reviewed to record chronic diseases requiring medication. The examining physician recorded current medication use by interviewing the participants. Additional information was derived from interviews, prescriptions and drug packages that the subjects had been asked to bring to the examination. Weight and height in light clothing without shoes were measured and body mass index (BMI) (kg m−2) was calculated by one of the nurses [5].

Orthostatic testing was performed by two trained nurses between 8 a.m. and 2 p.m., irrespective of the subjects having had meals. BP (2 mmHg accuracy) and heart rate (two-beat accuracy during 30 s) were measured after 5 min rest, the subject in a supine position, from the right arm, unless it was injured, using a mercury manometer with a 12.5 × 40 cm cuff and a stethoscope with a bell. After this the subject was asked to stand up from the lying position. The BP measurements were repeated after the subject had been standing, unassisted, for 1 and 3 min. OH-D was defined as a drop of diastolic BP ≥10 mmHg, and OH-S as a drop of systolic BP of ≥20 mmHg, either 1 or 3 min after standing up. Similarity of the BP recordings was confirmed in pretests performed by the nurses using a two-limb stethoscope [5].

Death certificates from the Central Statistical Office of Finland were used to record deaths due to codes 4100–4109 of ICD-9 (myocardial infarctions, MI) in the cohort from the day of orthostatic testing up to the end of the follow-up period, 31 December 1995. Data on nonfatal cases of MI were derived from the statistics of Oulu University Hospital during the same period and according to the same diagnoses. Data on MIs that occurred from 1976 to the start of the follow-up period were derived from the same statistics and were based on codes 410.00–410.99 (ICD-8) during 1976–1986 and codes 4100–4109 (ICD-9) from that time, onwards.

Statistical analysis

Student's t-test was used to compare continuous data, and the chi-square statistics to compare binary data. When appropriate, continuous data were dichotomized at one SD below or above the mean. Kaplan–Meier estimation was used to analyse survival distributions until the first fatal or nonfatal MI, and Cox regression modelling was used to analyse survival time. The proportionality assumption was tested by constructing interaction terms between the variables and time to first MI. Cox regression analyses showed no statistically significant interactions with time (each P > 0.05).

The results concerning the relative risks are summarized as hazard ratios (HRs) with 95% confidence intervals (95% CIs). Computation was carried out using commercially available software (BMDP Statistical Software Inc, Los Angeles, CA, USA) on a SUN/UNIX mainframe computer (Sun Microsystems, Inc., Palo Alto, CA, USA).


The characteristics of the study population according to presence of OH-D and OH-S are described in Table 1. More persons with OH-D than those without used a calcium antagonist or a diuretic medication. Subjects with OH-S had lower BMIs and higher diastolic and systolic BPs and more of them had diabetes mellitus than those without OH-S.

Table 1.  Characteristics of the study population (n = 792) according to the presence of diastolic and systolic orthostatic hypotensiona
Present (n = 64)Absent (n = 728)Present (n = 212)Absent (n = 580)
  1. Numbers are mean ± SD or n (%); BP, blood pressure.

  2. aDiastolic orthostatic hypotension (OH-D) defined as a drop of diastolic blood pressure ≥10 mmHg; systolic orthostatic hypotension (OH-S) defined as a drop of systolic blood pressure ≥20 mmHg either 1 or 3 min after standing up.

  3. bFrom the chi-square test (stratified variables) or t-test (continuous variables).

Age (years) 77 ± 6 76 ± 50.082 76 ± 5 76 ± 50.399
Body mass index (kg m−2) 27 ± 5 29 ± 50.077 28 ± 4 29 ± 50.003
Systolic BP (mmHg)160 ± 27157 ± 250.500167 ± 24154 ± 25<0.001
Diastolic BP (mmHg) 82 ± 14 81 ± 110.654 83 ± 11 80 ± 11<0.001
Supine heart rate (beats min−1) 70 ± 13 69 ± 110.698 68 ± 11 70 ± 110.051
Male sex 22 (34)277 (38)0.561 90 (42)209 (36)0.099
Chest pain 30 (47)264 (36)0.099 84 (40)210 (36)0.415
History of myocardial infarction 7 (11) 60 (8)0.458 18 (8) 49 (8)0.985
Diabetes mellitus 18 (28)140 (19)0.092 57 (27)101 (18)0.004
Current smoking 6 (9) 55 (8)0.610 16 (8) 45 (8)0.902
Use of nitrate 30 (51)309 (44)0.297 85 (41)254 (45)0.327
Use of calcium antagonist 19 (32)129 (18)0.009 45 (22)103 (18)0.275
Use of diuretic 33 (56)288 (41)0.024 83 (40)238 (43)0.604
Use of β-blocker 12 (20)127 (18)0.657 41 (20) 98 (18)0.433

Total mortality

A total of 167 persons died during the follow-up period. Forty subjects died of MI during initial hospitalization. Amongst the rest, 127 subjects the following main groups of death (ICD-9) were found: 36 (28%) neoplasms (140–239), one (1%) endocrine and metabolism disease (240–279), three (2%) mental disturbances (290–319), two (2%) nervous system diseases (320–389), 60 (47%) cardiovascular (CVD) diseases (390–459), 10 (8%) respiratory diseases (460–519), five (4%) digestive organ diseases (520–579), three (2%) urinary organ diseases (580–629), one (1%) congenital disorder (740–759) and five (4%) injuries (800–999). Of these subjects, 58 (46%) were admitted to hospital during the follow-up period because of CVD. Amongst CVDs only the diagnostic group 420–429 (other heart diseases) were associated with OH-D, as three (25%) subjects with OH-D and seven (6%) without OH-D belonged to this group; χ2 = 5.36, d.f. = 1, P = 0.0206. Of the individual diagnoses in this group, atrial fibrillation (427) was most strongly associated with OH-D; two (25%) vs. eight (7%); Fisher's exact test; P = 0.1212. OH-S was not associated with any of the CVD groups amongst the 127 subjects (P > 0.05 for each).

Amongst the persons who participated in the study (n = 833), those who could not perform orthostatic testing successfully (n = 41), had a poor prognosis, as 22 of these subjects died during the follow-up period; two (9%) neoplasms, two (9%) endocrine and metabolism diseases, one (5%) mental disturbance, two (9%) nervous system diseases, 10 (45%) CVDs, three (14%) respiratory diseases and two (9%) digestive organ diseases. Two additional persons had nonfatal MIs.

Myocardial infarctions

Altogether, 90 persons experienced a MI after a mean follow-up time of 756 (SD 446, range 14–1542) days. In 73 cases of MI, the diagnosis was based on autopsy, or standardized clinical criteria for MI. These included symptoms suggestive of MI, along with ECG findings and serial measurement of markers of myocardial damage (creatine kinase-mass and total creatine kinase activity). Only one death due to MI occurred outside the Oulu district, in the Central Hospital of Kotka, and also that MI diagnosis was certified with standard ECG and enzyme analyses.

The 40 subjects who died of MI during initial hospitalization did not differ from those who survived after MI longer (n = 50) as regards gender (P = 0.7145), OH-D (P = 0.3919) or OH-S (P = 0.3911).

Amongst the subjects with subsequent MI, OH-D was associated with diabetes mellitus; nine (60%) vs. 22 (29%) (P = 0.0282), chest pain; 12 (80%) vs. 35 (47%) (P = 0.0183), and higher age; mean 80 (SD 6) vs. 77 (SD 5) years (P = 0.0411). Correspondingly, OH-S was associated with use of nitrate medication; 15 (56%) vs. 17 (28%) (P = 0.0148) and higher systolic BP; 170 (SD 32) vs. 154 (SD 26) mmHg (P = 0.0135). Amongst the subjects without MI during the follow-up period, OH-D was associated with use of calcium antagonist; 14 (31%) vs. 107 (17%) (P = 0.0164). OH-S was associated with diabetes mellitus; 44 (24%) vs. 82 (16%) (P = 0.0160), higher systolic BP; 167 (SD 23) vs. 154 (SD 25) mmHg (P < 0.0001), higher diastolic BP; 84 (SD 11) vs. 80 (SD 12) mmHg (P < 0.0001), and lower BMI; 27 (SD 4) vs. 29 (SD 5) kg m−2 (P = 0.0006).

According to univariate Cox regression analyses, incidence of MI was associated with a history of MI: HR (95% CI) being 3.68 (2.26–6.00), diabetes mellitus; 2.31 (1.50–3.56), chest pain; 1.95 (1.29–2.95), and use of calcium antagonist; 2.01 (1.28–3.16), β-blocker; 1.92 (1.22–3.04), nitrate; 2.35 (1.52–3.63) and diuretic; 2.13 (1.39–3.26) medication. The risk of subsequent MI did not differ with regard to male sex; 1.38 (0.91–2.08), current smoking; 1.60 (0.83–3.09), high (≥80 years) age; 1.14 (0.70–1.84), high (≥31 kg m−2) BMI; 1.06 (0.53–2.13), high (≥181 mmHg) systolic BP; 1.53 (0.91–2.57) or high (≥91 mmHg) diastolic BP; 0.70 (0.35–1.39).

Risk of subsequent MI related to OH

Orthostatic diastolic BP reduction of 2 mmHg 1 min after standing up (OH-D1) was associated with subsequent MI, but the other BP reductions at 1 or 3 min were not. Consistently, when the variables representing OH were stratified according to the traditional definitions, only OH-D1 in magnitude ≥10 mmHg was associated with MI, whilst the other definitions were not (Table 2).

Table 2.  Study population with (n = 90) and without (n = 702) subsequent myocardial infarction (MI) as regards orthostatic blood pressure reductions
CharacteristicsWith MI (n = 90)Without MI (n = 702)P-valuea
  1. Numbers are mean ± SD or n (%).

  2. aUnivariate Cox regression analysis.

  3. ODBP, orthostatic diastolic blood pressure; OSBP, orthostatic systolic blood pressure.

ODBP reduction (mean ± SD mmHg at 1 min)−2 ± 10 −5 ± 110.029
ODBP reduction (mean ± SD mmHg at 3 min)−2 ± 11 −4 ± 100.121
OSBP reduction (mean ± SD mmHg at 1 min)9 ± 20 8 ± 160.541
OSBP reduction (mean ± SD mmHg at 3 min) 6 ± 20 4 ± 160.290
ODBP reduction [≥10 mmHg at 1 min, n (%)]12 (13) 33 (5)<0.001
ODBP reduction [≥10 mmHg at 3 min, n (%)] 8 (9) 38 (5)0.221
OSBP reduction [≥20 mmHg at 1 min, n (%)]22 (24)150 (21)0.504
OSBP reduction [≥20 mmHg at 3 min, n (%)]23 (26)125 (18)0.059

Risk of subsequent MI in regard to orthostatic diastolic BP change

The risk of MI did not increase linearly with greater orthostatic diastolic BP reductions, and the risk was increased only in the subjects with the most pronounced decrease in diastolic BP. The confidence intervals concerning the risk of MI as regards with these changes were wide, however (Fig. 1). In order to search for strongest cut-off of OH-D1 we performed Cox regressions at 2 mmHg intervals between the reductions of 4 and 12 mmHg. Using Bonferroni corrections for multiple testing (the level of statistical significance changed from P < 0.05 to P < 0.01) we found the following associations between the risk of subsequent MI and the reductions of OH-D1 of 4 (P = 0.062), 6 (P = 0.010), 8 (P = 0.001), 10 (P = 0.004) and 12 mmHg (P = 0.021). According to these results, OH-D1 reductions of ≥8 and ≥10 mmHg were associated with subsequent MI. The strongest association with MI as regards OH-D1 was found at ≥8 mmHg (OH-D18) (n = 62). The corresponding HR (95% CI) of MI as regards OH-D18 was 2.91 (1.69–4.99). The Kaplan–Meier curves for subsequent MI as regards presence and absence of OH-D18 are described in Fig. 2.

Figure 1.

Relative risks (hazard ratio) of myocardial infarction (MI) according to increasing diastolic blood pressure reductions 1 min after standing up.

Figure 2.

Kaplan–Meier plots on the probability of remaining free of myocardial infarction according to diastolic blood pressure fall of ≥8 mmHg (yes) and <8 mmHg (no) 1 min after standing up during a 4-year follow-up period.

Multivariate model results regarding the risk of MI

A multivariate Cox proportional hazards modelling, with all the statistically significant associates of MI entered in the model, showed that OH-D18, diabetes mellitus, and history of MI were predictors of MI (Table 3). The main result concerning the prediction of MI as regards OH-D18 did not change, when the nonsignificant covariates male sex, current smoking, high BMI, and high systolic BP were also entered in the model; HR (95% CI) being 1.82 (1.002–3.30). Further, we constructed interaction terms between OH-D18 and each covariate, and could not find any statistically significant interactions in association with subsequent MI (95% CI of HR crossing 1 in each case), when interactions terms, along with the individual terms, where entered in Cox proportional hazard analyses.

Table 3.  Risk factors of myocardial infarction in the elderly expressed as adjusteda hazard ratios (HR, 95% CI)
Risk factor (unit)CoefficientSEHRb (95% CI)
  1. aAdjusted for all variables in the Table.

  2. bCalculated against the unit expressed for each risk factor.

  3. OH-D18, diastolic blood pressure reduction of ≥8 mmHg 1 min after standing up.

OH-D18 (0, 1 = yes)0.690.302.00 (1.11–3.59)
History of myocardial infarction (0, 1 = yes)0.880.292.40 (1.37–4.22)
Diabetes mellitus (0, 1 = yes)0.560.231.75 (1.11–2.78)
Chest pain (0, 1 = yes) (0.70–1.81)
Use of calcium antagonist (0, 1 = yes) (0.75–2.04)
Use of β-blocker (0, 1 = yes) (0.79–2.12)
Use of nitrate (0, 1 = yes)0.350.271.41 (0.83–2.40)
Use of diuretic (0, 1 = yes)0.400.231.49 (0.95–2.34)

When we repeated the Cox modelling after exclusion of the 17 persons without autopsy or ECG confirmation of MI, the main result concerning the risk of MI as regards OH-D18 did not change, the adjusted HR (95% CI) being 2.20 (1.15–4.20).

Results of subgroup analyses

We examined the effect of severity of underlying disorders on the risk of MI associated with OH-D18. In persons without chest pain, in those experiencing chest pain once a week, and in those experiencing chest pain more frequently, we found HRs (95% CI) of 1.45 (0.45–4.68), 4.07 (1.83–9.06) and 2.52 (0.85–7.50), respectively. In subjects without diabetes, those with a diet-treated, and those with a tablet- or insulin-treated diabetes the respective figures were 1.67 (0.72–3.89), 8.00 (2.13–30.1) and 4.76 (1.71–13.3). According to increasing quartiles of systolic BP, the corresponding figures were 4.68 (1.77–12.3), 6.61 (2.29–19.0), 2.03 (0.60–6.95) and 1.30 (0.39–4.33).

Mortality risk associated with OH-D18

Twenty-three persons (37%) with prevalent and 144 (20%) with absent OH-D18 died during the follow-up period, the HR (95% CI) of death as regards OH-D18 being 2.10 (1.35–3.25). After exclusion of the 40 persons who had MI close to death, mortality between those with prevalent and those with absent OH-D18 did not differ; 11 (22%) vs. 97 (14%), the relative risk of death in this case being 1.65 (0.88–3.08).


Our results show that orthostatic testing offers a novel method to assess the risk of MI amongst elderly home-dwelling persons, in whom the test can be successfully performed. We found that failure to perform the orthostatic test was associated with very high mortality. Earlier investigations have stressed the importance of postural symptoms [4, 10], which could have contributed to failure to perform the orthostatic test in the present study.

Even if the subject is able to stand for 1 min during the orthostatic test, adequate performance of the test is difficult. The time of day, and meals and activities relative to the time of orthostatic testing may contribute to the results. The frequency of ≥20 mmHg or greater systolic OH is greatest in the morning, before breakfast. Early morning is the time when orthostatic testing should be, primarily, performed [11]. However, systolic OH, in particular, is highly variable over time [11]. High variability of systolic OH raises the risk of stroke in elderly nursing home residents, calling for repeated measurements [9], which, regrettably, were not done in the present study. Although variability of diastolic orthostatic BP change is less than that of systolic orthostatic BP change [11], we currently do not know the importance of the variability of diastolic OH. Nevertheless, a drop in diastolic BP immediately after standing up appears to identify elderly subjects at high risk of subsequent MI, with strongest such a prediction found when diastolic BP falls by ≥8 mmHg. We found also increased mortality risk associated with OH-D18, but MIs appeared to explain the excess mortality.

Orthostasis, standing upright, imposes a major stress on the cardiovascular system, which requires effective adjustments to ensure that the arterial pressure remains at a level adequate for perfusion of the vital organs. Usually, systolic BP is not changed or decreases slightly. Diastolic BP tends to increase in the upright posture, at least in younger subjects [4, 10, 12, 13]. In older subjects, however, there tends to be a substantial reduction in the supine cardiac volumes and early diastolic filling rates, which may predispose to OH [14]. The operation of arterial baroreflex is crucial in the maintenance of adequate BP [15]. Unloading of baroreceptors during orthostasis results in an increase in total peripheral resistance through a sympathetic response. Baroreflex also mediates the immediate increase in heart rate that compensates the reduction in stroke volume [12] in order to maintain cardiac output. Subjects with diabetic neuropathy [16] as well as those with OH [5, 6] have diminished heart rate changes after active rising. Impaired baroreflex increases total and cardiovascular mortality in persons with prior MI [17, 18], and it offers a hypothetical explanation why diabetes and diastolic OH were strongly associated just in patients with subsequent MI. Curiously, high systolic BP was only associated with systolic OH, and did not affect the risk of MI associated with diastolic OH. High systolic BP is known to decrease baroreflex sensitivity in older subjects, regardless of the type of systolic hypertension [19].

Interestingly, only the impaired immediate adjustments of diastolic, but not systolic, BP were related to the risk of MI. This may be related to the crucial role of preserved diastolic perfusion pressure on myocardial blood flow. Autoregulation maintains coronary flow nearly constant regardless of large reductions in perfusion pressures [20]. In the present study, the diastolic orthostatic BP reduction required to increase the risk of MI seemed high. Yet, studies with greater sample sizes are needed to verify more accurately the critical values of diastolic OH likely to increase the risk of MI. Not surprisingly, symptoms of chest pain were associated with diastolic OH in subjects with subsequent MI.

Some concerns related to registration of cases of MI during this study exist. First, we may have lost some silent MIs, which frequently occur in older adults, and have prognosis similar to that described for recognized infarctions [21, 22]. Secondly, although the treatment of all symptomatic MIs occurring in the present study area takes place at the Oulu University Hospital, some persons might have suffered nonfatal MIs whilst travelling in other districts. Thirdly, the ability of death certificates to accurately define cause of death amongst elderly patients is highly questionable [23]. However, removal of the cases with no autopsy or ECG verification did not change our main result, suggesting that the results were not biased by diagnostic errors.

We did not have comprehensive evaluation of the serum lipids which contribute to the risk of MI. Serum lipid concentrations are not associated with OH [7], but unfavourable lipid profiles predict CVD morbidity and mortality even in advanced ages [24]. Therefore, it is possible that lack of lipid data confounded our main result. One important limitation as regards clinical implications is that the reproducibility of the orthostatic test results is not known in home-dwelling older adults [11].

When adequately performed, orthostatic testing may provide a cheap clinical measure for the assessment of the risk of MI amongst the elderly. In a representative sample of older persons, a nearly threefold increase in the risk of MI can be observed by using diastolic BP reduction after active rising as a predictive measure. Patients with diabetes and coronary artery disease are at particularly high risk. Prevention of diabetes and adequate control of coronary symptoms may reduce the risk of MI associated with diastolic OH in older subjects.

Conflict of interest statement

No conflict of interest was declared.


This study was supported in part by the Department of Health and Security of Finland, the Oulu University Hospital (Oulu, Finland) and Tampere University Hospital (Tampere, Finland) Medical Research Funds.