Blood pressure and ageing: longitudinal cohort study

BMJ 1998; 317 doi: (Published 22 August 1998) Cite this as: BMJ 1998;317:513
  1. John M Starr (John.Starr{at}, senior lecturera,
  2. Susan Inch, research nursea,
  3. Susan Cross, research nursea,
  4. William J MacLennan, professora,
  5. Ian J Deary, professorb
  1. aGeriatric Medicine Unit, University of Edinburgh EH3 9EW
  2. bDepartment of Psychology, University of Edinburgh EH8 9JZ
  1. Correspondence to: Dr John M Starr, Royal Victoria Hospital, Edinburgh EH4 2DN
  • Accepted 26 March 1998

Arterial blood pressure rises throughout most of life in industrialised societies.1 In old age, however, blood pressure —especially diastolic— >falls.1 This may be because people who are included in epidemiological studies have diseases or take drugs that affect blood pressure. As greater disease free life expectancy is predicted, the prevalence of hypertension in the elderly may rise if a continued blood pressure increase occurs in those who remain healthy. The few studies examining this provide no consensus. 2,3 We investigated whether blood pressure changes in old age relate to health.

Subjects, methods, and results

We visited 602 (237 men, 365 women) untreated, healthy subjects (mean age 75.7 years, range 70-88 years) at home.4Educational attainment was recorded and occupation classed as standard occupational classification groups 1-4 (white collar occupations) and groups 5-9 (blue collar occupations).5 Blood pressure was measured after the subjects had rested for 25-30 minutes.(A longer version of the methods appears on our web site.)

At follow up (median period 4.20 years, range 3.23-5.23 years), 69 subjects (11.5%) had died and 105 (17.4%) were unavailable.4 Health status (documented and self reported) was recorded for the remaining 426 subjects (70.8%) and blood pressure remeasured at home. Measurements were of questionable reliability in 41 subjects (9.6%) for reasons such as interruptions. Of the remaining 385 subjects (90.4%), men had significantly lower systolic (p=0.01) and diastolic (p=0.017) pressures than women. Occupation significantly influenced diastolic (p=0.009), but not systolic (p=0.68), pressure. Education correlated neither with systolic (p=0.44) nor diastolic (p=0.44) pressure. Age correlated negatively with diastolic −align=baseline>0.124, p=0.01), but not with systolic (r=− align=baseline>0.039, p=0.2), pressure.

At follow up, 195 subjects (50.6%; 69 men, 126 women) remained healthy and untreated. Correlations between baseline and follow up for systolic and diastolic pressure were 0.49 and 0.24 respectively. When baseline blood pressure was controlled for, neither age nor education significantly affected blood pressure at follow up. When systolic and diastolic pressures were combined in a multivariate repeated measures analysis of variance, sex had no significant effect (p=0.43), but a significant effect was shown for occupation (p=0.021). A consistent pattern of rises in blood pressure occurred across occupational groups (table) except in groups 8 and 9. When subjects were stratified by occupation—white collar (n=119) and blue collar (n=66)—a significant rise in blood pressure occurred in white collar (p<0.001), but not blue collar (p=0.08), workers.

Incident disease (excluding 29 subjects with hypertension only) was reported or recorded in 163 subjects (42.3%; 76 men, 87 women). These subjects were older (p=0.005) than those who remained healthy and more commonly men (p=0.021), but there were no significant differences in education (p=0.40) or occupation (p=0.067). Mean blood pressure remained almost unchanged (table). No significant effects were detected for sex (p=0.46) or occupation (p=0.37) over time. Compared with healthy subjects there was a significant change in systolic pressure (p=0.003) and diastolic pressure (p=0.016) over time, corrected for age.


In the elderly, changes in blood pressure relate to health. In those who remained healthy blood pressure continued to rise by 1.5 mm Hg per year but fell in those who developed disease (representing probable heterogeneous effects of different medical conditions). In the healthy group no rise was seen in retired blue collar workers. possibly, retired manual workers have more unrecognised disease, consistent with a persistent effect of socioeconomic factors on health status in old age. For healthy individuals the most important predictor of blood pressure at follow up was baseline blood pressure, accounting for 24% of the variance of systolic, but only 6% of the variance of diastolic, pressure. In general, disease affected systolic more than diastolic pressure. Hence the poorer correlation between baseline and follow up diastolic pressure compared with systolic pressure is not easily explained by possible undetected disease in the healthy group and deserves further investigation.

Systolic and diastolic arterial blood pressure by standard occupational group at baseline and four years' follow up of healthy, untreated elderly people (mean age at baseline 74.4 years, mean education 11.2 years) compared with those with incident disease (mean age at baseline 75.8 years, mean education 10.9 years) or hypertension only (mean age at baseline 75.6 years, mean education 10.6 years). Values are mean (SD) unless stated otherwise

View this table:

Distribution of 134 cases and 134 controls according to selected characteristics and by matched sets

View this table:


We thank the patients and general practitioners.

Contributors: JMS designed and supervised the study, collected and analysed the data, interpreted the results, and reported the findings; he will act as guarantor for the paper. WJMacL and IJD designed and supervised the study, interpreted the results, and reported the findings. SI and SC collected the data, interpreted the results, and reported the findings.


  • Funding This study was supported by a grant from the chief scientist's office, Scotland.

  • Conflict of interest None.


View Abstract