High blood pressure has been clearly associated with an increased risk of negative cardiovascular outcomes. This is the case for both acute events such as a stroke, and slower changes in the vasculature such as left ventricular hypertrophy and increasing arterial stiffness . It is also well known that SBP tends to rise with increasing age, at least in most populations, whereas DBP tends to rise around age 50 and then fall . Over time, the focus with regard to risk factor classification and intervention has varied between SBP, DBP and pulse pressure (defined as the difference between SBP and DBP). Recent guidelines (2011) in the UK have provided target pressures for SBP and DBP but have also recommended the use of ambulatory blood pressure monitoring devices to allow measurement of blood pressure over 24 h (http://www.nice.org.uk/CG034). Although the diagnosis of hypertension has traditionally been made based on more than a single recording, this use of 24 h data also allows treatment choice to be made without being unduly influenced by ‘one-off’ high readings such as, for example those due to the phenomenon of ‘white-coat’ hypertension. White-coat hypertension has been defined as a clinic SBP at least 140 mmHg and/or a DBP at least 90 mmHg alongside an ambulatory pressure of less than 135/85 during activity . White-coat hypertension has also been associated with an increased risk of cardiovascular events and left ventricular hypertrophy, although not all studies agree [2,3]. More recently, blood pressure variability has been shown to be associated with an increased risk of stroke and cardiovascular events . Both the data relating to white-coat hypertension and that relating to blood pressure variability suggest that transient high blood pressure measures seen in general practice may not be benign. Despite this, long-term follow-up of those with transiently elevated blood pressure readings is rare and, at present, such transiently elevated recordings observed in general practice would not usually result in treatment. The General Practice Hypertension Study Group (GPHSG) was a study set up within general practice during the early 1970s measuring baseline blood pressures in 1974 [5–7]. Data were gathered with regard to those who had sustained diastolic hypertension, those who had transient diastolic hypertension and those who were normotensive. The primary focus at the time that the original study was set up was DBP; however, SBP data were also collected. Following the initial data collection, participants were registered with the Office of National Statistics and death certificate data collected until 2009. This allows an exploration of the possible effects of transiently elevated DBP, the relationship between DBP and SBP and cause of death over 29 years of follow-up in the original enrolled group.
The first analyses of these data were reported in 1992 after an average of 11 years of follow-up. In these analyses, a higher risk of later cardiovascular mortality was seen in those who were newly diagnosed with diastolic hypertension (both men and women). There were no significant relationships between transient diastolic hypertension and later cardiovascular mortality, although the point estimates were greater than 1 for men and less than 1 for women when compared with normotensive controls . An additional analysis after 18.7 years of follow-up reported that women with transient diastolic hypertension had significantly lower risk of cardiovascular mortality than controls. In men, the point estimate remained above 1, but the result missed significance .
Details of the GPHSG objectives and screening methods have been published previously [5–7]. Over 28 000 individuals aged between 18 and 65 years were screened from eight general practices. The practice population included were of rural areas (Pontesbury, Eaton Socon), provincial towns (Rugby, Ipswich, Darlington) and suburban and inner city areas (East Kilbride, Eccles, Southall). The participants were identified from the age-sex registers and were either invited by letter, identified from special screening clinics, or their notes were tagged to allow for screening when attending the practice for other reasons.
The screening was performed by practice nurses who had been trained to use random zero mercury sphygmomanometers. SBP and DBP (4th and 5th sound) were recorded after 5 min of resting in the sitting position. The DBP was used to define whether an individual should be considered hypertensive. For these analyses, the Korotkoff 4th sound reading was employed, as this was the reading used originally to define an individual as hypertensive. Data were also collected on smoking, height and weight.
Those with a DBP of less than 90 mmHg were considered to be normotensive. The majority (85%) of those screened fell into this category and were not recalled for further measurements. Those with DBPs of 90–129 mmHg were recalled on two further occasions at least 1 month apart. On the basis of these further readings, they were either classified as ‘new hypertensive patients’ (DBP of 90 mmHg or more on at least one subsequent visit) or ‘transient hypertensive patients’ (DBP less than 90 mmHg at both subsequent visits). Those already identified as being hypertensive on practice records were classified as ‘known hypertensive patients’. Any individual with a DBP of 130 mmHg or more at any time was referred on to hospital. The results of the screening programme have been previously published . Those classified as hypertensive or as having transient hypertension were age-matched and sex-matched to normotensive controls.
Seven of the original eight practices registered their participants with the Office of National Statistics (ONS) and copies of death certificates continued to be received until 2009. Causes of death were coded using the International Classification of Disease versions 9 and 10. These were additionally broadly categorized by cause, cancer, respiratory, cardiovascular, and so on: some overlap is possible in cases in which more than one disease area was given.
The relationship between baseline data and mortality was examined using Cox proportional hazard regression stratified by age and sex in order to ensure assumptions were not violated. Wherein sex was examined separately, stratification for age was used. The final censor date was taken as either the date of death, or, for those for whom no death certificate had been received as the most recent death (i.e. the most recent date of death in the dataset was used as the last date at which data were available from ONS and the cut off point for follow-up in those wherein no death was reported). If no death certificate was received, the participant was assumed to be alive at the given date. As the initial classification was based on DBP, further analyses were carried out with the transient hypertensive population after removing those participants who according to more recent guidelines would have been treated based on their SBPs (>140 mmHg) as recorded on the same occasion as their transiently high DBP. That is, to try to examine ‘pure’ transient diastolic hypertensive patients, those without high SBPs were selected and compared to their matched controls. Baseline data from the practice that did not register participants with ONS were not used in the analyses.
Over 20 000 patients were screened initially and final numbers included in the study (registered for collection of later death certificate data and available for analysis) were 850 with transient diastolic hypertension, 2342 with diastolic hypertension, (556 who were known to have diastolic hypertension and 1786 who were newly diagnosed as hypertensive) and 3117 normotensive individuals age-matched and sex-matched to those with diastolic hypertension and transient diastolic hypertension.
Baseline characteristics are shown in Tables 1–3, which detail the characteristics by blood pressure type and location of practice (rural, provincial town or inner city). As may be expected, those with different blood pressure types had different blood pressures; those with previously known diastolic hypertension were also more likely to be older, men and have higher BMI values. Those with transient diastolic hypertension had blood pressures that fell between the diastolic hypertensive and normotensive groups. All three practice location groups had broadly similar characteristics. Death certificate data were collected until 2009 with the most recent date of death 22 June 2009 giving a mean of 29 years since baseline. There were 1681 deaths overall, 908 men and 773 women. There were 6309 participants with sufficient data to be included in the survival analyses. Average age at time of entry was 47.9 years and age at time of death for those who died was 75.9 years (SD 10.2). Mean age at death in the normotensive group was 75.1 (SD 9.7) years for men and 79.3 (9.4) for women, for those with transient hypertension 72.1 (11.1) and 76.4 (9.7) and hypertension 72.9 (10) and 77.7 (10.4), respectively.
Cause of death was most commonly cardiovascular, recorded on 47.2% of male and 45.3% of female death certificates, followed by cancer at 34.3% men and 29.5% women and respiratory causes 28.2 and 29.4%, respectively. Twelve death certificates gave old age as the only cause of death and were not able to be ascribed to a disease area. There was a relationship between diastolic hypertension and total mortality, cardiovascular-related mortality but not cancer-related mortality. Compared to the normotensive group, having diastolic hypertension at baseline meant a 17% increased risk of all-cause mortality and a 34% increased risk of cardiovascular-related death. See Table 4 for details showing the adjusted results. Having transient diastolic hypertension at baseline was associated with a significantly increased risk of cardiovascular-related death (see Table 4). The category of cardiovascular death was rerun excluding mention of diabetes; however, this did not materially change the results.
The data were further examined by sex. No significant relationships remained for men for either diastolic hypertension or transient diastolic hypertension for any outcome (all-cause mortality, cardiovascular-related mortality or cancer-related death). In contrast, all relationships were significant in women for all-cause and cardiovascular-related mortality such that both diastolic hypertension and transient diastolic hypertension were associated with increased risk. Transient diastolic hypertension also just reached significance as did cancer-related mortality. Known diastolic hypertensive patients and new diastolic hypertensive patients were combined in these analyses; however, rerunning the analyses separately resulted in the same pattern of relationships for both classifications.
Additional examination of the data showed that 10.9% of men and 8.6% of women with transiently elevated DBP at baseline would also be likely to be classified as having systolic hypertension by current guidelines (≥140 mmHg) and so would currently probably be treated. The remaining 89.1% of men and 91.4% of women were classified as having ‘true’ transient diastolic hypertension (i.e. true transients even by current classification) and were compared to their matched controls. Wherein full baseline data were available for both case and control (n = 1188), comparison of these true diastolic transients with their controls resulted in similar relationships. Women showed a just significantly increased risk of cardiovascular mortality [hazard ratio 2.74 (95% confidence interval, CI 1.003–7.5) P = 0.049] and men a nonsignificant relationship [hazard ratio 1.56 (0.71–3.42)] both stratified by age.
These data suggest that transiently high readings of DBP in midlife may be related to an increased risk of later mortality and cardiovascular mortality, particularly in women. Although the classifications used in the original study were based on DBP, similar relationships with later mortality were seen when excluding those who, in accordance with today's guidelines, may be treated based on their SBP. This may suggest that a transiently elevated DBP recording is not benign. It may be that those with transiently elevated DBP may have been more likely to subsequently manifest either white-coat or even masked hypertension, they may also have had blood pressures that were in general tracking upwards towards a diagnosis of hypertension or were those who had more variable blood pressures all of which would potentially place them at higher risk of subsequent events [4,8–10]. This cannot be explored further in the current dataset, and it is also likely that participants went on to receive different treatments or practice different lifestyle behaviours, over subsequent years; however, despite this, these relationships retain significance in women. The difference between men and women is interesting and may indicate a greater propensity to treat men at the time and possibly subsequently. Certainly, this bias has been suggested by at least one other analysis in which an excess of readings of 88 mmHg DBP was noted, that is, a reading just below the threshold for the classification of hypertension. In women, this was associated with a corresponding increase in cardiovascular and total mortality at DBP 88–89 mmHg compared to DBP of 90–99 mmHg, whereas for men, the corresponding mortality rates were lower at 88–89 than 90–99 mmHg . The finding that women are at higher risk than their male counterparts after a mean follow-up of 29 years is contrary to results reported from earlier follow-up analyses in this population. Previous analyses after mean follow-up periods of 11 and 18.7 years found men to be at increased risk. Previous analyses also focused on the comparison between newly diagnosed diastolic hypertensive patients, transient diastolic hypertensive patients and normotensive individuals rather than those known to have diastolic hypertension. Our results are for combined diastolic hypertension categories (known diastolic hypertensive patients and newly diagnosed diastolic hypertensive patients), although when these were rerun separately, the pattern and direction of relationships remained the same. There may be sex differences in the impact of risk factors over time with differing critical periods or levels of impact. There has been some suggestion of this in the literature looking at antihypertensive treatment and blood pressure trajectory in relation to later dementia, but the literature is not yet comprehensive . In addition to this, a large ambulatory blood pressure monitoring study with over 9000 participants reported that although absolute risk was lower in women, the proportion of preventable cardiovascular complications was greater. Women showed a steeper relationship between blood pressure and cardiovascular outcomes . In the study reported here, it may also be that the men who were older at recruitment were at higher cardiovascular risk than the women of the same age, whereas the younger women were at higher risk or at least closer in risk to their male counterparts. Men may have been at generally higher cardiovascular risk at baseline with those at highest risk dying soonest leaving a survivor group. Women may have been at lower baseline cardiovascular risk overall and although their risk increased over time, those at highest risk and dying soonest may have been at more advanced age than their male counterparts. In each case, high DBP at baseline may identify a higher risk group, but in women the association took longer to occur.
Those men who survived longest may also have actively reduced their risk behaviours and/or been better controlled during the intervening years either in spite of or because of their hypertension as compared to men who died earlier during follow-up, or as compared to women.
Younger women may have had or developed behaviour patterns such as increasing rates of smoking or alcohol consumption that increased their cardiovascular and cancer risk over time. It may also be that women may take longer to manifest the results of the increased risk having benefited from additional cardiovascular protection prior to menopause. The association between transient diastolic hypertension in women and risk of cancer-related mortality [hazard ratio 1.51 (95% CI 1.01–2.27)] is more surprising. The direction of the relationship in women is the same for the sustained diastolic hypertensive patients, though not significant [hazard ratio 1.30 (95% CI 0.98–1.72)] and there is no suggestion of a relationship in men. A link between elevated SBP and DBP and increased cancer incidence and mortality has been suggested previously , although not universally confirmed or sex-specific . If it was the case that women had their blood pressure treated less systematically than men, it may also be that they had less contact with healthcare professionals and were diagnosed at later stages, or that those with transiently elevated DBPs had higher levels of predisposing factors that raised risk of both cardiovascular and cancer outcomes, for example alcohol consumption, smoking, and obesity. Although the smoking behaviour and BMI of participants were available at baseline and are adjusted for in these analyses, further data relating to longer term usage or weight gain and loss are not available.
Unfortunately, no further data are available with regard to other relevant risk factors or subsequent risk behaviour or treatment decisions over the years and these may have had a greater influence on longer term outcome. Other limitations include the classification of participants by DBP rather than the currently used SBP. This was consistent with current thinking at the time of study set up. It may also retain relevance, given the age of participants at entry. The use of death data also has inherent flaws not only in terms of accuracy but also with regard to change in diagnostic criteria over time and cultural biases, for example recording dementia on a death certificate. It may also be that death data were not accurately captured in some cases; however, this is the best understanding that we have given both the historical nature of the study and the mechanisms available. Strengths of the study include the length of follow-up and the baseline matching of cases and controls.
Overall, these data alongside those reported previously for this study suggest that it may be worth monitoring blood pressure at least in those with transiently elevated DBP readings in order to fully understand what is happening to their blood pressure over time and to target treatment most appropriately. Transient diastolic hypertension does not seem to be benign, particularly in women and may point to those at higher cardiovascular risk when seen in a general practice population.
The authors would like to acknowledge the general practitioners who made the original study possible and all the staff at Imperial College London who helped to collate and enter the death certificate data over the years.
No funding has been received for these analyses.
Conflicts of interest
There are no conflicts of interest.
Reviewers’ Summary Evaluations Reviewer 1
This study examines the impact of transient and sustained hypertension on later mortality. The follow-up is long, very long, may be too long (mean = 29 years) considering the fact that there is no information available on antihypertensive treatment and blood pressure control during the interval. Also, data on additional risk factors are provided at inclusion in the study, but no information is given thereafter. The co-existence of other risk factors might obviously greatly influence the cardiovascular outcome, may be more than a punctual diagnosis of transient or sustained hypertension based on office blood pressure readings.
The major strength of the study is the long-term follow-up. Furthermore, the cases and controls were matched at baseline. There are several potential limitations, but they were correctly identified by the authors of the manuscript. First, the participants were classified by DBP rather than the currently used SBP. Second, the authors were not able to collect data regarding treatment decisions and changes in risk factors during the follow-up. Third, death-related diagnostic criteria might have changed over time. These inherent flaws are counterbalanced by the strengths of the study. Overall, these findings suggest that truly long-term studies might be needed to explore complexity of cardiovascular risk, particularly in females.
This paper highlights the importance of elevations in DBP, both sustained and transient, as a predictor of cardiovascular morbidity and mortality. Its strength lies in the large sample size of patients drawn from a community screening program and in very long-term follow-up. Its main limitation is that the association with cardiovascular risk is only significant in women, though it does trend in the same direction in men.
1. US Department of Health and Human Services. The seventh report of the Joint National Committee on Prevention Detection Evaluation and Treatment of High Blood Pressure. Bethesda, MD: NIH; NIH Publication No. 04-5230; August 2004.
2. Purchades R, Ruiz-Nodar J, Blanco F, Rodríguez F, Gabrriel R, Suárez C. White-coat hypertension in the elderly. Echocardiographic analysis. A substudy of the EPICARDIAN Project. Rev Esp Cardiol 2010; 63:1377–1381.
3. Pierdomenico S, Cuccurullo F. Prognostic value of white coat and masked hypertension diagnosed by ambulatory monitoring in initially untreated subjects: an updated meta-analysis. Am J Hypertens 2011; 24:52–58.
4. Rothwell P, Howard S, Dolan E, O’Brien E, Donson J, Dahlöf B, et al. Prognostic significance of visit-to-visit variability, maximum systolic blood pressure, and episodic hypertension. Lancet 2010; 375:895–905.
5. Fletcher A, Bradley C, Broxton J, Bulpitt C, Davis A, Dyker G, et al. Survival of hypertensive subjects identified on screening: results for sustained and unsustained diastolic hypertension. Eur Heart J 1992; 13:1595–1601.
6. Wingfield D, Grodzicki T, Palmer A, Wells F, Bulpitt C. General Practice Hypertension Study GroupTransiently elevated diastolic blood pressure is associated with a gender-dependent effect on cardiovascular risk. J Hum Hypertens 2005; 19:347–354.
7. Mayhew S. Hypertension screening in general practice. Report on behalf of the General Practitioner Hypertension Study Group. J R Coll Gen Pract 1983; 33:434–437.
8. Wingfield D, Freeman G, Bulpitt C. On behalf of the General Practice Hypertension Study Group (GPHSG). Selective recording in blood pressure readings may increase subsequent mortality. Q J Med 2002; 95:571–577.
9. Franklin S, Thijs L, Hansen T, Boggia J, Kikuya M, Björklund-Bodegård K, et al. on behalf of the International Database on Ambulatory Blood Pressure in relation to Cardiovascular Outcomes InvestigatorsSignificance of white-coat hypertension in older persons with isolated systolic hypertension: a meta-analysis using the international database on ambulatory blood pressure monitoring in relation to cardiovascular outcomes population. Hypertension 2012; 59:564–571.
10. Verbeck W, Kessels A, Leeuw P. Prevalence, causes and consequences of masked hypertension: a meta-analysis. Am J Hypertens 2008; 21:969–975.
11. Hansen T, Kikuya T, Thijs L, Li Y, Boggia J, Björklund-Bodegård K, et al. Diagnostic thresholds for ambulatory blood pressure moving lower: a review based on a meta-analysis-clinical implications. J Clin Hypertens 2008; 10:377–381.
12. Joas E, Bäckman K, Gustafson D, Ostling S, Waern M, Guo X, Skoog I. Blood pressure trajectories from midlife to late life in relation to dementia in women followed for 37 years. Hypertension 2012; 59:796–801.
13. Boggia J, Thijs L, Hansen T, Li Y, Kikuya M, Björklund-Bodegård K, et al. Ambulatory blood pressure monitoring in 9357 subjects from 11 populations highlights missed opportunities for cardiovascular prevention in women. Hypertension 2011; 57:397–405.
14. Stocks Y, Hemelrijck M, Manjer J, Bjørge T, Ulmer H, Hallmans G, et al. Blood pressure and risk of cancer incidence and mortality in the Metabolic Syndrome and Cancer Project. Hypertension 2012; 59:802–810.
15. Batty G, Shipley M, Davey Smith G. Blood pressure and site-specific cancer mortality: evidence from the original Whitehall study. Br J Cancer 2003; 89:1243–1247.