Blood lead and blood pressure: evidence from the Health ... - Nature

Journal of Human Hypertension (1999) 13, 123–128  1999 Stockton Press. All rights reserved 0950-9240/99 $12.00 http://www.stockton-press.co.uk/jhh

ORIGINAL ARTICLE

Blood lead and blood pressure: evidence from the Health Survey for England 1995 L Bost1, P Primatesta1, W Dong1 and N Poulter2 1

Department of Epidemiology and Public Health, University College London Medical School, London WC1E 6BT; 2Cardiovascular Studies Unit, Imperial College School of Medicine, St Mary’s Campus, London W2 1PG, UK

Objectives: To examine the relationship between blood lead and blood pressure (BP) and to estimate the possible effects of a decrease in blood lead on BP. Methods: A 2-ml blood sample was collected from a sub-sample of those included in the Health Survey for England 1995, a cross-sectional survey of a nationally representative sample of the adult English population. Blood lead concentration was measured by atomic absorption spectrometry and three BP readings were taken under standardised conditions using the Dinamap 8100 monitor. Analyses were carried out using data on 2563 men and 2763 women aged 16 and over. Results: In stepwise multiple regression analyses adjusting for various confounders—age, body mass index, smoking status, social class, region of residence

and alcohol intake—blood lead was found to be significantly and positively associated with diastolic BP, and not systolic BP in men, but not in women. These findings were unaffected by the inclusion or exclusion of those on antihypertensive medication, by whether mean or median BP was used in the regression, or by the adjustment for alcohol consumption. A halving of currently prevalent blood lead levels is estimated to be associated with a decrease of between 0.8 to 1.1 mm Hg diastolic BP in men. Conclusion: These findings in the context of other published data are consistent with a small pressor effect of environmental lead levels on BP. They support recommendations for further efforts to reduce lead in the environment.

Keywords: blood lead; blood pressure; pressor effect

Introduction A link between high-level lead exposure and adverse cardiovascular and renal outcomes was recognised in the 19th century.1 More recently the possibility that chronic low-level lead exposure may induce small but potentially important increases in blood pressure (BP) has been the subject of much investigation, prompted by the decreasing environmental lead levels observed in the developed world.2 In 1988 a consensus meeting3 concluded that the weight of evidence established an association between lead exposure, usually measured as blood lead concentration, and increased BP, a view which was supported by a more recent National Academy of Sciences panel in the USA.4 A comprehensive review of relevant studies published between 1980 and 19925 and a subsequent meta-analysis by Schwartz of studies conducted in men6 also concluded that although the data were not definitive, the association was likely to be causal. However the author of one meta-analysis published in 1994 which reviewed 23 studies,7 many of them included in Schwartz’s meta-analysis,6 felt that the data did Correspondence: Dr P Primatesta, Epidemiology and Public Health, University College London Medical School, 1–19 Torrington Place, London WCIE 6BT, UK Received 30 July 1998; revised 20 August 1998; accepted 22 October 1998

not support the hypothesis of causation. Similarly, studies carried out on subgroups of the British population led to contradictory results.8–11 The Health Survey for England (HSE), an annual nationwide survey that monitors various aspects of the nation’s health, constituted a good opportunity to update information on lead levels in the general population and explore the possible links between blood lead and BP. This was achieved using a subsample of the HSE 1995.

Subjects and methods The Health Survey for England 1995 A detailed description of the HSE 1995 can be found elsewhere.12 In brief, the HSE 1995 is the fifth annual nationwide survey commissioned by the Department of Health, including data on various aspects of the nation’s health in a representative sample of the English population drawn from those living in private households. A multi-stage stratified probability sampling design was adopted and the sample designed so that the fieldwork conducted in each quarter of the year was carried out with a fully representative sub-set of the whole. Data were collected by an interviewer who administered a questionnaire and measured height and weight, and subsequently by a nurse who measured BP, recorded current use of prescribed medicines and took a blood sample from consenting respon-

Blood lead and blood pressure L Bost et al

124

dents. BP was measured using an automated device, the Dinamap 8100 monitor.13 Three BP readings were taken under standardised conditions on the right arm with the informant in a seated position after 5 min of rest. The data used in this study are based on the mean of the second and third measurements made on respondents who had three recordings completed. Blood lead levels were measured on a sub-sample of respondents, participating during April–June and September–December 1995; blood samples from a total of 6517 adults aged 16 and over (3119 men, 3398 women) were analysed. A 2 ml blood sample was collected into an EDTA (ethylene diamine tetra-acetic acid) tube and lead concentration was measured by atomic absorption spectrometry using micro-sampling flame atomisation.14 Statistics The relationship of systolic BP (SBP) and diastolic BP (DBP) with blood lead was initially determined using Pearson correlation coefficients. This was followed by multiple linear regression analyses which included various factors that have been shown to be associated with BP in the HSE:15,16 age, body mass index (BMI) defined as weight (kg)/height2 (m2), smoking status (non-current/current smoker), region of residence (North/Midlands/South), social class based on the occupation of the household head (manual/non-manual), and alcohol consumption (non/ex/occasional/moderate/heavy drinker). Occasional, moderate and heavy drinking were defined as ⬍1 unit/week, 1–21 units/week (men) or 1–14 units/week (women), and ⬎21 (men) or ⬎14 (women) respectively. Valid data for all the above variables were available on 2563 men and 2763 women. SBP, blood lead and BMI were log-transformed and described by the geometric mean. Separate regression analyses were carried out with and without adjustment for alcohol intake, and including and excluding those on antihypertensive medication (314 men and 369 women). In the analyses which included those on antihypertensive medication, two types of regression—‘ordinary’ (using mean BP) and ‘median’ (using median BP)—were conducted. The latter approach finds a line through the data that minimise the sum of the residuals rather than the sum of the squares of the residuals as in ordinary regression.17 In the median regression analyses, it was assumed that those on medication had underlying BP in the upper half of the distribution and were assigned BP values above the median. Regression models were fitted separately for SBP, DBP and pulse pressure (SBP–DBP), and for men and women in a stepwise process.

Results Selected characteristics of the men and women in the study are shown in Table 1. The regional distribution in our sample approaches the mid-1995 population estimates, although the proportion residing in the

Table 1 Characteristics of the men and women in the study Characteristic

Men

Women

Number

2563

2763

Geometric mean SBP Median SBP Median SBP adjusteda Mean DBP s.e. DBP Median DBP Median DBP adjusteda

138.0 136 137

133.0 130 131

77.8 0.24 77 79

73.2 0.23 72 73

3.7

2.6

Mean age s.e. age

47.5 0.34

47.7 0.33

Geometric mean BMI (kg/m2)

26.1

25.6

% smokers

25

22

% manual classes

50

46

% non/ex/occasional drinkers % men 1–21, women 1–14 units a week % men ⬎21, women ⬎14 units a week

12 58

30 56

29

14

26 21 53

26 20 54

Geometric mean blood lead (␮g/dl)

% residing in the North % residing in the Midlands % residing in the South a

see Methods; respondents on antihypertensive medication were assumed to have underlying blood pressure above the median.

South is slightly higher by around 1–2%. The sample which did not include those on antihypertensive medication had slightly lower mean age (45.0 for men and 44.8 for women), geometric mean SBP (136.6 and 130.4, respectively), and mean DBP (77.0 and 72.2, respectively), but had the same geometric mean blood lead and was largely similar in other characteristics to the whole sample (data not shown). Unadjusted mean SBP and DBP generally increased with increasing blood lead group in both sexes (Figure 1). Both BPs were significantly associated with blood lead (P ⬍ 0.001), with simple correlation coefficients being 0.11 in men and 0.19 in women for SBP, and 0.15 and 0.14 for men and women, respectively, for DBP. The associations between several variables and SBP or DBP stratified by sex in ordinary multiple regression analyses are presented in Tables 2–5. In all the analyses, blood lead was found to be significantly and positively associated with DBP, but not SBP, in men. A significant association with DBP was only found in women when there was no adjustment for alcohol intake and when those on antihypertensive medication were included (Table 4). Among all the variables entered in the regression, age and, to a lesser extent, BMI were consistently the most important determinants of BP as shown by the relative magnitude of their standardised coefficients. A halving of blood lead in men (from any level) was associated with an estimated decrease in mean DBP of between 0.78 and 1.07 mm Hg: 0.78 (95%


125

Figure 1 Mean systolic and diastolic blood pressure (unadjusted) by blood lead level. Table 2 Summary of the regression models: with adjustment for alcohol consumption, respondents on antihypertensive medication included

Table 3 Summary of the regression models: with adjustment for alcohol consumption, respondents on antihypertensive medication excluded

Variable

Variable

Standardised coefficients Log10 (Systolic)

Diastolic

Men

Women

Men

Women

No. Adjusted R2

2563 0.206

2763 0.405

2563 0.209

2763 0.171

Age Log10 (BMI) Heavy drinkerd Log10 (lead) Manual classes Northern resident Smoker

0.392a 0.164a 0.063a NS 0.041c NS NS

0.559a 0.204a NS NS NS 0.038b NS

0.367a 0.182a 0.063a 0.061a NS NS NS

0.362a 0.142a 0.040c NS NS NS NS


Diastolic

Men

Women

Men

Women

No. Adjusted R2

2249 0.186

2394 0.369

2249 0.216

2394 0.165

Age Log10 (BMI) Heavy drinkerd Log10 (lead) Manual classes Northern resident Smoker

0.369a 0.165a 0.060b NS 0.044c NS NS

0.516a 0.231a NS NS NS 0.043b NS

0.386a 0.175a 0.066a 0.055b NS NS NS

0.337a 0.177a 0.038c NS NS NS NS

a

P ⬍ 0.001; bP ⬍ 0.01; cP ⬍ 0.05; d⬎21 units for men and ⬎14 units for women. NS: not significant and not included in the model.

a

P ⬍ 0.001; bP ⬍ 0.01; cP ⬍ 0.05; d ⬎21 units for men and ⬎14 units for women. NS: not significant and not included in the model.

confidence interval of 0.01–1.55) in the model that excluded men on antihypertensive medication but adjusted for alcohol consumption, 0.88 (0.13–1.63) in the model that included men on antihypertensive medication and adjusted for alcohol consumption, 0.96 (0.23–1.70) in the model that excluded men on antihypertensive medication and did not adjust for alcohol consumption, 1.07 (0.37–1.78) in the model that included those on antihypertensive medication and did not adjust for alcohol consumption. The results for the median regression analyses

were largely similar to those of the above regression (data not shown). Blood lead was not significant in any of the models for DBP in women nor in any of the models for SBP in either sex. Reducing blood lead by half in men was associated with an estimated decrease in median DBP of 1 mm Hg, irrespective of adjustment for alcohol consumption. Blood lead was significantly associated with pulse pressure in women (P ⬍ 0.001) but not in men. However this association did not reach statistical significance after adjusting for various confounders.


126

Table 4 Summary of the regression models: without adjustment for alcohol consumption, respondents on antihypertensive medication included Variable


Diastolic

Men

Women

Men

Women

No. Adjusted R2

2563 0.202

2763 0.405

2563 0.204

2763 0.171

Age Log10(BMI) Log10(lead) Manual classes Northern resident Smoker

0.382a 0.169a NS 0.039c NS NS

0.559a 0.204a NS NS 0.038b NS

0.355a 0.187a 0.075a NS NS NS

0.347a 0.142a 0.038c NS NS NS

P ⬍ 0.001; bP ⬍ 0.01; cP ⬍ 0.05; d⬎21 units for men and ⬎14 units for women. NS: not significant and not included in the model

a

Table 5 Summary of the regression models: without adjustment for alcohol consumption, respondents on antihypertensive medication excluded Variable


Diastolic

Men

Women

Men

Women

No. Adjusted R2

2249 0.182

2394 0.369

2249 0.213

2394 0.164

Age Log10(BMI) Log10 (lead) Manual classes Northern resident Smoker

0.362a 0.170a NS 0.042c NS NS

0.516a 0.231a NS NS 0.043b NS

0.375a 0.180a 0.068a NS NS NS

0.334a 0.175a NS NS NS NS

P ⬍ 0.001; bP ⬍ 0.01; cP ⬍ 0.05; d⬎21 units for men and ⬎14 units for women. NS: not significant and not included in the model.

a

Discussion Overall, evidence in this study suggests a small but statistically significant association between blood lead and DBP in men. This concurs with the conclusion drawn in a recent comprehensive review.5 This study is the first to use a nationally representative sample of the English population whereas previous studies in the UK have investigated age–sexrestricted9,10 or occupational11 subgroups of the population, and the results of these studies were inconsistent and conflicting. By fitting several analytical models this study has attempted to address two hitherto controversial issues, adjustment for alcohol consumption,5,18 and the inclusion and/or adjustment for the use of antihypertensive medication. The assumption behind the use of regression analysis to adjust for the effects of certain variables is that they confound the association under investigation. The role of alcohol in the blood lead/BP association is not clear-cut since the mechanisms whereby alcoholic beverages, parti-

cularly wine which contain significant amounts of lead,19 elevate BP have not been definitely identified and may or may not involve lead. Furthermore, alcohol can influence lead absorption, its metabolism in the liver, and its secretion via the kidney.5 The handling of respondents on antihypertensive medication has varied in previous studies. Some studies have excluded those on treatment20 thus reducing the power of the regression analysis and possibly excluding those most susceptible to any lead-induced increase in BP. However most studies have included those on treatment, thereby introducing potential bias because the BP recorded for those on treatment is lower than their ‘true’ pressure were they not on medication. One approach to this problem which has been used in other BP studies21,22 is to use the median BP and assume that the underlying BP of those on medication was above the median. Interpretation is further complicated because some antihypertensive agents can affect blood lead levels23 although the size of this effect is not known. In this study all three approaches (using mean BP and including or excluding those on medication and, using median BP whilst including those on medication) were carried out and little difference in the results was found, perhaps because only 12% of men and 13% of women were on antihypertensive medication. The finding of a significant association with DBP, and not SBP, is consistent with other cross-sectional studies in Austria24 and Denmark.25 Other studies based on large surveys in the USA26 and UK27 found significant associations with both measures of BP, while some smaller studies from Wales28 and Denmark29 found no significant association with either DBP or SBP. The finding in this report of a significant association in men, and not in women, concurs with studies based on NHANES II26 and in the Cadmibel study.30 To our knowledge, no cross-sectional studies have shown significant associations in women only, and in Staessen et al’s meta-analysis7 mentioned previously (the other meta-analysis, by Schwartz,6 did not include women) a weak positive association in both SBP and DBP was apparent in both sexes. These inconsistencies in findings are what might be expected if the magnitude of the BP elevation due to lead is small. A disadvantage of this study is that it is cross-sectional and therefore the temporal sequence of the observed association cannot be established. However controlled studies of animal models including various rat strains, pigeons and dogs show that the ingestion of low to moderate lead doses in the diet are associated with BP elevation.31–33 Similarly, four longitudinal studies,29,34–36 all of which showed a small if inconsistent positive association, support the likelihood that higher lead levels induce increased BP and not vice-versa. This evidence from animal experimental data and longitudinal studies add consistency and coherence to the data observed in the present study and hence tend to support a causal basis for the relationship. The use of the Dinamap 8100 for BP measurement


has received some criticism.37 However, whilst several comparative studies have shown differences in BP measured by the Dinamap 8100 and mercury sphygmomanometers, the direction and magnitude of these differences are inconsistent, and in the setting of a large multicentre survey in which BPs are measured in the home, the Dinamap 8100 is considered to be suitable and sufficiently accurate.13 Furthermore, any non-systematic error due to the use of this machine would only be expected to dilute any association between blood lead levels and BP. Based on the HSE data, the estimated decrease in SBP due to halving blood lead in men ranged from 0.8 to 1.1 mm Hg. This reduction is small, but changes in BP of this magnitude at the population level are likely to be associated with important benefits in terms of reduced cardiovascular morbidity and mortality. Between 1984 and 1995 there has been a three to five-fold reduction in blood lead.38 A major contributor to this reduction has been the reduced emission of lead from petrol, sales of unleaded petrol having increased from 0.1% of total sales in 1986 to 64.5% of the total in 1995. Further reductions of lead in the environment by increased use of leadfree petrol could feasibly reduce blood lead levels by 50%. The other major environmental lead source is drinking water. A recent report by the House of Lords Select Committee on the European Communities (1996) concluded that lead in drinking water posed serious health risks, particularly to children, infants and foetuses, and that the elimination of lead from drinking water should be an ideal policy objective.39 In summary, the HSE data in the context of other available information suggest that, given that lead has no known health benefit, is toxic in large amounts and, in lower doses, may have an adverse health impact on a major cardiovascular risk factor, further efforts to reduce environmental lead levels should be encouraged.

Acknowledgements We are indebted to the Department of Health and the Department of Environment for funding of this study and to the Social and Community Planning Research for carrying out the project jointly with the University College London Medical School’s Department of Epidemiology and Public Health.

References 1 Tanquerel des Planches L. Traite des maladies de plomb ou saturnines. Paris: 1839, p 550. 2 Pirkle JL et al. The decline in blood lead levels in the United States: The National Health and Nutrition Examination Surveys (NHANES). JAMA 1994; 272: 284 –291. 3 Victery W et al. Summary of discussions: symposium on lead-blood pressure relationships. In: Victery W (ed). Symposium on lead-blood pressure relationships, April 1987, Environmental Health Perspective: Chapel Hill NC. 1988; 78: pp 139–155.

4 National Research Council. Measuring Lead Exposure in Critical Populations. National Academy Press: Washington DC, 1993, p 97. 5 Hertz-Picciotto I, Croft J. Review of the relationship between blood lead and blood pressure. Epidem Reviews 1993; 15: 352–373. 6 Schwartz J. Lead, blood pressure, and cardiovascular disease in men. Arch of Environ Health 1995; 50: 31–37. 7 Staessen SJ et al. Hypertension caused by low-level lead exposure: myth or fact? J Cardiovasc Risk 1994; 1: 87–92. 8 Beevers DG et al. Blood lead and hypertension. Lancet 1976; ii: 1–3. 9 Pocock SJ et al. Blood lead concentration, blood pressure, and renal function. Br Med J 1984; 289: 872–874. 10 Pocock SJ et al. The relationship between blood lead concentration, blood pressure, stroke and heart attacks in middle-aged British men. Environ Health Perspect 1988; 78: 23–30. 11 Staessen J et al. Blood lead concentration, renal function, and blood pressure in London civil servants. Br J of Ind Med 1990; 47: 442– 447. 12 Prescott-Clarke, Primatesta P (eds). Health Survey for England 1995. The Stationery Office: London, 1997. 13 Bolling K. Dinamap 8100 calibration study. OPCS. HMSO: London, 1994. 14 Delves HT. A microsampling method for the determination of lead in blood by atomic absorption spectrophophotometry. Analyst 1970; 95: 431– 438. 15 Dong W, Primatesta P, Bost L. Blood Pressure. In: Prescott-Clarke T, Primatesta P (eds). Health Survey for England 1995. The Sationery Office: London, 1997. 16 Dong W, Colhoun H, Lampe H. Blood pressure. In: Colhoun H, Prescott-Clarke T (eds). Health Survey for England 1994. HMSO: London, 1996. 17 STATA Reference Manual, Release 5, Volume 2. Stata Press: College Station, Texas, 1997. 18 Schwartz J. Lead, blood pressure, and cardiovascular disease in men. Arch of Environ Health 1995: 50: 31–37. 19 Rhainds M, Levallois P. Effects of maternal cigarette smoking and alcohol consumption on blood lead levels of newborns. Am J Epidemiol 1997; 145: 250–257. 20 Menditto A et al. Association of blood lead and blood pressure in men aged 55 to 75 years: effect of selected social and biochemical confounders. Environ Health Perspect 1994; 102: 107–111. 21 White I, Chaturvedi N, McKeigue P. Median analysis of blood pressure for a sample including treated hypertensives. Stat Med 1994; 13: 1635–1641. 22 Intersalt Cooperative Research Group. Intersalt: an international study of electrolyte excretion and blood pressure. Results for 24 hour urinary sodium and potassium excretion. Br Med J 1988; 297: 319–328. 23 Sharp DS et al. Elevated blood pressure in treated hypertensives with low-level lead accumulation. Arch Environ Health 1989; 44: 18–22. 24 Wolf C et al. Effect of blood lead on blood pressure in occupationally nonexposed men. Am J Indust Medic 1995; 27: 897–903. 25 Kirkby H, Gyntelberg F. Blood pressure and other cardiovascular risk factors of long-term exposure to lead. Scand J Work Environ Health 1985; 11: 15–19. 26 Harlan WR et al. Blood lead and blood pressure, relationship in the adolescent and adult US population. JAMA 1985; 253: 530–534. 27 Pocock SJ et al. The relationship between blood lead, blood pressure, stroke, and heart attacks in middleaged British men. Environ Health Perspect 1988; 78: 23–30.

127


128

28 Elwood PC et al. Blood pressure and blood lead in surveys in Wales. Am J Epidemiol 1988; 127: 942–945. 29 Grandjean P et al. Blood lead-blood pressure relations: alcohol intake and hemoglobin as confounders. Am J Epidemiol 1989; 129: 732–739. 30 Staessen J et al. The association between blood pressure, calcium, and other divalent cations: a population study. J Hum Hypertens 1991; 5: 485– 494. 31 Bogden JD et al. Dietary lead and calcium: effects on blood pressure and renal neoplasia in Wistar rats. J Nutr 1991; 121: 718–728. 32 Fouts PJ, Page IH. The effect of chronic lead poisoning on arterial blood pressure in dogs. Am Heart J 1942; 24: 329–331. 33 Perry HM, Erlanger MW. Pressor effects of chronically feeding calcium and lead together. In: Trace substances in Environmental Health-XII: Proceedings of the University of Missouri’s 12th annual Conference on trace substances in Environmental Health, Colmbia, MO. Hemphill DD (ed) University of Missouri: Columbia, MO, 1978, pp 268–275. 34 Staessen JA, Roels H, Fagard R. Lead exposure and

35

36 37

38

39

conventional and ambulatory blood pressure. JAMA 1996; 275: 1563–1570. Neri LC, Hewitt D, Orser B. Blood lead and blood pressure: analysis of cross-sectional and longitudinal data from Canada. Environ Health Perspect 1988; 78: 123–126. Weiss ST et al. The relationship of blood lead to blood pressure in a longitudinal study of working men. Am J Epidemiol 1986; 123: 800–880. O’Brien E, Mee F, Atkins N, O’Malley K. Short report: accuracy of the Dinamap portable monitor, model 8100 determined by the British Hypertension Society protocol. J Hypertens 1993; 11: 761–763. Delves HT. Overview of UK and international studies on trends in blood lead, and the use of lead isotope ratios to environmental sources. Recent UK Blood Lead Surveys. Institute for Environment and Health: Leicester, 1998. Thornton I. Trends in environmental lead levels: implications to human exposure. Recent UK Blood Lead Surveys. Institute for Environment and Health: Leicester, 1998.