Nonnarcotic Analgesic Use and the Risk of Hypertension in US Women
Acetaminophen, aspirin, and other nonsteroidal anti-inflammatory drugs (NSAIDs) are widely consumed. Each is theoretically capable of elevating blood pressure by altering prostaglandin homeostasis; however, there is little prospective information on the relation between these agents and physician-diagnosed hypertension. We examined the association between the use of aspirin, acetaminophen, or NSAIDs and incident hypertension in a prospective cohort study of 51 630 women 44 to 69 years of age in 1990 who had no history of hypertension or chronic renal insufficiency. Analgesic use was assessed in 1990 by a mailed questionnaire, and the women were followed for 8 years. The primary outcome was physician-diagnosed hypertension reported on a follow-up biennial questionnaire. During 381 078 person-years of follow-up, 10 579 incident cases of hypertension were identified. Compared with nonusers, women who used aspirin or acetaminophen at least 1 day per month or NSAIDs 5 or more days per month were at a significantly higher risk for development of hypertension. After adjusting for potential confounders, the odds ratios for women in the highest frequency of use category (≥22 days per month) compared with no use were as follows: aspirin, 1.21 (95% CI, 1.13 to 1.30); acetaminophen, 1.20 (1.08 to 1.33); and NSAIDs, 1.35 (1.25 to 1.46). For each analgesic type, there was a significant trend toward an increased risk of incident hypertension with increasing frequency of use (P<0.001). Given the observed odds ratios, biologic plausibility, and the sizeable population at risk, health professionals should consider potential hypertensive effects of aspirin, acetaminophen, and NSAIDs when counseling their patients about the use of nonnarcotic analgesics.
Aspirin, acetaminophen, and ibuprofen are the three most commonly used medications among adults 18 years of age and older, according to a national survey of US households conducted between 1998 and 1999.1 Seventeen percent of respondents reported using aspirin in the preceding week, 23% reported acetaminophen use, and 17% ibuprofen use. Prevalence of use was greater among women than men for acetaminophen and ibuprofen.
Short-term prospective studies suggest nonsteroidal anti-inflammatory drugs (NSAIDs) can cause acute elevations in blood pressure,2,3 but a diagnosis of hypertension was used as the primary outcome in only one case-control study.4 Although aspirin and acetaminophen also influence prostaglandin homeostasis,5–7 prospective data on their potential hypertensive effects have been sparse and inconclusive.8,9 A recent study of these effects in a large cohort of younger US women demonstrated an increased incidence of hypertension in users of NSAIDs and acetaminophen but not aspirin (OR 1.86 and 2.00, respectively, for the highest use category, P<0.001).10
Twenty-four percent of US adults and 50% to 70% of those >60 years of age have hypertension.11 Even small elevations in blood pressure caused by nonnarcotic analgesic use could affect cardiovascular morbidity and mortality.12 To examine this issue, we studied the association between the use of aspirin, acetaminophen, and NSAIDs and incident hypertension in a large cohort of US women.
The Nurses’ Health Study
The Nurses’ Health Study cohort was assembled in 1976 when 121 700 female registered nurses, 30 to 55 years of age, completed and returned a mailed questionnaire.13 Follow-up questionnaires have been mailed every 2 years to update information on health-related behaviors and medical events. In 1990, questions were included regarding frequency of use of acetaminophen, aspirin, and other NSAIDs.
The 1990 questionnaire was answered by 85 625 women. Women who reported a history of hypertension (n=27 344) or chronic kidney failure (n=10) on or before the 1990 questionnaire were excluded from the study, as were women who did not answer any of the questions on analgesic use (n=336). We also excluded subjects who did not report having had a physical examination between 1988 and 1990 (n=735) and at least one physical examination during follow-up from 1990 to 1998 (n=5570). Thus, 51 630 women were included in the current study.
Assessment of Analgesic Use
The 1990 biennial questionnaire contained questions on the use of aspirin, acetaminophen, and NSAIDs. The questions were worded: “On average, how many days each month do you take any of the following medications?” The response categories for each analgesic were “none,” “1 to 4 days per month,” “5 to 14 days per month,” “15 to 21 days per month,” and “≥22 days per month.” Information on dosage and prior duration of analgesic use was not ascertained.
Assessment of Covariates
Body mass index (BMI), smoking status, age, and physical activity were ascertained on the 1990 questionnaire and were updated with each subsequent biennial questionnaire. Alcohol and sodium intake were assessed in 1990 and again in 1994 with the use of a semiquantitative food frequency questionnaire. A woman was considered to have a history of diabetes if she had reported it on any biennial questionnaire through 1990. Information on family history of hypertension was obtained on the 1992 questionnaire.
Ascertainment of Incident Hypertension
On each biennial questionnaire, women were asked to indicate whether a physician had made a new diagnosis of hypertension in the preceding 2 years. A previous validation study in this cohort showed that self-reported hypertension was highly correlated with a documented diagnosis of hypertension in the medical record and was also predictive of subsequent adverse cardiovascular outcomes.14
Incidence rates were computed by dividing the number of new cases of hypertension by the number of person-years in that analgesic use category. The odds ratio was used as the measure of association. Multivariate pooled logistic regression was used to estimate odds ratios while simultaneously controlling for the following potential confounding variables: age (5-year increments); BMI (6 categories); sodium intake (quintiles); alcohol use (6 categories); physical activity (quintiles of METs); smoking status (current, past, or never); history of diabetes (yes/no); and family history of hypertension (yes/no). We also simultaneously controlled for concurrent use of the other analgesic types. Subjects with missing data were assigned to a missing category for that specific time. Period probability values for trend were generated by entering the frequency of analgesic use as a continuous variable in pooled logistic regression models. We examined the relation between frequency of analgesic use reported in 1990 and the development of hypertension after return of the 1990 questionnaire until May 31, 1998. Subjects were censored after being diagnosed with hypertension or at the time of death.
We also investigated whether age, BMI, or alcohol use modified the relation between analgesic use and incident hypertension. These potential effect modifiers were categorized as follows: age <60 or ≥60 years; BMI <25 or ≥25 kg/m2; and alcohol intake 0 to 4 g/d, 5 to 14 g/d, or >14 g/d. All probability values are 2-sided. Analyses were performed with SAS statistical software, version 7.0 (SAS Institute Inc).
Subject Characteristics and Baseline Analgesic Use
During 381 078 person-years of follow-up, we identified 10 579 cases of incident hypertension. The median age of the cohort was 55 years (interquartile range 49 to 61), and 50% had a BMI ≤24 kg/m2 (interquartile range 22 to 27). A family history of hypertension was reported by 41%, and 3% reported diabetes mellitus. Cohort characteristics by analgesic type and frequency of use are displayed in Table 1. For each analgesic type, 41% to 47% of women reported no regular use, and 15% did not use any of the 3 analgesics. Use frequency in 1990 and in 1992 were compared to assess the consistency of use over time. Of the subjects with complete data on frequency of use for both years, 62% to 65% remained in the same use category. If grouped according to low (0 to 4 days per month), medium (5 to 14 days per month), or high (≥15 days per month) use frequency, 76% to 82% remained in the same category on both questionnaires. Furthermore, only 0.3% to 2.1% changed from the lowest to the highest or from the highest to the lowest use frequency category between 1990 and 1992.
Analgesic Use and Incident Hypertension
The age-adjusted risk of hypertension was higher for users of aspirin, acetaminophen, or NSAIDs at all use frequencies compared with nonusers (Table 2). After further adjustment for other potential confounders, the risks remained significantly higher beginning at 1 to 4 days per month for aspirin (OR=1.08; 95% CI, 1.03 to 1.15) and acetaminophen (OR=1.07; 95% CI, 1.02 to 1.13) and beginning at 5 to 14 days per month for NSAIDs (OR=1.21; 95% CI, 1.12 to 1.31). For each analgesic type, there was a significant trend toward an increased risk of hypertension with increasing frequency of use (P<0.001). Among the highest frequency users (≥22 days per month), the odds ratio of hypertension was slightly higher for NSAIDs (OR 1.35; 95% CI:1.25 to 1.46) than for aspirin (OR 1.21; 95% CI:1.13 to 1.30) and acetaminophen (OR 1.20; 95% CI, 1.08 to 1.33). Adjusting for age in 1-year increments and BMI in deciles did not alter the results.
The odds ratios of hypertension for aspirin and NSAIDs use were higher in younger women (<60 years) compared with older women. However, the differences in the magnitudes of the odds ratios were generally <20%. The magnitudes of the odds ratios were not substantially modified by BMI or alcohol intake.
We observed an increasing risk of development of hypertension with increasing use of aspirin, acetaminophen, or NSAIDs, based on a single assessment of analgesic use. This risk appeared to be slightly attenuated in older women.
The balance between vasodilator (eg, PGI2 and PGE2) and vasoconstrictor prostaglandins (eg, PGF2α and thromboxane A2) is postulated to affect blood pressure through renal and systemic mechanisms.15–17 Aspirin,7 acetaminophen,5,6 and NSAIDS18 have each been shown to reduce formation of vasodilator prostaglandins from arachadonic acid. In the setting of volume depletion, PGI2 and PGE2 enhance glomerular filtration15,19 and reduce fractional sodium reabsorption.20,21 These actions help counterbalance the effects of constriction of the renal vasculature caused by angiotensin II and catecholamines22 under conditions such as chronic kidney failure.19 In addition, ibuprofen and indomethacin have been shown to preferentially inhibit extrarenal prostaglandin synthesis and to elevate blood pressure16,17 in part, perhaps, by increasing endothelin-1 production.23
Two cross-sectional studies in elderly populations suggest a link between NSAID use and hypertension. In a cohort of 133 community-dwelling elderly subjects taking antihypertensive medications,24 the odds of having a systolic blood pressure >140 mm Hg were higher in NSAID users (OR, 2.19; 95% CI, 1.33 to 3.61) than in control subjects matched for age, gender, and BMI. In the second study,25 NSAID use was significantly associated with hypertension only in subjects taking antihypertensive medications (OR, 1.4; 95% CI, 1.1 to 1.7). Users of aspirin and NSAIDs were pooled, and neither intensity of antihypertensive medication use nor concomitant use of acetaminophen was controlled for in this analysis.
A large case-control study of Medicaid beneficiaries4 found the odds of filling a first prescription for antihypertensive medication were 2 times higher in subjects who had filled a prescription for NSAIDs within the preceding year than in those who had not. The odds ratios were greater with more recent NSAID use and with an increasing daily dose of NSAIDs.
Intervention studies have been small with conflicting results; however, 2 meta-analyses2,3 found that NSAIDs significantly elevated mean arterial pressure an average of 4 to 5 mm Hg in subjects taking antihypertensive medications. None of the trials included in these meta-analyses exceeded 6 weeks in duration, so the effects of longer-term NSAID use on blood pressure could not be evaluated.
The effects of prolonged aspirin use on blood pressure have been evaluated in 2 prospective studies. A Swiss cohort study of 1040 women26 found that baseline aspirin use, determined by a single urine sample, was associated with a odds ratio of 1.3 (P=0.11) for developing hypertension over the following 20 years. This study did not comment on concurrent use of NSAIDs or acetaminophen and, apart from age, did not adjust for standard risk factors for hypertension. The second study8 was a double-blind, placebo-controlled trial in which 102 male patients were randomly assigned to receive either 1.5 g of aspirin or placebo daily for 2 years after myocardial infarction. After the 74 subjects who were taking antihypertensive medications were excluded from analysis, the treatment group (n=17) had significant increases in systolic and diastolic pressures (14 mm Hg and 9 mm Hg, respectively, P<0.05) during the first 6 months of follow-up relative to baseline. However, blood pressures returned to baseline values by 1 year of follow-up. There was no significant change in blood pressure in the placebo group (n=11).
We are aware of one intervention study that examined the effect of acetaminophen on blood pressure.9 In this randomized, placebo-controlled, crossover trial of 22 hypertensive patients, 1 g acetaminophen taken 4 times daily was associated with a 4–mm Hg increase in supine and standing systolic blood pressures (P<0.05) during 4 weeks of treatment. A recent prospective cohort study of >80 000 nurses 31 to 50 years of age found a higher incidence of hypertension among subjects who used acetaminophen or NSAIDs at least 22 days per month compared with nonusers (OR, 2.00 and 1.86, respectively, P<0.001).10 No statistical association was observed for aspirin. Analyses controlled for all covariates in the present study plus dietary magnesium and potassium, oral contraceptive use, and baseline blood pressure.
Although there are published data to support a potential hypertensive effect for aspirin, acetaminophen, and NSAIDs, our finding that the magnitudes of the odds ratios were similar across analgesic types by frequency of use deserves mention. It is unknown if analgesic users were more likely to be diagnosed with hypertension because of more frequent physician contact than nonusers. We minimized this possibility by excluding women who did not have a physical examination in the 2-year period before the assessment of analgesic use and at least one biennial physical examination during the follow-up period. Although the exact time course for developing hypertension after exposure to nonnarcotic analgesics is unknown, data from a similar study in a cohort of younger nurses10 showed statistically significant increases in the risk of hypertension by analgesic use frequency within the first 2 years of follow-up. Could residual confounding by BMI or physical activity have produced this effect? Our careful control for these and other known risk factors for hypertension in multivariate models makes this unlikely. Furthermore, we are unaware of any common conditions for which analgesics are used that are independently associated with the risk of hypertension. For these reasons, the frequency of doctor visits for conditions associated with analgesic use is not likely to be associated with the absolute risk of developing hypertension.
Although we hypothesized that the odds ratio of hypertension should be highest for NSAIDs and lowest for acetaminophen, this was based on data from studies of relatively short duration, which may have underestimated the more remote hypertensive effects of aspirin and acetaminophen. Although brief exposure to these drugs may influence the risk of development of hypertension, our data show that the majority of subjects remained in the same general use category when surveyed 2 years later. This suggests that current use may reflect long-term use patterns. In fact, variation in subjects’ frequency of analgesic use during follow-up would be expected to produce random misclassification of the exposure that would reduce our ability to detect an association. Our study was uniquely suited to detect these effects because of its length of follow-up and the broad range of analgesic use frequencies examined.
We observed a plateau in the incidence of hypertension at the third highest frequency of use category (15 to 21 days per month) among all 3 analgesic classes. This may indicate a physiological plateau beyond which increasing the frequency of analgesic use does not cause further blood pressure elevations.
Our study has several strengths. We determined baseline analgesic use frequency before the development of hypertension, and the follow-up interval was 8 years. In addition, we used reliable information on many known risk factors for hypertension and concurrent use of multiple analgesic categories to control for possible confounding. The latter point is important, given the previously reported hypertensive effects of NSAIDs and the possibility that an individual may use more than one type of analgesic concurrently.
Our findings suggest that use of aspirin, acetaminophen, and NSAIDs is associated with the risk of development of physician-diagnosed hypertension. This carries important public health implications, given the frequency of use of these medications in the general population and the relation between elevated blood pressure and cardiovascular death.12 Additional large, prospective cohort studies with frequent and uniform blood pressure surveillance are needed to confirm our findings and to determine precise temporal relations between analgesic use and hypertension.
This study was supported by grants DK52866 and CA87969 from the National Institutes of Health. The funding source had no involvement in the study design or in the collection, analysis, and interpretation of the data. The authors thank the participants of the Nurses’ Health Study and Dr Sharon Curhan, Elaine Coughlan, Christine Jones, Kristine Vernon, Barbara Egan, and Karen Corsano for their expert help.
Presented in part in abstract form at the American Society of Nephrology Annual Meeting on October 15, 2000, Toronto, Ontario, Canada.
- Received April 15, 2002.
- Revision received May 13, 2002.
- Accepted August 20, 2002.
Kaufman DW, Kelly JP, Anderson TE, Rosenberg L, Mitchell AA. A comprehensive population-based survey of medication use in the United States: the adult population. Pharmacoepidemiol Drug Saf. 2000; 9: S60.
Mattammal MB, Zenser TV, Brown WW, Herman CA, Davis BB. Mechanism of inhibition of renal prostaglandin production by acetaminophen. J Pharmacol Exp Ther. 1979; 210: 405–409.
Roth GJ, Stanford N, Majerus PW. Acetylation of prostaglandin synthase by aspirin. Proc Natl Acad Sci U S A. 1975; 72: 3073–3076.
Curhan G, Willett WC, Rosner B, Stampfer MJ. Frequency of analgesic use and risk of hypertension in younger women. Arch Intern Med. 2002. In press.
Burt VL, Whelton P, Roccella EJ, Brown C, Cutler JA, Higgins M, Horan MJ, Labarthe D. Prevalence of hypertension in the US adult population. Results from the Third National Health and Nutrition Examination Survey, 1988–1991. Hypertension. 1995; 25: 305–313.
MacMahon S, Peto R, Cutler J, Collins R, Sorlie P, Neaton J, Abbott R, Godwin J, Dyer A, Stamler J. Blood pressure, stroke, and coronary heart disease, I: prolonged differences in blood pressure: prospective observational studies corrected for the regression dilution bias. Lancet. 1990; 335: 765–774.
Colditz GA, Martin P, Stampfer MJ, Willett WC, Sampson L, Rosner B, Hennekens CH, Speizer FE. Validation of questionnaire information on risk factors and disease outcomes in a prospective cohort study of women. Am J Epidemiol. 1986; 123: 894–900.
Laneuville O, Breuer DK, Dewitt DL, Hla T, Funk CD, Smith WL. Differential inhibition of human prostaglandin endoperoxide H synthases- 1 and -2 by nonsteroidal anti-inflammatory drugs. J Pharmacol Exp Ther. 1994; 271: 927–934.
Ciabattoni G, Cinotti GA, Pierucci A, Simonetti BM, Manzi M, Pugliese F, Barsotti P, Pecci G, Taggi F, Patrono C. Effects of sulindac and ibuprofen in patients with chronic glomerular disease: evidence for the dependence of renal function on prostacyclin. N Engl J Med. 1984; 310: 279–283.
Kauker ML. Prostaglandin E2 effect from the luminal side on renal tubular 22Na efflux: tracer microinjection studies. Proc Soc Exp Biol Med. 1977; 154: 274–277.
Needleman P, Douglas JR Jr, Jakschik B, Stoecklein PB, Johnson EM Jr. Release of renal prostaglandin by catecholamines: relationship to renal endocrine function. J Pharmacol Exp Ther. 1974; 188: 453–460.
Chrischilles EA, Wallace RB. Nonsteroidal anti-inflammatory drugs and blood pressure in an elderly population. J Gerontol. 1993; 48: M91–M96.