Headaches and the Treatment of Blood Pressure
Results From a Meta-Analysis of 94 Randomized Placebo-Controlled Trials With 24 000 Participants
Background— Uncertainty exists over whether blood pressure–lowering drugs prevent headache.
Methods and Results— A meta-analysis was carried out of the 94 randomized placebo-controlled trials of 4 different classes of blood pressure–lowering drugs (thiazides, β-blockers, ACE inhibitors, and angiotensin II receptor antagonists) in fixed doses in which data on headache were reported. There were 17 641 participants who were allocated blood pressure–lowering drugs and 6603 who were allocated placebo. Treatment lowered systolic and diastolic blood pressures by 9.4 and 5.5 mm Hg, respectively, on average. One third fewer people on average reported headache in the treated groups (8.0%) than the placebo groups (12.4%) (odds ratio, 0.67; 95% CI, 0.61 to 0.74; P<0.001). About 1 in 30 treated persons benefited by having headache prevented. The prevalence of headache was reduced (P<0.001) in trials of each of the 4 classes of drugs.
Conclusions— Our results show that blood pressure–lowering drugs prevent a significant proportion of headaches. That this effect is seen with pharmacologically unrelated classes of drugs indicates that it is likely to be due to the reduction in blood pressure per se, the only recognized action that the drugs have in common. This in turn indicates that high blood pressure is a cause of headache, but this conclusion is not supported by observational studies of blood pressure and headache. The uncertainty over whether high blood pressure causes headache does not, however, detract from the practical benefits of the use of blood pressure–lowering drugs in preventing headaches and cardiovascular disease.
Received December 15, 2004; revision received July 25, 2005; accepted July 29, 2005.
The classic “hypertensive headache,” present on waking, throbbing in nature, and wearing off during the morning, was described 90 years ago by Janeway.1 His patients described it “so commonly that I have almost come to look on it as a typical hypertensive symptom.”1 Similar reports followed; about half of patients reported headaches.2 Janeway’s view, however, was rejected in favor of an interpretation that the headache was “a socio-psychological disorder precipitated by the recognition of the hypertension.”2 This followed the demonstration 50 years ago on 200 consecutive patients with hypertension that headache was common (74%) in 96 patients who had been told that they had high blood pressure but uncommon (16%) in 104 patients who had not (the 2 groups were comparable in blood pressure and other factors)3; that finding was confirmed later in a larger (n=3858) study.4 Subsequent explanations were that anger- or anxiety-induced hyperventilation may both cause headache and raise blood pressure.5,6 Cross-sectional studies have generally shown no association between blood pressure and headache,4,7–21 reinforcing the prevailing view that there is no link with headache22 or that, if there is, the headaches are attributable to anxiety and tension induced by being told one has “hypertension.”23 Recently, an editorial asked, “Why does the hypertension headache myth persist?”24
Against this negative observational data, it is striking that published intervention studies have suggested that blood pressure–lowering drugs prevent headache. Data from 7 randomized double-blind trials of 1 blood pressure–lowering drug, irbesartan, showed that it prevented headache25; a large randomized controlled trial (not double blind) of doctors using any drug to lower blood pressure showed fewer headaches in treated patients26; and 2 nonrandomized intervention studies showed that when patients diagnosed as hypertensive received treatment, headache became less prevalent.7,27 The randomized double-blind trial data, however, relate to only 1 drug,25 and the other results are susceptible to bias.
To assess whether blood pressure–lowering drugs prevent headache, we report here a meta-analysis of 94 randomized placebo controlled trials of 4 classes of blood pressure–lowering drugs in which data on the prevalence of headache were reported.
The analysis was adapted from our previously reported systematic review of 354 randomized placebo-controlled trials of 5 classes of blood pressure–lowering drugs in fixed doses28,29 in which we assessed their efficacy in lowering blood pressure and the prevalence of adverse effects according to dose.28 In this analysis, we excluded trials of calcium channel blockers because they can cause headache (through vasodilation).
The analysis was based on a systematic review of all randomized placebo controlled trials of any drug in 4 classes of blood pressure–lowering drugs (thiazides, β-blockers, ACE inhibitors, and angiotensin II receptor antagonists) in fixed doses that were published from 1966 to 2001. We used MEDLINE but also searched the Cochrane Controlled Trials Register and the Web of Science database, examined citations in the reports of the trials identified and in review articles, and asked pharmaceutical companies to identify trials of drugs that they manufactured. We used Medical Subject Headings (MeSH) terms that included trials (clinical trial, controlled clinical trial, randomized controlled trial, random allocation, double-blind method) and blood pressure–lowering drugs (antihypertensive agents; hypertension; blood pressure; diuretics; thiazide; adrenergic β-antagonists; ACE inhibitors; receptors, angiotensin/antagonists and inhibitors; tetrazoles or the generic and trade names of individual drugs in the 4 classes). We also searched for studies containing the text words randomized or randomised or the generic or trade names of the individual drugs.
We included all double-blind trials of ≥2-week duration in adults (≥18 years of age). We excluded trials with no placebo group, with <2-week duration, with dose titration (so that different patients received different doses), or in which some control patients were treated. We also excluded trials testing drugs only in combination with other drugs, crossover trials with nonrandomized order of treatment and placebo periods, and trials recruiting patients with heart failure or acute myocardial infarction (but otherwise included trials regardless of diseases of the participants).28 Trials were not excluded if similar proportions of treated and placebo participants took other blood pressure–lowering drugs besides the test drug, although in almost all trials this did not happen. In trials in which the numbers of treated and placebo participants reporting headache over the duration of the trial were published, we recorded these data. We also recorded the reductions in systolic and diastolic blood pressures as the change in sitting or supine blood pressure in the treated group minus that in the placebo group (in crossover trials, end-treatment minus end-placebo blood pressure).
The data were analyzed with STATA statistical software. Parallel-group trials and crossover trials yielded similar results, so we combined them. Results were analyzed on an intention-to-treat basis. The estimates from each trial of the ratio of the odds of developing headache among participants allocated active treatment to that among participants allocated placebo were combined through the use of a random-effects model based on the method of DerSimonian and Laird.30 We tested for heterogeneity using the I2 test statistic proposed by Higgins and Thompson.31
We performed 2 additional searches of the databases. To identify randomized placebo controlled trials of the preventive effect of the blood pressure–lowering drugs in patients with migraine, we combined the names of individual drugs or of the classes of drugs with migraine, all as MeSH terms or text words. To identify observational studies of the association between blood pressure and headache, we combined the terms hypertension or blood pressure with headache.
Table 1 summarizes details of the 94 trials identified. Table 2 shows the reduction in blood pressure and prevalence of headaches according to the class of blood pressure–lowering drug, with summary estimates for the 4 classes of drug combined. On average, across the 94 trials, treatment lowered systolic and diastolic blood pressures by 9.4 and 5.5 mm Hg, respectively, adjusted for the change in the placebo groups. One third fewer people reported headache in the treated group than in the placebo group (odds ratio, 0.67; 95% CI, 0.61 to 0.74; P<0.001). There was also a statistically highly significant (P<0.001) reduction in the prevalence of headache in trials of each of the 4 classes of blood pressure–lowering drug.
Separate analysis of trials of parallel-group and crossover design showed statistically significant reductions in the prevalence of headache in the treated groups compared with placebo in both. In the 84 parallel-group trials, 33% fewer treated than placebo participants reported headache on average (95% CI, 26 to 39; P<0.001), and in the 10 crossover trials, 45% fewer participants reported headache (95% CI, 5% to 68%; P=0.03). There was no statistically significant difference between the 2, validating a combined statistical analysis.
The Figure summarizes the individual trial data. For each of the 4 classes of blood pressure–lowering drugs (thiazides, β-blockers, ACE inhibitors, angiotensin II receptor antagonists), there is a separate data point for each of the larger trials (>10 participants reporting headache); smaller trials are combined into a single data point. That figure shows that almost all the individual trials recorded a directionally lower prevalence of headaches in persons allocated blood pressure–lowering drugs than placebo. In 17 individual trials, headache was statistically significantly less prevalent in treated than placebo groups (identifiable in the Figure as those in which the 95% CI does not cross the line of unity); in only 1 trial was headache significantly less prevalent in the placebo groups. These results and the numbers with headaches in each trial (see Figure) show that the overall association does not arise from a small number of trials. Omitting the few trials with the most extreme results from sensitivity analyses had trivial effects.
There was a statistically significant dose-response relation across trials for diastolic blood pressure (ie, a greater reduction in headaches in trials producing a greater reduction in diastolic blood pressure). On average, the reduction in the prevalence of headaches was 13% greater (95% CI, 5% to 20%; P=0.01) in a trial with a 5-mm Hg-greater diastolic blood pressure reduction, but this association depended on the influence of a single trial with an outlying result and was weaker with this trial omitted (reduction in headaches 8% greater; 95% CI, 0 to 17%; P=0.06). There was no relation with systolic blood pressure. The analysis, however, lacked the statistical power to demonstrate a dose-response relationship because all the trials tested a single blood pressure–lowering drug against placebo, so there was little variation across trials in the blood pressure reduction attained.
There was significant heterogeneity between trials in the proportional reduction of headaches in the treated groups (χ293=124; P=0.01). This may reflect in part the above tendency to a greater reduction in headaches with a greater reduction in diastolic blood pressure. In addition, there was a suggestion of heterogeneity across the 4 classes of blood pressure–lowering drug (χ23=7.4; P=0.06), reflecting a greater effect of β-blockers than the other 3 classes of drug (see the Figure). The reduction in headache with blood pressure lowering was unrelated to age.
The absolute difference in the proportion of people reporting headache between the treated and control groups was 3.5% on average (95% CI, 2.8 to 4.1; P<0.001); ie, treatment prevented headache in 3.5%, or about 1 person in 30. However, the absolute difference in prevalence between the treated and placebo groups increased with prevalence in the placebo group (r=0.66, P<0.001). The relative reduction (of one third) reported above has the advantage of being independent of the prevalence of headache in the placebo group (blood pressure reduction tended to reduce headaches by a constant proportion of their existing prevalence).
The 94 randomized placebo controlled trials show that blood pressure–lowering drugs reduce the prevalence of headache by a third, a result that is highly statistically significant. The result cannot reasonably be explained by either chance or bias. In particular, publication bias, whereby trials showing a reduction in the prevalence of headache are more likely to be published than other trials, is unlikely. Many of the trials in our original data set28 did not report on headache, but this was because they reported only on symptoms recognized as being caused by the drugs or did not record symptoms. In the 94 trials in which headache was reported, those data were among tabulated data on several common symptoms; there was no special focus on headaches or on the fact that they were less common in the treated group. There was no indication of publication bias on inspection of funnel plots for asymmetry or on performing Egger’s test,32 although these tests, based on showing that smaller trials are more likely to be positive, are relatively insensitive.
Therefore, this meta-analysis shows that blood pressure–lowering drugs prevent headache. Interestingly, it is already acknowledged that 1 form of headache, namely migraine, is prevented by blood pressure–lowering drugs. Randomized placebo controlled trials conducted in patients with migraine testing β-blockers,33–36 ACE inhibitors,37 angiotensin II receptor antagonists,38 and calcium channel blockers39–41 (we found no randomized trials of thiazides) all show statistically highly significant reductions in frequency of migraine attacks in the treated groups. Our result from the 94 trials that blood pressure–lowering agents have a general effect in preventing headache extends the prophylactic effect beyond the prevention of migraine alone.
Whether the prevention of headache is attributable to the blood pressure reduction or to separate pharmacological actions of the drugs other than lowering blood pressure is unresolved. It would appear likely that the effect is due to the blood pressure reduction because the reduction in headaches was statistically significant (P<0.001) for each of the 4 classes of blood pressure–lowering drugs. ACE inhibitors and angiotensin II receptor antagonists have related actions, but these 2 classes of drugs together, thiazides, and β-blockers have no identifiable actions in common by which they might prevent headache other than blood pressure reduction (eg, no recognized analgesic effect). One would need to postulate that each prevented headaches in a different way. The effect of β-blockers may have been greater than that of the other drugs (Figure), which might reflect the recognized action of β-blockers in reducing the somatic manifestations of anxiety, but it is not recognized that any of the other classes of drug share this action. The 1 action that the drugs are recognized to share, blood pressure reduction, should be the probable mechanism of effect.
The conclusion that the blood pressure–lowering drugs prevent headache is firm, and the conclusion that a higher blood pressure causes headache is likely but is not corroborated by the observational studies (generally cross-sectional in design) on blood pressure and headache. We identified 16 such studies: 4 showed clear associations between blood pressure and headache that were statistically significant,7–10 1 showed a weak association,11 10 showed no suggestion of an association,4,12–20 and 1 (the largest) showed a statistically significant inverse association.21
Because of these generally negative results, we carried out a cross-sectional analysis across the placebo groups of the 94 trials (ignoring the intervention). On average, headaches were 17% less prevalent (95% CI: − 31% to 1%) for a 5-mm Hg lower diastolic blood pressure, but the association did not formally reach statistical significance (P=0.06), and there was no association with systolic blood pressure. The weak observational association in this and other studies may be attributable in part to random error such as variation between trials in criteria for defining an episode as a headache and to 2 sources of bias that will dilute the association in the cross-sectional studies. These sources of bias are regression dilution bias arising from the fluctuation on blood pressure in each individual person42 and the failure of some studies to adjust for age in the statistical analysis, which introduces bias because blood pressure increases with increasing age but the prevalence of headaches declines.9,12,16,20
Whether these factors are sufficient to account for the absence of an association in many cross-sectional studies is uncertain. The conclusion that the 4 classes of blood pressure–lowering drugs reduce the prevalence of headache by about a third (whatever the mechanism) is firm, but the issue of whether higher blood pressure per se causes headache remains unresolved. This uncertainty, however, does not detract from the practical benefit that the increasing use of these drugs to prevent cardiovascular disease will prevent headache in several millions of people worldwide.
Drs Law and Wald have a patent interest in the Polypill.
The online-only Data Supplement can be found at http://circ.ahajournals.org/cgi/content/full/112/15/2301/DC1.
Janeway TC. A clinical study of hypertensive cardiovascular disease. Arch Intern Med. 1913; 12: 755–798.
Di Tullio M, Alli C, Avanzini F, Bettelli G, Colombo F, Devoto MA, Marchioli R, Mariotti G, Radice M, Taioli E, Tognoni G, Villella M, Zussino A, for the Gruppo di Studio Sulla Pressione Arteriosa Nell’ Anziano. Prevalence of symptoms generally attributed to hypertension or its treatment: study on blood pressure in elderly outpatients (SPAA). J Hypertens. 1988; 6 (suppl 1): S87–S90.
Badran RH, Weir RJ, McGuiness JB. Hypertension and headache. Scott Med J. 1977; 15: 48–51.
Ziegler DK, Hassanein RS, Couch JR. Characteristics of life headache histories in a nonclinic population. Neurology. 1977; 27: 265–269.
Wessely S, Nickson J, Cox B. Symptoms of low blood pressure: a population study. BMJ. 1990; 301: 362–365.
D’Alesandro R, Benassi G, Lenzi PL, Gamberini G, Sacquegna T, De Carolis P, Lugaresi E. Epidemiology of headache in the Republic of San Marino. J Neurol Neurosurg Psychiatry. 1988; 51: 21–27.
Benseñor IJM, Lotufo PA, Martins MA. Behaviour of blood pressure in patients with chronic daily headache. Cephalagia. 1997; 17: 282. Abstract.
Waters WE. Headache and blood pressure in the community. BMJ. 1971; 1: 142–143.
Hagen K, Stovner JL, Vatten L, Holmen J, Zwart J-A, Bovim G. Blood pressure and risk of headache: a prospective study of 22 685 adults in Norway. J Neurol Neurosurg Psychiatry. 2002; 72: 463–466.
Williams GH. Approach to the patient with hypertension. In: Fauci AS, Braunwald E, Isselbacher KJ, Kurt J, Wilson JD, Martin JB, Kasper DL, Hauser SL, Longo DL, eds. Harrison’s Principles of Internal Medicine. 14th ed. New York, NY: McGraw-Hill Co; 1998: 202–205.
Swales JD. The clinical features of essential hypertension. In: Wetherall DJ, Ledingham JGG, Warrell DA, eds. Oxford Textbook of Medicine. 3rd ed. Oxford, UK: Oxford University Press; 1996: 2534–2535.
Friedman D. Headache and hypertension: refuting the myth. J Neurol Neurosurg Psychiatry. 2002; 72: 431.
Law MR, Wald NJ, Morris JK, Jordan RE. Value of low dose combination treatment with blood pressure lowering drugs: analysis of 354 randomised trials. BMJ. 2003; 326: 1427–1431.
Law M, Wald N, Morris J. Lowering blood pressure to prevent myocardial infarction and strokes: a new preventive strategy. Health Technol Assess. 2003;7:80–94. Available at: http://www.ncchta.org/project.asp?Pjtld=880.
Egger M, Davey Smith G, Schneider M, Minder C. Bias in meta-analysis detected by a simple, graphical test. BMJ. 1997; 315: 629–634.
van de Ven LL, Franke CL, Koehler PJ. Prophylactic treatment of migraine with bisoprolol: a placebo-controlled study. Cephalalgia. 1997; 17: 596–599.
Hedman C, Andersen AR, Andersson PG, Gilhus NE, Kangasniemi P, Olsson J-E, Strandman E, Nestvold K, Olesen J. Symptoms of classic migraine attacks: modifications brought about by metoprolol. Cephalalgia. 1988; 8: 279–284.
Kangasniemi P, Andersen AR, Andersson PG, Gilhus NE, Hedman C, Hultgren M, Vilming S, Olesen J. Classic migraine: effective prophylaxis with metoprolol. Cephalalgia. 1987; 7: 231–238.
Schrader H, Stovner LJ, Helde G, Sand T, Bovim G. Prophylactic treatment of migraine with angiotensin converting enzyme inhibitor (lisinopril): randomised, placebo controlled, crossover study. BMJ. 2001; 322: 19–22.
Andersson KE, Brandt L, Hindfelt B, Ryman T. Migraine treatment with calcium channel blockers. Acta Pharmacol Toxicol. 1986; 58 (suppl 2): 161–167.
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, part 1: prolonged differences in blood pressure: prospective observational studies corrected for the regression dilution bias. Lancet. 1990; 335: 765–774.