Anthony G. Johnson

Do Nonsteroidal Anti-inflammatory Drugs Affect Blood Pressure? A Meta-Analysis

Hypertension is prevalent [1, 2] and is a major determinant for stroke and coronary heart disease [3-5]. Further, the higher the blood pressure, the more marked is the reduction in life expectancy [6]. Nonsteroidal anti-inflammatory drugs (NSAIDs) are the most commonly prescribed drugs worldwide when grouped by generic categories [7], accounting for 4% to 9% of all prescriptions in developed countries [8-12]. Since 1967, their use has increased steadily, particularly among elderly persons [13]. Certain NSAIDs have been reported in randomized, controlled trials to elevate blood pressure in some previously normotensive persons with [14] and without antihypertensive therapy [15-20], in patients with mild hypertension untreated [19, 21, 22] or treated with single doses of antihypertensive agents [23, 24], and in hypertensive persons whose blood pressure had been controlled by drug therapy [21, 25-39]. However, several well-designed studies have failed to show any effect of NSAIDs on blood pressure [40-51]. In addition, many randomized studies [52-63] have attempted to compare the effect of various NSAIDs on blood pressure control with widely differing results that have been thus far inconclusive. For most of the individual trials, the estimates may have been imprecise because the sample size was insufficient. In view of the anticipated increase in the prevalence of hypertension [64] and the substantial use of NSAIDs, any potential drugdrug (NSAID-antihypertensive agent) and drug-disease (NSAID-hypertension) interactions should be fully clarified. Consequently, we did a meta-analysis of randomized trials. Our primary goal was to produce a stable estimate of the overall effect of various NSAIDs on blood pressure; our secondary aims were to evaluate possible mechanisms by which NSAID therapy may alter blood pressure and to determine potential predisposing factors for this interaction. It has been proposed that NSAIDs may alter blood pressure through the effects of prostaglandin synthesis inhibition on body weight, cardiac output, or renal function. Consequently, where provided, we have included these data in the meta-analysis. Although NSAIDs have been associated with blood pressure elevation in normotensive persons and in both treated and untreated hypertensive persons, data pertaining to studies including persons in these subgroups were also analyzed separately to determine whether differential effects occurred in various population groups. Similarly, it was uncertain whether NSAIDs interacted with antihypertensive agents from different classes in the same manner and whether different NSAIDs altered blood pressure to the same degree. Consequently, we also studied these subgroups separately. Methods We did a MEDLINE search from 1966 to 1990. We also did the following additional searches: Embase (1974 to 1990), Biosis (1969 to 1990), Diogenes (1976 to 1990), Science Citation Abstracts (1972 to 1990), International Pharmaceutical Abstracts (1970 to 1990), IOWA Drug Information Service (1966 to 1990), and the Combined Health Information Database (1973 to 1990). For each search, key words relating to trial design (meta-analysis, research design, double-blind method, double-blind study, double, blind, random allocation, random, control, clinical trials) were crossed with the names of individual NSAIDs and the following terms: anti-inflammatory agents, nonsteroidal, hypertension, blood pressure. Articles reviewing the potential interaction between NSAIDs and blood pressure were selected from these searches, and their bibliographies were carefully checked for randomized trials not published elsewhere. We also reviewed textbooks on hypertension, clinical pharmacology, and NSAIDs, and checked reference lists of all randomized trials identified by any of the above means. We selected 194 articles as potentially fulfilling the entry criteria. Only 60 articles, however, described random allocation of one or more NSAIDs and measurement of the effect of NSAIDs on blood pressure. Of the 60 articles, 38 included randomized and controlled trials [14-51] (Appendix Tables 1 and 2). Twelve more articles [52-63] were reports of randomized but not placebo-controlled studies; in these studies, two or more NSAIDs were compared with each other rather than with placebo in terms of their effects on blood pressure control. The remaining 10 articles [65-74] could not be included: eight articles [65-72] because they did not provide measures of variance for the blood pressure data such as standard error or standard deviation, which was needed for our analysis; one article [73] because no blood pressure data were provided; and one article [74] because blood pressure measurement was not the primary goal of the study and because uncontrolled administration of morphine in that trial probably confounded blood pressure levels. The eight articles that did not provide variance data for the blood pressure measurements [65-72] and the article in which blood pressure data were not provided [73] were similar to the 50 articles included in the meta-analysis with respect to patient characteristics, relevant diagnoses, and treatment received. In addition, like the included trials, reporting of an effect of NSAIDs on blood pressure was inconsistent among these articles. However, many were completed several years ago, reducing the possibility of obtaining accurate, reliable data from the original authors; most were abstracts [66, 67, 69, 70, 73], and none had a large sample size with adequate follow-up. Consequently, we considered it appropriate to exclude these studies. Control and Measurement of Potential Bias To minimize selection bias, the Clinical Trials Coordinator coded the methods and results of articles that potentially fulfilled the entry criteria for the meta-analysis. The relevant methods and results sections were photocopied after any identifying information was blacked out. The decision to accept or reject articles was made by two authors, based on the methods alone without any knowledge of the source of the articles (authors or institution) and achieved by consensus. The same two authors subsequently developed an algorithm for assessing the methodologic rigor and scientific quality of the articles to be pooled and for calculating quality assessment scores (0 to 100 [the higher the score, the better the quality of the study assessed]), using only the methods and results sections. This algorithm was broadly based on a systematic evaluation of study quality published by Chalmers and colleagues [75]. Quality assessment scores by the two assessors highly correlated (r = 0.9), suggesting close interobserver agreement. In an attempt to control data-extraction bias, relevant data were drawn from papers by the assessors, who were blinded to all but the results sections, and in 89% of studies evaluated agreement in the data extracted was complete. One author provided a quality assessment score for all trials and extracted data from every trial included in the meta-analysis, whereas the other author reviewed a 20% random selection for quality assessment scoring and another 20% random selection for extraction of data. Appendix Tables 1 and 2 present the relevant features of all randomized, controlled trials with a placebo group. Table 1 provides details of patients at trial entry (age, sex, race) and key study features (sample size, percent dropouts, design, quality assessment score, geographic location, and source of support). Appendix Table 4 outlines the treatment received by patients in each trial (including diet, NSAID type and duration of use, and antihypertensive type and duration), and the average difference (in mm Hg) between mean blood pressure recorded on NSAID treatment and control treatment, respectively, adjusted for baseline mean blood pressure ([NSAID treatment mean blood pressure -baseline mean blood pressure] [control treatment mean blood pressure baseline mean blood pressure]). Appendix Tables 3 and 4 present similar information for randomized, controlled trials without a placebo group. Because only two trials [41, 51] used ambulatory blood pressure monitoring (both with same small sample sizes: n = 12) and all other trials assessed blood pressure using standardized mercury sphygmomanometers, the method of blood pressure measurement was not considered in the analysis. Table 1. Meta-analysis of Randomized, Placebo-Controlled Trials Appendix Table 1. Study Features of Randomized, Controlled Trials *Table OMITTED Appendix Table 2. Treatment Received in Randomized, Controlled Trials *Table OMITTED Appendix Table 3. Study Features of Randomized Trials without a Placebo Group *Table OMITTED Appendix Table 4. Treatment Received in Randomized Trials without a Placebo Group *Table OMITTED Combining the Trials The randomized, controlled trials assessing the effect of NSAIDs on blood pressure [14-51] may be readily subdivided into two broad categories (Appendix Table 3): those of predominantly white, middle-aged patients of both sexes with mild to moderate uncomplicated essential hypertension [29 trials] and those of predominantly white, healthy, young volunteers of both sexes with no history of hypertension and with normal blood pressure (15 trials). These broad categories may be further subdivided (Appendix Tables 1 and 2) into three groups of trials in hypertensive patients and two groups of trials involving normal volunteers based on whether or not antihypertensive therapy was coadministered and on its duration. Because all these trials were randomized and controlled, measured the effect of NSAIDs on blood pressure in patients with uncomplicated mild to moderate essential hypertension or healthy volunteers, and had no other apparent systematic differences between the trials in each of the five groups, we considered it clinically appropriate to pool their results. To assess whether these trials could be combined statistically, the randomized, contro

Blood pressure is linked to salt intake and modulated by the angiotensinogen gene in normotensive and hypertensive elderly subjects.

Objectives To evaluate salt sensitivity in elderly subjects with different forms of hypertension and controls and to investigate any modulation by genotype Design Randomized, double-blinded, placebo-controlled latin-square Setting Tertiary referral hospital Participants Community subjects (n = 46) aged ⩾ 60 years classified as isolated systolic hypertension [ISH; systolic blood pressure (SBP) ⩾ 160, diastolic blood pressure (DBP) < 90 mmHg, n = 19], diastolic ± systolic hypertension (SDH; DBP ⩾ 90 mmHg, n = 10) and normotension (SBP < 160, DBP < 90 mmHg, n = 17). Intervention Four 14 day treatments, 50, 100, 200 and 300 mmol/day of sodium chloride supplementation interspersed with 14 day washout periods on a salt-restricted diet. Main outcome measures The 24 h blood pressure, heart rate, weight, urinary sodium and creatinine clearance measured during baseline, treatment and washout periods and angiotensinogen (AGT) and angiotensin converting enzyme (ACE) genotypes. Results For the entire cohort, the mean ± standard error (SE) of change from baseline in SBP for 50, 100, 200 and 300 mmol/day salt was 7.7 ± 2.4, 12.1 ± 2.4, 16.6 ± 3.0, 18.5 ± 2.6 mmHg, respectively. For DBP, the respective changes were:− 0.1 ± 1.5, 2.4 ± 1.6, 3.0 ± 1.5, 5.8 ± 1.7 mmHg. The increase in SBP among ISH subjects was significantly higher than among subjects in the SDH and normotensive groups (P < 0.05). AGT genotype influenced the effect of salt dose on the change in DBP (P = 0.006) but not SBP (P = 0.7). Conclusions In healthy, older subjects, a linear increase in BP occurred with increasing salt dose, it appeared most pronounced in ISH subjects and could be modulated by AGT genotype.

The Problems and Pitfalls of NSAID Therapy in the Elderly (Part I)

SummaryNonsteroidal anti-inflammatory drugs (NSAIDs) are the most commonly prescribed drugs worldwide when grouped by generic categories and account for 3 to 9% of total prescription numbers in various countries. While NSAIDs are responsible for approximately 25% of all reported adverse drug reactions, aging may substantially increase the risk of NSAID-induced reactions.Several factors may contribute to NSAID-related toxicity in the elderly. The increase in morbidity associated with aging may result in consumption of a wide range of potent drugs, while inappropriate drug therapy and aberrant compliance are also capable of contributing to adverse drug reactions in geriatric patients. Age-related alterations in pharmacokinetics may influence the handling of NSAIDs in the elderly; in particular, dosage reduction is appropriate for azapropazone (apazone), naproxen, ketoprofen and salicylates administered to healthy aged patients, whereas the presence of renal disease may also necessitate dosage reduction of diflunisal, indomethacin, sulindac and mefenamic acid. Changes in NSAID pharmacodynamics with aging, such as increased CNS sensitivity to NSAIDs and impaired homeostasis, also predispose the elderly to NSAID-related adverse effects.It is undisputed that gastrointestinal toxicity due to NSAID therapy is a class effect. A significant association has been found between aspirin and uncomplicated gastric, but not uncomplicated duodenal ulcer, while nonaspirin NSAIDs are significantly associated with both uncomplicated gastric and duodenal ulceration. The use of NSAIDs is accompanied by a 2- to 5-fold risk of serious complications of peptic ulcer disease, i.e. haemorrhage or perforation, which increases in the elderly, particularly women. A broad range of renal side effects has been ascribed to NSAIDs, of which acute renal impairment is the most common in the elderly. Although most NSAIDs have been reported to cause hepatotoxicity, serious abnormalities of liver function are rare and are largely unpredictable. Other adverse effects due to NSAIDs have also been described, some of which (e.g. cardiovascular, CNS and haematological effects) may be more common in the elderly.

Arginine vasopressin and osmolality in the elderly

Objectives: To evaluate the influence of age on plasma arginine vasopressin (AVP) concentrations and the relationship between plasma AVP and serum osmolality in younger and older subjects, and in the elderly, to assess the effect of gender on plasma AVP concentration and to determine the impact of prostaglandin blockade on renal responsiveness to AVP.

Adverse drug interactions with nonsteroidal anti-inflammatory drugs (NSAIDs) : recognition, management and avoidance

SummaryThe prevalence and incidence of adverse drug interactions involving nonsteroidal anti-inflammatory drugs (NSAIDs) remains unknown. To identify those proposed drug interactions of greatest clinical significance, it is appropriate to focus on interactions between commonly used and/ or commonly coprescribed drugs, interactions for which there are numerous well documented case reports in reputable journals, interactions validated by well designed in vivo human studies and those affecting high-risk drugs and/or high-risk patients.While most interactions between NSAIDs and other drugs are pharmacokinetic, NSAID-related pharmacodynamic interactions may be considerably more important in the clinical context, and prescriber ignorance is likely to be a major determinant of many adverse drug interactions. Prescribing NSAIDs is relatively contraindicated for patients on oral anticoagulants due to the risk of haemorrhage, and for patients taking high-dose methotrexate due to the dangers of bone marrow toxicity, renal failure and hepatic dysfunction. Combination NSAID therapy cannot be justified as toxicity may be increased without any improvement in efficacy. Where lithium or anti-hypertensives are coprescribed with NSAIDs, close monitoring is mandatory for lithium toxicity and hypertension, respectively, and aspirin (acetylsalicylic acid) or sulindac are preferred. Phenytoin or oral hypoglycaemic agents may be administered with NSAIDs other than pyrazoles and salicylates provided that patients are monitored carefully at the initiation and cessation of NSAID treatment. Digoxin, aminoglycosides and probenecid may be coprescribed with NSAIDs, but close monitoring is required, particularly for high-risk patients such as the elderly. Indomethacin and triamterene should be avoided due to the risk of renal failure. High dose aspirin should be replaced by naproxen in patients on valproic acid (sodium Valproate) and care is required when corticosteroids are administered to patients taking salicylates long term in high dosage. Interactions between NSAIDs and antacids or cholestyramine are generally avoidable.Adverse drug interactions involving NSAIDs may be limited by rational prescribing and by careful monitoring, particularly for high-risk patients, drugs and therapy periods.

Evaluation of the potential interaction between NaCl and prostaglandin inhibition in elderly individuals with isolated systolic hypertension.

OBJECTIVEnTo evaluate whether prostaglandin inhibition with the non-steroidal anti-inflammatory drug (NSAID), indomethacin (I) interacts synergistically with different doses of salt (NaCl) in elevating systolic blood pressure (SBP).nnnDESIGN AND METHODSnThis randomized, placebo-controlled, double-blind, crossover study examined the interaction between NaCl and the prostaglandin inhibitor, I in 31 healthy elderly individuals with a mean age (+/- SD) of 68.7+/-5.7 years (range 61-85 years). Participants aged more than 60 years on a 140 mmol/day NaCl dose for 6 weeks were chosen with normal blood pressure [24-h SBP <148 mm Hg, diastolic blood pressure (DBP) <85 mm Hg on the Takeda Ambulatory Blood Pressure Monitor (TABPM); n = 15] and isolated systolic hypertension (ISH), [24-h SBP >148 mm Hg, 24-h DBP <85 mm Hg on TABPM; n = 16]. Exclusion criteria included uncontrolled hypertension (SBP >220 mm Hg and/or DBP >110 mm Hg), cardiac disease, creatinine clearance <60 ml/min, dementia and recent cerebrovascular accident or secondary hypertension. A 2x2 Latin square design was structured using four treatment groups [low salt (NaCl = 90 mmol/day) + I placebo, high salt (NaCl = 240 mmol/day) + I placebo, low salt + I (25 mg three times daily) and high salt + I] for 2 weeks each, balanced and interspersed with 2 week washout periods to minimize carryover effects. Twenty-four hour SBP, DBP and heart rate were measured and summarized using a moving interval averaging technique. The mean change in 24-h SBP, DBP, heart rate, urinary Na+, K+, protein and creatinine, creatinine clearance and serum electrolytes were compared across treatments in the total cohort and in ISH and control groups separately using ANCOVA (SAS).nnnRESULTSnIn the total cohort, compared with low NaCl, chronic high NaCl increased mean SBP (5.76 mm Hg; P = 0.0002) and DBP (3.36 mm Hg; P = 0.002). Indomethacin significantly increased mean SBP (2.66 mm Hg, P = 0.015) but not DBP (0.31 mm Hg, P = 0.419). High salt and I were additive (SBPT, DBPT) but there was no interaction (P = 0.795 and P = 0.739, respectively). Additionally, chronic high NaCl increased serum Na (P = 0.0001) and 24-h urinary Na (P = 0.0001) as expected. Indomethacin significantly decreased mean heart rate (P = 0.018). The effects of NaCl and I on SBP, DBP and heart rate were not modified by age, alcohol intake, serum K+, body mass index or treatment order. In the ISH group, NaCl dose significantly elevated SBP (9.87 mm Hg; P = 0.0001) and DBP (5.26 mm Hg, P = 0.006) but did not significantly alter blood pressure in the normotensive group. Indomethacin significantly elevated SBP (P = 0.03) in normotensive individuals but had no effect on blood pressure in the ISH group.nnnCONCLUSIONSnChronic high salt diet elevated blood pressure more than I in the total cohort of elderly individuals. No interaction was demonstrated and their effects were additive. In the ISH group, chronic high salt diet significantly increased SBP and DBP while I failed to alter blood pressure. In the normotensive group, I, but not salt, elevated SBP. Patients with ISH are sensitive to the pressor effect of NaCl but resistant to the pressor effect of prostaglandin inhibition in contrast to elderly normotensive control individuals where the reverse was found.