Intended for healthcare professionals

Practice Uncertainties Page

Should we prescribe diuretics for patients with prediabetes and hypertension?

BMJ 2008; 337 doi: (Published 21 August 2008) Cite this as: BMJ 2008;337:a679
  1. Bruce Arroll, professor ,
  2. Timothy Kenealy, associate professor ,
  3. C Raina Elley, senior lecturer
  1. 1Department of General Practice and Primary Health Care, University of Auckland, Private Bag 92019, Auckland, New Zealand 1142
  1. Correspondence to: B Arroll b.arroll{at}
  • Accepted 4 April 2008

The uncertainty arises from the findings of the large randomised controlled ALLHAT 2002 trial (31 512 people 55 years or older with hypertension and one other risk factor for cardiovascular disease), in which the thiazide-like diuretic chlortalidone seemed to increase some cardiac risk factors, including the rate of developing diabetes. After four years of follow-up in those who had a normal fasting blood glucose at baseline, 302 (11.6%) people taking chlortalidone, 154 (9.8%) of those taking amlodipine (a calcium channel blocker), and 119 (8.1%) of those taking lisinopril (an angiotensin converting enzyme inhibitor) had fasting blood glucose concentrations of ≥7 mmol/l (P<0.001).1 Diuretics are known to achieve long term cardiovascular results as good as, or better than, alternative antihypertensives—at least in people aged 55 years or older. But clinicians may not initially treat patients with prediabetes and hypertension with a diuretic because they worry that the induced diabetes might result in a worse outcome than if they prescribed a different class of antihypertensive.

What is the evidence of the uncertainty?

International guidelines give conflicting advice, presumably because of this uncertainty. The National Institute for Health and Clinical Excellence (NICE) advises general practitioners in the United Kingdom to use diuretics and calcium channel blockers as first line treatments in patients over 55 years and angiotensin converting enzyme inhibitors in those under 55. Diuretics are not recommended in the younger group partly because of concern about patients developing diabetes.2 In the United States, the JNC 7 guidelines advise doctors to start with a diuretic when treating hypertension and do not see prediabetes or diabetes as reasons for choosing another drug class above thiazides.3

A recent systematic review showed that compared with diuretics, β blockers do not induce diabetes, and calcium channel blockers, angiotensin converting enzyme inhibitors, and angiotensin II blockers (in that order) may even reduce the incidence of diabetes.4 The absolute difference in the incidence of diabetes caused by antihypertensives is modest, at around 3.6%.2

On the other hand, a systematic review undertaken in the NICE guideline compared cardiovascular outcomes for five commonly used antihypertensive drugs. Its comparison of angiotensin converting enzyme inhibitors with thiazide diuretics found that the only statistically significant difference was an increase in stroke associated with angiotensin converting enzyme inhibitors (relative risk 1.13, 95% confidence interval 1.02 to 1.25). A similar comparison between calcium channel blockers and thiazide diuretics found that the only statistically significant difference was an increase in congestive heart failure associated with calcium channel blockers (1.38, 1.25 to 1.53). This suggests that diuretics are at least as good as angiotensin converting enzyme inhibitors and calcium channel blockers in terms of cardiovascular outcomes.

The large ALLHAT study found that patients with impaired fasting glucose (and those with normoglycaemia and known diabetes mellitus) taking the thiazide-like diuretic chlortalidone had cardiovascular outcomes as good as or better than those taking amlodipine or lisinopril.5 For chlortalidone versus amlodipine, coronary heart disease, and combined coronary heart disease (myocardial infarction, coronary revascularisation, and hospital admission for angina) outcomes favoured chlortalidone (1.73, 1.10 to 2.72; 1.37, 1.00 to 1.87). No difference was seen for stroke, congestive heart failure, or cardiovascular disease combined with end stage renal disease. In addition, no differences were seen for all outcomes, including coronary heart disease alone and coronary heart disease combined with stroke, congestive heart failure, cardiovascular disease, and end stage renal disease for chlortalidone versus lisinopril.

The 14 year follow-up of the SHEP study (the original randomised trial lasted 4.3 years) found that although diabetes increased in the chlortalidone group, the cardiovascular outcomes of those with diuretic induced diabetes were not significantly worse and their prognosis was better than those with pre-existing diabetes.6 However, the large confidence intervals mean a large deleterious effect of diuretic induced diabetes cannot be excluded.

Another systematic review found no difference in cardiovascular outcomes across classes of antihypertensive drugs in people with and without diabetes. Evidence that lower blood pressure goals resulted in larger reductions in total major cardiovascular events was limited.7 Lowering blood pressure may be more important than inducing diabetes, all else being equal.

Diabetes caused by diuretics may be different from naturally occurring diabetes. Although diuretics increase the incidence of diabetes, they seem to be effective at lowering blood pressure and have benefit on cardiovascular outcomes that may outweigh their diabetogenic tendency, at least in the short term (most cardiovascular randomised trials are less than five years in duration and we can only speculate about longer terms effects.)

Is ongoing research likely to provide relevant evidence?

The definitive answer to this question requires a randomised controlled trial in people under 55 that compares a thiazide with one or more antihypertensive agent in patients with both forms of prediabetes—impaired fasting glucose and impaired glucose tolerance. We searched the World Health Organization international clinical trials registry (, which lists trials at various stages of planning and conduct from most major trial registers in the world. A trial in Japan (diuretics in the management of essential hypertension study; /NCT00131846) is investigating this uncertainty in 30-79 years olds by comparing a thiazide diuretic with no diuretic. The primary outcome is new onset type 2 diabetes and secondary outcomes include cardiovascular events and death, but the results are several years away.

What should we do in the light of the uncertainty?

On the basis of current evidence, it is justifiable to use a thiazide diuretic as first line treatment in people with hypertension and prediabetes (especially in resource poor settings because thiazides are relatively inexpensive). In people without diabetes, people with impaired fasting glucose, and those who develop diabetes during trials, cardiovascular outcomes from treating hypertension with diuretics are at least as good or better than with other antihypertensives. These results have been collected for up to 14 years and are mainly from those aged 55 or over. For younger patients with prediabetes the uncertainty remains, but the data from older patients are encouraging, and the key problem will probably be ensuring a consistently low blood pressure rather than deciding which drugs should be used to obtain it.


Cite this as: BMJ 2008;337:a679


  • This is a series of occasional articles that highlights areas of practice where management lacks convincing supporting evidence. The series advisers are David Tovey, editorial director, BMJ Knowledge, and Charles Young, editor of BMJ Clinical Evidence, and editor in chief, BMJ Point of Care.

  • Contributors: BA had the original idea and wrote the first draft. TK and CE wrote subsequent drafts and added references and comments. BA is guarantor.

  • Competing interests: BA is on the advisory board for Pharmac educational seminars. Pharmac is the government funded pharmaceutical purchasing agency in New Zealand. He is also on the primary care committee of the Future Forum, an educational foundation funded by Astra Zeneca (UK). He has been paid to talk to Astra Zeneca (NZ) staff and has received conference expenses from Sanofi Aventis (NZ).

  • Provenance and peer review: Not commissioned; externally peer reviewed.


View Abstract

Log in

Log in through your institution


* For online subscription