Article Text

other Versions

Cohort study
Lower blood pressure associated with higher mortality in retrospective study of patients with newly diagnosed type 2 diabetes
  1. Richard W Grant1,
  2. Deborah J Wexler2
  1. 1Division of Research, Kaiser Permanente Northern California, Oakland, California, USA
  2. 2Diabetes Center, Department of Medicine, Massachusetts General Hospital, Boston, Massachusetts, USA
  1. Correspondence to: Dr Richard W Grant
    Division of Research, Kaiser Permanente Northern California,  2101 Webster, Oakland, CA 93452, USA; richard.w.grant{at}

Statistics from

Commentary on Vamos EP, Harris M, Millett C, et al. Association of systolic and diastolic blood pressure and all cause mortality in people with newly diagnosed type 2 diabetes: retrospective cohort study. BMJ 2012;345:e5567


Cardiovascular disease (CVD) is the leading cause of death in type 2 diabetes, which has led to increased focus on CVD risk reduction for patients with diabetes that emphasises blood pressure and cholesterol control.1 For most patients, clinical trial evidence supports treatment up to <140/90 mm Hg. In patients with increased CVD risk, such as the additional risk conferred by diabetes, the recommendation has been to treat up to <130/80 mm Hg, with the rationale that lower was better (which has proven more or less true for low-density lipoprotein (LDL lowering)). However, treatment at lower blood pressure levels is not without risk, particularly in older patients and those with multiple concurrent conditions, thus careful consideration of treatment goals is necessary.


In this paper by Vamos and colleagues, the relationship between blood pressure level achieved in the first year after diabetes diagnosis and mortality over the next 3.5 years was examined. This was a retrospective cohort analysis of primary care patients whose data were prospectively entered into the UK General Practice Research Database. Relative death rates were compared by different levels of mean blood pressure levels in the first year, with results stratified by the presence of baseline CVD and examined separately for systolic and diastolic blood pressures. The very first reading for each patient was excluded to eliminate the bias towards the null that can happen when values are selected because they are above the goal. Mortality was derived from the UK registry, and potential baseline confounders such as age, gender, weight and socioeconomic status were controlled for using multivariate models.


The primary finding was the U-shaped relationship in univariate analyses between mean blood pressure in the first year and subsequent mortality, with levels either above or below 130–139 mm Hg systolic and 80–85 mm Hg diastolic associated with increased mortality. In adjusted analyses in both treated and untreated patients, the very lowest levels of blood pressure control (<120 mm Hg systolic and <75 mm Hg diastolic) but not the highest levels of systolic and diastolic blood pressure were associated with increased mortality. The authors conclude that ‘lower is better’ does not appear to hold true for blood pressure control in diabetes.


There are several critical points to consider regarding this analysis. First, the death rate was extremely high, with 29% of CVD patients and 19% of non-CVD patients dead within 3.5 years of diabetes diagnosis, a death rate higher than most diabetes clinical trials.2 The data were collected beginning in 1990, when screening and treatment practices were very different, and suggest that patients were much sicker and presumably had longer duration of (undiagnosed) diabetes and less effective risk factor control as newly diagnosed patients would have today. This is an important distinction because low blood pressure can be a consequence rather than a cause of impending mortality and because the benefit of ‘tight’ blood pressure control probably requires decades to accrue. Second, a careful distinction must be made between ‘treated’ and ‘untreated’ blood pressure levels as 77% of CVD and 51% of non-CVD patients were treated for hypertension. Detailed medication data are not given, and some medications used during this period, such as rosiglitazone for glycaemia, may have increased mortality. Finally, despite excluding patients with known heart failure, the very low blood pressure levels and high death rate in this obese diabetes cohort suggest that undiagnosed chronic heart failure may have been an unmeasured confounder.

The finding that very low blood pressure is associated with higher mortality risk is congruent with the results of ACCORD-BP and with a closer examination of clinical trial evidence in patients with diabetes.3 However, an equally valid conclusion from the Vamos study is that high blood pressure (even >160/95 mm Hg among treated patients) is not associated with increased mortality, a conclusion that directly contradicts clinical trials and underscores the fact that it can be difficult to infer treatment guidelines from observational data. The lesson here may be that while evidence-based guidelines are useful at the population level, treatment decisions must be made on an individual basis. Thus, a younger man with newly diagnosed diabetes will probably benefit from careful blood pressure control over the next several decades of life, whereas an older and frailer man may incur greater risks with little benefit from a similar approach. Further research is now needed to guide more individualisation of treatment goals to maximise the benefits of various available treatments.


View Abstract


  • Competing interests None.

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.