Current issues of ACP Journal Club are published in Annals of Internal Medicine


Review: Serum insulin levels are associated with hypertension

ACP J Club. 1993 Jan-Feb;118:27. doi:10.7326/ACPJC-1993-118-1-027

Source Citation

Denker PS, Pollock VE. Fasting serum insulin levels in essential hypertension. A meta-analysis. Arch Intern Med. 1992 Aug;152:1649-51.



To determine if an association exists between fasting serum insulin levels and blood pressure through a meta-analysis of pooled data from relevant studies.

Data sources

The MEDLINE database was searched for the years 1983 to 1991 for studies in English using the terms insulin and hypertension. Reference lists were scanned for additional citations.

Study selection

Studies were included if the enrolled patients had untreated essential hypertension or treated essential hypertension with an adequate washout phase. Data from patients with overt diabetes mellitus, impaired glucose tolerance, or serious intercurrent illness were excluded. 11 studies containing adequate data on the association between hypertension and serum insulin were selected from the 49 identified articles.

Data extraction

Data on fasting serum insulin and glucose levels, blood pressure, and patients' age and body mass index were extracted. Effect sizes (the difference between 2 group means divided by their pooled standard deviation) weighted by sample size were calculated for fasting serum insulin levels associated with diastolic blood pressure (DBP) and systolic blood pressure (SBP). An overall estimate of the mean effect size was calculated from the 11 studies. Results were tested for homogeneity across studies.

Main results

5858 patients were included in the 11 analyzed studies. In univariate analysis the overall mean effect size for the relation between serum insulin and blood pressure was 0.28 (95% CI 0.21 to 0.34) for DBP and 0.23 (CI 0.18 to 0.27) for SBP. Study results were heterogeneous for both DBP and SBP (P < 0.001), implying significant variation of effect sizes across studies. Age and body mass index were also associated with DBP (effect sizes 0.45 [CI 0.31 to 0.58] and 0.64 [CI 0.38 to 0.90], respectively); again the study results were heterogeneous (P < 0.001). Fasting plasma glucose level was not related to either DBP (effect size 0.12, CI -0.03 to 0.26, P > 0.05) or SBP (effect size 0.13, CI -0.02 to 0.28, P > 0.05).


Fasting serum insulin levels are associated with higher diastolic and systolic blood pressure measurements in normoglycemic persons with essential hypertension. The association may not be independent of the effects of other factors including age and body mass index.

Source of funding: Not stated.

For article reprint: Dr. P.S. Denker, Joslin Diabetes Clinic, Morton Plant Hospital, Clearwater, FL 34616, USA. FAX 813-493-2307.


The hypothesis implicating hyperinsulinemia (or insulin resistance, or both) as an important cause of essential hypertension is currently popular but unconfirmed. Confirmation of the causal mechanism would have a major effect on the clinical management of essential hypertension and the strategies for its prevention. Although the meta-analysis of the association between fasting serum insulin and blood pressure by Denker and Pollock provides an overview of recent studies, its clinical relevance is tempered by the limitations of the study.

First, insulin and blood pressure data from large population surveys, such as the Paris Prospective Study, Helsinki Policeman Study, and San Antonio Heart Study, were not included in the meta-analysis, presumably because of difficulty in extracting the required data. The exclusion of such important studies may bias or dilute the results of the meta-analysis. Second, although levels of both insulin and blood pressure are affected by variables such as age, body mass, and body fat distribution, data from the individual studies in the meta-analysis were not adjusted to control for such confounding variables.

Third, the 2 largest studies in the meta-analysis (from Israel and the Pacific Islands) included more than 5000 participants, whereas the remaining 9 studies included a total of only 671 participants. Because ethnic background appears to be an important determinant of the insulin-blood pressure relation, results based primarily on 2 relatively narrow population samples may have little relevance to other ethnic groups. In addition, the small number of persons in the remaining 9 studies limits the potential for subgroup analysis.

In summary, the results of this meta-analysis represent a small addition to the body of evidence supporting a causal role for insulin in essential hypertension. Most important, this study highlights the need for additional epidemiologic, clinical, and basic science investigation if the complex relation between insulin metabolism and blood pressure regulation is to be clarified.

Paul Levinson, MD
Memorial Hospital of Rhode IslandPawtucket, Rhode Island, USA