Is blood glucose on admission a predictor of mortality in adult acute pneumonia?

Authors


  • Authorship and contributorship

    FF and DG analysed data and wrote the paper. AC and EP collected data and contribute to design. AB designed the study, analysed data and wrote the paper.

  • Ethics

    Research is based on administrative data treated anonymously in the analysis. Research was conducted in compliance with the requirements and under the authorisation of the local Institutional Review Board.

  • Conflict of interest

    The authors have stated explicitly that there are no conflicts of interest in connection with this article.

Correspondence

Dario Gregori, PhD, Laboratories of Epidemiological Methods and BioStatistics, Department of Environmental Medicine and Public Health, Via Loredan, 18, 35121

Padova, Italy

Tel: +39 049 8275384

Fax: +39 02 700445089

email: dario.gregori@unipd.it

Abstract

Background

Even if hyperglycaemia is often identified as an independent risk factor for developing respiratory tract infection, only few studies have investigated this relationship. The aim of this study is to investigate if plasma glucose on admission is related with in-hospital mortality among patients with pneumonia and to identify the glycaemic range with significant reductions of mortality risks in non-intensive care patients.

Methods

Data come from administrative records of 1018 non-intensive care patients hospitalised with diagnosis of pneumonia. For every patient, administrative records were linked with the plasma glucose. A multivariate logistic regression model was performed in order to evaluate the associations between in-hospital mortality and a set of demographic and clinical variables. Plasma glucose was added to the model as restricted cubic spline; risk estimates for hypoglycaemic and hyperglycaemic patients have been derived on the basis of this nonlinear model and presented with two values of odds ratio (OR).

Results

The minimal risk of in-hospital mortality was found at plasma glucose levels of mean 86 mg/dL [95% confidence interval (CI) 61–102]. The adjusted OR of deaths for plasma glucose on admission for hypoglycaemic patients (below 86 mg/dL) is 0.78 (95% CI 0.62–0.98) for each 10 mg/dL of decrease, whereas for hyperglycaemic patients (above 86 mg/dL), the OR is 1.33 (95% CI 1.07–1.66) for each 10 mg/dL of increase in plasma glucose.

Conclusions

Our observations suggest that in non-intensive care patients, hypoglycaemia, as hyperglycaemia, is associated with in-hospital mortality.

Ancillary