Inadvertent hypothermia and mortality in critically ill adults: Systematic review and meta-analysis

Abstract

Objective: Considering that inadvertent hypothermia (IH) is common in Intensive Care Unit (ICU) patients and can be followed by severe complications, this systematic review identified, appraised and synthesised the published literature about the association between IH and mortality in adults admitted to the ICU.

Data sources: By using key terms, literature searches were conducted in Pubmed, CINAHL, Cochrane Library, Web of Science and EMBASE.

Review methods: According to PRISMA guidelines, articles published between 1980-2016 in English-language, peer-reviewed journals were considered. IH was defined as core temperature of <36.5°C or lower, present on ICU admission or manifested during ICU stay. Outcome measure included ICU, hospital or 28-day mortality. Selected cohort studies were evaluated with the Newcastle-Ottawa Scale. Extracted data were summarised in tables and synthesised qualitatively and quantitatively, with adjusted odds ratios (ORs) for mortality being combined in meta-analyses.

Results: Eighteen observational studies met inclusion criteria. All of them had high methodological quality. In twelve out of fifteen studies, unadjusted mortality was significantly higher in hypothermic patients compared to non-hypothermic ones. Likewise, in thirteen out of sixteen studies, IH or lowest core temperature was independently associated with significantly higher mortality. High severity and long duration of IH were also associated with higher mortality. Mortality was significantly higher in patients with core temperature <36.0°C (pooled OR 2.093, 95% CI 1.704-2.570), and in those with core temperature <35.0°C (pooled OR 2.945, 95% CI 2.166-4.004).

Conclusions: These findings indicate that IH predicts mortality in critically ill adults and pose suspicion that this may contribute to adverse patient outcome.

Keywords: Core temperature; Critically ill; Hypothermia; Intensive care unit; Mortality; Outcome.