A systematic search was conducted across PubMed, Embase, Web of Science, and the Cochrane Library databases from their inception to October 20, 2025, to identify all studies examining the association between DM and mortality risk in patients with sepsis. The search strategy combined Medical Subject Headings (MeSH) terms with free-text keywords, with the core query defined as: (“Sepsis” OR “Septic shock” OR “Severe sepsis” OR “Bacteremia”) AND (“Diabetes” OR “Diabetes mellitus” OR “Type 2 diabetes” OR “Type 1 diabetes” OR “Hyperglycemia”) AND (“Mortality” OR “Death” OR ‘Survival’ OR “Prognostic”). Additionally, relevant reviews, references, and conference abstracts were manually searched to supplement any omitted literature. All searches were restricted to English-language publications without limitations on region, study type, or publication date. The specific search strategy is detailed in Supplementary Table S1.

Inclusion Criteria:

Exclusion Criteria:

Two authors independently screened the literature for inclusion by importing the literature into endnote according to the literature inclusion and exclusion criteria, the final included studies were used for data extraction using excel software and if there was a dispute about the literature screening then it would be discussed, or a third person would be sought to adjudicate. The extracted data contained basic characteristics of the study (first author, year of publication, country, study design), basic characteristics of the population (sample size, gender, mean age) and types of DM, mortality, diagnosis of sepsis.

The Newcastle-Ottawa Scale (NOS) (17) was used in this study to evaluate the quality of the included observational studies. The scale is categorized into three main dimensions based on the type of study (cohort study or case–control study): selection of study subjects (0–4 points), comparability of comparison groups (0–2 points) and outcome measures (0–3 points) out of a possible 9 points. A score of ≥7 is considered high quality, 5–6 is considered moderate quality and ≤4 is considered low quality. The two evaluators scored independently, and any disagreements in scoring were resolved through negotiation or referred to a third party for validation.

All statistical analyses were performed using Stata 15.0 software. This study employed a binary data analysis approach, with patient mortality as the dichotomous outcome. For each included study, the number of deaths and total sample size in both the diabetic and non-diabetic groups were extracted. The relative risk (RR) and its 95% confidence interval (95% CI) were calculated to assess the association between DM and the risk of death from sepsis. Heterogeneity was assessed via Cochran’s Q test and the I² statistic. A fixed-effects model was employed when I² < 50%, indicating low heterogeneity; a random-effects model was used when I² ≥ 50%, signaling substantial heterogeneity warranting further investigation into its sources. Subgroup analyses were conducted by country, study design, diagnosis of sepsis, mortality to explore potential differences in effects across groups. Sensitivity analyses were performed to assess result robustness by sequentially excluding individual studies to observe outcome stability, thereby evaluating each study’s impact on overall results. Finally, funnel plots and Egger’s test were used to assess publication bias, detecting any asymmetry that might indicate bias in the included studies, if the funnel plot is asymmetric, further assessment of the stability of the results will be conducted using trim and fill.

As shown in Figures 1, a total of 8,374 articles were retrieved from PubMed (n = 1,844), Embase (n = 4,940), the Cochrane Library (n = 198), and Web of Science (n = 1,394). After excluding 1,415 studies, 6,939 were discarded based on title and abstract review, and 7 were excluded after full-text assessment. Ultimately, 13 articles (18–30) were included.

A total of 13 articles were included in the analysis, involving 1,209,263 patients with sepsis across 12 cohort studies and 1 case–control study. Three studies primarily originated from China, two from Greece, two from the United States, and two from the Netherlands. The age range of participants was approximately 59–75 years. Sepsis diagnosis adhered to either the Sepsis-3 criteria or Sepsis-2 criteria. Key baseline characteristics are summarized in Table 1.

This study employed the NOS scoring system for evaluation, with 8 studies scoring 8 points and 5 studies scoring 9 points. The overall article quality was rated as high. Detailed scores are presented in Table 2.

Thirteen articles discussed the association between DM and mortality due to sepsis. Among these, Angriman et al. (20) reported 28-day and 90-day mortality rates; Lin et al. (23) reported 28-day, ICU, and in-hospital mortality rates; Stegenga et al. (27) reported 28-day and 90-day mortality rates; van Vught et al. (28) reported ICU, 90-day, and in-hospital mortality rates, while Zohar et al. (30) reported 28-day and 90-day in-hospital mortality rates. Heterogeneity testing (I² = 98.6%, p = 0.001) was performed using a random-effects model. Results (Figure 2) suggest DM increase mortality in patients with sepsis [RR = 1.25, 95% CI (1.14, 1.38)]. Due to substantial heterogeneity, sensitivity analysis was conducted by sequentially excluding studies. Results (Supplementary Figure S1) indicate stable findings unaffected by individual studies.

The study conducted subgroup analyses based on the type of mortality. The results (Figure 3) indicated that DM increase in-hospital mortality among patients with sepsis [RR = 1.60, 95% CI (1.11, 2.31), n = 5], while it had no significant effect on 28-day mortality [RR = 1.34, 95% CI (0.87, 2.07), n = 5], 90-day mortality [RR = 1.10, 95% CI (0.88, 1.36), n = 5], or ICU mortality [RR = 1.40, 95% CI (1.00, 1.97), n = 3]. The study conducted subgroup analyses based on the country. The results (Figure 4) indicated that DM increase the mortality of patients with sepsis in China [RR = 1.79, 95% CI (1.19, 2.70), n = 5], the Netherlands [RR = 1.14, 95% CI (1.12, 1.17), n = 5], and Israel [RR = 1.81, 95% CI (1.55, 2.12), n = 3]. The study conducted subgroup analyses based on the study design. The results (Figure 5) indicated that DM increase cohort study [RR = 1.25, 95%CI (1.13, 1.38)] mortality among patients with sepsis. The study conducted subgroup analyses based on the diagnosis of sepsis. The results (Figure 6) indicated that DM increase Sepsis-3 criteria [RR = 1.59, 95%CI (1.24, 2.04), n = 12] mortality among patients with sepsis.

This study employed funnel plots and Egger’s test to assess publication bias. Results (Supplementary Figure S2) indicate asymmetry in the funnel plot for “Association Between DM and Mortality due to sepsis,” with an Egger’s p-value of 0.003, suggesting a high likelihood of publication bias. Consequently, trim-and-fill methods were applied to evaluate result stability. Results (Supplementary Figure S3) demonstrate that despite the presence of publication bias, the findings remain robust.

The strengths of this study lie in its systematic and comprehensive approach. First, it includes a broad patient population and diverse study designs, ensuring high representativeness and broad external validity of the findings. By employing random-effects models to synthesize results across studies, we account for heterogeneity more effectively, yielding robust conclusions. Second, we conducted detailed subgroup analyses to explore the impact of various factors—including different types of mortality, geographic regions, study designs, and sepsis diagnostic criteria—on the association between DM and mortality risk. This approach provided a deeper understanding of DM’s role in sepsis pathogenesis and offered more targeted guidance for clinical practice. Finally, sensitivity analyses were performed to assess the influence of individual studies on the overall results, further enhancing the reliability and robustness of our findings.

Although this study provides valuable conclusions, several limitations remain. First, despite using a random-effects model to pool results, the inclusion of observational studies precludes complete elimination of potential confounding factors. For instance, factors such as glycemic control in patients with DM, underlying comorbidities (cardiovascular disease, kidney disease), infection type, and sepsis severity may influence mortality rates. However, most studies did not perform detailed adjustments for these variables. The lack of sufficient multivariable analysis prevented us from further controlling these potential confounders, potentially affecting the accuracy of the findings and the ability to draw causal inferences. Second, substantial heterogeneity existed between studies (I² = 98.6%), likely attributable to differences in patient populations, study designs, diagnostic criteria, and other variables. Although sensitivity analyses and subgroup analyses adjusted for some factors, heterogeneity remains a limitation of this review. Third, most included studies did not perform detailed stratified analyses based on DM type (type 1 or type 2 diabetes), disease duration, or treatment modalities. Consequently, we could not evaluate the specific impact of these factors on mortality due to sepsis risk. Finally, publication bias within the studies may also compromise the accuracy of conclusions, as studies reporting non-significant associations might have been excluded from the analysis, potentially leading to an overestimation of the results.

This study demonstrates an association between DM and increased mortality in patients with sepsis, with a particularly evident relationship for in-hospital mortality. However, the findings should not be interpreted as suggesting that all patients with DM face uniformly high mortality risk during sepsis. Rather, we emphasize the importance of individualized risk assessment in patients with DM with sepsis. The impact of DM on mortality risk may vary significantly depending on factors such as the type of DM (type 1 vs. type 2), the degree of glycemic control, the presence of comorbidities, and the management strategies employed during hospitalization. Thus, it is crucial to consider these factors when evaluating the prognostic of patients with DM in sepsis, rather than assuming a higher mortality risk universally.

These findings contribute to the existing evidence base by highlighting DM as an important comorbidity that should be considered as part of a personalized risk stratification approach in septic patients. However, due to the observational nature of the included studies, definitive conclusions regarding optimal clinical management strategies for patients with DM with sepsis cannot be drawn. Future research should focus on elucidating how DM-related characteristics, including glycemic control and comorbidities, influence sepsis outcomes, and how these factors can be incorporated into individualized management strategies.