Association of history of metformin use with delirium and mortality: A retrospective cohort study

INTRODUCTION

Delirium is a severe medical illness, and its prevalence is increasing as our society ages (1-4). Delirium is associated with poor outcomes including extended length of stay, institutionalization after discharge from the hospital, and high mortality (1-3, 5). The risk of delirium increases with age, and also with medical conditions including infection and after surgery (1-4). At present, there is no solid understanding of the pathogenesis of delirium, and thus we do not have therapeutic or preventative methods to effectively improve care of patients with delirium. It has been shown that commonly prescribed antipsychotic medications are not helpful for treatment or prevention (6-8), and novel viewpoints to investigate this devastating illness are warranted.

Additional major risk factor for delirium is baseline dementia (4). Worse, after delirium, it is known that cognitive function further declines and dementia progression accelerates (9). In the recent literature, there is evidence showing that type 2 diabetes mellitus (DM) may share a key process of pathophysiology with dementia. DM and high glucose levels have been tied to increased cognitive decline and risk for dementia (10-12). This suggests that vascular and cellular damage induced by high blood glucose may mediate common pathological processes leading to dementia onset and progression (13). These evidences suggest that DM also increases the risk of delirium potentially through common underline mechanisms that increase dementia risk. However, the literature is not consistent with regard to the association between DM and delirium despite the close relationship between delirium and dementia (14, 15).

Of interest, among various anti-diabetic medications, it has been shown that metformin may decrease the risk of various forms of dementia, including Alzheimer’s disease (16-19), although the results from various studies show inconsistency (20). Because DM is associated with increased risk of dementia and cognitive decline, the association between anti-diabetic medication use and decreased risk of dementia was thought to be due to the better control of hyperglycemic mechanisms that could be a part of the pathogenesis of cognitive decline (21).

However, when metformin was compared to other anti-diabetic medications such as sulfonylureas (acetohexamide, chlorpropamide, glimepiride, glipizide, glyburide, tolazamide, and tolbutamide), the benefit for decreased risk of dementia and/or mortality was superior with metformin (18, 19, 22-24). To date, although multiple studies have replicated data showing that metformin seems to have benefits for decreased risk of dementia and mortality, there is very limited data about the potential role of metformin and its association with delirium, mortality, and DM.

Thus, in this report we aimed to investigate the relationship between DM and delirium risk with a focus on the influence from metformin. We hypothesized that history of metformin use is associated with lower risk for delirium. We were also interested in testing if history of metformin use can alter one of the most important patient outcomes, mortality.

METHODS

RESULTS

Delirium, DM, and history of metformin use

The prevalence of delirium in the DM-without-metformin group (36.0%) was significantly higher than it was in the non-DM group (27.7%) (p = 0.048, Chi-square test corrected by Holm method) (Figure 1). The prevalence of delirium in the DM-with-metformin group (29.2%) was lower than it was in the DM-without-metformin group (36.0%), but this result was not statistically significant (p = 0.25, Chi-square test corrected by Holm method) (Figure 1). Logistic regression analysis using only DM subjects showed that a history of metformin use reduced the risk of delirium in patients with DM (OR = 0.50, 95% CI: 0.32–0.79, p = 0.003) even after controlling for age, sex, dementia status, BMI, and history of insulin use (Table 2). When logistic regression analyses were performed separately for the DM subjects with dementia and DM subjects without dementia, the results showed that metformin use history reduced risk of delirium both in non-dementia group (OR: 0.54, 95%CI: 0.32–0.91, p = 0.02) (Supplementary Table 1) and dementia group (OR: 0.40, 95%CI: 0.13–1.21, p = 0.11) (Supplementary Table 2) after controlling for age, sex, CCI, BMI, and history of insulin use, although dementia group did not reach statistically significant level likely due to reduced sample size.

View this table:

• View inline
• View popup
• Download powerpoint

Table 2: Result of the Logistic regression in subjects with diabetes (N=506)

• Download figure
• Open in new tab

Figure 1

Prevalence of delirium by comparing three patient groups based on their DM status and history of metformin use. The Chi-square test corrected by Holm method showed significant difference between non-DM group and DM-without-metformin group (p=0.048).

Mortality risk factors; benefit of history of metformin use

First, 3-year mortality was compared between the following three groups: the non-DM group, DM-without-metformin group, and DM-with-metformin group. Mortality for the DM-without-metformin group (survival rate: 0.595, 95% CI: 0.512–0.669) was significantly higher than mortality for the non-DM group (survival rate: 0.715, 95% CI: 0.672–0.753) (p = 0.0036, log-rank test corrected by Holm method). On the other hand, mortality for the DM-with-metformin group (survival rate: 0.695, 95% CI: 0.604–0.770) was significantly lower than mortality for the DM-without-metformin group (p = 0.035, log-rank test corrected by Holm method). The mortality for DM-with-metformin group is almost exactly as good as that of non-DM patients (p = 0.91, log-rank test corrected by Holm method) (Figure 2). The same results were replicated in younger (age < 65) subgroup, aged (age ≥ 65) subgroup, female subgroup, male subgroup, non-dementia subgroup, and non-delirium subgroup (Supplementary Figures 2A-D, 3A, and 3C). Intriguing differences were found in dementia subgroup and delirium subgroup. Although it did not reach statistical significance, the mortality among DM patients who have used metformin were lower even when compared to that of non-DM group (Supplementary Figure 3B and 3D). Cox proportional hazard model showed that history of metformin use significantly decreased risk of 3-year mortality after adjustment for age, sex, CCI, BMI, history of insulin use, and delirium status (HR = 0.69, 95%CI: 0.48–0.98, p = 0.038) (Table 3). When cox proportional hazard models were performed separately for the DM subjects with dementia and DM subjects without dementia, the results showed that metformin use history did not reduce risk of mortality in non-dementia group (HR: 0.89, 95%CI: 0.59–1.35, p = 0.59) (Supplementary Table 3), but metformin exposure reduced risk of mortality among dementia group (HR: 0.40, 95%CI: 0.18– 0.87, p = 0.02) (Supplementary Table 4). The GLOBAL tests for the proportional hazards assumptions were not statistically significant (Table 3, Supplementary Table 3, and Supplementary Table 4).

View this table:

• View inline
• View popup
• Download powerpoint

Table 3:

Result of the Cox proportional hazard model in subjects with diabetes (N=506)

• Download figure
• Open in new tab

Figure 2

Kaplan-Meier cumulative survival curve over 3 years based on the three-group comparison.

DISCUSSION

This large-cohort study examined whether history of DM as well as metformin use are associated with delirium. Another investigation was if DM and history of metformin use are altering mortality risk. In the data presented here, higher prevalence of delirium and increased mortality were observed in DM patients without a history of metformin use compared to non-DM patients. On the other hand, DM patients with a history of metformin use showed lower prevalence of delirium in our data, and this could be a reason why past literature investigating relationship between DM and delirium was inconclusive, as most likely those subjects categorized with DM included those who were on metformin, and thus had less prevalence of delirium in average over DM patients with and without metformin use. Additionally, we found that a history of metformin use was associated with decreased risk of delirium and mortality even after adjusting for many potential confounding variables. Our data indicate the potential positive benefit of metformin on delirium risk and mortality, providing insights about additional pathophysiological mechanisms of delirium and potential therapeutic and preventative opportunities for this devastating illness commonly seen among the aged population. However, it still remains unclear whether diabetes increases the risk of delirium or mortality as we have not analyzed the direct effect of diabetes on the risk of delirium and mortality with prospective clinical study including various important confounding factors.

The association observed here in this report between DM and delirium itself is not new, although previous data were mixed (14, 15). However, the preventative role of metformin related to delirium is unique, and this is the first report showing such potential relationship. There has been little study investigating relationship between metformin and delirium. One study, which used the U.S. Food and Drug Administration adverse events reporting system (FAERS), reported metformin as a potential delirium-inducing drug, contrary to our result (40). The discrepancy in these results might be due to methodological difference because the data in the FAERS are not designed to specifically investigate relationship between metformin and delirium but rather screening of large variety of commonly prescribed medications (40). To solve this inconsistency, a future randomized clinical trial to test metformin for its effect on delirium risk would be of importance.

In addition, metformin has been reported to be beneficial for survival among various patient populations including cancer patients (41-43). Our data presented here showed its potential benefit for survival regardless of delirium status. It is of importance to note that DM patients without a history of metformin use showing higher mortality was not simply due to their having a severe form of DM, because subjects in this group are defined as those who have never been on metformin, and they are not the group that was diagnosed with DM, treated with metformin first, and was switched to insulin due to their poor control of DM while they were on metformin. In fact, in our data set, BMI and insulin use ratio were lower in DM-without-metformin group than ones in DM-with-metformin group, suggesting that DM-without-metformin group had potentially less severe diabetes.

Basic research studies have shown that metformin appears to target a number of aging-related mechanisms (44, 45). It was reported that metformin influenced pro-inflammatory cytokines such as IL-6 and TNF-alpha to suppress inflammation through modulating the NF-kB pathway (46, 47). Furthermore, metformin was reported to activate AMPK and inhibits mTOR (48, 49), which is known to influence the aging process (45, 49, 50). It is possible that these effects by metformin on inflammation and pathways involving AMPK and mTOR can decrease delirium risk and prolong patient’s life, although evidence for these effects in humans are limited.

Our data showed this potential benefit of metformin. The question is whether people without DM should start taking metformin. It is an important question that should be carefully explored, as metformin can have problematic side effects including vitamin B12 deficiency if taken for the long term (51). Then the next question is if patients at risk for delirium such as those going through major surgery (e.g., cardiac, orthopedic, or neurosurgery) should take metformin even for a short period of time preoperatively and postoperatively. This needs to be answered by future clinical trials, and we believe these are worth conducting to improve our patient care.

LIMITATIONS

We acknowledge the following limitations in this report. First, because history of metformin use was obtained from hospital records by retrospective chart review, it is possible to include false-positive cases when patients were prescribed with metformin but did not take it because of intolerable side effects or non-adherence to it, and false-negative situations such as metformin prescribed by other providers outside of our hospital network. Second, the definition of delirium is not based on the gold standard psychiatric assessment based on DSM-5 criteria (52). However, our categorization methods have been effective enough to show discrete mortality risk based on our clinical classification of delirium as shown in our previous reports (25-29). Also, Inouye et al. reported that retrospective chart review can capture delirium with reasonably good sensitivity and specificity (38). Third, the definition of DM is based on chart review as well, although our classification of DM clearly differentiates mortality risk, supporting the validity of our approach. Fourth, we did not control for metformin dose or duration. Total dose information might be important because duration or dose of metformin can alter its effect, as a previous animal study showed (53). Fifth, other anti-diabetic medication other than insulin were not recorded and incorporated in our investigation, as it has been repeatedly reported that when compared to metformin, other anti-diabetic medications did not show benefit for mortality (22-24). Lastly, our data does not necessarily show causal relationship of metformin use and risk for delirium or mortality. However, metformin was used prior to the occurrence of delirium as well as death, suggesting a strong possibility of a beneficial effect from metformin in decreasing risk for delirium and increasing chance of survival. To address this question more precisely, prospective clinical trials are needed. Despite all these potential limitations, we found significant associations among metformin, DM, and delirium, as well as all-cause mortality.

CONCLUSION

In this report, we showed the potential benefit of metformin in decreasing the risk of delirium and mortality.

Funding / Support

None.

Declaration of Interest

Gen Shinozaki is co-founder of Predelix Medical LLC and has pending patents as follows: “Non-invasive device for predicting and screening delirium”, PCT application no. PCT/US2016/064937 and US provisional patent no. 62/263,325; “Prediction of patient outcomes with a novel electroencephalography device”, US provisional patent no. 62/829,411; “Epigenetic Biomarker of Delirium Risk” in the PCT Application No. PCT/US19/51276, and in U.S. Provisional Patent No. 62/731,599. All other authors have declared that no conflict of interest exists.

Author Contributions

T.Y. organized and analyzed data, and edited of the final form of the manuscript. Z.E.A., M.M., G.C., and T.T. collected the clinical data, organized it and analyzed for initial preparation, and wrote the initial draft of the manuscript. P.S.M., N.E.W., K.J.C., E.J.S., S.S.J., K.R.C., C.C.A., and A.A.M. collected the clinical data. S.L. and E.S. organized data collection process and critically reviewed the manuscript. M.I. critically reviewed the manuscript. H.R.C. analyzed data. G.S. conceived ideas of the study, planned its design and coordination, wrote the initial draft and edited the final form of the manuscript.

Access to Data and Data Analysis

G.S. had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis

Data Availability

The data that support the findings of this study are available from the corresponding author, G.S., upon reasonable request.