Metformin and the risks of cellulitis, foot infections, and amputation in patients with type 2 diabetes

1. INTRODUCTION

Infection is an emerging complication of type 2 diabetes (T2D) in the 21st century.1 Chronic hyperglycemia and increased accumulated reactive oxygen species may deteriorate the immune function of patients with T2D and exacerbate the risk and severity of infection.1 Reports show that T2D is associated with 1.8 to 2.0 folds of cellulitis, 1.2 to 2.6 folds of pneumonia, 3.0 to 4.3 folds of urinary tract infection, and 2.0 to 3.3 folds of sepsis.2 Cellulitis is the deep dermal and subcutaneous infection caused by bacterial invasion through an impaired skin barrier. It is a common, potentially severe infection that has plagued humans for a long time.3,4 Old age, obesity, and diabetes increase the risk of cellulitis.4 The global number of deaths due to cellulitis increased 1.66 times from 42 555 in 1999 to 70 526 in 2019.5 About 60% of cellulitis occurs in the foot.6–8 Without proper treatment, foot infections may lead to foot ulcers; cellulitis may also spread to the bone leading to osteomyelitis, bacteremia, sepsis, and even leg amputation.7,8 Patients with T2D are reported to be at higher risk for foot infections and leg amputations.9 Appropriate treatment of cellulitis, foot ulcers, and infections through pharmacological or surgical methods can reduce the risk of sepsis and amputation, and improve patients’ quality of life.10,11

Metformin has been tested and used as an anti-malarial and anti-influenza agent since the 1940s.12 Preclinical researches have demonstrated that metformin can enhance the function of neutrophil and T cells and decrease the amount of proinflammatory cytokines by stimulating the adenosine monophosphate-activated protein kinase (AMPK), thus producing anti-inflammatory and antibacterial effects.13 Human studies have demonstrated that metformin may attenuate the risk of pneumonia, mycobacterial infection, sepsis, hospitalization, and mortality due to infection.2,13–16 No study has explored the effect of metformin on the risk of cellulitis and foot infections. We hypothesize that metformin may have an impact on the risk of cellulitis, foot infections, and leg amputation in patients with T2D. Therefore, we performed this cohort study to determine the risks of cellulitis, recurrent cellulitis, foot infections, and amputation between metformin use and no-use in patients with T2D.

2. METHODS 2.1. Data source

We recruited persons with new diagnosis of T2D from Taiwan’s National Health Insurance Research Database (NHIRD) since January 1, 2000, to December 31, 2017. The NHIRD is described in our previous study.17 Information of the insured on sex, age, residential areas, premiums, diagnoses, laboratory tests, medications, and clinical procedures are written in the NHIRD. Disease diagnosis is according to the International Classification of Diseases, 9th and 10th Revision, Clinical Modification (ICD-9 and 10-CM). This dataset switched from ICD-9 to ICD-10 coding in 2016. The NHIRD has linkage to the National Death Registry to get mortality data. We confirmed that all methods used were performed according to the Declaration of Helsinki. This research was approved by the Research Ethics Committee of China Medical University and Hospital [CMUH109-REC2-031(CR-2)]. The identifiable data of care providers and patients was enciphered and scrambled before release to avoid data leakage. Our study was permitted by the Research Ethics Committee to exempt for the informed consent of patients.

2.2. Study design and participants

In Taiwan, doctors will test the patient’s blood glucose and glycated hemoglobin according to the patient’s description. If the results are consistent with a diagnosis of T2D patients will be diagnosed with T2D and will receive diabetes education, medications, and regular follow-up. The Diabetes association of the Republic of China has established guidelines for T2D, including criteria for diagnosing T2D and recommending that patients’ hemoglobin A1c be monitored every 3 months and that low-density cholesterol, fundus, neurological, renal function, and microalbuminuria be checked at least once a year. We recruited patients from the NHIRD. They were diagnosed with T2D and taking antidiabetic drugs. T2D was diagnosed according to the ICD codes (Supplementary Table S1, https://links.lww.com/JCMA/A240) for ≧3 outpatient claims or 1 hospitalization. The method of taking ICD codes to define T2D was validated by previous research in Taiwan with acceptable accuracy (74.6%).18 Patients with the following conditions were excluded (Fig. 1): (1) missing gender or age, (2) age <20 or >80 years, (3) diagnosed type 1 diabetes, cellulitis, foot infections, or amputation at baseline, malignant cancers of the urinary tract, hematopoietic and lymphatic tissue, dialysis, hepatic failure, or immunosuppressant administered during the study, (4) index years not between enrollment dates and end of the research.

F1Fig. 1:

The flowchart of patient’s selection in this research. T2DM = type 2 diabetes mellitus.

2.3. Study procedures

When a patient finishes an office visit, he goes to the pharmacy to fill the prescription. The drug coverage is available for all ages within the Taiwan’s National Health Insurance. Patients who used metformin for ≧28 days within 1 year were defined as the metformin use study group, and those who did not use metformin during the study period were defined as the metformin nonuse control group. The first day of metformin use after the diagnosis of T2D was set as the index date, and the index date for the comparison group was set as the same time from T2D diagnosis to the index date of metformin usage. Some crucial variables assessed and matched between metformin use and no-use were age, sex, smoking, obesity, alcohol-related disorders, hypertension, dyslipidemia, stroke, coronary artery disease, atrial fibrillation, heart failure, chronic kidney disease, peripheral arterial disease, retinopathy and other retinal disorders, chronic obstructive pulmonary disease (COPD), rheumatoid arthritis, systemic lupus erythematosus, liver cirrhosis, psychosis, depression, cancers, and dementia diagnosed within 1 year before the index date. Prescriptions, such as the number and item of oral antidiabetic drugs, glucagon-like peptide-1 receptor agonists (GLP-1RAs), insulin, corticosteroids, nonsteroidal anti-inflammatory drugs (NSAIDs), statins, and aspirin, were also recorded before or during the index date. We counted the Diabetes Complication Severity Index (DCSI) and Charlson Comorbidity Index (CCI) scores to assess the disease burden of patients.19,20 But about the information of patient’s persistence and adherence to doctor’s prescribed medication is incomplete in this dataset.

2.4. Main endpoints

We assessed and compared the risk of cellulitis development, recurrent cellulitis, foot infections, and leg amputation between metformin use and no-use during the follow-up time.21 Cellulitis was diagnosed with the ICD codes for ≧3 outpatient claims or one hospitalization. Recurrent cellulitis was characterized by the second episode of cellulitis occurring more than 30 days after the initial event. The foot infections included gangrene, osteomyelitis, and cellulitis or abscess of the leg. Amputation was characterized by at least one hospitalization for amputation, excluding traumatic cases.

2.5. Statistical analysis

Propensity score matching was adopted for matching related variates between metformin use and no-use.22 Nonparsimonious multivariable logistic regressions were used to estimate the propensity score for every patient, with metformin use as the dependent variate, 42 clinical variates, including gender, age, obesity, smoking, comorbidities, DCSI, CCI scores, prescriptions, and duration of T2D, as the independent variates (Table 1). The nearest-neighbor algorithm was adopted to select pairs, and the control group was matched without replacement. We assumed the standardized mean difference (SMD) of ≦0.1 as a negligible difference between the study and comparison groups.

Table 1 - Baseline characteristics, comorbidities, and prescriptions in patients with T2D with and without metformin use Variable Before PSM After PSM Nonmetformin users Metformin users SMD Nonmetformin Users Metformin users SMDa n (%)/mean ± SD n (%)/mean ± SD All 25 615 78 581 23 234 23 234 Gender 0.1369 0.0079 Female 13 226 (51.63) 35 212 (44.81) 11 811 (50.83) 11 903 (51.23) Male 12 389 (48.37) 43 369 (55.19) 11 423 (49.17) 11 331 (48.77) Age group (y)  20-39 1928 (7.53) 6222 (7.92) 0.0147 1749 (7.53) 1553 (6.68) 0.0328  40-59 10 600 (41.38) 41 213 (52.45) 0.2231 9928 (42.73) 9660 (41.58) 0.0234  60+ 13 087 (51.09) 31 146 (39.64) 0.2316 11 557 (49.74) 12 021 (51.74) 0.0400 Age (y) 59.10 ± 12.40 56.25 ± 11.62 0.2371 58.74 ± 12.27 59.28 ± 12.03 0.0440 Comorbidities  Obesity 0.0050 0.0164   Yes 266 (1.04) 856 (1.09) 252 (1.08) 293 (1.26)  Smoking status 0.0237 0.0125   Yes 393 (1.53) 987 (1.26) 347 (1.49) 383 (1.65)  Alcohol disorders 0.0463 0.0097   Yes 715 (2.79) 1633 (2.08) 632 (2.72) 669 (2.88)  Hypertension 0.0657 0.0735   Yes 14 771 (57.67) 42 754 (54.41) 13 414 (57.73) 14 252 (61.34)  Dyslipidemia 0.2194 0.0711   Yes 14 666 (57.26) 36 427 (46.36) 13 190 (56.77) 14 003 (60.27)  Coronary artery disease 0.1922 0.0370   Yes 7415 (28.95) 16 254 (20.68) 6532 (28.11) 6922 (29.79)  Stroke 0.1884 0.0225   Yes 4170 (16.28) 7817 (9.95) 3551 (15.28) 3741 (16.10)  Atrial fibrillation 0.0923 0.0085   Yes 649 (2.53) 1001 (1.27) 533 (2.29) 563 (2.42)  Heart failure 0.1225 0.0065   Yes 1569 (6.13) 2756 (3.51) 1325 (5.70) 1360 (5.85)  Peripheral arterial disease 0.0729 0.0017   Yes 663 (2.59) 1219 (1.55) 567 (2.44) 561 (2.41)  Chronic kidney disease 0.2303 0.0116   Yes 2105 (8.22) 2327 (2.96) 1569 (6.75) 1502 (6.46)  Retinopathy and other retinal disorders 0.0589 0.0173   Yes 1587 (6.20) 3812 (4.85) 1426 (6.14) 1524 (6.56)  COPD 0.1801 0.0166   Yes 4526 (17.67) 8920 (11.35) 3867 (16.64) 4012 (17.27)  Rheumatoid arthritis 0.0654 0.0026   Yes 497 (1.94) 893 (1.14) 417 (1.79) 425 (1.83)  Systemic lupus erythematous 0.0469 0.0019   Yes 73 (0.28) 66 (0.08) 48 (0.21) 46 (0.20)  Liver cirrhosis 0.0953 0.0025   Yes 688 (2.69) 1059 (1.35) 583 (2.51) 592 (2.55)  Cancers 0.1478 0.0000   Yes 1295 (5.06) 1795 (2.28) 1041 (4.48) 1041 (4.48)  Psychosis 0.0353 0.0035   Yes 501 (1.96) 1176 (1.50) 426 (1.83) 437 (1.88)  Depression 0.1318 0.0056   Yes 1479 (5.77) 2412 (3.07) 1232 (5.30) 1203 (5.18)  Dementia 0.1292 0.0043   Yes 705 (2.75) 788 (1.00) 539 (2.32) 524 (2.26)  CCI   0 18 062 (70.51) 64 888 (82.57) 0.2876 16 775 (72.20) 16 435 (70.74) 0.0324   1 3000 (11.71) 7276 (9.26) 0.0801 2711 (11.67) 2909 (12.52) 0.0261   2+ 4553 (17.77) 6417 (8.17) 0.2890 3748 (16.13) 3890 (16.74) 0.0165  DCSI   0 9102 (35.53) 39586 (50.38) 0.3033 8559 (36.84) 7859 (33.83) 0.0631   1 4912 (19.18) 14 361 (18.28) 0.0231 4521 (19.46) 4746 (20.43) 0.0242   2+ 11 601 (45.29) 24 634 (31.35) 0.2898 10 154 (43.70) 10 629 (45.75) 0.0411 Medications  Numbers of oral antidiabetic agents   <2 25 199 (98.38) 75 659 (96.28) 0.1302 22 821 (98.22) 22 692 (97.67) 0.0391   2-3 411 (1.60) 2904 (3.70) 0.1305 408 (1.76) 535 (2.30) 0.0388   >3 5 (0.02) 18 (0.02) 0.0023 5 (0.02) 7 (0.03) 0.0054  GLP-1RAs 0.0097 0.0131   No 25 615 (100.00) 78 581 (100.00) 23 234 (100.00) 23 234 (100.00)  Insulins 0.2097 0.0391   Yes 5340 (20.85) 10 234 (13.02) 4775 (20.55) 5147 (22.15)  Corticosteroids 0.4994 0.0403   Yes 12 025 (46.95) 18 686 (23.78) 10 448 (44.97) 10 914 (46.97)  Statins 0.2611 0.0477   Yes 7232 (28.23) 13 649 (17.37) 6377 (27.45) 6877 (29.60)  NSAIDs 0.8137 0.0830   Yes 21 078 (82.29) 36 272 (46.16) 18 709 (80.52) 19 447 (83.70)  Aspirin 0.2571 0.0477   Yes 6911 (26.98) 12 937 (16.46) 6119 (26.34) 6613 (28.46) Duration of T2D (y) 3.92 ± 3.54 1.91 ± 2.94 0.6157 3.74 ± 3.39 3.91 ± 3.76 0.0470

CCI = Charlson comorbidity index; COPD = chronic obstructive pulmonary disease; DCSI = Diabetes complications severity index; GLP-1RAs = glucagon-like peptide-1 receptor agonists; NSAIDs = nonsteroidal anti-inflammatory drugs; PSM = propensity score matching; SMD = standardized mean difference; T2D = type 2 diabetes.

aA SMD ≤0.1 indicates a negligible difference between the two cohorts.

The incidence rate of endpoints was calculated by the time-scale of 1000 person-years during the traced period. Crude and multivariable adjusted Cox proportional hazards models with robust sandwich standard error estimates were used to compare outcomes between metformin use and no use. The results are displayed as hazard ratio (HR), adjusted hazard ratio (aHR) and 95% CI for metformin use versus no-use. To assess the observed risk, we traced patients till the date of respective endpoints, mortality, or at the end of follow-up time on December 31, 2017, whichever happened first. Log-rank test and Kaplan-Meier method were utilized to describe and measure the cumulative incidence of incident cellulitis, foot infections, and amputation between metformin use and no-use during the traced time. We also evaluated the average cumulative duration of metformin usage for the risk of cellulitis, recurrent cellulitis, foot infections, and amputation compared with metformin no-use. We have performed a sensitivity analysis by using full cohort of unmatched patients through inverse probability of treatment weighting (IPTW). Time-varying exposure of metformin analysis was done to account for the changes of metformin use over time in practice. We also included “cellulitis and oral soft tissue abscess” in the definition of cellulitis, “necrotizing fasciitis,” and “pyomyositis” in the broad definition of foot infections, and performed the multivariable adjusted analysis (Supplementary Table S2, https://links.lww.com/JCMA/A240). Using the outcome-related risk factors as adjustment variables, we performed multivariate analysis of different models. That is, model I to adjust for age and sex, model II to adjust for age, sex, obesity, smoking and alcohol, model III to adjust for age, sex, obesity, smoking, alcohol disorders, peripheral arterial disease, chronic kidney disease, retinopathy and other retinal disorders, COPD, liver cirrhosis, rheumatoid arthritis, systemic lupus erythematosus, corticosteroids, duration of T2D (Supplementary Table S2, https://links.lww.com/JCMA/A240). To avoid confounding by the high risk of cardiovascular mortality in diabetes. We used mortality as a competing risk to perform the competing risk analysis for the risks of cellulitis, foot infections, and amputation between metformin use and no use (Supplementary Table S3, https://links.lww.com/JCMA/A240).

We considered a two-tailed p value less than 0.05 as statistically significant. Statisticians, Kai-Chieh Hu and Teng-Shun Yu, performed data organization, and used SAS (version 9.4; SAS Institute, Cary, NC) for statistical analysis.

3. RESULTS 3.1. Study population

From January 1, 2000, to December 31, 2017, we found 276 415 patients with newly diagnosed T2D; 206 046 patients used metformin, and 70 369 patients did not use metformin (Fig. 1). After excluding unsuitable participants, one to one propensity score matching was adopted to select 23 234 pairs of patients with metformin use and no-use. All critical variables were matched well between the study and comparison groups with the SMD ≦0.1 (Table 1). In matched cohorts, 51.03% of patients were female. The mean (SD) age of patients with metformin use and no-use was 59.28 (12.03) and 58.74 (12.27) years, and the mean follow-up period was 6.31 (3.93) and 5.54 (3.97) years, respectively.

3.2. Main endpoints

After propensity score matching, 6310 (27.16%) metformin users and 5070 (21.82%) nonusers had cellulitis during the follow-up period (incidence rate: 49.99 vs 45.00 per 1000 person-years; Table 2). The multivariable models presented that metformin users had a significantly (8%) higher risk of incident cellulitis (aHR, 1.08; 95% CI, 1.04-1.12) than nonusers (Table 2). Patients with male sex, young age (20-39 years), alcohol-related disorders, heart failure, coronary artery disease, retinopathy and other retinal disorders, liver cirrhosis, rheumatoid arthritis, chronic kidney disease, psychosis, COPD, dementia, CCI ≧2, DCSI score ≧2, insulins, and NSAIDs use had a significantly higher risk of cellulitis; while patients with dyslipidemia, and statin use showed a significantly lower risk of cellulitis (Supplementary Table S4, https://links.lww.com/JCMA/A240). Multivariable models also showed that metformin use had a 33% higher risk of recurrent cellulitis (aHR, 1.33; 95% CI, 1.14-1.55), a 91% higher risk of foot infections (aHR, 1.91; 95% CI, 1.75-2.09), and an 88% higher risk of amputation (aHR, 1.88; 95% CI, 1.35-2.62) than metformin no-use (Table 2).

Table 2 - Incidence and risk of outcomes associated with metformin use in patients with T2D Variable Nonmetformin users Metformin users Crude Adjusteda Event Person-years IR Event Person-years IR HR (95% CI) p HR (95% CI) p Cellulitis 5070 112 654 45.00

留言 (0)

沒有登入
gif