Type 2 diabetes mellitus (DM) plus chronic heart failure (CHF) is a common but lethal combination and therapeutic options are limited. Metformin is perceived as being relatively contraindicated in this context, although mounting evidence indicates that it may be beneficial. This study was carried out to investigate the use of metformin therapy for treating patients with DM and CHF in a large population-based cohort study. The Health Informatics Centre–dispensed prescribing database for the population of Tayside, Scotland (population ∼400,000) was linked to the Diabetes Audit and Research in Tayside Scotland (DARTS) information system. Patients with DM and incident CHF from 1994 to 2003 receiving oral hypoglycemic agents but not insulin were identified. Cox regression was used to assess differences in all-cause mortality rates between patients prescribed metformin and patients prescribed sulfonylureas with adjustment for co-morbidities and other therapies. Four hundred twenty-two study subjects (mean ± SD 75.4 ± 0.5 years of age, 46.2% women) were identified: metformin monotherapy (n = 68, mean age 75.5 ± 1.1 years, 48.5% women), sulfonylurea monotherapy (n = 217, mean age 76.7 ± 0.7 years, 45.2% women), and combination (n = 137, mean age, 73.4 ± 0.7 years, 46.7% women). Fewer deaths occurred in metformin users, alone or in combination with sulfonylureas, compared to the sulfonylurea monotherapy cohort at 1 year (0.59, 95% confidence interval 0.36 to 0.96) and over long-term follow up (0.67, 95% confidence interval 0.51 to 0.88). In conclusion, this large observational data suggest that metformin may be beneficial in patients with CHF and DM. These findings need to be verified by a prospective clinical trial.
Patients with a combination of chronic heart failure (CHF) and type 2 diabetes mellitus (DM) have limited drug therapy options for their DM. Because insulin resistance is a key pathophysiologic process in these patients, it would seem intuitive to administer an insulin-sensitizing agent such as thiazolidinediones or metformin. However, thiazolidinediones have the potential to exacerbate CHF in patients with poor cardiac reserve and are contraindicated in patients with CHF in New York Heart Association functional class III or IV. Although metformin is a widely prescribed drug in the management of DM, its use in patients with DM and CHF has previously been discouraged due to concerns over the risk of lactic acidosis originating from previous experience with phenformin. However, clinical experience and the literature suggest that risk of metformin-associated lactic acidosis is very low and similar to that of other antidiabetic drugs. Observational data suggest that metformin may actually be beneficial for CHF. A retrospective cohort study in a group of Medicare recipients discharged from United States hospitals with a principal diagnosis of CHF suggests that metformin and thiazolidinediones are not associated with increased mortality and may improve outcomes in older patients with DM and CHF. It has been proposed by several groups that compensated CHF can no longer be upheld as contraindication for metformin. The beneficial effect of metformin in CHF may extend to patients with other manifestations of cardiovascular disease; an observational study in the United Kingdom demonstrated that cardiovascular risk was lower in metformin users compared to sulfonylureas users. Further evidence for the potential benefit of metformin in patients with DM and CHF may result in expanding the therapeutic options for managing this important common and complex group of patients. Therefore, the aim of this study was to add to the growing body of evidence on the safety of metformin and to explore benefits of its use in patients with DM and CHF in Tayside, Scotland. In this study, we specifically studied patients with DM who were treatment naive who developed CHF and examined the efficacy of metformin and sulfonylureas in this context.
Methods
This study was carried out in the population (approximately 400,000) of Tayside in Scotland, using the Diabetes Audit and Research in Tayside Scotland (DARTS) information system and the dispensed prescribing database maintained by the Health Informatics Centre (HIC) University of Dundee. This contains records of all dispensed community prescriptions dating back to 1993. The DARTS dataset contains detailed clinical information on every patient diagnosed with DM in Tayside. Clinical information is collected according to the national clinical dataset for the care of diabetic patients in Scotland and includes diabetes type, date of diagnosis, duration, therapy, hemoglobin A 1c , presence (and date) of microvascular and macrovascular diabetic complications, and cardiovascular risk factors. Datasets available from HIC also include hospital discharge data, biochemistry data, mortality data, sociodemographic descriptors, and other data that are linked by a unique 10-digit patient identifier, the community health index number, that is used for all health care activities in Tayside. The first 6 digits are the date of birth, with a figure coding gender (male gender, even female gender) in the remaining 4 distinguishing digits. This number facilitates high accuracy record linkage at the level of the patient across the region. All research data are robustly made anonymous and approved by the Tayside National Health Service Caldicott Guardians. The study was granted ethical approved by the Tayside committee on medical research ethics.
The study population was defined as residents of Tayside who were registered with their general practitioner during the study period 1994 to 2003. Using the DARTS database, we identified all Tayside residents who were diagnosed with DM before December 2003. We then identified all those who were newly treated with oral hypoglycemic agents during the study period (January 1994 to December 2003). Any patients who received oral hypoglycemic agent prescriptions before 1994 or received insulin at any point during the study period were excluded.
We identified patients who had incident CHF during the study period and defined a date of CHF diagnosis for each patient. This was the earliest date of patients fulfilling any 1 of the following criteria. (1) Patients with a hospital admission International Classification of Diseases, Ninth Revision and 10th Revision (ICD-9/10) diagnostic code for CHF during the study period (ICD-9 code 428, ICD-10 code 50). The date of admission was defined as the date of CHF diagnosis. (2) Patients commenced on CHF medications defined as a combination of loop diuretics and ACE inhibitors. Co-prescribing had to occur within a 90-day period, and the date of the second drug was defined as the date of CHF diagnosis. (3) Patients who had ≥1 admission for myocardial infarction and then received loop diuretic medication. The date of diuretic medication was defined as the date of CHF diagnosis. Patients were excluded if their plasma creatinine concentration was >200 μmol/L before the prescribing of a loop diuretic. This was to exclude patients with renal hypertension without CHF who might receive loop diuretics and ACE inhibitors. We previously used similar criteria to identify CHF from large datasets in Tayside.
To be eligible for the study, the date of CHF diagnosis had to occur after the date of diagnosis of DM. The patient also had to receive a first prescription for metformin or sulfonylureas within 1 year after the date of CHF diagnosis. Patients were then divided into 3 cohorts: those who received prescriptions for metformin only (metformin monotherapy), those who received prescriptions for sulfonylureas only (sulfonylureas monotherapy), and those who received prescriptions for metformin and sulfonylureas (combination).
All subjects were prospectively followed up from their index date (date of CHF diagnosis) until the primary outcome, termination of HIC health care coverage, or termination of the study. Study outcomes were all-cause mortality at 1 year (short term) and by the end of the follow-up period (long term). All-cause mortality was determined from death certification records of the General Register Office of Scotland. We compared survival between cohorts for each outcome using Kaplan-Meier survival plots and used Cox regression analyses to estimate the relative risks of each outcome for patients in the study cohorts. For this analysis, the metformin monotherapy cohort and the combination cohort were merged into a larger cohort, with the sulfonylureas monotherapy cohort as the reference group. Survival times were censored if patients left Tayside or at the end of the study period. The following confounding variables were investigated: gender, age at index date, duration of DM at index date, creatinine, and hemoglobin A 1c . We also accounted for whether the patient had been admitted to hospital with an ICD-9/10 diagnostic code for a major cardiovascular event (myocardial infarction, coronary heart disease, or stroke) before the diagnosis of CHF. We calculated the proportions of patients in each cohort who had received prescriptions for any of 4 drug types—ACE inhibitors, aspirin, diuretics, or β blockers—and adjusted for baseline differences in medication. Continuous covariates were categorized into quartile groups where appropriate and all covariates were included in the final models only if these were statistically significant in univariate analyses (p <0.05), to produce adjusted risk estimates for all covariates. All analyses were conducted using SPSS 16.0 (SPSS, Inc., Chicago, Illinois).
Results
There were 1,141 patients who were diagnosed with DM before December 2003, who received oral hypoglycemic agents during the study period (1994 to 2003), but who were not on insulin. All these patients had been admitted to hospital with a diagnostic code for CHF, although 218 patients were not eligible for the study because this hospital admission occurred outside the study period. It was not possible to identify a date of diagnosis of CHF for 9 patients who were also excluded from the study. From the remaining 914 patients, we identified 769 whose date of CHF diagnosis occurred after date of diagnosis of DM.
Four hundred ninety patients were prescribed metformin or sulfonylureas after their date of CHF diagnosis. However, we excluded a further 59 patients whose first oral hypoglycemic agent was prescribed >365 days after date of CHF diagnosis, and 9 without a valid date of diagnosis of DM. Of the remaining 422 patients (mean age 75.4 ± 0.5 years), 68 were prescribed metformin only, 217 were prescribed sulfonylureas only, and 137 received prescriptions for metformin and sulfonylureas. Characteristics of these patients are presented in Table 1 . We compared patients who had received any metformin (metformin monotherapy cohort or combination cohort) to patients in the sulfonylureas monotherapy cohort. Although they were slightly younger and more likely to be women, the only statistically significant differences were that they had lower mean creatinine and a larger proportion was treated with aspirin and ACE inhibitors.
| Variable | Sulfonylureas Monotherapy (n = 217) | Metformin Monotherapy (n = 68) | Metformin + Sulfonylurea Combination (n = 137) | Any Metformin: Metformin Monotherapy and Combination (n = 205) | p Value ⁎ |
|---|---|---|---|---|---|
| Women | 45.2% | 48.5% | 46.7% | 47.3% | 0.66 † |
| Age (years), mean ± SD | 76.7 ± 9.8 | 75.5 ± 8.9 | 73.4 ± 8.7 | 74.1 ± 8.8 | 0.19 ‡ |
| Mean diabetes duration (years), mean ± SD | 6.7 ± 6.2 | 6.7 ± 6.2 | 7.7 ± 5.8 | 7.4 ± 5.9 | 0.86 ‡ |
| Previous hospital admission (%) | 33.2% | 33.8% | 27.0% | 29.3% | 0.39 † |
| Mean last creatinine (mmol/L), mean ± SD | 170.8 ± 139.4 | 135.1 ± 71.0 | 133.2 ± 82.8 | 133.8 ± 78.0 | <0.001 ‡ |
| Mean last hemoglobin A 1c (%), mean ± SD | 7.2 ± 1.6 | 7.4 ± 1.5 | 7.6 ± 1.9 | 7.5 ± 1.8 | 0.56 ‡ |
| Angiotensin-converting enzyme inhibitor treatment | 32.6% | 41.8% | 43.3% | 42.9% | 0.03 † |
| Aspirin treatment | 51.6% | 68.7% | 67.6% | 68.0% | 0.001 † |
| Diuretic treatment | 21.9% | 14.9% | 17.6% | 16.7% | 0.13 † |
| β-blocker treatment | 11.6% | 19.4% | 15.4% | 16.7% | 0.13 † |
⁎ Comparison is between sulfonylurea monotherapy and any-metformin category.
One-year and long-term mortalities were higher in the sulfonylureas monotherapy cohort ( Figure 1 , Table 2 ) compared to patients prescribed metformin. In Cox regression analysis, unadjusted hazard ratios were 0.56 (95% confidence interval 0.38 to 0.84) and 0.53 (95% confidence interval 0.33 to 0.67) for 1-year and long-term mortalities, respectively, when the metformin and combination cohorts were grouped together and compared to the sulfonylurea monotherapy cohort ( Table 2 ). After adjusting for baseline differences between the 2 groups, users of metformin, alone or in combination, had a 30% to 40% lower risk of the outcomes ( Table 2 ).
| 1-Year Mortality | Long-Term Mortality | |||
|---|---|---|---|---|
| Unadjusted | Adjusted ⁎ | Unadjusted | Adjusted ⁎ | |
| Cohort | ||||
| Sulfonylurea monotherapy | 1.00 | 1.00 | 1.00 | 1.00 |
| Metformin monotherapy + combination | 0.56 (0.38–0.84) | 0.60 (0.37–0.97) | 0.53 (0.33–0.67) | 0.67 (0.51–0.88) |
| Male gender | 1.00 | 1.00 | 1.00 | 1.00 |
| Female gender | 0.82 (0.67–1.00) | 0.47 (0.29–0.77) | 1.16 (0.93–1.45) | 0.64 (0.49–0.84) |
| Age (years) | ||||
| <60 | 1.00 | 1.00 | 1.00 | 1.00 |
| 60–69 | 5.60 (0.76–41.4) | 4.62 (0.62–34.64) | 1.15 (0.57–2.32) | 1.11 (0.52–2.39) |
| 70–79 | 5.07 (0.70–36.9) | 4.20 (0.59–34.64) | 1.66 (0.85–3.28) | 1.57 (0.75–3.25) |
| 80–89 | 7.06 (0.97–51.5) | 7.33 (0.98–54.86) | 2.82 (1.43–5.56) | 2.61 (1.24–5.49) |
| >89 | 8.57 (1.09–67.7) | 12.39 (1.52–101.06) | 3.24 (1.51–6.94) | 3.44 (1.47–8.05) |
| Duration (years) | ||||
| <5 | 1.00 | — | 1.00 | 1.00 |
| 5–9 | 1.44 (0.91–2.28) | — | 1.26 (0.97–1.64) | 0.92 (0.67–1.26) |
| 10–14 | 1.70 (1.00–2.91) | — | 1.28 (0.92–1.78) | 1.13 (0.77–1.64) |
| >14 | 1.38 (0.74–2.58) | — | 1.80 (1.25–2.58) | 1.77 (1.17–2.65) |
| Previous hospital admission | ||||
| Yes vs no | 1.23 (0.83–1.83) | 1.08 (0.67–1.74) | 1.01 (0.79–1.28) | 0.97 (0.73–1.27) |
| Creatinine ⁎ | ||||
| Quartile 1 | 1.00 | 1.00 | 1.00 | 1.00 |
| Quartile 2 | 0.54 (0.30–0.96) | 0.46 (0.26–0.84) | 0.55 (0.38–0.80) | 0.52 (0.35–0.76) |
| Quartile 3 | 0.29 (0.14–0.59) | 0.23 (0.11–0.47) | 0.62 (0.43–0.88) | 0.53 (0.36–0.77) |
| Quartile 4 | 0.67 (0.38–1.17) | 0.51 (0.28–0.91) | 1.09 (0.78–1.53) | 0.87 (0.61–1.23) |
| Hemoglobin A 1c † | ||||
| <7.5 | 1.00 | — | 1.00 | — |
| 7.5–9.49 | 1.14 (0.70–1.83) | — | 1.08 (0.83–1.40) | — |
| >9.49 | 1.06 (0.48–2.34) | — | 1.03 (0.67–1.57) | — |
| Angiotensin-converting enzyme inhibitor | ||||
| Yes vs no | 0.56 (0.36–0.86) | 0.57 (0.34–0.97) | 0.66 (0.52–0.84) | 0.76 (0.58–1.00) |
| Aspirin | ||||
| Yes vs no | 0.61 (0.41–0.89) | 0.67 (0.42–1.08) | 0.65 (0.52–0.82) | 0.80 (0.61–1.05) |
| Diuretic | ||||
| Yes vs no | 1.14 (0.72–1.81) | — | 1.09 (0.83–1.43) | — |
| β blocker | ||||
| Yes vs no | 0.46 (0.22–0.94) | 0.41 (0.16–1.04) | 0.52 (0.36–0.74) | 0.52 (0.34–0.80) |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
