This case-control study uses data from a nationwide claims database to investigate the risk of developing age-related macular degeneration in patients without diabetes who take metformin.
Key Points
Question
Is there an association between the use of metformin and protection against the development of age-related macular degeneration (AMD) among patients without diabetes?
Findings
In this case-control study of 231 142 cases with AMD and 232 879 matched controls, none of whom had a diagnosis of diabetes, metformin use was associated with 17% lower odds of AMD development. This association was not dose dependent.
Meaning
Findings of this study suggest that metformin may be useful as a therapeutic tool for protection against AMD development in patients without diabetes.
Abstract
Importance
Metformin use may protect against the development of age-related macular degeneration (AMD) based on results from observational studies. However, its potential effectiveness among patients without diabetes remains unclear.
Objective
To assess the association between metformin use and the development of AMD in patients without diabetes.
Design, Setting, and Participants
This case-control study used data from 2006 to 2017 in the Merative MarketScan Research Database, a nationwide insurance claims database that includes between 27 and 57 million patients in the US with primary or Medicare supplemental health insurance. Cases with AMD and controls without AMD aged 55 years or older were matched 1:1 by year, age, anemia, hypertension, region, and Charlson Comorbidity Index score. Then, cases and matched controls without a diagnosis of diabetes were selected. In subgroup analyses, cases with dry AMD and their matched controls were identified to explore the association between metformin use and AMD staging in patients without diabetes. Data were analyzed between March and September 2023.
Exposures
Exposure to metformin in the 2 years prior to the index date (ie, date of AMD diagnosis in cases and date of a randomly selected eye examination for controls) was assessed from the claims database and categorized into quartiles based on cumulative dose (1-270, 271-600, 601-1080, and >1080 g/2 y). Exposure to other antidiabetic medications was also noted.
Main Outcomes and Measures
Odds of new-onset AMD development as assessed by multivariable conditional logistic regression after adjusting for known risk factors for AMD, including female sex, hyperlipidemia, smoking, and exposures to other antidiabetic medications. Asymptotic Cochran-Armitage tests for trend were also performed.
Results
We identified 231 142 patients with any AMD (mean [SD] age, 75.1 [10.4] years; 140 172 females [60.6%]) and 232 879 matched controls without AMD (mean [SD] age, 74.9 [10.5] years; 133 670 females [57.4%]), none of whom had a diagnosis of diabetes. The sample included 144 147 cases with dry AMD that were matched to 144 530 controls. In all, 2268 (1.0%) cases and 3087 controls (1.3%) were exposed to metformin in the 2 years before their index visit. After data adjustment, exposure to any metformin was associated with reduced odds of any AMD development (adjusted odds ratio [AOR], 0.83; 95% CI, 0.74-0.87), specifically in the dosing quartiles of 1 to 270, 271 to 600, and 601 to 1080 g/2 y. Any metformin use was also associated with a reduced odds of developing dry AMD (AOR, 0.85; 95% CI, 0.79-0.92), specifically in the dosing quartiles of 1 to 270 and 271 to 600 g/2 y. Adjusted odds ratios for any AMD and dry AMD development did not differ across the dosing quartiles. Asymptotic Cochran-Armitage tests for trend revealed 2-sided P = .51 and P = .66 for the any and dry AMD samples, respectively.
Conclusions and Relevance
In this case-control study of a population without a diagnosis of diabetes, metformin use was associated with reduced odds of developing AMD. This association does not appear to be dose dependent. These findings provide further impetus to study metformin’s usefulness in protecting against AMD in prospective clinical trials.
Introduction
In recent years, metformin use has been associated with reduced odds of developing age-related macular degeneration (AMD) in patients with diabetes.1,2,3,4 This protective association has been inconsistently reproduced, and several studies failed to identify such an association.5,6 Furthermore, the protective association of metformin with AMD development reported in observational studies does not necessarily imply causality, which would require proof, for example, from randomized clinical trials. In a prior case-control study of AMD cases and matched controls without AMD, patients with diabetes comprised approximately one-quarter of the study population and represented over 95% of the patients exposed to metformin.2 Thus, the potential protection afforded by metformin may not have been accurately captured among the 75% of AMD cases without diabetes.2
Prescribing metformin to patients without diabetes is neither novel nor approved for this indication by regulatory authorities, such as the US Food and Drug Administration. Physicians might consider off-label use because metformin does not cause hypoglycemia and is prescribed off-label to treat polycystic ovarian syndrome.7 Metformin is associated with reduced mortality from cardiovascular disease and reduced incidence of dementia,8 although it is not used clinically to treat these conditions. Therefore, we investigated the association between metformin use and AMD development in a subset of patients without a diabetes diagnosis.
Methods
The University of Chicago Institutional Review Board provided an exemption for this study because patient data were deidentified. We followed the STROBE reporting guideline. This case-control study was designed after the publication of a peer-reviewed case-control study of patients ages 55 years and older with newly diagnosed AMD of any subtype (hereafter, any AMD) between 2008 and 2017 in the Merative MarketScan Research Databases.2 These annual databases include between 27 and 57 million patients in the US with primary or Medicare supplemental health coverage through private insurance. Cases with AMD and controls were matched 1:1 based on year, age, presence of anemia, presence of hypertension, region, and Charlson Comorbidity Index (CCI) score. The CCI score is a risk factor for mortality from comorbid conditions and is used to measure burden of disease, with higher scores reflecting higher burden of disease.9 Cases and controls were required to have a minimum of 2 eye examinations in the 12 months prior to the index date, which was defined as the date of AMD diagnosis for cases and the date of a randomly selected eye examination for controls. Metformin exposure in patients with newly diagnosed AMD (cases) was compared with that in patients without AMD (controls). Exposure to any metformin was ascertained in the 2 years prior to the index visit to capture the range of treatment courses and was categorized by quartile (1-270, 271-600, 601-1080, and >1080 g/2 y).2 We excluded cases and controls with diabetes, as identified by an outpatient or inpatient claim with a relevant diagnosis code (International Classification of Disease, Ninth Revision [ICD-9] codes 250.XX and 362.0X or International Statistical Classification of Diseases and Related Health Problems, Tenth Revision [ICD-10] codes E10.X and E11.X) in the 12 months prior to the index visit. This lookback period was short, as diabetes is likely to be regularly captured in the medical record due to its chronic course.
We used conditional logistic regression to account for the matched pairs (ie, case-control) study design.10 We performed multivariable logistic regression to test the association between metformin use and the odds of developing any AMD after adjusting for known risk factors for AMD, including female sex, hyperlipidemia, smoking, and exposures to other antidiabetic medications. Adjusted odds ratios (AORs) across dosing quartiles were calculated. In addition, we performed asymptotic Cochran-Armitage tests for trend among the dosing quartiles to determine 2-sided P values. Analyses were repeated in the subset of cases with dry AMD and their matched controls to explore the association between metformin use and AMD staging in patients without diabetes. Dry AMD included patients with early AMD, intermediate AMD, or geographic atrophy. Data were analyzed between March and September 2023 using SAS, version 9.4 (SAS Institute Inc).
Results
We identified 231 142 patients with any AMD (mean [SD] age, 75.1 [10.4] years; 140 172 females [60.6%] and 99 970 males [39.4%]) and 232 879 matched control individuals without AMD (mean [SD] age, 74.9 [10.5] years; 133 670 females [57.4%] and 99 209 males [42.6%]); this sample included 144 147 cases with dry AMD matched to 144 530 controls (Table 1). In all, 2268 cases (1.0%) and 3087 controls (1.3%) were exposed to metformin (Table 2). Exposures to other antidiabetic medications are shown in Table 2. In adjusted analyses, any metformin use was associated with a 17% decrease in the odds of developing any AMD (AOR, 83; 95% CI, 0.78-0.88). Compared with no metformin use, 2-year cumulative metformin doses of 1 to 270, 271 to 600, and 601 to 1080 g were associated with 20%, 19%, and 12% lower odds for developing AMD, respectively (1-270 g/2 y: AOR, 80 [95% CI, 0.74-0.87]; 271-600 g/2 y: AOR, 0.81 [95% CI, 0.73-0.91]; and 601-1080 g/2 y: AOR, 0.88 [95% CI, 0.78-1.00]; Table 3). Any metformin use was associated with a 15% decrease in the odds of developing dry AMD (AOR, 0.87; 95% CI, 0.79-0.92). Compared with no metformin use, 2-year cumulative metformin doses of 1 to 270 and 271 to 600 g were associated with 18% and 14% lower odds, respectively, for dry AMD development (1-270 g/2 y: AOR, 0.82 [95% CI, 0.74-0.91]; 271-600 g/2 y: AOR, 0.86 [95% CI, 0.75-0.99]).
Table 1. Characteristics of Cases With Age-Related Macular Degeneration (AMD) and Matched Controls All Without a Diabetes Diagnosis.
| Characteristic | No. (%) | |||
|---|---|---|---|---|
| Cases with any AMD | Matched controls | Cases with dry AMD | Matched controls | |
| Total No. | 231 142 | 232 879 | 144 147 | 144 530 |
| Age group, y | ||||
| 55-64 | 49 504 (21.4) | 51 638 (22.2) | 32 970 (22.9) | 34 265 (23.7) |
| 65-74 | 54 377 (23.5) | 54 837 (23.6) | 36 792 (25.5) | 36 830 (25.5) |
| 75-84 | 79 097 (34.2) | 78 582 (33.7) | 48 609 (33.7) | 48 008 (33.2) |
| ≥85 | 48 164 (20.8) | 47 822 (20.5) | 25 776 (17.9) | 25 427 (17.6) |
| Age, mean (SD), y | 75.1 (10.4) | 74.9 (10.5) | 74.3 (10.3) | 74.1 (10.3) |
| Sex | ||||
| Female | 140 172 (60.6) | 133 670 (57.4) | 87 475 (60.7) | 82 354 (57.0) |
| Male | 90 970 (39.4) | 99 209 (42.6) | 56 672 (39.3) | 62 176 (43.0) |
| US Census Bureau region | ||||
| Northeast | 44 995 (19.5) | 45 487 (19.5) | 29 156 (20.2) | 29 845 (20.4) |
| North Central | 69 435 (30.0) | 69 520 (29.9) | 43 405 (30.1) | 43 339 (30.0) |
| South | 75 786 (32.8) | 76 327 (32.8) | 47 893 (33.2) | 47 915 (33.2) |
| West | 40 926 (17.7) | 41 545 (17.8) | 23 693 (16.4) | 23 791 (16.5) |
| Study year | ||||
| 2008-2011 | 83 572 (36.2) | 83 016 (35.7) | 51 388 (35.7) | 50 751 (35.1) |
| 2012-2014 | 86 493 (37.4) | 87 520 (37.6) | 52 495 (36.4) | 52 857 (36.6) |
| 2015-2017 | 61 077 (26.4) | 62 343 (26.8) | 40 264 (27.9) | 40 922 (28.3) |
| CCI score | ||||
| 0 | 120 787 (52.3) | 121 519 (52.2) | 76 959 (53.4) | 77 090 (53.3) |
| 1 | 47 512 (20.6) | 47 739 (20.5) | 29 183 (20.3) | 29 187 (20.2) |
| 2 | 33 825 (14.6) | 34 212 (14.7) | 20 695 (14.4) | 20 877 (14.4) |
| ≥3 | 29 018 (12.6) | 29 409 (12.6) | 17 310 (12) | 17 376 (12) |
| CCI, mean (SD) | 0.99 (1.4) | 1 (1.41) | 0.96 (1.38) | 0.97 (1.39) |
| Smoking | 12 354 (5.3) | 9440 (4.1) | 7909 (5.5) | 6058 (4.2) |
| Hyperlipidemia | 104 592 (45.3) | 102 927 (44.2) | 65 888 (45.7) | 64 400 (44.6) |
| Anemia | 12 642 (5.5) | 12 527 (5.4) | 7543 (5.2) | 7423 (5.1) |
| Hypertension | 139 957 (60.6) | 142 415 (61.2) | 85 994 (59.7) | 87 512 (60.3) |
Abbreviation: CCI, Charlson Comorbidity Index.
Table 2. Medication Exposure and Rates of Age-Related Macular Degeneration (AMD) Among Cases and Matched Controls All Without a Diabetes Diagnosis.
| Medication | No. (%) | |||
|---|---|---|---|---|
| Cases with any AMD (n = 231 142) | Matched controls (n = 232 879) | Cases with dry AMD (n = 144 147) | Matched controls (n = 144 530) | |
| Any metformin | 2268 (1.0) | 3087 (1.3) | 1442 (1.0) | 1906 (1.3) |
| Metformin dose, g/2 y | ||||
| 0 | 228 874 (99.0) | 229 792 (98.7) | 142 705 (99.0) | 142 624 (98.7) |
| 1-270 | 993 (0.4) | 1328 (0.6) | 636 (0.4) | 834 (0.6) |
| 271-600 | 552 (0.2) | 765 (0.3) | 349 (0.2) | 460 (0.3) |
| 601-1080 | 472 (0.2) | 627 (0.3) | 303 (0.2) | 392 (0.3) |
| >1080 | 251 (0.1) | 367 (0.2) | 154 (0.1) | 220 (0.2) |
| Insulin | 265 (0.1) | 485 (0.2) | 162 (0.1) | 289 (0.2) |
| Sulfonylureas | 898 (0.4) | 1439 (0.6) | 535 (0.4) | 880 (0.6) |
| Glitazones | 546 (0.2) | 797 (0.3) | 341 (0.2) | 487 (0.3) |
| Meglitinides | 58 (0) | 97 (0) | 32 (0) | 58 (0) |
| Other diabetes medicationsa | 333 (0.1) | 456 (0.2) | 198 (0.1) | 288 (0.2) |
| Statins | 108 213 (46.8) | 111 118 (47.7) | 67 152 (46.6) | 68 777 (47.6) |
Other diabetes medications include exenatide, sitagliptin, and pramlintide.
Table 3. Multivariable Conditional Logistic Regression Models for Age-Related Macular Degeneration (AMD) Cases and Matched Controls All Without a Diabetes Diagnosis.
| Characteristic | Adjusted odds ratio (95% CI) | |||
|---|---|---|---|---|
| Cases with any AMD and matched controls (n = 464 021) | Cases with dry AMD and matched controls (n = 288 677) | |||
| With any metformin use | Without metformin use | With any metformin use | Without metformin use | |
| Risk factors for AMD | ||||
| Female sex | 1.15 (1.13-1.16) | 1.15 (1.13-1.16) | 1.17 (1.15-1.19) | 1.17 (1.15-1.19) |
| Smoking | 1.39 (1.35-1.42) | 1.39 (1.35-1.42) | 1.38 (1.33-1.43) | 1.38 (1.33-1.43) |
| Hyperlipidemia | 1.08 (1.07-1.10) | 1.08 (1.07-1.10) | 1.09 (1.07-1.11) | 1.09 (1.07-1.11) |
| Medications | ||||
| Any metformin | 0.83 (0.78-0.88) | NA | 0.85 (0.79-0.92) | NA |
| Metformin dose, g/2 y | ||||
| 0 | NA | 1 [Reference] | NA | 1 [Reference] |
| 1-270 | NA | 0.80 (0.74-0.87) | NA | 0.82 (0.74-0.91) |
| 271-600 | NA | 0.81 (0.73-0.91) | NA | 0.86 (0.75-0.99) |
| 601-1080 | NA | 0.88 (0.78-1.00) | NA | 0.91 (0.78-1.07) |
| >1080 | NA | 0.86 (0.73-1.02) | NA | 0.90 (0.73-1.12) |
| Insulin | 0.66 (0.57-0.77) | 0.66 (0.57-0.77) | 0.69 (0.56-0.84) | 0.69 (0.56-0.83) |
| Sulfonylureas | 0.74 (0.68-0.81) | 0.74 (0.67-0.81) | 0.71 (0.64-0.80) | 0.71 (0.63-0.80) |
| Glitazones | 0.89 (0.79-0.995) | 0.88 (0.78-0.99) | 0.92 (0.79-1.06) | 0.91 (0.78-1.06) |
| Meglitinides | 0.75 (0.54-1.04) | 0.75 (0.54-1.04) | 0.68 (0.44-1.05) | 0.67 (0.43-1.04) |
| Other diabetes medicationsa | 0.99 (0.85-1.15) | 0.99 (0.85-1.15) | 0.91 (0.76-1.11) | 0.91 (0.75-1.10) |
| Statins | 0.95 (0.94-0.96) | 0.95 (0.94-0.96) | 0.95 (0.93-0.96) | 0.95 (0.93-0.96) |
Abbreviation: NA, not applicable.
Other diabetes medications include exenatide, sitagliptin, and pramlintide.
The odds of developing AMD did not differ significantly across the dosing quartiles in any AMD or dry AMD samples (eTable in Supplement 1). Asymptotic Cochran-Armitage tests for trend among the quartiles revealed 2-sided P = .51 and P = .66 for the any and dry AMD samples, respectively.
Discussion
Metformin use was associated with lower odds of developing AMD in patients without a diabetes diagnosis (AOR, 0.83; 95% CI, 0.78-0.88), consistent with prior literature describing similar associations in patients with diabetes.2,3,4 This AOR is lower than that reported in a previous case-control study among a subset of patients with diabetes, which found an AOR of 0.95 (95% CI, 0.93-0.97).2 These differences in odds reduction may be due to differences in the cohorts evaluated or in confounding by diabetes severity, which should have been lower in the current study because patients with diabetes were excluded. These results support the hypothesis that metformin may protect against AMD development in patients without diabetes.3,4 In our study, the odds reduction of AMD associated with metformin use was observed in all but the highest cumulative dose quartile in the any AMD group and, in the dry AMD group, was observed in the lower 2 quartiles of metformin use. Comparisons of the AORs and the results of the tests for trend suggest that the benefits of metformin may not be dose dependent.
Notably, the reduced odds of AMD associated with metformin use was also observed in participants taking insulin, sulfonylureas, statins, or glitazones. This finding may be due to a true causal association but may also be a result of confounding factors. In vitro studies have shown that glyburide, a sulfonylurea, has retinal neuroprotective properties.11 Observational studies of statins have also yielded similarly encouraging results, further bolstered by those of a small randomized clinical study.12,13 Metformin, however, is of particular promise due to its safety profile, which makes it an ideal candidate for widespread use in the general population. Other antidiabetic medications pose a greater risk for complications, including hypoglycemia.
Limitations
The study has several limitations. A substantial limitation is the potential misclassification of patients without diabetes. Given that a small portion of cases and controls were exposed to other antidiabetic medications, these patients’ claims may have had diabetes diagnosis codes entered 1 to 2 years prior to their index visit. To be included in our study, however, these patients would have then needed to have had no diabetes diagnosis codes in the 12 months prior to their index visit, which is unlikely given the chronic nature of diabetes. Similarly, there is a risk of incorrect AMD coding, although this was minimized by methodology requiring 2 eye examinations in the 12 months prior to the index date without a code for AMD.2 We were unable to adjust for risk factors of AMD not captured by ICD9/10 coding, such as the amount of drusen and presence of pigment.14 We did account for well-established clinical and demographic risk factors for AMD. Due to limitations in ICD-9/10 coding, we were also unable to explore the association of metformin with specific dry AMD subtypes. This is a notable limitation, as geographic atrophy is a particularly debilitating form of AMD for which treatment is scarce. There is a need to determine if these results can be generalized to a healthy population without comorbid indications for metformin, which will further minimize confounding by indication.
Conclusions
This case-control study found that metformin use was associated with a reduced odds of any AMD and dry AMD in a sample of patients without diabetes. This association does not appear to be dose dependent. Randomized clinical trials are warranted to help determine if the association between metformin and AMD among patients without diabetes is causal and if the management of AMD should change.
eTable. Odds Ratios for Developing AMD Across Metformin Dosing Quartiles
Data Sharing Statement
References
- 1.Romdhoniyyah DF, Harding SP, Cheyne CP, Beare NAV. Metformin, a potential role in age-related macular degeneration: a systematic review and meta-analysis. Ophthalmol Ther. 2021;10(2):245-260. doi: 10.1007/s40123-021-00344-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Blitzer AL, Ham SA, Colby KA, Skondra D. Association of metformin use with age-related macular degeneration: a case-control study. JAMA Ophthalmol. 2021;139(3):302-309. doi: 10.1001/jamaophthalmol.2020.6331 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Brown EE, Ball JD, Chen Z, Khurshid GS, Prosperi M, Ash JD. The common antidiabetic drug metformin reduces odds of developing age-related macular degeneration. Invest Ophthalmol Vis Sci. 2019;60(5):1470-1477. doi: 10.1167/iovs.18-26422 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Stewart JM, Lamy R, Wu F, Keenan JD. Relationship between oral metformin use and age-related macular degeneration. Ophthalmol Retina. 2020;4(11):1118-1119. doi: 10.1016/j.oret.2020.06.003 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Gokhale KM, Adderley NJ, Subramanian A, et al. Metformin and risk of age-related macular degeneration in individuals with type 2 diabetes: a retrospective cohort study. Br J Ophthalmol. 2023;107(7):980-986. doi: 10.1136/bjophthalmol-2021-319641 [DOI] [PubMed] [Google Scholar]
- 6.Lee H, Jeon HL, Park SJ, Shin JY. Effect of statins, metformin, angiotensin-converting enzyme inhibitors, and angiotensin II receptor blockers on age-related macular degeneration. Yonsei Med J. 2019;60(7):679-686. doi: 10.3349/ymj.2019.60.7.679 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Corcoran C, Jacobs TF. Metformin. In: StatPearls. StatPearls Publishing; 2023. Accessed July 22, 2023. [Google Scholar]
- 8.Mohammed I, Hollenberg MD, Ding H, Triggle CR. A critical review of the evidence that metformin is a putative anti-aging drug that enhances healthspan and extends lifespan. Front Endocrinol (Lausanne). 2021;12:718942. doi: 10.3389/fendo.2021.718942 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Quan H, Li B, Couris CM, et al. Updating and validating the Charlson Comorbidity Index and score for risk adjustment in hospital discharge abstracts using data from 6 countries. Am J Epidemiol. 2011;173(6):676-682. doi: 10.1093/aje/kwq433 [DOI] [PubMed] [Google Scholar]
- 10.Breslow NE, Day NE, Halvorsen KT, Prentice RL, Sabai C. Estimation of multiple relative risk functions in matched case-control studies. Am J Epidemiol. 1978;108(4):299-307. doi: 10.1093/oxfordjournals.aje.a112623 [DOI] [PubMed] [Google Scholar]
- 11.Berdugo M, Delaunay K, Naud MC, et al. The antidiabetic drug glibenclamide exerts direct retinal neuroprotection. Transl Res. 2021;229:83-99. doi: 10.1016/j.trsl.2020.10.003 [DOI] [PubMed] [Google Scholar]
- 12.Ma L, Wang Y, Du J, Wang M, Zhang R, Fu Y. The association between statin use and risk of age-related macular degeneration. Sci Rep. 2015;5(1):18280. doi: 10.1038/srep18280 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Guymer RH, Baird PN, Varsamidis M, et al. Proof of concept, randomized, placebo-controlled study of the effect of simvastatin on the course of age-related macular degeneration. PLoS One. 2013;8(12):e83759. doi: 10.1371/journal.pone.0083759 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Ferris FL III, Wilkinson CP, Bird A, et al. ; Beckman Initiative for Macular Research Classification Committee . Clinical classification of age-related macular degeneration. Ophthalmology. 2013;120(4):844-851. doi: 10.1016/j.ophtha.2012.10.036 [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
eTable. Odds Ratios for Developing AMD Across Metformin Dosing Quartiles
Data Sharing Statement
