Skip to main content
Advertisement
Browse Subject Areas
?

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

  • Loading metrics

Statins and renal disease progression, ophthalmic manifestations, and neurological manifestations in veterans with diabetes: A retrospective cohort study

  • Ishak A. Mansi ,

    Roles Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Project administration, Software, Supervision, Validation, Writing – original draft, Writing – review & editing

    ishak.Mansi@va.gov

    Affiliations Department of Education, Orlando VA Healthcare System, Orlando, Florida, United States of America, Department of Internal Medicine, University of Central Florida, Orlando, Florida, United States of America, Department of Internal Medicine, University of Texas Southwestern Medical Center, Dallas, Texas, United States of America

  • Matheu Chansard,

    Roles Data curation, Formal analysis, Methodology, Software, Validation

    Affiliation Department of Anesthesiology and Pain Management, University of Texas Southwestern Medical Center, Dallas, Texas, United States of America

  • Ildiko Lingvay,

    Roles Conceptualization, Formal analysis, Investigation, Methodology, Writing – review & editing

    Affiliations Department of Internal Medicine, University of Texas Southwestern Medical Center, Dallas, Texas, United States of America, Department of Population and Data Sciences, University of Texas Southwestern Medical Center, Dallas, Texas, United States of America

  • Song Zhang,

    Roles Conceptualization, Formal analysis, Methodology, Writing – review & editing

    Affiliation Department of Population and Data Sciences, University of Texas Southwestern Medical Center, Dallas, Texas, United States of America

  • Ethan A. Halm,

    Roles Conceptualization, Formal analysis, Methodology, Supervision, Writing – review & editing

    Affiliation Department of medicine, Robert Wood Johnson Medical School, New Brunswick, NJ, United States of America

  • Carlos A. Alvarez

    Roles Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Resources, Supervision, Validation, Writing – review & editing

    Affiliations Department of Population and Data Sciences, University of Texas Southwestern Medical Center, Dallas, Texas, United States of America, Department of Pharmacy Practice, Texas Tech University Health Sciences Center, Dallas, Texas, United States of America

Abstract

Background

Statins increase insulin resistance, which may increase risk of diabetic microvascular complications. Little is known about the impact of statins on renal, ophthalmologic, and neurologic complications of diabetes in practice. The objective of this study was to examine the association of statins with renal disease progression, ophthalmic manifestations, and neurological manifestations in diabetes.

Methods

This is a retrospective cohort study, new-user active comparator design, that included a national Veterans Health Administration (VA) patients with diabetes from 2003 to 2015. Patients were age 30 years or older and were regular users of the VA with data encompassing clinical encounters, demographics, vital signs, laboratory tests, and medications. Patients were divided into statin users or nonusers (active comparators). Statin users initiated statins and nonusers initiated H2-blockers or proton pump-inhibitors (H2-PPI) as an active comparator. Study outcomes were: 1) Composite renal disease progression outcome; 2) Incident diabetes with ophthalmic manifestations; and 3) Incident diabetes with neurological manifestations.

Results

Out of 705,774 eligible patients, we propensity score matched 81,146 pairs of statin users and active comparators. Over a mean (standard deviation) of follow up duration of 4.8 (3) years, renal disease progression occurred in 9.5% of statin users vs 8.3% of nonusers (odds ratio [OR]: 1.16; 95% confidence interval [95%CI]: 1.12–1.20), incident ophthalmic manifestations in 2.7% of statin users vs 2.0% of nonusers (OR: 1.35, 95%CI:1.27–1.44), and incident neurological manifestations in 6.7% of statin users vs 5.7% of nonusers (OR: 1.19, 95%CI:1.15–1.25). Secondary, sensitivity, and post-hoc analyses were consistent and demonstrated highest risks among the healthier subgroup and those with intensive lowering of LDL-cholesterol.

Conclusions

Statin use in patients with diabetes was associated with modestly higher risk of renal disease progression, incident ophthalmic, and neurological manifestations. More research is needed to assess the overall harm/benefit balance for statins in the lower risk populations with diabetes and those who receive intensive statin therapy.

Introduction

Type 2 diabetes mellitus has been considered a “cardiovascular risk equivalent” [1], resulting in a universal recommendation of statins for all patients with diabetes aged 40 to 75 with LDL-cholesterol 70 to 189 mg/dL for primary prevention of cardiovascular diseases (CVD) [2]. Despite their cardiovascular benefits, statins have also been shown to increase insulin resistance [38], which is thought to be a main driver of the pathogenesis of diabetic microvascular complications [913].

There is a paucity of data on the effects of statins on diabetic microvascular complications. The landmark cardiovascular randomized controlled trials (RCTs) that support the current guidelines for statin use for primary prevention did not, a-priori, evaluate their potential impact on diabetic microvascular complications. A handful of observational studies reported an increased risk of diabetes microvascular complications associated with statin use [1417], of which two studies were significantly larger (60,455 patients followed for a mean of 4 years and 25,970 patients followed for 6.4 years) [14, 15] than RCTs which established statins’ safety. However, other observational studies found no association between statin use and increased risk of diabetic neuropathy and/or diabetic retinopathy [1820] and two small trials (less than 50 participants each) associated statin use with improvement in diabetic retinopathy [21, 22].

Reasons for these conflicting results include inadequate adjustment for baseline confounders and the short duration of follow up. On one hand, statin use may be falsely associated with better outcomes because of healthy-user bias, and being a surrogate for higher quality of care, or better access to healthcare [23, 24]. Alternatively, statin use may be falsely associated with worse outcomes because of more exposure to healthcare resulting in ascertainment bias or confounding by indication [24].

The objective of this study was to address these methodological concerns by employing a new user design with active comparators to examine the association of statin therapy with incidence of renal disease progression, and diabetes with ophthalmic and neurological manifestations in a large national cohort of patients with diabetes in the Veterans Affairs (VA) health system who had significant longitudinal follow-up and who have detailed data on healthcare utilization, medical encounters, medication history, vital signs, and laboratory investigations to minimize confounding.

Methods

Study design

This is a retrospective cohort study using the national VA Corporate Data Warehouse (CDW), which encompasses inpatient and outpatient diagnosis/procedure codes, pharmacy, vital signs, and laboratory data. CDW catalogues its data according to published protocols (S1 File) [25]. This study cohort was assembled from a national VA cohort with diabetes identified using a validated algorithm [26] that has been described previously [27]. Briefly, we assembled a cohort of statin users and nonusers (overall cohort), aged 30 years or older, and who are regular VA users. We defined regular VA users as having all of the followings during each of the baseline and the follow-up periods: 1) at least one VA health care encounter; 2) blood pressure and weight measurements; 3) pharmacy records of medications; and 4) laboratory data that included blood/serum glucose, creatinine, and LDL-cholesterol. Available data for included patients encompassed all encounters from fiscal year (FY) 2003 to FY2015 (10/1/2002 to 9/30/2015) regardless of the date in which the patients were diagnosed with diabetes.

We used an active comparator, new user design, to mitigate the risk of immortal time bias and minimize confounding due to unmeasured characteristics [28]. We used newly initiated H2-blockers or proton pump-inhibitors (H2-PPI) as an active comparator to identify statin nonusers if they were not concurrently prescribed statins. Statin users were also newly initiated on statins. We excluded patients who previously filled prescriptions of either medication class within 12 months from cohort entry.

Index date was the date of the first prescription of statins or H2- PPI in their perspective groups. Since the study data included all available encounters from FY 2003 to FY 2015 regardless when patients were diagnosed with diabetes, the index date could have preceded, coincided, or followed their diagnosis of diabetes.

Study intervals

The study encompassed two intervals. The baseline period, which was used to describe baseline characteristics, included the year preceding the index date. The follow-up period, which was used to ascertain outcomes, started from the index date and continued until either: (1) the last available date of VA care, or (2) end of study period, (3) death, or (4) date of statin initiation in active comparators who subsequently used a statin; among this subset of patients who entered the cohort as active comparators but subsequently used a statin, the follow up period ended as active comparators (or outcomes were censored in time-to-event analysis) at date of statin initiation and were subsequently allowed entry into the cohort as statin users starting from the date of their statin initiation as a new index date for the statin user group. We excluded patients with fewer than 60 days of follow up duration from both groups since our main outcomes would be highly unlikely to be due to fewer than 60 days of statin exposure.

Outcomes

We used a combination of International Classification of Diseases, 9th Revision, Clinical Modification [ICD-9-CM] codes and laboratory investigations to identify outcomes. As previously published, to increase specificity of chronic diseases diagnoses using ICD-9-CM codes, we required each diagnosis to be present in ≥ 2 separate encounters [26, 29, 30].

Primary outcomes.

These incident outcomes occurred during follow up period but not at baseline

  1. Renal disease progression composite outcome: This dichotomous outcome comprised the presence of any of the following:
    1. Doubling of mean serum creatinine during the last year of follow up in comparison to mean serum creatinine during baseline.
    2. Incident stage 5 chronic kidney disease (CKD): Incident decrease in mean estimated glomerular filtration rate (eGFR) during the last year of follow up to <15 mL/min/1.73m2 (stage 5) [31], using The Modification of Diet in Renal Disease (MDRD) equation (S1 File) [32].
    3. Incident renal replacement therapy (S1 File).
    4. Incident diabetic nephropathy: As defined by the Agency for Health Research and Quality Clinical Classifications Software (AHRQ-CCS) multilevel diagnosis category 3.3.2 (S1 File) [33].
    5. Incident CKD: As defined by AHRQ-CCS diagnosis categories 156 and 158. Administrative diagnostic codes for renal events have been widely used to identify kidney diseases [34, 35] and their specificity was high (95–99%) [36].
  2. Incident Diabetes with ophthalmic manifestations: As defined by AHRQ-CCS multi-level category 3.3.3 [33].
  3. Incident Diabetes with neurological manifestations: As defined by AHRQ-CCS multi-level category 3.3.4 [33]. Administrative codes were commonly used in identifying patients with diabetic neuropathy in clinical research and utilization studies [3740].

Overall, administrative codes are useful for identifying diabetic complications [4145]. The sensitivity and specificity of ICD-9 codes for diagnosing diabetes with complications were 63.6% and 98.9%, respectively [46]. Diabetic complications codes are essential components in calculation of the Charlson comorbidity index [47] and the Elixhauser comorbidity score [48] from administrative data; both these scores are widely used [49].

Secondary outcomes.

  1. All individual components of the composite renal disease progression outcome.
  2. Change in mean creatinine (mg/dL) of individual patients during the last year of follow up in comparison to baseline.

Negative control outcomes.

To ensure that our findings were not due to unidentified confounders [50], we used two other outcomes that should not be affected by statins: 1) Chronic obstructive pulmonary disease (COPD) and 2) Suicide (S1 File) [51, 52].

Cohort characterization

Patients’ baseline characteristics [47], Charlson Comorbidity Index [47], and cardiovascular risk [53] were defined (S1 File). We created a propensity score to match statin-users and nonusers in the overall cohort at a ratio of 1:1 using 99 variables chosen a priori. Using the routine of Leuven and Sianesi, we performed multivariable logistic regression to estimate the propensity score and perform nearest number matching with a caliper of 0.0008 with no replacement (S1 File) [54, 55].

Primary analysis

We compared our primary, secondary, and negative control outcomes in the propensity score matched overall cohort using conditional logistic regression.

Secondary analyses

We compared our primary outcomes in the following prespecified cohorts (S1 File):

  1. The Overall cohort: Included all eligible patients before propensity score matching.
  2. Healthy cohort: Included only patients with a Charlson comorbidity index of zero at baseline.
  3. Intensive cholesterol lowering statin users in comparison to nonusers in the overall cohort [2].
  4. Medium-intensity cholesterol lowering statin users in comparison to nonusers in the overall cohort [2].
  5. Low-intensity cholesterol lowering statin users in comparison to nonusers in the overall cohort [2].
  6. Time-to-event analysis in the propensity score matched cohort: We estimated the hazard ratio (HR) in statin users in comparison to nonusers using survival regression analysis of the following outcomes: a) Incident CKD; b) Incident diabetes with ophthalmic manifestations; and c) Incident diabetes with neurological manifestations. We performed a separate regression analysis for each of these outcomes.
  7. Time-to-event analysis in the propensity score matched cohort with death as a competing risk: We estimated the subhazard ratio (SHR) in statin users in comparison to nonusers using survival regression analysis using similar outcomes to previous analysis.

Sensitivity analysis

We examined the odds of primary outcomes after excluding those who were diagnosed with incident diabetes, incident diabetic complications, or incident cardiovascular disease within 60 days of the index date. Since it is highly unlikely that statins influenced any of these outcomes within 60 days, excluding those patients further mitigated confounding by indication or residual confounding [5658].

Post-hoc analysis

We performed several post-hoc analyses:

  1. Propensity score-matched prevalent diabetes cohort: In this analysis, we restricted analysis to subjects with prevalent diabetes at index date. We, thereafter, created a propensity score to match statin-users and nonusers in this restricted cohort at a ratio of 1:1 using the same technique used earlier. We achieved balance in between comparison groups using a caliper of 0.00002 with no replacement.
  2. Ever user vs never user cohort: Excluded patients who started as active comparators and crossed over to statin users group.
  3. Incident diabetes complications cohort: Excluded patients who had any component of diabetes complications at baseline.
  4. Statin duration-based analysis: We stratified statin users by duration of statin use as < 3 year of statin use, or > 3 years of statin use. Each stratum of statin users was compared to nonusers for risk of each outcome in a separate logistic regression model adjusting for the propensity score and duration of follow up.
  5. Survival regression analysis with death as a competing risk in the intensive cholesterol lowering statin users in comparison to nonusers in the overall cohort and adjusting for propensity score. We performed this analysis because our secondary analysis showed that this cohort had the highest risk of complications among all other cohorts.
  6. Any retinopathy and its complications: Rather than using AHRQ-CCS codes, we used a different set of ICD-9-CM codes used by other researchers (S1 Table in S1 File) [30, 59, 60].

Other statistical analysis details

Dichotomous variables were compared using χ2 and continuous variables were compared using t-test. When Kolmogorov-Smirnov test indicated unequal distribution, we used the Wilcoxon Mann-Whitney test. We performed a separate logistic regression model for each dichotomous outcome in secondary and sensitivity analyses where the outcome was the dependent variable and statin use was an independent variable adjusting for the propensity score. Statistical significance was defined as two-tailed p-values < 0.05. Statistical analyses were performed using STATA version 15 (College Station, TX). The study was approved by the VA North Texas Health Care System and Texas Tech University Health Sciences Center Institutional Review Boards, which waived informed consent since data were fully anonymized before being accessed by the investigators. The study followed Reporting of Observational Studies in Epidemiology (STROBE) guidelines.

Results

Cohort assembly is shown in S1 File. A total of 705,774 patients fulfilled the study criteria (595,579 statin users and 110,195 active comparators); Cohort baseline characteristics are described in S1 File. We successfully matched 81,146 pairs of statin users and active comparators (nonusers) on all baseline characteristics and duration of follow-up period (approximately 4.8 ± 3.0 years) with the exception of proportions of some racial minorities and ethnicity (Table 1). Although all the cohort had diabetes by the end of the study, at baseline period not all patients were diagnosed with diabetes yet. In the propensity score-matched cohort, statin users and nonusers had similar proportions of patients diagnosed with diabetes, all diabetes complications, and similar glycemic control. Additionally, statin users and nonusers had similar utilization of glucose lowering agents, and similar creatinine and eGFR. Baseline lipid levels were higher in statin users than nonusers (Table 1). Overall, 63% of the statin prescriptions were for simvastatin, 12% for atorvastatin, 11% for rosuvastatin, 10% for pravastatin. As expected, statin users had a greater decrease in LDL-cholesterol during follow-up compared to nonusers (mean [SD] -25.2 [31.5] mg/dL in statin users and -0.9 [23.6] mg/dL in nonusers, p<0.001)–(S1 File).

thumbnail
Table 1. Baseline characteristics of propensity score-matched statin users and active comparators.

https://doi.org/10.1371/journal.pone.0269982.t001

Primary analysis

Statin use was associated with increased odds of renal disease progression (OR: 1.16, 95%CI: 1.12–1.20), ophthalmic manifestations (OR: 1.35, 95%CI: 1.27–1.44), and neurological manifestations (OR: 1.19, 95%CI: 1.15–1.25); (Table 2).

thumbnail
Table 2. Risk of outcomes during follow up period in propensity score matched cohort of statin users in comparison to active comparators.

https://doi.org/10.1371/journal.pone.0269982.t002

Statin users also had higher odds of incident Stage 5 CKD (OR: 1.14, 95%CI: 1.03–1.28); incident renal replacement therapy (OR: 1.11, 95%CI: 1.0–1.22); incident diabetic nephropathy (OR: 1.25, 95%CI: 1.15–1.37); and incident CKD (OR: 1.20, 95%CI 1.16–1.25). There was no difference in odds of doubling mean serum creatinine.

Negative control outcomes were similar between statin users and nonusers (Table 2); OR of chronic obstructive pulmonary diseases was 1.0 (95%CI: 0.98–1.02) and OR of suicide was 0.98 (95%CI: 0.93–1.04).

Secondary analysis

Statin users had higher ORs of all primary outcomes that were consistent throughout all cohorts (Table 3). The propensity score matched prevalent diabetes cohort showed overall results consistent with the primary analysis. The healthy cohort had the highest OR for all complications. Similarly, intensive cholesterol lowering statin users relative to nonusers also had the highest numerical OR of all diabetes complications than less intense cholesterol lowering cohorts. Sensitivity and survival analysis supported our main analysis for all outcomes (Table 4).

thumbnail
Table 3. Secondary analysis and sensitivity analysis comparing outcomes during follow between statin users vs active comparators.

https://doi.org/10.1371/journal.pone.0269982.t003

thumbnail
Table 4. Hazard/Subhazard ratio of outcomes in statin users in comparison to nonusers.

https://doi.org/10.1371/journal.pone.0269982.t004

Post-hoc analysis

All post-Hoc analyses had OR that were generally consistent in direction and magnitude with primary and secondary analyses (Tables 24). However, renal disease progression composite outcome in the ever user vs never user design had OR of 1.05, which was lower than other analyses (Table 3).

Statin duration analysis showed that statin users for < 3 years have no increased risk of the primary outcomes and may be decreased risk of incident diabetes with neurological manifestation. However, statin users for 3 years or more had increased risks of all outcomes.

Discussion

In this study of a national cohort of veterans with diabetes, statin users compared to nonusers had modest, but significantly higher risks of incident renal, ophthalmic and neurologic complications. The results were consistent across all analyses. Moreover, there was a dose-response relation to intensity of LDL-cholesterol lowering: statin users with intensive cholesterol lowering having the highest risk for all outcomes, which strengthens our confidence in these associations. The lack of associations with the negative control outcomes (COPD, suicide) adds to specificity of these findings.

Of specific interest is that patients without comorbidities (healthy cohort) had the highest increase in odd of adverse outcomes associated with statin use. For instance, OR of renal disease progression was 1.31 (vs. 1.17 in the overall cohort), that of ophthalmic manifestations was 2.09 (vs. OR = 1.33 in the overall cohort), and that of neurological manifestations was 1.48 (vs OR = 1.17 in the overall cohort). Our findings are consistent with those from a propensity score matched cohort of a healthy Tricare population (which contains both active duty soldiers and their families) who used statins as their only prescription medication [16]. In this study, OR of diabetes with complications was 2.15 in statin users compared to nonusers, but when the analysis was restricted to healthy active duty soldiers who are expected to be healthier and more physically active, the odds was even higher at 2.47 [17].

There is biological plausibility for these associations. Statins may increase the risk of diabetes microvascular complications through increasing insulin resistance [6] and inducing mitochondrial dysfunction resulting in more toxic effects of oxygen radicals [61]. Evidence from in-vitro studies, a Mendelian randomization study, and observational studies have demonstrated that statin therapy is associated with insulin resistance [68]. Insulin resistance is associated with increased risk of diabetic complications [9, 10], endothelial dysfunction, inflammation, hypercoagulability, and increased platelet reactivity [1113]. In presence of hyperglycemia, the availability of excessive intracellular glucose for oxidization in the tricarboxylic cycle results in production of larger amounts of electrons [62, 63]. Excessive electron burden generating superoxide may lead to diabetic complications [62, 63]. Statin therapy was associated with mitochondrial dysfunction [61, 64], which may compound the effects of superoxide. Beyond these basic science findings, we recently reported, using the same population, that statin initiation was associated with increased risk of diabetes treatment escalation and hyperglycemic events [27].

Our findings are also concordant with a recent metaanalysis of RCTs reporting that statins were associated with increased risk of renal insufficiency (OR: 1.14, 95%CI: 1.01–1.28) [65]. Yet, our study findings contrast with some of the scarce studies that examined this topic. A recent study reported that diabetic polyneuropathy in patients with newly diagnosed diabetes was similar in statin users compared to nonusers [20]. However, in that cohort 39% of new statin users discontinued their statin and 45% of statin nonusers initiated statins during the follow-up period. When the investigators censored patients at time of either initiating or discontinuing statins, there was an increased risk of diabetic polyneuropathy (HR = 1.17) [20]. Another nested matched study (with a median follow-up of 2.7 years) compared the risk of diabetic microvascular complications between patients who received statins prior to being diagnosed with diabetes and statin nonusers. Statin users had lower incidence of diabetic retinopathy (HR: 0.60) and diabetic neuropathy (HR: 0.66), but not diabetic nephropathy [18]. However, this study lacked several critical baseline characteristics such as body weight, obesity, a measure for comorbidity, or critical laboratory values, which along with the short duration of follow-up raise concerns for presence of confounders. In another retrospective cohort study, statin users had a lower rate of diabetic retinopathy (HR: 0.86) [60], however, the study excluded patients with LDL cholesterol <100 mg/dL. The study also lacked several important baseline characteristics including predictors of diabetic retinopathy such as diastolic blood pressure and glycemic control [66]. Additionally, our duration-based analysis may offer an insight into some aspects of the conflicting results in the literature since it suggests that statin use in our study was associated with risk of outcomes in those who used statins for longer duration (≥ 3 years) but not shorter duration (< 3 years). However, since our duration-based analysis was secondary post-hoc analysis, interpreting its findings should be considered exploratory indicating necessity of prospectively designed further research with this analysis defined a priori.

Any adverse effects of statins should be put in context of their well-demonstrated cardiovascular benefits. Table 5 compares the calculated number needed to be exposed for one additional harm (NNEH) based on the data from this study using previously published formula [67] and number needed to treat (NNT) for cardiovascular benefit from other studies [68, 69]. We recognize that not all events are clinically equivalent, so comparing the absolute NNEH v. NNT needs to be interpreted in a larger context. For example, a disabling stroke is clearly more morbid than doubling serum creatinine. However, a retinopathy resulting in blindness can be more devastating than revascularization following angina.

thumbnail
Table 5. Number needed to be exposed for one additional harm (NNEH) from this study and number needed to treat (NNT) for cardiovascular benefit from other studies.

https://doi.org/10.1371/journal.pone.0269982.t005

Weighing the balance of risks to benefits of statins would seem to be most important in the case of primary prevention where the absolute cardiovascular benefits are more modest, so higher risks of impactful non-cardiovascular outcomes might change decision making. Unfortunately, placebo -controlled RCTs of statins for primary prevention in the general population, which exclusively enrolled patients with diabetes or intended to specifically enroll more patients with diabetes, are limited to four RCTs (S1 File) [70].

Overall, these studies were of relatively modest size (<3000 patients in any study), were of relatively short duration (2.4–4.8 years), enrolled patients with multiple risk factors (other than diabetes), minimally (if any) assessed diabetic microvascular complications, and none of them showed a benefit (or did not report) on total mortality [7174]. Additionally, most of these studies were done in the past century where pharmacologic agents for diabetes control, blood pressure control, and smoking were different from the present era. Recently, the incidence of acute coronary events has been declining in developed countries [75] whereas the incidence of diabetes-related complications resurged [76]. As such, it is important to incorporate all available information and critically reassess the overall harm/benefit balance of statins on all outcomes. It might be possible that, especially in the lowest risk group where the absolute cardiovascular benefits are small, the subtle adverse metabolic effects associated with statin use might tip the net balance differently than currently assumed.

This study, to our knowledge, is the largest study to date that examined the association of statin use with risk of renal diseases progression, ophthalmologic, and neurological manifestations of diabetes. Several limitations are worth noting. Although we used several methodological techniques to mitigate immortal time bias, confounding by indication, and extensively described relevant baseline characteristics, residual confounding is always a concern in observational studies. This study may have underestimated the magnitude of the outcomes since a significant proportion of its population did not have diabetes at baseline, hence, their follow up may not be long enough to manifest diabetes complications. Additionally, some studies have associated use of PPI, which we used as a control group in our study, with modest increase in renal diseases, [77, 78] some neurological conditions [79], or ophthalmic conditions [80]. Though we had detailed longitudinal data on patients within the VA healthcare system, we did not have information on potential care outside the VA system. However, it is unlikely that care outside the VA would differentially affect statin users and nonusers. Finally, VA patients are predominantly males, which may limit generalization of our data; however, research shows that male VA patients have similar health characteristics as individuals with other insurance coverage suggesting greater generalizability [81].

In conclusion, we found that among patients with diabetes, statin use was associated with a modest but significant increased risk of renal, ophthalmic and neurologic manifestations. This risk was more pronounced with intensive LDL-cholesterol lowering and in healthier populations. Further research in the use of statins for primary prevention of CVD in patients with diabetes, in which renal, ophthalmic and neurologic outcomes are specifically evaluated as primary outcomes is needed to reliably assess the overall risk benefit ratio of statins in this large segment of the population. The ethos of primary prevention should be “first do no harm”.

Acknowledgments

Disclaimer

The views expressed herein are those of the authors and do not reflect the official policy or position of the Department of the Army, Department of Defense, VA Administration, or the US Government. The VA Health Care System, the University of Texas Southwestern and NIDDK had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication. One of the authors (IM) is an employee of the US government. This work was prepared as part of his official duties and, as such, there is no copyright to be transferred.

References

  1. 1. Executive summary of the third report of the National Cholesterol Education Program (NCEP) expert panel on detection, evaluation, and treatment of high blood cholesterol in adults (adult treatment panel iii). JAMA. 2001;285(19):2486–97. pmid:11368702
  2. 2. Stone NJ, Robinson JG, Lichtenstein AH, Bairey Merz CN, Blum CB, Eckel RH, et al. 2013 ACC/AHA guideline on the treatment of blood cholesterol to reduce atherosclerotic cardiovascular risk in adults: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines. Journal of the American College of Cardiology. 2014;63(25 Pt B):2889–934. pmid:24239923.
  3. 3. Nakata M, Nagasaka S, Kusaka I, Matsuoka H, Ishibashi S, Yada T. Effects of statins on the adipocyte maturation and expression of glucose transporter 4 (SLC2A4): implications in glycaemic control. Diabetologia. 2006;49(8):1881–92. pmid:16685502
  4. 4. Liew SM, Lee PY, Hanafi NS, Ng CJ, Wong SS, Chia YC, et al. Statins use is associated with poorer glycaemic control in a cohort of hypertensive patients with diabetes and without diabetes. Diabetology & metabolic syndrome. 2014;6:53. pmid:24782916; PubMed Central PMCID: PMC4003286.
  5. 5. Sukhija R, Prayaga S, Marashdeh M, Bursac Z, Kakar P, Bansal D, et al. Effect of statins on fasting plasma glucose in diabetic and nondiabetic patients. Journal of investigative medicine: the official publication of the American Federation for Clinical Research. 2009;57(3):495–9. pmid:19188844.
  6. 6. Swerdlow DI, Preiss D, Kuchenbaecker KB, Holmes MV, Engmann JE, Shah T, et al. HMG-coenzyme A reductase inhibition, type 2 diabetes, and bodyweight: evidence from genetic analysis and randomised trials. Lancet. 2015;385(9965):351–61. pmid:25262344; PubMed Central PMCID: PMC4322187.
  7. 7. Henriksbo BD, Lau TC, Cavallari JF, Denou E, Chi W, Lally JS, et al. Fluvastatin causes NLRP3 inflammasome-mediated adipose insulin resistance. Diabetes. 2014;63(11):3742–7. Epub 2014/06/12. pmid:24917577.
  8. 8. Mitchell P, Marette A. Statin-Induced Insulin Resistance Through Inflammasome Activation: Sailing Between Scylla and Charybdis. Diabetes. 63(11):3569–71. pmid:25342725
  9. 9. Kilpatrick ES, Rigby AS, Atkin SL. Insulin resistance, the metabolic syndrome, and complication risk in type 1 diabetes: "double diabetes" in the Diabetes Control and Complications Trial. Diabetes care. 2007;30(3):707–12. pmid:17327345.
  10. 10. Buse MG. Hexosamines, insulin resistance, and the complications of diabetes: current status. Am J Physiol Endocrinol Metab. 2006;290(1):E1–E8. pmid:16339923; PubMed Central PMCID: PMC1343508.
  11. 11. Kernan WN, Viscoli CM, Furie KL, Young LH, Inzucchi SE, Gorman M, et al. Pioglitazone after Ischemic Stroke or Transient Ischemic Attack. N Engl J Med. 2016;374(14):1321–31. pmid:26886418; PubMed Central PMCID: PMC4887756.
  12. 12. Mather KJ, Steinberg HO, Baron AD. Insulin resistance in the vasculature. J Clin Invest. 2013;123(3):1003–4. pmid:23454764; PubMed Central PMCID: PMC3582147.
  13. 13. Semenkovich CF. Insulin resistance and atherosclerosis. J Clin Invest. 2006;116(7):1813–22. pmid:16823479; PubMed Central PMCID: PMC1483180.
  14. 14. Mansi I, Frei CR, Wang CP, Mortensen EM. Statins and New-Onset Diabetes Mellitus and Diabetic Complications: A Retrospective Cohort Study of US Healthy Adults. Journal of general internal medicine. 2015;30(11):1599–610. pmid:25917657; PubMed Central PMCID: PMC4617949.
  15. 15. Mansi IA, Frei CR, Halm EA, Mortensen EM. Association of statins with diabetes mellitus and diabetic complications: role of confounders during follow-up. J Investig Med. 2017;65(1):32–42. Epub 2016/08/31. pmid:27574296.
  16. 16. Mansi IA, English J, Zhang S, Mortensen EM, Halm EA. Long-Term Outcomes of Short-Term Statin Use in Healthy Adults: A Retrospective Cohort Study. Drug safety: an international journal of medical toxicology and drug experience. 2016;39(6):543–59. Epub 2016/03/17. pmid:26979831.
  17. 17. Mansi IA, English JL, Morris MJ, Zhang S, Mortensen EM, Halm EA. Statins for primary prevention in physically active individuals: Do the risks outweigh the benefits? J Sci Med Sport. 2017;20(7):627–32. Epub 2017/02/12. pmid:28185810.
  18. 18. Nielsen SF, Nordestgaard BG. Statin use before diabetes diagnosis and risk of microvascular disease: a nationwide nested matched study. Lancet Diabetes Endo. 2014;2(11):894–900. WOS:000347779800016. pmid:25217178
  19. 19. Zhang J, McGwin G Jr., Association of statin use with the risk of developing diabetic retinopathy. Arch Ophthalmol. 2007;125(8):1096–9. Epub 2007/08/19. pmid:17698757.
  20. 20. Kristensen FP, Christensen DH, Callaghan BC, Kahlert J, Knudsen ST, Sindrup SH, et al. Statin Therapy and Risk of Polyneuropathy in Type 2 Diabetes: A Danish Cohort Study. Diabetes care. 2020. Epub 2020/10/02. pmid:32998990.
  21. 21. Gordon B, Chang S, Kavanagh M, Berrocal M, Yannuzzi L, Robertson C, et al. The effects of lipid lowering on diabetic retinopathy. Am J Ophthalmol. 1991;112(4):385–91. pmid:1928239.
  22. 22. Sen K, Misra A, Kumar A, Pandey RM. Simvastatin retards progression of retinopathy in diabetic patients with hypercholesterolemia. Diabetes Research and Clinical Practice. 2002;56(1):1–11. pmid:11879715
  23. 23. Dormuth CR, Patrick AR, Shrank WH, Wright JM, Glynn RJ, Sutherland J, et al. Statin adherence and risk of accidents: a cautionary tale. Circulation. 2009;119(15):2051–7. Epub 2009/04/08. CIRCULATIONAHA.108.824151 [pii] 10.1161/CIRCULATIONAHA.108.824151. pmid:19349320; PubMed Central PMCID: PMC2744446.
  24. 24. Mansi I, Mortensen E. The controversy of a wider statin utilization: why? Expert opinion on drug safety. 2013;12(3):327–37. pmid:23488561.
  25. 25. Corporate Data Warehouse (CDW); Health Services Research & Development. avaliable at: https://www.hsrd.research.va.gov/for_researchers/vinci/cdw.cfm. Acessed October 22, 2020.
  26. 26. Miller DR, Safford MM, Pogach LM. Who has diabetes? Best estimates of diabetes prevalence in the Department of Veterans Affairs based on computerized patient data. Diabetes care. 2004;27 Suppl 2:B10–21. pmid:15113777.
  27. 27. Mansi IA, Chansard M, Lingvay I, Zhang S, Halm EA, Alvarez CA. Association of Statin Therapy Initiation With Diabetes Progression: A Retrospective Matched-Cohort Study. JAMA Intern Med. 2021. Epub 2021/10/05. pmid:34605849.
  28. 28. Lund JL, Richardson DB, Sturmer T. The active comparator, new user study design in pharmacoepidemiology: historical foundations and contemporary application. Curr Epidemiol Rep. 2015;2(4):221–8. Epub 2016/03/10. pmid:26954351; PubMed Central PMCID: PMC4778958.
  29. 29. Kim SY, Servi A, Polinski JM, Mogun H, Weinblatt ME, Katz JN, et al. Validation of rheumatoid arthritis diagnoses in health care utilization data. Arthritis Res Ther. 2011;13(1):R32. Epub 2011/02/25. ar3260 [pii] 10.1186/ar3260. pmid:21345216; PubMed Central PMCID: PMC3241376.
  30. 30. Brown JB, Pedula KL, Summers KH. Diabetic retinopathy: contemporary prevalence in a well-controlled population. Diabetes Care. 2003;26(9):2637–42. Epub 2003/08/28. pmid:12941732.
  31. 31. Kidney Disease: Improving Global Outcomes CKDMBDUWG. KDIGO 2017 Clinical Practice Guideline Update for the Diagnosis, Evaluation, Prevention, and Treatment of Chronic Kidney Disease-Mineral and Bone Disorder (CKD-MBD). Kidney Int Suppl 2017;7(1):1–59.
  32. 32. Stevens LA, Coresh J, Feldman HI, Greene T, Lash JP, Nelson RG, et al. Evaluation of the modification of diet in renal disease study equation in a large diverse population. J Am Soc Nephrol. 2007;18(10):2749–57. Epub 2007/09/15. pmid:17855641.
  33. 33. Elixhauser A, Steiner C, Palmer L. Clinical Classifications Software (CCS) for ICD-9-CM. Databases and Related Tools from the Healthcare Cost and Utilization Project (HCUP) [Internet]. 2012 01/01/2012:[Appendix A p.]. Available from: http://www.hcup-us.ahrq.gov/toolssoftware/ccs/ccs.jsp (last accessed March 16, 2015).
  34. 34. Pasternak B, Wintzell V, Melbye M, Eliasson B, Svensson A-M, Franzén S, et al. Use of sodium-glucose co-transporter 2 inhibitors and risk of serious renal events: Scandinavian cohort study. BMJ. 2020;369:m1186. pmid:32349963
  35. 35. Winkelmayer WC, Schneeweiss S, Mogun H, Patrick AR, Avorn J, Solomon DH. Identification of individuals with CKD from Medicare claims data: a validation study. American journal of kidney diseases: the official journal of the National Kidney Foundation. 2005;46(2):225–32. Epub 2005/08/23. pmid:16112040.
  36. 36. Grams ME, Plantinga LC, Hedgeman E, Saran R, Myers GL, Williams DE, et al. Validation of CKD and related conditions in existing data sets: A systematic review. American journal of kidney diseases: the official journal of the National Kidney Foundation. 2011;57(1):44–54. Epub 2010/08/10. pmid:20692079; PubMed Central PMCID: PMC2978782.
  37. 37. Gore M, Tai K-S, Zlateva G, Bala Chandran A, Leslie D. Clinical Characteristics, Pharmacotherapy, and Healthcare Resource Use among Patients with Diabetic Neuropathy Newly Prescribed Pregabalin or Gabapentin. Pain Practice. 2011;11(6):528–39. pmid:21435162
  38. 38. Fincke BG, Miller DR, Turpin R. A classification of diabetic foot infections using ICD-9-CM codes: application to a large computerized medical database. BMC health services research. 2010;10:192–. pmid:20604921.
  39. 39. Hartsfield CL, Korner EJ, Ellis JL, Raebel MA, Merenich J, Brandenburg N. Painful Diabetic Peripheral Neuropathy in a Managed Care Setting: Patient Identification, Prevalence Estimates, and Pharmacy Utilization Patterns. Popul Health Manag. 2008;11(6):317–28. pmid:19108647.
  40. 40. Labovitz JM, Shofler DW, Ragothaman KK. The impact of comorbidities on inpatient Charcot neuroarthropathy cost and utilization. Journal of Diabetes and its Complications. 2016;30(4):710–5. pmid:26850144
  41. 41. Washington R, Andrews R, Mutter R. Emergency Department Visits for Adults with Diabetes, 2010. HCUP Statistical Brief #167. November 2013. Agency for Healthcare Research and Quality, Rockville, MD. Available at http://www.hcup-us.ahrq.gov/reports/statbriefs/sb167.pdf (last accessed March 16, 2015).
  42. 42. Wier L, Witt E, Burgess J, Elixhauser A. Hospitalizations Related to Diabetes in Pregnancy, 2008. HCUP Statistical Brief #102. December 2010. Agency for Healthcare Research and Quality, Rockville, MD. http://www.hcup-us.ahrq.gov/reports/statbriefs/sb102.pdf (last accessed March 16, 2015).
  43. 43. One in 16 Women Hospitalized for Childbirth Has Diabetes: AHRQ News and Numbers, December 15, 2010. December 2010. Agency for Healthcare Research and Quality, Rockville, MD. http://archive.ahrq.gov/news/newsroom/news-and-numbers/121510.html (last accessed March 16, 2015)Jan 28, 2015.
  44. 44. Fraze T, Jiang H, Burgess J. Hospital Stays for Patients with Diabetes, 2008. HCUP Statistical Brief #93. August 2010. Agency for Healthcare Research and Quality, Rockville, MD. http://www.hcup-us.ahrq.gov/reports/statbriefs/sb93.pdf (last accessed March 16, 2015).
  45. 45. Hines A, Barrett M, Jiang H, Steiner C. Conditions With the Largest Number of Adult Hospital Readmissions by Payer, 2011. HCUP Statistical Brief #172. April 2014. Agency for Healthcare Research and Quality, Rockville, MD. http://www.hcup-us.ahrq.gov/reports/statbriefs/sb172-Conditions-Readmissions-Payer.pdf (last accessed March 16, 2015).
  46. 46. Quan H, Li B, Saunders LD, Parsons GA, Nilsson CI, Alibhai A, et al. Assessing validity of ICD-9-CM and ICD-10 administrative data in recording clinical conditions in a unique dually coded database. Health services research. 2008;43(4):1424–41. pmid:18756617; PubMed Central PMCID: PMC2517283.
  47. 47. Deyo RA, Cherkin DC, Ciol MA. Adapting a clinical comorbidity index for use with ICD-9-CM administrative databases. Journal of clinical epidemiology. 1992;45(6):613–9. Epub 1992/06/01. 0895-4356(92)90133-8 [pii]. pmid:1607900.
  48. 48. Elixhauser A, Steiner C, Harris DR, Coffey RM. Comorbidity measures for use with administrative data. Medical care. 1998;36(1):8–27. Epub 1998/02/07. pmid:9431328.
  49. 49. Li B, Evans D, Faris P, Dean S, Quan H. Risk adjustment performance of Charlson and Elixhauser comorbidities in ICD-9 and ICD-10 administrative databases. BMC health services research. 2008;8:12. pmid:18194561; PubMed Central PMCID: PMC2267188.
  50. 50. Prasad V, Jena AB. Prespecified Falsification End Points: Can They Validate True Observational Associations? JAMA. 2013;309(3):241–2. pmid:23321761
  51. 51. Criner GJ, Connett JE, Aaron SD, Albert RK, Bailey WC, Casaburi R, et al. Simvastatin for the Prevention of Exacerbations in Moderate-to-Severe COPD. New England Journal of Medicine. 2014;370(23):2201–10. pmid:24836125.
  52. 52. Lilly SM, Mortensen EM, Frei CR, Pugh MJ, Mansi IA. Comparison of the risk of psychological and cognitive disorders between persistent and nonpersistent statin users. The American journal of cardiology. 2014;114(7):1035–9. pmid:25212545.
  53. 53. D’Agostino RB Sr., Vasan RS, Pencina MJ, Wolf PA, Cobain M, Massaro JM, et al. General cardiovascular risk profile for use in primary care: the Framingham Heart Study. Circulation. 2008;117(6):743–53. Epub 2008/01/24. pmid:18212285.
  54. 54. Becker S, Ichino A. Estimation of average treatment effects based on propensity scores. The Stata Journal. 2002;2(4):358–77.
  55. 55. Leuven E, Sianesi B. PSMATCH2: Stata module to perform full Mahalanobis and propensity score matching, common support graphing, and covariate imbalance testing. available at: http://ideas.repec.org/c/boc/bocode/s432001.html. version 4.0.5 ed 2003.
  56. 56. Danaei G, Rodriguez LA, Cantero OF, Logan R, Hernan MA. Observational data for comparative effectiveness research: an emulation of randomised trials of statins and primary prevention of coronary heart disease. Statistical methods in medical research. 2013;22(1):70–96. pmid:22016461; PubMed Central PMCID: PMC3613145.
  57. 57. Jee SH, Sull JW, Park J, Lee SY, Ohrr H, Guallar E, et al. Body-mass index and mortality in Korean men and women. The New England journal of medicine. 2006;355(8):779–87. pmid:16926276.
  58. 58. Allison DB, Faith MS, Heo M, Townsend-Butterworth D, Williamson DF. Meta-analysis of the effect of excluding early deaths on the estimated relationship between body mass index and mortality. Obesity research. 1999;7(4):342–54. pmid:10440590.
  59. 59. Bearelly S, Mruthyunjaya P, Tzeng JP, Suner IJ, Shea AM, Lee JT, et al. Identification of patients with diabetic macular edema from claims data: a validation study. Arch Ophthalmol. 2008;126(7):986–9. Epub 2008/07/16. pmid:18625948.
  60. 60. Kang EY, Chen TH, Garg SJ, Sun CC, Kang JH, Wu WC, et al. Association of Statin Therapy With Prevention of Vision-Threatening Diabetic Retinopathy. JAMA ophthalmology. 2019;137(4):363–71. Epub 2019/01/11. pmid:30629109; PubMed Central PMCID: PMC6459113.
  61. 61. Golomb BA, Evans MA. Statin adverse effects: a review of the literature and evidence for a mitochondrial mechanism. American journal of cardiovascular drugs: drugs, devices, and other interventions. 2008;8(6):373–418. Epub 2009/01/23. pmid:19159124; PubMed Central PMCID: PMC2849981.
  62. 62. Brownlee M. The pathobiology of diabetic complications: a unifying mechanism. Diabetes. 2005;54(6):1615–25. pmid:15919781.
  63. 63. Korshunov SS, Skulachev VP, Starkov AA. High protonic potential actuates a mechanism of production of reactive oxygen species in mitochondria. FEBS letters. 1997;416(1):15–8. pmid:9369223.
  64. 64. De Pinieux G, Chariot P, Ammi-Said M, Louarn F, Lejonc JL, Astier A, et al. Lipid-lowering drugs and mitochondrial function: effects of HMG-CoA reductase inhibitors on serum ubiquinone and blood lactate/pyruvate ratio. British journal of clinical pharmacology. 1996;42(3):333–7. Epub 1996/09/01. pmid:8877024; PubMed Central PMCID: PMC2042680.
  65. 65. Cai T, Abel L, Langford O, Monaghan G, Aronson JK, Stevens RJ, et al. Associations between statins and adverse events in primary prevention of cardiovascular disease: systematic review with pairwise, network, and dose-response meta-analyses. Bmj. 2021;374:n1537. Epub 2021/07/16. pmid:34261627.
  66. 66. Klein BE, Moss SE, Klein R, Surawicz TS. The Wisconsin Epidemiologic Study of Diabetic Retinopathy. XIII. Relationship of serum cholesterol to retinopathy and hard exudate. Ophthalmology. 1991;98(8):1261–5. Epub 1991/08/01. pmid:1923364.
  67. 67. Bender R, Blettner M. Calculating the "number needed to be exposed" with adjustment for confounding variables in epidemiological studies. Journal of clinical epidemiology. 2002;55(5):525–30. Epub 2002/05/15. pmid:12007557.
  68. 68. Vijan S, Hayward RA, American College of P. Pharmacologic lipid-lowering therapy in type 2 diabetes mellitus: background paper for the American College of Physicians. Ann Intern Med. 2004;140(8):650–8. Epub 2004/04/21. pmid:15096337.
  69. 69. Tonelli M, Lloyd A, Clement F, Conly J, Husereau D, Hemmelgarn B, et al. Efficacy of statins for primary prevention in people at low cardiovascular risk: a meta-analysis. CMAJ: Canadian Medical Association journal = journal de l’Association medicale canadienne. 2011;183(16):E1189–202. pmid:21989464; PubMed Central PMCID: PMC3216447.
  70. 70. Executive Summary: Standards of Medical Care in Diabetes—2008. Diabetes care. 2008;31(Supplement 1):S5–S11.
  71. 71. Sever PS, Poulter NR, Dahlof B, Wedel H, Collins R, Beevers G, et al. Reduction in cardiovascular events with atorvastatin in 2,532 patients with type 2 diabetes: Anglo-Scandinavian Cardiac Outcomes Trial—lipid-lowering arm (ASCOT-LLA). Diabetes care. 2005;28(5):1151–7. Epub 2005/04/28. pmid:15855581.
  72. 72. Knopp RH, d’Emden M, Smilde JG, Pocock SJ. Efficacy and safety of atorvastatin in the prevention of cardiovascular end points in subjects with type 2 diabetes: the Atorvastatin Study for Prevention of Coronary Heart Disease Endpoints in non-insulin-dependent diabetes mellitus (ASPEN). Diabetes care. 2006;29(7):1478–85. pmid:16801565.
  73. 73. Colhoun HM, Betteridge DJ, Durrington PN, Hitman GA, Neil HA, Livingstone SJ, et al. Primary prevention of cardiovascular disease with atorvastatin in type 2 diabetes in the Collaborative Atorvastatin Diabetes Study (CARDS): multicentre randomised placebo-controlled trial. Lancet. 2004;364(9435):685–96. pmid:15325833.
  74. 74. Collins R, Armitage J, Parish S, Sleigh P, Peto R, Heart Protection Study Collaborative G. MRC/BHF Heart Protection Study of cholesterol-lowering with simvastatin in 5963 people with diabetes: a randomised placebo-controlled trial. Lancet. 2003;361(9374):2005–16. pmid:12814710.
  75. 75. Institute of Medicine (US) Committee on Preventing the Global Epidemic of Cardiovascular Disease: Meeting the Challenges in Developing Countries. In: Fuster V, Kelly BB, editors. Promoting Cardiovascular Health in the Developing World: A Critical Challenge to Achieve Global Health Available from: http://wwwncbinlmnihgov/books/NBK45693/. The National Academies Collection: Reports funded by National Institutes of Health. Washington (DC)2010.
  76. 76. Gregg EW, Hora I, Benoit SR. Resurgence in Diabetes-Related Complications. JAMA. 2019. Epub 2019/04/16. pmid:30985875.
  77. 77. Guedes JVM, Aquino JA, Castro TLB, Augusto de Morais F, Baldoni AO, Belo VS, et al. Omeprazole use and risk of chronic kidney disease evolution. PLoS One. 2020;15(3):e0229344. Epub 20200304. pmid:32130255; PubMed Central PMCID: PMC7055824.
  78. 78. Al-Aly Z, Maddukuri G, Xie Y. Proton Pump Inhibitors and the Kidney: Implications of Current Evidence for Clinical Practice and When and How to Deprescribe. American journal of kidney diseases: the official journal of the National Kidney Foundation. 2020;75(4):497–507. Epub 20191010. pmid:31606235.
  79. 79. Makunts T, Alpatty S, Lee KC, Atayee RS, Abagyan R. Proton-pump inhibitor use is associated with a broad spectrum of neurological adverse events including impaired hearing, vision, and memory. Sci Rep. 2019;9(1):17280. Epub 20191121. pmid:31754136; PubMed Central PMCID: PMC6872761.
  80. 80. Schonhofer PS, Werner B, Troger U. Ocular damage associated with proton pump inhibitors. BMJ. 1997;314(7097):1805. pmid:9224084; PubMed Central PMCID: PMC2126950.
  81. 81. Wong ES, Wang V, Liu CF, Hebert PL, Maciejewski ML. Do Veterans Health Administration Enrollees Generalize to Other Populations? Med Care Res Rev. 2016;73(4):493–507. Epub 2015/11/22. pmid:26589675.