Introduction: We investigated the associations between antecedent all-cause CVD diagnoses, cause-specific CVD diagnosis, and CVD medication prescriptions with the risk of developing amyotrophic lateral sclerosis (ALS). Materials and Methods: We conducted a population-based case-control study of U.S. Medicare enrollees from 2006 to 2013. The final sample included 3,714 incident ALS cases and 18,570 controls (matched on age, sex, enrollment length, and county). Information was collected from Medicare Parts A, B, and D administrative claims data on hypertension, ischemic heart disease, heart failure, acute myocardial infarction, atrial fibrillation, prescriptions of angiotensin-converting enzyme inhibitors, angiotensin II receptors blockers, calcium channel blockers, beta blockers, and antiarrhythmics. Associations were evaluated using conditional logistic regression adjusting for age, sex, race/ethnicity, geographical location, alcohol and tobacco use, and socioeconomic status. Results: The odds ratio (OR) for having one or more ICD-9 codes for any cardiovascular disease diagnosis at least 24 months prior to the date of ALS diagnosis was 0.85 (95% confidence interval [CI]: 0.78–0.92). Cardiovascular conditions that were inversely associated with ALS included heart failure (OR = 0.79; 95% CI 0.70–0.89), atrial fibrillation (OR = 0.81; 95% CI 0.77–0.92), and hypertension (OR = 0.91; 95% CI 0.84–0.98). Exposures to several classes of cardiovascular medications were inversely associated with ALS risk even after adjusting for confounding by indication, including ACE inhibitors (OR = 0.84, 95% CI 0.77–0.91), calcium channel blockers (OR = 0.64, 95% CI 0.59–0.70), and beta blockers (OR = 0.76, 95% CI 0.71–0.83). Discussion/Conclusion: These findings merit additional research, including animal studies and pilot clinical trials, to further evaluate and evidence the effects of ACEIs, CCBs, and BBs on the risk of developing and clinical expression of ALS.

© 2022 The Author(s). Published by S. Karger AG, Basel

Introduction

Amyotrophic lateral sclerosis (ALS), commonly known as Lou Gehrig’s disease, is a motor neuron disease (MND) characterized by progressive neurodegenerative changes to upper and lower motor neurons. The median survival time of ALS symptom onset to death ranges from 20 to 48 months, and approximately 10–20% of ALS patients survive longer than 10 years [1]. In 2015, the estimated prevalence of ALS in the USA was 5.2 per 100,000 [2]. The incidence and prevalence of ALS increase with age, and as the proportion of older persons in the U.S. population increases, the number of individuals affected with ALS is expected to increase by 34% from 2015 to 2040 in the USA and by 69% globally [3]. This expected increase in ALS cases will place an immense economic burden on healthcare systems, in addition to the personal socioeconomic costs and health-related quality of life impacts among ALS patients [4, 5].

Despite decades of research, little is known about the nongenetic etiology of ALS. Given that middle and older-aged individuals are at particular risk for ALS, there is considerable interest in whether antecedent conditions, or the medications taken for these conditions, may be associated with the risk of developing ALS. Cardiovascular diseases (CVDs) and associated medications, if associated with ALS, would have important public health implications for pathogenesis, prevention, and possibly treatment. To date, nine population-based studies have assessed the association between ALS diagnosis and CVD conditions, with inconsistent findings among studies [6-16]. Antecedent prescriptions of cardiovascular medications have been inversely associated with ALS in population-based case-control studies in the USA, Taiwan, and Italy [17-19].

A comprehensive examination of antecedent CVD conditions and medications and their association with ALS among U.S. patients has not been carried out. Using a population-based case-control design, we studied the relationship between hypertension, ischemic heart disease, heart failure, acute myocardial infarction, and atrial fibrillation with the risk of developing ALS in a U.S. population. Additionally, we investigated the association between classes of medications prescribed to treat these conditions including angiotensin-converting enzyme inhibitors (ACEIs), angiotensin II receptors blockers (ARBs), calcium channel blockers (CCBs), beta blockers (BBs), and antiarrhythmics (AAs) and ALS risk.

MethodsStudy Population

We constructed a retrospective cohort of adults for the years 2006–2013 using a random 20% nationwide sample of Medicare fee-for-service beneficiaries using research-identifiable files from the Centers for Medicare and Medicaid Services (CMS). Inclusion criteria required (1) continuous enrollment in fee-for-service Medicare Part A (hospital) and B (outpatient) for a minimum of two years to allow for the assessment of prior health conditions; (2) age ≥65 years at entry into Medicare; and (3) simultaneous coverage by Part D (drug benefits) so that medication prescriptions could be ascertained. Beneficiaries entered the study cohort when they enrolled in fee-for-service Medicare Parts A, B, and D, and they were followed from the date of first enrollment until death, loss to follow-up, or the end of the study period.

ALS Case Ascertainment

For the same time period and using the same cohort selection criteria as for the overall cohort, we obtained a 100% sample of all possible patients with ALS on the basis of having one or more ALS or MND International Classification of Disease ICD-9 diagnostic codes. We did this to maximize the sample size of cases with this relatively rare condition. We applied the ALS case definition criteria used by the National ALS Registry, previously validated against El Escorial criteria, which has 88.7% positive predictive value, 65.9% sensitivity, and 99.8% specificity [2, 20]. The algorithm identifies all ICD-9 codes for any MND, and incorporates information on prescriptions for riluzole (a glutamate antagonist prescribed primarily for ALS) and provider types associated with the MND diagnosis.

Control Selection

We used incidence density sampling to randomly select 5 controls for each ALS case from among cohort members who were alive when the ALS case was diagnosed but had no prior diagnostic codes for ALS and had not been treated with riluzole. Control participants were individually matched to cases on sex, year of birth (±1), date of cohort enrollment, and county of residence at case’s diagnosis date. When 5 controls could not be identified for a given case, county was relaxed to state. If 5 controls were still not found, matching on year of birth was further relaxed. Relaxed matching criteria were required for only 12% of controls. Each control was assigned the same index date as their matched case, which ensured that the length of follow-up was the same in both the case and his or her matched controls. This design feature is very important for achieving “equal opportunity” for assessing exposure to antecedent factors and is the most rigorous method for carrying out such analyses in case-control studies [21].

Antecedent Cardiovascular Conditions

Primary and secondary CVD hospital admissions were identified using evidence-based adjudicated algorithms recommended by CMS based on the Chronic Conditions Warehouse (CCW) [22]. The CCW algorithms were used to identify cases and controls with ICD-9 and National Drug Code (NDC) codes for each condition for each year from 2006 to 2013 in any of the three claims sources (Medicare Parts A, B and D). CVD conditions included hypertension, ischemic heart disease, heart failure, acute myocardial infarction, and atrial fibrillation. Details on the algorithms and qualifying ICD-9 diagnosis for each cardiovascular condition can be found in Online Supplementary Table 1 (for all online suppl. material, see www.karger.com/doi/10.1159/000526982).

Antecedent Cardiovascular Disease Medications

CVD medications (including ACEIs, ARBs, CCBs, BBs and AAs) were identified using filled outpatient prescriptions contained in Part D Medicare data. Each filled prescription included information on date dispensed, days supplied, quantity dispensed, and doses. For the primary analysis, individuals were categorized as having a CVD medication prescription if they had at least one prescription filled for the given medication more than two years before the index date. For individuals with overlapping prescriptions, we calculated exposure days as the sum of the day’s supply of prescriptions if the overlap was ≤30 days. If the overlap was >30 days, we considered the more recent prescription, assuming that the new prescription had replaced the previous one.

Covariates

Previous studies have shown certain CVD risk factors associated with ALS in a pooled analysis to be less common among ALS cases than controls prior to disease development including diabetes (meta-analysis pooled relative risk 0.68; 95% confidence interval [CI], 0.55–0.84) [23] and obesity (odds ratio [OR] 0.73; 95% CI 0.55–0.96) [24, 25]. Smoking has been identified as a risk factor that is positively of large cohort studies (relative risk 1.4 for former and current smokers) [26]. In all multivariate analyses, we controlled for diabetes, obesity as a surrogate for body mass index (BMI), tobacco use, and socioeconomic status. CCW algorithms (Online Suppl. Table 1) were used to identify any diabetes-related hospital admissions, obesity, and tobacco use. Socioeconomic status was based on the presence/absence of one or more Medicare low-income subsidies (LIS) codes in the period between study entry and index date.

Statistical Analysis

We used conditional logistic regression to estimate crude and adjusted ORs with 95% CIs associated with each factor of interest, and in multivariate analyses adjusted for diabetes, obesity, tobacco use, and SES. We conducted a dose-response analysis for each medication class that was significantly associated with ALS (never prescribed, <2 years, 2–4 years, >4 years). We evaluated the presence of effect modification by sex, age, and race/ethnicity using multiplicative terms in unconditional logistic regression models. The primary statistical analyses for associations between CVD conditions, medications, and ALS utilized a claims exclusionary period of 24 months prior to the ALS diagnosis date (for cases) or index date (for matched controls) (Fig. 1). The basis for excluding 24 months immediately prior to ALS diagnosis from consideration is that there is a period of time prior to clinical recognition of ALS where the disease is likely present but not clinically recognized; prior research reports the average time between symptom onset and definitive diagnosis of ALS is approximately 14 months [27]. It is important to exclude this time period because factors occurring after disease onset cannot be considered risk factors, and conditions newly diagnosed during this period may simply be misattributions of early ALS symptoms. In addition to the primary analysis, we conducted two sensitivity analyses to determine the effects excluding one year and no time prior to an index date, enabling us to compare our results with those of past studies that examined the entire clinical history up to the diagnosis dates of ALS cases. To address confounding by indication for medications that were significantly associated with ALS, we stratified by any CVD condition and CVD medication use, and fit interaction terms between any CVD diagnosis and medication prescriptions. All statistical analyses were conducted using SAS version 9.4 (SAS Institute Inc.). This study was determined to be exempt by the Stanford University Institutional Review Board because it relied on de-identified claims data.

Fig. 1.

Claims exclusionary periods for statistical analyses of antecedent medical conditions and medications prior to clinical recognition of ALS. The primary analysis excluded the 24-month period immediately preceding the ALS index date, which was the first ALS/MND diagnostic code (for ALS cases) and a matching date for matched controls. Secondary analyses were conducted employing two alternate exclusionary periods (excluded the 12-month period immediately preceding the ALS diagnosis date and excluding no time prior the index date).

Results

The final study population consisted of 3,714 ALS cases and 18,570 matched controls. Cases and controls each contributed an average of 54.8 months of person-time for our primary analysis. The distributions of matching factors for cases and controls are presented in Table 1.

Table 1.

Characteristics of ALS cases and controls in a population-based nested case-control study in the U.S. Medicare population, 2006–2013

Table 2 presents the proportions of ALS cases and controls with antecedent all cause CVD, cause-specific CVD conditions, and CVD medications. The prevalence of any antecedent CVD condition during the period ending 24 months prior to clinical recognition of ALS was found to be 72.9% in ALS cases, as compared with 75.7% in controls. Having one or more CVD conditions was associated with a 15% reduction in ALS risk in matching variable-adjusted models (OR = 0.85; 95% CI: 0.78–0.92) and with a 7% reduction in risk in covariate and matching variable-adjusted models (OR = 0.93; 95% CI: 0.86–0.98). The associations were similar for heart failure, atrial fibrillation, and hypertension, which were associated with a 32%, 21%, and 18% reduction in ALS risk, respectively. Secondary analyses showed no evidence of effect modification by gender, age, or race/ethnicity (data not shown). However, in secondary analyses using the 12- and 0-month gaps between antecedent condition assessment period and index date, the matching variable-adjusted associations for any CVD condition with ALS in the 12- and 0-month gap analyses were no longer statistically significant (Online Suppl. Tables 5, 6).

Table 2.

Prevalence of antecedent cardiovascular condition and medication use, and association with amyotrophic lateral sclerosis in a population-based nested case-control study in the U.S. Medicare population, 2006–2013a

In matching variable-adjusted analyses, we found evidence of inverse associations between antecedent ACEI, CCB, BB, and AA prescriptions and ALS risk (Table 2). In models additionally adjusting for obesity, diabetes, and tobacco use, the inverse associations persisted for these medications. For example, having been prescribed a CCB was associated with a 36% reduction in ALS risk in fully adjusted models (adjusted OR = 0.64; 95% CI: 0.59, 0.70). Moreover, in secondary analyses using the 12- and 0-month gaps between the antecedent condition assessment period and the date of earliest ALS codes, we found very similar results for the inverse associations of ALS with ACEIs, CCBs, and BBs as we did when we used the 24-month exclusionary period. We did not find a significant association between the prescriptions of ARBs and ALS, therefore they were not included in dose-response or confounding by indication analyses.

When examining dose-response relationships with duration of medication prescriptions, ALS risk decreased with increased duration of ACEIs, CCBs, BBs, and antiarrhythmic prescriptions (Table 3). For example, having been prescribed a CCB for less than 2 years was associated with a 25% reduction in ALS risk compared to those never prescribed CCBs, while CCB prescriptions of 2–4 years and more than 4 years were associated with 44% and 70% reductions in risk, respectively (p value for trend <0.0001). Having been prescribed a BB for less than 2 years was associated with an 18% reduction in ALS risk, while BB prescriptions of 2–4 years and more than 4 years were associated with 25% and 51% reductions in risk, respectively (p for trend <0.0001).

Table 3.

Observed estimates for the dose-response relationship of cardiovascular medications at least 3 years prior to amyotrophic lateral sclerosis diagnosis in a population-based nested case-control study in the U.S. Medicare population, 2006–2013a

Interaction analyses examining the main and joint effects of each CVD medication class and any CVD condition beginning at least 2 years prior to ALS diagnosis are presented in Table 4. The referent group had no CVD diagnoses and no use of antecedent CVD medications. ACEI prescriptions were significantly inversely associated with the risk of ALS among subjects who had one or more codes for CVD conditions (OR = 0.83; 95% CI: 0.76, 0.90) and those who had no coded CVD diagnoses (OR = 0.64; 95% CI: 0.46, 0.91). Similarly, CCB and BB prescriptions were significantly inversely associated with ALS risk among subjects who had one or more codes for CVD conditions and those who had no coded CVD diagnoses. In contrast, AAs were borderline statistically significant after adjustment for confounding by indication and were inversely associated with ALS risk only among those with a CVD indication. Online Supplementary Tables 2–4 present the main and joint effects of ACEIs, CCBs, and BBs according to CVD-specific indications and largely confirm the results presented in Table 4.

Table 4.

Main and joint associations of cardiovascular medications and any cardiovascular condition at least 3 years prior to amyotrophic lateral sclerosis diagnosis in a population-based nested case-control study in the U.S. Medicare population, 2006–2013

Discussion

To our knowledge, our study is the first to comprehensively examine the effect of associations between antecedent CVD conditions and cardiovascular medications on ALS risk in a population-based study in the USA. Having one or more CVD conditions was associated with a 15% reduction in ALS risk, a finding that was robust to covariate adjustment. When examining CVD conditions separately, inverse associations were observed for heart failure, hypertension, and atrial fibrillation. Prior ischemic heart disease and myocardial infarction were not associated with ALS. We observed inverse associations with several classes of CVD medications (ACEIs, CCBs, and BBs), with a significant dose-response with increasing duration of use for all three classes. These findings occurred among those who did not have an apparent CVD indication and persisted even after adjusting for confounding by indication.

Our findings are consistent with the interpretation that people who develop ALS have had a more favorable cardiovascular profile than controls, as past studies have indicated they are less likely to be obese or have diabetes prior to disease onset [21, 23, 24], and are possibly more likely to have led a physically vigorous life [27, 28]. A few epidemiologic studies of antecedent CVD have largely focused on using administrative claims data because patient self-report can be incomplete and subject to recall bias [8]. An early case-control study conducted in 1991 in Rochester, Minnesota found antecedent hypertension was less frequent among male patients with ALS than male control subjects (OR = 0.10, 95% CI: 0.01–0.88) but was not associated with ALS among women [29]. Results from a recent large-scale national linkage study conducted in England were consistent with our findings. The relative risk for any CVD condition associated with ALS was 0.88 [95% CI: 0.82–0.95], very similar to our study findings [14]. In contrast, a large-scale Danish study found a modest positive association of antecedent CVD with ALS at least three years prior to ALS diagnosis (OR = 1.15; 95% CI 1.04–1.27); however, results varied across CVD cause-specific admissions and there was no association with hypertensive or cerebrovascular diseases [7]. The Danish study sample was largely based on patients identified through hospitalizations. This could have created a spurious positive association of CVD with ALS due to Berkson’s bias, a type of selection bias that occurs when factors associated with the risk of hospitalization appear to be associated (i.e., both CVD and ALS increase the risk of hospitalization) [30-32].

Our finding that three classes of medications (ACEIs, CCBs, and BBs) are inversely associated with ALS in a dose-response fashion, even after adjusting for the presence of CVD, and even among patients who were prescribed these medications for non-CVD conditions, raises the possibility that those medications could have a biological effect on reducing the risk of ALS or delaying its clinical expression. Three previous case-control studies in Taiwan [18], Italy [17], and the USA [19] have examined the association of certain classes of CVD medications with ALS. A study in Taiwan during 2002–2008 (729 ALS cases, 14,580 controls) reported a significant dose-dependent inverse association between cumulative defined daily dose (cDDD) of ACEIs and ALS risk [18], with a 17% reduction in risk in the group with a cDDD lower than 449.5 (OR = 0.83; 95% CI, 0.65–1.07) and 57% reduction for those cDDDs greater than 449.5 (OR = 0.43; 95% CI, 0.26–0.72) [18]. In contrast, a population-based case-control study of a very similar design conducted in Northern Italy from 2000 to 2010 found no significant association of ALS with antecedent ACEIs (OR = 0.9, 95% CI 0.8–1.1) or ARBs (OR = 1.0, 95% CI 0.8–1.2) [17]. A subsequent large drug screen of U.S. Medicare beneficiaries during 2008–2014 (10,450 ALS cases, 104,500 controls), after excluding the period 3 years prior to ALS diagnosis from consideration, found significant inverse associations for 4 of 15 hypertension medications (ACEI [lisinopril] OR 0.86; calcium channel blocker [amlodipine] OR 0.85; a beta blocker [metoprolol] OR 0.84; diuretic [furosemide] OR 0.75) [19]. A recent systematic review and meta-analysis of ACEI studies reported significant inverse associations for ACEIs (OR 0.81, 95% CI 0.74–0.89), CCBs (OR 0.85, 95% CI 0.79–0.93), beta blockers (OR 0.82, 95% CI 0.76–0.90), and diuretics (OR 0.87, 95% CI 0.81–0.93) [33]. Our findings were very similar for ACEIs (OR 0.84), and even more strongly inversely associated for CCBs (OR 0.64) and BBs (0.76). Adding to the findings of previous studies, our study suggests a possible biological protection for ACEIs, CCBs, and BBs that merits investigation in future observational and laboratory studies.

It is important to consider noncausal explanations of the inverse associations we observed, particularly for the three classes (ACEIs, CCBs, and BBs) for which we observed a significant dose-response with increasing duration of use. A noncausal reason for observing an inverse association with those who have no apparent indication would be if the medical indications themselves were less common among ALS patients than controls. This explanation is implausible, however, since off-label indications for these medications are relatively limited and include migraine headaches (CCBs, BBs), Raynaud’s (CCBM, BBs), tremors (BBs), and diabetic renal disease (ACEI). We know of no evidence suggesting that these conditions occur less frequently among ALS cases than controls. Another noncausal explanation would be if these medications suppressed symptoms of ALS which could delay clinical detection; however, we know of no plausible mechanisms by which these agents would mask symptoms such as muscle weakness. Lastly, it is possible that patients treated with CVD medications are likely to have more severe CVD; however, we cannot suggest a plausible mechanism whereby having more severe CVD would be a protective factor for ALS.

A variety of biologic mechanisms have been implicated in ALS pathogenesis that could explain possible neuroprotective mechanisms of ACEIs, CCBs, and BBs. For example, increased brain ACE activity is associated with activation of microglia and astrocytes at an early stage of the disease in SOD1G37R mice [34-36]; ACEI medications could inhibit those effects [37]. Other possible mechanisms of ACEIs include blocking glutamate-induced neurotoxicity [38], protecting against oxidative stress [39], and exerting neurotrophic actions on spinal motor neurons [40]. Our finding that the use of calcium channel blockers was inversely associated with ALS, with a very strong inverse dose-response relationship, suggests that CCBs could be exerting a biologically plausible protective effect. The specific vulnerability of motor neurons in ALS is thought to be due to calcium dysregulation in ALS, evidenced by Ca2+permeability of AMPA receptors causing excessive Ca2+ influx during glutamergic neurotransmission [41] and low cytosolic Ca2+ buffering properties of mitochondria [42]. Animal studies have shown that therapeutic agents targeting either Ca2+-permeable AMPA channels [43, 44] or mitochondrial calcium signaling [45] slow the loss of motor neurons in SOD1 animals. Therefore, therapeutic agents targeting calcium dysregulation provide important targets for neuroprotection in ALS [45, 46]. No previous studies have investigated the potential ALS-modifying effects of beta blockers; however, a recent study in the transgenic G93A SOD1 ALS model found that the non-selective lipophilic beta blocker (S-oxprenolol) that crosses the blood-brain barrier extended survival by 33% compared to placebo, and that mice treated with the highest daily dose of S-oxprenolol (20 mg/kg) had significantly longer survival than mice given riluzole [47]. Together with our findings, these studies implicating plausible biological mechanisms lead to the conclusion that future studies investigating the role of central acting ACEIs, CCBs, and BBs on ALS pathogenesis and preclinical disease progression are strongly warranted.

Our study is the first to examine several classes of CVD medications as antecedent factors in the risk of developing ALS; therefore, additional studies are needed before proceeding with randomized trials to evaluate whether the medications would be effective in the prevention and/or treatment of ALS. Trials of these agents could be carried out within a large cohort of individuals at genetic risk for developing ALS (i.e., who have affected family members). However, before conducting such trials, an effort should be made to investigate these medications in animal models of ALS to determine if they have an effect in delaying the pathophysiologic or clinical expression of the disease. Another approach would be to use a target trial emulation framework with existing observational datasets to determine whether CVD medications slow the disease course in patients with ALS [48]. Such a protocol would emulate the key design aspects of randomized trial design (i.e., inclusion/exclusion criteria, follow-up period, intention-to-treat, and per-protocol effects) to evaluate the effects of the CVD medications in preventing or treating ALS. This approach has been used to evaluate the effects of statin use on the risk of developing dementia [49].

Our study has some limitations. ALS and CVD misclassification is a possibility. Nevertheless, the use of validated claims algorithms for CVD conditions with known sensitivities and specificities in the U.S. Medicare population lends considerable weight to the evidence presented for the relationship between ALS incidence and CVD. Moreover, we expect any bias to be towards the null, not finding associations.

To our knowledge, this study represents the largest and most comprehensive investigation to date into the association between ALS and prior CVDs and medication prescriptions. Notable strengths of this study include utilizing a large comprehensive and representative integrated national claims database, the close match of controls to cases length of health claim records, and the exclusion of time periods during which early symptoms of ALS were likely to have been present (i.e., avoiding reverse causation). This is especially important as the median time between onset of ALS symptoms and diagnosis is estimated to be 11–14 months [9]. Additional studies, including animal studies and pilot clinical trials, are needed to further evaluate and evidence the effects of ACEIs, CCBs, and BBs on the risk of developing and clinical expression of ALS.

Statement of Ethics

Study approval statement: This study was determined to not qualify as human subject research by the Stanford University Institutional Review Board, therefore Institutional Review Board approval was not required. This study was determined to be exempt by the Stanford University Institutional Review Board because it relied on de-identified claims data.

Conflict of Interest Statement

Dr. Lorene Nelson receives grants from the National Institutes of Health, National MS Society, and Centers for Disease Control. Dr. Nelson receives compensation for serving as a consultant to Acumen, Inc. Dr. Kasarskis receives support for clinical ALS trials from the Healey platform trial program, AB Science, Alexion, and Amylx. Dr. Kasarskis also receives support from the Crispen and Team 7 endowments for ALS research. All other authors have no conflicts of interest to declare.

Funding Sources

This study was funded by a grant from Centers for Disease Control and Prevention (Agency for Toxic Substances and Disease Registries), Grant No. R01TS000249.

Author Contributions

Hoda S. Abdel Magid, Stanford University in Stanford, CA: primary data analysis and drafting and revision of manuscript. Barbara Topol, Stanford University in Stanford, CA: data analysis/interpretation, and revision of manuscript. Valerie McGuire and Jessica Hinman, Stanford University in Stanford, CA, and Edward Kasarskis, University of Kentucky in Lexington, KY: interpretation of data and revision of manuscript. Lorene Nelson, Stanford University in Stanford, CA: study concept and design, interpretation of data, and revision of manuscript.

Data Availability Statement

The original de-identified Medicare data for this study were obtained through the Research Data Assistance Center (ResDAC) at the University of Minnesota, which is authorized by the CMS to provide data to qualified researchers. For reasons of privacy protection, the authors are unable to distribute the raw data. Interested researchers desiring access to the raw data that were used to generate the analytical files should contact ResDAC directly regarding the possibility of licensing the data for the associated fees.

This article is licensed under the Creative Commons Attribution-NonCommercial 4.0 International License (CC BY-NC). Usage and distribution for commercial purposes requires written permission. Drug Dosage: The authors and the publisher have exerted every effort to ensure that drug selection and dosage set forth in this text are in accord with current recommendations and practice at the time of publication. However, in view of ongoing research, changes in government regulations, and the constant flow of information relating to drug therapy and drug reactions, the reader is urged to check the package insert for each drug for any changes in indications and dosage and for added warnings and precautions. This is particularly important when the recommended agent is a new and/or infrequently employed drug. Disclaimer: The statements, opinions and data contained in this publication are solely those of the individual authors and contributors and not of the publishers and the editor(s). The appearance of advertisements or/and product references in the publication is not a warranty, endorsement, or approval of the products or services advertised or of their effectiveness, quality or safety. The publisher and the editor(s) disclaim responsibility for any injury to persons or property resulting from any ideas, methods, instructions or products referred to in the content or advertisements.