Patients with multiple chronic conditions are less likely to use generic drugs, even when facing lower co-payments.
Objectives: The impact of pharmacy benefit redesigns on increasing the use of generic drugs is not well studied in individuals using the most drugs: those with multiple chronic conditions. The objective of this study was to analyze whether generic substitution occurs in certain drug classes following a change in co-payments for working-aged adults with chronic conditions.
Study Design: This study uses pharmacy and medical claims data from Maryland’s high-risk pool from 2009 to 2011. An interrupted time series design exploits a natural experiment in plan drug benefit redesign that occurred in 2010. The pool lowered co-payments on generic drugs and raised them on preferred and nonpreferred brands.
Methods: Generalized estimating equations were used to analyze the impact of the policy change on the percentage of generics utilized in the most common chronic disease medication classes.
Results: Individuals’ generic use decreases as their number of chronic conditions increases. Antidepressant use increased 9% as a result of the policy, but this was not different for those with varying numbers of chronic conditions. The generic utilization rate remained unchanged for most other classes in the quarter immediately following the policy change.
Conclusions: The policy change impacted the generic utilization rate of antidepressants only, and across all classes and time points, generic use was lower among those individuals with more chronic conditions. Understanding why generic use decreases with more chronic conditions will be important in designing health insurance policies to encourage the use of generics, as the number of individuals with multimorbidities continues to increase.
Am J Pharm Benefits. 2017;9(2):-0
Generic medication use has grown substantially in the United States over the last 3 decades and now accounts for 83% of all prescriptions filled.1 Despite this overall increase, generic use is not this high across all plans, drug classes (with generic equivalents), or specific subpopulations.
As such, health plans continue to experiment with ways to increase the use of lower-cost generic drugs in the hopes of decreasing overall pharmacy expenditures. Newer value-based designs seek to incorporate the clinical effectiveness of drugs—not just their cost—when creating cost-sharing structures, such as lowering co-payments on highly effective drugs to treat chronic conditions.2,3
The savings to health insurers and patients can be substantial from increased generic substitution. One study estimated that the savings on just 3 drugs could be $100 million for state Medicaid programs.4 Another estimated that the savings to Medicare’s Part D program could be as much as $1 billion for every 10% increase in the use of generics.5 Using employer data, Liberman and Roebuck found that a 1% increase in the use of generics could lower plan expenditures for pharmaceuticals by 2.5%.6
Despite the potential for savings, the use of generics varies across particular drug classes. The Office of the Inspector General of the Department of Health and Human Services used the Medicare Part D program to examine the generic substitution rate across several drug classes in Part D plans (number of generic fills divided by the total number of generic plus multisource brand fills).
Across the Part D drug plans studied, the use of generics varied widely, within class across plans. The generic utilization rate could vary across drug classes over time as new brand drugs come onto the market or older ones go off patent. However, when generics were available, they accounted for 75% to 98% of diuretic prescriptions and for only 33% to 77% of diabetes therapies.7
To increase generic usage, one of the primary formulary tools has been to lower co-payments on generic drugs, or raise them on brand name drugs, or both, in some combination. In much of the literature on the impact of co-payments on prescription drug use, consumers have been shown to be responsive to co-payments when filling prescriptions.8
Several studies have examined the overall shift to generics from increased cost sharing for particular drug classes; however, they have neither focused on how the co-payment changes may impact those with multimorbidities differently, nor on whether they are more or less likely to shift to generics when brand name prices rise.9,10
Drug use is highest in those with multiple chronic conditions,11 and therefore, their responses may be different than those found in previous studies: these patients could be more responsive because they are taking more drugs or less responsive if they feel particularly committed to their current mix of medications. There is some indication from previous studies that adults with multiple chronic conditions may be less responsive to shifts in cost sharing even though co-payment increases can raise their out-of-pocket costs, sometimes substantially.12,13
If those with more chronic conditions are less responsive to shifts in co-payment overall, they may be more resistant to switching to generics for reasons other than cost, such as perceptions of effectiveness, provider characteristics, or regimen stability. Shrank et al conducted a patient survey on attitudes toward generic usage and found that although 70% of respondents said the generics were a better value, only 38% agreed that they would rather take generics.14
Physicians may also affect generic usage if they do not realize the variation in out-of-pocket costs each patient faces for various medications.15 Unfortunately, few studies have looked at predictors of generic usage in those with multiple chronic conditions and whether increased co-payments are a significant predictor of increased generic usage.
The objective of this paper is to examine whether the percentage of generic fills increased within particular drug classes for chronic diseases after changes in drug co-payment requirements. In July 2010, Maryland’s high-risk pool reduced co-payments on generic drugs and raised them on preferred and nonpreferred brand name drugs, creating a natural experiment.
The policy change was designed to increase incentives for people to substitute generics. Prior to the Affordable Care Act (ACA), high-risk pools offered coverage to those with pre-existing conditions who did not have affordable coverage in the private, individual market. High-risk pool enrollees were a group with chronic conditions, ideal for studying the impact of co-payment changes in generic usage for those with multiple morbidities.
Understanding how to control spending while maintaining health for those with multiple chronic conditions will be key in helping policy makers and insurers design better health plans, particularly as high-risk pools again enter the national policy discussion as one possible part of current reform proposals.
Data Source: High-Risk Pools
The ACA established preexisting condition plans to help those individuals with chronic conditions gain health insurance, that ended as the marketplaces came online in 2014.16 High-risk pools, however, existed in 35 states before the ACA—some for more than 30 years. Maryland had the country’s fourth largest high-risk pool until high-risk enrollees began entering the health insurance exchanges in 2014.17
The reforms to the rating rules for setting premiums in the individual market made the high-risk pools obsolete. Insurers are required to cover those with preexisting conditions, and there are premium tax credits that cap premium payments to a percentage of income if the person purchases the second-lowest cost silver plan. Several recent health care reform proposals at the federal level would re-instate the high-risk pool mechanism, making the data presented in this study helpful in understanding the potential health care utilization of high-risk pool enrollees.
To have qualified for a high-risk pool prior to the ACA, the person must have attempted to purchase coverage on the private individual market and been denied because of preexisting medical conditions. The Maryland Health Insurance Plan (MHIP) used funds from a statewide hospital tax to subsidize health insurance premiums, which were set at about 125% of the average premium in Maryland’s individual market.
For the average individual in 2012, the premium on the individual market was about $500 per month. High-risk pool enrollees did not have the option to enroll in other government programs, either because they had incomes higher than Medicaid eligibility levels or were not eligible for Medicare.
In July 2010, MHIP lowered the co-payments for generic drugs by 33% to 50% depending on the formulary tier and raised the co-payments on brand name and nonpreferred brand name drugs as much as 37%. The co-payments on the newly created specialty tier increased 40% for most of the MHIP plans. Table 1 shows the difference in co-payments for the various pharmacy drug tiers and for the different plan types within MHIP.
An interrupted time series study design was used to analyze the changes in the generic utilization rate (GUR) for selected classes, using administrative claims data from July 2009 to June 2011. The sample consisted of those who met the following requirements: continuously enrolled for the 2-year analysis period, aged 18 to 64 years, and had at least 1 fill in the selected classes in the year before the policy change.
Drugs were assigned to classes using Multum’s Lexicon database, which groups National Drug Codes into therapeutic classes. Sixteen classes were initially selected: antihyperlipidemic agents; antidiabetic agents; beta-adrenergic blocking agents (beta-blockers); angiotensin-converting enzyme (ACE) inhibitors; antihypertensive combinations; sex hormones; thyroid hormones; bronchodilators; diuretics; calcium channel blockers; leukotriene modifiers; antidepressants; anxiolytics, sedatives, and hypnotics; anticonvulsants; antipsychotics; and central nervous system (CNS) stimulants.
Due to the high proportion of generics already being used in this sample for diuretics, calcium channel blockers, ACE inhibitors, and beta-blockers in the pre-period, these classes were removed from consideration for further analysis. Leukotriene modifiers were also removed from further analysis because of the absence of generic alternatives.
The particular classes in this analysis are used to treat chronic conditions and therefore should be taken regularly; they were also the classes of drugs with the greatest volume of fills. Two exclusions should be noted: analgesics and antivirals were among the top classes by volume.
Analgesics were removed because of their use in acute pain management. A separate program in Maryland (and similar ones in many other states) paid some cost sharing for patients with HIV/AIDS, so these were also removed from the analysis.
The main covariates used are age, gender, number of chronic conditions, and plan type. Age is measured continuously, and gender is binary (1 = female). The number of chronic conditions is continuous. International Classification of Disease, Ninth Revision, Clinical Modification (ICD-9-CM) codes were translated into categories of conditions and chronic diseases using the Clinical Classifications Software from the Agency for Healthcare Research and Quality.18 For a condition to be counted as chronic, the ICD-9-CM code had to appear in at least 2 outpatient visits or at least 1 inpatient visit.
Plan type, such as preferred provider organization (PPO) or health maintenance organization (HMO), was also added as a series of dummy variables, which represent differences across plan benefit structure. Plan type also reflects differences in income, since MHIP provides a second set of plans, called MHIP Plus (MHIP+), for low-income individuals.
Prior to the implementation of the ACA, low-income childless adults in Maryland were also not eligible for Medicaid, and parents were only covered to 116% to 122% of the Federal Poverty Level (FPL).19 To qualify for MHIP+, the person must have had an annual income below 200% of the FPL ($21,660 in 2010).
The main outcome is the percentage generic of all drugs filled in a particular class. The GUR is calculated as the number of generic fills in the class divided by the total number of fills in the class. Other work in this area also calculates the generic substitution rate, which is the number of generics in the class divided by the total number of generics and multisource brands in a class.
This particular dataset did not have enough multisource brand fills to create a meaningful measure. An additional measure that could have been used would have been, at a more granular level, the number of days supplied. Because the objective of the paper was to focus on chronic disease medications that should be taken regularly, this outcome is saved for future analyses.
An interrupted time series design using individual-level data was used. The key variable of interest was the policy change indicator, where the null hypothesis is that in the absence of the policy change, the trend of generic utilization would have remained constant:
Yit = ƒ(β0 + β1timet + β2policyt + β3post timet + β4chronici + β5chronic × policyit + β6chronic × post timeit + Xitλ)
β1timet is the time trend, measured quarterly since July 2009 to account for secular increases in the use of generic drugs. MHIP runs on a fiscal year from July to June. The policy change is included as a binary indicator equal to 0 in the pre-period and 1 in the post period.
The post time variable measures the quarters continuously since the policy change, so it is equal to 0 in the pre-period and begins at 1 in July 2010. β2policyt + β3post timet represents the impact of the policy change for the whole sample, including both the initial impact and the change in the trend after the intervention.
β4chronici represents how patients with different numbers of chronic conditions will react. β5chronic × policyit represents how the policy impact changes across groupings of numbers of chronic conditions, and β6chronic × post timeit represents whether the trend in the post period is different for different numbers of chronic conditions after the policy change. Xitλ represents the vector of control covariates: age, gender, and plan type.
The outcomes are analyzed using generalized estimating equations (GEEs) specifications to analyze the impact of the policy change, because having a prescription in 1 month is highly correlated with having a prescription drug in the next month, particularly for patients with a chronic disease. GEE allows for the specification of the family (ie, Gaussian) and link function (ie, identity or log) for the mean, making this a very flexible regression model.
The correlation structure was assessed through examining the correlation between the outcome overtime. Binomial models were used given the distribution of the outcomes (a proportion varying from 0 to 1).
The final sample consisted of 6215 individuals, all of whom were aged 18 to 64 years, were continually enrolled across the 2-year study period, and filled at least 1 prescription in the selected classes in the pre-year (Table 2). Their average age was 51 years, with just over half female.
Table 2 also shows that as the number of chronic conditions increases, the number of different drug classes the person used increases. For example, individuals with 6 or more chronic conditions used more than 10 different classes of medications each quarter.
Table 3 shows the average GURs in the pre- and post-period. Without controlling for the general increase in generics over time, this table suggests whether increases in generic usage arose in particular classes after the policy change.
The percentage of generics used in the year before the policy change ranged from a low of 5% in the bronchodilators class to a high of 84% for the anxiolytics/sedatives. The quarterly GUR increased for most classes in the post period, but by small amounts.
Anxiolytics/sedatives, antidepressants, CNS stimulants, antihypertensives, antihyperlipidemics, sex hormones, and thyroid medications are classes that demonstrated significant increases in generic percentage in the post period (P <.05). For example, the GUR for antidepressants increased from 62% to 70% in the post period. The GUR of antihypertensives increased from 57% to 65% in the post period. Nevertheless, these unadjusted results do not control for the general increase in generic use over time.
Figure 1 shows the quarterly trend in generic usage separated by the number of chronic conditions. This figure shows that once a person has a chronic condition, they are less likely to be using generic drugs. The graph also shows that the usage of generic drugs is increasing steadily over time across all chronic conditions. However, visual inspection does not detect a marked shift in the trend or the level of generic usage at the policy change time point (marked with the vertical line).
The policy does not have a differential impact on those with different numbers of chronic conditions. Across most of the drug classes, the interaction of the number of chronic conditions and the policy change is not statistically significant, nor is the odds ratio on the interaction of the number of chronic conditions and the post time trend.
The eAppendix (available at ajpb.com) shows the full regression results for each drug class. For the classes of mental health medications, antidepressants are the only class with a statistically significant increase in the GUR in the post period, according to the significance level of the coefficient on the policy change in the regressions.
For the other coefficients in the model, plan type (eg, HMO, PPO) varies as to whether it is a significant driver in determining the GUR. The $500 PPO group was held as the reference group to examine the impact on the low-income plans in MHIP+.
The expectation is that plans less generous in terms of cost sharing, such as the high-deductible health plan (HDHP) and the HMO, may result in a higher GUR. However, the coefficient is not consistently significant in magnitude or direction across classes.
Gender is a significant predictor for many of the drug classes, although not the same direction for every class. For example, being female increases the use of sex hormones (largely birth control) and decreases the rate for thyroid hormones. Age was also a significant predictor for most classes.
A secondary effect that can be seen with this data is that across almost all drug classes, the percentage of generic drugs used is lower as the number of chronic conditions increases. In the immediate post period, the predicted use of generic antidepressants is 61% for those with 14 chronic conditions compared with 68% for those with no chronic conditions.
Because the predicted GUR will depend on all of the covariates used in a binomial model, the marginal effects are calculated to show the predicted probability of generic use within each class for several levels of chronic conditions and at 2 postperiod time points. (Marginal effects can be found for all classes in the eAppendix.)
Because it may take time for the policy change to impact the people taking the drugs, the trend may be different in the immediate post period versus a year later, so postperiod time points are shown. Figure 2 shows the marginal effects by number of chronic conditions and time period, clearly demonstrating a downward trend in the GUR as the number of chronic conditions increases.
This relationship holds for nearly all of the classes studied except for bronchodilators. For antidepressants and antihypertensives, the largest increase occurred in the quarter just after the policy change, with the increase in generic use growing more slowly over time.
Over the 2-year study period, generic utilization increased steadily in most classes for Maryland’s high-risk pool, reflecting a similar trend across the United States.20 The co-payment policy change did not impact the use of generics for most classes studied, nor were those with multiple chronic conditions more likely to be affected than those with fewer conditions. The policy change was associated with a significant increase in the GUR for antidepressants while the policy had no effect on many of the other drug classes.
Some studies have shown similar findings. A study by Wang et al is one of the few to examine the impact of being sicker on medication utilization when facing co-payment changes. The authors found that those with higher disease burden are less likely to reduce their adherence to medications than those who are healthier.13 Remler and Atherly construct a theoretical model for this behavior, and confirm with empirical data that those with more morbidities are less elastic in their response to price increases.12
Even evidence in the value-based design literature finds that those who are already on a regimen are less likely to see major increases in adherence to drugs than those who were nonadherent before.21 These findings indicate that individuals with more conditions may have legitimately less response to changes in co-payments.
A secondary, but significant, finding of the current study is that in almost all classes, except bronchodilators and anxiolytics, the effect of the person having more chronic conditions is to decrease the percentage generic used. Sensitivity analyses show the trend is unchanged, even after removing those who were over the out-of-pocket maximum in the second year. Plan types were also not significant in the regression models, except for antipsychotics, and the percentage of individuals with each grouping of chronic conditions was relatively evenly distributed across plans.
There is growing literature explaining why adults may be less likely to choose generic drugs, which includes reasons such as the availability of generics within a class, preferences for certain brand name drugs, patient and physician characteristics, and the formulary design of the health plan.11
Although the FDA requires that all generic drugs have the same active molecule, in real-world practice there are disputes as to whether the generic versions are exactly bioequivalent. Some drugs may have different inert ingredients to which some individuals may be allergic. The cost of switching drugs, real or perceived, may be high for some patients.
Some older, generic, mental health medications that are therapeutic substitutes for newer mental health medications such as antidepressants or antipsychotics are perceived to have particularly adverse effects, such as weight gain, tardive dyskinesia (involuntary movement), and metabolic problems.23,24
For many other drugs, studies have confirmed that, generally, there is little evidence indicating that generic versions of active ingredients operate differently than their brand name counterparts; however, this perception may still impact patient and prescriber behavior.22
Although these studies may explain the effect among all adults, it is not clear why those with multiple chronic conditions are not using more generics, particularly since they would have a financial incentive to switch to generics given their overall medication burden. It could be that they are sicker and are reluctant to switch once on a given regimen.
In a survey of adult prescription drug users, Shrank et al (2009) found that just under 40% of respondents agreed with the statement that they would prefer to take generics over brands, even though almost 70% of respondents agreed with the statement that generic drugs are a better value than brand name drugs; however, these findings are not explored by the number of chronic conditions. A small study in Portugal found patient preferences for generics were lower when the diseases were perceived as more serious.25
This may indicate that those with less serious conditions feel safer in switching to generics. Providers also have preferences for prescribing certain medications and may continue prescribing their currently preferred treatments if unaware that the patient has substantial co-payments.15,26
Another possible explanation is that once established on a regimen, individuals with multiple chronic conditions are less likely to switch to other drugs for fear of drug interactions or other destabilizations. Although polypharmacy—the notion of taking many drugs at once—has become a topic of interest for the elderly in recent years,27 there is little similar literature in the working-age adult population.
Working-age adults could have the same constellations of chronic conditions as the elderly, but may be better functioning at younger ages (ie, less frailty or cognitive decline). Due to the rising levels of multimorbidity in developed and developing countries, more work is needed in this area to understand generic drug use.11,28
In this study, there may be another possible explanation for the decreasing use of generics among those with higher numbers of chronic conditions and the lack of response to the policy change overall. It could be that those with more conditions could have selected to enroll in more generous health plans and are therefore more protected from the impact of cost sharing.
The distribution of chronic conditions across plan types is relatively similar in the pre-period, with the less generous plans (eg, HDHPs and HMOs), attracting a slightly lower percentages of those with 10 or more chronic conditions (18% and 17%, respectively) than the MHIP+, $500 PPO, and $1000 PPO (26%, 25%, and 24%, respectively). However, the models already control for plan type, and these coefficients are not a significant predictor, except for the antipsychotics class.
With an interrupted time series analysis, there is a concern that simultaneous patent expiries could be causing the shift in use, rather than the co-payment change. As a sensitivity analysis, any new drug (either brand or generic) with an approval date during the study period was dropped.
The antipsychotic class does have a noticeably higher proportion of generics in the pre- and post periods due to the entrance of olanzapine (Zyprexa) in late 2009. However, the regression results on this sample were nearly identical to the main results. The models also were not sensitive to the inclusion of the total number of drug fills per person per quarter as a covariate.
This analysis is limited in causal implications because there is no control group to rule out any further co-occurring changes that may impact generic drug utilization, such as the increasing use of co-payment cards that allow patients to access medications with no or reduced cost sharing.
At the time they were operating, high-risk pools contained members with preexisting conditions, making comparisons to a commercial population different. Furthermore, Maryland’s high-risk pool was one of the largest in the country at the time and there were no high-risk pools of similar size in the mid-Atlantic region with which to compare. Other populations, such as Medicare dual-eligibles, have a high number of comorbidities, but are elderly and face very different cost-sharing designs in Medicare Part D, thereby making this population a poor control as well.
Given how the GUR is calculated, it is possible that the generic percentage filled may appear to increase because the number of brand name medications drops while the number of generics filled stayed unchanged. If brand name drug use drops while generic use stays unchanged, then this suggests that some enrollees stop taking those brand name medications altogether, instead of substituting with a generic version.
For example, in the case of bronchodilators for asthma, where the main drug (Advair) was still under patent until 2012, there is a small dip in the number of brand name medications filled while the generic fills do not change; this suggests there may be a decrease in adherence. However, this is not the case in all classes. For antidepressants, there does appear to be a corresponding increase in generics as brand name drug fills decrease, indicating a substitution effect.
With claims data, there are substantial limitations in identifying other unmeasured factors that may also influence use. The sample was limited to 1 year continuously enrolled in the pre- and post periods surrounding the policy change. Those who dropped coverage during the year tended to be healthier and younger.
However, this pattern is not markedly different in the post period versus the pre-period. Finally, this analysis only examines the high-risk pool in 1 state, Maryland, so this may further limit the generalizability of the findings. Because those in high-risk pools opted to purchase the insurance, it could indicate this is a more price-insensitive cohort, and this could explain the effects seen in this study. If this explanation were true, it would limit the external validity of this study. However, other early work in this field with very different populations found similar price insensitivities.
Future research will use a larger dataset to substantiate the idea that those with multiple chronic conditions are not very responsive to co-payment changes and to confirm that those with higher numbers of chronic conditions are less likely to use generic drugs. This analysis highlights the difficulties in understanding why generic usage might be lower among those with multiple comorbidities, which is a limitation of any claims data analyses. Work in this area to further explore why generic drug usage is lower among persons with multiple morbidities could include survey or qualitative work to better understand this relationship.
The co-payment increase for this pool increased the generic usage of antidepressants, but did little to increase the trend toward generic usage for almost all other classes studied. There were several classes already at their maximum generic usage, particularly the cardiovascular-related medications.
Therefore, this policy did little to encourage those with multiple chronic conditions to switch to generics. Adjusting cost sharing alone may be of limited use in populations with multiple chronic conditions because there are additional factors impacting drug usage, other than price alone. Other interventions to increase the use of generics may be needed, such as those that would increase pharmacist or physician counseling regarding the patients’ mix of drugs.
The author would like to thank Gerard F. Anderson for his comments on many iterations of this draft.
Author Affiliation: RAND Corporation (CB), Arlington, VA.
Source of Funding: This work was funded with a dissertation grant from the Jayne Koskinas Ted Giovanis Foundation for Health and Policy.
Author Disclosures: The remaining authors report no relationship or financial interest with any entity that would pose a conflict of interest with the subject matter of this article.
Authorship Information: Concept and design (CB); acquisition of data (CB); analysis and interpretation of data (CB); drafting of the manuscript (CB); critical revision of the manuscript for important intellectual content (CB); statistical analysis (CB); provision of patients or study materials (CB); obtaining funding (CB); administrative, technical, or logistic support (CB); and supervision (CB).
Address Correspondence to: Christine Buttorff, PhD, RAND Corporation, 1200 S Hayes St, Arlington, VA 22202. E-mail: firstname.lastname@example.org. 703.413.1100 x5154.
1. Medicines use and spending in the US—a review of 2015 and outlook to 2020. IMS Health website. http://www.imshealth.com/en/thought-leadership/quintilesims-institute/reports/medicines-use-and-spending-in-the-us-a-review-of-2015-and-outlook-to-2020#form. Published April 2016. Accessed April 2017.
2. Fendrick AM, Smith DG, Chernew ME, Shah SN. A benefit-based copay for prescription drugs: patient contribution based on total benefits, not drug acquisition cost. Am J Manag Care. 2001;7(9):861-867.
3. Fendrick AM, Chernew ME. Value based insurance design: maintaining a focus on health in an era of cost containment. Am J Manag Care. 2009;15(6):338-343.
4. Shrank WH, Choudhry NK, Agnew-Blais J, et al. State generic substitution laws can lower drug outlays under Medicaid. Health Aff (Millwood). 2010;29(7):1383-1390. doi: 10.1377/hlthaff.2009.0424.
5. Hoadley JF, Merrell K, Hargrave E, Summer L. In Medicare Part D Plans, low or zero copays and other features to encourage the use of generic statins work, could save billions. Health Aff (Millwood). 2012;31(10):2266-2275. doi: 10.1377/hlthaff.2012.0019.
6. Liberman JN, Roebuck MC. Prescription drug costs and the generic dispensing ratio. J Manag Care Pharm. 2010;16(7):502-506.
7. Generic utilization in the Medicare Part D program. Office of the Inspector General, HHS website. https://oig.hhs.gov/oei/reports/oei-05-07-00130.pdf. Published November 2007. Accessed May 1, 2014.
8. Goldman DP, Joyce GF, Zheng Y. Prescription drug cost sharing: associations with medication and medical utilization and spending and health. 2007;298(1):61-69. doi: 10.1001/jama.298.1.61.
9. Gibson TB, Ozminkowski RJ, Goetzel RZ. The effects of prescription drug cost sharing: a review of the evidence. Am J Manag Care. 2005;11(11):730-740.
10. Gibson TB, McLaughlin CG, Smith DG. Generic utilization and cost-sharing for prescription drugs. Adv Health Econ Health Serv Res. 2010;22:195-219.
11. Anderson GF. Chronic care: making the case for ongoing care. Robert Wood Johnson Foundation website. http://www.rwjf.org/en/library/research/2010/01/chronic-care.html. Published February 2010. Accessed May 1, 2014.
12. Remler DK, Atherly AJ. Health status and heterogeneity of cost-sharing responsiveness: how do sick people respond to cost-sharing? Health Econ. 2003;12(4):269-280. doi: 10.1002/hec.725.
13. Wang PS, Patrick AR, Dormuth C, et al. Impact of drug cost sharing on service use and adverse clinical outcomes in elderly receiving antidepressants. J Ment Health Policy Econ. 2010;13(1):37-44.
14. Shrank WH, Cox ER, Fischer MA, Mehta J, Choudhry NK. Patients’ perceptions of generic medications. Health Aff (Millwood). 2009;28(2):546-556. doi: 10.1377/hlthaff.28.2.546.
15. Shrank WH, Liberman JN, Fischer MA, Girdish C, Brennan TA, Choudhry NK.
Physician perceptions about generic drugs. Ann Pharmacother. 2011;45(1):31-38. doi: 10.1345/aph.1P389.
16. H.R.3590—Patient Protection and Affordable Care Act. Congress.gov website. https://www.congress.gov/bill/111th-congress/house-bill/3590. Accessed May 1, 2014.
17. High-risk pools for uninsurable individuals. Kaiser Family Foundation website. http://kff.org/health-reform/issue-brief/high-risk-pools-for-uninsurable-individuals/. Published August 1, 2016. Accessed February 2017.
18. Elixhauser A, Steiner C, Palmer L. Clinical Classifications Software (CCS). Agency for Healthcare Research and Quality website. http://www.hcup-us.ahrq.gov/toolssoftware/ccs/ccs.jsp. Published 2012. Accessed May 1, 2014.
19. Medicaid eligibility for adults as of January 1, 2014. Kaiser Family Foundation website. http://kff.org/medicaid/fact-sheet/medicaid-eligibility-for-adults-as-of-january-1-2014/. Published October 1, 2013. Accessed May 1, 2014.
20. IMS Institute for Health Informatics. The use of medicines in the United States: review of 2011. Environmental Health News website. http://www.environmentalhealthnews.org/ehs/news/2013/pdf-links/IHII_Medicines_in_U.S_Report_2011-1.pdf. Published April 2012. Accessed May 1, 2014.
21. Farley JF, Wansink D, Lindquist JH, Parker JC, Maciejewski ML. Medication adherence changes following value-based insurance design. Am J Manag Care. 2012;18(5):265-274.
22. Kesselheim AS, Misono AS, Lee JL, et al. Clinical equivalence of generic and brand-name drugs used in cardiovascular disease: a systematic review and meta-analysis. JAMA. 2008;300(21):2514-2526. doi: 10.1001/jama.2008.758.
23. Clinical Antipsychotic Trials of Intervention Effectiveness (CATIE). National Institute of Mental Health website. http://www.nimh.nih.gov/funding/clinical-trials-for-researchers/practical/catie/index.shtml. Published 2014. Accessed May 27, 2014.
24. Lieberman JA, Stroup TS, McEvoy JP, et al. Effectiveness of antipsychotic drugs in patients with chronic schizophrenia. N Engl J Med. 2005;353(12):1209-1223.
25. Figueiras MJ, Marcelino D, Cortes MA. People’s views on the level of agreement of generic medicines for different illnesses. Pharm World Sci. 2008;30(5):590-594. doi: 10.1007/s11096-008-9247-y.
26. Hermes ED, Sernyak M, Rosenheck R. Impact of a program encouraging the use of generic antipsychotics. Am J Manag Care. 2012;18(8):e307-e314.
27. Ballentine NH. Polypharmacy in the elderly: maximizing benefit, minimizing harm. Crit Care Nurs Q. 2008;31(1):40-45. doi: 10.1097/01.CNQ.0000306395.86905.8b.
28. Barnett K, Mercer SW, Norbury M, Watt G, Wyke S, Guthrie B. Epidemiology of multimorbidity and implications for health care, research, and medical education: a cross-sectional study. Lancet. 2012;380(9836):37-43. doi: 10.1016/S0140-6736(12)60240-2.