Among insured patients with type 2 diabetes, copayments and glycosylated hemoglobin values were associated with increased rates of drug switching.
As healthcare expenditures continue to rise, insurers seek to slow the rate of growth through the implementation of cost-containment measures. A common cost-containment measure is the incentive formulary, which attempts to disincentiviz
e patients from choosing more expensive
brand-name drugs.1,2 Previous studies have shown
that cost-containment measures including tiered formularies
succeed in reducing overall consumption of prescription
drugs by health plan members, reducing health plan
spending on pharmaceuticals, and encouraging the choice
of less expensive over more expensive medications.3-6 Possible
effects of cost sharing include switching to lower-cost
drugs, treatment interruptions, and taking medications less
frequently than prescribed to extend the period of use for
prescription drugs.7-9 The majority of studies to date have focused
on only a few therapeutic classes of drugs, most notably
angiotensin-converting enzyme inhibitors, proton pump
inhibitors, and statins.10-12 Several studies have examined
hypertension,7,13,14 although other diagnoses have not been
Diabetes mellitus, with an estimated total prevalence of 7% and medical expenditures totaling $174 billion,15 is a serious public health problem that disproportionately affects ethnic subgroups such as non-Hispanic blacks, American Indians, Alaska Natives, Native Hawaiians, and Filipinos.16,17 In Hawaii, the setting of this study, the prevalence of diabetes among Native Hawaiians is 11.5% and 10.4% among Filipinos.16 Although previous studies have examined the relationship between cost sharing and medication use among diabetic patients,5,18 to our knowledge, no studies to date have looked at the interaction between drug switching and cost sharing, an issue with potential implications for diabetes management.
To address this gap, we examined the relationship between copayment level and drug switching among patients who newly initiated a medication regime for the treatment of type 2 diabetes. The conceptual model guiding this investigation was
the economic framework for healthcare demand proposed
by Gibson et al.1 According to this framework, raising the
cost-sharing level for prescription drugs is expected to result
in (1) reduced consumption of drugs overall; (2) substitution
of less expensive, equivalent drugs; and (3) a decrease
in consumption of “nonessential” drugs (eg, discretionary
drugs like antihistamines).1 We hypothesized that higher copayments
would increase the rate of switches to drugs with
lower copayments, and also that copayment levels might
have a differential effect on drug switching depending on
Study Design and Population
The design was a retrospective observational study. The study population was drawn from members of a single health plan in the state of Hawaii that provides coverage to approximately 650,000 members, or about 50% of the state’s total insured population.14 Inclusion in the study required patients to have (1) a diagnosis of type 2 diabetes and (2) a first filled prescription for a diabetes-related medication after at least 1 full year of prior enrollment. Patients were excluded if (1) they had type 1 diabetes or (2) they filled 2 different prescription diabetes medications on their first or index date. A small number of patients (n = 779) began on polytherapy and were excluded from the final study sample. Data were extracted for eligible members during the study period from January 2002 through August 2007. No universal changes in copayments occurred for this insured population during the study period. However, the study population was limited to patients who newly initiated pharmacologic treatment of their diabetes, in order to isolate “new” diabetes patients using antidiabetic medications from chronic antidiabetic medication users.
As previous studies have documented a lack of communication between patients and providers about the out-of-pocket costs for medications, we did not assume that discussions about cost occurred prior to the time the first prescription was filled.19,20 In addition, other studies have shown that clinicians often are unaware of the cost of different medications and therefore do not take cost into account when prescribing.21 A patient’s data were included from the date of the first prescription until the end of the study period or the end of continuous enrollment, whichever was sooner. This study was approved by the University of Hawaii Committee on Human Studies.
The medical claims database provided a limited data set on diagnoses, patient utilization, type of insurance plan (fee-for-service or HMO plan), demographic information, and morbidity levels (Johns Hopkins Adjusted Clinical Groups Case-Mix System; see www.acg.jhsph.edu) for the 3-month period preceding the index prescription.22 The pharmacy claims database provided data on medications including prescription fill dates, days of supply, copayment (out-of-pocket cost to the patient), and formulary tier. Medications were categorized by therapeutic class (all classes of oral hypoglycemics as well as insulins) within formulary tier (generic, preferred brand, and nonpreferred brand). Glycosylated hemoglobin (A1C) levels were obtained from laboratory reports from the various diagnostic laboratories. The laboratory value that most closely followed the index prescription was included in analyses. Ethnicity and education were obtained from member satisfaction surveys that are mailed annually to a random sample of the insurer’s members with response rates of 40% to 50%.
Patients were considered to have switched drugs if they discontinued use of the starting drug and filled a prescription either (1) for a drug in a different diabetes therapeutic class or (2) for a drug in the same therapeutic class but in a different tier. Discontinuation was defined as having a gap between prescriptions greater than twice the length of the first prescription. Similarly, the study window to switch to a second drug was twice the length of the first prescription. Patients who discontinued therapy completely (ie, stopped filling their current prescription drug without adding another drug) were treated as censored from the end date of their last prescription. Gaps were calculated as the difference between the end date of the first prescription (based on the days of supply) and the fill date for the second prescription.Similar definitions of drug switching have been used inprevious studies.7,23
This study focused on switching patterns related to the patient’s initial drug. If patients subsequently added drugs of a different diabetes therapeutic class while continuing use of the starting drug, the additional drugs were not examined for switches. Drugs added after the patient discontinued use of their initial drug, however, were followed for switches. Multiple drug switches were included in the analyses for such patients.
Copayments were continuous and were standardized in units of $5 per 30-day period. Glycosylated hemoglobin values were reported as a continuous variable in units of percent hemoglobin that was glycosylated, as a dichotomy between A1C >7% and A1C <7%, or in more refined categories with unit increases from 6% to 10%. Switches to more expensive drugs or drugs of equal or lesser cost were based on the difference between the out-of-pocket copayment for the initial drug and the out-of-pocket copayment of the drug the patient switched to. In other words, they were based on the relevant out-of-pocket costs to the patient.
Data were analyzed using multistate proportional hazards models.24,25 The multistate model extends thetraditional proportional hazard model by allowing transitionsfrom multiple starting to multiple ending states.“States” were defined as periods of continuous druguse as specified above. Drug switches were modeled aschanges in states. Copayment amounts and prescriptiondrug choice were reset in the data set at the date of eachdrug switch. The effect of covariates on a transition suchas from state i to state j assumed a proportional hazardsmodel on the transition hazards. The transition hazard λij(t;z) for transition from state i to state j is given byλij (t;z) = λij,0t exp(β1z1 + … + βpzp),where λij,0t is the baseline hazard for an individual withcovariates of 0 for the transition from state i to state j, andz is a vector of covariates (ie, z1,…zn).
Interactions between A1C levels and copayment were modeled using A1C dichotomized between >7% and <7%, and continuous copayment amounts. Interactions between certain ethnicities and copayment also were examined and are illustrated by a graph of hazard ratios (HRs) by copayment amounts. All analyses were completed using SAS version 9.1 (SAS Institute Inc, Cary, NC).
Demographic and Clinical Characteristics
A total of 9260 adults with diabetes were included in this study; analyses by ethnic subgroup included the 3537 members with ethnicity data available. Nearly 60% of the patients were between 45 and 64 years of age and 53% were male (
). Of participants with known ethnicity, the majority (40%) self-identified as Japanese, with Filipinos (18%) and Native Hawaiian or Pacific Islanders (18%) comprising the next largest ethnic subgroups. Although exact comparisons are difficult, this ethnic breakdown is roughly consistent (both in rank and percentage) with the general population of diabetic patients in the state of Hawaii, with the exception of a higher proportion of patients with self-reported Japanese ethnicity.26 However, it is unclear whether this discrepancy represents a higher response rate from this ethnic group or a greater proportion of Japanese enrollees within this insured population. Most Caucasians (47%) and Chinese (42%) reported having at least a college degree, while the majority of Filipino (46%), Native Hawaiian or Pacific Islanders (49%), and those choosing other ethnicities (51%) reported having a high school diploma or less (data not shown). Educational attainment was approximately evenly distributed for Japanese, with 31% reporting a high school diploma or less, 35% reporting some college, and 35% reporting at least a college degree. Almost 80% of patients had fee-for-service insurance.
Morbidity among study participants was nearly equally distributed between low, medium, and high levels, with slightly fewer patients having high morbidity levels (28%) and slightly more having medium levels (38%). More than half (52%) of all diabetic patients were at target A1C levels of less than 7%. Approximately 90% of all copayments fell within the range of $5 to $35. The most common
copayment amount for either the first (49%) or second
(46%) filled antidiabetic prescription was $5. More than
15% of antidiabetic medications prescribed (either first
or second filled prescription) had out-of-pocket costs at
the highest copayment tier of more than $15. Participants
initially filled prescriptions for 1 of 14 drugs. Therapeutic
classes of drugs utilized by participants during the study
period were as follows: sulfonylureas, sulfonylureas/biguanides,
biguanides, thiazolidinediones, thiazolidinediones/
biguanides, thiazolidinedione/sulfonylureas, incretin
mimetics, dipeptidyl peptidase-4 inhibitors, dipeptidyl
peptidase-4 inhibitor/biguanides, meglitinides, glucosidase
inhibitors, and insulins. A total of 1376 patients
switched to different drugs (13%), and 4408 (48%) discontinued
therapy completely by the end of the study period.
Copayments and A1C
Compared with the lowest copayment level (<$5) for the initial antidiabetic medication, higher copayment levels were associated with increased rates of drug switching, with an almost 2-fold increased rate seen for copayments of more than $15 (HR = 1.95, 95% confidence interval [CI] = 1.50, 2.53) (
). Similarly, increasing A1C levels were monotonically associated with increased rates of switching drugs. Compared with patients who had A1C values of less than 6%, patients with A1C values of 10% or greater had a more than 2-fold increased rate of switching drugs (Table 2).
shows the direction of switches to drugs with higher copayments or copayments of equal or lesser cost relative to the cost of the starting drug (model 1) and stratified by A1C values (model 2). For each $5 copayment increase over a standard 30-day period, study participants were significantly less likely to switch to a more expensive drug (HR = 0.49, 95% CI = 0.43, 0.56) and significantly more likely to switch to a drug of equal or lesser cost (HR = 1.04, 95% CI = 1.03, 1.05) (Table 3). The relationship between copayment and drug-switching rate also was examined within subgroups by levels of glycemic control. Patients with optimal glycemic control (A1C <7%) were more likely to switch to a drug of equal or lesser cost (HR = 1.10, 95% CI = 1.07, 1.13). A similar relationship on a reduced scale (HR = 1.03, 95% CI = 1.02, 1.04) also was found for patients with suboptimal glycemic control (A1C >7%). Comparisons of the HRs for patients with optimal versus suboptimal glycemic control revealed that switches to drugs of equal or lesser cost differed significantly (P <.001), suggesting that patients with A1C >7% were less affected by increases in copayment than patients with A1C <7%. The results from the above models were not significantly altered when patients on insulin were excluded from the data set.
Ethnicity and Education
In analyses by ethnic subgroup, Japanese, Caucasian, Chinese, Filipino, and Native Hawaiian/Pacific Islander populations did not differ significantly in their rates of drug switching (P >.05), and rates of drug switching were not significantly influenced by educational attainment (
). However, compared with Japanese (the largest ethnic group among study participants) as the referent group, significant interactions were detected between copayment level and indicators of Chinese (P = .04), Filipino (P <.0001), and Native Hawaiian or Pacific Islander (P = .01) ethnicities. The modifying effects were in different directions. Filipino and Native Hawaiian or Pacific Islander patients had the greatest difference in rates of drug switching compared with Japanese patients at low copayments (
). Interestingly, for Chinese patients, the pattern was reversed and Chinese participants had the greatest rates of switching at higher copayment amounts.
Understanding the health impact of drug switching and cost sharing is a topic of ongoing interest among patients, providers, and healthcare insurers, particularly within the context of escalating healthcare and pharmaceutical costs. In this study, higher copayments were associated with increased rates of drug switching even after adjusting for other covariates. As expected, participants were more likely to switch to a drug that was less expensive than their first prescription. Our findings are consistent with other studies that have found that increases in cost-sharing are associated with decreased use of higher-cost diabetes medications.27-30
Previous studies also have found that patients may exhibit different patterns of drug switching depending on whether their medications fall into the “essential” versus “discretionary” therapeutic class.18,29,31 In this study, we examined only patients who were taking antidiabetics, an essential class of medications, and found that the associations between copayment amounts and drug-switching patterns were associated with A1C levels and patient ethnicity. Patients with higher A1C levels were significantly more likely to switch drugs than patients with lower A1C levels. However, this relationship may be influenced by other unmeasured variables such as healthcare providers who may switch patients’ therapies to improve glycemic control rather than because of cost alone (ie, copayments).
When the relationship was examined between cost sharing and switches to drugs with more or less expensive copayments (relative to the cost of the starting drug), significant associations were found between copayment level and both decreased rates of switching to more expensive drugs and increased rates of switching to drugs of equal or lesser cost. Although the increased rate of switching to drugs of equal or lesser cost was modest (HR = 1.04, 95% CI = 1.03, 1.05), the magnitude of this effect must be evaluated within the larger context of the number of patients enrolled in various health systems nationally rather than solely within the limited scope of an individual insurer.
Another finding of this study was the varied effect of copayment level on drug switches between patients with high versus low A1C values. For each $5 difference in out-of-pocket costs, patients with better glycemic control (A1C <7%) were 7% more likely to switch to drugs with equal or lesser copayments in the next 30-day period than patients with poorer glycemic control (A1C >7%). This effect was magnified as copayment amounts increased. Patients with a $20 higher copayment were 28% more likely to switch to a drug of equal or lesser cost. These results may reflect differences in the willingness of patients with less controlled disease to disrupt their existing therapy by switching to a new drug, even within an essential class of medications. Conversely, the findings also may suggest that patients with well-controlled diabetes are considering factors such as long-term costs and exploring less expensive alternatives.
Multiple clinical trials have established the importance of optimal glycemic control in preventing microvascular complications32; thus, the clinical implications of these findings are relevant to the long-term management of type 2 diabetes. In addition, the variation in drug-switching patterns by levels of glycemic control suggests that other studies might have been strengthened by examining similar markers of disease severity, for example, blood pressure in studies of antihypertensives.
Patterns of medication use are significantly affected by a number of different factors at multiple different levels including the individual, physician, and health plan. A 2004 study found that patient attitudes and physician prescribing patterns may significantly influence the decision to change medications.33 Ethnic subgroups also may respond differently to cost pressures.6,34 For example, Steinman et al reported that nonwhite Americans were more than 3 times as likely to have taken less medication than prescribed because of cost (relative risk = 3.4, 95% CI = 2.4, 4.7), even after controlling for age, sex, income, education, marital status, out-of-pocket prescription drug costs, and 3 measures of health status.34 Our study found that copayment level had a differential effect on Chinese, Filipino, and Native Hawaiian or Pacific Islander patients compared with Japanese patients. Given the increasing body of literature on the role of ethnicity in health and disease,35 the results of this study suggest that further analyses of the influence of these important factors on drug switching should be considered.
Finally, regarding methodology, previous studies examined drug switches either in terms of differences in the proportion of participants who were on a particular drug in a pre/post comparison,11,27,29 or in terms of the length of time that patients remained on their initial drug in survival analyses.7,36 This study used a more powerful multistate methodology that modeled multiple drug switches over time. Thus, our findings extend the existing literature by providing confirmatory evidence of trends over time for multiple successive drug switches rather than restricting the event of interest to the first drug switch only.
This study has a number of limitations. Using data from a single insurer limits the generalizability of the findings. In addition, although the study included A1C as a measure of disease severity, only initial A1C values were included; thus, the effects of changing A1C levels over time were not addressed. In addition, the data did not include the dates of diagnosis of diabetes and consequently did not allow a consideration of differences in the length of time between the diagnosis and the first filled prescription. Data on income levels were not available in this administrative data set, although models with ethnicity were adjusted for education as a proxy measure for income. We also chose to exclude patients who did not initiate therapy, although the choice not to initiate pharmacologic treatment also is an important aspect of patient behavior and may be influenced by issues related to cost sharing. Finally, data were not available on the clinical appropriateness of drug switching. Thus, it is unclear whether drug switches were positive or negative events, an issue that might be addressed in the future.
Strengths of this study include the availability of data from approximately half of the insured population in a state that has a stable, ethnically diverse patient population with little out-migration. In examining only new initiators of medication regimes for the treatment of diabetes, the study design avoided mixing new and
continuing users, groups that may have different responses
to cost sharing. The use of multistate modeling
allowed the separate investigation of drug switches to
more expensive drugs or to drugs of equal or lesser
cost. In addition, the analyses considered the influence
of A1C levels, ethnicity, education, and interactions between
these variables and copayment amount on the
rates of drug switching. Previous studies of drug switching
have seldom examined the importance of sociodemographic
factors other than age and sex.
This study provides confirmatory evidence that copayments are associated with changes in patterns of drug switching. It also extends the existing literature by examining drug switching within a previously understudied disease, diabetes, and among understudied populations of Native Hawaiians and Asian/Pacific Islanders. Although an understanding of the direct clinical implications of this study requires further exploration of the consequences of drug switches, clinicians should be aware that prescribing more costly medications may influence patients’ desire to switch drugs. Thus, accurate, accessible prescribing guides and other related materials are needed to aid physicians in making these decisions. Future research should consider clinical and sociodemographic factors including measures of disease severity over time, ethnicity, and income, as well as the clinical implications of drug switches such as effects on medication adherence and glycemic control.