Sharon Glave Frazee, PhD, MPH

Chronic myeloid leukemia, also called chronic myelogenous leukemia (CML), is a slow-growing cancer of the white blood (myeloid) cells in bone marrow and blood. It is estimated that there will be 5920 new cases of CML and 610 deaths due to CML in the United States in 2013. The incidence of CML increases with age, with half of all new cases of CML diagnosed in individuals 65 years or older.

 As record numbers of American baby boomers (born between 1946 and 1964) reach their mid-60s, the number of new patients diagnosed with CML is expected to increase.

Untreated, CML takes an average of 4 years to become more aggressive and reach the blastic stage, when it is usually fatal. However, the prognosis and survival of CML patients have improved dramatically since imatinib, an orally administered chemotherapeutic agent, was approved by the US Food and Drug Administration in 2001. Five-year survival rates for newly diagnosed patients with CML increased from 40% between 1996 and 2000 to more than 55% between 2001 and 2007.

Long-term studies have shown dramatically decreased mortality and improved survival among CML patients taking imatinib.

 Consequently, CML has effectively been transformed into a chronic disease with patients using imatinib and other CML medications over extended periods of time.

 Despite imatinib’s life-extending benefits, many patients struggle to remain adherent to it. A large body of scientific evidence has demonstrated that optimum outcomes from drug therapy are reliant on adherence.

Although previously published studies have defi ned optimum adherence to imatinib as 90% or above, actual average adherence among US commercially insured CML patients is reported at about 80%,

 with only 54% of patients taking imatinib achieving adherence of 90% or higher.

CML patients face many challenges influencing their medication adherence, including coordination with physicians and other healthcare providers, managing complex treatment regimens, medication side effects, and healthcare costs. Common side effects of imatinib include diarrhea, muscle cramps, nausea, skin reactions, swelling, and vomiting. More serious adverse effects such as myelosuppression (decreased ability of bone marrow to produce blood cells) and elevated liver transaminase levels (indicative of liver damage) may require treatment interruptions.

 Additionally, with the annual cost of therapy reported to be $92,000 in 2012, the economic burden of imatinib is substantial.

 Couple the clinical challenges with the out-of-pocket drug cost facing cancer patients, and it is not surprising that medication adherence is lower than desired, posing a serious concern to patients, clinicians, and payers.

The need to improve medication adherence among patients using oral oncology medications such as imatinib is widely recognized.

 According to a National Comprehensive Cancer Center Task Force, oncology-specific support from specialty pharmacies improves adherence, encourages communication between patients and pharmacists, identifies potential safety concerns, helps prevent unwarranted drug expenditures, and ensures appropriate use of medications.

 Given the suboptimal adherence rates to imatinib therapy, patients given this medication, especially those who are new to imatinib therapy,

 are ideal candidates for therapy-related services commonly provided by specialty pharmacies such as refill reminder programs and oncology care management.

Our hypothesis was that adherence to imatinib would be higher among patients starting a new course of therapy with a specialty pharmacy instead of other dispensing channels. Our study objective was to compare imatinib adherence between patients using specialty pharmacy and those using other dispensing channels.

A claims-based, retrospective study was conducted using de-identified prescription data collected by a large national pharmacy benefi ts manager (PBM) with an in-house specialty pharmacy. Patients receiving imatinib through the specialty pharmacy had access to supportive services designed to enhance clinical outcomes, increase drug safety, manage side effects, and help patients stay on therapy. Nurses and pharmacists specifically trained in oncology provided education and care management through proactive patient outreach throughout the course of therapy at clinically meaningful intervals. Telephonic prescription refill reminders were timed to be made when the patient’s quantity of imatinib reached a specifi ed level prior to depletion, to ensure an uninterrupted supply of medication.

Pharmacy claims data for the period July 1, 2010, to August 31, 2012, were analyzed. Irrespective of the type of pharmacy or dispensing channel used, the claims data set included all patients for whom the PBM processed claims for oncology drugs. Thus, patients in the study could obtain imatinib from any combination of the inhouse specialty pharmacy, other specialty pharmacies, retail pharmacies, or home delivery pharmacies. Patients who obtained a simple majority of their 30-day adjusted imatinib prescriptions from the PBM’s in-house specialty pharmacy were assigned to the specialty pharmacy group. Patients receiving a majority of their imatinib supply from other channels were assigned to the other pharmacy group. The design of this study was not submitted to an institutional review board, as only de-identified administrative data were used. All regulations related to the Health Insurance Portability and Accountability Act were followed.

Patient records selected for the study showed at least 1 prescription claim for imatinib (Medi-Span Generic Product Identifier code 21-53-40-35-10) during the index period between January 1, 2011, and August 31, 2011. Analysis was limited to patients new to therapy, defined as patients with no imatinib claims in the 6 months prior to the index imatinib claim. To accurately assess newto-therapy status and postinitiation adherence, study patients were required to be continuously eligible for pharmacy benefits 6 months prior to and 12 months following the index imatinib claim. Medicare and Medicaid beneficiaries were excluded from the analysis since these patients’ pharmacy benefit design is different from that of commercial patients. Patients who paid 100% of the out-of-pocket cost for imatinib were also excluded from the analysis because it was likely that imatinib was not covered by the pharmacy benefi t for these patients. We may not have captured all transactions for patients who paid 100% of their medication cost. These patients may not have used the pharmacy benefit and hence generated no claims. Each study patient was followed for 365 days after the index imatinib claim.

The outcome of interest was optimum imatinib adherence, a dichotomous variable defi ned as a proportion of days covered (PDC) of 90% or higher during follow-up. The PDC was calculated as the number of days the patient had medication on hand during the 365-day followup period divided by 365.

When the days of supply from a new fill overlapped the last 7 days of supply of imatinib from a previous fill, the fill date for the new fill was adjusted forward to the depletion date of the earlier fill. The patients were thus credited for finishing the supply of imatinib from the previous fill before using the supply of imatinib from the subsequent fill. In situations where multiple claims for imatinib representing different drug strengths (100 mg and 400 mg) occurred concurrently on the same fill date, it was assumed that medication from both claims would be used concurrently and no adjustment was made. The PDC was also evaluated as a continuous variable.

A literature review identifi ed adherence channel confounders, which were included in multivariate analysis. Demographic confounders included patient sex and age at the time of the index imatinib claim.

 Because urbanicity may impact access to pharmacies and therefore  adherence,

 we created a dichotomous indicator for whether the 5-digit zip code of the patient’s residence was inside the urban center of a US Census—defined Core Based Statistical Area.

Patient out-of-pocket cost burden was determined by averaging the amounts paid by the patient (copayment, coinsurance, and deductible) per 30-day supply of imatinib during the follow-up period.

 was the number of therapy classes (as established by the number of unique 2-digit Medi-Span Generic Product Identifi er codes) used by the patient during the follow-up period.

 Because the use of more days of supply per prescription may also improve adherence,

 the average number of days of supply per unadjusted imatinib prescription was included as a covariate. Finally, an indicator was created to identify patients who had at least 1 imatinib claim paid by a plan sponsor whose pharmacy benefit design included a days-of-supply optimization program for specialty medications including imatinib.

As part of this program, patients starting a new course of imatinib therapy were provided medication in 30-day increments for the initial 90 days of therapy, after which patients transitioned from 30-day prescriptions to 90-day prescriptions of imatinib for the remainder of follow-up. Although the program was intended to maximize appropriate use, it may have affected the imatinib PDC in 2 ways. First, patients who completed the first 90 days of therapy would have a higher PDC based on having more days of supply per prescription. Second, those patients who discontinued therapy before completing the first 90 days of therapy would have a lower PDC based on having fewer days of supply per prescription.

Differences in age and imatinib PDC between the specialty pharmacy group and other pharmacy group were evaluated using 

 tests. Differences in out-of-pocket pharmacy spending per 30-day adjusted imatinib claim, medication burden, and days of supply per unadjusted imatinib prescription were evaluated using nonparametric Wilcoxon rank sum tests because the distributions of these variables were highly skewed. Differences in categorical variables were evaluated with χ

 tests. Logistic regression models were created with dichotomous imatinib adherence as the primary dependent variable, dispensing channel as the primary independent variable, and other covariates. Multivariate adjustment for continuous imatinib PDC was performed using ordinary least squares regression with PDC as the dependent variable, dispensing channel as the primary independent variable, and other known confounders as covariates.

The Figure outlines the patient selection procedure. A total 6121 patients had at least 1 prescription claim for imatinib between January 1, 2011, and August 31, 2011. Excluded were 1550 patients who were not continuously eligible for pharmacy benefi ts 6 months before and 12 months after the index imatinib claim, 3556 patients who had a claim for imatinib in the 6 months prior to the index claim, 2 patients with inconsistencies in their eligibility information, 292 whose pharmacy costs were paid by Medicare Part D, 3 who were Medicaid beneficiaries, and 14 who paid the entire cost of imatinib out of pocket. The fi nal study sample consisted of 704 patients, of which 433 were in the specialty pharmacy group and 271 in the other pharmacy group.

Selected characteristics of the study patients are presented in 

. Patients in the 2 study groups had similar distributions of age, sex, and urbanicity. Compared with the other pharmacy group patients, patients in the specialty pharmacy group had lower out-of-pocket imatinib costs and a lower medication burden. Patients in the specialty pharmacy group had on average about 15 additional days of imatinib per prescription. More patients in the specialty pharmacy group had at least 1 fill of imatinib through a benefit that encouraged 90-day fills only for patients stabilized on imatinib therapy.

During the follow-up period, patients in the specialty pharmacy group had greater odds of being adherent (a PDC of at least 90%) (unadjusted odds ratio [OR] = 1.87, 95% confi dence interval [CI], 1.36-2.58) compared with patients in the other pharmacy group. After adjusting for confounders, patients in the specialty pharmacy group were signifi cantly more likely to achieve optimum imatinib adherence (adjusted OR = 1.46, 95% CI, 1.02-2.09; 

). When we examined the second outcome, continuous PDC, we found that the unadjusted average PDC for imatinib was greater for the specialty pharmacy group than for patients in the other pharmacy group (75.78% vs 60.37%; 

 <.0001). After multivariate adjustment using ordinary least squares, controlling for known confounders, and setting all model coeffi cients to their mean values, the adjusted PDC for specialty pharmacy patients was 74.14% versus 63.34% in the other pharmacy group (

Our study found that patients who used a specialty pharmacy that provided refill reminders and comprehensive clinical care management programs to fill a majority of their imatinib prescriptions had a significantly higher likelihood of attaining optimum imatinib adherence compared with those using other dispensing channels for a majority of their 30-day adjusted imatinib prescriptions.

Adherence to imatinib has been associated with lower healthcare costs. A study of US commercially insured CML patients starting imatinib therapy found that those with medication possession ratios (MPRs) of 85% and above incurred $17,727 less in total healthcare costs over a 12-month follow-up period than patients with lower MPRs.

Another study of CML patients found that a 10% using a retail pharmacy. In the same study, specialty pharmacy patients had 13% lower total healthcare costs (medical and pharmacy) compared with retail pharmacy patients over a 12-month follow-up period.

Goals proposed by the National Comprehensive Cancer Network Task Force on Specialty Pharmacy include maximizing adherence, optimizing clinical outcomes, and improving economic outcomes.

 The specialty pharmacy investigated in this study offered patients a  number of interventions and programs to attain these goals. From the initiation of imatinib therapy, patients using the specialty pharmacy had access to oncology-trained nurses and pharmacists for disease and drug education and support. Planned, proactive interactions also allowed for detection of potential side effects, medication errors, and premature discontinuation of therapy. In addition, adherence monitoring and refi ll reminders affected the imatinib adherence of specialty pharmacy patients.

There are several limitations to this study, a number of which result from the exclusive use of pharmacy claims data for this analysis. Because integrated medical claims and patient chart data were unavailable, patients could not be selected or stratifi ed based on medical diagnosis or cancer stage. Disease severity could not be controlled for in the multivariate adjustment. The PDC calculation did not adjust for time spent in the inpatient setting among those patients who were admitted. A proxy to control for the impact of medication burden on adherence was created, but this proxy did not completely capture comorbidity burden. A possession-based measure of adherence was used as the study outcome, assuming that any imatinib possessed by the patient was taken as prescribed. Similar measures have been used in earlier studies.

 A 6-month preindex period was used to select imatinib patients starting a new course of therapy. It is possible that all study patients may not have been imatinib naïve at the time of the index imatinib claim in this study, which may have had an infl uence on imatinib adherence.

The findings from this study are generalizable only to commercially insured patients with PBM-administered pharmacy benefits. Additionally, some patients in the other pharmacy group may have obtained the majority of their medications from another specialty pharmacy. However, it has been reported that specialty pharmacies have a wide range of pharmacy delivery models and offer a variety of oncology care management interventions.

 Because the specific care model of each external specialty pharmacy or retail pharmacy could not be determined, all patients using one of these channels for a majority of their imatinib prescriptions were combined into a single comparator group. Patients in the other pharmacy group may have been provided with intensive and comprehensive oncology care through their pharmacy; thus, this analysis represents a conservative estimate of the impact of specialty pharmacy on imatinib adherence. Finally, this study may be biased because of the “healthy adherer” effect. Patients who generally espouse healthy behaviors may select specialty pharmacy over other channels to take advantage of the more intensive and disease-focused approach of this channel and may also be more motivated to be adherent to medications. It is possible that part of the superior adherence observed in specialty pharmacy may be explained by a greater proportion of healthy adherers using specialty pharmacy, rather than channel characteristics.

This study found that commercially insured patients using specialty pharmacy integrated with refi ll reminders and a comprehensive oncology care management program were 46% more likely to achieve optimum adherence to imatinib compared with patients who did not use the specialty pharmacy. With an increasing number of oral medications for cancer entering the marketplace,

patient adherence is an increasingly important determinant of treatment success. Plan sponsors, payers, and patients may benefit from a pharmacy model used by the specialty pharmacy evaluated in this study for the dispensing and use of oral oncology medications.

Articles

The authors examined trends in the utilization of prescription opioid medication following introduction of abuse deterrent technology for Oxycontin.

Specialty Pharmacy Improves Adherence to Imatinib

Hepatitis C virus (HCV) infection affects between 1.3% and 1.9% of Americans and an estimated 3.2 million suffer from chronic HCV infection.

 Hepatitis C virus is a common cause of cirrhosis and hepatocellular carcinoma as well as the most common cause for liver transplantation in the United States.

The economic burden of HCV infection is considerable. Recent studies have found the presence of HCV was associated with significantly more hospital admissions, emergency department visits, and liver transplantations.

 A 2009 study found that patients with HCV infection had an average incremental annual cost of $9681 compared with patients who had no diagnosis of HCV infection. The average incremental cost per patient per year increased to $27,845 for HCV patients with decompensated cirrhosis, $43,671 for HCV patients with hepatocellular carcinoma, and $93,609 for HCV patients with a liver transplant.

 In addition, HCV infection has been shown to result in significant indirect costs in terms of increased presenteeism, absenteeism, and work impairment.

Compared with interferon monotherapy, combination therapy with interferon and ribavirin in hepatitis C patients has been shown to signifi cantly improve sustained viral response (SVR). Such improvement has been established in multiple subgroups of hepatitis C patients including treatment- naïve patients, relapsers, nonresponders, and patients with cirrhosis.

 Adherence to this regimen has been shown to be critical for treatment success. A study of patients infected with genotype 1 HCV showed that the patients who could be maintained on more than 80% of their interferon and ribavirin dosage for the duration of treatment had enhanced sustained response rates.

 In May 2011, 2 new novel protease inhibitors, telaprevir and boceprevir, entered the market. Recent guidelines have recommended the use of novel protease inhibitors in combination with peginterferon and ribavirin for HCV genotype 1 patients, but not for HCV genotype 2 or 3 patients.

Whether on the 3-drug regimen or the 2-drug regimen, which remains applicable treatment for genotype 2 or 3, patient medication adherence remains a challenge.

Hepatitis C patients face many challenges in managing their treatment, including complex dosing regimens, coordination with physicians’ offices for injections, and training to use self-injectable medications. In addition, patients often have to cope with unpleasant side effects for extended periods of time. While retail and other pharmacies continue to dispense these medications, specialty pharmacies that employ teams of trained healthcare professionals including pharmacists and nurses who can support the patient and coordinate care with prescribers are beginning to take the lead in dispensing these medications.

 Services offered by specialty pharmacy are aimed toward patient assistance and support; they help patients to better navigate hepatitis C regimens, manage adverse events, liaison with prescribers to better coordinate care, and encourage adherence to therapy. The objective of this study was to compare hepatitis C regimen adherence between patients enrolled in a specialty pharmacy that had integrated hepatitis C care management and patients using retail pharmacy.

A claims-based, retrospective study design was used to evaluate hepatitis C drug regimen adherence between specialty pharmacy and retail patients.

Pharmacy claims and eligibility data were extracted from a large, national pharmacy benefi t management (PBM) company database. The PBM also operates an independent specialty pharmacy that serves patients across the entire country. The claims database contains claims for patients from all over the United States for whom the PBM serves as the network claims processor. Thus, patients in the study could obtain their HCV medications from the PBM-operated specialty pharmacy, external specialty pharmacies, or retail pharmacies. Claims for the period from 2007 to 2009 were used in this analysis. Although more recent claims data from 2010 were available, large numbers of patients delaying treatment in 2010 were reported in anticipation of the approval of novel protease inhibitors.

The channel used to dispense each claim was identified by the source of origin of the claim that was populated by the PBM at the time of claim adjudication. For the purposes of this analysis, nonspecialty pharmacies were classifi ed as retail or mail-order. Specialty pharmacies were classifi ed into the PBM’s in-house specialty pharmacy and other specialty pharmacies. 

Patients were included in the study if they started a new 2-drug hepatitis C regimen (ribavirin plus interferon), defined as having index claims for ribavirin and interferon within 14 days of each other between July 1 and December 31, 2008. A new start on the regimen was defined as an index claim for either drug after a gap of at least 6 months since any previous claim for ribavirin or interferon. Hepatitis C medications were selected based on Medi-Span Generic Product Identifi er (GPI) codes. (For a list of codes and medications selected, see the 

.) A maximum gap of 14 days between interferon and ribavirin index claims was selected to identify patients on concurrent interferon and ribavirin therapy, after sensitivity analysis revealed that 95% of patients started interferon and ribavirin within 11 days of each other. This ensured that the index claims were reasonable markers for the beginning of a new regimen.

Patients with any observed claims for antiretrovirals (GPI code 12-10-xx) or hepatitis B (GPI code 12-35-20-xx) were excluded because they were at higher risk for treatment discontinuation.

 Finally, patients using multiple dispensing channels to obtain their medication were excluded. This was done to capture all channel-related services that could influence medication adherence conducted at each stage ofthe long and complex hepatitis C regimen.

For each patient, the length of therapy for each drug (ribavirin and interferon) was calculated separately and defined as the days between the first fill date in the index period and the last depletion date for that medication.

To avoid the possibility of capturing claims from multiple, successive hepatitis C regimens while still allowing for delayed filling of prescriptions in the regimen under study, patients whose total length of therapy for either HCV drug exceeded 365 days (52 weeks) after the index claim were excluded from the analysis. The observed length of regimen was defined as the average of the length of therapy for interferon and ribavirin. All patients included in the study were continuously eligible for pharmacy benefi ts from 6 months before index claim to end of follow-up period. Based on patient genotype and response-guided dosing protocols, the recommended length for an interferon plus ribavirin regimen for hepatitis C is either 24 or 48 weeks.

 In the absence of genotype or rapid viral response information, assumptions were made about the expected length of the regimen based on the observed length of the regimen. It was assumed that patients whose regimen was observed to be more than 36 weeks long (ie, closer to 48 weeks than 24 weeks) had an expected regimen length of 48 weeks. Alternately, if the observed length of the regimen was equal to or less than 36 weeks (ie, closer to 24 weeks than 48 weeks), it was assumed that the expected length of the regimen for these patients was 24 weeks. Thus, a patient whose observed regimen was 28 weeks long would be expected to be on a 24-week regimen, but a patient whose observed regimen was 40 weeks long was expected to be on a 48-week regimen.

Regimen adherence was defi ned as the proportion of days covered by both ribavirin and interferon divided by the expected length of the regimen period (24 or 48 weeks). Based on evidence that 80% or better adherence is associated with significantly improved rates of SVR, regimen adherence was dichotomized, with 80% or better adherence considered optimum adherence and less than 80% adherence considered suboptimal. Gap days were defined as the number of days between the earliest fill date and latest depletion date for the regimen for which patients did not have an available supply of both ribavirin and interferon.

Unadjusted differences between the retail and specialty pharmacy were evaluated using 

 tests for categorical variables. The alpha was set a priori to .05. Multivariate models were created to estimate multivariate adjusted regimen adherence in each of the comparison groups. To account for the variable expected length of the hepatitis C regimen, a weighted Poisson regression model was created to estimate multivariate adjusted adherence.

 The number of days covered by both ribavirin and interferon was the dependent variable, and pharmacy type (ie, specialty vs retail) was the primary predictor along with other covariates. The expected length of therapy was used as the weighting variable (ie, offset). Least squares mean regimen adherence, holding all other variables at their mean values, was compared between patients using retail and specialty pharmacy for hepatitis C medication. A Poisson regression model was used to estimate multivariate adjusted differences in gap days, with the number of gap days as the dependent variable, pharmacy type (specialty vs retail) as the primary predictor, and other covariates. Finally, a logistic regression model was created with optimal adherence (defi ned as regimen adherence 

80%) as the dichotomous dependent variable, pharmacy type as the primary predictor, and other covariates. Statistical analyses were performed using SAS version 9.2 (SAS Institute Inc, Cary, North Carolina).

Variables that are known covariates of the relationship between adherence and channel were identifi ed on the basis of a literature review, and covariates for these factors or their proxies were generated from the pharmacy claims data and included in multivariate analysis. Demographic confounders included patient age at the time of the first index claim and patient sex. Because urbanicity may impact access to pharmacies as well as adherence, a dichotomous indicator for whether the 3-digit zip code of the patient’s residence was at least in part inside a metropolitan statistical area was created as a proxy for the whether the patient resided in an urban setting.

 Patient out-of-pocket payment was defi ned as the average of the patient-paid amount (sum of copayment, coinsurance, and deductible) per 30 regimen days for hepatitis C medications during follow-up. Patient medication burden was proxied by the number of medications filled by the patient during the 6-month baseline period as defi ned by unique 10-digit GPI codes.

 Patients suffering from depression have great difficulty adhering to any medication regimen.

 We used evidence of treatment for depression as a proxy for depression. An indicator variable was created for the presence of any pharmacy claim for selective serotonin reuptake inhibitors and serotonin—norepinephrine reuptake inhibitors as identified by claims for medications with GPI codes starting with 56-16 or 56-18 in the baseline period.

The patient selection procedure is outlined in the

. A total of 23,798 patients with at least 1 claim for 1 or more hepatitis C drugs from 2007 to 2009 were identified. Of these, 6087 (25.6%) had the first hepatitis C claim between July 1, 2007, and December 31, 2008. However, 167 (2.7%) were excluded because they had claims for antiretrovirals or hepatitis B medications. A further 4997 (84.4%) out of 5920 patients had claims for both interferon and ribavirin, and of those 4774 (80.6%) began interferon and ribavirin within 14 days of each other. An additional 710 (14.9%) patients were excluded from the target sample because they were observed to continue hepatitis C therapy beyond 52 weeks after therapy initiation. Of the remaining 4064 patients, 392 (9.6%) patients were not continuously eligible throughout the expected length of therapy, and 969 (26.4%) additional patients were not continuously eligible throughout the 6-month period before the index claim, leaving a total of 2703 patients eligible for inclusion. Of the remaining patients, 463 (17.1%) used multiple dispensing channels (including patients who may have used multiple specialty pharmacies) and were excluded from the analysis. Ten patients (0.4%) exclusively used mail order pharmacy, a cohort that was too small to use in this analysis. Our comparison of adherence was restricted to 1489 (55.1%) exclusive users of a single specialty pharmacy and 741 (27.4%) exclusive users of retail pharmacy.

The demographic and clinical characteristics of the selected patients are presented in 

. On average patients in both groups were approximately 49 years old and more than half were male. Most patients in either group resided in a metropolitan statistical area, but therewere significantly more urban patients in the retail group than in the specialty group (87.8% vs 82.6%, 

 = .002). Out-of-pocket spending per 30-day adjusted hepatitis C prescription was signifi cantly higher in the retail group ($62.10 vs $39.25, 

 <.001). Retail patients had a signifi - cantly higher medication burden (14 vs 12 medications,

 = .002) and more frequent indications for depressive symptoms (24.6% vs 19.7%, 

 = .009) at baseline compared with specialty patients.

Compared to specialty pharmacy patients, a higher proportion of retail patients had an expected length of therapy of 24 weeks (81.4% vs 73.9%). A significantly higher proportion of specialty patients had 80% or better regimen adherence compared with retail patients (53.9% vs 40.6%).

Unadjusted regimen adherence was 66.8% in the retail group (95% confi dence interval [CI] 64.6%-69.2%) and 76.8% in the specialty group (95% CI 75.2%-78.6%). Multivariate adjusted regimen adherence was estimated to be 69.5% in the retail group (95% CI 66.7%-72.4%) and 78.1% in the specialty group (95% CI 75.7%-80.5%) (

). The retail and specialty groups had an average of 76 gap days (95% CI 71-81 days) and 58 gap days (95% CI 55-61 days), respectively. After multivariate adjustment, the retail and specialty groups had an adjusted average of 70 gap days (95% CI 64-77 days) and 55 gap days (95% CI 51-59 days), respectively (

). Patients in the specialty group had higher odds of attaining optimum adherence (adjusted odds ratio [AOR] 1.71; 95% CI 1.43-2.04). After adjusting for confounders, patients in the specialty group had a signifi cantly higher likelihood of achieving optimum adherence (AOR 1.59; 95% CI 1.32-1.91).

The primary objective of this study was to determine whether there are differences in hepatitis C regimen adherence between specialty and retail pharmacy patients. Our findings show that after adjusting for a number of known confounders, patients who used a single specialty pharmacy exclusively had on average 8.6% higher regimen adherence and 15 fewer gap days than patients who used retail pharmacy exclusively. Our adherence metric takes into account adherence to the complete regimen, rather than measures of adherence for individual medications aggregated across both drugs in the regimen. This approach makes our fi ndings more readily interpretable in the context of fi ndings that regimen adherence is a major modifiable factor governing the effectiveness of hepatitis C treatment.

We found that patients using specialty pharmacy had nearly 60% higher odds of achieving SVR than patients using retail pharmacy. Suboptimal adherence to ribavirininterferon regimens is highly prevalent in US hepatitis C patient populations. Other researchers who used claims data from a US managed care population and similar possession-based adherence metrics also reported adherence to the ribavirin-interferon regimen in the range of 57% to 76%.

 Another study of self-reported adherence from a cross-sectional survey of patients at 7 US urban clinics found that 7% of patients reported missing at least 1 injection of pegylated interferon in the last 4 weeks and 21% reported missing at least 1 dose of ribavirin in the last 7 days.

The addition of drugs such as novel protease inhibitors to the 2-drug hepatitis C regimen adds complexity, cost, and side effects to an already difficult, lengthy, and complex regimen. Although the newer 3-drug regimens may deliver improved cure rates,

 they also pose an additional barrier to achieving optimum regimen adherence in real-life environments. In addition to adequate therapy adherence, timely follow-up and intermediate viral load testing are required for accurately guiding responsedriven dosing. Patients must feel comfortable in the use of self-injectable medications and manage signifi cant sideeffects such as severe flu-like symptoms and neuropsychiatric effects.

 Given the increasingly complex nature of hepatitis C treatment, higher levels of care including

disease-specific and drug-specific patient counseling, instruction and training on self-injection, and assistance with administrative issues including payment processes and financial aid will be needed. Dispensing channels such as specialty pharmacy that provide such services may be positioned to drive superior outcomes for hepatitis C patients.

This study has several limitations, several of which are derived from the exclusive use of pharmacy claims data. We could not observe any medical claims or diagnosis codes associated with such claims for the primary condition (hepatitis C), confounders, correlates, and measures of comorbidity derived from medical claims data. However, claims for hepatitis C medications are almost exclusively dispensed in the outpatient pharmacy setting. We were able to generate proxies for important confounders such as baseline comorbidity (measured as medication burden) and diagnosis of baseline depressive symptoms (proxied as pharmacy claims for antidepressants). We did not have patient genotype information or any information about the prescribed length of ribavirin and interferon therapy or response-guided changes to such therapy. Our assumptions regarding the expected length of therapy might have miscategorized early discontinuers of 48-week therapy as having an expected length of therapy of 24 weeks. We used a possessionbased measure of regimen adherence, assuming that medication possessed by the patient would be used as prescribed. An earlier retrospective study of adherence to hepatitis C medication defi ned medication adherence as the ratio of the days of supply in the patient’s possession divided by the number of days in a fi xed follow- up period based on assumptions about the length of hepatitis C regimens similar to those in this study.

We could not distinguish between treatment-naïve and experienced patients except to ensure that no patient had been on an earlier hepatitis C regimen 6 months prior to the study regimen. For experienced patients, we were unable to ascertain the treatment status from previous rounds of treatment. Finally, although patient race has been shown to be a determinant of treatment response and tolerability,

 information about patient race was unavailable and thus could not be controlled for as a confounder in multivariate analysis. Owing to issues with a small sample size and heterogeneity in specialty pharmacy interventions, we restricted the specialty pharmacy to only a single specialty pharmacy. Our analysis might also have been confounded by patients with worse prognosis, patients who generally espouse healthy behaviors, or other patients who were more motivated to be adherent and successfully complete treatment, who selected specialty pharmacy over retail to take advantages of the additional services offered in this dispensing channel. Finally, we could not determine what specifi c services offered by specialty pharmacy were used by patients or whether such services were being used by retail pharmacy patients.

Findings from our study suggest that in real-life practice settings, patients exclusively using a specialty pharmacy had about 8.6% higher adherence, 2 fewer weeks of gap days, and a 60% higher likelihood of achieving optimum adherence compared with patients exclusively using retail pharmacy. Given the increasing complexity of hepatitis C regimens, dispensing channels such as specialty pharmacy that offer services to better assist patients and support them through the treatment regimens may drive improved adherence. Future research should not only describe differences in adherence between 2-drug and 3-drug hepatitis C regimens but also evaluate the effect of dispensing channel and specific services and programs for hepatitis C patients on regimen adherence.

Researchers found that patients who used specialty pharmacy had significantly higher adherence to hepatitis C drug regimens than patients who used retail pharmacy.