key: cord-0931339-df66hn17
authors: Zuckerman, Autumn D; DeClercq, Josh; Choi, Leena; Cowgill, Nicole; McCarthy, Kate; Lounsbery, Brian; Shah, Rushabh; Kehasse, Amanuel; Thomas, Karen; Sokos, Louis; Stutsky, Martha; Young, Jennifer; Carter, Jennifer; Lach, Monika; Wise, Kelly; Thomas, Toby T; Ortega, Melissa; Lee, Jinkyu; Lewis, Kate; Dura, Jillian; Gazda, Nicholas P; Gerzenshtein, Lana; Canfield, Scott
title: Adherence to self-administered biologic disease-modifying antirheumatic drugs across health-system specialty pharmacies
date: 2021-08-18
journal: Am J Health Syst Pharm
DOI: 10.1093/ajhp/zxab342
sha: 1090d7739b0e4004eb85b6e0e5f9b1eef1afe9b5
doc_id: 931339
cord_uid: df66hn17

DISCLAIMER: In an effort to expedite the publication of articles related to the COVID-19 pandemic, AJHP is posting these manuscripts online as soon as possible after acceptance. Accepted manuscripts have been peer-reviewed and copyedited, but are posted online before technical formatting and author proofing. These manuscripts are not the final version of record and will be replaced with the final article (formatted per AJHP style and proofed by the authors) at a later time. PURPOSE: Adherence to self-administered biologic disease-modifying antirheumatic drugs (bDMARDs) is necessary for therapeutic benefit. Health-system specialty pharmacies (HSSPs) have reported high adherence rates across several disease states; however, adherence outcomes in rheumatoid arthritis (RA) populations have not yet been established. METHODS: We performed a multisite retrospective cohort study including patients with RA and 3 or more documented dispenses of bDMARDs from January through December 2018. Pharmacy claims were used to calculate proportion of days covered (PDC). Electronic health records of patients with a PDC of <0.8 were reviewed to identify reasons for gaps in pharmacy claims (true nonadherence or appropriate treatment holds). Outcomes included median PDC across sites, reasons for treatment gaps in patients with a PDC of <0.8, and the impact of adjusting PDC when accounting for appropriate therapy gaps. RESULTS: There were 29,994 prescriptions for 3,530 patients across 20 sites. The patient cohort was mostly female (75%), with a median age of 55 years (interquartile range [IQR], 42-63 years). The original(ie, prereview) median PDC was 0.94 (IQR, 0.83-0.99). Upon review, 327 patients had no appropriate treatment gaps identified, 6 patients were excluded due to multiple unquantifiable appropriate gaps, and 420 patients had an adjustment in the PDC denominator due to appropriate treatment gaps (43 instances of days’ supply adjusted based on discordant days’ supply information between prescriptions and physician administration instructions, 11 instances of missing fills added, and 421 instances of clinically appropriate treatment gaps). The final median PDC after accounting for appropriate gaps in therapy was 0.95 (IQR, 0.87-0.99). CONCLUSION: This large, multisite retrospective cohort study was the first to demonstrate adherence rates across several HSSPs and provided novel insights into rates and reasons for appropriate gaps in therapy.

(eg, tumor necrosis factor α inhibitors, interleukin antagonists) as well as a growing number of newer oral agents (eg, Janus kinase inhibitors). Biologic DMARDs are often selected for patients with ongoing moderate or high disease activity after treatment with conventional DMARDs (eg, methotrexate), given their demonstrated ability to slow disease progression, induce remission, and improve radiologic outcomes. 1 Despite the benefits of bDMARDs, previous research has found that medication adherence within the RA population ranges from 44% to 83%. [2] [3] [4] [5] The optimal level of medication adherence and preferred method for evaluating adherence in RA are not established; however, nonadherence to antirheumatic treatments is associated with poor 28-joint Disease Activity Scores, indicating worse disease activity. 6, 7 Though a specific threshold for adherence correlated to clinical outcomes has not been reported, the Pharmacy Quality Alliance, a public-private cooperative founded to promote appropriate medication use, recommends an adherence threshold of 0.8, calculated as proportion of days covered (PDC), for patients with RA taking noninfused biologic medications. 8 PDC is determined using pharmacy claims data to calculate the amount of medication a patient has in hand (covered days) divided by the number of days in a timeframe. 9 As with all adherence calculations using pharmacy claims data, PDC is a surrogate endpoint of actual patient administration and is used to estimate adherence. 9 Alongside adherence challenges, drug spend on biologic products used in the treatment of RA and other inflammatory conditions is significant and was identified as the highest per-A c c e p t e d M a n u s c r i p t member-per-month contributor to annual drug expenditures for top commercial pharmacy benefit managers throughout a recent 5-year period (2015-2019). 10 Specialty pharmacies play a key role in assisting patients and prescribers who utilize bDMARDs by providing insurance navigation, care coordination, medication dispensing, and longitudinal patient monitoring services. Health-system specialty pharmacy (HSSP) programs have grown rapidly in recent years. As of 2019, over 26% of all hospitals and 89% of larger hospitals (over 600 staffed beds) reported operating a specialty pharmacy. 11 Within the HSSP care model, pharmacists and pharmacy technicians provide integrated care for patients alongside physicians and other clinical care team members. These integrated programs have demonstrated their ability to optimize patient adherence as measured by PDC and medication possession ratio (MPR) in a variety of specialty disease categories, including oncology, multiple sclerosis, and pulmonary arterial hypertension. [12] [13] [14] A recent study demonstrated high rates of bDMARD adherence in 675 patients served by one HSSP. 15 However, more data is needed to assess if the growing number of HSSPs have similar rates of high adherence.

Health-system specialty pharmacists have several touch points with patients on specialty medications due to their integration within clinics. This entails frequent communication with patients, often in person and via audio and/or video communications. The HSSP care model includes shared documentation systems within the electronic health record (EHR), whereby the entire multidisciplinary care team documents treatment decision making, disease progression, and treatment outcomes. This integration also allows for comprehensive tracking of patient adherence patterns, including documentation of instances of clinically appropriate medication holds. Although interruption in treatment is common in RA due to active infections or surgical interventions, these holds would not be A c c e p t e d M a n u s c r i p t accounted for in traditional PDC calculations because the clinical details of the treatment timeline are not incorporated. The PDC method and other claims-based adherence calculations suffer from a lack of industry standardization and transparency and from reliance on pharmacy claims data alone to approximate adherence. 9 The lack of clinical details in the traditional PDC calculation can cause appropriate gaps in therapy to be labeled as periods of nonadherence. The impact of the HSSP model on adherence in RA across multiple HSSP programs has not been previously explored. Additionally, given HSSP access to clinical data, there is a previously unreported opportunity to understand the frequency of appropriate medication holds and their impact on traditionally calculated PDC compared to an adjusted PDC accounting for appropriate gaps in claims.

The objectives of the multisite study described here were to evaluate adherence to specialty medications in patients with RA receiving care within integrated HSSP models and to investigate the frequency at which retrospectively reviewing integrated care team documentation of a suboptimal PDC (ie, <0.8) revealed an opportunity to correct PDC calculation to more accurately reflect true medication adherence. Figure 1 illustrates the shared and unique roles of clinic staff as reported by participating HSSPs and the number and type of clinic staff within participating HSSPs. Similarities in the HSSP practice model existed, allowing for the ability to combine and report adherence data. HSSPs had access to their respective health-system EHRs, enabling comprehensive patient clinical review and communication with prescribing providers. When HSSPs received a referral for a new specialty medication, pharmacy staff performed a benefits investigation and assessed the patient's ability (based on payer and manufacturer restrictions) and willingness to fill the prescription at the HSSP. Depending on this determination and the HSSP's practice model, the HSSP then assisted with insurance approval and obtaining financial assistance for the patient as needed. If the HSSP was unable to fill the prescription or the patient preferred an alternate pharmacy, the prescription was triaged to the patient-preferred or insurance-or manufacturer-mandated pharmacy. Due to their integration, HSSPs serve as a resource for specialty patients regardless of whether they receive drugs dispensed from the HSSP, often answering drug information questions and helping address and mitigate adverse effects resulting from specialty therapy.

PDC was calculated by generating a supply diary for each patient from the time of the index prescription (the date of first fill of an included medication within the study period) to the date of the last fill, omitting the days' supply in the patient's last fill. A fill was defined as the "sold date" within the pharmacy dispensing software. Excess supply due to overlapping refills was shifted forward, never backward, and oversupply at the end of the time period was truncated from the total supply. Among patients who switched therapy, it was assumed that patients stopped the first drug before starting the next drug, and thus excess supply A c c e p t e d M a n u s c r i p t from the first drug was not carried forward when a new drug was filled. A single PDC value was calculated for each patient, with all fills for any included medication for a single patient analyzed in aggregate.

Identification of appropriate reasons for therapy gaps and adjustment of PDC. After an initial PDC calculation, sites were provided a list of study IDs and gap dates for patients with a PDC of <0.8. Sites then reviewed EHR and pharmacy dispensing data for patients with a PDC of <0.8 to assess for a reason for gaps in fill data. Patients could have had more than one appropriate gap and thus more than one reason. Reasons for appropriate gaps in therapy (as documented in claims data) were discussed and agreed upon by all sites a priori.

Sites were responsible for reviewing individual patient data and determining whether an appropriate gap existed and into which category the reason for the gap was assigned.

Unique instances were discussed among all sites and final categorization was agreed upon.

Patients were excluded from the final PDC analysis if there were multiple appropriate extended gaps for which dates of gaps could not be quantified. If the reason for gaps in fill data was discordance between physician directions and the prescription-specified days' supply, the days' supply (PDC denominator) was adjusted to mirror physician administration instructions and the PDC recalculated based on the new days' supply data. If sites found fills that were not present upon the first data extraction, the missing fill was added to the dataset and the PDC recalculated with the new fill's days' supply in the numerator and denominator. Sites recorded the number of days that accounted for appropriate therapy gaps due to the following reasons: infections, physician-directed drug holiday, external fills, patient utilized samples and/or patient assistance program enrollment, transition to intravenous therapy, transition to an oral or infused medication that was not included in the study list of medications, allergic reaction, held for pregnancy. For these instances, the A c c e p t e d M a n u s c r i p t number of days of the appropriate gap were removed from the PDC denominator.

Therefore, each patient's adjusted PDC resulted from accounting for adjusting days' supply based on physician instructions, adding missing fills, and removing appropriate gap days from the PDC denominator. Patients were grouped into 3 PDC categories (>0.5, 0.5-0.8, and <0.5) before and after PDC adjustment.

Statistical analysis. Descriptive statistics were used to describe data. Proportions were calculated for categorical variables, while the mean, median, standard deviation (SD), and IQR were used to describe continuous variables. The primary outcome was PDC across all sites. Secondary outcomes included reasons for apparent nonadherence or treatment gaps in patients with a PDC of <0.8 and the impact of adjusting PDC when accounting for appropriate gaps in therapy.

With the summary data for PDC from 20 sites, such as mean (SD) values as well as sample size for each site, we performed a meta-analysis using a random-effects model. The metafor package for R (R Foundation for Statistical Computing, Vienna, Austria) was utilized to perform the meta-analysis and make a forest plot to present the results.

There were 29,994 prescriptions from 3,530 patients across the 20 sites included. Limitations of using pharmacy claims to evaluate patient adherence have previously been described. 9, 25, 26 A small study of specialty pharmacy patients serviced by an HSSP found that up to 40% of patients with a PDC of <0.8 had appropriate reasons for gaps in therapy, primarily due to provider-directed medication holds (69%). 25 However, our study is among the first to quantify rates and reasons for low PDC rates that may inaccurately describe true medication adherence in a rheumatology population. Accounting for treatment gaps is important when evaluating bDMARD adherence, as patients are often directed to hold therapy due to infections, surgery, or other illnesses. EHR review revealed that over half of patients with a PDC of <0.8 had an appropriate reason for gaps in pharmacy claims, most commonly due to clinically appropriate holds. Due to the large sample size and its inclusion of few patients with an original PDC of <0.8, the overall impact of adjusting PDC based on appropriate gaps was minimal, with the median increasing from 0.94 to 0.95.

However, after adjusting for appropriate gaps, fewer patients remained in the lower-PDC categories (<0.5 and 0.5-0.8, with decreases from 2.8% to 1.2% and from 17.5% to 11%, A c c e p t e d M a n u s c r i p t respectively), while more patients entered the higher-PDC categories (0.8-1 and 1, with increases from 59.2% to 65.1% and from 20.6% to 22.8%, respectively). Based on these findings, we believe additional clinical data, not just pharmacy claims alone, is likely needed to accurately assess adherence in an RA population. The clinical information to accurately identify these scenarios is available within the HSSP model due the integrated care model and shared documentation, but such information is unlikely to be accessible to or accurately depicted by nonintegrated specialty pharmacies. 26 These results also demonstrate the limitations of assessing adherence using pharmacy claims generated at the pharmacy level, as 79 patients appeared nonadherent due to filling medications external to the HSSP during the study period. As adherence is often a quality measure for accreditation and contracting standards to which pharmacies are accountable, it is important to note that this limitation of data availability may falsely lower calculated adherence rates at the pharmacy level. 9 The study was not without limitations. Patients with a PDC of >0.8 were not reviewed for potential appropriate reasons for nonadherence. Additionally, other potential reasons for appropriate nonadherence, such as a patient having a sufficient supply on hand or waiting to start treatment, were identified upon chart review; however, these were not accounted for as they were not agreed-upon valid reasons a priori. These limitations could have led to bias such that reported PDC rates were lower than the true adherence rates. For patients with PDC of <0.8, site reviewers assessed the EHR for evidence of reasons for interruption in therapy, and the number of days removed from the PDC denominator to account for an appropriate gap was dependent on the reviewer's interpretation of the clinical information. Therefore, review bias may have occurred during evaluation of the patient's EHR.

A c c e p t e d M a n u s c r i p t

The large, multisite retrospective cohort study was the first to demonstrate bDMARD adherence rates across several HSSPs and demonstrate the benefit of the HSSP model in supporting high adherence rates. Additionally, the results provide novel insights into rates and reasons for appropriate gaps in bDMARD therapy that are otherwise unaccounted for by common methods of approximating medication adherence. Accounting for appropriate gaps in pharmacy claims is an important element of evaluating true nonadherence in specialty disease states in which clinically appropriate therapy holds are common. However, accessibility to this data is often limited beyond the HSSP model, which represents a unique opportunity for HSSP practices to further investigate the optimal methods for quantifying medication adherence.

We acknowledge Bridget Lynch, PharmD, for her assistance with data collection and assimilation. Additionally, we acknowledge the integrated health-system specialty pharmacists and pharmacy technicians who provide optimal patient care at each of our institutions.

Dr. Zuckerman, Dr. DeClercq, and Dr. Choi report research support from Sanofi, Inc., unrelated to the work described here within the last 36 months. M a n u s c r i p t 8.

Pharmacy Quality Alliance Adherence: PQA measure overview. Accessed November M a n u s c r i p t 

 This large, multisite retrospective cohort study is the first to demonstrate adherence rates to biologic disease-modifying antirheumatic drugs across several health-system specialty pharmacies, demonstrating the benefit of this increasingly common model.

 This study is among the first to utilize the review of health-system integrated clinical documentation to investigate gaps and reasons for inappropriate and appropriate gaps in pharmacy claims data.

A c c e p t e d M a n u s c r i p t  The study results demonstrate the impact on proportion of days covered when the calculation is modified to correct for clinically appropriate gaps in refill history.

A c c e p t e d M a n u s c r i p t 

American College of Rheumatology guideline for the treatment of rheumatoid arthritis

Impact of specialty pharmacy on treatment costs for rheumatoid arthritis

Compliance with biologic therapies for rheumatoid arthritis: do patient out-of-pocket payments matter?

Adherence, persistence, and expenditures for high-cost anti-inflammatory drugs in rheumatoid arthritis: an exploratory study

Utilization and adherence patterns of subcutaneously administered anti-tumor necrosis factor treatment among rheumatoid arthritis patients

Impact of inadequate adherence on response to subcutaneously administered anti-tumour necrosis factor drugs: results from the Biologics in Rheumatoid Arthritis Genetics and Genomics Study Syndicate cohort

Adherence and health literacy as related to outcome of patients treated for rheumatoid arthritis: analyses of a large-scale observational study

Navigating the Wild West of medication adherence reporting in specialty pharmacy

Express Scripts

ASHP national survey of pharmacy practice in hospital settings: prescribing and transcribing-2019

Adherence to disease-modifying therapies at a multiple sclerosis clinic: the role of the specialty pharmacist

Integrated clinical and specialty pharmacy practice model for management of patients with multiple sclerosis

High rates of medication adherence in patients with pulmonary arterial hypertension: an integrated specialty pharmacy approach

Rheumatoid arthritis medication adherence in a health system specialty pharmacy

Analyzing the costs of developing and operating an integrated health-system specialty pharmacy: the case of a centralized process for specialty clinic patients

Pharmacist interventions to improve specialty medication adherence: study protocol for a randomized controlled trial

Modeling specialty medicine access: understanding key health system processes and players

Specialty pharmacy and the medication access dilemma

Benefits of developing a collaborative, outcomes-based specialty pharmacy program

Advancing patient care through specialty pharmacy services in an academic health system

Clinical and economic impact of a specialty care management program among patients with multiple sclerosis: a cohort study

The effect of an integrated health system specialty pharmacy on HIV antiretroviral therapy adherence, viral suppression, and CD4 count in an outpatient infectious disease clinic

Risk factors for medication nonadherence to self-injectable biologic therapy in adult patients with inflammatory bowel disease

Closing the gap: identifying rates and reasons for nonadherence in a specialty population

Improving the validity of determining medication adherence from electronic health record medications orders

A c c e p t e d M a n u s c r i p t