Assessment of Medication Regimen Complexity of COPD Regimens in Individuals Visiting Community Pharmacies

Introduction

Chronic obstructive pulmonary disease (COPD) is in the top ten causes of death worldwide¹ and is an important source of disability, morbidity, and health-care costs from medications and urgent care.² Impact increases with age and prevalence of comorbidities including diabetes and cardiovascular disease.² Non-adherence in COPD is high and includes primary non-adherence, poor device technique and non-persistence.^3,4 Patients may be routinely prescribed complex medication regimens which can lower adherence.

The Medication Regimen Complexity Index (MRCI) is a validated tool to quantify the difficulty of correctly taking medications as prescribed.^5,6 It takes between 2–8 minutes to assess a regimen and has high inter-rater and test-retest reliability.⁵ See Figure 1. The total of the three sub-scores can compare the regimen intricacy between individuals for the same or between conditions. A disease-specific score could be calculated (e.g., COPD regimen only) or the total regimen MRCI (i.e., all medications). An expert panel proposed using the MRCI to identify patients who would likely benefit from medication regimen review by pharmacists.⁶

Figure 1 Calculation of the medication regimen complexity index. Note: aAs another medication is already in an inhaler, no additional device points are added for the second medication in an inhaler.

Regimen complexity might lower adherence and worsen outcomes including mortality, exacerbation frequency and health-care use. An MRCI score over 8 was associated with 30-day hospital readmissions.⁷ Other data associated higher MRCI scores with increased mortality, non-adherence and hospitalizations in the elderly.⁸ Total MRCI score was related to 30-day readmissions and urgent care use for chronic diseases including COPD.⁹ Those readmitted or using urgent care had a total MRCI score of 30.8 but this was only 26.3 in those not requiring such care (p<0.01). In another study of chronic conditions, the MRCI predicted the potential for adverse drug events and unplanned 20-day hospital readmissions.¹⁰ Authors suggested that MRCI might identify those potentially benefiting from transitions of care interventions.

Data on MRCI COPD-specific regimens are very limited.^11,12 Complex medication regimens were shown to be associated with COPD severity and comorbid conditions.¹¹ An MRCI of 6 or more has been suggested as a screen for non-adherence.¹²

Objective

To perform a secondary analysis of previously collected data to calculate MRCI scores for COPD-specific regimens of participants from community pharmacies across Missouri and assess relationship to symptom and disease severity.

Materials and Methods

Study Design

For the initial study, inclusion criteria included participants over 40 years old with self-reported COPD, chronic bronchitis and emphysema who were recruited from 19 independent and 16 small-chain Missouri community pharmacies between October 10, 2016 and April 30, 2017. Dispensing records had screened for COPD medications in the previous 12 months including short-acting muscarinic antagonists (SAMA), long-acting muscarinic antagonists (LAMAs), long-acting beta agonists (LABAs), roflumilast and/or theophylline. Inclusion criteria also required that participants spoke English, were able to provide informed consent, complete a written questionnaire, and visit the community pharmacist investigator (CPI) within 30 days. The CPIs recruited 25 participants per pharmacy. Exclusion criteria were cognitive impairment limiting ability to participate in a face-to-face study visit, non-English speaking, and previously participating in this study at a different pharmacy site.

Participants were offered assistance with the questionnaire if the response to the Single-Item Literacy Screen (SILS) question¹³ indicated a risk for low health literacy based on self-reported confidence in completing medical forms. Demographic information, COPD assessment test (CAT) symptom score, antibiotic and oral corticosteroid usage in the last year, selected co-morbidities, estimated out-of-pocket monthly expenses for COPD medicines and all medicines, and insurance type were collected.

Participants listed all breathing-related medications and current instructions after an open-ended prompt. The CPI further prompted for unmentioned COPD medicines on the dispensing record. If the medicine was still taken, it was added to the current COPD regimen. Additional prompts were given for samples and patient assistance medications. De-identified 12-month dispensing reports were used to calculate the individual proportions of days covered (PDCs). Detailed methodology for recruitment and data collection are reported with the initial study.¹⁴ The St. Louis College of Pharmacy Institutional Review Board approved the protocol as consistent with the Nuremberg Code, Declaration of Helsinki, and Belmont Report. A waiver of consent for reviewing recruitment phase prescription data was also provided.

Data Analysis

Pharmacy prescription fill data were used to calculate the PDC for any one COPD maintenance medication rather than for each individual medication using previously described methods.¹⁵ Adherence was defined as PDC ≥0.80. Symptoms of COPD were assessed using the COPD Assessment Test Score (CAT) and categorized as low symptoms (<10), medium symptoms (10–20), or very symptomatic (>20).^16,17

The participant’s Global Initiative for Chronic Obstructive Lung Disease (GOLD) 2022 ABCD group was classified by primary investigators using CAT scores and risk for exacerbations² based on history of hospitalizations and oral corticosteroid and antibiotic use in the last year. Self-reported use of either an oral corticosteroid or antibiotic were considered “moderate” exacerbations. Use of an antibiotic and corticosteroid once each (or one of either) in the last 12 months was considered one exacerbation, use twice was counted as two exacerbations, and episodes associated with hospitalization or urgent care were classified as “severe”.

The self-reported list of current COPD medicines (including samples and patient assistance medicines) was used as the COPD regimen. Medicines self-discontinued (i.e., participant reported providers were unaware of discontinuation) were included in the current regimen, but not medicines reported as discontinued by providers. A medication count in the regimen consisted of the number of unique medications in the regimen (e.g., albuterol by inhaler and nebulizer are considered one medication); a combination product containing fluticasone and salmeterol were counted as two medications. The formulation count in the regimen consisted of the number of products (e.g., fluticasone plus salmeterol was counted as one formulation). However, albuterol inhaler and albuterol nebulizing solution were counted as two formulations. The number of maintenance medications included any regularly scheduled medication (including scheduled SABA and SAMA). The number of distinct devices was also calculated. Inhaled device polypharmacy (IDP) was identified as using three or more different respiratory devices.¹⁸

The MRCI for each COPD regimen was calculated using validated, previously published methods.⁵ See Figure 1. The University of Colorado’s electronic MRC Data Capture Tool and accompanying MRCI additional instructions¹⁹ were used to calculate MRCIs for the COPD regimen including as-needed medications. To update for newer respiratory devices, Respimat^® was given a device weighting score of 3;¹¹ Neohaler^® was weighted as 3 (consistent with Handihaler^® capsule device). Pressair^® and Ellipta^® were weighted as “other dry powdered inhalers” (+3 weight). Ellipta^® and Diskus^® are not identical, so were counted as separate devices of equal complexity (+3 weight). MRCI regimen scores were categorized as low (≤4), medium (5–8) or high (>8).¹⁸ Descriptive statistics were employed for demographic and regimen-related data. Chi-squared compared categorical data with significance defined as p<0.05.

Results

Baseline demographic data are summarized in Table 1. Most participants had a history of prior or current tobacco use, co-morbidities and some form of insurance. Self-reported spending on COPD and all medications was widely variable. The participant population was highly symptomatic (mean CAT 21.5 ± 8.4; 90% in GOLD categories B or D) and at high risk for exacerbations (59% GOLD categories C or D). Adherence to even one maintenance COPD medication was extremely low (average PDC 0.43 ± 0.37); PDC >80% for one medication (28.7%). Approximately one third of participants were at risk for low health literacy and under the care of a pulmonologist. The most common medications in regimens were LABA, ICS and SABA.

Table 1 Baseline Demographics and Clinical Characteristics of Participants

Data related to the MRCI are reported in Table 2. The average COPD MRCI score was 14.05 ± 5.4, with sub-score A (7.73 ± 3.2) and B (4.71 ± 2.2) contributing the majority. The COPD MRCI scores increased across the GOLD categories A to D. Over 90% reported an inhaler device with 30–40% using other devices as well. Close to 30% used three or more devices (i.e., inhaled device polypharmacy). The medication count averaged 3.79 ± 1.5 different medications with similar averages across CAT symptom and GOLD categories. Considering combination inhalers, the number of different formulations was 2.54 ± 1.1.

Table 2 Medication Regimen Complexity Data

Statistical analysis of MRCI, inhaled device polypharmacy, and health literacy is reported by GOLD category and CAT scores in Table 3 and Table 4. High MRCI scores (>8) increased in GOLD Category C and D as did IDP (p<0.05). Similarly, MRCI increased and IDP increased with CAT score (p<0.05). No MRCI difference was detected based on risk for low health literacy for either GOLD category or CAT scores.

Table 3 Medication Regimen Complexity, Inhaled Device Polypharmacy and Health Literacy by GOLD Severity Classificationa (N = 709)

Table 4 Medication Regimen Complexity, Inhaled Device Polypharmacy and Health Literacy by CAT Score (N= 709)

Discussion

In this analysis, scores for COPD MRCI appear high and regimens averaged almost four medications. This is consistent with a smaller COPD MRCI sampling previously reported.¹¹ See Table 5. The MRCI for COPD regimens also appears higher compared with other disease-specific MRCI reported. These participants’ COPD MRCI averaged 14.05 ± 5.4 compared with congestive heart failure (7.2 ± 3.4)²⁰, HIV (4.93 ± 2.12)¹⁹, geriatric depression (3.03 ± 1.1)¹⁹ and hypertension (3.53 ± 1.47).¹⁹

Table 5 Comparison of Reported MRCI for Chronic Disease Regimens

Diabetes might be expected to have a higher MRCI than other chronic conditions due to regimens containing multiple oral and injectable medications. The diabetes MRCI in an Ethiopian study ranged from 2–10.¹⁸ (See Table 5). The percentages in MRCI categories of low (<4), medium (5–8) or high (> 8) complexity were 31.3%, 46.4 and 22.2% respectively. They reported 70% medication adherence using the 8-point Morisky Medication Adherence Scale (score >8) with 43% having good glycemic control (fasting blood glucose between 70–130 mg/dL). High MRCIs correlated to lower adherence and poorer control; low or medium MRCIs did not affect either. By comparison, close to 85% of participants in this study had a high MRCI (>8). This study lacked participant-specific PDC data, but adherence overall was very low (only about 29% adherent to even one maintenance medication). While possible, it is unknown whether higher MRCI COPD regimens in this study correlated to lower adherence.

In a US-based study,¹⁹ diabetes regimens were reported to have a mean MRCI of 6.28 ± 3.4 (range 1–15) with a median of almost 2 medications. By comparison, in this COPD sampling the total and range of MRCI as well as number of disease-specific medications was much higher.

Data available on COPD-specific MRCI are limited.^11,12 For 222 Australians with COPD, the number of COPD-specific medications and MRCI are reported in Table 5.¹¹ Sub-scales A and B had more impact on the total COPD MRCI than sub-score C. The MRCI scores were significantly higher for participants in GOLD category D than A, B or C (15.5 vs 13.5, 12.5 and 13, respectively); the total medication number and devices were also higher in GOLD group D. Device polypharmacy was seen in 30.6%. The COPD MRCI scores, MRCI A and B sub-scores’ relative contribution to the total MRCI, and incidence of IDP were similar to this analysis.

However, the Australian study¹¹ also looked at the participants’ MRCI for all diseases and relationship to COPD outcomes. Participants with cardiovascular, gastrointestinal and neurologic comorbidities had a higher total MRCI with non-COPD MRCI making a greater contribution to total MRCI. Whereas both COPD and non-COPD medications contributed to higher MRCI scores in participants with metabolic or psychiatric comorbidities. COPD-specific MRCI and COPD-specific medication count were significantly correlated to COPD outcomes, such as CAT symptom scores, St. George Respiratory Questionnaire scores, 6-minute walk test, and prior year exacerbation and hospitalization history.

A US study assessed the association between total MRCI with adherence and COPD outcomes of 400 participants with COPD plus either hypertension or diabetes or both.¹² The total MRCI was calculated for all maintenance medications, but as-needed medications were not included. Adherence over 15 months was measured by self-report (Medication Adherence Rating Scale) and device dose counters. Severity was measured by COPD Severity score. The total MRCI was not significantly different between those with low and adequate adherence by self-report (33.5 ± 13.4 vs 32 ± 13.6) or by device dose counter (32.7 ± 13.6 vs 31.5 ± 13.6). There was a low correlation between COPD Severity scores and total MRCI score (r = 0.26, p<0.0001) and medication counts (r = 0.19; p<0.0001). Interestingly, there was no correlation between the total MRCI score and either blood pressure or diabetes control. Low COPD medication adherence (less than 80%) was 45% (self-report) and 52.6% (dose count); low adherence was less frequent for hypertension (28.9%) and diabetes (38.8%). Unfortunately, COPD-specific MRCI scores were not reported, but some associations between total MRCI scores and COPD severity and adherence were reported. In comparison, adherence by self-report and dose counters were higher than in this analysis using PDC. Fewer participants (24%) were at risk of low health literacy (also assessed by the Single Item Literacy Screener) than in this analysis.

These data^11,12 along with this analysis support further investigation of the relationship between COPD MRCI scores and outcomes. For example, does an increasing MRCI score contribute to poor outcomes (e.g., lower adherence, more exacerbations)? Or does undetected non-adherence result in additional medication prescribing causing higher MRCI scores? Are increases in MRCI scores secondary to COPD progression over time? Additionally, COPD MRCI scores should be studied in the context of an individual’s total medication regimen MRCI score. Those with COPD appear to have a high incidence of comorbidities, likely contributing to the total regimen MRCI scores.

An MRCI score has potential clinical applications for COPD patients. For example, prescribers and pharmacists can intervene when encountering patient regimens with high MRCI scores and low adherence based on PDC. An initial screen of COPD patient regimens can identify those with both a high total MRCI score (for all medications) and a high COPD MRCI score (>8). Patients with several medications (maintenance and as needed), various devices, and multiple daily dosing schedules may easily have a COPD MRCI score in the range of 15–20. Combining very high symptoms (e.g., CAT >20) with a high COPD MRCI score (e.g., >15) and a low PDC could provide a useful screen to target regimens for interventions. Health-care professionals have opportunities to intervene during medication reconciliation at transitions of care to lower regimen complexity, potentially improving adherence and hospital readmission rates.

As the largest portion of the MRCI score comes from A and B sub-scales, logical interventions would include decreasing the number of devices and formulations and by continuing a device the patient already knows and uses correctly (i.e., device continuity).²¹ In considering how to simplify COPD regimens, Table 6 includes factors to consider and Figure 2 lists FDA-approved medications in available devices. For example, a patient has prescriptions for a SABA in a metered dose inhaler (MDI), and maintenance medicines an ICS/LABA, and a LAMA in two other devices. Device polypharmacy (i.e., three devices) exists. But all drug classes are available in MDIs. By recommending a switch to a combination ICS/LABA/LAMA MDI (budesonide/glycopyrronium/fluticasone furoate) only an MDI device is used for both maintenance and SABA and only one maintenance formulation. The device MRCI sub-score for maintenance medicines decreases from 10 to 4 and combined frequency and directions sub-scores decrease from 4 to 3 (decreasing the total COPD MRCI by 7). (See Figure 1.) It may be more commonplace to replace 3 or 4 devices with 2. For example, a regimen consists of LAMA twice daily and LABA once daily in different dry powdered formulations plus albuterol MDI as needed. A once daily combination LABA/LAMA dry powdered product (and continuing albuterol MDI as needed) could decrease both the number of devices and doses. Both examples lowered the MRCI score, however the total MRCI score would still be “high” at >8. On the other hand, providers and pharmacists may identify highly symptomatic individuals despite adherence to therapy. In such cases it would be useful to consider “stepping up” therapy with minimal MRCI score increases (e.g., adding a combination formulation or an additional medication in the same device vs another drug in a different device). When brainstorming, simplification strategies such as co-pays, formulary status, patient device preferences and ability to use devices correctly are also important to consider (Table 6).

Table 6 Factors to Consider Before Recommending Device Regimen Changes

Figure 2 FDA approved COPD medications and available devicesa. Abbreviations: ACL, aclidinium; ALB, albuterol; BEC, beclomethasone; BUD, budesonide; CIC, ciclesonide; FF, formoterol furoate; FP, fluticasone propionate; GLY, glycopyrrolate; IPRA, ipratropium; LABA, long-acting beta agonist; LAMA, long-acting muscarinic antagonist; LEV, levalbuterol; MDI, metered-dose inhaler; MOM, mometasone; OLO, olodaterol; SABA, short-acting beta agonist; SAL, salmeterol; SAMA, short-acting muscarinic antagonist; TIO, tiotropium; UM, umeclidinium; VIL, vilanterol. Note: aWhen making product switches, consider if a similar dose (e.g., low, medium, high) is available in new device.

In designing interventions, not all factors affecting adherence, adverse events or poor health outcomes are included in the MRCI score. For example, warfarin orally once daily may have a low MRCI score, but the underlying condition and bleeding may result in adverse outcomes despite good adherence. Current data do not indicate a relationship between MRCI and low health literacy. However, in individuals with low health literacy, high total or COPD MRCI may require more caregiver home support including use of medication organizational tools such as pill boxes, medication alarms, phone apps, etc.

Strengths

Our secondary analysis of previously collected data had a larger participant size than other MRCI reports for COPD or other chronic diseases. Including both rural and urban participants with COPD from across the state lends some external validity. A uniform method of collecting medication histories was used. The COPD regimen was collected from the participant perspective likely reflecting the participants’ “real world” perception of their regimen complexity. The MRCI was calculated using validated methods and an automated program for uniformity and accuracy. This report adds to the limited data available on the MRCI in COPD and matches MRCI scores to COPD symptoms and GOLD severity categories (A, B, C, D). The risk for low health literacy was evaluated in relation to COPD MRCI scores.

Limitations

This secondary analysis does have drawbacks. First, participants’ self-reported COPD without spirometric confirmation; however, these data could apply to those with respiratory diseases. Second, MRCI was not matched to patient-specific PDCs; adherence was calculated for the entire cohort. However, overall adherence was extremely poor for even one maintenance medication. Future studies would benefit from calculating the total MRCI. As co-morbidities were common, other medications (e.g., metformin, insulin, lisinopril, atorvastatin) could add substantially to the MRCI total score, adversely affecting adherence and outcomes. Participants’ self-reported medication costs were collected and risk for low health literacy was assessed. However, it would be useful in future studies to attempt to capture actual and perceived burden of medication costs. It would be important to consider health literacy and medication costs when interpreting the impact of lowering MRCI scores on adherence and COPD outcomes.

Future Research

There is a need for further investigation of MRCI scores in COPD. Prospective studies are needed to assess if decreasing COPD-specific or total MRCI scores improves adherence or COPD outcomes (e.g., lowers symptoms, exacerbation frequency, urgent care use, hospital admissions or readmissions) or decreases health-care and medication costs. Nothing has been reported about the effect of MRCI score on patients’ experience. E.g., what is the patient’s perception of the burdensomeness of taking medications? Do patients feel more empowered if included in shared decision-making regarding device selection?

Another avenue to explore would be comparing patient-specific PDC to MRCI scores. This would help assess the individual impact of COPD as well as the entire medication regimen on individual adherence. Including non-prescription medications in the calculation of total MRCI score has been suggested as these can add to the overall score.²² An integrated assessment of MRCI score into the electronic record or prescription dispensing systems could also help identify high MRCI²³ scores making it easier for prescribers and pharmacists to identify and simplify complex regimens. Prescription dispensing systems that calculate MRCI scores (both total and COPD specific) could help pharmacists identify patients for therapeutic and transition of care interventions.

Of note GOLD 2023 recommendations²⁴ adjusted the ABCD categories. Categories C and D were collapsed into a single category of those at high risk for exacerbations (category E). Due to the small number of our participants in the C category, this would not be expected to alter our overall results. However, the new categories (A, B, E) should be used in future studies.

Conclusion

COPD MRCI scores are higher than for other chronic diseases; COPD MRCI scores increase with symptom scores and GOLD severity categories. The MRCI could be used to identify complex regimens and patients who may benefit from prescriber and pharmacist interventions. More research is needed to assess whether prospectively lowering MRCI scores could improve adherence and COPD outcomes.