If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Department of Pediatrics, Sections of Allergy and Immunology and Pulmonary and Sleep Medicine, University of Colorado School of Medicine and the Breathing Institute, Children's Hospital Colorado, 13123, East 16th Ave, Aurora, CO, 80045, USA
Department of Pediatrics, Sections of Allergy and Immunology and Pulmonary and Sleep Medicine, University of Colorado School of Medicine and the Breathing Institute, Children's Hospital Colorado, 13123, East 16th Ave, Aurora, CO, 80045, USA
Department of Pediatrics, Sections of Allergy and Immunology and Pulmonary and Sleep Medicine, University of Colorado School of Medicine and the Breathing Institute, Children's Hospital Colorado, 13123, East 16th Ave, Aurora, CO, 80045, USA
Following cohort adherence trajectories over time based on initial adherence is possible.
•
Adherence declined in all adherence groups, with the largest relative decreases in the suboptimally adherent groups.
•
Rescue use also declined in all groups.
•
Electronic Medication Monitoring is a tool that may be useful in promoting patient driven care.
•
Further study is needed on how to longitudinally engage patients.
Abstract
Non-adherence to treatment regimens in asthma is well described, however less is known about temporal patterns of medication use. We monitored 20 weeks of controller medication use and analyzed these patterns in patients ≥4 years of age with self-reported asthma enrolled in a digital health program. At baseline, approximately 20%, 28%, 25% and 27% of patients had optimal, moderate, sub-optimal and poor adherence, respectively. Medication adherence decreased in all groups in this study. The largest absolute decreases in adherence (−32%) were observed for moderately adherent patients. Certain adherence patterns which demonstrated greater declines, that, once identified, could be intervened upon.
]. While non-adherence is a known problem, less is known about temporal patterns of individual medication use.
Electronic medication monitors (EMMs) for inhaled asthma medications allow for real-time, objective and passive collection of medication use data. Using a digital health database with EMM-recorded inhaler use, the primary objective was to define controller medication adherence subgroups and describe adherence over time in patients with asthma. The secondary objective was to compare patient-level characteristics by those subgroups.
1. Methods
The study included patients ≥4 years old with self-reported asthma enrolled in a digital health platform and using EMM-compatible controller medications between August 2012 and February 2018. Patients were required to have ≥20 weeks of controller medication monitoring including the first week of enrollment (the training week).
EMMs (Propeller Health, Madison, WI) were provided to patients for their inhaled controller (including dry powder and metered dose inhalers) and short-acting beta-agonist (SABA) medications [
]. EMMs passively captured the date and time of actuations. Electronic notifications were available to alert patients and providers of missed doses of controller medication and to warn of SABA overuse via a dedicated smartphone application and website.
Analyses included data analyzed from weeks 2–20 after enrollment (week 1 was excluded as a run-in week where the participant was acclimating to the digital platform). Data on patient age and the type of controller inhalers were collected. Adherence to controller medications was calculated weekly as the number of actuations divided by the number prescribed, capped at 100% to avoid inflating adherence due to overuse. Symptom directed therapy is not widely prescribed in the US, so this was not considered as a separate treatment pattern. Patient-level adherence subgroups were determined using mean adherence at baseline, during weeks 2–6. We a priori classified the adherence subgroups as: optimal (>80%) [
], moderate (>50 to ≤80%), sub-optimal (>20–50%), and poor (≤20%). Mean weekly SABA actuations were calculated using EMM data.
Patient demographics were summarized using counts (%) and means (standard deviations [SD]) and compared across adherence subgroups using chi-square tests for categorical and t-tests for interval-scale variables (α = 0.05). Next, binomial (for binary variables) and linear (for interval-scale variables) regressions were used to estimate differences between patient demographics in optimal vs. the other adherence groups and poor vs. the other adherence groups. Finally, using linear regression, changes in adherence (i.e. the slope) during baseline and the overall study period were quantified for each group. Patients without 20 weeks of complete controller medication monitoring data and complete patient demographic data were excluded from analyses.
This protocol was determined to be exempt by the Copernicus Institutional Review Board (PRH1-18-132).
2. Results
The study included 1745 patients (median age: 24 years, interquartile range: 27 years) prescribed ICS only (58%), ICS/LABA only (36%) or other combinations of controller medications (7%) (Supplementary Table 1). EMMs were used for SABA inhalers for 1493 patients (83%). Mean (SD) patient adherence during baseline (weeks 2–6) and during the entire study period (weeks 2–20) was 48% (32%) and 38% (30%), respectively.
At baseline, approximately 20%, 28%, 25% and 27% of patients had optimal, moderate, sub-optimal and poor adherence, respectively. All four adherence groups experienced decline in adherence during the study (Fig. 1). The largest absolute decreases in baseline (week 2 vs. 6) adherence were observed for moderately adherent patients. Overall (week 2 vs. 20) there were larger absolute decreases in the optimal, moderate and sub-optimal groups, with the lowest change in the poor adherence group, likely due to already low adherence levels during week 2 (Supplementary Table 2). However, there were lower relative decreases in adherence overall (week 2 vs. 20) in the optimal (−28%) group than in moderate (−46%), sub-optimal (−58%) and poor (−50%) adherence groups.
Fig. 1Adherence trajectories from week 2–20, stratified by adherence phenotype.
Compared with patients in the lower adherence groups, patients with optimal adherence were on average 4.6 (95% CI: 2.6, 6.5; P < 0.01) years older (mean [SD]: 30.9 [20.0] vs. 26.3 [16.7]). The poor adherence group had an 8.1% (95% CI: 2.9%, 13.3%; P < 0.01) higher proportion of patients prescribed ICS only compared to those in the other adherence groups (63.7% vs. 55.6%). Mean weekly rescue use during the study period was lowest among patients with poor adherence (data not shown).
3. Discussion
Patients with asthma were classified into adherence subgroups determined by their adherence at baseline, during weeks 2–6. Average age differed across the adherence groups, previous studies have shown that increasing age in adult patients may be associated with improved adherence [
]. Approximately 75% of the population demonstrated suboptimal adherence as determined by our a priori and admittedly arbitrary adherence categories.
Rescue use declined in all groups over time, with lowest use among patients with poor adherence. While the decline in rescue use was not surprising across the population [
Effectiveness of population health management using the propeller health asthma platform: a randomized clinical trial. The journal of allergy and clinical immunology in practice.
], the observed low rescue use in patients poorly adherent to controllers was not expected. Low SABA use in the those poorly adherent may reflect poor perception of airway obstruction, patients with mild asthma or recently well controlled asthma who self-identified that both rescue and daily controller therapy is unnecessary, or those who received EMMs but did not attach them to their medications.
Adherence also declined across each group. While this is common [
]. Thus, our findings suggest that early intervention, especially among patients with moderate and sub-optimal adherence where the rate of decline was the greatest, might provide insight on patterns of long-term adherence and early opportunities for intervention. Additionally, we may need to consider alternative interventions in the most poorly adherent groups who may perceive either lack of need (either poor perception or actual good control), or whose poor engagement with the monitoring system is reflective of their engagement with the healthcare system as a whole.
Importantly, adherence in all groups declined over time, indicating that continued individual and population level initiatives will be key across all groups, though targeting the groups with the largest levels of decline may be desirable, especially if there are indicators of periodic loss of control. That said, more work is needed to understand how the introduction of a digital monitoring tool may result in a temporary increase in adherence. While we attempted to control for this by examining only the data after week 2, it's possible some of the effect remained beyond the initial two weeks and thus resulted in slightly inflated decreases across the subgroups. Another possibility for this decline could be regression to the mean. This may indicate a low level of interest or engagement in this group, which will be a challenge for future interventions. Initiation of medication monitoring and early intervention by providers may be key in leveraging patient engagement and improving long-term adherence, though cost effectiveness would need to be addressed and ethical considerations may include payors utilizing this information to dictate care, as well as privacy concerns.
This study is limited by a lack of clinical characterization, including other characterization of asthma control other than measured rescue use, and no measures of baseline adherence levels (prior to the initiation of monitoring) were available.
Monitoring controller therapy and identifying the respective subgroup should then prompt a more detailed assessment of overall asthma control. The next step would be to evaluate rescue therapy and other measures of asthma control, such as serial lung function, and factors that may impact asthma management, such as patient-provider communication, poor perception of airway obstruction, medication costs, or an evaluation of social determinants of health. This approach is in line with methods to encourage patient engagement and shared decision making.
Sources of support
This work was supported by Propeller Health. Propeller Health employees were involved in design, data analysis and writeup.
Funding
This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.
CRediT authorship contribution statement
Heather E.H. De Keyser: Conceptualization, Resources, Writing - original draft, Writing - review & editing. Leanne Kaye: Conceptualization, Resources, Writing - review & editing. William C. Anderson: Conceptualization, Resources, Writing - review & editing. Rahul Gondalia: Conceptualization, Formal analysis, Data curation, Writing - review & editing. Ben Theye: Formal analysis, Data curation. Stanley J. Szefler: Conceptualization, Resources, Writing - review & editing, Supervision. David A. Stempel: Conceptualization, Resources, Writing - review & editing, Supervision.
Declaration of competing interest
The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: Heather Hoch has served as a consultant to Astra Zeneca, and has received research support from Propeller Health, the Children’s Hospital Colorado Research Institute, and the Colorado Department of Public Health and Environment Cancer, Cardiovascular and Pulmonary Disease Program. William Anderson has served as a consultant to Astra Zeneca. Stanley Szefler has consulted for Astra Zeneca, Boehringer-Ingelheim, GlaxoSmithKline, Novartis, Propeller Health, Regeneron and Sanofi and has received research support from the National Institutes of Health, the National Heart, Lung and Blood Institute and the Colorado Department of Public Health and Environment Cancer, Cardiovascular and Pulmonary Disease Program. Leanne Kaye, Rahul Gondalia, Ben Theye, and David A. Stempel are employees of Propeller Health and receive compensation and stock ownership.
Appendix A. Supplementary data
The following are the Supplementary data to this article:
Effectiveness of population health management using the propeller health asthma platform: a randomized clinical trial. The journal of allergy and clinical immunology in practice.
30054-8&id=gr1.jpg){kind=link}