A Micro-longitudinal Approach to Measuring Medication Adherence in Pediatric Inflammatory Bowel Diseases

Abstract

Measuring medication adherence in pediatric inflammatory bowel diseases (IBD) is challenging because of complexities in personalized treatment regimens and increased use of biologic mono- and combination therapy. Objective measurement of adherence via electronic monitoring is the gold standard; however, it is not useful for daily monitoring when multiple medication formulations (eg, pills, injections, infusions) as well as vitamins/supplements are prescribed. Although validated subjective measures are available, they are not designed for daily use and do not capture day-to-day variation in adherence. In the following article, a new approach to measuring adherence regardless of a patient’s specific medication regimen is presented. Utilizing a micro-longitudinal design, 30 days of daily self-reported medication adherence data was collected from youth with IBD via text message. Results reflect mean adherence rates from studies utilizing pill counts and electronic monitoring, suggesting promise for the use of self-reported daily diaries to assess medication adherence in pediatric IBD.

Adolescents and young adults (AYAs) with inflammatory bowel diseases (IBD) struggle to adhere to medication with nonadherence rates ranging from 50% to 88% (1,2). Consequences include disease relapse (3), increased healthcare utilization (4), and treatment escalation (5). Nonadherence may result from the complexity of the regimen (6), which consists of multiple medications across dosing schedules and formulations (7). With the introduction and common use of injectable and intravenous biologic therapies over the past 15 years (8), fewer youth with IBD are prescribed daily oral medications alone.

Pediatric IBD adherence research has utilized objective (eg, pill counts, electronic monitoring, bioassays) and subjective measures (eg, validated measures, visual analog scales) to examine nonadherence (7,9). Although objective measures are considered the “gold standard,” they are not useful across medication regimens broadly. Pill counts and electronic pill bottles exclude medications administered via injection and infusion. Therapeutic drug levels obtained via bioassays do not provide information on the amount of medication taken, are affected by variability in pharmacokinetics and pharmacogenomics, and are not available for all pediatric IBD medications. Additionally, validated measures do not capture day-to-day variability in adherence, but rather summarize medication-taking behaviors over a designated time period (eg, 2 weeks, 30 days) (10).

Accurate and comprehensive assessment of adherence is imperative for improving clinical care so that treatment decisions are made under the most informed conditions. Micro-longitudinal designs are used to collect a large amount of data over a short time, and have been used in other pediatric populations (eg, diabetes, asthma) to measure a variety of health behaviors (11), but have not yet been utilized to measure adherence in pediatric IBD.

The present study describes feasibility of a novel, micro-longitudinal approach to measuring adherence in AYAs with IBD that can be utilized across medication regimens; reports adherence rates in AYAs with IBD utilizing a micro-longitudinal design; and compares adherence from the present study to previously published adherence obtained via objective measures.

METHODS

Procedure

Institutional review board approval was secured before initiating study procedures. Eligible individuals were identified via chart review and contacted during an IBD clinic appointment. Upon providing consent/assent, participants received a link to baseline questionnaires followed by daily diaries via text message for 30 consecutive days. Baseline surveys and daily diaries were hosted on REDCap and formatted for ease of completion via mobile phone. Participants received $20 after completing baseline questionnaires. At the end of the 30-day study period, participants received $1 for each daily diary completed, regardless of adherence, with those completing all 30 daily diaries receiving an additional $30.

In addition to recruitment, we utilized published data from 3 independent studies using objective adherence measurement in AYAs with IBD (1,12,13). Detailed procedures are described in the references cited. Greenley et al in 2016 included an adherence promotion intervention, therefore only baseline adherence data was analyzed to avoid intervention effects.

Participants

Inclusion criteria were: medical diagnosis of IBD; ages 16 to 22; access to a personal mobile phone with text messaging and internet access, and sufficient data or regular WiFi access to accommodate daily text messaging and internet use; prescribed at least 1 medication for IBD management to be taken during the 30-day study period. Exclusion criteria were lack of English fluency, cognitive impairment that would preclude questionnaire completion, or participation in other research protocols that might interfere with the present study (eg, participation in an adherence promotion intervention).

Measures

Demographic and Disease Information

Participants completed a study-developed questionnaire assessing age, sex, and race/ethnicity. Diagnosis and medication regimen were verified via chart review. Physician’s global assessment ratings were obtained via chart review along a 4-point scale from 0 (quiescent) to 4 (severe) (14).

Medication Adherence

Study-developed questions about medication adherence included how much of their IBD medications they took in the past 24 hours (Overall Adherence), and self-report of which IBD medications they were prescribed in the past 24 hours (eg, pills, vitamins/supplements, injections, infusions). For each type of medication reported, participants responded “yes” or “no” to whether they took it. If participants did not have medications to take on a given day, they selected “I did not have to take any IBD medications in the past 24 hours.”

Statistical Analyses

Descriptive analyses were conducted to summarize sample characteristics. For Overall Adherence, “all” was coded as adherent for that day. For medication-specific adherence, “yes” was coded as adherent for that medication on that day. Mean adherence was calculated for Overall Adherence and by medication type. The number of participants reporting <80% adherence (eg, adherent on less than 24 of 30 days), ≥80% adherence (eg, adherent on at least 24 of 30 days), and the number of participants reporting 100% adherence (eg, adherent on all 30 days) were reported. Nonadherence prevalence was determined by the proportion of participants reporting <80% adherence. The 80% cutoff was used based on clinical trials in pediatric and adult IBD (15). For those who did not complete all 30 days of daily diaries (n = 16), the denominator used to calculate adherence percentages was based on the number completed. For those who reported they were not prescribed daily medication, the denominator used to calculate adherence percentages was based on number of days, participants reported having medication to take.

Chi-squared tests were used to compare mean adherence and nonadherence prevalence from the present study to published data obtained via objective measures (16). Data from the present study was compared with the full samples of Hommel et al (1) and Greenley et al (13). LeLeiko et al in 2013 reported results by age group, therefore, data from the present study was compared with a subsample of youth above age 15. Nonadherence prevalence was not reported in Greenley et al (13), therefore, these analyses are only reported for Hommel et al (1) and LeLeiko et al (12).

RESULTS

Sample Characteristics

Participants’ (N = 50) were 16 to 22 years old (M[SD] = 18.86[1.69]) and 58% male (n = 29) with most identifying as European Origin/white (n = 42, 84%). Thirty-six (72%) were diagnosed with Crohn disease and 14 were diagnosed with ulcerative colitis (28%). Physician’s global assessment ratings at baseline reflected most having quiescent disease (n = 45.90%). Thirty-five (70%) were prescribed biologics, most of whom were prescribed biologic monotherapy (n = 24.48%). Three (6%) were prescribed systemic steroids.

Feasibility

Two participants (4%) did not complete any daily diaries and were excluded from further analyses. Forty-eight participants by 30 days of adherence data equaled a total of 1440 cases available for analyses with missing adherence data for 8% (n = 120). On a given day during the 30-day study period, daily diaries were completed for an average of 44 (92%) participants. On average, participants completed 27.46 of 30 daily diaries (SD = 5.89, range: 2–30) with 92% of participants completing at least 15 daily diaries. An average of 35 (73%) participants were prescribed at least 1 IBD medication on a given day during the study period. On average, participants were prescribed medications on 21.88 of the 30 days (SD = 11.38, range: 1–30) with 71% of participants reporting they were prescribed IBD medication on at least 15 days.

Overall Adherence

Mean adherence was 66.71% (SD = 40.29%). Sixty percentage (n = 29) reported taking all their IBD medications at least 80% of the time and 40% (n = 19) reported taking all their IBD medications 100% of the time (Table 1).

TABLE 1.

Mean adherence and nonadherence prevalence for previously published studies using objective measures and the present study

	Hommel et al (2009)	LeLeilco et al (2013)	Greenley et al (2015)	Current Study
Total sample size	N=42	N=79	N=76	N=48
Participant ages	13 to 17	8 to 17	11 to 18	16 to 22
Study duration	30 days	6 months	2-week baseline monitoring period	30 days
Adherence monitoring	Pill counts (30-day supply); bioassays (6-TGN and 6-MMPN)	Electronic monitoring	Electronic monitoring	Daily self-report
Medications monitored	5-ASA and 6-MP/AZA	81% 5-ASA 46% 6-MP 34% 5-ASA and 6-MP	66% Immunomodulator 45% aminosalicylate 8% corticosteroid 5% antibiotic	Overall Adherence* Pills, vitamins/supplements, injections, infusions†
Adherence calculation	(Doses consumed/doses prescribed) × 100	(Number of times the bottle was opened/number of expected openings) × 100	(Number of actual openings during specified period/number of expected openings during specified period) × 100	Overall adherence: (number of days reported taking “all” medication/number of days reported having medication prescribed)
				Pills, vitamins/supplements, injections, infusions: (doses reported as taken/doses reported prescribed) × 100
Definition of nonadherence	<80% of a single medication or subtherapeutic 6-TGN levels/unquantifiable 6-MMPN levels	<80%	Lower scores reflect lower medication adherence	<80%
Mean adherence rate	Pill counts: 5-ASA: 51% 6-MP/AZA: 62%	Full sample: 5-ASA: 71% 6-MP: 65% > 15 y.o.: 5-ASA: 54% 6-MP: 51%	83%	Overall adherence: 67% Pills: 91% Vitamins/supplements: 92% Injections: 85% Infusions: 100%
Nonadherence prevalence	Pill counts: 5-ASA: 88% 6-MP/AZA: 64% Bioassays: 36%	Full sample: 5-ASA: 45% 6-MP: 58% > 15 y.o.: 5-ASA: 75% 6-MP: 85%	–	Overall adherence: 40% Pills: 15% Vitamins/supplements: 15 % Injections: 23% Infusions: 0%

^*Overall Adherence refers to participants’ response to how much of their prescribed IBD medications they took in the past 24 hours with a response of “all” indicating adherence on that day

^†Adherence to pills, vitamins/supplements, injections, and infusions was determined by whether a participant selected that they were prescribed a given type of medication in the past 24 hours and responded “yes” or “no” to whether they took it with responses of “yes” indicating adherence on that day.

Adherence by Inflammatory Bowel Disease Medication Type

Pills

Mean adherence for the subsample of participants reporting being prescribed pills (N = 34) was 91.20% (SD = 14.42%). Of this subsample, 85% (n = 29) reported taking their pills when prescribed at least 80% of the time and 47% (n = 16) reported taking their pills when prescribed 100% of the time.

Vitamins/Supplements

Mean adherence for the subsample of participants reporting being prescribed vitamins/supplements (N = 26) was 91.86% (SD = 17.06%). Of this subsample, 85% (n = 22) reported taking their vitamins/supplements when prescribed at least 80% of the time and 62% (n = 16) reported taking their vitamins/supplements when prescribed 100% of the time.

Injections

Mean adherence for the subsample of participants reporting being prescribed injections (N = 26) was 85.25% (27.10%). Of this subsample, 77% (n = 20) reported taking their injections when prescribed at least 80% of the time and 69% (n = 18) reported taking their injections when prescribed 100% of the time.

Infusions

All participants reporting having an infusion scheduled during the study period (n = 9) reported receiving their infusion (M[SD] = 100%[0%]), and therefore, were not compared to previously published oral medication adherence rates (Table 1).

Adherence Compared to Objective Measures

Mean Adherence

Overall Adherence

Mean overall adherence in the present study was statistically similar to mean adherence obtained via pill count and electronic monitoring in 2 of the 3 comparison studies, respectively (Table 2).

TABLE 2.

Mean adherence rates and nonadherence prevalence compared to previously published studies using objective measures

	Hommel et al (2009)				LeLeiko et al (2013)				Greenley et al (2015)
Mean adherence rates	MD	95% CI	χ2	P	MD	95% CI	χ2	P	MD	95% CI	χ2	P
Overall
5-ASA	16%	−4.16%, 34.61%	2.35	0.125	13%	−9.85%, 35.32%	1.14	0.286
6-MP	5%	−14.25%, 24.04%	0.24	0.623	16%	−11.54%, 42.76%	1.11	0.292
Various									16%	0.70%, 31.67%	4.19	0.040
Pills
5-ASA	40%	21.64%, 55.49%	17.71	<0.001	37%	15.72%, 56.80%	12.87	<0.001
6-MP	29%	11.65%, 45.00%	10.67	0.001	40%	13.75%, 64.69%	11.14	<0.001
Various									8%	−5.28%, 19.31%	1.56	0.212
Vitamins/supplements
5-ASA	41%	22.81%, 56.36%	18.83	<0.001	38%	16.87%, 57.69%	13.90	<0.001
6-MP	30%	12.83%, 45.87%	11.62	<0.001	41%	14.87%, 65.60%	12.14	<0.001
Various									9%	4.04%, 20.12%	2.02	0.155
Injections
5-ASA	34%	14.84%, 50.36%	12.00	<0.001	31%	9.02%, 51.48%	8.02	0.005
6-MP	23%	4.80%, 39.84%	6.13	0.013	34%	7.18%, 59.23%	6.73	0.010
Various									2%	−12.37%, 14.46%	0.09	0.770
Nonadherence prevalence	MD	95% CI	χ2	P	MD	95% CI	χ2	P
Overall
5-ASA	48%	28.75%, 62.31%	21.75	<0.001	35%	10.61%, 53.12%	7.74	0.005
6-MP	24%	3.33%, 41.95%	5.11	0.024	45%	14.71%, 61.64%	8.15	0.004
Bioassay	4%	−15.68%, 23.02%	0.15	0.698
Pills
5-ASA	73%	54.78%, 83.06%	47.23	<0.001	60%	36.42%, 74.99%	24.97	<0.001
6-MP	49%	29.29%, 63.79%	22.61	<0.001	70%	40.35%, 83.17%	23.51	<0.001
Bioassay	21%	3.02%, 37.86%	5.24	0.020
Vitamins/supplements
5-ASA	73%	54.78%, 83.06%	47.23	<0.001	60%	36.42%, 74.99%	24.97	<0.001
6-MP	49%	29.29%, 63.79%	22.61	<0.001	70%	40.35%, 83.17%	23.51	<0.001
Bioassay	21%	3.02%, 37.86%	5.24	0.020
Injections
5-ASA	65%	46.08%, 76.76%	37.60	<0.001	52%	27.88%, 68.18%	17.73	<0.001
6-MP	41%	20.64%, 56.99%	15.27	<0.001	62%	31.92%, 76.50%	16.74	<0.001
Bioassay	13%	−5.68%, 30.92%	1.82	0.178

Results from chi-squared tests comparing mean adherence and nonadherence prevalence from the present study are presented (see Table 1). As Hommel et al in 2009 and LeLeiko et al in 2013 reported mean adherence and nonadherence prevalence for 5-ASA and 6-MP, data from the present study was compared with both medications. Adherence data from the present study was compared with data from the various medications measured in Greenley et al (2015). Last, data from the present study was also compared with nonadherence prevalence based on bioassay results from Hommel et al (2009).

Bolded results represent cases in which data mean adherence or nonadherence prevalence was statistically similar to data from previously published studies of pediatric IBD adherence using objective measures. CI = confidence interval; MD = mean difference.

Pills, Vitamins/Supplements, Injections

Mean oral medication, vitamin/supplement, and injection adherence in the present study was statistically similar to the mean oral medication adherence obtained via electronic monitoring in 1 of the 3 comparison studies (Table 2).

Nonadherence Prevalence

Overall Adherence

Nonadherence prevalence measured in the present study was significantly different from nonadherence prevalence via pill count and electronic monitoring in each of the comparison studies, respectively. Nonadherence prevalence in the present study was statistically similar to the nonadherence prevalence determined via bioassays (Table 2).

Pills, Vitamins/Supplements, Injections

Nonadherence prevalence for oral medication, vitamins/supplements, and injections measured in the present study were significantly different from nonadherence prevalence via pill count and electronic monitoring in each of the comparison studies, respectively. When compared with nonadherence prevalence via bioassay, nonadherence prevalence for oral medication and vitamins/supplements were statistically different whereas nonadherence prevalence for injections was statistically similar. See Table 2.

DISCUSSION

To our knowledge, this is the first study to pilot a micro-longitudinal design and collect daily adherence data via text messaging in pediatric IBD. The micro-longitudinal approach yielded minimal missing data, suggesting preliminary feasibility. Second, mean adherence rates collected via an overall adherence question were statistically similar to those obtained in previous pediatric IBD adherence studies via pill counts and electronic monitoring. Additionally, mean adherence rates for specific medication types were statistically similar to a mean adherence rate from a previous study using electronic monitoring to capture adherence across a variety of oral IBD medications. Last, nonadherence prevalence collected via an overall adherence question and an injection-specific question was statistically similar to nonadherence prevalence obtained via bioassays.

Utilizing a micro-longitudinal approach to measuring daily adherence across regimens in pediatric IBD shows promise; however, specific aspects of this methodology may be more or less appropriate given the adherence behavior of interest. Measuring adherence to primary oral maintenance medications may be better captured by an overall adherence question while formulation-specific questions may be better for assessing other medications, including antibiotics, steroids, vitamins/supplements, or injections. Additional benefits of this methodology include capturing relationships within and between everyday health behaviors, reducing recall bias, and allowing for between- and within-subjects analyses (17). Further, this approach may be particularly useful in identifying patterns of nonadherence for personalized behavioral intervention delivery. Future adherence and self-management research, including clinical trials involving multiple medications across varying dosing schedules, may consider utilizing these methods for estimations of adherence as well.

Regarding limitations, although statistically similar in some cases, adherence in the present study was consistently higher than in the comparison studies. This may reflect a generally adherent sample, which may be associated with low rates of missing data. Due to the pilot nature of this work, objective measures were not included in the present study offering opportunities for future studies to further validate these methods. Receiving a daily text message, while not designed as an intervention, may have served as a medication reminder, although the same could be said for utilizing an electronic pill bottle or expecting a pill count during a study visit. Last, payment contingent on completing daily diaries may not be adaptable for nonresearch contexts; however, was necessary to collect as much data as possible to examine the utility of this approach. Although the comparison studies utilized passive adherence monitoring via objective measures, 2 of the 3 of our comparison studies published their participant compensation information [eg, $25 gift cards (1), up to $90 (13)]. Future researchers utilizing this methodology may consider recruiting youth struggling with adherence and testing this approach with different incentive structures for diary completion with the goal of refining this methodology for eventual integration into clinical practice without monetary incentives. Last, larger studies using this methodology are needed to detect whether demographic covariates, medication regimen, or other daily health behaviors are associated with adherence behavior.

Overall, using daily diaries to assess medication adherence via a micro-longitudinal design is feasible in pediatric IBD. More research is needed to validate this novel methodology against objective adherence monitoring techniques and with respect to short- and long-term health outcomes.

What Is Known

• Medication nonadherence in youth with inflammatory bowel disease is poor, in part because of the complexity of the medication regimen.

• Existing objective and subjective measures of adherence in pediatric inflammatory bowel disease are not adequate for capturing adherence across regimens or day-to-day variability in adherence behavior.

What Is New

• Micro-longitudinal design utilizing self-reported daily diaries via text message is a novel, and feasible method of measuring medication adherence.

• Self-reported adherence data obtained via daily diaries may be appropriate for assessing adherence to complex medication regimens in youth with inflammatory bowel disease.