Medication Management and Caregiving Stress or Benefit in Children with Medical Complexity

Abstract

This study quantified medication management experiences for children with medical complexity and associations with caregiver stress or benefit. Using data from the MedS@HOME trial, involvement of professional caregivers was associated with increased caregiver benefit, and greater network collective efficacy was associated with increased caregiver benefit and decreased caregiver stress.

Trial registration

ClinicalTrials.gov, ID NCT05816590.

Caregivers of children with medical complexity (CMC) confront a range of positive and negative experiences navigating substantial daily care demands.1, 2, 3, 4, 5 Recent research suggests that sharing caregiving responsibilities with others can both promote caregiver benefit (ie, perception of beneficial psychological and social impact of caregiving on the caregiver)⁶ and minimize negative consequences formed by systematic barriers facing this population.^5,7,8

One specific aspect of care of CMC, medication management, is particularly challenging, and errors have been tied to harm.9, 10, 11, 12 The complexity of medication management at home is underscored in an American Academy of Pediatrics policy statement, which contains 24 distinct recommendations for safe medication administration.¹³ Despite growing evidence about the impacts of caregiving for CMC, the role of medication management in positive and negative caregiving experiences remains poorly understood.

As part of a larger clinical trial to improve medication safety for CMC,¹⁴ our objectives in this study were to quantitatively describe medication management and caregiving network collective efficacy and to determine associations with caregiver stress or benefit. We hypothesized that assistance from professional caregivers and greater perceived collective efficacy of caregivers involved in the care of CMC would be associated with lower caregiver stress and higher caregiver benefit.

Methods

Study Design, Setting, and Population

This secondary analysis used cross-sectional baseline data from the MedS@HOME randomized controlled trial,^14,15 which enrolled CMC who receive care at the University of Wisconsin Health American Family Children's Hospital (Figure). Eligible participants were self-identified primary caregivers of children with ≥2 complex chronic conditions (CCCs),¹⁶ taking ≥1 high-risk medication, and ≤16 years old. High-risk medications were defined from prior literature^10,11,17, 18, 19 as medications with a serious risk of consequences for incorrect or missed doses or drug-drug interactions (antiarrhythmics, anticoagulants, antiepileptics, immunosuppressants, opioids, psychotropics, pulmonary hypertension medications, sleep aids, stimulants, and tone or spasticity medications). Caregivers were aged ≥18 years, were English-speaking, and had access to a mobile device. Data for this study were obtained from trial enrollment surveys and electronic health records.

Figure.

MedS@HOME study consort diagram through baseline data collection.

Outcomes, Exposures, and Confounders

Outcomes were caregiver stress and benefit, measured by the University of Washington's Caregiver Stress Scale 6+2-Item Short Form V.2.0 and the Caregiver Benefit Scale 6-Item Short Form V.2.0. These measures have evidence of reliability and validity^2,20 and were developed with caregivers of CMC to quantify the stress/benefit caregivers attribute to caring for their child.² Population norms were established with a sample of US caregivers of children lacking serious health conditions.² Caregivers use 5-point Likert scales to describe how they usually feel (“not at all” to “very much”). Responses were scored according to the developer user guide,^6,21 in which nonmissing summary scores are converted to T-scores, centered on a mean of 50 and an SD of 10. To facilitate interpretation, we categorized respondent scores ≥1 SD below, within 1 SD of, or ≥1 SD above population means. Higher scores corresponded to greater stress/benefit.

Exposures included medication administration indicators (number of medications and number of people giving medications). Caregiving experience was evaluated with the Perceived Collective Efficacy Scale,²² in which caregivers used 5-point Likert responses (“strongly disagree” to “strongly agree”) to rate agreement with 4 statements about how well their child's caregivers work together. Higher summed scores (range, 4-20) indicated greater collective efficacy. Confounders selected a priori by the study team included demographic variables (child insurance, caregiver education, and household income) and severity of illness variables determined through survey data (number of medications and medication deliverers) and electronic health record data (CCC and prior-year emergency department visits).23, 24, 25, 26 Medication adherence, determined in the past research to be associated with caregiver experience,²⁷ was assessed using the 12-item Adherence to Refills and Medications Scale (ARMS)²⁸ with 4-point Likert responses. Higher summed scores (range, 12-48) indicated better adherence.

Statistical Analyses

We calculated descriptive statistics, followed by bivariate and multivariate multinomial regression models to identify associations between exposures and the benefit or stress scores (≥1 SD below the population mean, within 1 SD, or ≥1 SD above). Two models were used, focusing on either benefit or stress. We adjusted for confounders with bivariate P values <.10 and used benefit or stress scores within 1 SD of the population mean as the reference for comparisons. All analyses were conducted using Stata (version 19.0; StataCorp LLC). P values <.05 were considered statistically significant. This study was approved by the UW Institutional Review Board (#2022-1620), registered on clinicaltrials.gov (#NCT05816590), and followed the Strengthening the Reporting of Observational studies in Epidemiology (STROBE) reporting guidelines for cross-sectional studies.²⁹

Results

Child, Caregiver, Medication Administration, and Caregiving Network Characteristics

The 167 CMC enrolled (Table I) had a mean age of 9.7 years, were mostly (76%) non-Hispanic White, took a median (IQR) of 4 (3-7) scheduled medications, and 63 (38%) had ≥4 CCCs. Enrolled caregivers had a median (IQR) age of 41 (36-47) years and were mostly women (84%). On average, 2.7 (SD, 1.3) different individuals regularly administered medications, which could include self-administration by CMC. Professional caregivers administered medications to 31% of CMC. The median (IQR) collective efficacy score was 19.5 (16-20) on a 4-20 scale. The median (IQR) ARMS score was 45 (44-47) on a 12-48 scale.

Table I.

Characteristics of caregivers, children, medication administration, and caregiving networks in the MedS@HOME trial

Characteristic	Statistics
Child characteristics
Age, y, n (%)
0-5	33 (19.8)
6-10	56 (33.5)
11-15	63 (37.7)
≥16	14 (8.4)
Missing	1 (0.6)
Sex, n (%)
Female	74 (44.3)
Male	91 (54.5)
Missing	2 (1.2)
Race and ethnicity, n (%)
Asian, non-Hispanic	11 (6.6)
Black, non-Hispanic	7 (4.2)
Hispanic	13 (7.8)
Other, non-Hispanic	4 (2.4)
White, non-Hispanic	126 (75.5)
Missing	6 (3.6)
Primary insurance, n (%)
Private	56 (33.5)
Public	58 (34.7)
Private and public	51 (30.5)
Missing	2 (1.2)
CCCs, n (%)
2	72 (43.1)
3	32 (19.2)
≥4	63 (37.7)
Missing	0 (0.0)
ED visits, year before enrollment, n (%)
0	100 (59.9)
1	37 (22.2)
≥2	30 (18.0)
Missing	0 (0.0)
Hospitalizations, year before enrollment, n (%)
0	106 (63.5)
1	36 (21.6)
≥2	25 (15.0)
Missing	0 (0.0)
Caregiver characteristics
Age, median (IQR)	41 (36-47)
Sex, n (%)
Man	25 (15.0)
Woman	140 (83.8)
Nonbinary	1 (0.6)
Missing	1 (0.6)
Education, n (%)
High school or GED	19 (11.4)
Some college	62 (37.1)
College	48 (28.7)
Advanced degree	34 (20.4)
Missing	4 (2.4)
Household income, n (%)
≥$100 000	69 (41.3)
$50 000-99 999	45 (27.0)
$35 000-49 999	14 (8.4)
<$35 000	24 (14.4)
Missing	15 (9.0)
Medication administration and caregiving network characteristics
Number scheduled medications per day, median (IQR)	4 (3-7)
High-risk medication categories, n (%)
Antiarrhythmics	30 (18.0)
Anticoagulants	8 (4.8)
Antiepileptics	70 (41.9)
Immunosuppressants	36 (21.6)
Opioids	2 (1.2)
Psychotropics	49 (29.3)
Pulmonary hypertension medications	55 (32.9)
Sleep aids	12 (7.2)
Stimulants	19 (11.3)
Tone or spasticity medications	15 (9.0)
High-risk medication number, n (%)
1	86 (51.5)
2	50 (29.9)
≥2	31 (18.6)
Locations where medications are delivered, n (%)
Community	21 (12.6)
Extended family or friends' homes	31 (18.6)
Home	165 (98.8)
Respite	10 (6.0)
School	18 (10.8)
Number people giving child medications, median, (IQR)	2 (2-3)
Child self-delivers medications, n (%)	42 (25.2)
Grandparent delivers medications, n (%)	51 (30.5)
Professional caregivers deliver medications, n (%)	51 (30.5)
Caregiving network collective efficacy,∗ median (IQR)	20 (16-20)
ARMS,† median (IQR)	45 (44-47)
Caregiver benefit scale, n (%)
>1 SD above population mean	40 (24.2)
Within 1 SD of the population mean	90 (54.6)
>1 SD below population mean	35 (21.2)
Caregiver Stress Scale, n (%)
>1 SD above population mean	31 (18.8)
Within 1 SD of the population mean	101 (61.2)
>1 SD below population mean	33 (20.0)

ED, emergency department; GED, general educational development.

^∗Caregiving network collective efficacy is a summed scale score of 4 items having 5-point Likert responses (range, 4-20), with higher scores reflecting greater collective efficacy.

^†ARMS is a summed scale score of 12 items having 4-point Likert responses (range, 12-48), with higher scores reflecting greater adherence.

Associations with Caregiver Benefit and Stress

About 1 in 5 (21.2%) had caregiver benefit scores at least 1 SD below the population mean. Similarly, 1 in 5 (20.0%) had caregiver stress scores at least 1 SD above the population mean (Table I). In adjusted models, involvement of professional caregivers was associated with a higher relative risk of reporting caregiver benefit ≥1 SD above the population mean (relative risk reduction [RRR], 4.40; 95% CI, 1.36-14.23). Higher collective efficacy was also associated with a higher relative risk of reporting benefit ≥1 SD above the population mean (1.42; 1.10-1.82) and a lower relative risk of reporting caregiver stress ≥1 SD above the population mean (0.61; 0.45-0.82). Higher ARMS scores were associated with a lower relative risk of reporting stress ≥1 SD above the population mean (0.74; 0.58-0.93) (Table II).

Table II.

Medication administration and caregiver experience associations with caregiver stress and benefit in adjusted^∗ models

Characteristic	Caregiver stress RRR (95% CI)		Caregiver benefit RRR (95% CI)
	≥1 SD below population mean vs within 1 SD of population mean	≥1 SD above population mean vs within 1 SD of population mean	≥1 SD below population mean vs within 1 SD of population mean	≥1 SD above population mean vs within 1 SD of population mean
Number scheduled medications per day	0.96 (0.78-1.17)	1.08 (0.94-1.24)	-	-
High-risk medication number
1	1 [reference]
2	0.27 (0.07-1.09)	2.63 (0.63-10.92)	-	-
≥3	0.81 (0.13-4.95)	4.01 (0.72-22.23)	-	-
Child self-delivers medications
Yes (vs no)	1.01 (0.97-1.05)	1.04 (1.01-1.07)	-	-
Professional caregivers deliver medications
Yes (vs no)	0.27 (0.05-1.53)	1.21 (0.33-4.41)	1.88 (0.63-5.65)	4.40 (1.36-14.23)
Medication delivery locations
Community, yes (vs no)	-	-	0.50 (0.09-2.83)	1.74 (0.45-6.70)
Adherence to refills and medications scale	1.43 (1.08-1.89)	0.74 (0.58-0.93)	-	-
Caregiving network collective efficacy scale	1.49 (1.02-2.18)	0.61 (0.45-0.82)	0.96 (0.79-1.17)	1.42 (1.10-1.82)
Child primary insurance
Private	1 [reference]
Public	-	-	0.72 (0.19-2.75)	1.26 (0.37-4.36)
Private and public	-	-	0.41 (0.14-1.16)	0.07 (0.02-0.32)
Caregiver education
College	1 [reference]
Advanced degree	0.25 (0.05-1.27)	22.08 (2.96-164.56)	-	-
Some college	0.72 (0.20-2.65)	7.97 (1.29-49.28)	-	-
High school or GED	0.66 (0.09-4.96)	8.12 (0.63-105.33)	-	-
Household income
≥$100 000	1 [reference]
$50 000-99 999	0.80 (0.23-2.77)	0.53 (0.11-2.53)	0.29 (0.09-0.96)	2.97 (0.91-9.62)
$35 000-49 999	1.36 (0.16-11.59)	1.71 (0.14-21.08)	0.26 (0.04-1.47)	0.35 (0.06-2.18)
<$35 000	0.94 (0.11-8.38)	3.05 (0.44-20.85)	0.07 (0.01-0.69)	0.46 (0.09-2.30)
Missing	-	1.46 (0.14-14.83)	1.26 (0.23-6.82)	4.22 (0.66-26.83)
CCCs
2	1 [reference]
3	1.03 (0.28-3.74)	1.48 (0.18-12.47)	-	-
≥4	0.53 (0.10-2.70)	9.12 (1.85-45.02)	-	-
ED visits, year before enrollment
0	1 [reference]
1	0.39 (0.09-1.64)	1.20 (0.24-5.98)	2.12 (0.81-5.56)	0.34 (0.08-1.37)
≥2	1.44 (0.19-10.84)	0.51 (0.07-3.80)	0.66 (0.15-2.88)	0.72 (0.22-2.30)

ED, emergency department; GED, general educational development.

^∗Models adjusted for all variables, with P < .10 in univariable models. Variables not included in caregiver stress or benefit models are represented with a dash.

Discussion

In this cross-sectional study using baseline data from the MedS@HOME randomized clinical trial,¹⁴ we observed that caregivers of CMC with professional caregivers reported greater caregiver benefit. Additionally, higher network collective efficacy was associated with greater caregiver benefit and lower caregiver stress. Our analysis is among the first to evaluate quantitative links with medication management.

Managing medications for CMC is highly complicated work.9, 10, 11 We were, therefore, not surprised that professional caregiver involvement was associated with increased caregiver benefit. Despite evidence of professional caregiver involvement improving health (eg, emergency department visits) and financial outcomes for CMC,^7,30 parent caregivers frequently face challenges in accessing these services.^7,30,31 Consistent with these studies, we observed relatively infrequent access to professional caregivers in our cohort. Providers should use clinical interactions as opportunities to provide families with resources in obtaining in-home professional caregiving support.

Including caregiving networks in general CMC research is increasingly recognized as necessary to best understand outcome variation.^8,32 Our study adds to this body of work by demonstrating how network collective efficacy correlates with caregiver outcomes. This finding aligns with past research showing that communal coping is linked to increased caregiver benefit³³ and that social support is linked to less burnout and stress.³⁴ Since individual self-efficacy is important for resilience among CMC caregivers,⁵ our study findings suggest that caregiving network's collective efficacy might similarly contribute to CMC caregiver resilience. Creating programs to foster interpersonal relationships may encourage the development of caregiving networks in complementary ways. Network interventions such as MedS@HOME might influence collective efficacy as a mechanism for improving health outcomes, illustrating a potential future network research opportunity.

Our study has important limitations. Due to the nature of the study design, causality and the direction of associations cannot be assumed. Nonresponse bias is relevant, as caregivers with less time may be less likely to participate in research while being more likely to face higher stress, biasing results toward smaller effect sizes.^35,36 As a single-site study with demographics reflecting the regional population (mostly White, non-Hispanic caregivers), generalizability may be limited. Results may differ in more racially and ethnically diverse communities. While caregiving collective efficacy may experience ceiling effects, detecting associations between collective efficacy and the primary outcomes suggests that important relationships may exist. An ongoing study with a larger, more inclusive sample may yield greater variation. Additionally, the study was powered for the clinical trial outcomes and may be underpowered to detect some relationships in this analysis. Variables with weak relationships in this study may still benefit from ongoing exploration in future studies. Future studies also would benefit from including qualitative data to contextualize caregivers’ perspectives of factors driving stress/benefit.

Despite these limitations, our study offers novel data correlating caregiver stress and benefit with caregiving supports. Interventions that promote access to professional caregivers to assist with medication delivery and high network collective efficacy may be promising future directions to support caregivers of CMC.

CRediT authorship contribution statement

Casey O'Hare: Writing – review & editing, Writing – original draft, Investigation, Formal analysis, Data curation, Conceptualization. Nicole Werner: Writing – review & editing, Supervision, Project administration, Funding acquisition, Conceptualization. Makenzie Morgen: Writing – review & editing, Investigation, Data curation, Conceptualization. Sophie Kooiman: Writing – review & editing, Project administration, Methodology, Data curation, Conceptualization. James Feinstein: Writing – review & editing, Supervision, Formal analysis, Conceptualization. Michelle A. Chui: Writing – review & editing, Supervision, Conceptualization. Gemma Warner: Writing – review & editing, Supervision, Project administration, Data curation, Conceptualization. Anna Jolliff: Writing – review & editing, Methodology, Conceptualization. Brittany Storhoff: Writing – review & editing, Data curation, Conceptualization. Kristan Sondegren: Writing – review & editing, Investigation, Data curation, Conceptualization. Danielle Gerber: Writing – review & editing, Methodology, Conceptualization. Jens Eickhoff: Writing – review & editing, Formal analysis, Conceptualization. Ryan J. Coller: Writing – review & editing, Writing – original draft, Supervision, Software, Project administration, Methodology, Investigation, Funding acquisition, Formal analysis, Data curation, Conceptualization.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this article. This research was funded by the Agency for Healthcare Research and Quality (R18HS028409), grant UL1TR000427.