High Parental Concern in Children With Medical Complexity: An Early Indicator of Illness

The study involved CMC aged between 1 and 21 years and their parents who were enrolled during hospitalization, in the randomized pilot trial, to use MyChildCMC at home for 3 months postdischarge from a 289-bed, tertiary care children’s hospital in the US Intermountain West region, to which CMC comprise 30% of admissions and 53% of hospital charges.

This was a secondary analysis of data collected from CMC (and primary caregivers) who participated in the pilot trial, which assessed the feasibility of use and preliminary impact of MyChildCMC.²¹ 

Participants were parents of children who met the following criteria:

1. CMC, defined as children with multisystem, chronic conditions characterized by having ≥3 organ systems affected, ≥3 health care providers involved in their care, ≥2 hospital admissions in the previous year, and technology dependence;

2. CMC between 1 and 21 years of age;

3. parent owns a smartphone or tablet with Internet access; and

4. parent speaks English.

The exclusion criteria included critically ill CMC (because our pilot trial main objective was our app feasibility, we did not want to enroll CMC during a critical hospitalization stage when parents were likely highly stressed), parents of CMC not discharging to home, non-English–speaking parents, and families without a smartphone/tablet with Internet access. The University of Utah institutional review board approved the study.

The pilot study was conducted between June 10, 2019, and December 19, 2019. Each morning, research coordinators contacted the hospitalist attending physicians to identify CMC admissions and potential study participants, and reviewed the medical records of potential participants to confirm eligibility. Participants were consented by research coordinators during hospital admissions, enrolled in the study, and randomized to either monitor their children’s health with MyChildCMC (intervention group) or usual care (control group), which included encouragement to monitor their child’s vital signs and symptoms on paper. Parents used the app (or paper) to monitor their child’s health daily for 3 months postdischarge. Participants randomized to MyChildCMC received help to download the app and training on its use. This current study is a secondary analysis of the intervention group data.

After enrollment, parents randomized to MyChildCMC completed a sociodemographic questionnaire and received daily prompts to use the app to record child’s vital signs (temperature, heart rate, respiratory rate, oxygen saturation), symptoms (pain, seizure status, fluid intake/feeding, mental status) and oxygen requirement (increased yes/no from baseline), and received real-time feedback about the child’s health status from the app.

Sociodemographic information (of the parent or primary caregiver) included: race/ethnicity, marital status (married, divorced, never married, widowed, unmarried couple, unknown), education level (grades <9, grades 9–11, grade 12–general educational development test, college 1–3 years, college ≥4 years, masters, doctoral), employment status (unemployed, employed, unknown), household income (<$25K, $40–49K, $50–74K, $75–99K, $100–124K, $125–149K, >$150K), and household parental structure (single, biological, adoptive, combined, other).

MyChildCMC app supports the home monitoring of CMC’s vital signs and symptoms that often lead to ED/hospital admissions, and promotes early detection of warning signals of acute health changes.^21,22  Temperature, heart rate, and respiratory rates were measured and recorded daily. Parents documented their child’s level of pain by using the Numeric Rating Scale,²³  a 1-item parent proxy scale with scores categorized as mild (0–3), moderate (4–7), and severe (8–10) pain. Feeding/fluid intake (1-item) is recorded as 25%, 50%, 75%, or 100% of usual food and fluid intake. Mental status is measured using the behavior (1-item) question of the pediatric early warning scores,²⁴  scored as: (1) playing/appropriate, (2) sleepy, (3) irritable/agitated, and (4) lethargic/unresponsive. Seizures are recorded as (1) fewer than usual, (2) usual, (3) more than usual, and (4) much more than usual. Parents indicated whether the child’s health that day was typical or “normal,” because CMC’s normal baselines may be classified as abnormal in a typically developing child. Lastly, parents completed a daily question indicating their level of concern about the child’s health status, on a 4-point Likert scale (1 = none, 2 = mildly, 3 = moderately, 4 = very concerned). Data collected via MyChildCMC are stored behind the firewall within a secured Health Insurance Portability and Accountability Act-compliant database.

To facilitate analysis and on the basis of data distribution of study participants, we grouped race/ethnicity into 3 categories (non-Hispanic white, Hispanic, and Asian American), and dichotomized sociodemographic variables: education level (≥4 years of college versus not), household income (>$75K versus not), employment status (employed versus not), marital status (married versus not) and parental household structure (biological parent household versus not). Similarly, we condensed food/fluid intake into 25% to 50% and 75% to 100% of usual food/fluid intake. On the basis of uneven distribution of the dependent variable parental concern scores, we condensed scores into 2 groups: high concern (scores 3–4) and low concern (scores 1–2).

We conducted univariate analysis to describe associations between individual factors and parental concern using mixed-effects logistic regression. We included variables with a P value <.20 into a multivariable mixed-effects logistic regression (backward elimination) with stopping rule set at P value ≤.05, to explore factors associated with parental concern. We included parent as a random effect and applied sine/cosine to time to account for the time of the day effects on measurements. Similar to use of sine/cosine transformations to address the seasonality effects, when subjects are measured repeatedly across different times of the day, sine/cosine transformations of time are used to account for potential effect of time of the day (eg, night, morning, evening, etc).²⁵  We reported results as odds ratios (ORs) and 95% confidence intervals (CIs) with P values. All analyses were conducted with Stata/IC version 16.1.

We analyzed 1223 daily measurements from 24 CMC/caregiver dyads who were randomized to the intervention group (use of MyChildCMC), including 113 (9.2%) instances of high parental concern. The distribution of parental concern scores was: 41.7% (n = 581 instances in 19 patients) for score 1, 43.3% (n = 529 instances in 18 patients) for score 2, 7.4% (n = 91 instances in 17 patients) for score 3, and 1.8% (n = 22 instances in 9 patients) for score 4. The overall median age of enrolled children was 5 years (range 1–17), and 37.5% (9 of 24) were female. Table 1 describes the demographic and clinical conditions of the CMC. On average, parents use the daily app on 78 of 91 possible days (median, 89; range, 13–91).

In the univariate analysis (Table 2), we found that a child living in a biological parent household was the only sociodemographic variable associated (OR, 9.51; 95% CI, 1.29–70.19; P = .03) with parental concern. Yet, several clinical factors of the CMC were significantly associated with parental concern, including: increased temperature (OR, 1.39; 95% CI, 1.11–1.74; P < .01), increased respiratory rate (OR, 1.02; 95% CI, 1.00–1.04; P = .04), increased oxygen requirement (OR, 22.46; 95% CI, 8.17–61.72; P < .01), moderate to severe pain (OR, 7.29; 95% CI, 4.11–12.92; P < .01), reduced intake (OR, 33.42; 95% CI, 8.37–133.45; P < .01) of usual food/fluid to 25% to 50%, worsened mental status (OR, 4.88; 95% CI, 2.84–8.38; P < .01), and increased seizures (OR, 2.62; 95% CI, 1.31–5.23; P < .01). Conversely, a parent report of a child having a typical day was associated with low concern (OR, 0.06; 95% CI, 0.03–0.14; P < .01).

After multivariable mixed-effects logistic regression analysis, clinical factors significantly associated with parental concern were increased pain (OR, 5.10; 95% CI, 2.53–10.29; P < .01), increased oxygen requirement (OR, 28.91; 95% CI, 10.07–82.96; P < .01), reduced nutrition/fluid intake (OR, 71.58; 95% CI, 13.01–393.80; P < .01) of usual food/fluid to 25% to 50%, and worsened mental status (OR, 2.15; 95% CI, 1.10–4.17, P = .02). We found no association with parent/caregiver related sociodemographic variables.

Our analysis of data gathered during the MyChildCMC pilot study reveals that having a child with moderate to severe pain, increased oxygen requirements, reduced food/fluid intake, and worsened mental status was associated with higher odds of having high parental concern. Although a caregiver being a biological parent was associated with high parent concern in the univariate analysis, no sociodemographic factor was significant in multivariate analysis.

Our findings suggest that health care providers (HCPs) might more thoroughly explore levels of parental concern for their CMC during routine and unplanned clinical encounters as a means to better render care in outpatient rather than ED or hospital settings. Although many parents of CMC believe that hospitalizations are largely unavoidable, Nelson et al emphasize the importance of parental self-efficacy in decisions to seek emergency care.²⁶  Remote monitoring for CMC and their families might promote parental caregiving self-efficacy. Our findings highlight the importance of including parental concern in remote monitoring as a possible early indicator of acute changes of CMC’s condition. Mobile apps or devices that support remote monitoring have the potential to extend health care beyond regular clinical encounters, improve communication between HCPs and parents, and improve CMC outcomes.²⁷ 

Meier et al reported that about one-third of ED visits for CMC with tracheostomies may be avoidable.²⁸  A key priority for CMC is ongoing assessment and early identification of changes in health and technology status with home and community-based care programs, and timely interventions to reduce avoidable ED and hospital visits. Strategies to reduce these visits include enrolling in a medical home,^29,30  allowing real-time access to knowledgeable ambulatory HCPs with expedited appointments,³¹  reducing medical technology complications,^10,32  using interdisciplinary ambulatory teams of HCPs familiar with CMC’s unique needs,²⁸  using telemedicine,^33,34  and implementing chronic care management and complex care coordination programs with home visiting and individualized, goal-oriented care plans.^30,35–38  Despite the impact of these interventions, none has specifically addressed parental concern as a driver of ED/hospital use among CMC. Current health care systems tend not to provide families with all the necessary tools and training needed to sustain CMC care at home. Even parents with caregiving confidence and competency lack resources to care for CMC at home. As a possible solution, adult health care services are developing “Hospital at Home” models, offering nursing and increased resources during acute intervals of need. The MyChildCMC app is an inexpensive, real-time approach using home monitoring of vital signs, symptoms, and parental concern as markers of acute illness that might lead to ED/hospital use. Implementing such tools may require restructuring the health care systems to meet patients’ needs without undue burden to parents.

It is invaluable that parents of CMC recognize and report changes in their child’s health status. Although parents may struggle to objectively describe clinical changes, they can clearly express their high concern. Yet, parents can be frustrated when their concerns are not acknowledged, or when they are made to feel as though they are overly concerned. Our findings suggest that high parental concern may accompany objective physiologic changes in their child’s baseline health status. Our study also suggests that, since parents were prompted to simultaneously record vital signs/symptoms of their CMC and their level of concern, changes in clinical parameters were associated with reporting of high parental concern. Although subtle changes in a child’s health may be difficult for parents to describe, we see that high concern, especially when it is the primary parent complaint, may indicate that the child’s health is deteriorating and there is an opportunity for intervention. Similarly, parental concern has been used to prompt evaluation and recognition of sepsis in hospitalized children.³⁹  Lim et al demonstrated that providing tools for parents to track their child’s health at home and to recognize serious issues increased parental confidence and reduced health-seeking behaviors and ED visits for self-limiting illnesses during the coronavirus disease 2019 lockdown.⁴⁰  Remote monitoring may support highly valued parent–provider partnerships in the care of CMC.⁴¹  Our near-term plan is to conduct a larger randomized controlled trial to determine the actual impact of the app and identify its potential for improving care at home.

Although our study was not designed to determine the association of parental concern with ED and hospital visits, findings from previous studies in children with and without chronic conditions justify efforts to evaluate drivers of parental concern. For instance, parents who perceive urgency for care and advantages of ED visits (faster and more efficient care) are predicted to display ED care-seeking behaviors for children.⁴²  Socioemotional factors, such as seeking reassurance and having excessive fear of bad consequences with inability to assess severity of the child’s underlying illness, were associated with ambulance use for nonemergent problems.⁴³  Associations between maternal concern for the health of children with chronic diseases and acute health care utilization was associated with maternal emotional functioning, mental health, personality, and adjustment.^20,44,45  Feigelman et al reported an association between parental concern and seeking ED care for nonurgent medical problems, highlighting the role of parental health beliefs.⁴⁵  Further, perceived emergency of the child’s health and lack of awareness of alternatives were associated with unnecessary ED care.⁴⁶  In comparison, parents of CMC have intensive and longitudinal experiences with their child’s health and health care systems, which may moderate many of the above drivers, leaving measurable physiologic deterioration as a more likely primary driver for acute health care utilization.

Sociodemographic factors can predict parent decisions to seek acute care for their children.⁴⁴  For instance, difficulties in accessing primary care services and lack of insurance were associated with ED visits for children with minor illnesses.⁴²  Lower parental socioeconomic status contributed to ED care-seeking for their children.⁴⁶  Specifically, children from food-insecure households and with no married parents were more likely to visit an ED for ambulatory-sensitive conditions.⁴⁷  Although a few studies have explored factors associated with excessive parental concern in children with other chronic conditions,^20,44,45  our study is novel in that we explored drivers of parental concern in CMC, and identified the role of child’s physiologic/clinical parameters.

Our study has several limitations. Although vital signs, symptoms, and caregiver concern were recorded multiple times for each CMC during study, providing multiple measurements (n = 1223) and opportunities to assess the associations in the analysis, parent/caregiver sociodemographics were collected once for the 24 participants. The lack of associations between parental concern and sociodemographic characteristics may be because of limited variation. Secondly, the study participants lacked racial/ethnic diversity, and non-English–speaking parents were excluded. Although 13% of our participants identified themselves as Hispanic, nearly matching the 15% of Utah’s representation for Hispanic people (2020 Utah Census), we aim to be inclusive and develop a Spanish version of the app. A third potential limitation is that we dichotomized parental concern as low (1–2) or high (3–4). Because of uneven data distribution, dichotomizing parental concern from the most frequent responses of none (1) and mildly concerned (2) to the less frequent responses of moderately (3) and extremely concerned (4) is the preferred, albeit imperfect, but common approach in epidemiologic studies when analyzing ordinal rather than continuous variables.⁴⁸  Fourth, the wide range (13–91) of app use days may mean that some parents were using the app only when the child was sicker or when they had high concern, or daily app use is associated with some parent or child characteristics.

Although measuring clinical parameters itself might contribute to high concern, our design does not allow us to determine causality. In a future study, we will determine if high parental concern is an actual marker of impending acute deterioration of heath and a potential driver of outpatient visits and ED/hospital admissions for CMC. If there is no association, a final iteration of our app might omit assessment of parental concern. Also, study participants were recruited during acute hospitalizations. Although parents started monitoring their CMC’s health after discharge, the hospitalization may have heightened their level of concern, and results may have differed if participants were recruited in the ambulatory setting. Further, access to medical home or home health care may reduce parental concern, yet we did not collect this information. Lastly, our study was conducted in a single hospital and generalizability of our results to other locations may be limited.

Changes in child clinical parameters were associated with high parental concern, supporting parental concern as a possible early indicator of acute change in the health of CMC, especially if it is the primary parent complaint, with increased pain, worsening mental status, increased oxygen requirement, and reduced nutrition/fluid intake having significant associations. These findings can be used to inform interventions (eg, remote monitoring) or supplement existing care programs to sustain CMC care at home when parental concern is voiced, and possibly reduce avoidable ED visits and hospital admissions for CMC. Future studies will examine the association on a study population with wider variability in sociodemographic factors.

We thank the research coordinators (Angela Zhu and Grace Perry) of the initial pilot trial for their excellent support provided for the study implementation, participant enrollment, participant follow-ups, and data collection. We also thank the wonderful parents of our CMC who so willingly participated in this study and provided input and guidance.