Parental Perspectives on Continuous Pulse Oximetry Use in Bronchiolitis Hospitalizations

As part of a multicenter, randomized trial (www.clinicaltrials.gov [identifier NCT03354325]) comparing follow-up strategies after bronchiolitis hospitalizations, we conducted an ancillary cross-sectional survey on parental perspectives on CPOX use. Participants underwent a structured interview led by a member of the research team during their child’s hospitalization (in-hospital interview) and received a follow-up structured interview by phone approximately one week after hospital discharge (postdischarge interview). Half of the participants were randomly assigned to receive an additional statement (Supplemental information) during the postdischarge interview on the potential harms of CPOX use to determine if conveying the concept of overdiagnosis of hypoxemia can change parental preferences on the frequency of monitoring.

We identified children <2 years of age hospitalized with a primary diagnosis of bronchiolitis during the respiratory seasons (December through April) of 2017–2018 and 2018–2019. Children with the following comorbidities were excluded: chronic lung disease, complex or hemodynamically significant heart disease, immunodeficiency, or neuromuscular disease. For the purpose of the primary clinical trial, enrollment was limited to patients for whom parents and medical providers were comfortable with random assignment to either a scheduled or as-needed follow-up plan. The sample size was based on the number of patients enrolled in the original trial. Given our fixed sample size of 306 patients who were ultimately contacted by phone, we had 80% power (α = .05) to detect a 10% absolute risk difference in desire for CPOX monitoring (eg, 30% vs 40%) between patients who were provided the intervention and those who were not.

This study took place within the inpatient units of 2 freestanding tertiary children’s hospitals (Lucile Packard Children’s Hospital Stanford, Palo Alto, CA; Primary Children’s Hospital, Salt Lake City, UT) and 2 affiliated community hospitals (Packard El Camino Hospital, Mountain View, CA; Riverton Hospital, Riverton, UT). Of note, select patients from the Utah hospitals were discharged on home oxygen therapy.¹⁷  Although these patients were not included in the primary trial, they were included in our ancillary study on parental views. The local Institutional Review Boards at Stanford University and the University of Utah approved this study.

Eligibility confirmation and self-reported sociodemographic factors were obtained from the electronic medical record. Screening and study enrollment occurred while the child was hospitalized for bronchiolitis and within 48 hours of hospital discharge. Signed, informed consent was obtained from the participating parent.

Each structured interview was conducted in either English or Spanish and responses were transcribed into Research Electronic Data Capture. For the in-hospital questions about monitors, research coordinators oriented the family by pointing to the monitoring devices as a prompt. We provided parents a $25 gift card at the completion of the first interview with an additional $25 after study completion.

Interview questions were either adapted from existing surveys regarding parental views on other methods of continuous monitoring, including home apnea monitors for premature infants,¹⁸  and fetal heart monitoring during birth^19,20  or created de novo. De novo questions were designed to explore parental perspectives on pulse oximetry use and their experiences with the machine. Areas of focus were based on previous studies that have identified parental preference for¹⁰  or even dependence on continuous monitoring.²¹  Question creation and development were under the consultation of a local survey design expert and research psychologist (B. Halpern-Felsher, PhD; see the Acknowledgments section). To assure clarity, accuracy, and consistency of questions, initial surveys were reviewed and modified by Lucile Packard’s Family Advisory Council. The survey was piloted on 10 recruited participants in October 2017, at which time cognitive interview pretesting was performed on all newly written or modified questions.

The in-hospital interview contained questions regarding parental views on vital signs, hospital monitors, and monitor alarms. The postdischarge interview contained only questions specific to the pulse oximetry machine including their overall preference for CPOX monitoring frequency. These monitoring preferences were obtained under the hypothetical situation of another bronchiolitis hospitalization. To reflect current recommendations of transition from continuous to intermittent pulse oximetry monitoring, parents were given the specific scenario in which their child is improving and no longer requiring oxygen support.

Because overdiagnosis is a relatively understudied concept for most patients and families,²²  efforts to educate parents about the overdiagnosis of hypoxemia or other conditions were not readily available. A brief intervention was designed to provide information on the potential harms of CPOX use, similar to the use of decision aids in shared decision-making.^23,24  The specific content on hypoxemia was initially drafted by several members of the authorship team (A.R.S. and E.R.C.) who have previously explored overdiagnosis in the context of bronchiolitis hospitalizations.^2,5,6,25  Similar to the interview instrument, the educational intervention was reviewed by a local survey design expert and research psychologist (B. Halpern-Felsher), modified by the Packard Family Advisory Council and piloted along with the rest of the interview on 10 families at Lucile Packard Children’s Hospital Stanford.

Descriptive statistical techniques were performed on demographic and baseline characteristics of survey participants. Responses to 5-point Likert questions on the degree of agreement were dichotomized to agree (agree and strongly agree) and disagree (uncertain, disagree, and strongly disagree). A similar dichotomy was applied to questions on the degree of comfort. For parental perceptions of vital sign importance, Likert responses were uniquely dichotomized to extremely important or not extremely important (all other responses) given the highly skewed distribution of parents assigning importance to all vital signs.

Responses from 5 questions in the postdischarge interview were combined to measure parental desire for CPOX monitoring. From these questions, an aggregate variable, representing “CPOX affinity,” was created on a matching Likert scale of 1 through 5, with 1 representing the lowest parental desire for continuous monitoring and 5 representing the greatest. Cronbach’s α with the standard minimum threshold of 0.7 was used to assess internal consistency among the 5 questions.

To determine if various parent or patient characteristics could predict CPOX affinity, we conducted bivariate linear regressions with the following predictor variables: parental education, parental anxiety levels, patient race, patient insurance type, hospital LOS, and whether the patient required ICU level care. We examined potential predictors of CPOX affinity only in the control group (those who did not receive the educational intervention) given that predictors in the intervention group may not be generalizable to the broader population. Because no predictor variables had a P value <.1, we did not perform multivariable regression. A Kruskal-Wallis test was used to assess potential differences in CPOX affinity across clinical sites.

The comparison of parental preferences for monitoring with and without the educational intervention was analyzed by χ² or Fisher’s exact test as appropriate. A 2-sided P value of <.05 was considered statistically significant. All analyses were conducted by using Stata 14.1 (Stata Corp, College Station, TX).

Of the 548 patients considered for enrollment, 357 (65%) patients were ultimately enrolled in the study (Fig 1). There were 306 parent responses to the postdischarge interview, representing 86% retention approximately one week after hospitalization. The same parent was interviewed at follow-up for 99% of cases. Patient and parental demographics are summarized in Table 1. There were no statistical differences in demographic variables between included and excluded patients.

Among vital signs, parents of children hospitalized with bronchiolitis viewed respiratory rate (87%) and oxygen saturation (84%) as “extremely important” (Table 2). During the in-hospital interview, nearly all parents either strongly agreed or agreed that having hospital monitors on their child was helpful (98%) and made them feel secure (94%) (Table 3). At the same time, some parents also strongly agreed or agreed that having the hospital monitors on their child was annoying (23%) and made them feel anxious (25%). When asked about monitor alarms, 140 of 336 (42%) parents who spent the night strongly agreed or agreed that the alarms woke them up from sleep, with 49% reporting >1 awakening in the night. However, only 36 of 357 (10%) parents felt that the monitor alarms woke their child up from sleep.

During the postdischarge phone interview, 215 of 306 (70%) parents reported that they were told what is considered a low oxygen level for their child. When asked to identify the oxygen saturation level at which additional oxygen support was warranted, 41% of parents reported that they “did not know.” Of the 181 who did report a value, the median response was an oxygen saturation level of 89% (range 30%–98%; interquartile range 88%–90%). To explore (post hoc) whether this median level varied by sites, we compared responses from the California hospitals (low altitude) and Utah hospitals (high altitude) and found no differences.

The effects of the randomly assigned, brief educational intervention explaining the potential harms of CPOX monitoring on parental preferences are summarized in Table 4. When compared with parents who did not receive educational messaging, the intervention resulted in a decrease in the proportions of parents who still preferred continuous monitoring should their child be hospitalized again with bronchiolitis (40% vs 20%; P < .001) and an increase in parents who either strongly disagreed or disagreed with the statement, “Even if I knew it might lead to a longer hospital stay, I would prefer continuous pulse oximetry monitoring at all times,” (18% vs 42%; P < .001). A similar increase was observed in those who strongly disagreed or disagreed with the option for a home pulse oximetry machine after discharge (18% vs 32%; P = .003). Of note, there was no difference between the control and intervention groups (11% vs 10%; P = .74) among parents who strongly disagreed or disagreed with the statement, “There is no harm to checking the oxygen level continuously.”

For the aggregate CPOX affinity variable, the minimum Cronbach’s α value of 0.7 was achieved by using all 5 questions outlined in Table 4. There were no statistically significant predictors of CPOX affinity, including being discharged on home oxygen, and affinity did not differ across clinical sites (P = .49). Post hoc, we explored the hypothesis that the efficacy of the educational intervention varied by college education. Among parents who had a college education or higher, the impact of the intervention on the CPOX affinity scale was −0.51 (95% confidence interval [CI] −0.75 to −0.27), whereas the impact on parents who did not have a college education was only −0.2 (95% CI −0.46 to 0.04).

In this multicenter, cross-sectional investigation of parental perspectives on CPOX use, we observed several important findings. Most parents reported that the hospital monitors gave them a sense of reassurance. However, a brief description of the potential harms of CPOX use can influence parental monitoring preferences in postdischarge telephone interviews.

In the last few years, the national discourse on reducing health care spending has identified specific areas of overtreatment in the field of pediatrics.^13,25,26  One of the most frequently cited opportunities for improvement is the management of viral bronchiolitis, which has an estimated annual cost of >$1.7 billion.²⁷  CPOX has been associated with unnecessary and prolonged hospitalizations,^2,3  but the practice of transitioning from continuous to intermittent pulse oximetry may cause anxiety for families. The majority of parents in our study reported that hospital monitors “were helpful” and “made them feel secure.” These findings are consistent with one single-institution study in Qatar by Hendaus et al¹⁰  in which 80% of parents supported the idea of CPOX monitoring in children hospitalized with bronchiolitis. In the same article, authors speculated that parental preference for CPOX may be a barrier to the widespread adoption of intermittent usage, calling for further exploration of parental perspectives and emphasizing the need for improved parental education.

In this multicenter study, we validate and expand the current understanding of parental perspectives on CPOX monitoring. Although parents appreciate the use of CPOX, they also acknowledge that hospital monitors “made them feel anxious” and “were annoying.” Among the parents who spent the night with their hospitalized child, 40% of parents reported that alarms woke them up from sleep. However, despite these sentiments, many parents still prefer CPOX, suggesting that the security of continuous monitoring outweighs many negative aspects of its use. Although these preferences do not imply that parents refuse intermittent monitoring when it is recommended, they do represent an important consideration when engaging in shared decision-making in the hospital and at times of recommended transition to intermittent monitoring.

In the second aim of our study, we explored whether parental counseling can change monitoring preferences. Recent frameworks for approaching overtreatment highlight the importance of incorporating patient factors and experiences, acknowledging that patient preferences may conflict with efforts to reduce waste.^28,29  In 2001, the Institute of Medicine stated that high-quality care should be customized to individual patient needs and values.³⁰  The challenge, however, is that overdiagnosis is a difficult concept to communicate.^31,32  Various strategies have been proposed in an attempt to improve both patient and clinician understanding of overdiagnosis but few have proven to minimize overtreatment in practice.³³  We found that a brief educational intervention on the potential harms of CPOX use could decrease parental affinity for continuous monitoring in postdischarge interviews. This study provides one example in which the concept of overdiagnosis was appreciated by some parents, demonstrating the impact of a brief educational intervention on family members. Similar studies exploring parental perspectives on antibiotic use for respiratory infections^34,35  have revealed that communication strategies that address parental concern and inform parents about antibiotic stewardship can increase parental acceptance of treatment recommendations.^36,37  This further highlights the potential value of decision aids for parents, particularly ones that identify potential harms of treatment.

Interventions that target patients and family members, however, require thoughtful consideration of various patient factors. In particular, we observed that word choice could have a direct impact on how parents interpret information that is provided to them. When parents were asked specifically to agree or disagree with the statement, “There is no harm to checking the oxygen level continuously,” the majority of parents agreed with the statement regardless of whether they received the educational intervention. The fact that many parents changed their preferences for monitoring after intervention implies that although they can appreciate the negative aspects of CPOX, they still did not equate an unnecessary and prolonged hospitalization as “harm” to their child. Another factor for consideration is the diverse spread of health literacy among patients. When stratified by parental education, we saw that the impact of our intervention had a greater effect size in parents who were college educated. This finding suggests that although overdiagnosis is a complex concept to understand, there may be opportunities to reframe our interventions to communicate effectively with a broader audience. Future endeavors hoping to incorporate parental education in HVC efforts require an astute understanding of parental health literacy and vocabulary.

Our study has several limitations. The educational intervention in this study was provided one week after hospital discharge. This decision was made both to prevent contamination of the primary study but also to avoid interfering with the care and information provided by the primary team. Although the intervention changed preferences for monitoring under a hypothetical situation, it will be important to replicate the efficacy of similar interventions during the acute stress of a hospitalization in real time, eliminating social desirability bias in this study and directly impacting monitoring frequencies in practice. Another limitation is the applicability of our educational intervention. For parents who did not change their preferences after intervention, it is possible that a different way of presenting the same information or perhaps additional facts would have been effective. Focus groups on this particular subset of parents may reveal what educational information, if any, could have changed their monitoring preferences. Educational interventions like decision aids can also vary in format from pamphlets to presentations.³⁸  With future iterations, there is an opportunity to create a validated decision aid for CPOX use based on International Patient Decision Aid Standards.

Parents find security in CPOX use, but a brief statement on the potential harms of overdiagnosis had an impact on stated preferences for monitoring. Parental perspectives are important to consider because they may influence the implementation of intermittent monitoring and other HVC practices. These findings in the setting of bronchiolitis could have broader implications for strategies to combat overdiagnosis and better engage families in HVC efforts.

We acknowledge Carrie Rassbach, Nivedita Srinivas, Amit Singh, Natalie Pageler, and Daniel Imler for providing feedback and oversight for this project. Survey development was supported by a local survey design expert and research psychologist, Bonnie Halpern-Felsher. We also thank the individual study coordinators for their contributions to this study. The Stanford University coordinators included Alicia Harnett, Juliana Moreno-Ramirez, Nick Bondy, Stephanie Escobar, and Trinh Nguyen. The University of Utah coordinators included Alexander Platt-Koch, Stacy Erickson, Evan Heller, Eun Hea Unsicker, Jillian Ivie, Karee Nicholson, Kelsee Stromberg, Lindsay Curtis, Megan Jenkins, Sharon Noorda, Tammi Lewis, and Taylor Mathie. We also thank Greg Stoddard for his advice in statistical analysis.

CI

confidence interval

CPOX

continuous pulse oximetry

HVC

high-value care

LOS

length of stay