PCP indicates primary care practitioner.
eTable 1. Comparison of Study Participants and Nonparticipants
eTable 2. Reasons that Inpatient Team Gave for Wanting Scheduled Follow-up
eTable 3. Comparison of Participants Included in Analysis vs Those Lost to Follow-up
Data Sharing Statement.
Customize your JAMA Network experience by selecting one or more topics from the list below.
Coon ER, Destino LA, Greene TH, Vukin E, Stoddard G, Schroeder AR. Comparison of As-Needed and Scheduled Posthospitalization Follow-up for Children Hospitalized for Bronchiolitis: The Bronchiolitis Follow-up Intervention Trial (BeneFIT) Randomized Clinical Trial. JAMA Pediatr. 2020;174(9):e201937. doi:10.1001/jamapediatrics.2020.1937
Is an as-needed posthospitalization follow-up visit noninferior to a scheduled follow-up visit for reducing anxiety in parents of children hospitalized for bronchiolitis?
In this randomized clinical trial of 304 children hospitalized for bronchiolitis, an as-needed follow-up visit led to 62% fewer follow-up visits and was noninferior to a scheduled follow-up visit with respect to reducing parental anxiety. No differences were noted in secondary outcomes, including readmissions and duration of symptoms.
In the absence of specific concerns, medical practitioners can recommend as-needed follow-up to parents of children hospitalized for bronchiolitis.
Posthospitalization follow-up visits are prescribed frequently for children with bronchiolitis. The rationale for this practice is unclear, but prior work has indicated that families value these visits for the reassurance provided. The overall risks and benefits of scheduled visits have not been evaluated.
To assess whether an as-needed posthospitalization follow-up visit is noninferior to a scheduled posthospitalization follow-up visit with respect to reducing anxiety among parents of children hospitalized for bronchiolitis.
Design, Setting, and Participants
This open-label, noninferiority randomized clinical trial, performed between January 1, 2018, and April 31, 2019, assessed children younger than 24 months of age hospitalized for bronchiolitis at 2 children’s hospitals (Primary Children’s Hospital, Salt Lake City, Utah, and Lucile Packard Children’s Hospital, Palo Alto, California) and 2 community hospitals (Intermountain Riverton Hospital, Riverton, Utah, and Packard El Camino Hospital, Mountain View, California). Data analysis was performed in an intention-to-treat manner.
Randomization (1:1) to a scheduled (n = 151) vs an as-needed (n = 153) posthospitalization follow-up visit.
Main Outcome and Measures
The primary outcome was parental anxiety 7 days after hospital discharge, measured using the anxiety portion of the Hospital Anxiety and Depression Scale, which ranged from 0 to 28 points, with higher scores indicating greater anxiety. Fourteen prespecified secondary outcomes were assessed.
Among 304 children randomized (median age, 8 months; interquartile range, 3-14 months; 179 [59%] male), the primary outcome was available for 269 patients (88%). A total of 106 children (81%) in the scheduled follow-up group attended a scheduled posthospitalization visit compared with 26 children (19%) in the as-needed group (absolute difference, 62%; 95% CI, 53%-71%). The mean (SD) 7-day parental anxiety score was 3.9 (3.5) among the as-needed posthospitalization follow-up group and 4.2 (3.5) among the scheduled group (absolute difference, −0.3 points; 95% CI, −1.0 to 0.4 points), with the upper bound of the 95% CI within the prespecified noninferiority margin of 1.1 points. Aside from a decreased mean number of clinic visits (absolute difference, −0.6 visits per patient; 95% CI, −0.4 to −0.8 visits per patient) among the as-needed group, there were no significant between-group differences in secondary outcomes, including readmissions (any hospital readmission before symptom resolution: absolute difference, −1.6%; 95% CI, −5.7% to 2.5%) and symptom duration (time from discharge to cough resolution: absolute difference, −0.6 days; 95% CI, −2.4 to 1.2 days; time from discharge to child reported “back to normal”: absolute difference, −0.8 days; 95% CI, −2.7 to 1.0 days; and time from discharge to symptom resolution: absolute difference, −0.6 days; 95% CI, −2.5 to 1.3 days).
Conclusions and Relevance
Among parents of children hospitalized for bronchiolitis, an as-needed posthospitalization follow-up visit is noninferior to a scheduled posthospitalization follow-up visit with respect to reducing parental anxiety. These findings support as-needed follow-up as an effective posthospitalization follow-up strategy.
ClinicalTrials.gov Identifier: NCT03354325
Contrary to adults, most hospitalized children are otherwise healthy and experiencing an acute, self-limited condition, such as a viral infection, cellulitis, or dehydration.1 After hospitalization and initiation of treatment, these children are expected to experience rapid and complete recovery, with low risk (<5%) of hospital readmission2 or any permanent disability. Despite such a reassuring prognosis, previous work3 has found that 85% of children discharged after a hospitalization for bronchiolitis were routinely instructed to attend a posthospitalization follow-up visit. In fact, recommending a posthospitalization follow-up visit for bronchiolitis has been proposed4 and implemented5 as a quality metric.
To our knowledge, no randomized clinical trials (RCTs) have examined the effects of a posthospitalization follow-up visit, and observational studies6-15 have come to conflicting conclusions. The intuitive appeal of a posthospitalization visit is that it might smooth the transition from hospital to home, providing an opportunity for a primary care practitioner (PCP) to verify that a patient is following an expected course and to provide additional reassurance, education, and/or treatments to support a patient’s recovery. On the other hand, particularly in the setting of an acute, self-limited condition, it is possible that these visits do not benefit children and their families and may be a source of unnecessary medical care.
We designed the Bronchiolitis Follow-up Intervention Trial (BeneFIT) to examine the value of the posthospitalization follow-up visit among children with bronchiolitis, the leading cause of hospitalization for young children.1,16 We chose parental anxiety as a primary outcome given the survey results from a pilot study3 of 198 children hospitalized for bronchiolitis in which parents who attended a follow-up visit reported that “reassurance provided” was the most valuable aspect of the visit. Furthermore, given the self-limited nature of bronchiolitis and the paucity of interventions available to modify the disease course,17 other outcomes, such as duration of symptoms and hospital readmission, were deemed less appropriate for a primary outcome. This study therefore tested the hypothesis that as-needed posthospitalization follow-up is noninferior to scheduled posthospitalization follow-up with respect to reducing parental anxiety after hospital discharge.
BeneFIT was an open-label, noninferiority RCT, performed between January 1, 2018, and April 31, 2019, that compared a scheduled follow-up visit with an as-needed follow-up visit among children being discharged after a bronchiolitis hospitalization. Randomization occurred at the patient level in a 1:1 allocation ratio between groups (Figure). The study protocol in Supplement 1 was registered before enrollment of any patients. A parent of each enrolled child provided written informed consent. Data were not deidentified. Ethics approval was obtained from the institutional review boards at each participating site. This study followed the Consolidated Standards of Reporting Trials (CONSORT) reporting guideline.
Participating hospitals included 2 free-standing children’s hospitals (Primary Children’s Hospital, Salt Lake City, Utah, and Lucile Packard Children’s Hospital, Palo Alto, California) and 2 community hospitals (Intermountain Riverton Hospital, Riverton, Utah, and Packard El Camino Hospital, Mountain View, California) with small pediatric units. Hospitalized children younger than 24 months with an attending diagnosis of bronchiolitis were eligible for recruitment if the child did not have a history of chronic lung disease, complex or hemodynamically significant heart disease, immunodeficiency, or neuromuscular disease and if the child was not being discharged home with a narcotic or benzodiazepine for treatment of withdrawal. Eligible patients were excluded for the following reasons: the inpatient team or PCP wanted a scheduled posthospitalization follow-up visit for the child; the child had a sibling concurrently hospitalized for bronchiolitis; neither parent spoke English or Spanish; routine follow-up, including a well-child check, was already scheduled for shortly after discharge; a PCP or follow-up clinic had not been identified for the child before randomization; or the child was being discharged with home oxygen therapy.18 The decision to discharge with oxygen is made at the time of discharge, necessitating exclusion of some enrolled patients, before randomization. Study investigators screened potential participants for eligibility by reviewing the medical records for children hospitalized with bronchiolitis and through further discussion with the patient’s inpatient medical team. Enrollment occurred during 2 winter seasons at each participating hospital.
Parents of children who were randomized to a scheduled posthospitalization follow-up visit were instructed to have their child follow up with the child’s PCP within 4 days of discharge, regardless of improvement and/or symptom resolution. Parents of children who were randomized to an asneeded posthospitalization follow-up visit were instructed to seek a PCP follow-up if the child’s symptoms did not resolve or worsened, but that a visit was not necessary if they remained comfortable with the trajectory of their child’s illness. Randomization was performed using randomly permuted blocks of random sizes stratified by hospital site. Randomization sequence generation was performed by study investigators (T.H.G. and G.S.) who were not involved in the actual assignment of enrolled patients to interventions. Allocation concealment was maintained by a central, automated computer platform (REDCap). Owing to the nature of the intervention, blinding of participants and their medical pracitioners was not possible. Even if it were possible, we would have chosen not to blind parents to their intervention group because doing so would effectively block the main mechanism for a possible effect of the intervention. That is, the primary mechanism by which an as-needed follow-up recommendation might affect parental anxiety and secondary outcomes is through parental knowledge of the recommendation.
All inpatient care occurred at the discretion of the inpatient team responsible for enrolled patients. Plans for imminent hospital discharge, as communicated by the inpatient team to study coordinators, prompted randomization by study coordinators, who then notified a patient’s parents, inpatient care practitioners, and PCP (via facsimile and/or telephone call) of the child’s assignment. Demographic and clinical characteristics were collected by medical record review and in-person interview of parents. Study coordinators measured outcomes by contacting parents via telephone every 7 days after hospital discharge until symptom resolution or 50 days after discharge, whichever came first, with the exception that all participants received a 28-day telephone interview regardless of symptom resolution. Scripts and computerized data entry were used for all telephone interviews. Symptom resolution was defined as parental report that the child’s cough had resolved and that the child was “back to normal.” Attendance of a scheduled posthospitalization follow-up visit was defined by parental selection of the reason for visit as routine hospital follow-up visit. This definition was applied regardless of the child’s randomization assignment and excluded visits for new concerns and visits to specialists.
The prespecified primary outcome was parental anxiety 7 days after hospital discharge, measured using the anxiety portion of the Hospital Anxiety and Depression Scale (HADS), which ranged from 0 to 28 points, with higher scores indicating greater anxiety.19 Choice of this tool was motivated by its use in a recent RCT of hospitalized patients with bronchiolitis,20 the anxiety scores of which informed our sample size calculation (trial protocol in Supplement 1). Parental anxiety was measured at 2 time points: (1) within the 2 days that preceded hospital discharge but before randomization (baseline parental anxiety) and (2) the 7-day posthospital discharge telephone call (7-day parental anxiety).
Prespecified secondary outcomes are detailed in the study protocol in Supplement 1 and included measures of symptom duration, ambulatory health care use, hospital re-use, and family-centered outcomes, such as satisfaction, missed work, and the relationship between parent and PCP. Satisfaction and dissatisfaction with care questions were identified from the Patient Satisfaction Questionnaire Short Form.21 Parental relationship with the child’s PCP was measured by the Patient-Doctor Depth-of-Relationship Scale.22 Ambulatory care charges were obtained as allowed charges for a subset of patients who, at the time of hospital discharge, were members of 1 regional, nongovernment health insurance plan in the West. Owing to the small subset of patients for whom charges could be obtained, between-group comparisons were not performed.
Sample size was estimated based on the primary outcome of 7-day parental anxiety as measured by HADS. We designated a noninferiority margin of 2 points on the 28-point scale. On the basis of the follow-up patterns observed in the pilot study,3 we estimated that 87% of parents instructed to schedule a follow-up visit would actually attend the visit and that 33% of parents instructed to follow up on an as-needed basis would nevertheless choose to attend a scheduled posthospitalization follow-up visit. Hence, the noninferiority margin for our primary intention-to-treat analysis that included all randomized patients was adjusted to be 2 × (1 − 0.33 − 0.13) = 1.1 points on HADS. Assuming that measurement of the primary outcome 7 days after hospitalization would be obtained for at least 90% of participants, a total sample size of 294 patients (147 patients in each group) would provide 80% power with 1-sided α = .025 to demonstrate noninferiority of as-needed follow-up if in fact as-needed and scheduled follow-up lead to equal levels of parental anxiety.
Analyses of the primary and secondary outcomes were performed in an intention-to-treat manner. The primary outcome was compared between groups via an analysis of covariance (ANCOVA) comparing 7-day parental anxiety scores, controlling for baseline parental anxiety scores and hospital site. For secondary outcomes, ANCOVA, log-rank tests, Cochran-Mantel Haenszel tests, and negative binomial models were used to compare continuous, time-to-event, dichotomous, and count outcomes, respectively, with hospital site included as a covariate (ANCOVA and negative binomial model) or a stratification factor (log-rank test or Cochran-Mantel Haenszel test). Symptom resolution and secondary outcomes measured until symptom resolution were right-censored at 50 days after discharge or at the time of unavailability for study follow-up. One parent answered only 6 of 7 baseline HADS questions. We imputed a value for the seventh response by using the mean of the parent’s responses to the other 6 questions. Otherwise, missing data were left as missing, without imputation.
In the absence of 100% adherence and complete study follow-up, the intention-to-treat analysis estimates the effect of a recommendation for scheduled vs as-needed follow-up. We supplemented the intention-to-treat analysis with an instrumental variable analysis, which estimates the effect of actually attending a scheduled follow-up visit, assuming the exclusion assumption is valid.23 In this analysis, 7-day parental anxiety was the dependent variable, and a patient’s randomization group was the instrument. Whether a patient actually attended a scheduled follow-up visit was the endogenous variable. Baseline parental anxiety, patient race/ethnicity, intensive care unit stay, and maximum respiratory support during hospitalization were included in the analysis as covariates. A 1-sided P < .05 was considered to be statistically significant. All analyses were performed using Stata, version 14 (StataCorp).
A total of 548 patients were considered for enrollment, and 304 underwent randomization (Figure). Compared with participants, nonparticipants were younger (47 [16%] vs 57 [23%] <2 months of age) and had a longer mean (SD) duration of illness at the time of discharge (6.0 [2.7] vs 6.5 [3.4] days) (eTable 1 in Supplement 2). Reasons that the inpatient team wanted a scheduled follow-up visit are listed in eTable 2 in Supplement 2.
Demographic and clinical characteristics were well balanced between the intervention groups (Table 1). A total of 106 children (81%) in the scheduled follow-up group attended a scheduled posthospitalization visit compared with 26 children (19%) in the as-needed group (absolute difference, 62%; 95% CI, 53%-71%). Median time to attendance of a scheduled follow-up visit was 3 days (interquartile range, 2-4 days) from hospital discharge, and 103 (78%) of these visits occurred with the patient’s PCP.
Measurement of the primary outcome was available for 269 patients (88% of those randomized). Compared with included participants, participants who were unavailable for study follow-up were more likely to have a history of prematurity and less likely to have a parent who graduated college (eTable 3 in Supplement 2). The mean (SD) 7-day parental anxiety score was 3.9 (3.5) among the as-needed follow-up group and 4.2 (3.5) among the scheduled follow-up group (absolute difference, −0.3 points; 95% CI, −1.0 to 0.4; P = .39) (Table 2). The upper bound of the 95% CI of the absolute difference in mean 7-day parental anxiety scores was within the prespecified noninferiority margin of 1.1. The estimated effect of actually attending posthospitalization follow-up from the instrumental variable analysis was also not statistically significant (mean difference, −0.6 points; 95% CI, −1.8 to 0.5). A total of 103 baseline parental anxiety scores (99%) were collected within the 2 days that preceded hospital discharge, and the median time of 7-day parental anxiety score collection was 8 days (interquartile range, 7-11 days) after hospital discharge.
Compared with patients in the scheduled follow-up group, patients in the as-needed follow-up group attended fewer clinic visits (absolute mean difference, −0.6 visits per patient; 95% CI, −0.4 to −0.8 visits per patient) (Table 2) and were less often exposed to ambulatory testing (almost exclusively pulse oximetry; absolute difference, −44.7%; 95% CI, −34.1 to −55.4%). Readmission rates were 2.2% among patients in the as-needed follow-up group and 3.8% among patients in the scheduled follow-up group (absolute difference, −1.6%; 95% CI, −5.7% to 2.5%). Of the 8 readmitted patients, 3 received care during their readmission in the intensive care unit (2 patients in the scheduled follow-up group and 1 patient in the as-needed follow-up group), with all 3 patients staying less than 2 days in the intensive care unit and none requiring intubation. Among patients in the as-needed follow-up group, 13 (9.4%) received a new ambulatory medication compared with 21 patients (16.0%) in the scheduled follow-up group (absolute difference, −6.6%; 95% CI, −14.6% to 1.3%). New ambulatory medications prescribed after hospital discharge included antibiotics (n = 22), albuterol (n = 7), and inhaled corticosteroids (n = 6). Acute otitis media was the indication for 19 (86%) of the new antibiotic prescriptions. Among 55 patients for whom ambulatory charges were obtained, the mean (SD) charge for a primary care visit was $92 ($21).
Among children hospitalized for bronchiolitis, as-needed posthospitalization follow-up was noninferior to scheduled posthospitalization follow-up with respect to the primary outcome of parental anxiety 7 days after hospital discharge. Although children randomized to scheduled follow-up attended more clinic visits and were more often exposed to pulse oximetry testing, the 2 groups did not differ for other secondary outcomes.
To our knowledge, this is the first RCT to examine the effect of scheduled posthospitalization follow-up on patient outcomes. Observational studies of adult patients have focused on the association between posthospitalization follow-up and hospital revisit, with some studies suggesting benefit7-12 and others finding no reduction in revisits.13-15 Compared with adults, children are less often burdened by chronic conditions and more often hospitalized for diseases for which complete resolution is expected,1,24 which translates into a lower risk of untoward posthospitalization outcomes, including hospital revisit and mortality.2,25 This generally reassuring prognosis may explain why studies6,26-28 of scheduled follow-up among children have rarely demonstrated a benefit.
For self-limited conditions from which complete recovery is anticipated and for which morbidity or mortality after hospitalization is unlikely, the expected benefit of routinely recommending more medical care in the immediate posthospitalization period should be considered carefully. By definition, disease-modifying interventions are less available for self-limited conditions, such as bronchiolitis,17 constraining the effect a practitioner can have on facilitating symptom resolution at a scheduled follow-up visit. Any benefit from a scheduled follow-up visit in this setting may be less tangible and tied to the broader physician-patient or physician-parent relationship. The present RCT did not find improvements in any relationship-related outcomes, including parental anxiety, satisfaction with care, or depth of relationship with the child’s PCP.
One risk of providing greater opportunity for patients to receive medical care is that health care can beget more health care, without necessarily improving outcomes. For example, multiple studies6,15,29 have documented a paradoxical increase in hospital revisits after efforts to augment posthospitalization follow-up, without otherwise improving patient outcomes. The current RCT was not powered to detect differences in revisits to the hospital, but the point estimate for readmission was higher among children randomized to scheduled follow-up. Although also not statistically significant, the point estimates suggest an absolute 7% increase in receipt of ambulatory medications among children randomized to scheduled follow-up. Given that medications are generally not indicated for treatment of bronchiolitis17 and that symptom duration was similar between the 2 groups, this excess receipt of ambulatory medications may represent overtreatment. Other potential risks of scheduled follow-up visits not measured in our trial include transmission of viruses between the convalescing child and other patients at the clinic30 and an increased burden on outpatient offices that may already be taxed during the busy respiratory season.
Our findings suggest that posthospitalization follow-up should not be routinely scheduled for children with bronchiolitis and should be reconsidered as a quality metric. For most cases, giving parents the autonomy to determine when and if their child needs follow-up can simplify their child’s recovery without causing parents undue worry. More broadly, given the time and cost that follow-up visits require of families and the health care system, parties responsible for developing quality measures and improving patient transition from hospital to home should consider whether incentives to increase routine follow-up for self-limited conditions are prudent.
Important study limitations should be considered. First, we did not standardize the posthospitalization follow-up visit, and variation likely exists in how outpatient practitioners administer follow-up. Second, our findings may not generalize to all patients hospitalized with bronchiolitis. We excluded patients with significant underlying comorbidities and those for whom a medical practitioner wanted scheduled follow-up. Third, measurement of the primary outcome was missing for 12% of participants, although we found that participants unavailable for study follow-up were largely similar to participants included in the analysis. Fourth, this was an open-label trial, making it more susceptible to bias. However, given the generally positive societal view of follow-up, we would expect lack of blinding to bias our findings toward inferiority of as-needed follow-up. Fifth, we did not measure PCP perspectives regarding the 2 follow-up strategies. Successful posthospitalization care requires shared decision-making that involves the family, hospitalist, and PCP, and alternative approaches, such as follow-up via telemedicine, warrant consideration. Sixth, our trial was not powered to detect a potential benefit from scheduled follow-up visits in terms of earlier detection of respiratory distress and prevention of respiratory arrest, which are rare outcomes in bronchiolitis, especially during convalescence after discharge.
This multicenter RCT found that as-needed posthospitalization follow-up for children with bronchiolitis was noninferior to scheduled follow-up. The intuitive appeal of scheduled visits has propelled the widespread acceptance and adoption of this practice and its inclusion as a quality metric. However, our findings should trigger reconsideration of this practice for bronchiolitis and other acute, self-limited conditions.
Accepted for Publication: March 29, 2020.
Corresponding Author: Eric R. Coon, MD, MS, Department of Pediatrics, Primary Children’s Hospital, University of Utah School of Medicine, 100 N Mario Capecchi Dr, Salt Lake City, UT 84113 (email@example.com).
Published Online: July 6, 2020. doi:10.1001/jamapediatrics.2020.1937
Author Contributions: Dr Coon had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Concept and design: Coon, Destino, Greene, Schroeder.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: Coon, Stoddard.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Coon, Greene, Stoddard.
Obtained funding: Coon, Schroeder.
Administrative, technical, or material support: Coon, Destino, Vukin.
Conflict of Interest Disclosures: Dr Coon reported receiving grants from Intermountain Stanford Collaboration during the conduct of the study. No other disclosures were reported.
Funding/Support: This study was funded under a joint program between Intermountain Healthcare and Stanford School of Medicine.
Role of the Funder/Sponsor: The funding sources had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Data Sharing Statement: See Supplement 3.
Additional Contributions: Marc Brewer, MD (Palo Alto Medical Foundation, Palo Alto, California), Neal Davis, MD (Intermountain Healthcare, Murray, Utah), Robin Drucker, MD (Palo Alto Medical Foundation, Palo Alto, California), and Kevin Lash, MD (Wasatch Pediatrics, West Jordan, Utah), served as nonpaid ambulatory pediatrician consultants for the design of the study.