Characteristics of Emergency Room and Hospital Encounters Resulting From Consumer Home Monitors Free

This study was reviewed by the hospital’s institutional review board and determined to meet exemption criteria. We analyzed documentation from all EDs (n = 293 591) and direct hospital admissions (n = 9413) for infants 0 to 12 months between January 1, 2013 and December 31, 2019 (total n = 303 004). Using structured query language and the relational database extract of our electronic health record (Epic, Verona, WI), we searched the free text of all initial ED and direct admission notes to identify clinical documentation in which a CHM keyword was mentioned.

The query was conducted in several phases. First, based on clinical documentation experience, the study team created a preliminary keyword list and queried all ED documentation during the study period. Keywords appeared in 757 encounters by 408 unique patients. Next, a representative sample of note sections was reviewed, (15%, 20 keyword-containing note sections per study year). An additional 10 keywords were identified, yielding an expanded keyword list (Supplemental Information). Using this list, all ED and direct admission documentation during the study period were queried; this query was restricted to infants 0 to 12 months. Keywords appeared in 1103 encounters by 870 unique patients. To further expand keyword capture, a representative sample of notes from different note author roles (ED nurse, provider, etc) were randomly selected and reviewed (15%, 50 notes from each role). This secondary review did not identify additional keywords, indicating the second query had approached maximum sensitivity.⁹ 

Documentation identified by a CHM keyword underwent independent review by 2 physicians. Reviewers categorized documentation of CHM use (documented CHM, no CHM documented, or indeterminate CHM documentation) and listed etiologic diagnoses from a multiple-choice list of common diagnoses. Encounters in which only 1 reviewer confirmed CHM documentation (n = 12) or there were discrepant discharge diagnoses (n = 5) were either discussed by the reviewers or evaluated by a third reviewer. To assure consistent coding, 2 example texts from each category were discussed by the entire study team to create the coding definitions (coding definitions provided in Supplemental Information).

Encounters with documented CHM use were further evaluated by 2 physician reviewers to classify the role of CHM (primary cause of visit, secondary role, no role, or indeterminate) and categorize the discharge diagnoses. In cases where CHM was determined to play a primary or secondary role, clinical characteristics and measures of resource utilization were abstracted. Costs were extracted from the Pediatric Health Information System database (Children’s Hospital Association, Lenexa, Kansas), an administrative database with deidentified information from 50 US tertiary care pediatric hospitals. The Pediatric Health Information System data include standardized wage-index-adjusted charges by department (room and board, pharmacy, laboratory, clinical, supply, and imaging) and cost-to-charge ratios. Treatment costs were estimated as wage-index-adjusted charges multiplied by the corresponding cost-to-charge ratio, inflated to 2019 US dollars using the Medical Services component of the Consumer Price Index.

Among the ED and direct admission encounters for infants 0 to 12 months old during the study period (n = 303 004), keyword queries identified 1103 potentially CHM-related encounters. Review confirmed documentation of CHM use in 39 cases (Fig 1). In 36 encounters by 34 unique patients, CHM had a primary (n = 25, 69%) or secondary role (n = 11, 31%) in prompting the visit. Encounters mentioning CHM increased over the study period, more than doubling in the final year of the study (2019) compared with the prior year.

All patients were under the age of 9 months at time of initial presentation (Table 1), with majority of presentations occurring within the first 3 months of life (n = 24, 67%). Over half of patients (n = 18, 53%) were born either preterm (<37 weeks; 8 patients, 24%) or early term (37 weeks to 38 weeks 6 days; 10 patients, 29%) per consensus definitions by multiple professional organizations.¹⁰  Most patients (n = 27, 79%) were non-Hispanic White, and all had English-speaking caretakers.

Most encounters were initiated in the ED (29, 80%) as opposed to direct admission (7, 20%). Median length of stay (LOS) for ED-only encounters was 2.8 hours (interquartile range [IQR]: 2.2–4.2 hours). Patients admitted through the ED had a median hospital LOS of 1.4 days (IQR: 1.2–4.2 days; max 1.1 weeks). Median LOS was longer among direct admissions at 2.3 days (IQR: 1.5 days–5.6 days; max 1.5 weeks).

Patients were hospitalized in 20 (56%) encounters including 8 ICU admissions. As summarized in Table 2 and listed for each patient in the Supplemental Table 3, evaluation included laboratory testing in 16 encounters (44%), procedures (eg, EKG or 4-channel pneumogram) in 18 encounters (50%), and radiographic studies in 15 encounters (42%). In 3 encounters the infant received a neonatal sepsis evaluation; none were diagnosed with invasive bacterial infection. The leading discharge diagnoses were viral infection (13, 36%), gastroesophageal reflux (8, 22%), and false positive alarm (6, 17%). Diagnoses for the 8 ICU admissions included supraventricular tachycardia (2, 25%), viral infections (5, 63%), and fall or trauma (1, 13%). In both admissions for supraventricular tachycardia, the infants appeared well before the CHM alarm but became symptomatic with poor feeding, inconsolability, and a dusky appearance shortly after the alarm. Both infants converted to sinus rhythm with adenosine and were ultimately discharged on oral propranolol. Other potentially life-threatening presentations included an infant whose CHM alarm activated following an unwitnessed fall and an infant who was ultimately diagnosed with interstitial lung disease.

Cost information was available for 35 encounters. Median cost for ED only visits was $233 (IQR: $118–$384; max $1510). Median costs were higher among direct admissions ($11 675, IQR: $3453–$26 254) compared with ED-initiated admissions ($6593, IQR: $3124–$21 721).

Using a rigorous EHR note text-mining approach, we identified CHM use contributing to 36 ED visits or direct admissions in a sample of over 300 000 encounters during a 7 year period. The rate of CHM-related encounters rose over time and occurred primarily in predominately non-Hispanic White infants with English-speaking caregivers. Hospital admission rates and resource utilization were high in CHM-related encounters, particularly among direct admissions. Therefore, even though the overall annual rate of CHM-related encounters was low and did not indicate a major public health burden, for individual families who present to the ED or hospital for concerns related to CHMs, there may be important adverse financial and emotional consequences.

This study has several important limitations. First, we do not know the extent of CHM use in this hospital’s catchment area or the proportion of monitor users in the catchment area that would present to this hospital with CHM-related concerns. Therefore, we cannot calculate population-based utilization rates. Second, notes are imperfect reflections of encounters. It is possible that some of the notes categorized as “indeterminate” might have been CHM-related. There may also be encounters instigated by CHM alarms in which CHMs were not discussed in documentation. However, we attempted to be expansive in our keyword search methodology, and no additional keywords were identified after a second iteration of the keyword search. The observed increase in CHM-related encounters over time also mirrors increasing sales of these devices, suggesting our methodology reflects CHM use patterns. Third, further research is needed to understand the effect of CHM on families (who may experience relief or anxiety associated with CHM use), and the challenges CHM pose to outpatient pediatricians.

In conclusion, while the overall hospital resource utilization burden of CHM was low, it is increasing over time. When patients presented for monitor-related concerns, resource utilization and hospital cost for the individual patient was high. Potentially actionable diagnoses were sometimes made, consistent with a prior report.²  It is unknown whether there would have been clinically important health impacts to these patients in the absence of CHM use or if they represent a form of overdiagnosis. While there are other compelling reasons for pediatricians to recommend against using CHM, the use of these products does not appear to be a frequent cause of ED or hospital visits in our catchment area.