During the coronavirus disease 2019 pandemic, technology-dependent children are at risk of encountering barriers to hospital discharge because of limits to in-home services. Transition difficulties could increase length of stay (LOS). With this study, we aim to (1) evaluate change in LOS and (2) describe barriers to hospital discharge between prepandemic and early pandemic periods for technology-dependent children.
A retrospective chart review of technology-dependent children discharged from an acute and specialty pediatric hospital within a single urban area between January 1 and May 28, 2020 was conducted. Technology dependence was defined by using a validated complex chronic condition coding system. Patients discharged prepandemic and during the pandemic were compared. Outcomes included LOS and the number and type of discharge barriers (a factor not related to a medical condition that delays discharge). Multivariate regression modeling and parametric and nonparametric analysis were used to compare cohorts.
Prepandemic, 163 patients were discharged, and 119 were discharged during the early stages of the pandemic. The most common technology dependence was a feeding tube. The unadjusted median LOS was 7 days in both groups. After adjusting for patient-level factors, discharge during the pandemic resulted in a 32.2% longer LOS (confidence interval 2.1%–71.2%). The number of discharge barriers was high but unchanged between cohorts. Lack of a trained caregiver was more frequent during the pandemic (P = .03).
Barriers to discharge were frequent for both cohorts. Discharge during the pandemic was associated with longer LOS. It was more difficult to identify a trained caregiver during the pandemic.
Children with medical complexity (CMC) constitute <1% of all children in the United States but account for 33% percent of pediatric Medicaid spending and 71% of pediatric hospital readmissions.1,2 The CMC population encompasses children with chronic medical conditions, functional limitations, and a high level of health care utilization.3 For pediatric academic medical centers, these children may represent up to 80% of the total inpatient hospital days.3 The hospital discharge process is a crucial period to coordinate safe and efficient care delivery at home. CMC experience discharge delays up to 30% of the time because of the extensive coordination required at discharge. These delays can increase health care costs and prolong length of stay (LOS).4,5 Discharge can be complicated by communication failures and competing socioeconomic priorities of primary caregivers after the patient transitions home.5–9
In March 2020, the advent of coronavirus disease 2019 (COVID-19) necessitated public health emergency declarations, including stay-at-home orders, that may have further complicated hospital discharge for this population.10 Pandemic-related restrictions may have limited the availability of in-person services, including home visits, homecare nursing, outpatient follow-up, and the acquisition of life-sustaining medical equipment and medications.10,11 Caregivers for CMC may have been especially hard hit with a lack of reliable child care for siblings, economic pressures, and a desire to protect their medically fragile child from illness.12,13 Initial studies of care disruptions during the COVID-19 pandemic indicate that those with multiple medical conditions, including CMC, are at the highest risk of interruptions in their usual medical care.11,14
Technology-dependent CMC who rely on life-sustaining medical devices may be at particularly high risk of hospital discharge barriers during the pandemic due to the need for trained in-home caregivers and durable medical equipment.15 Discharge delays for otherwise stable children who have technology dependence could significantly impede hospital flow, contributing to the critical shortage of hospital beds during the pandemic.4,16 There has been no systematic investigation into the number and types of barriers to hospital discharge encountered by this population during the COVID-19 pandemic. With this study, we aim to (1) evaluate change in LOS and (2) describe barriers to hospital discharge between prepandemic and early COVID-19 pandemic periods for technology-dependent children.
Materials and Methods
Setting
This was a 2-center cohort study using a retrospective review of hospitalized technology-dependent patients. The first center was a tertiary care, freestanding children’s hospital in an urban city on the east coast of the United States. The second center was a smaller pediatric specialty hospital in the same geographical area. This study was approved by the hospital’s institutional review board.
Study Period
Technology-dependent patients were eligible for our study if they had a hospital discharge from January 1, 2020 to May 28, 2020. The prepandemic cohort was defined as those with a discharge between January 1, 2020 and March 11, 2020. Those with a discharge between March 12, 2020 and May 28, 2020 were defined as being discharged during the early stages of the COVID-19 pandemic. The dates were defined on the basis of local and state public emergency declarations and stay-at-home orders.17–19
Study Population
A cohort of technology-dependent patients’ hospital discharges was identified by using complex chronic condition International Classification of Diseases, 10th Revision (ICD-10) discharge codes mapping to technology dependence.20 All pediatric patients aged 0 to 21 who were discharged from any unit or service in each hospital were included. Encounters were excluded if the patient died during the hospitalization, the patient was found to have no technology dependence after chart review, or the patient was admitted to the acute hospital from the specialty hospital. After analysis, we identified a sizable percentage of patients in our cohorts who experienced multiple hospital discharges during our time frame. Only the first discharge from each time period per patient was included. The study team assumed that discharge barriers were likely not independent for the same patient within time frames but may have changed during the pandemic because of public health policies.
Data Collection
Demographic information collected included age, sex, race and ethnicity, state of residence, insurance type, preferred language, type of technology dependence, and family structure (including the presence of other children in the home and multiple adult caregivers if documented in a social work note within 1 year of admission). Charts were reviewed to determine if technology was newly acquired during the admission because these patients may be at higher risk of discharge barriers because of the establishment of new in-home services. Severity of illness measures were included, specifically ICU admission and number of home medications, for each hospital discharge. Hospital LOS in days was collected for each patient encounter for comparison across time periods.
Discharge barriers were defined after literature review and group consensus as factors not related to the patient’s medical condition that delay discharge from the hospital.16 The number and type of discharge barrier were collected for each discharge via chart review of inpatient case management, social work notes, and the physician-authored discharge summary. Barriers were grouped into 8 categories on the basis of a literature review. The categories identified in the literature were considered by the authors on the team with experience caring for CMC and determined to be clinically relevant for our population. The discharge barrier categories used were home nursing, housing, appropriate patient caregiver, durable medical equipment (DME), medications, follow-up appointments, transportation, and financial.16
The primary investigator (MR) initially reviewed 10% of the identified cohort of discharges, and other investigators reviewed 10 charts each to provide feedback on data collection methodology. Research assistants (RAs) were employed to continue and complete a chart review. Each RA reviewed 5 charts, with subsequent feedback after rereview from the primary investigator (M.R.). An additional 10 charts were then assigned to each RA, with rereview from a senior investigator (A.K.) to ensure reliability. Identified areas of common discrepancy included a definition of newly acquired technology, a definition of subacute care, and locating a description of family structure in the chart. Approximately 20% of charts assigned to an RA had these commonly discrepant fields (patients with newly acquired technology, who required subacute care, or when the family description could not be located) and underwent secondary review by PI (M.R.) to ensure the accuracy of data. While reviewing eligible encounters, 14 discharges for technology-dependent patients within the study period were observed that were not identified by using the ICD-10 coding strategy. This was thought to be due to variations in the use of discharge codes. Most of these encounters were patients with multiple scheduled readmissions, such as for chemotherapy, in which it is likely that their technology dependence was not recorded as a separate discharge code during each time. These encounters were reviewed and added to the cohort. All patients identified by the coding strategy but deemed to have no technology dependence after chart review were also rereviewed by the PI (MR) to verify accuracy.
Analysis
The prepandemic and during-pandemic cohorts’ demographics were compared by using the student’s t test for quantitative variables and either the χ2 test or Fisher’s exact test for categorical variables. The frequency of any discharge barrier and the frequency of specific barriers were also compared prepandemic and during the pandemic by using either the χ2 test or Fisher’s exact test. Subgroup analysis was conducted for discharges in which patients acquired new technology during that hospitalization. Negative binomial multivariable regression modeling was used to determine the factors significantly impacting LOS. Univariate modeling was first used looking at all demographic factors and indicators of disease severity. Each variable was evaluated for interactions. Those identified as significant with a P value of <.05 in univariate analysis (type of technology, age, family structure, insurance type, and indicators of illness severity) were added to the multivariate regression model.
Results
We identified 399 total discharges for technology-dependent patients from our hospital and associated specialty hospital from January 1, 2020 until May 28, 2020 (Fig 1). We identified 364 discharges from the tertiary care center and 35 from the specialty hospital. We excluded 117 encounters. Of the remaining discharges, 163 patients were discharged prepandemic and 119 during the pandemic. Patients in both groups were demographically similar, although male patients were overrepresented in those discharged during the pandemic (Table 1).
. | Pre-COVID-19, n (%) (n = 163) . | COVID-19, n (%) (n = 119) . | P . |
---|---|---|---|
Age** | 3.0 (2.0) | 2.8 (2.0) | .531 |
Sex | |||
Male | 74 (45.7%) | 72 (60.5%) | .014* |
Female | 88 (54.3%) | 47 (39.5%) | |
Race | |||
Asian | 2 (1.2%) | 4 (3.4%) | .370 |
Black or African American | 53 (32.5%) | 35 (29.4%) | |
Hispanic or Latino | 12 (7.4%) | 6 (5.0%) | |
Native Hawaiian or other Pacific Islander | 1 (0.6%) | 0 (0%) | |
Caucasian | 43 (26.4%) | 42 (35.3%) | |
Other or not documented | 52 (31.9%) | 32 (26.9%) | |
Ethnicity | |||
Hispanic or Latino | 46 (29.9%) | 24 (22.9%) | .212 |
Non-Hispanic or Latino | 108 (70.1%) | 81 (77.1%) | |
State of residence | |||
Maryland | 93 (57.4%) | 69 (58.0%) | 1.000 |
Virginia | 36 (22.2%) | 26 (21.9%) | |
Washington DC | 28 (17.3%) | 20 (16.8%) | |
West Virginia | 3 (1.9%) | 2 (1.7%) | |
Other | 2 (1.2%) | 2 (1.7%) | |
Type of insurance | |||
Self-pay or blank | 13 (8.1%) | 8 (7.0%) | .062 |
Medicaid | 107 (66.9%) | 63 (54.8%) | |
Private | 40 (25.0%) | 44 (38.3%) | |
Language | |||
English | 121 (74.2%) | 96 (80.7%) | .334 |
Spanish | 32 (19.6%) | 16 (13.5%) | |
Arabic | 4 (2.5%) | 5 (4.2%) | |
Other or Not Documented | 6 (3.7%) | 2 (1.7%) | |
Family structure | |||
1 caregiver | 24 (14.7%) | 25 (21.0%) | .169 |
2 caregivers | 99 (60.7%) | 76 (63.9%) | .593 |
No other children at home | 64 (39.3%) | 57 (47.9%) | .148 |
Extended family in area | 29 (17.8%) | 28 (23.5%) | .236 |
Foster home | 2 (1.2%) | 1 (0.8%) | 1.000 |
Group home | 6 (3.7%) | 1 (0.8%) | .245 |
CPS case | 7 (4.3%) | 6 (5.0%) | .768 |
Unknown | 23 (14.1%) | 13 (10.9%) | .429 |
Other | 7 (4.3%) | 7 (5.9%) | .544 |
Types of technology | |||
Tracheostomy | 44 (27.0%) | 24 (20.2%) | .186 |
Mechanical ventilator | 26 (59.1%) | 16 (66.7%) | .539 |
Noninvasive respiratory support | 28 (17.2%) | 13 (10.9%) | .141 |
Enteral feeding tube | 131 (80.4%) | 87 (73.1%) | .151 |
Central line | 18 (11.0%) | 16 (13.6%) | .523 |
Ventricular shunt | 26 (16.0%) | 27 (22.7%) | .153 |
Other | 37 (22.7%) | 22 (18.5%) | .390 |
Total no. of technologies** | 1.9 (1.0) | 1.7 (0.9) | .105 |
>5 meds | 108 (67.1%) | 79 (66.4%) | .903 |
. | Pre-COVID-19, n (%) (n = 163) . | COVID-19, n (%) (n = 119) . | P . |
---|---|---|---|
Age** | 3.0 (2.0) | 2.8 (2.0) | .531 |
Sex | |||
Male | 74 (45.7%) | 72 (60.5%) | .014* |
Female | 88 (54.3%) | 47 (39.5%) | |
Race | |||
Asian | 2 (1.2%) | 4 (3.4%) | .370 |
Black or African American | 53 (32.5%) | 35 (29.4%) | |
Hispanic or Latino | 12 (7.4%) | 6 (5.0%) | |
Native Hawaiian or other Pacific Islander | 1 (0.6%) | 0 (0%) | |
Caucasian | 43 (26.4%) | 42 (35.3%) | |
Other or not documented | 52 (31.9%) | 32 (26.9%) | |
Ethnicity | |||
Hispanic or Latino | 46 (29.9%) | 24 (22.9%) | .212 |
Non-Hispanic or Latino | 108 (70.1%) | 81 (77.1%) | |
State of residence | |||
Maryland | 93 (57.4%) | 69 (58.0%) | 1.000 |
Virginia | 36 (22.2%) | 26 (21.9%) | |
Washington DC | 28 (17.3%) | 20 (16.8%) | |
West Virginia | 3 (1.9%) | 2 (1.7%) | |
Other | 2 (1.2%) | 2 (1.7%) | |
Type of insurance | |||
Self-pay or blank | 13 (8.1%) | 8 (7.0%) | .062 |
Medicaid | 107 (66.9%) | 63 (54.8%) | |
Private | 40 (25.0%) | 44 (38.3%) | |
Language | |||
English | 121 (74.2%) | 96 (80.7%) | .334 |
Spanish | 32 (19.6%) | 16 (13.5%) | |
Arabic | 4 (2.5%) | 5 (4.2%) | |
Other or Not Documented | 6 (3.7%) | 2 (1.7%) | |
Family structure | |||
1 caregiver | 24 (14.7%) | 25 (21.0%) | .169 |
2 caregivers | 99 (60.7%) | 76 (63.9%) | .593 |
No other children at home | 64 (39.3%) | 57 (47.9%) | .148 |
Extended family in area | 29 (17.8%) | 28 (23.5%) | .236 |
Foster home | 2 (1.2%) | 1 (0.8%) | 1.000 |
Group home | 6 (3.7%) | 1 (0.8%) | .245 |
CPS case | 7 (4.3%) | 6 (5.0%) | .768 |
Unknown | 23 (14.1%) | 13 (10.9%) | .429 |
Other | 7 (4.3%) | 7 (5.9%) | .544 |
Types of technology | |||
Tracheostomy | 44 (27.0%) | 24 (20.2%) | .186 |
Mechanical ventilator | 26 (59.1%) | 16 (66.7%) | .539 |
Noninvasive respiratory support | 28 (17.2%) | 13 (10.9%) | .141 |
Enteral feeding tube | 131 (80.4%) | 87 (73.1%) | .151 |
Central line | 18 (11.0%) | 16 (13.6%) | .523 |
Ventricular shunt | 26 (16.0%) | 27 (22.7%) | .153 |
Other | 37 (22.7%) | 22 (18.5%) | .390 |
Total no. of technologies** | 1.9 (1.0) | 1.7 (0.9) | .105 |
>5 meds | 108 (67.1%) | 79 (66.4%) | .903 |
CPS, child protective services.
The 2 cohorts are similar across multiple domains. More male patients were discharged during the pandemic.
Indicates statistical significance at P <.05.
Age and number of technologies are presented as: mean (standard deviation) rather than n (%).
Length of Stay
Median LOS was 7 days both prepandemic and during the pandemic (prepandemic interquartile range varied from 4 to 20 days; during the pandemic interquartile range varied from 3 to 23.5 days). When using regression modeling to account for variables that may impact LOS, discharge during the pandemic was statistically associated with longer LOS (estimated increase of 32%, 95% confidence interval 2.1%–71.2%, P = .03). Other factors that were associated with an increase in LOS were younger age, having extended family in the area, number of technologies required, >5 home medications, the use of noninvasive respiratory support at home, an ICU stay, or discharge from the pediatric specialty hospital. Ventricular shunts were associated with a shorter LOS (Table 2).
. | % Change in LOS . | 95% CI . | P . |
---|---|---|---|
Pandemic period | 32.2 | 2.1 to 71.2 | .03* |
Age | −11.8 | −17.3 to −5.93 | <.01* |
Type of insurance | |||
Private | Ref | Ref | .06 |
Self-pay or blank | 40.7 | −16.8 to 138.0 | |
Medicaid | −16.4 | −37.2 to 11.5 | |
Family structure | |||
Extended family in area | 53.6 | 11.1 to 112.6 | .01* |
Types of technology | |||
Noninvasive respiratory support | 44.7 | −0.2 to 109.9 | .05** |
Enteral feeding tube | 0.4 | −29.7 to 43.4 | .98 |
Ventricular Shunt | −44.7 | −62.4 to −18.6 | .01* |
Total no. of technologies | 12.5 | −4.6 to 32.8 | .16 |
ICU during admission | 106.1 | 51.2 to 180.9 | <.01* |
>5 meds | 174.6 | 54.4 to 388.4 | <.01* |
Admission to pediatric rehabilitation hospital | 232.2 | 116.4 to 409.7 | <.01* |
. | % Change in LOS . | 95% CI . | P . |
---|---|---|---|
Pandemic period | 32.2 | 2.1 to 71.2 | .03* |
Age | −11.8 | −17.3 to −5.93 | <.01* |
Type of insurance | |||
Private | Ref | Ref | .06 |
Self-pay or blank | 40.7 | −16.8 to 138.0 | |
Medicaid | −16.4 | −37.2 to 11.5 | |
Family structure | |||
Extended family in area | 53.6 | 11.1 to 112.6 | .01* |
Types of technology | |||
Noninvasive respiratory support | 44.7 | −0.2 to 109.9 | .05** |
Enteral feeding tube | 0.4 | −29.7 to 43.4 | .98 |
Ventricular Shunt | −44.7 | −62.4 to −18.6 | .01* |
Total no. of technologies | 12.5 | −4.6 to 32.8 | .16 |
ICU during admission | 106.1 | 51.2 to 180.9 | <.01* |
>5 meds | 174.6 | 54.4 to 388.4 | <.01* |
Admission to pediatric rehabilitation hospital | 232.2 | 116.4 to 409.7 | <.01* |
CI, confidence interval.
Factors identified with univariate modeling predicting LOS are listed. Multivariate modeling reveals the pandemic was a factor in predicting LOS when accounting for technology type, local family, and indicators of illness severity.
Indicates P value <.05.
Indicates significant before rounding.
Discharge Barriers
Discharge barriers were experienced by >1 in 4 of our patients (26% prepandemic, 27% during the pandemic P = .83). Home nursing was the most frequently identified barrier to discharge in both time frames. Difficulty identifying an appropriately trained caregiver for home was more likely to be a discharge barrier during the pandemic (P = .03) (Table 3). Forty percent of patients who experienced a discharge barrier had >1 barrier documented. There was no change in the frequency of having >1 discharge barrier during the pandemic.
. | Prepandemic (n = 163) . | During Pandemic (n = 119) . | P . |
---|---|---|---|
Lack of nursing | 15 (9.2%) | 15 (12.6%) | .36 |
Lack of housing | 7 (4.3%) | 3 (2.5%) | .53 |
Lack of caregiver | 4 (2.5%) | 10 (8.4%) | .03* |
Difficulty obtaining DME | 14 (8.6%) | 7 (5.9%) | .39 |
Difficulty obtaining medication | 7 (4.3%) | 7 (5.9%) | .54 |
Inability to obtain follow-up | 2 (1.2%) | 3 (2.5%) | .65 |
Transportation difficulty | 6 (3.7%) | 9 (7.6%) | .15 |
Financial difficulty | 6 (3.7%) | 4 (3.4%) | 1.00 |
Any barrier during encounter | 42 (25.8%) | 32 (26.9%) | .83 |
. | Prepandemic (n = 163) . | During Pandemic (n = 119) . | P . |
---|---|---|---|
Lack of nursing | 15 (9.2%) | 15 (12.6%) | .36 |
Lack of housing | 7 (4.3%) | 3 (2.5%) | .53 |
Lack of caregiver | 4 (2.5%) | 10 (8.4%) | .03* |
Difficulty obtaining DME | 14 (8.6%) | 7 (5.9%) | .39 |
Difficulty obtaining medication | 7 (4.3%) | 7 (5.9%) | .54 |
Inability to obtain follow-up | 2 (1.2%) | 3 (2.5%) | .65 |
Transportation difficulty | 6 (3.7%) | 9 (7.6%) | .15 |
Financial difficulty | 6 (3.7%) | 4 (3.4%) | 1.00 |
Any barrier during encounter | 42 (25.8%) | 32 (26.9%) | .83 |
>25% of patients with technology dependence experience a barrier to discharge. Lack of a caregiver for discharge from the hospital was more frequently noted in the pandemic period.
Indicates P value <.05.
Subgroup analysis was completed to identify discharge barriers for patients who acquired new technology during hospitalization. Prepandemic, 45 patients were discharged with new technology, and 41 were discharged with new technology during the pandemic. Barriers to discharge were more prevalent in this subgroup, with 47% experiencing at least 1 discharge barrier compared with 17% for those who did not have a new technology requirement (P <.01). Patients with new technology most frequently encountered problems related to home nursing and DME. There was no difference in the number or type of discharge barriers when comparing this group prepandemic and during the pandemic.
Discussion
Our study illustrates that there was no change in unadjusted median LOS in this group of patients across cohorts; however, when accounting for patient-level factors, discharge during the pandemic was associated with a slightly longer LOS. In addition, >1 in 4 patients with technology dependence had at least 1 barrier to hospital discharge documented in their medical records. Identifying a trained and available caregiver for discharge was statistically more likely to be a barrier to hospital discharge during the pandemic.
Of note, the number of hospital discharges for technology-dependent patients at the 2 institutions studied were similar prepandemic and during the pandemic, despite an overall reduction in pediatric hospitalizations during this time.21 Although technology-dependent children may be at higher risk of clinical decompensation, additional factors, such as caregiver exhaustion, lack of home nursing services, and unanticipated delays in follow-up care, likely also contributed to the relatively high prevalence of technology-dependent pediatric discharges during this time frame6,12,14,21 Further study to investigate the reasons for hospital admission in this cohort are needed to determine the likely multifactorial etiology of this observation.
LOS for technology-dependent children is multifactorial and may not be directly linked to medical complexity.5 Our data are consistent with previous literature revealing large variation in LOS; this finding is likely driven by many factors, including more frequent ICU needs and less standardization in care than the general pediatric population.5 The LOS during the pandemic had a wider interquartile range, indicating more variability. This makes it difficult to draw conclusions on the basis of data that do not account for these individual factors. Our adjusted results indicate that the pandemic was associated with a longer LOS. We did not find evidence that technology-dependent children admitted during the pandemic were sicker because there was no increase in the frequency of ICU admission between cohorts and no difference in the prevalence of patients discharged with >5 medications. Previously published literature has revealed no difference in LOS for common pediatric diagnoses during this time frame; however, we found that children with technology dependence were affected differently at our study sites when adjusting for patient-level factors.22
There was a sense of urgency in hospital discharge during the early stages of the pandemic, which could have increased productivity in addressing discharge barriers. Families may have felt strongly about expedient hospital discharge during the initial stages of the pandemic because of fear of exposure to COVID-19 in a hospital. Conversely, there was significant uncertainty early in the pandemic regarding follow-up and in-home care availability. In addition, many hospital workers transitioned from in-person to virtual work. These changes could have increased LOS. The small increase in adjusted LOS associated with the pandemic likely reflects complex pressures surrounding hospital discharge early in the pandemic.
Additional factors associated with longer LOS, including indicators of illness severity and younger age, are consistent with previous literature.2 Specific technologies associated with longer (noninvasive respiratory support) or shorter (ventriculoperitoneal shunt) LOS could be related to the intensity of care required for these devices at home. It was interesting to note that those who have extended family in a local area also had a longer LOS. This finding was counterintuitive for the authors; however, one explanation could be that extended family may request additional training, leading to prolonged hospitalization. There may be a reporting bias because those with longer LOS may have more detailed documentation in their inpatient notes. If a patient was hospitalized far away from their home during the early stages of the pandemic, there may have been pressure to return closer to home quickly.
The high prevalence of barriers to hospital discharge is contributing to the health care utilization of CMC.4–6 The observation that an identified and trained caregiver was more likely to be a discharge barrier during the pandemic may reflect the competing demands of caregiving parents as childcare facilities and schools closed and families lost social support in attempts to limit COVID-19 spread.23 Previous studies indicate that there is a high caregiver burden during the hospital-to-home transition process for families of CMC, and the pandemic likely worsened this experience.24,25 Visitor limitation policies in the hospital may have played a role in the difficulty with the completion of required caregiver training for technology-dependent children.26 Lack of home nursing was the most frequently encountered barrier but was unchanged during the pandemic. Home nursing shortages were prevalent before the pandemic and our experience indicates that staffing shortages have worsened as the pandemic has continued12,27,28 It is possible that, if data were collected later in the pandemic, we could see a different outcome.
Patients with a newly acquired technology were more likely to experience a discharge barrier than those with chronic technology use, and this did not change during the pandemic. This is consistent with previous literature evaluating discharge delays for patients with new mechanical ventilation needs.29 Nearly half of this group encountered at least 1 discharge barrier, often related to acquiring new DME or home nursing initiation. We did not further characterize this group between those who had preplanned technologies (such as new surgical g-tubes) and those who urgently required new technology. Further study is needed to better characterize this group with a high burden of barriers to hospital discharge to identify solutions for streamlining their hospital-to-home transition.
Our study had a few limitations. First, this project was limited by the retrospective nature of the review and included only what was documented in the medical record. It is possible that additional barriers were encountered but not documented, which could lead to an underestimation of actual discharge barriers. There could have been a difference in documentation due to fluctuations in patient volume between the time frames. In addition, the move to virtual work for roles early in the pandemic could have changed the quality of the documentation. Second, items such as financial stability, housing, and transportation barriers may be underreported by families because some families may not be forthcoming about their personal hardships to their medical team. Third, the discharge barriers collected were categorized on the basis of previous work from an adult academic hospital and the authors’ clinical experiences.4,29,30 It is possible that different categories of barriers could be relevant to the discharge of the technology-dependent pediatric population because there is not yet a universally accepted framework for discharge barriers in this population.31 Fourth, there was no quantification of the magnitude of time and effort required to overcome barriers to discharge by hospital staff or families. Fifth, we used discharges from January to March 2020 as our control group. Comparing groups from different months could have introduced seasonal variation. Sixth, our analysis assumed that discharge barriers for the same patient may have changed during the pandemic. It is possible that discharges between time frames for the same patient were not independent. Finally, this study was conducted at 2 regional pediatric centers within the same hospital system, which limits generalizability.
Conclusions
This study illustrates the complexity of hospital discharges for technology-dependent children, especially during the COVID-19 pandemic, when home services were limited. LOS for this group of patients was high prepandemic, and the pandemic was associated with a longer LOS when adjusting for patient-level factors. The number of discharge barriers was unchanged during the pandemic. Home nursing was the most frequently encountered barrier to discharge and identifying an available and trained caregiver for home was more difficult during the pandemic. A standardized process to identify caregivers and provide appropriate training early in admission may improve the discharge process for CMC during the pandemic. Barriers to timely discharge place additional burdens on caregivers, impact hospital flow, and increase health care costs. Health policies that address the need for enhanced in-home services for technology-dependent children are required to ensure optimal care for these children. Our quantification of LOS and barriers to hospital discharge during the pandemic illustrates avenues for system-based improvement in the discharge process for CMC. Future work to identify if discharge barriers have a causative impact on LOS would be useful to aid in the identification of interventions that may help CMC and their families transition home. In addition, analysis of the effort required to overcome discharge barriers for both hospital staff and families could provide insight for future systematic improvements in the care of CMC.
Drs Rush and Khan conceptualized and designed the study, drafted the initial manuscript, coordinated supervision of data collection, and reviewed and revised the manuscript; Drs Bloom, Anspacher, Fratantoni, and Parikh designed and tested the data collection instrument, collected data, provided clinical expertise, and input on study design, and reviewed and revised the manuscript; Mr Barber conducted data analysis and drafted, reviewed, and revised the analysis sections of the manuscript; and all authors approved the final manuscript as submitted and agree to be accountable for all aspects of the work.
FUNDING: Funded by the National Institutes of Health (NIH). This project was supported by awards UL1TR001876 and KL2TR001877 from the NIH National Center for Advancing Translational Sciences. Its contents are solely the responsibility of the authors and do not necessarily represent the official views of the National Center for Advancing Translational Sciences or the NIH. The NIH had no role in the design and conduct of this study.
CONFLICT OF INTEREST DISCLOSURES: The authors have no conflicts of interest to report.
Comments