Causes of readmissions for patients discharged on enteral nutrition

Abstract

Background and Aims:

Patients discharged with enteral nutrition through an enteral access device (DCEN) are noted to have increased hospital readmissions but data on the causes for readmissions are limited. We assessed the proportion of these readmissions attributed to enteral nutrition (EN) and determined the contributing factors to readmissions.

Methods:

Using electronic health record data, we conducted a retrospective cohort study of all hospital encounters in an academic, urban hospital from 7/2017 - 12/2019 with DCEN to find all unplanned readmissions at the same hospital within 90-days. For each readmission, we conducted chart review to evaluate whether discharge documentation identified the primary cause of readmission to be EN-related and evaluated for EN-related plan adjustments upon discharge.

Results:

Over the 30-month period, there were 224 and 442 readmissions occurring within 30 and 90 days of DCEN respectively. EN-related readmissions accounted for 20.5% of 30-day readmissions and 16.7% of 90-day readmissions. Among EN-related readmissions, 44.6 % (33/74) documented enteral access device (EAD) issues, 40.5% (30/74) cited gastrointestinal symptoms the team attributed to EN, and 14.9% (11/74) attributed sodium imbalance. The EN plan was changed in 97.3% (72/74) of EN-related readmissions and 32.3% (119/368) of EN-unrelated admissions. 52.0% of 90-day readmissions were within 90 days of initiating EN.

Conclusion:

More than 20% of patients discharged with EN have readmissions related to EN, many within 90 days of initiating EN. Quality improvement interventions targeting post-discharge care to address these causes may decrease hospital readmissions in this high-risk and medically complex population of patients.

Introduction

Enteral nutrition (EN) delivered via an enteric access device (EAD) is employed to treat severe malnutrition as well as feed patients with pathology restricting oral intake. A 2017 study estimated that approximately 250,000 adults in the United States are using EN at home¹, which is a dramatic increase from an estimate of 152,000 total adult and pediatric patients in 1992². While there are many studies evaluating the role of EN during hospitalization³, there are limited data on the clinical outcomes of patients discharged from the hospital with EN, despite this growing population. Patients discharged with EN are noted to have increased hospital readmissions⁴ but data on the causes for readmissions are limited. Understanding the causes may allow for designing interventions to reduce readmissions.

The primary aim of this study was to assess the frequency, timing and proportion of readmissions related to EN in patients discharged with EN through an enteral access device (DCEN), and categorize the causes for these readmissions. Regarding non-EN related readmissions, a secondary aim was to assess the frequency of EN plan adjustment as a surrogate for the need for EN plan optimization.

Methods

Study setting and population

Using electronic health record (EHR) data, we conducted a retrospective, cohort study of all hospital encounters at the Hospital of the University of Pennsylvania (HUP) from 7/2017 - 12/2019 with DCEN to find all readmissions at HUP within 90-days. We excluded encounters that were planned readmissions (i.e., for scheduled surgeries or scheduled infusions) or readmissions at hospitals other than HUP. For patients transferred from other institutions, we maintained the date of initial presentation. These were collected by using an existing registry of all hospital discharges and evaluating for readmissions.

HUP is an academic, tertiary care 800-bed hospital with an on-site Clinical Nutrition Support Service (CNSS), a multidisciplinary team including registered dietitians that manages all inpatients referred for nutrition-related needs. All patients initiated with EN are evaluated and managed by the CNSS until discharge. A nutrition-focused data registry collects discrete data on EN-related care, including provider assessments and recommendations. All recommendations regarding care are documented in progress notes by the team in the EHR as well, and the care plan upon discharge is documented by the primary provider in the discharge summary. There is no coordinated outpatient EN support service within the healthcare system.

Data collection

Each readmission was chart reviewed by a physician (SP) and research assistant (AP) to obtain the chief complaint and hospital course summary, days to readmission from prior discharge, demographics (age, sex), and co-diagnosis of malnutrition during initial assessment by CNSS⁵. We evaluated whether discharge documentation identified the primary cause of readmission to be attributed to EN.

For EN-related readmissions, we obtained from the documentation: presence of malnutrition co-diagnosis; initial admission chief complaint; days to readmission; days since EAD insertion as a proxy for EN initiation; EAD- related procedures (surgery or interventional radiology); and EN-related plan adjustments and actions taken to address the causes of readmission upon discharge (procedure, medications, formula type or rate change, water flush change). From the chief complaint and documentation, we categorized the cause of admission to be EAD issues, GI symptoms attributed to EN (nausea, vomiting, diarrhea, abdominal pain), sodium imbalance (hyponatremia or hypernatremia), constipation (which was not clearly attributed to EN), and other.

For non-EN related readmissions, we obtained: cause of readmission (aspiration pneumonia, bacteremia, other infection, stroke, GI bleeding, other) and whether there was a documented change in the EN care plan upon discharge in the categories as described above.

Study data were collected and managed using REDCap (Research Electronic Data Capture) tools hosted at the University of Pennsylvania^6,7.

Outcome variables

For the primary aim, the primary outcome variables were whether the readmission was related to EN, the number of days to date of readmission from EAD placement as a proxy for EN initiation, and the categorized cause for readmission. For the secondary aim regarding non-EN related readmissions, the outcome variable was whether there was an EN-related plan adjustment upon discharge.

Statistical Analysis

To evaluate the demographic and clinical differences between EN-related and non-EN related readmissions, we used a Wilcoxon rank-sum test for continuous variables and Fisher’s exact test for binary and categorical variables. All analyses were performed using STATA version 13.0 (College Station, TX) and Excel. This study was reviewed and determined to qualify as Quality Improvement by the University of Pennsylvania’s Institutional Review Board.

Results

Over the 30-month period, there were 224 encounters for unplanned readmissions within 30-days of DCEN and 442 within 90-days of DCEN. EN-related readmissions accounted for 20.5% (46/224) of 30-day readmissions and 16.7% (74/442) of 90-day readmissions.

For 90-day readmissions, the median number of days from EN initiation to readmission was 86, where 52.0% (230/442) were within 90 days and 16.0% (71/442) were within 30 days of EN initiation. For EN-related readmissions, 56.9% (41/72) and 20.8% (15/72) were within 90 and 30 days of EAD placement respectively. Median length of stay of all readmissions was 7 days (IQR 4-14) (table 1). 90-day readmissions that were EN- related, compared to those that were not, had a higher proportion of females (p < 0.001) and a higher proportion of a co-diagnosis of malnutrition (0.034).

Table 1:

Demographics of all readmissions within 90 days of DCEN. Values for age, days to readmission, days since EAD placement, and readmission length of stay are the median (IQR). The following statistical tests were utilized: Wilcoxon rank-sum test (age, days to readmission, days since EAD placement, and readmission length of stay), and Fisher’s exact test (sex, malnutrition co-diagnosis, discharge location).

		ALL	NON-EN RELATED	EN-RELATED	P-VALUE
N		442	368	74
Age		63 (50, 72)	63 (49, 71)	63 (55, 73)	0.39
Sex (female)		171 (38.7%)	127 (34.5%)	30 (59.5%)	<0.001
Co-diagnosis of malnutrition		287 (64.9%)	231 (62.8%)	56 (75.7%)	0.034
Discharge location
Home with homecare		169 (38.2%)	138 (37.5%)	31 (41.9%)	0.68
Skilled nursing facility		119 (26.9%)	103 (28.0%)	16 (21.6%)
Home		86 (19.5%)	70 (19.0%)	16 (21.6%)
Rehab center		64 (14.5%)	54 (14.7%)	10 (13.5%)
Other/unknown		4 (0.9%)	3 (0.8%)	1 (1.4%)
Days to readmission		27 (8, 50)	30 (8, 51)	19 (8, 36)	0.15
Days since EAD placement		86 (44, 245)	86 (45, 252)	74.5 (35.5, 238)	0.28
Readmission length of stay		7 (4, 14)	7 (4, 14)	7 (5, 14)	0.51

EN-related admissions

Among EN-related readmissions, 44.6% (33/72) documented EAD related issues, 40.5% (30/74) cited symptoms the team attributed to EN, and 14.9% (11/74) attributed sodium imbalance from dehydration (Figure 1). Four patients with GI symptoms also had EAD issues. One admission with GI symptoms also had sodium imbalance, while one other admission with GI symptoms also had constipation. Only 5.6% of all EN-related readmissions (4/72) were not attributed to any of these three issues (3 for failure to thrive not attributed to symptoms, and 1 for hypoglycemia in setting of EN not running).

Figure 1:

Causes for readmission. X axis represents number of encounters, of total N = 74. EAD: enteric access device. Each encounter can have more than one cause listed.

During the readmission, 44.6% (33/74) addressed the cause of readmission with a procedure regarding the EAD (figure 2). This included tube exchange to address malpositioning or local symptoms (20), conversion in the tube entry site (8), size modification (2), and declogging (4). Oral medications used were primarily anti-emetics (12/74), and also included laxatives, loperamide, aprepitant, and gabapentin. Bedside procedure includes one nasogastric tube placement and two EAD declogging.

Figure 2:

Methods utilized during hospitalization to address readmission cause. X axis represents number of encounters, of total N = 74. Each encounter can have more than one method listed.

EN-unrelated readmissions

There were 368 readmissions with a primary indication not related to EN. Causes for readmission included bacteremia (11.1%), aspiration pneumonia (5.2%), other infection (27.8%), gastrointestinal bleeding (3.2%), and stroke (2.0%).

Of these EN-unrelated readmissions, 32.3% (119/368) involved a change in the EN plan by CNSS by time of discharge. 23.5% (28/119) involved a change in oral diet recommendations, where 25 patients had their diet advanced and 3 had their diet more restricted. EAD related procedures occurred in 21.8% (26/119). Changes in formula or water flush occurred in 43 cases, of which 13 involved increasing the rate to meet caloric needs better. CNSS recommended symptom management medications in 13 patients, including 9 for constipation.

Discussion

Hospital readmissions adversely impact patients’ quality of life and are costly. As such, hospital readmissions are an established quality measure by many groups including the Centers for Medicare and Medicaid Services (CMS)⁸, and strategies are needed to reduce readmission rates. Patients with DCEN are at increased risk for readmissions⁴, but there are very limited data on the causes for readmissions or how to prevent them. In this study, we evaluated the causes of readmission for patients with DCEN with the goal to provide data that can lead to implementing interventions designed to reduce them.

Many hospital and healthcare networks, including ours, lack coordinated outpatient nutrition support services for EN care. Many of the methods utilized during the readmissions to address EN-related readmission causes, like EAD related procedures and formula and water flush adjustments, do not require the hospital setting, and may be routinely performed in the outpatient setting. Our study suggests that close follow up within the first 90 days of initiating EN may identify 52% of the problems leading to readmissions and allow for outpatient interventions that may prevent these readmissions.

Our strict criteria for assessing whether an admission was EN-related is likely to have underestimated the role of EN in readmissions. A high proportion of this population has a co-diagnosis of malnutrition⁴, which also has been associated with higher likelihood for readmissions^9,10 as well as risk for infection^11,12. Among the large proportion of infection-related readmissions, it is feasible that some were partially correlated with their nutrition status, and some may have been avoided by better EN care over a longer period. A sizeable proportion of patients with non-EN related admissions had their nutrition plans adjusted to increase caloric intake, thus likely leading to better addressing nutritional needs and patient quality of life for those with advancement in oral feeds.

A major strength to our study is that this was performed in a large tertiary-care hospital with an active data registry, in which we were able to identify a sizeable population of patients with DCEN and subsequent readmission and evaluate the details of the readmission using chart review. Our results are likely to be generalizable to other tertiary care centers in major metropolitan areas. Whether these results are generalizable to other settings is unknown. Furthermore, the retrospective and observational nature of the study limited the ability to assess the impact of other parameters or confounding factors, like practice differences between providers, discharge location, and tube type.

In conclusion, more than 20% of 30-day readmissions for patients discharged on EN are due to causes attributed to EN. A majority of these readmissions is within 90 days of initiating EN. The three most common causes, accounting for nearly 95% of readmissions, were GI symptoms of intolerance, issues with the EAD, and sodium imbalance attributed to dehydration. A significant proportion of those with readmissions unrelated to EN also had EN plan adjustments, suggesting there was need for EN plan optimization. Quality improvement interventions targeting post-discharge care to address these causes may decrease hospital readmissions in this high-risk and medically complex population.

Clinical Relevancy Statement.

A recent study confirmed that patients discharged with enteral nutrition are noted to have increased hospital readmissions. In this study evaluating for the causes of readmission, we found that more than 20% of 30-day readmissions were related to enteral nutrition, and the three most common causes, accounting for nearly 95% of those readmissions, were GI symptoms of intolerance, issues with the EAD, and sodium imbalance attributed to dehydration. Quality improvement interventions targeting post-discharge care to address these causes may decrease hospital readmissions in this high-risk and medically complex population.

Data availability:

Data and analytic methods can be made available upon request to the corresponding author