Frequency of Hospital Use Before and After Home-Delivery Meal by Meals On Wheels, of Tarrant County, Texas

Abstract

Background

Patients recently discharged from the hospital are vulnerable and are at high risk for readmission. Home-delivered meals may be beneficial in improving their health and facilitating independent living in the community. The purpose of this study was to identify the association between home-delivered meals and use of hospital services.

Methods

This study includes 120 clients recently discharged from an inpatient hospital stay or from an emergency department (ED) visit who received meal services from Meals On Wheels, Inc., of Tarrant County. Healthcare utilization data was extracted from the Dallas-Fort Worth Hospital Council Foundation, a regional all claims database used by over 90% of hospitals in Dallas-Fort Worth area. Signed tests and generalized linear models (GLM) were performed.

Results

A total of 16,959 meals were delivered from March 2013 through March 2014. Each client received an average of 6.19 meals per week. The average number of ED visits decreased from 5.03 before receipt of meals to 1.45 after receipt of meals, z = -5.23, p <.001. The average number of hospitalizations decreased from 1.33 to.83, z = -7.29, p <.001. The average length of stay per hospitalization decreased from 5.47 days to 2.34 days, z = -5.84, p <.001. Clients who received more meals were less likely to experience ED visits and hospitalizations after controlling for demographic characteristics and levels of physical functioning.

Conclusion

The findings of this study indicate that home-delivered meals services may contribute to a reduction in hospital based care services among frail and vulnerable adults. Additional studies should consider the short and long-term effects of home-delivered meals services on healthcare utilization and the potential to decrease healthcare costs.

Introduction

With increasing life expectancies, the number of older adults is expected to grow globally over the next several decades, from 841 million in 2013 to more than 2 billion in 2050 (1, 2, 3). With the increasing number of older adults, there is an increased prevalence of diseases and disability (4) and increased healthcare utilization in this population. Older adults represent the majority of hospitalized inpatients with 40% of hospitalized patients in 2013 aged 65 or older in U.S. (5, 6, 7).

Patients discharged from hospital, especially older adults, are more likely to face the burden of diseases, functional decline, and risk factors for frailty, disability, re-hospitalization, and mortality (8, 9, 10, 11, 12, 13). Covinsky and colleagues found that many older adults at hospital discharge showed decreased functioning in activities of daily living (ADL) compared to levels at hospital admission (14). Brown et al. also concluded that low mobility is common in hospitalized older patients and is associated with adverse outcomes such as mortality (8). Nutritional deficits are often observed in patients admitted for hospital care (15, 16, 17, 18) and significantly contribute to muscle dysfunction, increased falls, and loss of independence (8, 19, 20). Several reports indicate that decreased food intake during hospitalization also contributes to weight loss, depression, functional decline, and re-admission (21, 22).

Given the multiple threats from under/malnutrition, several studies have demonstrated the benefits of optimizing energy and protein intake at discharge (21, 22, 23, 24). Furthermore, a recent review found that after hospital discharge, nutritional interventions have positive effects on nutritional status (25). In this regard, a home-delivered meals program can deliver nutrition to improve health and well-being at the community level.

Meals on Wheels (MOW) is the largest national organization to support older adults and their nutritional status. The Older Americans Act (OAA) Nutrition Program partially funded the home-delivered meals program, which served meals to more than 840,000 clients all across the U. S. in 2013 (26). The purpose of the OAA Nutrition Program is “1) to reduce hunger and food insecurity among older individuals, 2) to promote socialization of older individuals, 3) to promote the health and well-being of older individuals, and 4) to delay adverse health conditions for older individuals” (27). Home-delivered meals by MOW have helped to increase independent living (28), maintain health, and facilitate recovery from disease among homebound or frail older adults (29).

Earlier studies have addressed the importance of collaboration between medical and community-based systems and/or services and its benefits in health promotion and preventive care at the community level. For example, Sahyoun and colleagues (30) emphasized that improving transitions along the continuum of nutrition care for older adults may contribute to reducing healthcare costs and avoidance of repeated hospitalization or re-institutionalization. Therefore, there is a need to improve transitional and supportive care to reduce risk of adverse outcomes including frequent readmission (31) and to further explore the direct association between healthcare utilization and meal services from community-based organizations.

The goal of this article is 1) to link home-delivered meals data from Meals On Wheel Inc., of Tarrant County, Texas (MOWI) with healthcare utilization data from a regional claims database for patients recently discharged from an inpatient hospitalization or emergency department visit, 2) to document clients' healthcare utilization before and after receiving meal services for six months, and 3) to examine the relationship between meal delivery and healthcare utilization for the six month period.

Methods

Participants

This study included individuals served by MOWI of Tarrant County with funds provided by Meals On Wheels Association of America (MOWA) in support of the “Expand the Vision” initiative of MOWA, which aims to end senior hunger by 2020 (32). The study targeted individuals who were discharged from a hospital stay or emergency department care. Home-delivery meals were provided within two weeks of discharge. All individuals served met the agency's standard eligibility criteria of being homebound, mentally or physically disabled, and unable to prepare meals for themselves. A total of 120 clients were identified and enrolled for this study. Diagnoses were identified from the claims data: chronic conditions (e.g., coronary heart disease, diabetes, and heart failure), acute conditions such as back pain, neck pain, or regional infection for ED visits; chronic conditions (e.g., chronic kidney disease, coronary heart disease, hip/pelvic fracture, and hypertension), mental disorder, and acute pulmonary diseases for inpatients.

Data Sources

This study explored the relationship between data on home-delivered meals collected by a local MOWI with healthcare administrative data (hospital reported utilization data) in a group of individuals residing in the Dallas Fort Worth metropolitan area of Texas. MOWI served as the data source for individual demographic information and meal delivery data (e.g., number and types of meals). Administrative data on individual hospital based care was extracted from the Dallas-Fort Worth Hospital Council (DFWHC) Foundation. The DFWHC Foundation securely houses the information of 8 million regional patients and their 28 million hospital encounters in the claim records of 90% of the hospitals in North Texas. The claims record contains the patient's demographic data, payer type, diagnostic codes, procedure codes, charges, Current Procedural Terminology (CPT) codes, severity of disease and other information available in the claims data warehouse (33). Identifiable information (e.g., first and last name, address, zip codes, date of birth) of clients enrolled in the grant were sent from MOWI to the DFWHC Foundation who matched MOWI clients with individuals' hospital claims data. All 120 MOWI clients were matched. The research team received de-identified data including demographic information of clients, types and number of meals, and claims data for the analysis. Institutional Review Board (IRB) approval was obtained at Baylor Scott & White Health to analyze the deidentified secondary data.

Measures

Healthcare utilization. Three variables were used to measure healthcare utilization: frequency of ED visits, frequency of inpatient hospital stays, and average length of stay per inpatient hospital stay (if applicable). Variables were captured in the 6 months prior to and after home-delivered meals provided by MOWI. For clients within an inpatient admission in the 6-month period prior to initiating services, data regarding length of stay was accessed for the 6-month period before services and 6-month period after services.

Age and sex. Age was coded as a continuous variable based on a participants' birth year and ranged from 42 to 94. Sex was scored 0 if the participant was male and 1 if the participant was female.

Meal history. Total number of meals provided was counted. Each participant had different duration of services. The number of meals served for each individual from March 2013 through March 2014 ranged from 1 to 636. Seven types of meals were also identified: noon meals, breakfast meals, holiday meals, weekend meals, shelf stable meals, frozen meals, and supplement.

Emergent care assessment score. Emergent care score was collected using the Risk Factors for Hospitalization and Emergent Care Assessment tool (34) when clients enrolled in the MOWI service. The tool ranks 22 items in three categories (prior pattern, chronic conditions, and risk factors). The tool scores as yes (1) or no (0) for each item; a total score of six or more indicates high risk for emergent care.

Nutritional risk screening. Nutritional status among clients was assessed by the National Screening Initiative (NSI) Checklist (35, 36) upon enrollment in the MOWI service. The NSI was developed through a collaborative effort between the American Diabetic Association, the American Academy of Family Physicians, and the National Council on Aging. The checklist, composed of 10 items, has been utilized as an awareness and educational tool for an elderly population in various settings (37). Three risk levels can be determined by the summary score: 0-2: good; 3-5 moderate nutritional risk; 6+: high nutritional risk.

Functional impairment. Functional impairment was measured by 14 questions assessing activities of daily living (ADL) and instrumental activities of daily living (IADL). Community-based organizations supported by the Texas Department of Aging and Disability and Disability Services (TDADS) use the 2060 Form to assess individual capacity for self-care (38). Questions on ADL measure the ability to perform one or more of the following six activities of daily living without personal assistance, stand-by assistance, supervision or cues: eating, dressing, bathing, toileting, transferring in and out of bed/chair, and walking. IADL refers to the inability to perform one or more of the following eight instrumental activities of daily living without personal assistance, stand-by assistance, supervision or cues: preparing meals, shopping for personal items, medication management, managing money, using telephone, doing heavy housework, doing light housework, and transportation ability (38). All 14 items were scaled so that 0 = no functional impairment; 1 = mild or minimal impairment; 2 = severe or extensive impairment; or 3 = total impairment. Functional impairment was scaled so that higher scores indicate higher levels of impairment.

Data Analysis

Descriptive analyses were conducted for personal characteristics and number of meals. The Signed tests for non-parametric variables (i.e., ED visits, hospitalization, average length of stay during hospitalization) were employed to examine the change in healthcare utilization before and after meal services. Statistical significance was examined at the level of 0.05. Generalized linear models with a Poisson distribution were also used to evaluate the association between healthcare utilization and the number of meals.

Results

Sample Characteristics

Table 1 summarizes characteristics of study participants. The range of age of the participants is from 42 to 94 years with an average of 71.5. The majority of participants were female (60.0%), and over three quarters of the participants were White/Caucasian. More than 30% were widowed. A quarter of the participants did not have impairment performing 6 activities of daily living; 70% mild impairment; and 5% severe impairment. Only 1.7% of participants had no impairment performing 8 instrumental activities of daily living, while a majority of participants (90.0%) had minimal or mild functional impairment, and 8.3% had severe impairment. The average Emergent Care score was 6.27, suggesting clients were at high risk of emergent care needs. Majority of participants (61.7%) reported 6 or more scores. The average nutritional risk score was 8.90 (sd = 4.00). Only one client had good nutritional status. Almost 80% of clients (79.1%) were at high risk (6+) in nutritional status. A total of 16,959 meals were delivered to the clients. The average number of meals that each individual received was 141 meals (sd = 128.4). Over 95% of participants received noon meals, and 80.0% and 69.2% of participants received holiday meals and breakfast meals, respectively. Over half of the participants (50.8%) also received shelf-stable meals. Although not shown in Table 1, participants with higher levels of impairment in ADL and IADL received more meals (189.33 meals with severe impairment in ADL; 209.70 meals with severe impairment in IADL) compared to their counterparts (138 meals with no impairment and mild impairment in ADL; 90.50 meals with no impairment in IADL; 135.94 meals with mild impairment in IADL).

Table 1.

Characteristic of the Study Participants

Characteristics	Frequency	(%)
Age [Range: 42–94]	M = 71.5	(sd = 12.6)
Sex
Male	48	(40.0%)
Female	72	(60.0%)
Marital Status*
Married	25	(22.1%)
Widowed	35	(31.0%)
Divorced	30	(26.5%)
Single	16	(14.2%)
Other	7	(6.2%)
Race/Ethnicity
White/Caucasian	91	(75.8%)
Hispanic	6	(5.0%)
Black/African American	23	(19.2%)
Nutritional Status* [Range: 2–19]	M = 8.90	(sd = 4.00)
Good (0–2)	1	(0.9%)
Moderate risk (3–5)	23	(20.0%)
High risk (6+)	91	(79.1%)
Emergent care assessment [Range: 0–10]	M = 6.27	(sd = 2.10)
0–5	46	(38.3%)
6+	74	(61.7%)
Activities of Daily Living (ADL)
No functional impairment	30	(25.0%)
Minimal/mild functional impairment	84	(70.0%)
Severe/extensive functional impairment	6	(5.0%)
Characteristic of the Study Participants
Instrumental Activities of Daily Living (IADL)
No functional impairment	2	(1.7%)
Minimal/mild functional impairment	108	(90.0%)
Severe/extensive functional impairment	10	(8.3%)
Total Number of Meals Provided	16,959
Number of Meals Received at Individual [Range: 1–636]a	M = 141	(sd = 128.4)
1-37 meals	40	(25.0%)
38-94 meals	40	(25.0%)
101-198 meals	24	(20.0%)
205-696 meals	36	(30.0%)
Type of Meals Received at Individual
Breakfast meals	83	(69.2%)
Holiday meals	96	(80.0%)
Weekend meals	17	(14.2%)
Shelf stable meals	61	(50.8%)
Frozen meals, Supplements	11	(9.1%)

^*Note. Sum of frequencies may vary due to missing; a. The number of meals was served from March 2013 through March 2014

Changes in Healthcare Utilization

To examine the association of meal services and healthcare utilization, the Signed Test was used for ED visits, hospitalizations, and the average length of stay during hospitalization. The average number of ED visits decreased from 5.03 before receiving meals to 1.45 after receiving meals. Specifically, 60 participants (50.0%) showed reduced ED visits, 14 participants (11.7%) had an increase in ED visits, and 46 participants (38.3%) remained unchanged in the number of ED visits (z = -5.23, p < .001). The average number of hospitalizations decreased from 1.33 to 0.83. Seventy-one participants (59.2%) showed reduced hospitalizations; 6 participants (5.0%) had increased hospitalizations, and 43 participants (35.8%) remained unchanged in their number of hospitalizations when comparing before and after meal services (z = -7.29, p < .001). The average length of stay during hospitalization also decreased from 5.47 days before receiving meals to 2.34 days after receiving meals. Eighty one participants (67.5%) showed reduced average length of stay per hospitalization; 21 participants (17.5%) had increased length of stay, and 3 participants (2.5%) remained unchanged in their length of stay when comparing before and after meal services (z = -5.84, p < .001) (Table 2).

Table 2.

Comparisons of Mean Scores in Healthcare Utilization 6 month Before and After Meal Service

Healthcare Utilization	n	Mean (SD)		p-value
		Before	After
ED visits	120	5.03 (±10.71)	1.45 (±1.98)	.000a
Hospitalization	120	1.33 (±.85)	.83 (±1.48)	.000a
Length of Stay	105	5.47 (±3.33)	2.34 (±3.60)	.000a

^aSign test was used; Note. ED: Emergency Department

Table 3 shows the comparisons of distribution of healthcare utilization 6 months before initiation of meals services and after completion of meal services. Over 32% of participants did not visit the ED; 24.2% visited once; 10.8% twice, and 32.5% visited the ED 3 or more times for 6 months before meal services. The distribution changed after meal services. Fifty participants (41.7%) did not visit the ED, and only 19% visited 3 or more times after meal services. Specifically, the proportion of non-hospitalized participants was 12.5% before meal services, which increased to 61.7% after meal services. One hundred five participants stayed at least one day before meal services; less than half of participants (N= 46, 43.8%) stayed at the hospital after meal services.

Table 3.

Comparisons of Healthcare Utilization 6 month Before and After Meal Service

Healthcare Utilization	Before	After
ED visits	120	120
None	39 (32.5%)	50 (41.7%)
Once	29 (24.2%)	24 (24.2%)
Twice	13 (10.8%)	18 (15.0%)
3+ times	39 (32.5%)	23 (19.2%)
Hospitalization	120	120
None	15 (12.5%)	74 (61.7%)
Once	63 (52.5%)	24 (20.0%)
Twice	31 (25.8%)	10 (8.3%)
3+ times	11 (9.2%)	12 (10.0%)
Average Length of Stay	105	105
None	-	59 (56.2%)
1-2 days	17 (16.2%)	8 (7.6%)
3-4 days	35 (33.3%)	19 (18.1%)
5-6 days	23 (21.9%)	7 (5.8%)
7+ days	25 (28.6%)	12 (11.4%)

Note. ED: Emergency Department

Table 4 shows the results of generalized linear models with a Poisson distribution for ED visits, hospitalizations, and length of stay. Total number of meals served was significant for change in ED visits (OR = .996 per meal; 95% CI: 0.993 – 0.999) and hospitalizations (OR = .998 per meal; 95% CI: 0.997 – 1.000). Receiving one meal decreased the risk of visiting the ED by 0.4% and decreased the risk of being hospitalized by 0.2%. Total number of meals was not significant for the length of stay.

Table 4.

Factors Associated with Healthcare Utilization 6 month Before and After Meal Service

Variables			ED visits					Hospitalization					Length of Stay
	B	SE	OR	95% CI		B	SE	OR	95% CI		B	SE	OR	95% CI
Age	-.018	.013	.982	.958	1.007	-.022	.008	.978**	.962	.994	-.004	.006	.996	.984	1.008
Sex (Female = 1)	-.445	.346	.641	.325	1.263	-.078	.165	.925	.669	1.280	-.059	.126	.943	.736	1.208
Marital Status (Married = 1)	-.474	.327	.622	.328	1.181	.092	.173	1.096	.781	1.539	.108	.148	1.115	.835	1.488
Race/Ethnicity (White = 1)	-.489	.456	.613	.251	1.499	.424	.203	1.527*	1.026	2.273	.152	.162	1.165	.847	1.600
ADL	.026	.053	1.027	.925	1.139	.010	.037	1.010	.940	1.085	.039	.031	1.040	.978	1.106
IADL	-.134	.051	.874**	.791	.966	.018	.040	1.018	.942	1.101	-.050	.032	.951	.893	1.014
Number of Meals	-.004	.002	.996*	.993	.999	-.002	.001	.998*	.997	1.000	-.001	.001	.999	.999	1.000

^*p < .05

^**p < .01

***p < .001; Note. ADL: Activities of Daily Living; IADL: Instrumental Activities of Daily Living; ED: Emergency Department; Age, gender, marital status, race/ethnicity, and ADL/IADL were controlled.

Discussion

Using a sample of clients who recently discharged from an inpatient hospitalization or emergency department visit, we linked home-delivered meals data from Meals On Wheel Inc., of Tarrant County, Texas (MOWI) with healthcare utilization data from a regional claims database. In addition, we examined clients' healthcare utilization before and after receiving home-delivered meals services for 6 months. Participants who received home-delivered meals showed a significant decrease in the average number of ED visits and the average number of hospitalizations from the 6 month period before receiving meals to the 6 month period after completion of receiving meals. The average length of stay among participants hospitalized before receiving meals decreased from 5.47 days to 2.34 days after receiving meals. Moreover, receiving more meals was positively associated with reducing ED visits and hospitalizations among patients recently discharged from area hospitals.

This study is significant for a variety of reasons. Firstly, the results of this study show that rates of healthcare utilization are lower for clients after they receive home-delivered meals from MOWI compared to their rates before receiving meals. The home-delivered meals program is a popular program for delivery of nutrition and supportive services in the U.S. (39) As shown in a recent review, among a variety of improved outcomes from the home-delivered meals program, the biggest benefit is receiving meals and nutrition, a well-known and critical factor for health at any age stage (40). Nevertheless, the relationship between home-delivered meals services and healthcare utilization has rarely been studied. Results of our study suggest that individuals receiving more meals are less likely to be hospitalized or go to the ED. Almost 20% of Medicare patients discharged from a hospital are readmitted within 30 days (41). In 2004, unplanned readmissions accounted for 17% of total hospital payments from Medicare at a cost of 17.4 billion (41, 42). Our results potentially have significant implications for the healthcare system as a whole. Home-delivered meals through local supportive services could provide balanced and nutritious meals to home-bound older adults, which can optimize energy, reduce functional decline, and allow participants to continue to live independently in their community (29, 42, 43, 44, 45, 46, 47, 48, 49). This would prevent potential readmissions, reduce healthcare costs, (28, 49), and eventually contribute to improved quality of life among patients (49, 50).

Second, this study utilized objective measures on healthcare utilization extracted from a regional claims database for hospitalizations or emergency department visits. To our knowledge, no study has investigated healthcare utilization by using objectively collected data. Many community-based programs and/or studies have used self-reported measures and/or outcomes such as chronic conditions or number of ED visits within the previous week or month; however, some researchers have identified as a study limitation in that it might produce recall bias (40). Furthermore, Thomas (39) suggested an increasing need for a variety of valid and standardized measures to provide evidence for the impact of home-delivered meals programs. Measures from different sources (e.g., using existing data) would be beneficial for a variety of reasons (39). The impact of home-delivered meals programs on standardized outcomes such as health, functional, and healthcare-related outcomes would demonstrate the direct effect of the program. Linking home-delivered meals program data and other outcomes would be informative to payers and policymakers. As such, this study linked meal service history to administrative claims data, which can provide valid and comprehensive healthcare utilization patterns and contribute to establishing a plan of care for patients recently discharged from the hospital.

Finally, this study also supports the idea that receiving home-delivered meals may be beneficial in reducting healthcare utilization for participants with higher level of impairments. As shown in Table 3, the impairment level in IADL was significant for change in ED visits (OR = .966 per score; 95% CI: 0.993 – 0.999) although it was controlled in the GLM. The IADL measures the inability to perform 8 instrumental activities of daily living without personal assistance and one of the activities is preparing meals. In other words, this may indicate an indirect effect of receiving meals on healthcare utilization, especially ED visits. This result is consistent in existing literature (51, 52). Individuals who are incapable of doing their own shopping or preparing meals are more likely to be in a poor nutritional status and have an increased risk of adverse health outcomes (53). Thus, delivery of meals with nutritious and dietary variety would be beneficial for those with high level of impairment to avoid physical and mental burden of preparing meals which eventually might contribute to preventing ED visits.

Limitations

Although findings from this study have important implications for healthcare utilization and community supportive services in home-delivered meals, it is important to note the limitations of this study. First, this study does not include a control group so it was impossible to examine the impact of home-delivered meals on healthcare utilization or draw a causal relationship between home-delivered meals and healthcare utilization. Future projects should consider including a control group to assess the causal relationship. Second, the study sample was derived from a single geographic area of the United States. Other clients in different regions may present different patterns of healthcare utilization and meal services. Furthermore, while the data source does capture 90% of all healthcare utilization within the DFW area, healthcare data for services rendered by providers outside of the DFWHC Foundation area were not available and could therefore not be included. We also acknowledge the probability of underestimating prevalence or rates of healthcare utilization beyond the 6 month window because we examined only 6 months prior to and after receiving meals. Examination on meal delivery services using claims data with a variety of timeframes would be considered in the future projects. Third, the data from MOWI was collected by the MOWI staff for a variety of service delivery, tracking, billing and health information purposes. Healthcare utilization data was also administered by a regional data repository institution. We used their existing data and linked two data sources together, but were not involved in data collection. Therefore, we are not able to assess variables that can explain compounding effects on the relationship between meal services and healthcare utilization such as quality of services, severity of chronic conditions, psychological wellbeing (e.g., depression or loneliness), social support, living arrangement, and socioeconomic status of patients (39, 54, 55). Future projects would be considered to design a comprehensive and research-oriented study plan.

Conclusion

This study provides evidence that home-delivered meals services may contribute to reductions in healthcare utilization among vulnerable MOWI clients. As the importance and demand of home-delivered meals for MOWI clients grows, home-delivered meals services could be beneficial in many ways, including improving nutrition and quality of life and reducing healthcare costs. Additional studies should consider the long-term effects of home-delivered meals services on healthcare utilization and the potential to decrease healthcare costs. This would contribute to decisions about how to maximize limited resources at the community level as well as national level.

Conflict of Interest

All authors declare no conflicts of interest.

Ethical standard

This study was approved by the Institutional Review Board (IRB) at Baylor Scott & White Health.

Sources of Funding

This study included individuals served by MOWI of Tarrant County with funds provided by Meals On Wheels Association America (MOWA) in support of the “Expand the Vision” initiative of MOWA.