Determinants of Frail Elders' Lengths of Stay in Meals on Wheels

Copyright 1999 byThe Gerontological Society of America

The CerontologistVol. 39, No. 4, 397-404

Home-delivered meal programs such as Meals on Wheels are low-cost long-term care servicesthat have potential for contributing to the delay of costly institutionalization and to the

maintenance of self-sufficiency and quality of life among community-dwelling frail elders. In thisstudy, reasons for elders' termination from a Meals on Wheels program and determinants of

their lengths of stay in the program are analyzed. The findings show that the reasons for elders'termination are largely associated with their deteriorating health. But it was also found that a

significantly higher proportion of African American elders than White elders discontinued theirparticipation due to their dissatisfaction with the meals offered or their poor appetite. The Cox

proportional hazard regression analysis also confirmed that elders' health status, race, andappetite are significantly associated with the hazard of elders' termination. Strategies to

improve the acceptability of the services are discussed.Key Words: Frail elders, Nutrition, Home-delivered meals, Service acceptability

Determinants of Frail Elders' Lengthsof Stay in Meals on Wheels

Namkee G. Choi, PhD

Nutritionally balanced meals are essential for eldersto maintain good health and to mitigate existing healthproblems. Especially among frail elders, inadequate in-take of necessary nutrients is likely to accelerate theirchronic health problems and the deterioration of theirfunctional ability, leading to a loss of their indepen-dence and eventually to institutionalization.

Home-delivered meal programs like Meals onWheels, supported by Title Ill-C of the Older Ameri-cans Act (OAA) and other public and private funds,have served the nutritional needs of millions of frail,homebound elders since their inception in 1972. In1994 alone, these programs delivered 113 million mealsto 877,000 homebound elders (Administration onAging [AoA], 1995). By providing at least one nutri-tionally balanced meal daily several days each weekto elders who are unable to obtain or prepare foodwithout assistance, home-delivered meal programs havebeen viewed as having the potential for contributingto elders' ability to maintain self-sufficiency and to delayor completely avoid costly and much-disliked institu-tionalization. Thus, home-delivered meal programs havebecome essential, relatively cheap long-term care ser-vices that are geared to facilitating elders' ability tocontinue living in the community and improving qualityof life among community-dwelling frail elders.

Despite the significance of home-delivered meal pro-grams as major community-based long-term care ser-vices, little research has been done on them, otherthan studies dealing with overall sociodemographics

This research was funded by the Baldy Center for Law and SocialPolicy of SUNY at Buffalo. The author thanks Barbara Burns for her data-collection efforts and Richard Gehring, Benjamin Gair, and Nancy Vigykianfor their support of the study.

Address correspondence to Dr. Namkee C. Choi, Associate Professor,School of Social Work, 359 Baldy Hall, State University of New York atBuffalo, Buffalo, NY 14260-1050. E-mail: [email protected]

and with the satisfaction levels of participants and thedietary effects of the meals on the participants. Spe-cifically, only a couple of studies have been done onfactors associated with elders' utilization of and con-tinued use of home-delivered meal programs. In thisstudy, I analyze reasons for elders' termination from aMeals on Wheels program and determinants of theirlength of stay in the program. Although much of thefocus of previous research has been on current par-ticipants, little is known about why certain elders dis-continue their participation in Meals on Wheels. Byidentifying reasons for their termination and factorsassociated with their length of stay, I intend to dis-cover non-health-related barriers to continued utiliza-tion of the services. Implications of the findings forprogram improvement are discussed.

Previous Studies

Nutritional Problems Among Elders.—Older peoplein general are more likely to be at nutritional risk thanyounger people because of elders' impaired digestion,absorption, or utilization of nutrients due to chronicdisease or drug-nutrient interactions (Horwath, 1991).Age-related changes in gastrointestinal organs canaffect food intake and interfere with digestion andabsorption of nutrients; changes in oral health canaffect nutritional status; the basal metabolic rate de-clines with age; and chronic diseases can affect el-ders' ability to digest, absorb, and utilize nutrients (Ponza,Ohls, & Posner, 1994). Many prescription and over-the-counter medications can interfere with absorptionand utilization of water-soluble vitamins and, to a lesserdegree, of fat-soluble vitamins, which are better storedin the body (Ahmed, 1992). Chronic illnesses and re-sulting physical and functional disabilities also restrict

Vol. 39, No. 4, 1999 397

Downloaded from https://academic.oup.com/gerontologist/article-abstract/39/4/397/550120by gueston 14 April 2018

elders' mobility and, thus, their ability to shop forgroceries and prepare nutritionally adequate meals.Chronic illness ana the depression that frequently ac-companies such illness also decrease appetite. Stressfrom illness or infection may also deplete existing sub-optimal stores of nutrients (Ahmed, 1992).

In addition to these physiological, metabolic, andmedical causes of nutritional risk in old age, poverty,social isolation, and ignorance of the need for a bal-anced diet also put low income elders and elders wholive alone at greater risk for nutritional problems (Posner,Jette, Smith, & Miller, 1993). Analysis of data from thesecond National Health and Nutrition Examination Sur-vey (NHANES II) showed that low-income elders andelderly individuals living alone tend to have lowermedian dietary intakes of several nutrients, subop-timum blood levels of vitamin C and iron, and anincreased prevalence of hypertension (Kuczmarski,1993). In conjunction with their lack of sufficient finan-cial resources, low-income elders are more likely tolack transportation and, therefore, are more likely tohave logistical difficulty with shopping and are less likelyto participate in activities associated with intake of avariety of foods and an adequate quantity of fruits andvegetables (Horwath, 1991). It is also difficult to changeelders' personal food preferences and unhealthy eat-ing habits built up over a lifetime (Ponza et al., 1994).

Study findings also suggest that African Americanelders are at higher risk for poor nutritional status thantheir White counterparts (Bernard, Anderson, & Forgey,1995; Kuczmarski, 1993). Considering that a signifi-cantly higher proportion of African American elderslive at or below the poverty line and that they havemore physical and functional health problems, the find-ing that dietary and nutritional deficiencies are morelikely among tnem is not unexpected.

Effectiveness of Home-Delivered Meals.—The Eld-erly Nutrition Programs (ENPs), funded under Titles IIIand VI of the OAA, target services to elders (60 yearsor older) with the greatest economic or social need.Especially, Title III-C2 home-delivered meal servicestarget frail homebound elders who are unable to pre-pare or who have difficulty preparing meals on tneirown. The 1995 ENP evaluation study conducted byMathematica Policy Research found that the averageage of participants in home-delivered meal programswas 78 years; 48% of the participants were living onincomes below 100% of the official poverty line, and90% of the participants were living on incomes below200% of the official poverty threshold; 16% were low-income (below 100% of the poverty line) minorities;and 60% of the participants were living alone. Thestudy also found that 59% of the participants had threeor more chronic health conditions; 43% had had ahospital or nursing home stay during the previous year;61% took three or more prescription or over-the-counterdrugs daily; and 77% either were unable to performor had difficulty performing one or more activities ofdaily living (ADLs) or instrumental activities of dailyliving (lADLs; AoA, 1995). These socioeconomic andhealth status characteristics of the participants indicatethat they are much more disadvantaged than the gen-

eral U.S. population group aged 60 and older, point-ing to the participants' increased risk for nutritionalproblems. The study found that 88% of the partici-pants had characteristics associated with moderateto high nutritional risk (AoA, 1995).

Home-delivered meals are highly effective in miti-gating the nutritional risk of these frail and vulnerableelders by providing them with approximately 40% to50% of their daily intake of most nutrients (AoA, 1995).Because each meal delivered is required to provideat least 33% of the Recommended Dietary Allowances(RDA), participants who receive two or three mealseach day can eat higher percentages of their RDA.The positive effects of home-delivered meals on par-ticipants' nutritional status has been proven in otherstudies based on community samples (see Coulston,Craig, & Voss, 1996; Locher, Burgio, Yoels, & Ritchie,1997). The AoA study and other previous studies alsofound that the participants were highly satisfied withprogram meal services (see Smith, Mullins, Mushel,Roorda, & Colquitt, 1994).

Home-delivered meal programs provide partici-pants more than just nutritious meals for their physi-cal sustenance. The programs also offer them manyvaluable ancillary services, such as nutrition screeningand assessment, education and counseling, and infor-mation on and referrals to other health and social ser-vices. Volunteers who deliver meals also constituteregular social contacts for the elderly participants, whoare usually confined to and isolated in their homes.Because volunteers are trained to recognize problemsthat might need to be addressed, they function as aregular in-person check-up system for frail elders andare able to detect signs of problems. Locher and col-leagues' (1997) study also found that home-deliveredmeals help participants build and maintain ties withinthe community by promoting their social interaction.That is, participants were found to share their home-delivered meals with their neighbors and receive ap-preciation in return.

Considering these health and social-support bene-fits, home-delivered meal services need to be providedto more frail, homebound elders. But the AoA (1995)study found that 41% of the home-delivered meal pro-grams had a waiting list, with a mean of 85 and amedian of 35 elders on the list. The mean length oftime an elder spends on the waiting list ranged be-tween two and three months, and the median lengthof time on the list was one month. The study alsofound that a substantial number of elders who appearto be eligible to participate in a home-delivered mealprogram do not do so. A study of reasons for nonpar-ticipation by program eligibles found that more than80% of them did not participate because they did notbelieve they needed the meals or other program ser-vices. Only a small proportion (2-7%) of the nonpar-ticipants reported such factors as food preferences, dis-comfort with application, sense of stigma, and lack ofprogram awareness as reasons (Burt, 1993, cited inPonza et al., 1994). Thus, lack of perception of need,rather than lack of information about and acceptabil-ity of the services, appears to be a primary reason fornonparticipation.

398 The Gerontologist


Factors Associated With the Utilization of and Con-tinuance on Home-Delivered Meals.—The analysis ofdata from the National Long-Term Care ChannelingDemonstration, 1982-1984, shows that need factors,such as high numbers of ADL impairments and physi-cal illnesses and history of institutionalization and liv-ing alone, were significant predictors of frail elders'use of Meals on Wheels services (Mui, Choi, & Monk,1998). Apparently, frail elders who live alone needthe service to be able to continue to live in the com-munity. But Mui and colleagues also found that a rela-tively small proportion of frail elders were consumersof Meals on Wheels services.

In addition to the problems of waiting lists andnonparticipation by the eligibles, the effectiveness ofthe meal programs can be compromised if participantsdrop out prematurely. Previous studies of reasons fortermination and factors associated with the length ofstay in home-delivered meal programs are almost non-existent. Only one study (Frongillo, Williamson, Roe,& Scholes, 1987) analyzed factors associated with con-tinued participation in home-delivered meal programs.Based on data from 2,002 elders receiving home-delivered meals in 23 counties of New York in 1984-1985, the researchers found that recent discharge fromhospital, having cancer, and having a nonhip fracturewere associated with probability of discontinuancewithin 6 months. That is, elders with temporary needfor home-delivered meals during recuperation fromnonfatal, nondebilitating illness and elders with ter-minal illness that resulted in death were likely to re-ceive the service for a short duration, whereas elderswith chronic debilitating illness and disabilities (e.g.,stroke, diabetic, and chairbound) were likely to stayon for a longer duration. The study also found that mi-norities continued on the program substantially longerthan Whites.

Because of its cross-sectional data set, Mui and col-leagues' (1998) study could not analyze factors associ-ated with continuance or discontinuance on the pro-gram. In the case of the study by Frongillo and col-leagues (1987), limiting the time frame to 6 monthsrestricts the utility of the findings. In contrast, this studypresented here utilizes longitudinal data that allowanalysis of participants' length of stay in Meals on Wheelsacross a longer period of time.

Data Source and Sample Characteristics

The data for this study were drawn from a county-wide Meals on Wheels program in western New YorkState. Serving a large urban area and surroundingsuburban and rural areas, the program delivers a hotnoonday meal and a cold evening meal to an averageof 1,600 elders per day. Due to a growing demandfor home-delivered meals, most of the elders wereenrolled in the program after a waiting period thathad lasted anywhere from a couple of days to a coupleof months. About 75% of the elderly participants re-ceive the meals 5 days per week, and 9-10% of themalso receive weekend meals. The program offers mealsthat meet the requirements of a large variety of spe-cial diets. The program also delivers meals for nonelderly

disabled persons, but they constitute a small propor-tion of all participants. For this study, a 33% system-atic random sample of all persons who began receiv-ing meals between 1994 and 1996 was selected. Ofthe 545 participants thus chosen, 36 nonelderlydisabled persons were excluded from the study. Thefinal sample of the study comprised 509 persons aged60 and older. An absolute majority of the 509 partici-pants were newcomers to the program; however, asmall number of them had been served by and termi-nated from the same program prior to 1994, but theyhad reenrolled between 1994 and 1996. The sourcesof data consist of initial social work, nutrition assess-ment, and periodic reassessment, done by the pro-gram's social work and nutrition staff, and a summarylog that contained dates of referral, intake and initialassessment, start of meal delivery, reassessment, andtermination.

Of the 509 elders, 328 (64.4%) were no longer par-ticipating at the time of the data collection (June 1997)and 181 (35.6%) were currently receiving the mealservices. Twenty-four of these current participants hadenrolled, terminated, and reenrolled in the programbetween 1994 and June 1997. By the same token,the 328 past participants included a small number ofelders wno had terminated, reenrolled, and termi-nated again during the same period. (The term "par-ticipants" refers to all 509 elders, whereas the term"current participants" refers to those who were cur-rently receiving meals in June 1997, and the term "pastparticipants" refers to those who were no longer re-ceiving the meals in June 1997.)

As shown in Table 1, the average age of the samplewas 78.4 (5D = 7.8) years, with a median of 79 years;64.8% were female, and 35.2% were male; 76.2%were White, 12.0% were Black, and 11.8% were otherraces (including Hispanics). The female participantswere slightly older than the male participants (79.0versus 77.2 years, p< .02), but there was no signifi-cant age difference among races. There was no sig-nificant racial difference in gender distribution either.Marital status distribution showed 57.4% widowed,22.3% married or in a long-term relationship, and 20.3%in other categories (never married, divorced, or sepa-rated). As expected, the married group was significantlyyounger than the widowed and the other unmarriedgroup (73.8 years for the married, 79.8 years for thewidowed, and 79.0 years for the others, p < .05),and a significantly higher percentage of men (31.8%)than women (17.0%) were married. Of the women,68.5% were widowed, compared with 36.9% of themen. The majority of unmarried participants livedalone.

With respect to the health problems of the par-ticipants, data in Table 1 show that 43.2% of the samplehad hypertension. A sizable proportion had diabetes,arthritis, congestive heart failure, chronic pulmonarydisease, joint disorder, and carotid artery disease. Eachparticipant reported having, on average, 3.25 physi-cal health problems. Approximately half of the partici-pants reported that they had been hospitalized duringthe 6 months preceding their enrollment in the pro-gram, and 2.2% had been in a nursing home during

Vol. 39, No. 4, 1999 399


Table 1. Key Sociodemographicand Health Characteristics of Sample

SociodemographicsAge (years)

Mean (SD)Median

Gender (%)FemaleMale

Race (%)WhiteBlackOther

Marital Status (%)Married/long-term relationshipWidowedOther

Health StatusPhysical Health Problems (%)

HypertensionHeart diseaseDiabetesArthritisPulmonary diseaseStroke

Number of ADLs with difficultyMean (SD)Median

Number of lADLs with difficultyMean (SD)Median

Hospitalized in 6 months prior to enrollment (%)Institutionalized in 6 months prior to enrollment (%)Weight change pattern in 6 months prior

to enrollment (%)Stable/positive changeNegative change

Required special diet (%)Degree of appetite

PoorSmallAverageLarge

78.4 (7.8)79

64.835.2

76.212.011.8

22.257.420.4

43.225.424.618.813.610.6

1.4 (1.4)1

5.0 (1.4)5

50.42.2

54.945.152.1

8.527.854.59.2

the same period. Seventy percent of the sample haddifficulty performing at least one ADL, but more than98% of the sample reported they had difficulty per-forming at least one IADL. Only a little more than20% of the participants indicated that they could gooutside the house without assistance. The mean com-bined number of ADL and IADL tasks performed withdifficulty shows that Black elders were more likely tobe functionally disabled than were White elders (7.1for Blacks versus 6.2 for Whites, p < .05). Those inthe "other" racial group, which averaged 6.6 tasks withdifficulty, were not significantly different from eitherWhites or Blacks. Mean weight showed that Blackswere significantly heavier than Whites (163 poundsfor Blacks versus 139 pounds for Whites, p < .05).The others, at 148 pounds, were not significantlydifferent from the Whites and Blacks. Approximately55% of the participants reported that their weight wasstable or had changed in a positive direction (losingweight in the case of being overweight or gaining weight

in the case of being underweight), but 45% reporteda negative weight change, mostly undesirable weightloss, during the 6 months preceding their enrollmentin the program. There were no racial or gender dif-ferences in the weight-change pattern. Approximatelyone third of the participants reported that their appe-tite was poor or small, and 50.7% of Whites, 69.4%of Blacks, and 45.5% of those in the other racial grouprequired a special diet, p < .03. Only a little morethan 20% had their own teeth, more than 70% hadpartial or total dentures, and the rest had no teeth orhad difficulty chewing or swallowing.

Overall, the sociodemographic status of our samplewas very similar to that of the national sample selectedfor the 1994 ENP evaluation discussed earlier, butthe health status of our sample appears to have beeneven more disadvantaged than that of the nationalsample. Moreover, the health status of Black partici-pants appears to have been significantly worse thanthat of their White and other counterparts.

Variables and Method of Analysis

To analyze reasons for termination or case closure,I did bivariate analyses comparing racial and genderdifferences. Length of stay by reason for terminationwas analyzed and compared.

To analyze the predictors of length of stay, I adoptedthe Cox proportional hazard regression model for allpast and current participants (Model I) as well as forpast participants only (Model II), because current par-ticipants' length of stay was censored at the time of thedata collection and the model for past participants wassubject to selection bias. As the dependent variable,length of stay in the program was measured in weeks.Covariates entered in the model were participants'sociodemographics—age; gender (female = 1, male =0); race (White = 1, Black = 2, other = 3); and maritalstatus (married/in long-term relationship = 1, single =0); participants' health status at the time of enroll-ment—whether the participant had been hospitalizedin the 6 months preceding his/her enrollment in theprogram (yes = 1, no = 0); combined number ofADLs/IADLs with difficulty (on a scale of 0-14); weightpattern in the 6 months preceding enrollment in theprogram (no change or positive change = 1, negativechange = 0); and self-reported appetite (on a scale of1 = poor to 5 = large); type of services the participantreceived—number of days per week that meals weredelivered, and whether the participant required aspecial diet; and monthly amount of contributionthe participant made to the program (divided by thenumber of days that he/she received meals).

The monthly amount of contribution—which asocial worker had negotiated with each participant basedon his/her ability to pay—was used as a proxy for theparticipant's economic status. (In 1997, the maximumsuggested daily contribution was $4.70.) Further analysisshowed that there was no gender difference, but therewas a significant racial difference in the average amountof contribution: The "other" racial participants madea significantly higher contribution than either Whiteor Black participants (p < .05).



Findings

The elders who no longer received the meals inJune 1997 (n = 328) had stayed in the program, onaverage, for 47.9 {SD = 41.9) weeks, and those whostill received meals in June 1997 (n = 181) had beenin the program, on average, for 94.5 {SD = 46.3) weeks.Bivariate results show that both White and Black par-ticipants—including all past and current—stayed in theprogram significantly longer than the "other" racialgroup (68.0 [SD = 50.8] weeks for whites and 65.0[SD = 51.2] weeks for Blacks vs 46.9 [SD = 43.6]weeks for the "other" group, p < .05). Elders whohad been hospitalized before their enrollment in theprogram had a significantly shorter stay than thosewho had not been hospitalized (58.0 [SD = 48.7] weeksvs 73.0 [SD = 51.1] weeks, p < .05). No gender ormarital status difference was found. Nor was there anydifference between those who received special dietmeals and those who did not.

Reasons for Termination.—As shown in Table 2, slightlymore than a quarter of the case closures were due tothe elder's prolonged hospitalization or placement ina supervised facility. (Although a small number of el-ders [n = 24] currently participating in the programhad returned to it after they had terminated, they wereincluded in the analysis of reasons for termination.)The next most frequently recorded reason for termi-nation was the elder's having moved in with a rela-tive who could prepare his or her meals or havingmade an arrangement for someone to cook for himor her. Another 10% of the elders discontinued theirparticipation because they had relocated to an areaoutside the geographic boundary served by the pro-gram; these elders might also have moved in with rela-tives. The elder's improved health and regaining ofability to prepare his or her own meals was a reasonfor termination for about 15% of the closed cases. Theelder's death was a reason for 14% of the closed cases.Most interestingly, 15% of the discontinued eldershad terminated because they were dissatisfied withthe quality of the meals, they did not like the foodsthat the program delivered (different food preference),and/or they had a poor appetite. (Although each of

these three reason categories was listed as a separateitem in the termination form, the elders tended tochoose any one or more of these three categories toindicate their dissatisfaction with the quality, variety,and/or taste of the foods delivered.)

Bivariate analysis found significant racial as well asgender differences in the reasons for termination. Asignificantly lower proportion of Black than White and"other" elders were discontinued due to their pro-longed hospitalization or placement in a supervisedfacility, or due to their improved health. The data showthat "other" elders were the most likely of all threegroups to be discontinued due to their placement ina supervised facility. A much higher proportion (28.2%)of Black than White and "other" elders stopped re-ceiving the meals because of their dissatisfaction withthe quality of the meals, different food preferences,and/or poor appetite. With respect to gender differ-ence, a significantly higher proportion of women thanmen discontinued from the program because of theirimproved health, dissatisfaction with the quality of themeals, or different food preferences, whereas a sig-nificantly lower proportion of them terminated becausethey moved in with a relative, made an arrangementfor someone to cook for them, or died.

Of all the terminated cases, those individuals whodiscontinued due to their dissatisfaction with the qual-ity of the meals, different food preferences, or a poorappetite and those who discontinued due to thetheir improved health had a significantly shorter lengthof stay in the program than tnose who discontinueddue to moving out of the service area. The lengths ofstay of elders who terminated for reasons other thanthese three were not significantly different from oneanother.

Further analysis of elders who terminated a secondtime (n = 40) showed that only 7.5% cited improvedhealth as a reason. Dissatisfaction with the quality ofthe food was a reason for only 5.0% of them; deathwas the reason for 17.5%; and prolonged hospitaliza-tion, placement in a supervised facility, moving in witha relative, arranging for someone to cook, and mov-ing outside the service area were the reasons for therest (results not reported in the table). Thus, it ap-pears that, as time went by, deteriorating health

Table 2. Racial and Gender Differences in Reasons for Termination (%)

Race" Gender**

White

15.85.6

19.918.810.212.813.9

3.0

Black

5.12.6

10.320.510.328.215.4

7.7

Other

21.30

36.28.54.3

14.912.8

2.1

Male

9.37.0

17.123.310.911.618.6

2.3

Female

18.83.1

23.314.3

8.916.611.24.0

Improved health (able to prepare own meals)Prolonged hospitalizationPlacement in a supervised facilityMove to a relative's home or finding someone to cookRelocation to a nonservice areaDissatisfaction with meals/food preferences/poor appetiteDeathOther

Note: n = 352. Included 328 elders who had ceased receiving meals by June 1997 as well as 24 current participants who had beenterminated after their enrollment between 1994 and 1996 but had reenrolled before June 1997.

*p < .02; ** p< .01: Denote significant racial and gender differences in reasons for termination.

Vol. 39, No. 4, 1999 401


and death became even more prevalent reasons fortermination.

Determinants of the Length of Stay.—As shown inModel I of Table 3, other things being equal, olderages, being White (as opposed to being "other" race),and having a better appetite were negatively associ-ated with the hazard of elders' termination from theprogram, whereas the number of ADLs/IADLs performedwith difficulty was positively associated with the haz-ard of termination. The significance of age may havebeen due to the tendency of older participants to stayin the program longer than younger ones, because thelatter may be more likely to go off the program owingto improved health. The negative association betweenbeing White, as opposed to being "other" race, andthe hazard of termination from the program is likelydue to the fact that a higher proportion of "other"than White elders terminated due to improved healthand placement in a supervised facility. Elders who hadmore difficulty carrying out ADLs/IADLs were likely tostay in the program for a shorter period, because tneirpoor health (as represented by tne number of ADLs/lADLs with difficulty) may have required them to movein with a relative, to be in a hospital for a long time,or to be placed in a nursing home. Poor health couldalso, of course, have resulted in death. Level of appe-tite may be a measure of health at the time of enroll-ment as well as an indication of an elder's tolerancefor the foods offered. In other words, elders who hadhad poor appetites to begin with may have been morelikely to be dissatisfied with the quality, variety, and/or taste of the foods and, thus, less likely to stay inthe program. The Cox regression results show that a

1-year increase in age decreased the hazard rate by2.1%, whereas a one-unit increase in the numberof ADLs/IADLs performed with difficulty increasedthe hazard rate by 5.3%. A one-unit increase in thelevel of self-reported appetite (e.g., from poor to fairappetite) decreased the hazard rate by 17%.

Elders' gender, marital status, being Black as com-pared to being "other" race, an episode of hospital-ization in the 6 months preceding enrollment in theprogram, weight-change pattern prior to enrollmentin the program, number of days of the week that themeals were delivered, and the amount of the elders'monetary contributions to the program had no signifi-cant association with the hazard of their terminationfrom the program.

The Cox proportional hazard model was also used todiscover the predictors of the length of stay amongterminated elders. The Cox regression coefficients (inModel II) indicate that, in addition to older ages, beingWhite, and having a better appetite, an episode ofhospitalization in the 6 months preceding enrollment inthe program was negatively associated with the hazardof termination from the program among the terminatedelders. An episode of hospitalization was a significantdeterminant of the length of stay among the terminatedelders, apparently because those who had been hospi-talized, as opposed to those who had not been hospi-talized, may have stayed longer in the program duringrecuperation. The data in Table 3 show that beingWhite, as opposed to being "other" race, decreased thehazard rate by 26%, and the episode of hospitalizationdecreased the hazard rate by about 19% among termi-nated elders. A one-unit increase in the level of self-reported appetite decreased the hazard rate by 21%.

Table 3. Relationship of Covariates to the Length of Stay on the Program: The Cox Proportional Hazard Regression Coefficients

AgeGenderMarital statusRace

WhiteBlack

Hospitalization in 6 monthsprior to enrollment

Weight-change pattern in6 months prior to enrollment

Number of ADLs/IADLs withdifficulty

Level of appetiteSpecial diet requirementNo. of days of week meals

deliveredAmount of contribution

Chi-square (df)-2LL change (df)

* p < .08; **p < .05; ***p <

(Both past and

B(SE)

-.021 (.008)***-.041 (.061)-.043 (.069)

-.154 (.089)*-.182 (.133)

-.097 (.059)

.089 (.059)

.051 (.019)***-.185 (.080)**

.028 (.059)

.007 (.064)-.007 (.009)

42.41(12)***39.66(12)***

.01.

Model Icurrent participants, n = 495)

Exp. (B)

.980

.857

1.053.831

402

95% Clfor Exp. (B)

(965, 994)

(.721, 1.020)

(1.015, 1.092)(.710, .973)

Model II(Past participants only,

B (SE)

-.023 (.008)***-.052 (.064)-.029 (.072)

-.304 (.094)***.049 (.141)

-.205 (.063)***

.025 (.062)

.031 (.020)-.229 (.078)***

.083 (.061)

-.027 (.067).005 (.010)

46.01(12)***44.51(12)***

Exp. (B)

.978

.738

.815

.795

The

n = 321)

95% Clfor Exp. (B)

(.964, .992)

(.614, .887)

(.721, .921)

(.682, .927)

Gerontologist


Discussion and Recommendations

Given that participants in home-delivered meal pro-grams are frail, homebound elders who are at highrisk for progressive morbidity and for mortality, it isnot unexpected that the reasons for their terminationfrom the meal program under study were largely as-sociated with circumstances generated by their dete-riorating health. Prolonged hospitalization, placement ina supervised facility, moving in with a relative-caregiver,arrangement for someone to cook for them, and deathwere the primary reasons participants terminated.

Considering research evidence that Black elders, de-spite their worse health status, are less likely to beinstitutionalized than are their White counterparts, thefindings of this study that a lower proportion of Blackthan White and "other" elders terminated due to theirplacement in a supervised facility are not surprising.But what is surprising is that a much higher propor-tion (28.2%) of African Americans discontinued par-ticipation due to their dissatisfaction with the qualityof the meals, different food preferences, or their poorappetite. In other words, these elders found that theservice was not acceptable to them, and thus theywere deterred from continued use of it (see Wallace,1990, for a discussion of service acceptability amongelders).

The basis for the acceptability problem may be alack of variety of food items that are palatable to theelders. However, considering that food habits amongelders represent one of the most culturally embeddedand enduring patterns (Read & Schlenker, 1993), thesignificantly higher level of dissatisfaction among Blacksmay also mean that the foods offered are not cultur-ally appropriate for them. Although the program of-fers a wide variety of diet menus and kosher meals, itdoes not offer ethnic-sensitive menus for Blacks, His-panics, or other minorities and does not allow indi-vidual choices of foods other than those offered onthe diet menus. The lack of choice of ethnic foodsmay have been a cause of the low acceptability ofthe service among Black elders.

It should be noted that 12.8% of White elders and14.9% of "other" elders also discontinued their par-ticipation due to dissatisfaction with the food, differ-ent food preferences, or poor appetite. These percentagesare by no means minuscule and point to the need forthe programs to allow more individual choices in themenu items in order to help these elders stay on theprogram longer by improving the acceptability of theservices. Special efforts to cater to these elders' foodpreferences might have made it less likely that theywould drop out of the program prematurely.

The Cox proportional hazard regression results con-firmed that elders' poor appetite, as well as their healthstatus (as represented by tneir number of ADLs/IADLswith difficulty), were indeed significant determinantsof the hazard of elders' termination from the program.If elders decide to discontinue participation in the pro-gram while remaining in the community, they may beat increased risk of malnutrition, considering that theyare unable or only partially able to obtain or preparenutritious meals. Low-income frail elders are especially

unlikely to be able to afford alternative food-cateringservices, which are usually much more costly than Mealson Wheels. Thus, they either are on their own or haveto depend on their informal support system, whichmay have already been overburdened by other caregivingresponsibilities. Moreover, for elders who live alone,the decision not to participate in the program deprivesthem of other valuable benefits of a home-deliveredmeal program—their regular contact with volunteerswho also function as a check-up system for them.

Based on these findings, I present the following tworecommendations aimed at improving the acceptabil-ity of the program:

1. Allow more individual choice of menu items, sothe participants can check off their preferred fooditems on a standardized daily or weekly menu inadvance. The availability of individual meal choiceswould increase attractiveness of the program forall potential and current participants.

2. Provide more ethnic-sensitive menus for minorityelders in order to encourage their enrollment andcontinued participation in the program. Especiallyin areas where minority elders are concentrated,their food preferences can easily be incorporatedin the meal-preparation process without increasingthe productions costs, owing to the economy ofscale. Although the 12% share of Blacks among allprogram participants reflects their share of the gen-eral population, it indicates a low utilization rateamong African American elders. This is true be-cause of their greater need for the service, whichis due to their higher number of health problemsand lower level of economic resources, comparedwith their White counterparts.

As discussed earlier, information and accessibilityare not serious barriers to participation in home-delivered meal programs. Previous studies also indi-cated that the low rate of utilization is more likely toreflect a lack of perception of need than a lack ofacceptability of the services. But this study found thatacceptability was a significant deterrent to continuedparticipation among elders who utilized the service.Although more research is needed to discover whetheracceptability is a factor in eligible elders' decision notto participate at all, a culturally sensitive approach inan increasingly diverse, multicultural aging society wouldsurely be a sensible one to reach underserved ethnic/racial minority frail elders. Such an approach wouldat least help enhance the acceptability among eldersalready enrolled in the program.

This study has obvious limitations in terms of itsapplicability to meal participants outside the specificprogram service area. More studies based on nation-ally representative samples are needed to examine thereasons for termination and determinants of length ofstay in order to identify any non-health-related carri-ers to frail elders' continued participation in home-delivered meal programs.

Continued participation needs to go hand in handwith enrollment of more eligibles in home-deliveredmeal programs. This study did not deal in depth with

Vol. 39, No. 4,1999 403


the problems of waiting lists. But efforts to ensure con-tinued participation of eligible current participantswould not make sense when a significant proportionof home-delivered meal programs, including this Mealson Wheels program, have waiting lists. Eliminating wait-ing lists as well as enrolling all eligible nonparticipantsin the programs are tasks as significant as preventingparticipants from dropping out.

References

Administration on Aging (1998). Serving elders at risk, the Older Ameri-cans Act nutrition programs, national evaluation of the elderly nutri-tion program, 1993-1995. [On-line]. Available: http://www.aoa.dhhs.gov/aoa/n utreva l/fu I Itext. htm I

Ahmed, F. E. (1992). Effect of nutrition on the health of the elderly. Jour-nal of the American Dietetic Association, 92, 1102-1108.

Bernard, M. A., Anderson, C , & Forgey, M. (1995). Health and nutri-tional status of old-old African Americans. Journal of Nutrition for theElderly, 74(2/3), 55-67.

Burt, M. R. (1993). Hunger among the elderly: Local and national compari-sons. Final report of a national study on the extent and nature of foodinsecurity among American seniors. Washington, DC: Urban Institute.

Coulston, A. M., Craig, L, & Voss, A. C. (1996). Meals-on-wheels appli-cants are a population at risk for poor nutritional status. Journal of theAmerican Dietetic Association, 96, 570-573.

Frongillo, E. A., Williamson, D. F., Roe, D. A., & Scholes, J. E. (1987).Continuance of elderly on home-delivered meals programs. AmericanJournal of Public Health, 77, 1176-1179.

Horwath, C. C. (1991). Nutrition goals for older adults: A review. TheCerontologist, 31, 811-821.

Kuczmarski, M. F. (1993). Nutritional status of older adults. In E. D. Schlenker(Ed.), Nutrition in aging (2nd ed., pp. 236-254). St. Louis: Mosby.

Locher, J. L, Burgio, K. L, Yoels, W. C , & Ritchie, C. S. (1997). Thesocial significance of food and eating in the lives of older recipients ofMeals on Wheels. Journal of Nutrition for the Elderly, 17(2), 15-33.

Mui, A. C , Choi, N. C , & Monk, A. (1998). Long-term care and ethnicity.Westport, CN: Auburn.

Ponza, M., Ohls, J. C , & Posner, B. M. (1998). Elderly nutrition program:Evaluation literature review [MPR Reference No. 8161-040]. Reportsubmitted to the Administration on Aging, Department of Health andHuman Services by Mathematica Policy Research, Princeton, NJ. [On-line], Available: http://www.fiu.edu/~nutreldr/ENPentire3.html

Posner, B. M., Jette, A. M., Smith, K. W., & Miller, D. R. (1993). Nutri-tion and health risks in the elderly: The nutrition screening initiative.American Journal of Public Health, 83, 972-978.

Read, M. & Schlenker, E. D. (1993). Food selection patterns among theaged. In E. D. Schlenker (Ed.), Nutrition in aging (2nd ed., pp. 284-312). St. Louis, MO: Mosby.

Smith, R., Mullins, L., Mushel, M., Roorda, J., & Colquitt, R. (1994). Anexamination of demographic, sociocultural, and health differences be-tween congregate and home diners in a senior nutrition program. Journalof Nutrition tor the Elderly, 14(1), 1-21.

Wallace, S. P. (1990). The no-care zone: Availability, accessibility, andacceptability in community-based long-term care. The Cerontologist,30, 254-261.

Received June 8, 1998Accepted April 13, 1999



Charles C ThomasP U B L I S H E R L T D .

Leader In Rehabilitation and Long-Term Care Publications• denotes new publication

• Ma, Grace Xueqin & George Henderson—RETHINK-ING ETHNICITY AND HEALTH CARE: ASociocultural Perspective. '99, 362 pp. (7 x 10), 2 il , 12tables.

• Wayne, James A.-^TRATEGIC MARKETING OFYOUR LONG-TERM CARE FACILITY: A How-ToApproach. '98, 250 pp. (8 1/2 x 11), 19 il., $39.95, spiral(paper).

Goldberg, Phyllis Z—HOW TO TOLERATE LAC-TOSE INTOLERANCE: Recipes and a Guide forEating Well Without Dairy Products. '98, 118 pp. (7 x10), 8 il., 4 tables, $19.95 spiral (paper).

Falk, Ursula Adler & Gerhard Falk—AGEISM, THEAGED AND AGING IN AMERICA: On Being Old inan Alienated Society. '97, 288 pp. (7 x 10), $69.95, cloth,$49.95, paper.

Cohen, Jane G. & Marilyn Wannamaker—EXPRES-SIVE ARTS FOR THE VERY DISABLED ANDHANDICAPPED FOR ALL AGES. (2nd Ed.) 96, 236pp. (8 1/2 x 11), 135 il., $42.95, spiral (paper).

Nemapare, Prisca & Richard Neumann—FUNDAMEN-TALS OF MEAL MANAGEMENT. '95, 242 pp. (7 x10), 1 il., 16 tables, $55.95, cloth, $38.95, paper.

Gordon, George Kenneth & Ruth Stryker—CREATIVELONG-TERM CARE ADMINISTRATION. (3rd Ed.)'94,390 pp. (7 x 10), 11 il., 6 tables, $69.95, cloth, $45.95,paper.

Royse, David—HOW DO I KNOW IT'S ABUSE?Identifying and Countering Emotional Mistreatmentfrom Friends and Family Members. '94, 248 pp. (7 x10), $54.95, cloth, $37.95, paper.

Oberley, Edith T. & Neal R. Glass—UNDERSTAND-ING KIDNEY TRANSPLANTATION. '87, 160 pp., 35il., $35.95.

Wood, Marion N—COPING WITH THE GLUTEN-FREE DIET. '82, 164 pp., 2 il., 2 tables, $35.95, spiral(paper).

Barrett, Shirley L.—PARTIES WITH A PURPOSE: AHandbook for Activity Directors. 80, 128 pp., $33.95,paper.

• Portnoy, Enid J.—MAKE THEIR DAYS: Activities forResidents in Long-Term Care. 99, 112 pp. $18.95,(paper).

Everstine, Louis—THE ANATOMY OF SUICIDE:Silence of the Heart. '98, 170 pp. (7 x 10), 6 il., $42.95,cloth, $27.95, paper.

Carter, Betty Woerner—I CAN'T HEAR YOU IN THEDARK: How to Learn and Teach Lipreading. '98, 226pp. (8 1/2 x 11), $34.95, spiral (paper).

Greenberg, Samuel I.—EUTHANASIA AND ASSIST-ED SUICIDE: Psychosocial Issues. '97, 188 pp. (7 x 10),1 table, $53.95, cloth, $38.95, paper.

Bashshur, Rashid L., Jay H. Sanders, & Gary W.Shannon—TELEMEDICINE: Theory and Practice."97, 454 pp. (7 x 10), 22 il., 15 tables, $89.95, cloth,$69.95, paper.

Jefferies, William McK—SAFE USES OFCORTISOL.(2nd Ed.) '96,228 pp. (7 x 10), 8 il, $55.95, cloth, $39.95,paper.

Thorman, George—COUNSELING OLDER PER-SONS: A Professional Handbook. '95, 110 pp. (7 x 10),1 il., 6 tables, $36.95, cloth, $23.95, paper.

Locke, Shirley A—COPING WITH LOSS: A GUIDEFOR CAREGIVERS. '94, 238 pp. (7 x 10), $51.95,cloth, $35.95, paper.

Clair, Jeffrey Michael, David A. Karp & William C.Yoels—EXPERIENCING THE LIFE CYCLE: ASocial Psychology of Aging. (2nd Ed.) 93, 274 pp. (7 x10), 2 il., $57.95, cloth, $38.95, paper.

Weinhold, Virginia Beamer—INTERIOR FINISHMATERIALS FOR HEALTH CARE FACILITIES: AReference Source for All Installations Where DurableSurfaces are Needed. '88, 376 pp. (7 x 10), 41 il., 25tables, $93.95, cloth, $53.95, paper.

Stryker, Stephanie—SPEECH AFTER STROKE: AManual for the Speech-Language Pathologist and theFamily Member. (2nd Ed.) '81, 442 pp. 179 il., $49.95,spiral (paper).

Grant, Roy E—SING ALONG—SENIOR CITIZENS.73, 108 pp. (8 1/2 x 11), $23.95, spiral (paper).

Books sent on approval • Shipping charges: $5.50 U.S. /$6.50 Canada • Prices subject to change without notice

Contact us to order books or a free catalog with over 800 titlesCall 1-800-258-8980 or 1-217-789-8980 or Fax 1-217-789-9130

2600 South First Street • Springfield • Illinois • 62704Complete catalog available at www.ccthomas.com • [email protected]


Determinants of Frail Elders' Lengths of Stay in Meals on Wheels

Documents