| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
| ||||||||||||||||||||||||||||||||
Correspondence: Joseph R. Sharkey,1, MPH, RD, Department of Nutrition, School of Public Health, University of North Carolina at Chapel Hill, CB#7400, Chapel Hill, NC 27599-7400. E-mail: aggie1{at}mindspring.com.
Decision Editor: Eleanor S. McConnell, RN, PhD
 |
Abstract |
|---|
 TOP Abstract MethodsResultsDiscussionReferences |
|---|
Key Words: Nutrition screening • Telephone surveys • Community-based nutrition programs
Community-based service providers grapple with the challenges of a dramatically increasing older population: divergent aging patterns, needs, and expectations (
Balsam and Bottum 1997
;
Hoogenboom, Spangler, and Crose 1998), as well as increased risk for poor nutrition and functional disabilities (
Galanos, Pieper, Cornoni-Huntley, Bales, and Fillenbaum 1994;
Jensen, Kita, Fish, and Heydt 1997;
Ponza, Ohls, and Millen 1996). The challenges to community-based service providers are especially monumental in rural areas, where a disproportionate percentage of older persons have poorer health, suffer from more chronic diseases, and are more vulnerable to poor nutritional status (
AbuSabha et al. 1997;
Blazer, Landerman, Fillenbaum, and Horner 1995;
Mainous and Kohrs 1995;
Schoenberg, Coward, Gilbert, and Mullens 1997). Compounding these challenges are concerns about accessibility of services, availability of resources, service coordination, quality of care, individual characteristics, and integration of services (
Arcury, Quandt, Bell, McDonald, and Vitolins 1998;
Magilvy 1996). And these concerns are linked to determining which older persons are most likely to be at nutritional and functional risk, and how best to provide assistance (
Arcury et al. 1998;
Benjamin-Coleman and Alexy 1999). This article reports our efforts to examine the feasibility of telephone surveying to identify nutritional and functional risk among noninstitutionalized older persons in a predominately rural North Carolina county.
The available survey methods for determining nutritional risk and functional status include face-to-face, direct mail, and telephone-administered surveys, each of which has trade-offs with regard to response rates, expense, respondent burden, anonymity, and completion rates (
Gibson, Koepsell, Diehr, and Hale 1999). Although the increasing popularity of telephone surveys is the result of many factors, including increased telephone coverage and documented similarity of response between telephone and face-to-face interviews (
Fox, Heimendinger, and Block 1992;
Herzog and Rodgers 1988;
Slutske et al. 1998), telephone interviews with older adults have been utilized primarily to recruit older subjects for research studies (
AbuSabha et al. 1997;
Taylor-Davis, Smiciklas-Wright, Davis, Jensen, and Mitchell 1998) and to collect data from previously recruited study populations (
Lahmann and Kumanyika 1999;
Schaffer, Coates, Caan, Slattery, and Potter 1997). Little is known about the use or costs of telephone surveying by community-based service providers, especially in rural areas, to determine current and future nutritional and functional needs of older adults in the community who do not currently receive services. At present, for most providers of nutrition services for the elderly population, the only available information on nutritional risk and functional status is through data routinely collected on program participants (
Ponza et al. 1996;
Sharkey, Haines, and Zohoori 2000). Without an understanding of the components of nutritional and functional risk among the vast majority of nonparticipating older persons, nutrition programs are limited in their ability to assist an increasing number of older persons to remain independent in the community; to plan future programs; to procure additional resources; to target interventions; and to tailor programs to the disparate needs within their community.
We extended this understanding by examining the response and participation rates, time requirements, and costs for a two-stage process (postal contact followed by telephone surveys) for obtaining comprehensive information on general health, eating habits, living environment, and functional status from a probability sample of noninstitutionalized older persons.
|
Methods |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
).
The decision was made a priori to recruit until 150 interviews were completed. A personalized advance letter (
Dillman 1978) from the Chatham County Council on Aging was sent to potential respondents 2 weeks prior to the telephone interview. The letter introduced the Council on Aging as a nonprofit source of Senior Services in Chatham County, emphasized the importance of the respondent's participation, and indicated that the survey results would assist in planning future services. The letter further described the survey topics, provided assurance of confidentiality and voluntary participation, and listed a contact number at the Council on Aging for any questions concerning the survey. In addition, both the executive director of the Council on Aging and the survey coordinator (J.R.S.) signed each letter. The probability sample was partitioned into sequential mailing groups. The first group of letters was mailed; the mailing was followed by telephone contact with the first group. At the time the first group was contacted by telephone, the next group of letters was mailed. As the members of that group were contacted, a third group of letters was mailed. The goal of 150 completed interviews was reached prior to contact with everyone in the third group (119 individuals were never contacted). The participants were not offered compensation for their participation in the study.
The telephone interviewers were trained in conducting interviews with older individuals. Included in the training were the importance of the introduction of the telephone interview, conducting the interview, answering respondent's questions, completing the survey form, and completing the phone logs. Calls were attempted between the hours of 1:00 p.m. and 8:00 p.m., Monday through Saturday. At the request of the Council on Aging, no calls were made on Sundays. A standard script for the introduction, which included the use of human subjects' disclosure and informed consent agreement, guided the interviewers (
Taylor-Davis et al. 1998). The telephone interviews were conducted from March 1999 to May 1999.
A tracking system was developed in which all potential respondents contacted by telephone were identified as "contacted for interview." Of those who were contacted, "participants" completed the survey and "nonparticipants" refused to complete the survey. Potential respondents were eliminated from the sample if they were unable to be contacted for any of the following reasons: no telephone, telephone not working or wrong number, deceased or transferred to a nursing home, or three unsuccessful attempts to contact. Using the most recent countywide telephone directories, every effort was made to locate telephone numbers for entries missing a number. A prospective respondent was eliminated from the sample due to no telephone when there were both a missing telephone number on the voter registration list and no listed telephone number by name or address in the telephone directory.
Age, sex, race, and residence (as determined by voting precinct) data were kept on all "contacted for interview" and "unable to contact" respondents for comparison. The actual response rate was defined (
Last 1995) as the number of completed interviews (n = 152) divided by the total number of persons who would have been surveyed if all had participated (n = 322). This excludes those who were never contacted because the study achieved the goal of 150 completed interviews (n = 119). The participation rate was determined to be the percentage of persons contacted for an interview (n = 201) who completed the survey; the nonparticipation rate was the percentage of persons contacted who refused to participate in the survey.
Telephone logs were maintained by the interviewers and collected by the survey coordinator each calling day. The information on the logs included date of call, time of call, telephone number, attempt number, person called, survey number, outcome, and pertinent notes (e.g., message left on machine). The outcome was coded for the following events: no answer, busy, answering machine, telephone not in service, telephone temporarily disconnected, wrong number, respondent unavailable, respondent refused to participate, appointment scheduled for survey, and completed interview. The pertinent notes section was used to indicate a scheduled appointment time, message left, and any other information the interviewer felt was relevant.
Survey Instrument
The 67-item survey instrument was adapted from the Nutrition Screening Initiative's Level I and II Screens (
Nutrition Screening Initiative 1996) and partitioned into four domains: general health, eating habits, living environment, and functional status. The 9 questions of the general health domain asked about weight change, medication, tobacco and alcohol use, and chronic health conditions. Eating habits included 16 questions that addressed availability of food, eating alone, chewing or swallowing difficulties, shopping and meal preparation, and normal intake of specific food group items and fluids. Living environment was evaluated with 16 questions concerned with income, household composition, security, adequate heating, cooling or refrigeration, transportation, and daily social contacts. The functional status questions referred to the individual components of the Activities of Daily Living (
Katz, Ford, Moskowitz, Jackson, and Jaffe 1963), the Instrumental Activities of Daily Living (
Lawton and Brody 1969), and any hospitalizations in the previous 12 months.
Costs
The costs for recruitment of respondents were calculated in a similar manner as reported by
Taylor-Davis and colleagues 1998. The costs of mailing included supplies for envelopes, letterhead paper, labels and printing, labor for creating the labels and personalized introductory letters, assembly of letters, and postage. Telephone screening costs included wages for interviewer training, attempts to contact prospective respondents, survey time, supplies (e.g., telephone logs and feedback forms), and actual telephone charges. All wage costs were determined at a rate of $10 per hour. Because the purpose of this study was to determine the feasibility of community-based telephone screening, the costs did not include expenses associated with development and printing of the screening instrument or the expenses associated with data entry and analysis of the screening instrument. Because the Chatham County Council on Aging is a nonprofit service provider, there was no charge for the age-specific (ages 
60 years) voter registration list. We assume this to be the case for other nonprofit service providers.
Data Management and Analysis
The survey instrument provided age, sex, and race of respondent, along with starting and ending times for the survey. All information was key entered from hard copy into a relational database, using Microsoft® Access 97 for Windows (
Microsoft Corporation 1997). Proportions by age, sex, and race; participation rate; number of attempts required for contact; and length of participant calls were calculated. Chi-square and Fisher's exact tests were used to compare men with women, participants with nonparticipants, contacted with unable to contact, and Blacks with Whites. The Student's t test was used to compare the mean age of participants with nonparticipants and contacted with unable to contact. The chi-square and Fisher's exact tests, t tests, and tabulations were performed using Stata Statistical Software, Release 6.0 (
StataCorp 1999).
|
Results |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
|
|
Table 2 shows the mailing and telephone costs for completing 152 interviews. Because this article focuses on telephone interviewing as a method for information gathering, the costs do not include survey instrument design, the cost of printing the actual survey, data entry of the results of the completed surveys, and survey data analysis. Therefore, the total cost for 152 participants was $1,619.45, with an estimated cost of $10.65 per completed interview.
|
|
Discussion |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
). These may include targeted programs of screening, nutrition intervention (i.e., meals, home care services, dental services, and pharmacist advice), nutrition education, and counseling. However, community-based nutrition programs for older adults often lack sufficient information to determine nutritional and functional service needs among their nonclient community. The findings suggest that telephone-administered surveying may be an appropriate component of an overall service strategy, especially in rural areas.
We believe this is the first report to characterize time and costs ($10.65 per completed survey) for participant recruitment and survey completion from a probability sample of noninstitutionalized older adults in a predominately rural county. Prior research has focused on the costs for participant recruitment from a patient database of a tertiary-care hospital into a subsequent study (
Taylor-Davis et al. 1998). In 1999-adjusted dollars (
Federal Reserve Bank of Minneapolis 2000),
Taylor-Davis and colleagues 1998 estimated a cost of $9.36 per recruited participant for an introductory letter and telephone recruitment into a future study of home-based nutrition education. By comparison, we calculated a cost in 1999 dollars of $10.65 for the introductory letter, telephone recruitment, and survey completion.
Our overall response rate of 47% and participation rate of 76% for recruitment and study completion are higher than previously reported for recruitment into research studies (
Boult, Boult, Morishita, and Pirie 1998;
Herzog and Rodgers 1988;
Stuck et al. 1995;
Taylor-Davis et al. 1998). While
Taylor-Davis and colleagues 1998 reported a 37% overall response rate (480/1300) and 45% participation (recruitment) rate,
Stuck and colleagues 1995 reported a 37% recruitment response using a voter registration list. In a telephone follow-up to an existing project,
Benjamin-Coleman and Alexy 1999 reported a 64% participation rate.
This study required an average of 3.6 calling attempts (555 total attempts) in order to complete 152 interviews, with 58% of the surveys completed on the first attempt and another 30% on the second attempt. This is a lower average number of calling attempts per recruitment and higher percentage of recruitment on the first calling attempt than previously reported (
Taylor-Davis et al. 1998). Our higher participation rate and lower average number of calling attempts may be the result of a number of factors: (a) the individually signed and personalized introductory letters from the Council on Aging; (b) collaboration with a well-respected community-based provider of services for older adults; and (c) minimal respondent burden. Because study participation occurred at the time of the calling attempt, potential respondents were not burdened with the prospect for a future in-home or clinic visit.
Although electoral rolls are a convenient frame from which to choose random samples for telephone surveys (
Smith, Mitchell, Attebo, and Leeder 1997), all older adults are not listed in this sampling frame. In Chatham County, approximately 72% of all older persons are registered voters (73% of the estimated older White and 62% of estimated older Black populations). Little has been published about the characteristics of older adults excluded from research because they are not registered voters. However, other sources of potential respondents are problematic for community-based studies: driver registration provides limited numbers of elders, especially in rural areas; senior centers or other community-based services are not widely used by most elders; and random-digit dialing (
Mainous and Kohrs 1995;
Olsen and Mandel 1988) is expensive, with wasted effort for checking out unused and nonresidential phones (
Dillman 1978) and the residential phones of nontarget persons. There were advantages to the use of electoral rolls in our study: (a) the rolls were provided to the Chatham County Council on Aging at no cost; and (b) the rolls provided information on age, sex, race, and precinct of county residence, which could be used to compare responders with nonresponders and provide addresses for alternative methods of contact.
We were unsuccessful in surveying minority elders in proportion to their representation in the overall older adult population. This presents another limitation to the exclusive use of electoral rolls: a large number of older adults, especially minority elders, are apparently without a telephone number. In addition to not having a telephone, other reasons for an absence of a telephone number could include refusal to provide a telephone number or access to telephone service through family or friends. From the entire Chatham County voter list, we found a large number of older persons who did not indicate the presence of a telephone (39% of Blacks and 20% of Whites). There also appears to be a difference according to ethnic background for lack of a telephone number by age group: the largest percentage of Blacks without a telephone number were aged 60–74 (41% of Blacks); Whites were aged 85 years and older (29% of Whites). After examining all area telephone directories by name and by address, we reported 13.7% of our study respondents were without a telephone. Our estimated telephone coverage is in sharp contrast to the reported 3.3% of potential respondents, aged 60 years and older, from the third National Health and Nutrition Examination Survey (NHANES III) and the 2.7% of persons aged 65 years and older from the 1994 National Health Interview Survey (NHIS) Access to Health Care Supplement (
Anderson, Nelson, and Wilson 1998;
Ford 1998). Both studies reported that socioeconomic variables, region, and locale might explain part of the differences in telephone coverage. In the NHANES III report, those without a telephone were characterized as more likely to be non-White, less educated, poorer, and to live outside an urban area (
Ford 1998). On the basis of the 1994 NHIS Access to Health Care Supplement,
Anderson and colleagues 1998 found the following indicators of lowest telephone coverage: living in South region, being Black, living in nonmetropolitan areas, being poor, and being Black below the poverty level.
Economic factors may not be the only explanation for the absence of a telephone number for so many registered voters. Many may refuse to provide a telephone number or may live with someone else and not have a telephone listed in their name. Although telephone surveys may represent a practical and economical method for the expansion and development of community interventions based on the assessment of local needs (
Fox et al. 1992;
Posner, Borman, Morgan, Borden, and Ohls 1982), the exclusive reliance on telephone-administered surveys will possibly exclude some individuals most in need of supplemental aging services, including the very poor (
Fox et al. 1992;
Mainous and Kohrs 1995), non-English speakers, cognitively impaired elders, minorities, and persons with moderate or high limitations because of chronic conditions including the hearing impaired and persons too ill or restricted in mobility to operate a telephone for moderate periods of time (
Ford 1998;
Marcus and Crane 1986). Given that low-income populations have a high proportion of disconnected numbers or lack telephones (
Gibson et al. 1999), and that a disproportionately high percentage of the rural older adult population may be without a telephone, it is necessary to complement the telephone screening with another method of survey administration for older adults at elevated risk for poor nutrition and functional disability. Additional approaches could include limited mailings, generally considered less expensive than telephone contact, and face-to-face contact, which is generally more expensive than telephone surveys (
Morris, Colditz, and Evans 1998;
Taylor-Davis et al. 1998).
There are several implications of these findings for rural nutrition service providers. First, telephone-administered surveys provide a feasible method to cover a broader range of older adults. Second, telephone-administered interviewing allows providers to quickly target specific subsets of the older population (e.g., geographic area, income, and minority status). Third, electoral rolls provide a source for potential respondents, which can be sorted by geographic area, race, age, and gender. Fourth, telephone-administered surveys may alert providers to non-clients who may be at elevated risk for adverse health outcomes. Fifth, there are serious financial implications. Scarce resources, especially in rural areas, may make it difficult to consider spending $10.65 per participant for interviewing an entire county. Limited resources may require a method for identifying individuals at greater risk through stratification by income level, educational attainment, or other criteria. Finally, no single method will identify individuals most at risk of inadequate nutritional intake. Although telephone-administered interviewing, as demonstrated in this study, is effective, many older adults considered to be at greatest risk of poor nutrition are without a telephone. Although one alternative may be the use of mailed-response surveys,
Morris, Mendes de Leon, Tangney, Evans, and Bienias 1999 found that personal contact would be required for participation among Blacks. Considering that many older persons will be without a telephone or have difficulty or reluctance in completing a mailed survey, some face-to-face interviewing may be required.
The results of this study have contributed to the understanding of the feasibility of community-based, telephone-administered surveying of older adults by service providers. Although this study reports an encouraging participation rate as well as efficient and effective time and costs for a voter list–based telephone interview, we recommend three specific changes in our methods to further increase the coverage and response rate: (a) increase the number of calling attempts beyond three; (b) attempt to convert refusals; and (c) study supplemental methods in order to target older adults who lack a working telephone or are absent from voter registration lists, such as determining the availability of alternate sources of names of noninstitutionalized older persons, and working with churches, ministerial associations, senior centers, and other service providers.
The telephone-administered survey of nutritional and functional health has the potential to assist community-based service providers in meeting the challenges of a burgeoning older population with ever-increasing needs for home- and community-based nutrition services. The estimation of the constituents of nutritional risk, by geographic area, economic status, or ethnicity, will aid in providing accurate and proactive estimates of service needs.
Practice Concepts
The Forum
Book Reviews
|
Acknowledgments |
|---|
|
Footnotes |
|---|
Received for publication January 3, 2001. Accepted for publication April 27, 2001.
|
References |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
 |
|
 |
|
  B. E. Fries, M. James, S. S. Hammer, L. R. Shugarman, and J. N. Morris Is Telephone Screening Feasible? Accuracy and Cost-Effectiveness of Identifying People Medically Eligible for Home- and Community-Based Services Gerontologist, October 1, 2004; 44(5): 680 - 688. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 H. Iredell, T. Shaw, P. Howat, R. James, and J. Granich Introductory postcards: do they increase response rate in a telephone survey of older persons? Health Educ. Res., April 1, 2004; 19(2): 159 - 164. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||
| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
Never contacted after 3 unsuccessful attempts.