Reasons among older Swedish women of not participating in a food survey

Abstract

Objectives: To (a) examine participation rate as a function of municipality, age group and living status; and (b) investigate the main reasons for exclusion and declining as stated by the women themselves.

Design: Analysis of participation rate and content analysis of statements given in phone calls explaining exclusion or declination from a project in which 24 h recalls and food-diaries were used.

Subjects: Self-managing Swedish women (n=505) were systematically selected from a stratified random sample covering single living and cohabiting women aged 64–68, 74–78 and 84–88 y living in three municipalities.

Results: No significant differences were found among included and declining women when municipalities and living status was analysed, but significantly more women in the oldest group were excluded (P>0.01). Among those in their 80s living at home, the usual reasons for exclusion were illness, disability or dementia, and many lived in institutions for old people. The four most used explanations to decline participation were ‘lack of time’, ‘tired, fragile, sick or having bad memory’, ‘not willing to participate in scientific studies’ and ‘too old and nothing to contribute’.

Conclusions: The participation rate was, compared with other food surveys in the older generation, fairly good, especially among those in their 80s. However, the most active and the very ill and disabled did not participate.

Sponsorship: The Swedish Council for Social Research, the Swedish Council for Forestry and Agricultural Research, the Swedish Foundation for Health and Care Sciences and Allergy Research and Uppsala University.

Introduction

With the purpose of studying—from nutritional and cultural perspectives—the food and meal habits of older women living at home, the project Meals, Eating habits and Nutritional intake among Elderly Women (MENEW) was carried out. Dietary methods were used as well as qualitative interviews. Self-managing women were included in study I of the project and disabled women in study II. This article reports from study I, where random sampling was used. Disabled women not managing cooking themselves were excluded, as they were investigated in study II.

Eating habits are personal matters that most people are unwilling to report in detail (Isaksson, 1998). Participation or response rates in dietary surveys are often found to be low and, in studies performed in the Nordic countries (Steingrimsdottir et al, 1991; Kleemola et al, 1994; Becker, 1994; Andersen et al, 1996; Johansson et al, 1997), ranged between 58 and 76%. Participation rate decreases with increasing age, and Johansson et al (1997) found that the mean participation rates in the age groups 70–74 and 75–79 y were 50 and 40%, respectively. This corresponds to figures from the Euronut SENECA study, performed in 12 European countries among people aged 70–75 y, where the response rate was 51% (van't Hof et al, 1991). In one British study, response rates were found to decrease from 67% for those aged under 70 y to 20% for those over 85 y (Wright et al, 1995). In general, the lowest response rates in dietary surveys are thus found among older women both in the Nordic countries and the United States (Harries et al, 1989; Becker, 1994; Andersen et al, 1996).

Low participation rates in dietary studies among old persons increase the risk of positive bias. Health-conscious and active persons as well as people who are motivated and more interested in their diet in general will be over-represented in the study sample, and those with poorer health status will be less likely to participate (Holcomb, 1995; van't Hof & Burema, 1996; Berglund, 1998). Persons aged ≥75 y are considered to have a potentially high risk for nutritional problems and, in studies such as these, they may be missed (Harries et al, 1989). In order to evaluate participation, both the Third National Health and Nutrition Examination Survey (NHANES III, Harries et al, 1989) and the Euronut SENECA studies (van't Hof et al, 1991) specifically investigated non-participation. NHANES III included five or six questions on health status and health habits, and Euronut SENECA included a few questions from the general questionnaire (Harries et al, 1989; van't Hof et al, 1991). A more common way to compare non-participants with participants is to use data from the total random sample to analyse whether participants are representative (Harries et al, 1989; Johansson et al, 1997). Random sampling is also considered the optimum method to achieve representative, generalizable results (Berglund, 1998).

Different methods to increase participation rates in dietary studies have been tested. Johansson et al (1997) studied different distribution methods for the Quantitative Food Frequency Questionnaire (QFFQ), and found that postal distribution combined with administration of the QFFQ by interviewers gave the highest participation rate, as did offering participation in a lottery. In NHANES III, the methods to increase participation were publicity in advance, local endorsements, easy-to-read/eye-catching materials and, furthermore, participants received health tips while responding and a certificate of participation upon completion (Harries et al, 1989).

Another discussion surrounding dietary surveys deals with obtaining valid data. Isaksson (1998), thus, points out that the best result is obtained when a dietician completes the form during a personal visit. Some researchers, however, have found that older people were less likely to participate in home interviews (Harries, 1989), whereas in a Swedish investigation among men and women aged 70 or more, 66% participated in such in-home dietary interviews (Rothenberg et al, 1993). Dietary studies among old persons, especially women, seem to be the most difficult to perform. However, the majority of the oldest households in the Western world consist of women living on their own (Dirren, 1994). Loneliness and bereavement have been found to impair food choices and nutritional intake among women living on their own, whereas women still living with their spouse do not change meal habits (Webb & Copeman, 1996; Lyon & Colquhoun, 1999; Wylie et al, 1999; Sidenvall et al, 2000). In Sweden, municipal home service has been reduced such that it now includes only personal care, and only those judged as severely disabled are offered meals on wheels (Szebehely, 1998). Consequently, most old women buy and cook their own food (Sidenvall et al, 2001), and studies on their nutritional well-being are, thus, essential. Few such studies have focused on self-managing women, and since women are responsible for cooking, it is interesting to study their management strategies.

The MENEW Study I invited older women from different age groups and living conditions spread over three municipalities to participate. After the invitation, the researchers made personal contact with each woman to assess their autonomy in terms of buying food and cooking meals. Consequently, the prospect of obtaining data on non-participants was extraordinarily good. The objectives of the present study were to (a) examine participation rate as a function of municipality, age group and living status, and (b) investigate the main reasons for exclusion and declining as stated by the women themselves.

Methods

Description of the MENEW Study I

The Menew Study I was carried out in the mid-eastern part of Sweden. Two dietary methods were used, repeated 24 h recall and food diary. Stratified random sampling, from which women were systematically selected, was used to include self-managing older women who were single living or cohabiting; the women were divided into three groups aged 64–68 (60s), 74–78 (70s) and 84–88 (80s). The women were living in one larger and two smaller towns or the surrounding rural area. Random sampling was chosen in order to cover such characteristics as attitudes to food, education, economic situation, type of housing etc, factors that vary in the community. The sample was obtained from the official register of Sweden.

Five-hundred and seventy women were sent a letter inviting them to participate in the study. The letter focused on the tradition of cooking among women in the older generation and on the researchers' interest in investigating women's everyday practices. Furthermore, the letter informed about the method of surveying eating habits using two dietary methods, of which one was a food diary that the women were expected to fill in for three consecutive days. The women were informed that data collection was to be performed through home visits, during which height and body weight would also be measured. One week after the letter was sent, a phone call was made to each woman to discuss the letter and ask whether she was willing to participate. Five researchers (BS, CF, JA, KG, MN) divided the selected women among them for telephone contact and subsequent interviewing. In addition, retirement status, living conditions, and ability to buy, prepare and cook food independently were checked. If the woman was willing to participate, she was asked to describe the way to her home. Through this conversation, the woman's mental ability and ability to manage herself were assessed. The procedure of first inviting the women, then assessing their willingness to participate was a gradual process performed to ensure an equal number of women in the six groups described above. In the study design, it was planned that 48 women from each age group would participate, ie 144 women, but as five researchers worked in parallel to invite the women and willingness to participate was difficult to predict, 159 women were included in the study.

Study objects

The study was performed in urban and rural settings. The largest municipality, A, had 187 302 inhabitants, and the smaller municipalities, B and C, had 21 748 and 36 121, respectively (SCB Statistics Sweden, 1999). The proportion of elderly people was comparatively low in municipality A, where two universities, several senior high schools, as well as central, regional and local governments are located. There are businesses in all three towns, and in the rural areas agriculture and forestry.

Sample

The study population comprised all women invited in writing (n=570) minus those who were impossible to reach by phone (n=65), thus a total of 505 women. The inclusion criteria were: retired; mentally oriented; manages food shopping and cooking self-sufficiently. The women were contacted by telephone during the years 1997–1998. The study was approved by the Research Ethics Committee at the Faculty of Medicine, Uppsala University.

Data collection

The telephone calls could be characterized as informal conversations. The researcher gave information about the study, and the women related to this by describing her present situation. Some women had difficulty deciding whether they wanted to participate, and thus wanted to think it over. In these cases, the researcher called a second time and in a few cases also a third time before decisions about participation, exclusion or declining were definite. Many women reported their eating and/or private situation before declining participation and gave a reason for not wanting to participate. Other women just said ‘no thanks’, and in these cases the researcher asked why this decision was made. The calls thus lasted from a few to 30 min. If the women did not fulfil the inclusion criteria, the reasons for exclusion were documented and, if she declined participation, her explanations were documented. All statements were recorded as literally as possible for each woman.

Data analysis

Inclusion rate was calculated by dividing the number of included women by the potential number of participants, ie those who were contacted by phone minus the excluded women. Chi-square (χ2) tests were used to analyse differences in participation, exclusion and declining among different categories of women. Content analysis was used to analyse reasons for exclusion and women's stated explanations for their declining to participate. Each woman's most dominant explanation was used in the analysis. Based on the women's statements, the first author developed one categorization system including six categories covering reasons for exclusion and another categorization system including 11 categories for explanations used to decline participation. Each category was developed by grouping explanations of the same kind and naming the category. This allowed independent raters to partition recorded statements into mutually exclusive categories. The developed categories were discussed in the research group and found to be relevant. Finally, a research nurse (UN), not belonging to the research group, but familiar with the culture of old people, and the first author categorized the statements independently. Interrater reliability was calculated using the following equation: number of agreements/(number of agreements+disagreements) (Polit & Hungler, 1999).

Results

One-hundred and six of the women contacted by telephone were excluded. The number of exclusions did not differ significantly between municipalities, but the proportion of exclusions was significantly higher among the group in their 80s, χ2(DF4)=15.5, P>0.01. Low numbers of single living women in their 60s and cohabiting women in their 70s declined participation, and consequently the highest participation rates (50%) were found among them. These differences were not significant, and the participation rate for the whole sample was 40% (Table 1).

Table 1 Excluded, potential sample, declining, included and participation rate among old women related to municipality, age group and living status

Sixty-seven percent of the exclusions were assigned to the first and second categories ‘Illness, disability or dementia but living at home’ and ‘Living in an institution for old people’. In both these categories, women in their 80s dominated. Also in the category ‘Living at home but not managing cooking’, women in their 80s were more numerous than the other age groups. Among women in their 60s, especially those living alone, the most common reason for exclusion was that they had not yet retired. All these reasons were predetermined criteria for exclusion, as we wished to study self-managing women. Inter-rater reliability for categorization of reasons for exclusion was 0.99. Exclusion categories and age groups are shown in Table 2.

Table 2 Reasons for exclusion

Two-hundred and forty women declined participation. The percentage of declining women in municipally C was somewhat higher than in A and B, however not significantly higher. There were no significant differences between decline by single-living and cohabitant women.

The most common explanations for declining participation were assigned to the first three categories shown in Tables 3 and 4. ‘Lack of time’ was most often expressed among women in their 60s, and ‘Tired, fragile, ill or having a bad memory’ among those in their 80s. ‘Unwillingness to participate in scientific studies’ was one type of declining explanation most often used by single-living women. Together these three categories (of the total 11) cover 57% of the total 240 explanations given. Thirty-two percent of the explanations belonged to the next three categories: ‘Too old and nothing to contribute’, ‘Declining participation without explanation’ and ‘Unwillingness to report eating’. The first of these three categories was most common among women in their 80s, the second covered explanations mostly given by single-living women in all age groups, and the third category was also most common among women in their 80s. Inter-rater reliability for categorization of reasons for declining participation was 0.98. In Table 3, the categories for women's explanations for declining participation are shown as a function of age group and family situation, and in Table 4, each category is explained and illustrated using citations.

Table 3 Explanations for declining participation
Table 4 Categories and quotations explaining declined participation

Discussion

In this study of self-managing older women, the participation rate was 40% (the number of women included divided by the potential sample invited minus those who were impossible to reach by phone and those excluded). Compared with those reported in other studies, which have also included ill older persons, this rate was relatively good. The figure for cohabiting women in their 70s corresponds to the SENECA study as well as to the Norwegian study (Johansson et al, 1997; van't Hof et al, 1991). Compared to Wright et al (1995), who reported a response rate of 20% among those aged over 85 y, the present rate for women in their 80s (36% for single-living and 38% for cohabiting) was encouraging, especially as only older women participated and they are known to have extremely low participation rates. In the MENEW Study I, however, the proportion of excluded women was significantly higher among the group in their 80s. They would probably have belonged to the group of non-participants had the inclusion criteria not been self-managing women. This argument is supported by the fact that the categories ‘Illness, disability or dementia but living at home’ and ‘Living in an institution for old people’ dominate reasons for exclusion in this group. In other words, these women probably represent a group that society should support, but they have been passed over in the MENEW study as well as in other surveys. This result was not unexpected, as arthritis, diabetes, dementia, cardiovascular and pulmonary diseases increase with rising age (Kaye, 1998; Harris et al, 1989; Nyth et al, 1991). Wright et al (1995) found that 78% of older elderly women (74–90 y) used prescribed medicines, indicating poor health. The proportion of excluded women was highest among cohabiting women, indicating that they receive help from their husbands, enabling them to live at home despite disabling disease.

The most commonly used explanation for declining participation was ‘Lack of time’. In Sweden, this phrase is often used to avoid saying ‘no’ and it says very little about the actual cause. Both those interested and uninterested in food may have used it. Presumably they did not want to spend time participating in a study, as they had many other things to do, which is evident in the quotations in Table 4. This explanation was also the most commonly used in the younger group among those who were active.

Women in their 80s, on the other hand, most frequently used explanation numbers 2 and 4 to decline participation, ie ‘Tired, fragile, ill or having bad memory’ and ‘Nothing to contribute and too old’. These two categories probably include women that were ill, and their ability to shop for food and cook is uncertain. As disability in performing these duties has been shown to cause food insecurity (Stitt et al, 1995; Wolfe et al, 1996; Wylie et al, 1999), and malnutrition is common at admittance to hospital (Larsson et al, 1990), it is problematic that this vulnerable group is often under-represented in food surveys. In the MENEW project, study II will focus on disabled women.

The statement ‘Nothing to contribute’ and the explanations ‘Unwillingness to participate in scientific studies’ and ‘Unwillingness to report eating’ indicate that many invited women may have found reporting eating habits problematic. Perhaps they found that their eating was not in accordance with recommendations and were unwilling to report, as Isaksson (1998) argued. Consequently, health-conscious, active persons interested in their diet may be over-represented in this study sample, as Harries et al (1998), Holcomb (1995), van't Hof and Burema (1996) and Berglund (1998) have discussed. This problem is common to all studies based on voluntary participation, and no method to increase participation, such as those described by Harries et al (1989), was used, as they are viewed as unethical in Sweden.

In the Euronut SENECA study, 75% of the explanations for declining participation were assigned to ‘No time/interest’ (van't Hof, 1991). This finding corresponds to our results if explanation numbers 3–7 in Table 4 are included as well. Furthermore, van't Hof et al (1991) and van't Hof and Burema (1996) reported ‘Illness’ as an important reason for non-participation, which corresponds to our second category in Table 4, but above all to the first two categories of reasons for exclusion. However, no other studies referred to in this paper have included only self-managing women and excluded those with illnesses, disability and dementia, which is why comparisons should be made with caution.

Fear of home visits may be hidden in several of the declining explanations given. Harries et al (1989) also found that older people were unwilling to participate in home interviews. On the other hand, Rothenberg et al (1993) had fairly good participation rates in such a study also performed among women. The in-home dietary method was chosen, as we wanted to obtain valid data, which is possible when participants are able to demonstrate their portion sizes and household measures. Furthermore, documentation of eating and drinking in the food diary could be demonstrated and discussed, which was also a way to ensure valid data. Isaksson (1998) also favoured this method of data collection.

Municipality B had the lowest number of inhabitants and was the most sparsely populated region among the study areas (SCB Statistics Sweden, 1999). Here the lowest number of phone calls was made and only 23% were excluded, but the percentage of women declining was almost equal to that of the other two municipalities. These differences are difficult to interpret, but the question is whether the older women in municipally B were healthier than were those in the other municipalities. In contrast, the proportions of excluded and declining women were highest in municipality C, which we attribute to police authorities in this municipality having cautioned older people not to allow strangers into their homes.

To enhance representativeness, stratified random sampling was used. The women were divided into homogeneous subsets, making the probability of selecting a markedly deviant subset low (Polit & Hungler, 1999). The fact that participation rates did not differ significantly among the three municipalities, the three age groups or between single-living and cohabiting women indicates that the representativeness was good. Consequently, unknown characteristics such as attitudes to food, housing, education and financial status were most likely similarly distributed across the subsets. The high inter-rater reliability for categorization of main reasons for exclusion and declining shows the simplicity of interpreting the statements.

The conclusion is that the present participation rate, compared with those of other food surveys among the older generation, was relatively good. The content analysis of women's stated reasons for declining participation indicates that active younger old people, on the one hand, and tired, fragile, ill or frail older old people as well as those with bad memories, on the other, have declined participation in this study. Consequently, as the most active and the very ill and disabled did not participate, most participants belong to an intermediate layer of self-managing old Swedish women. Non-participation studies from other projects in Norway, Europe and the United States reveal the same inclusion problems. The difference is that, in the case of the MENEW Study I, the objective of the random selection method was to include self-managing women, thereby excluding those suffering from diseases and living in institutions. To gain knowledge about disabled older old women's nutritional status, specially directed studies are necessary. When surveys on samples obtained from official registers are performed, non-participants should be studied. This article argues for the importance of using telephone contact to procure information on the particulars of each woman's reasons for declining participation.

References

  1. Andersen NL, Fagt S, Groth MV, Hartkopp HB, Mller A, Ovesen L, Warming DL . 1996 Danish dietary habits 1995 (English summary). National Food Authority, Report no. 235 Copenhagen: Quickly Tryk A/S

    Google Scholar 

  2. Becker W . 1994 Dietary habits and nutrient intake in Sweden 1989. Methods and results (English summary) Uppsala: National Food Administration

    Google Scholar 

  3. Berglund L . 1998 Dietary investigations—what are the effects of invalid selection procedures and measurement errors? Scand. J. Nutr./Näringsforskning 42: 60–62

    Google Scholar 

  4. Dirren HM . 1994 EURONUT-SENECA: a European Study of Nutrition and Health in the Elderly Nutr. Rev. 8: 38–43

    Google Scholar 

  5. Harries T, Woteki C, Briefel RR, Kleinman JC . 1989 NHANES III for older persons: nutritional content and methodological considerations Am. J. Clin. Nutr. 50: 1145–1149

    Article  Google Scholar 

  6. Holcomb CA . 1995 Positive influence of age and education on food consumption and nutrient intakes of older women living alone J. Am. Diet. Assoc. 95: 1381–1386

    CAS  Article  Google Scholar 

  7. Isaksson B . 1998 Methodology problems of dietary assessment in analytical studies (English summary) Scand. J. Nutr./Näringsforsk-ning 42: 50–55

    Google Scholar 

  8. Johansson L, Solvoll K, Björneboe G-E Aa, Drevon CA . 1997 Dietary habits among Norwegian men and women Scand. J. Nutr./Näringsforskning 41: 63–70

    Google Scholar 

  9. Kaye JA . 1998 Diagnostic challenges in dementia Neurology 51: S45–S52

    CAS  Article  Google Scholar 

  10. Kleemola P, Virtanen M, Pietinen P . 1994 The 1992 dietary survey of Finnish adults Publication of the National Public Health Institute, B2/1994 Helsinki: National Public Health Institute

    Google Scholar 

  11. Larsson G, Unosson M, Ek A-C, Nilsson L, Thorslund S, Bjurulf P . 1990 Effect of dietary supplement on nutritional status and clinical outcome in 501 geriatric patients—a randomised study Clinical Nutr. 9: 179–184

    CAS  Article  Google Scholar 

  12. Lyon P, Colquhoun A . 1999 Home, hearth and table: a centennial review of nutritional circumstances of older people living alone Aging Society 19: 53–67

    Article  Google Scholar 

  13. Nyth AL, Gottfries CG, Blennow K, Bråne G, Wallin A . 1991 Heterogeneity of the course of Alzheimer's disease: a differentiation of subgroups Dementia 2: 18–24

    Google Scholar 

  14. Polit FD, Hungler BP . 1999 Nursing Research. Principles and Methods Philadelphia, PA: JB Lippincott

    Google Scholar 

  15. Rothenberg E, Bosaeus I, Steen B . 1993 Intake of energy, nutrients and food items in an urban elderly population Aging Clin. Exp. Res. 5: 105–116

    CAS  Article  Google Scholar 

  16. SCB Statistics Sweden. 1999 Statistical Yearbook of Sweden Örebro: Publication Services

  17. Sidenvall B, Nydahl M, Fjellström C . 2000 The meal as a gift. The meaning of cooking among retired women J. Applied Gerontol. 19: 405–423

    Article  Google Scholar 

  18. Sidenvall B, Nydahl M, Fjellström C . 2001 Managing food shopping and cooking: the experiences of older Swedish women Ageing Society 21: 135–152

    Article  Google Scholar 

  19. Steingrimsdottir L, Torgeirsdottir H, Aegisdottir S . 1991 Könnun a mataraedi Islenndinga 1990. 1. Helstu nidurstödur (in Icelandic) Reykjavik: National Nutrition Council Prentsmidjan Gutenberg hf

    Google Scholar 

  20. Stitt S, O'Connell C, Grant D . 1995 Old, poor and malnourished Nutr. Health 10: 135–154

    CAS  Article  Google Scholar 

  21. Szebehely M . 1998 Hjälp i hemmet i nedskärningstid—hemtjänstens och anhörigas insatser för gamla kvinnor och män In Åt var och en efter behov ed. Sandqvist A-M Stockholm: Kommentus förlag

    Google Scholar 

  22. van't Hof MA, Burema J . 1996 Assessment of bias in the SENECA study Eur. J. Clin. Nutr. 50: (Suppl 2): S4–S8

    PubMed  Google Scholar 

  23. van't Hof MA, Hautvast JGAJ, Schroll M, Vlachonikolis IG . 1991 Design, methods and participation Eur. J. Clin. Nutr. 45: (Suppl 3): S5–S22

    Google Scholar 

  24. Webb GP, Copeman J . 1996 The Nutrition of Older People London: Arnold and Age Concern

    Google Scholar 

  25. Wolfe WS, Olson CM, Kendall A, Frongillo EA Jr . 1996 Understanding food insecurity in the elderly: a conceptual framework J. Nutr. Educ. 28: 92–100

    Article  Google Scholar 

  26. Wright AJA, Southon S, Bailey AL, Finglas PM . 1995 Nutrient intake and biochemical status of non-institutionalized elderly subjects in Norwich: comparison with younger adults and adolescents from the same general community Br. J. Nutr. 74: 453–475

    CAS  Article  Google Scholar 

  27. Wylie C, Copeman J, Kirk SFL . 1999 Health and social factors affecting the food choice and nutritional intake of elderly people with restricted mobility J. Hum. Nutr. Diet. 12: 375–380

    Article  Google Scholar 

Download references

Author information

Affiliations

Authors

Corresponding author

Correspondence to B Sidenvall.

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Sidenvall, B., Fjellström, C., Andersson, J. et al. Reasons among older Swedish women of not participating in a food survey. Eur J Clin Nutr 56, 561–567 (2002). https://doi.org/10.1038/sj.ejcn.1601359

Download citation

Keywords

  • participation
  • elderly
  • women
  • food survey

Further reading

Search