M Brustad 1 *, G Skeie 1, T Braaten 1, N Slimani 2 and E Lund 1

Transcription

1 ORIGINAL COMMUNICATION in the assessment of dietary intake by the 24 h recall EPIC SOFT program the Norwegian calibration study M Brustad 1 *, G Skeie 1, T Braaten 1, N Slimani 2 and E Lund 1 (2003) 57, ß 2003 Nature Publishing Group All rights reserved /03 $ Institute of Community Medicine, University of Tromsø, Norway; and 2 Unit of Nutrition and Cancer, International Agency for Research on Cancer, Lyon, France Objective: To compare food group intakes in grams, total energy and energy contribution from macronutrients between two random samples of women, using a standardized computer program (EPIC SOFT) with either a face-to-face or a telephone 24 h recall interview design. Methods: Two groups of Norwegian women living in Tromsø were drawn at random from the Norwegian Women and Cancer Study (NOWAC) cohort. The EPIC SOFT computer program was used to conduct single 24 h dietary recalls either by telephone or face to face. For the latter, 160 women were invited, of whom 111 responded positively (crude response rate ¼ 69.4%) and 102 were interviewed. For the telephone option, 180 were invited to participate, 109 responded positively (crude response rate ¼ 60.6%) and 103 were interviewed. Results: The two groups of women were similar with respect to age, body mass index, basal metabolic rate, smoking, education, physical activity and special diet status. No statistically significant difference in dietary intake was found between interviews conducted by telephone and face-to-face, except for egg and egg products (P < 0.01), for which intake was higher in the telephone group. Total energy intake differed, but the difference was attributable to one interviewer. The distribution of energy-contributing nutrients was not significantly different between groups, except for reported energy percentage from protein (P ¼ 0.02), which was significantly higher in the face-to-face group. Conclusion: The telephone vs face-to-face interview design did not influence recalled diet when using the EPIC SOFT program. (2003) 57, doi: =sj.ejcn Keywords: 24 h dietary recall; telephone; dietary intake methodology; interviewer effect; EPIC SOFT Introduction The European Prospective Investigation into Cancer and Nutrition (EPIC) is a large multi-centre prospective cohort study involving more than subjects in 10 European countries (Germany, The Netherlands, UK, France, Spain, Italy, Greece, Sweden, Denmark and Norway; Riboli & *Correspondence: M Brustad, Institute of Community Medicine, University of Tromsø, N-9037 Tromsø, Norway. magritt.brustad@ism.uit.no Guarantor: M Brustad. Contributors: MB wrote the manuscript and performed the statistical analysis. GS contributed to the writing of the paper. NS developed the general concept and design of the EPIC-SOFT program and contributed to the writing of the paper. TB supervised the statistical analysis. EL designed and supervised the study. Received 14 September 2001; revised 26 March 2002; accepted 28 March 2002 Kaaks, 1997). Different dietary questionnaires and strategies for data collection have been used to assess usual dietary intake of individual subjects in each country cohort. In order to compare and pool dietary data obtained from these different baseline dietary assessments, additional single 24 h dietary recalls, used as common calibration measurements, were collected from a representative sub-sample of the study populations (Kaaks & Riboli, 1997). A highly standardized 24 h dietary recall computer program named EPIC SOFT, adapted to each of the 10 European countries participating in EPIC, was developed for the so-called EPIC calibration study (Slimani et al, 1999, 2000). In all countries except Norway, the 24 h recall interviews were conducted face-to-face. However, in Norway, the cohort was drawn at random from the total population, and because of the wide geographical distribution of participants throughout a sparsely populated country, the face-to-face method used in the

2 108 other EPIC centres was modified for telephone interview to collect the 24 h dietary recalls needed for the calibration study. Studies to compare 24 h face-to-face dietary recalls with telephone recalls to assess mean energy and protein intakes (Bogle et al, 2001), macronutrients and micronutrients (Galasso et al, 1994) and total energy intake and energy expenditure (Tran et al, 2000) have yielded satisfactory results. Traditionally, 24 h dietary recall has been recorded by a face-to-face interview, but interviews conducted by telephone have become increasingly common (Andersson & Rossner, 1996; Bogle et al, 2001; Tran et al, 2000; Fox et al, 1992; Casey et al, 1999) and are considered as a feasible alternative to expensive and time-consuming face-to-face methods (Fox et al, 1992; Posner et al, 1982). To evaluate the validity of the method of telephone interviews used for the Norwegian 24 h dietary recalls vs the face-to-face approach used in the other EPIC countries, a comparison study was undertaken based on random sampling of two groups of women from the Norwegian cohort. The aim was to compare reported food group intakes in grams, total energy intake and energy contribution from macronutrients using EPIC SOFT with either a face-to-face or a telephone 24 h recall interview design similar to the interviews in the Norwegian part of the EPIC calibration study. Methods Sample Two groups of women living in Tromsø, the major city in northern Norway were drawn at random from the Norwegian Women and Cancer cohort, NOWAC (Lund & Gram, 1998; Hjartaker & Lund, 1998). The two groups were randomly selected among the 494 eligible women in the city, who remained after the subjects had been sampled for the EPIC calibration study. This comparison study was restricted to Tromsø to avoid practical problems of long travel distances in northern Norway. For the face-to-face interviews, 160 women were invited to participate. One woman had died shortly before the start of the study. A total of 111 women responded positively after two reminders, giving a crude response rate of 69.4%. Nine women were absent due to disease or holidays. Thus, 102 interviews were completed. For the telephone interviews, a total of 180 women were invited, of whom two had moved from the city. A total of 109 answered positively after one reminder, giving a crude response rate of 60.4%. However, only 103 were interviewed, since the desired number of interviews conducted in each group was set at 100 and interviewing was stopped when the group sizes were equal. The Norwegian picture booklet The methodology of the standardized 24 h recall based on the EPIC software has been described in detail elsewhere (Slimani et al, 1999). For the Norwegian telephone design, some adjustments were needed. The original EPIC SOFT picture book (Van Kappel et al, 1994), with 146 series of pictures used to quantify food portion sizes, contained photos for all EPIC countries. Not all these pictures were relevant in a Norwegian context. The EPIC SOFT picture book was too large to be mailed to the 3000 women invited to participate in the Norwegian calibration study. Therefore, an adapted picture booklet was made based on a sample of 45 photo series from the original EPIC SOFT picture book, with different foods and dishes with different portion sizes relevant for the Norwegian diet. In addition, four series of pictures of different bread-shapes were added to the Norwegian booklet together with five series of photos of glasses, cups and plates of different size. The final adapted picture booklet had 36 pages. Data collection procedures The face-to-face interviews were performed at the University Hospital of Tromsø in the Department of Clinical Research. For the face-to-face interviews, plates, cups, glasses and spoons were used for quantification purposes, together with the Norwegian food quantification booklet that was mailed to the participants in the calibration study. As part of a feasibility study, blood samples were collected from women who were interviewed face to face. The women interviewed by telephone were mainly called at home, but a few were called at work, according to their own choice. The quantification booklet was mailed to the group interviewed by telephone and the interview routines were similar to the recalls in the calibration study. All interviews in both groups were conducted between Monday and Friday, from February to June The EPIC SOFT did not contain questions on education status, physical activity and smoking. These variables were therefore based on data from a NOWAC questionnaire answered by the women in Data on height and weight were recorded in the 24 h recall interviews. Interviewers The interviews were conducted by seven interviewers who had all been trained by the nutritionist responsible for the Norwegian part of the EPIC calibration study. The nutritionist monitored the interviews, provided feedback to the interviewers, answered questions as they arose, and edited the final interviews. Three trained nurses working at the University Hospital of Tromsø in the Department of Clinical Research conducted all face-to-face interviews. Three students at the University in Tromsø and one dietician conducted the telephone interviews.

3 Statistical methods In EPIC SOFT, food items are classified into 17 common food groups. The analysis was conducted using this food classification system that is common across the EPIC countries. The energy intake was calculated based on the country-specific food composition figures already implemented in the Norwegian version of the EPIC SOFT program. These figures where mainly based on the Norwegian Food Composition Table (National Nutrition Council & Norwegian Food Control Autority, 1995). Basal metabolic rate (BMR) for each subject was estimated based on sex, age, weight, and height according to FAO=WHO=UNU (1985). Statistical analyses were performed using the SPSS software package, version Parametric methods were used when analysing distribution of background variables among the study groups and total energy intake. Intakes at food group level and energy distribution were, however, not normally distributed, mainly due to the large number of non-consumers associated with each food group. Thus, non-parametric methods were used when analysing these variables. Statistical comparisons between groups were performed by Student s t-test, Kruskal Wallis test, Mann Whitney test, linear regression analysis, or chi-square statistics when appropriate. Results The two randomized groups of women were comparable with respect to the characteristics outlined in Table 1. The two groups did not differ significantly according to weekdays of interview (Pearson chi-square, P ¼ 0.56). The numbers of subjects reporting illness at the recording day were significantly different in the two groups (Pearson chi-square, Table 1 Distribution of background variables among the study groups Face-to-face Telephone Variable n ¼ 102 n ¼ 103 P-values Age at recall, y (mean) a Basal metabolic rate, kcal (mean) a Body mass index, kg=m 2 (mean) a Special diet (n (%)) yes 11 (10.8) 12 (11.7) no 91 (89.2) 91 (88.3) 1.00 b Diet on recorded day considered representative (n (%)) yes 68 (66.7) 79 (76.7) no 34 (33.3) 24 (23.3) 0.12 b Education (y) (n (%)) 7 7 (7) 3 (3) (14) 11 (12) (26) 29 (32) > (53) 48 (53) 0.54 b Physical activity (n (%)) low 14 (16) 14 (16) moderate 69 (78) 65 (73) high 5 (6) 10 (11) 0.41 b Smoking (n (%)) yes 36 (39) 26 (31) no 57 (61) 58 (69) 0.34 b a Student t-test. b Pearson chi-square. P ¼ 0.03), being two and nine subjects in the face-to-face and telephone groups, respectively. For the analyses of total energy and energy distribution, these subjects were not included. No statistically significant difference in reported dietary intakes was found between subjects interviewed by telephone or face-to-face, except for egg and egg products (P < 0.01), of which intake was higher in the telephone group (Table 2). For reported intake of potatoes and tubers, there was a difference of borderline significance (P ¼ 0.052), with a somewhat higher reported intake among the face-to-face group than in the telephone group. Thus, only one of 17 food groups differed significantly between the women interviewed face-to-face and by telephone. As shown in Table 3, the group interviewed by telephone reported significantly lower energy intake than the face-toface group (P ¼ 0.02). The energy contribution from macronutrients did not differ significantly between the groups, except for the energy percentage from protein, which was significantly higher in the telephone group (P ¼ 0.02). When mean energy intake per interviewer was assessed as the dependent variable by the Kruskal Wallis test, the difference attributed to the interviewers was of borderline significance (P ¼ 0.056), as shown in Table 4. The interviewer recording the highest mean energy had kcal which was kcal more than the interviewer recording the second highest mean energy intake. When analysing difference in energy between interviewers by linear regression the parameter estimates did not change when adjusting for body mass index and physical activity. By excluding the interviewer who recorded the highest energy intakes the P-values increased from to The difference in total energy intake between the two groups of women disappeared when the same interviewer was excluded from the analysis (P ¼ 0.46). Table 5 shows the impact of excluding one interviewer at a time on mean energy intake in the face-to-face vs the telephone group. Only exclusion of interviewer no. III caused an obvious reduction in difference between the two groups. The mean estimated energy intake : basal metabolic rate (EI : BMR) for the face-to-face and telephone groups was 1.39 and 1.27, respectively (P ¼ 0.04; subjects reporting illness on recall day were excluded). There was no significant difference in EI : BMR between the face-to-face group and the telephone group nor any difference between interviewers when the interviewer with the significantly higher energy was excluded. Discussion In this paper, we have presented the results of a study to compare the impact on reported diet and energy intake of telephone interviews vs face-to-face interviews for 24 h dietary recalls. The major conclusion of this study is that the EPIC SOFT 24 h dietary recall program can be administered by telephone or in-person interviews without a significant 109

4 110 Table 2 Comparison of reported intake of food groups (in grams) for the two interview approaches Food groups Face-to-face (n ¼ 102) Mean median (25th and 75th percentiles) Telephone (n ¼ 103) Mean Median (25th and 75th percentiles) P-value a Potatoes and tubers (0.0, 142.0) 39.2 (0.0, 117.6) Vegetables (41.1, 156.5) (50.0, 205.5) Legumes Fruits (0.0, 258.5) (0.0, 250.5) Dairy products (73.8, 391.6) (48.0, 374.5) Cereals and cereal products (99.5, 259.3) (108.6, 232.0) Meat and meat products (11.5, 143.1) 45.0 (13.0, 130.2) Fish and shellfish (0.0, 126.3) 5.0 (0.0, 119.5) Eggs and egg products (0.0, 0.0) 0.0 (0.0, 16.3) Fat (3.6, 21.6) 9.7 (2.3, 20.0) Sugar and confectionery (3.9, 75.0) 21.0 (0.0, 52.5) Cakes (0.0, 71.8) 0.0 (0.0, 77.0) Non-alcoholic beverages (1322.7, ) (1400.0, ) Alcoholic beverages Condiments and sauces (0.0, 50.0) 10.0 (0.0, 61.7) Soups and bouillon Miscellaneous a Mann Whitney test. Table 3 Intake of total energy and energy contribution from macronutrients a Energy contribution Face-to-face (n ¼ 100) Telephone (n ¼ 94) P-values Total energy (kcal) b Energy from fat 35.7% 33.9% 0.07 c Energy from protein 15.1% 16.9% 0.02 c Energy from carbohydrates 47.9% 46.5% 0.35 c Energy from alcohol 1.5% 2.8% 0.28 c a Subjects reporting illness on the recall day were not included in the analysis. b Student s t-test. c Mann Whitney test. Table 4 Interview setting Mean energy intake reported by interviewer a Interviewer Number of interviews Mean energy per interviewer (kcal) Face-to-face I II III Telephone IV V VI VII P-value b 0.06 P-value, excluding interviewer III 0.88 a Subjects reporting illness on the recall day were not included in the analysis. b Kruskal Wallis test. difference in estimated dietary intake. Mean total energy intake did differ, but this difference was entirely attributable to one interviewer. Several studies have concluded that telephone and faceto-face diet recalls appear to be interchangeable (Galasso et al, 1994; Tran et al, 2000; Casey et al, 1999; Krantzler et al, 1982). Fox et al (1992) reviewed telephone surveys as a method for obtaining dietary data and indicated that well designed and well administered telephone surveys are as good as, and may be better than, other dietary assessment methods. Morgan et al (1987) examined nine different methods for assessing dietary status, including in-person interview, telephone, mail and combinations of these meth-

5 Table 5 Single interviewers impact on mean energy intake a Face-to-face Mean energy Telephone Excluded interviewer (kcal) n (kcal) n P-value b All interviewers included Interviewer I excluded Interviewer II excluded Interviewer III excluded Interviewer IV excluded Interviewer V excluded Interviewer VI excluded Interviewer VII excluded a Subjects reporting illness on the recall day were not included in the analysis. b Student t-test. ods. It was concluded that 24 h dietary recall by telephone involved less effort and costs, and that the dietary data obtained were similar to those collected by other methods. Face-to-face interviews are time-consuming and expensive, especially for large population-based dietary surveys (Fox et al, 1992; Marcus & Crane, 1986). The cost of 24 h dietary recalls conducted by telephone is less than that of face-toface interviews (Derr et al, 1992), by as much as a factor of two (Weeks et al, 1983). Although most comparisons of 24 h dietary recalls conducted by telephone vs face-to-face interview have been performed using rather small samples (Galasso et al, 1994; Tran et al, 2000; Krantzler et al, 1982); Casey et al (1999) compared 700 telephone 24 h dietary recalls with 550 faceto-face interviews. This study concluded that 24 h dietary recalls conducted by telephone are a practical and valid tool for use in national food consumption surveys. One of the main advantages of conducting interviews by telephone compared with face-to-face is that the sampling is not geographically restricted (Marcus & Crane, 1986). Since the NOWAC study is a nationwide survey, conducting faceto-face interviews for the Norwegian calibration study, as in the rest of the EPIC countries, would have been logistically complicated. Thus, it was essential to develop an alternative design for the 24 h recall to be used in the calibration study for Norway. The design of our study was based on comparison of independent groups, whereas previous similar analyses have been based on paired observations performed on the same subjects (Galasso et al, 1994; Tran et al, 2000; Krantzler et al, 1982). The design of the present study where two different groups of women were compared has its strengths and limitations. The main advantage was that data collected where not biased by a learning effect. However, this could be avoided by randomizing the succession of the two interview settings for each participating subject. Another advantage of the design was that it was less demanding on each individual subject. Further, it is likely that a crossover design could have been associated with more practical problems, eg having to reach each subject twice causing a dropout problem. Seasonal variation is less likely to affect a design where the interviewing of the two separate groups runs parallel in time, compared to a crossover design where a randomization with respect to season is needed. The two groups were similar with respect to basic characteristics, indicating a satisfactory randomized sampling procedure. A higher response rate in the face-to-face interview than in the telephone group has been found in other studies (Marcus & Crane, 1986; Fox et al, 1992). Although the face-to-face interview in our study was more demanding on the interview subjects than the telephone interview, a greater proportion of those invited to the face-to-face interview attended. This may indicate that people tend to consider an in-person interview to be associated with greater integrity than a telephone interview. However, the higher response rate among the in-person interview group was most likely due to the fact that, to reach the required 100 participants in each group, two reminders were sent to this group, while only one reminder was sent to the subjects invited for telephone interview. One concern about telephone interviews is non-covering bias due to the exclusion of subjects without a telephone (Fox et al, 1992; Bogle et al, 2001). Since nowadays almost all households in Norway have a telephone, this is unlikely to have caused selection bias in the present study. However, it is likely that people reached by telephone differ from those attending a face-to-face interview. This was demonstrated by the significantly higher number of subjects that reported illness on the recording day among the subjects interviewed by telephone compared with the face-to-face group. It is likely that ill people will be less willing to make the effort to attend a face-to-face interview, especially if it requires going out, travel, etc. In our study, no controlled registration of the number of changed appointments due to illness was made. Due to the unequal number of people with illness on the recall day, these subjects were excluded from the analysis on reported energy, thus illness could not explain the difference in energy between groups. The significantly higher mean energy intake recalled in the face-to-face interviewed group was attributed to one interviewer. When this interviewer was excluded from the analyses, the difference disappeared. This change in statistical significance could be caused by the lower statistical power resulting from the smaller sample size. This seems unlikely, however, since exclusion of any of the other interviewers did not have the same effect (Table 5). Identifying over- or under-reporting in energy intake in single 24 h dietary recalls at the individual level is associated with uncertainties due to day-to-day variation in diet. Several repeated measurements would be needed to estimate accurately the long-term usual individual intakes. Therefore, we assessed the mean recalled EI : BMR at group level. The mean EI : BMR for the two groups was below the lower confidence limit defined by Goldberg s cut-off* calculated 111

6 112 *Lower 95% CI ¼ 1.46 (light mean physical activity level (PAL) ¼ 1.55 for the group). according to Black (2000). Under-reporting has been considered a typical characteristics for the 24 h dietary recall method (Carter et al, 1981) and this was supported by our study. Based on our data there is no indication that underreporting is more common in either a face-to-face or a telephone approach since the difference in EI : BMR in the two groups in our study was most likely due to an interviewer effect. The 24 h dietary recall method s vulnerability to an interviewer effect has been found elsewhere (Marcus & Crane, 1986). Slimani et al (2000) have assessed, by analysis of variance, the mean energy intake for the 90 interviewers in the EPIC study. The difference in energy intake observed among interviewers in some EPIC centres was entirely attributable to one interviewer. Slimani et al addressed the need for development of more comprehensive statistical models to correct for, among other things, errors attributable to the interviewers. The interviews assessed by Slimani et al were all in-person interviews. It might be that a face-to-face approach is more vulnerable to an interviewer effect than a telephone interview, since the interviewer has a greater ability to influence the interview in the former situation (Fox et al, 1992). The present study supports this assumption, but further research is needed to address issues related to interviewer effects in telephone interviews and in-person interviews. The fact that different interviewers conducted the interviews in both methods can be considered a weak point in the study design because interviewers caused differences between groups in dietary data collected. However, the EPIC SOFT program is a standardized interview program and should therefore help minimize interviewer effects. When different interviewers were conducting face-to-face interviews and telephone interviews, we were not only assessing the impact on the setting (face-to-face vs telephone), but also the impact of the sum of the setting and individual interviewers effects on dietary data collected. The objective for our study was to assess whether or not the Norwegian 24 h dietary recalls conducted by telephone collected for the EPIC calibration study could be considered comparable with the face-to-face interviews from rest of the EPIC centres. Thus, our study setting, with different interviewers in the telephone and face-to-face group, was close to the real situation regarding the Norwegian 24 h dietary recall data and our participation in the EPIC calibration study. The interviewers conducting the telephone interviews seemed to take less time than those conducting face-to-face interviews (data not shown). To record duration of each interview by EPIC-SOFT, the interview program had to be opened the moment each interview started and saved when each interview was done. This was not done systematically by the interviewers. Due to inadequate routines for measuring time spent on each interview, the time variable is not included in the analysis, but it is likely that the duration of the interview could have affected the quality of data collected. It is possible that spending more time probing and questioning would help the interviewee to remember more. On the other hand, if the longer time was due to the interviewer s comparatively poor interviewing routines, the interviewee might lose interest and hence less accurate data might be collected. One important difference between the telephone and inperson interviews in the present study was that data collected in telephone interviews was not biased by the subjects knowing in advance the exact day for the interview. Although the subjects interviewed face-to-face were not aware that they would be asked to recall their previous day s diet, the surprise effect is more present in a telephone interview than in an in-person interview with a set appointment for the interview (Derr et al, 1992). The quantification methods did differ between the two interview settings. While the subjects interviewed by telephone used only the food quantification booklet, the subjects in the face-to-face interviews additionally used glasses, plates, spoons etc. Lyu et al, (1998) found that participants in a telephone-administered 24 h recall were able to use photographs for estimating food amounts. In our study, the photographs in the booklet were smaller than in the EPIC picture book, while the participants for the in-person interview could use models of the plates and glasses used in the booklet. This could have influenced the recalled portion sizes. However, since the recalled dietary intake did not differ between groups, there is no indication that the different quantification methods used did affect the quality of the dietary data collected. A weak point in the study design was that the Norwegian picture booklet was used in both the face-to-face and telephone interviews. If the original EPIC picture book had been used in the face-to-face interviews, issues related to differences in methodology between Norway and the other EPIC centres, such as the impact of restricted number of pictures and smaller photos on mean estimates, could have been assessed. In conclusion, although in the Norwegian part of the EPIC calibration study the 24 h diet recalls were conducted by telephone instead of by in-person interviews as in the rest of the EPIC countries, the data collected by telephone appear to be interchangeable with dietary data collected face-toface. However, the results of this study confirm that the 24 h dietary recall method is vulnerable to an interviewer effect. The benefit of using the telephone as an alternative to the face-to-face approach should be considered in the further implementation of the EPIC-SOFT program. Acknowledgements The work described in this paper was carried out with support from Europe against Cancer Programme of the European Commission, and Department of Clinical

7 Research, University Hospital of Tromsø. The authors thank all the interviewers who conducted the interviews and Dr John Cheney for revising the English language. References Andersson I & Rossner S (1996): Meal patterns in obese and normal weight men: the Gustaf study. Eur. J. Clin. Nutr. 50, Black AE (2000): Critical evaluation of energy intake using the Goldberg cut-off for energy intake: basal metabolic rate. A practical guide to its calculation, use and limitations. Int. J. Obes. Relat. Metab. Disord. 24, Bogle M, Stuff J, Davis L, Forrester I, Strickland E, Casey PH, Ryan D, Champagne C, McGee B, Mellad K, Neal E, Zaghloul S, Yadrick K & Horton J (2001): Validity of a telephone-administered 24-hour dietary recall in telephone and non-telephone households in the rural lower Mississippi delta region. J. Am. Diet. Assoc. 101, Carter RL, Sharbaugh CO & Stapell CA (1981): Reliability and validity of the 24-hour recall. J. Am. Diet. Assoc. 79, Casey PH, Goolsby SL, Lensing SY, Perloff BP & Bogle ML (1999): The use of telephone interview methodology to obtain 24-hour dietary recalls. J. Am. Diet. Assoc. 99, Derr JA, Mitchell DC, Brannon D, Smiciklas WH, Dixon LB & Shannon BM (1992): Time and cost analysis of a computer-assisted telephone interview system to collect dietary recalls. Am. J. Epidemiol. 136, FAO=WHO=UNU (1985): Energy and protein requirements. Report of a joint FAO=WHO=UNU consultation. Technical Report Services. Geneva: World Health Organization. Fox TA, Heimendinger J & Block G (1992): Telephone surveys as a method for obtaining dietary information: a review. J. Am. Diet. Assoc. 92, Galasso R, Panico S, Celentano E & Del-Pezzo M (1994): Relative validity of multiple telephone versus face-to-face 24-hour dietary recalls. Ann. Epidemiol. 4, Hjartaker A & Lund E (1998): Relationship between dietary habits, age, lifestyle, and socio-economic status among adult Norwegian women. The Norwegian Women and Cancer Study. Eur. J. Clin. Nutr. 52, Kaaks R & Riboli E (1997): Validation and calibration of dietary intake measurements in the EPIC Project: methodological considerations. European Prospective Investigation into Cancer and Nutrition. Int. J. Epidemiol. 26(Suppl 1), S15 S25. Krantzler NJ, Mullen BJ, Schutz HG, Grivetti LE, Holden CA & Meiselman HL (1982): Validity of telephoned diet recalls and records for assessment of individual food intake. Am. J. Clin. Nutr. 36, Lund E & Gram IT (1998): Response rate according to title and length of questionnaire. Scand. J. Soc. Med. 26, Lyu LC, Hankin JH, Liu LQ, Wilkens LR, Lee JH, Goodman MT & Kolonel LN (1998): Telephone vs face-to-face interviews for quantitative food frequency assessment. J. Am. Diet. Assoc. 98, Marcus AC & Crane LA (1986): Telephone surveys in public health research. Med Care 24, Morgan KJ, Johnson SR, Rizek RL, Reese R & Stampley GL (1987): Collection of food intake data: an evaluation of methods. J. Am. Diet. Assoc. 87, National Nutrition Coucil & Norwegian Food Control Authority (1995): The Norwegian Food Composition Table Oslo: Universitetsforlaget. Posner BM, Borman CL, Morgan JL, Borden WS & Ohls JC (1982): The Validity of a telephone-administered 24-hour dietary recall methodology. Am. J. Clin. Nutr. 36, Riboli E & Kaaks R (1997): The EPIC Project: rationale and study design. European Prospective Investigation into Cancer and Nutrition. Int. J. Epidemiol. 26(Suppl 1), S6 S14. Slimani N, Deharveng G, Charrondiere RU, van Kappel AL, Ocke MC, Welch A, Lagiou A, van Liere M, Agudo A, Pala V, Brandstetter B, Andren C, Stripp C, van Staveren WA & Riboli E (1999): Structure of the standardized computerized 24-h diet recall interview used as reference method in the 22 centers participating in the EPIC Project. European Prospective Investigation into Cancer and Nutrition. Comput. Methods Programs Biomed. 58, Slimani N, Ferrari P, Ocke M, Welch A, Boeing H, Liere M, Pala V, Amiano P, Lagiou A, Mattisson I, Stripp C, Engeset D, Charrondiere R, Buzzard M, Staveren W & Riboli E (2000): Standardization of the 24-hour diet recall calibration method used in the European Prospective Investigation into Cancer and Nutrition (EPIC): general concepts and preliminary results. Eur. J. Clin. Nutr. 54, Tran KM, Johnson RK, Soultanakis RP & Matthews DE (2000): Inperson vs telephone-administered multiple-pass 24-hour recalls in women: validation with doubly labeled water. J. Am. Diet. Assoc. 100, Van Kappel AL, Amoyel J, Slimani N, Vozar B & Riboli E (1994): EPIC- SOFT Picture Book for Estimation of Food Portion Sizes. Lyon: International Agency for Research on Cancer. Weeks MF, Kulka RA, Lessler JT & Whitmore RW (1983): Personal versus telephone surveys for collecting household health data at the local level. Am. J. Public Health 73,