MARTINI/C:04 LORD_c

MARTINI/C:04 LORD_c

20/11/08

10:55

Page 1

The Journal of Nutrition, Health & Aging© Volume , Number , 2008

THE JOURNAL OF NUTRITION, HEALTH & AGING©

VALIDATION OF A FOOD FREQUENCY QUESTIONNAIRE FOR WOMEN WITH OSTEOPOROSIS G.A.P. PEREIRA1, P.S. GENARO1, L.C. SANTOS1, K.S. SARKIS1, M.M. PINHEIRO2, V.L. SZJENFELD2, N.J. SCHUCH, L.A. MARTINI1 1. Nutrition Department – School of Public Health- São Paulo University; 2. Reumathology Division – São Paulo Federal University. Corresponding author: Lígia A Martini, PhD, Departamento de Nutrição, Av Dr Arnaldo, 715, São Paulo – SP CEP 01246-904, Brazil

Abstract: The importance of nutrient intakes in osteoporosis prevention in treatment is widely recognized. The objective of the present study was to develop and validate a FFQ for women with osteoporosis. The questionnaire was composed of 60 items, separated into 10 groups. The relative validation was accomplished through comparison of the 3-Day Food Record (3DR) with the FFQ. The 3DR was applied to 30 elderly women with confirmed osteoporosis, and after 45 days the FFQ was administrated. Statistical analysis comprised the Kolmogorov–Smirnov, Student T test and Pearson correlation coefficient. The agreement between two methods was evaluated by the frequency of similar classification into quartiles, and by the Bland-Altman method. No significant differences between methods were observed for the mean evaluated nutrients, except for carbohydrate and magnesium. Pearson correlation coefficients were positive and statistically significant for all nutrients. The overall proportion of subjects classified in the same quartile by the two methods was on average 50.01% and in the opposite quartile 0.47%. For calcium intake, only 3% of subjects were classified in opposite extreme quartiles by the two methods. The Bland-Altman analysis demonstrated that the differences obtained by the two methods in each subject were well distributed around the mean of the difference, and the disagreement increases as the mean intake increases. These results indicates that the FFQ for elderly women with osteoporosis presented here is highly acceptable and is an accurate method that can be used in large-scale or clinical studies for evaluation of nutrient intakes in a similar population. Key words: Dietary investigation, validation, nutrient intakes, bone metabolism, elderly.

Introduction

questionnaire must be validated or calibrated for each population (5). The relative validation of the FFQ is generally evaluated by the comparison of results with a reference method, for example a multiple 24 hour recall, a food record, or a duplicate portions analyses, since a gold standard is not available to estimate the food intake (6). Taking into consideration the importance of an adequate evaluation of nutrient intakes with regards to monitoring the relationship between diet and disease, the present study was undertaken in order to validate a Food Frequency Questionnaire for osteoporotic women.

Osteoporosis is a generalized skeletal disorder characterized by low bone mineral density (BMD), deterioration of the microarchitecture of bone tissue and susceptibility to fracture. Most frequently it occurs in postmenopausal women and the aged. It is a chronic condition of multifactorial etiology and is a major global healthcare problem in developed countries and rising in developing countries (1). Fractures due to osteoporosis considerably increase mortality and morbidity rates, reducing the mobility and quality of life in this population (2). The role of nutrient intakes in osteoporosis prevention and treatment have been investigated for decades, with calcium, vitamin D, phosphorus, magnesium, and protein being the most investigated, since adequate intake of these positively contribute to bone metabolism. The evaluation of nutrient intakes is a complicated task, due to the weaknesses of data-gathering techniques, human behavior, the natural tendencies of an individual’s nutrient intakes to vary considerably from day to day, and the limitations of nutrient composition tables and databases (3). The food frequency questionnaire (FFQ) is considered the most practical and informative method of dietary food intake assessment and essentially important for epidemiological studies that relate food with the occurrence of nontransmissible chronic diseases (4). However, incomplete or excessive food lists can compromise its validity, so the

Patients and methods Patients and study design The study comprised 30 postmenopausal women with osteoporosis, attending São Paulo Hospital outpatient clinic. The study protocol was approved by the institution’s ethics committee from São Paulo University and São Paulo Federal University. All patients provided signed informed consent forms. Bone mineral density of the femoral neck, spine, and total body were measured with a model DPX dual-energy X-ray absorptiometer (Lunar Radiation Corp, Madison, WI). Software version 3.6z was used for data acquisition and version 3.6z for analysis. The CVs of the measurements were 3.5% (femoral 1

MARTINI/C:04 LORD_c

20/11/08

10:55

Page 2


VALIDATION OF A FOOD FREQUENCY QUESTIONNAIRE FOR WOMEN WITH OSTEOPOROSIS neck), 2.0% (spine), and 0.6% (total body). The WHO (1994) criteria for osteoporosis were adopted (7). The relative validation of the present study was accomplished through comparison of the reference method data, comprised of 3-Day Food Records (3DR) with the data obtained by the FFQ. The patients were instructed to fill in a 3 non-consecutive days Food Record, with one of the days being on the weekend. The 3DR was returned to a trained nutritionist on the week following that when the record was filled and was checked for further clarifications about the food consumptions and quantities. After 45 days a quantitative FFQ was applied. The FFQ was previously designed taking into consideration the foods/preparations most consumed by Brazilian elderly women. The FFQ is composed of 60 foods/preparations, separated into 10 groups, and measures the consumption of food and nutrients over the last 6 months. Food frequency consumption of each item was evaluated using 7 categories of answer: never, less than once a month, 1 to 3 times a month, once a week, 2 to 4 times a week, once a day, and 2 or more times a day. The questionnaire also included the portion size for each food item: small, medium or large. All questionnaires were applied by the same, previously trained interviewer. The 3DR was analyzed by the software Nutrition Data System for Research (University of Minnesota, 2005) and the FFQ was analyzed by the NutWin (UNIFESP, 2002). The food intake evaluation consisted of the assessment of consumption of calories and the following nutrients: carbohydrates, proteins, lipids, calcium, phosphorus and magnesium. Since the predominant source of vitamin D bioavailability is sun exposure, this nutrient was not measured by the FFQ.

presented osteoporosis at least in one site. The mean T-Score at lumbar spine was –2.78 (0.79), and at femoral neck – 2.03 (0.88). The BMI varied from 15.18 to 36.03 Kg/m² with an average of 25.73 ± 4.72 Kg/m², indicating that most patients were overweight. Only 3 patients were observed to be underweight. Table 1 General characteristics of participants (n=30) Characteristics Age (years) Weight (kg) Height (cm) BMI (Kg/m²)

Mean ± SD

Range

62.79 ± 8.04 59.59 ± 10.53 152.42 ± 7.21 25.73 ± 4.72

51 - 85 36 - 80 132 - 166 15.18 - 36.03

BMI: Body Mass Index

Table 2 presents the mean nutrient intakes by the FFQ and the 3DR. No significant differences were observed between nutrient intakes by the two methods, with the exception of the intakes of carbohydrate and magnesium, which were significantly higher for the FFQ. There was a positive correlation between the two methods when measuring energy, macronutrients, calcium, phosphorus and magnesium. Table 2 Comparison of nutrient intake from the 3DFR and the FFQ Food Intake

Energy (Kcal/d) Protein (g/d) Lipids (g/d) Carbohydrate (g/d) Calcium (mg/d) Phosphorus (mg/d) Magnesium (mg/d)

Statistical Analysis The mean nutrient intakes by FFQ were compared with the results of the 3DR using parametric tests since the variables were normally distributed (Kolmogorov-Smirnov test). The mean intake data of energy and nutrients obtained from both methods were compared using the Student’s t-test for paired samples. Correlations between the two methods were assessed by Pearson’s correlation coefficients. The agreement of the two methods in classifying a subject according to the intake of energy, macro and micronutrients was examined by determining the frequency of similar classification into quartiles. The overall percentage of individuals classified into same or opposite quartiles was calculated. The Bland and Altman statistical method was used to evaluate the agreement between the two dietary intake assessment methods. Data were analyzed using the software SPSS 11.0 for Windows (SPSS Inc, Chicago IL, USA).

3DR

FFQ

Mean ± SD

Mean ± SD

1443.67 ± 530.19 62.39 ± 21,66 49.63 ± 22.01 195.01 ± 72.99 648.50 ± 323.43 853.79 ± 316.38 213.51 ± 100.59

1570.97 ± 571.24 64.42 ± 27.60 45.64 ± 15.39 237.66 ± 98.06* 722.66 ± 269.79 889.46 ± 380.01 239.05 ± 80.17*

Pearson correlation coefficient r p value 0.765 0.869 0.498 0.788 0.508 0.520 0.550

0.000 0.001 0.005 0.000 0.004 0.003 0.002

* Significantly different from 3DR at p< 0,01

The Pearson´s correlation coefficients for nutrients involved in bone metabolism (Figure 1) ranged from 0.508 for calcium to 0.869 for protein. The agreement between 3DR and FFQ in classifying individual intakes was evaluated by quartiles of nutrients in both methods. This cross- classification of nutrient intakes based on quartiles by 3DR is presented in Table 3. The overall proportion of subjects classified in the same quartile by the two methods was on average 50.01% and in the opposite quartile 0.47%. The proportion of individuals in the lowest quartile of 3DR and in the FFQ was on average 67.3%. For calcium intake this proportion was 71.4%. Additionally, the proportion of individuals in the highest quartile for two methods was on average 53.04%. However, 14.3% of individuals classified in

Results General characteristics of the study population are presented in Table 1. All women were in postmenopausal state, and 2

MARTINI/C:04 LORD_c

20/11/08

10:55

Page 3


THE JOURNAL OF NUTRITION, HEALTH & AGING© the highest quartile for calcium intake in the 3DR were classified in the lowest quartile when evaluated by FFQ.

Figure 2 Differences between energy (A), calcium (B), phosphorus (C) and magnesium (D) intake from the 3DFR and FFQ plotted against the mean from the 2 methods

Figure 1 Plot of energy (A), calcium (B), phosphorus (C) and magnesium (D) intake from FFQ vs 3DFR with regression line

Energy Intake (kcal/day) derived by

Calcium Intake (mg/day) derived by

Magnesium Intake (mg/day) derived by

Phosphorus Intake (mg/day) derived by

Table 3 Cross-classification of nutrient distribution quartiles from 3DFR and FFQ

Nutrients

Energy (Kcal/d) Protein (g/d) Lipids (g/d) Carbohydrate (g/d) Calcium (mg/d) Phosphorus (mg/d) Magnesium (mg/d)

Discussion

Lowest quartile Highest quartile Overall 3DR 3DR Lowest Highest Highest Lowest Exact Opposite quartile quartile quartile quartile agreement (%) FFQ(%) FFQ(%) FFQ(%) FFQ(%) (%) 85.7 57.1 71.4 57.1 71.4 71.4 57.1

0 0 0 0 0 0 0

71.4 57.1 57.1 57.1 42.9 57.1 28.6

0 0 0 0 14.3 0 0

66.7 50.0 56.7 46.7 43.3 46.7 40.0

Osteoporosis has been recognized as the most frequent chronic condition occurring in postmenopausal women and the aged, is a major global healthcare problem in developed countries and is on the rise in developing countries. Optimization of bone health by appropriate dietary and lifestyle practices is a key factor in management of the disorder. Considering this information, and because the consequences of osteoporosis are multidimensional, a cost-effective method to identify those with inadequate nutrient intake should be developed. The present study developed and validated a food frequency questionnaire in order to evaluate nutrient intakes related to osteoporosis in elderly women. Several investigators have designed and validated FFQs for assessing calcium intake in adults, the elderly and even the general population (8, 9, 10, 11). However, nutrients such as phosphorus, magnesium, protein, as well as total energy intake play an important role in osteoporosis. In the present study, the FFQ was validated for these nutrients. Although results are mainly relevant to the Brazilian population, since dietary habits are particular, they can be of some use for studies in similar populations. Different dietary assessment methods often over- or underestimate mean intakes due to several factors, such as weaknesses in data-gathering techniques, use of different nutrient composition tables and databases, errors related to each single method, and, more important, the natural tendency of an

0 0 0 0 3.3 0 0

The Bland-Altman analysis (Figure 2) demonstrated the 95% limits of agreement between 3DR and FFQ for energy (2A), calcium (2B), phosphorus (2C), and magnesium (2D). The graphs provide a clear idea of the magnitude of disagreement between methods. For calcium and phosphorus intakes, the disagreement between the methods increases as the mean intake increases. The same pattern was observed for energy. There were two measurements outside the range of mean difference ± 2SD for calcium, two for energy, two for magnesium, and three for phosphorus.

3

MARTINI/C:04 LORD_c

20/11/08

10:55

Page 4


VALIDATION OF A FOOD FREQUENCY QUESTIONNAIRE FOR WOMEN WITH OSTEOPOROSIS individual’s nutrient intake to vary considerably from day to day. The food frequency questionnaire, when validated for a specific population group, has several advantages: It can provide high-quality data on large group of respondents, data may be more representative of usual intake than a few days of diet records, and the questionnaire is considered by several authors as the method of choice for research on diet-disease relationships (12). By using several statistical approaches to validate a FFQ for nutrient evaluation in women with osteoporosis, the present study demonstrated that the FFQ has a high consistency in estimating energy, macronutrients, calcium, phosphorus, and magnesium. The Pearson correlation coefficient, widely used in nutritional research, can be a good measure of linear association. When comparing the correlation coefficients for calcium intake from other studies similar results were observed (9, 11). Chee et al. (2002) in comparing calcium intake from a food frequency questionnaire with three-day food record found a correlation coefficient of 0,563 (9). In a validation of a food frequency questionnaire for calcium intake in the general population, Magkos et al. (2006), found a correlation coefficient of 0.639 (11). On the other hand, Montomoli et al. (2002), observed a correlation coefficient for calcium intake of 0.90 (10). Differences in validation methods, study populations and distribution of nutrients among population make the comparison difficult. Montomoli et al. (2002), in a study of Italian women, compared the food frequency questionnaire with a 14 day estimated diet record (10). Chee et al. (2002), when examining Malaysian women, used a three-day food record (9). It is well known that the number of days of dietary observation potentially contribute to the accurate estimate of nutrient intake (12). The number of days that data must be collected is related to intra- and inter-individual variation, gender and age of the group surveyed and the nutrient of interest. Aside from the nutrients of interest for osteoporosis, the number of days that should be collected for the food record varies from 7 for calcium, 5 for phosphorus, and 3 for total energy (3). The fact that we studied a homogeneous group in regards to gender, age and socio-economic level probably contributed to findings of similar correlation coefficients observed in other validation studies (9,11,13) where the number of evaluation days was higher. The categorization of subject in intakes quartiles for nutrients of interest for osteoporosis, demonstrated that, on average, only 3% of subjects were misclassified for calcium intakes, or in other words, classified in opposite extreme quartiles by the three day dietary records, and the food frequency questionnaire. For energy, macronutrients, calcium, phosphorus, and magnesium, the exact agreement between methods ranged from 40% to 66.7%, being considered satisfactory. Similar results were observed by Grootenhuis et al (1995)

for men and women aged 50-75 years regarding calcium intakes, when the overall proportion of subjects classified into opposite quartiles was 2.7% (13). For adolescents, Bertoldi (2005) observed a misclassification of calcium intakes in 2% of the studied population (14). The possible explanation for such results consists in the higher intake of dairy products during 3DR by some individuals, probably not observed in the FFQ that evaluates habitual food intakes. The presentation of data in cross-classification has an advantage of correlation coefficients, that is the differences in under and over-reporting, since it provides information concerning the capacity of both methods to allocate individuals according to the dietary intake distribution. In considering these issues, the present study demonstrated good individual classification, since, for the majority of nutrients, approximately 50% of subjects were in the same quartile of classification, ranging from 40-66.7%. Respective proportions in literature vary from 20-67% (4,6,11,14,15,16) for exact agreement. The agreement between methods evaluated by Bland-Altman analysis, demonstrated that for energy, calcium, phosphorus, and magnesium, the differences obtained by the two methods in each subject were well distributed around the mean of the difference, and they were, on average within 2 standard deviations. The fact that the disagreement increases as the mean intake increases (for calcium and phosphorus) do not invalidate the data, but indicate that mean intakes higher than that observed by general population should be evaluated with caution. Similar trends using Bland-Altman analysis was observed by Magkos (2006), in a development and validation of a FFQ for assessing dietary calcium intake in the general population (11). They observed that a quantitative accuracy for an individual spanned over approximately 1300 mg/day, and indicate that the FFQ cannot be considered appropriate for estimating actual calcium intake of an individual. These facts elucidate another important issue: the FFQ is an important dietary assessment method for a specific group of individuals or epidemiological studies. The individual intakes should be evaluated by more precise/accurate methods such as dietary records, weight food records or diet history. Regarding the sequence of administration of the reference and the tested method there is not a clear consensus. In theory, the FFQ and the reference method should assess diet over the same span, since the FFQ was validated to assess usual intake over the past 6 months, the reference method should be administered in intervene (11,16,17). A period of 45 days between the two methods was given, in order to avoid, or minimize a possible memory effect on the respondents’ behalf. A general limitation of FFQ validation studies is represented by the fact that there is no gold standard method for the measurement of dietary intake as a whole; therefore, an alternative method has to be chosen. In theory, the measurement errors of the FFQ and reference method should be unrelated (12,17,18). In this case, the best method would 4

MARTINI/C:04 LORD_c

20/11/08

10:55

Page 5


THE JOURNAL OF NUTRITION, HEALTH & AGING© probably be a weighted record, since they have the least correlated errors with FFQ. However, this method is expensive, time consuming, and generally suitable only for individuals or small groups of cooperative volunteers (14). Thus, the 3DR was preferred over other methods under the circumstances of being more cost-effective, a fairly accurate technique, and for not relying upon memory. It is known that a sufficient number of days of dietary information are needed to reasonably describe an individual’s diet (typically 14 to 28 days). However, few studies manage to achieve such a large number of good quality dietary information from their subjects and therefore most use between two and five days per subject (17). Additionally, research indicates that there is a significant increase in incomplete records as more days of records are kept because of respondent fatigue. The validity of the collected information decreases in the later days of a 7-day recording period in contrast to collected information in the earlier days (19). Another limitation that should be acknowledged to the present study is the sample size. According to Magnos et al. (2006) sample sizes of validation studies have ranged from approximately 20-60 up to a 100. It is important for future research to evaluate different populations, and also perform the reproducibility of the FFQ. In summary, the good agreement of mean intakes of nutrients, satisfactory correlation coefficients and similar distribution of subjects among quartiles, indicate that the presented FFQ for elderly women with osteoporosis is highly acceptable and an accurate method. Therefore, can be used in large-scale epidemiological studies or in clinical settings for evaluation of energy, macronutrients, calcium, phosphorus and magnesium intake in a similar population.

3. 4. 5. 6.

7.

8. 9.

10.

11.

12. 13.

14.

15.

16.

17. 18.

19.

Acknowledgments: Research suppoted by Fundação de Amparo a Pesquisa do Estado de São Paulo (FAPESP) under agreement no 03/06238-7 (LAM) and 04/14247-9 (GAPP)

References 1. 2.

Kulenovic I, Rasic S, Kulenovic E. Osteoporosis current trends in diagnosis and management. Bosn J Basic Med Sci. 2006; 6: 24-8. Gennari, C. Calcium and vitamin D nutrition and bone disease of the elderly. Public Health Nutr. 2001; 4: 547-59.

5

Lee RD, Nieman DC. Measuring diet. In: Lee RD & Nieman DC, editors. Nutritional Assessment. 2nd ed. St. Louis: Mosby, 1996. p.91-156 Slater B. et al. Validação de Questionários de Freqüência Alimentar – QFA. Rev. Bras. Epidemiol. 2003; 3: 200-08. Lopes, ACS. et al. Ingestão alimentar em estudos epidemiológicos. Rev. Bras. Epidemiol. 2003; 6: 209-19. Kroke, A et at. Validation of a self-adminiistered food-frequency questionnaire administered in the European Prospective Investigation into Cancer and Nutrition (EPIC) Study: comparison of energy, protein, and macronutrient intakes estimated with the doubly labeled water, urinary nitrogen, and repeated 24-h dietary recall methods. Am J Clin Nutr. 1999; 70: 439-47. World Health Organization. Study group on assessment of fracture risk and its application to screening for postmenopausal osteoporosis. Report. Geneva; 1994. (WHO Technical Report Series) Angbratt, M. Questionnaire about Calcium Intake: Can We Trust the Answers? Osteoporosis Int.1999; 9: 220-25. Chee, WSS et al. Dietary calcium intake in postmenopausal Malaysian women: comparison between the food frequency questionnaire and three-day food records. Asia Pacific J Clin Nutr. 2002; 11: 142-46. Montomoli, M et. al. Validation of a food frequency questionnaire for nutritional calcium intake assessment in Italian women. European Journal of Clinical Nutrition. 2002; 56: 21-30. Magkos, F et al. Development and validation of a food frequency questionnaire for assessing dietary calcium intake in the general population. Osteoporos Int. 2006; 17: 304-12. Willet, W. Food frequency methods. In Nutritional Epidemiology. Nutritional epidemiology. 2nd ed. Oxford: Oxford University Press, 1998; 74-100. Grootenhuis, PA et al. A semiquantitative food frequency questionnaire for use in epidemiologic research among the elderly: validation by comparison with dietary history. J Clin Epidemiol. 1995; 48: 859-68. Bertoli S et al. Validation of a Food Frequency Questionnaire for Assessing Dietary Macronutrients and Calcium Intake in Italian Children and Adolescents. J Pediatr Gastroenterol Nut. 2005; 40: 555-60. Cardoso MA. Kida AA, Tomita LY, Stocco PR. Reproducibility and validity of food frequency questionnaire among women of Japanese ancestry living in Brazil. Nutr Resear 2001; 21: 725-33. Osowski JM, Beare T, Specker B. Validation of a food frequency questionnaire for assessment of calcium ad bone-related nutrient intake in rural populations. J Am Diet Assoc 2007; 107: 1349-55. Cade J, Thompson R, Burley V, Warm D. Development, validation and utilization of food-frequency questionnaires – a review. Publ Heath Nutr 2002/ 5(4): 567-87. Decarli A, Franceschi S, Ferraroni M, Gnagnrella P, Parpinel MT, Vecchia C et al. Validation of a food-frequency questionnaire to assess dietary intakes in cancer studies in Italy. Ann Epdemiol 1996; 6: 110-18. Thompson FE, Byers T. Dietary assessment resource manual. J Nutr 1994; 124 suppl: 2245-317.