• Open Access

Combining food records with in-depth probing interviews improves quality of dietary intake reporting in a group of South Asian women

Authors


Correspondence to: Rozanne Kruger, Senior Lecturer in Human Nutrition, Institute of Food, Nutrition and Human Health, Massey University, Private Bag 102 904, North Shore Mail Centre, Auckland 0745, New Zealand; e-mail: R.Kruger@massey.ac.nz

Abstract

Objective: To investigate if the addition of an in-depth interview focused on cultural dietary practices could improve the quality of dietary data from food records among South Asian women in New Zealand.

Methods: Cross-sectional data were collected from 134 South Asian women (≥20 years), living in Auckland. Dietary data were collected using four-day food records. Nutritional analysis revealed 33.6% under-reporting of energy intakes. All women were recalled for an in-depth probing interview focused on culture-specific foods and dietary practices.

Results: The interview revealed extensive use of dairy products and plant oils. The nutrient content of the food record alone and the food record plus interview were compared; median energy intakes were 6,852 kJ vs 7,246 kJ (p<0.001); under-reporting decreased by 14.2%, and total fat and protein intakes (g/day) increased (p<0.001). Estimates of poly- and mono-unsaturated fatty acids increased significantly (p<0.001) due to greater use of plant oils due to greater use of plant oils replacing saturated fatty acid-rich fats in food preparation. A significant increase (17%) (p<0.001) in calcium intake reflects the higher dairy intake identified with the interview.

Conclusion: The addition of an in-depth probing interview to a four-day food record enhanced food intake reporting. Self-reported dietary assessments in immigrant population groups require quality control for accuracy.

Implications: Methods to ensure high-quality dietary data are essential to assess health outcomes and to inform public health interventions, especially in immigrant populations.

The health status of immigrant or minority groups in any country is often challenging, as these groups typically have higher rates of chronic diseases of lifestyle or non-communicable diseases such as cardiovascular disease, cancer or diabetes mellitus.1 India has the highest rank for diabetes cases,2 and Asian Indian immigrants around the world experience high mortality from cardiovascular disease and increasing rates of non-communicable diseases.1,3–6 This is apparent in New Zealand (NZ), where South Asian adults have the second highest prevalence rate (95%CI) of 6.5% (5.4–7.7) of type 2 diabetes.7 Hospitalisation and mortality in South Asians due to ischaemic heart disease are significantly higher than the total NZ population.8 Environmental factors, including dietary practices, may play an important role in the increased incidence of type 2 diabetes and cardiovascular disease in the South Asian immigrant population in NZ.8 Within the diet-disease relationship, an understanding of dietary intake, food choices and behaviours is essential to design preventive strategies. A potential problem in addressing these issues is that methods used in assessing dietary intakes or in designing nutritional prevention strategies may not always be culturally appropriate and may fail to recognise the importance of cultural factors in food choice.9,10 It is critical to add a socio-cultural perspective to the nutritional scientific approach of dietary assessment in order to gain an understanding of the nature and significance of dietary practices in immigrant communities.10

Little is known about the dietary intakes of the South Asian immigrant population in NZ and information on dietary practices of individuals of Asian Indian origin is limited.11 Measuring the dietary intakes of this group is a challenge due to their strongly embedded cultural practices and related food choices, varying food preparation methods, late additions and garnishes to dishes.4,12 Most dietary assessment methods have been criticised for their lack of accuracy and reliability,13 but transcultural dietary assessment presents even further challenges, including lack of knowledge of the researcher/interviewer regarding cultural practices, preferred food choices, cultural habits, and even subtle meanings related to food intake. Research strategies that involve observation or dialogue will reduce the differences between the researcher and subject, contributing to data quality. This was found to be effective in a multi-method strategy for transcultural nutrition research focused on feeding and weaning practices, knowledge and attitudes towards nutrition of Tswana mothers/caregivers of 0–3 year old children in a South African population.14

The SURYA study investigated the diet, lifestyle and health factors in women of South Asian origin (n=235) living in Auckland, NZ and is described in detail elsewhere.15 Although one of the best methods of dietary assessment (four-day food record)16 was used to assess usual intakes of individuals in the SURYA study, it was observed that a limited number of foods were recorded in the food records and that several of the known practices typical of this cultural group4,12 were not evident in the data collected from the food records. In addition, a large proportion of women (33.6%) under-reported their energy intake (using the Goldberg cut-off method17,18 as described in the methods section). The dietary data for the SURYA study were required to assess dairy and fat contents of the diets specifically as diet related to bone health and cardiovascular disease risk was investigated in that study. Previous research has shown that using in-depth food-based probing interviews enhanced dietary data assessment.9,19,20 The aim of this study was to investigate if and how the addition of a focused in-depth probing interview method on dietary habits could improve the quality of the calcium and fat reporting in a completed four-day food record in women of South Asian origin living in Auckland, NZ.

Methods

A cross-sectional study design was used to collect data from a sub-set of 134 South Asian immigrant women (the subject, both parents or grandparents must have been born on the Indian subcontinent), aged 20 years or older, living in Auckland (NZ) who participated in the SURYA study.15 Data on dietary intakes were collected using four-day self-conducted food records, including at least one weekend or feast day.

Estimated rather than weighed food records were chosen to improve adherence and reduce respondent burden.16 Detailed instructions (oral and written) and measuring tools were provided to participants, all of whom spoke English. They were asked to give detailed descriptions of all foods eaten and to estimate amounts using natural measures (e.g. pieces, slices), household measuring aids (e.g. cups, spoons, bowls) or by weighing if possible. Experienced dietitians followed a standardised protocol including a manual on household measure conversion21 to convert estimated amounts to weights. Nutrient analysis was done using FoodWorks 2007 (Xyris software, Queensland, Australia). New foods and traditional recipes were added with food composition data taken from an Indian food composition database where relevant.22 However, despite these efforts, it was noted that the food records of these women living in Auckland did not conform to some of the identified culturally-based food practices of South Asians.4,12 In an attempt to improve the quality of the dietary data, all the women (n=134) were recalled for an in-depth probing interview. Telephone interviews were conducted with the women (n=44) who were unable to attend in person. This in-depth probing interview was conducted by an experienced dietitian, using visual aids including food models and photographs or prompting with household utensils and appropriate food packaging materials to estimate portion sizes and to obtain correct food item information. Each interview lasted approximately 30 minutes. A set of probing questions, rather than a specific interview script, was used to elicit the information required. This method was designed to improve the quality of the completed four-day food records rather than obtaining a new record of dietary intake. Furthermore, it was designed to include elements of dietary behaviour that quantitative methods might be insensitive to.19 These were previously identified in a pilot study that focused on finding food items and ingredients frequently consumed and used in meal preparation by Asian cultures in NZ. Two of the authors of this paper conducted an exploratory investigation (including probing interviews with South Asian women, food product investigation and literature searches) as part of a contract with the NZ Ministry of Health. This provided information to design appropriate culturally-oriented 24h-recall probe questions for Asian cultures to be used for the 2009 NZ National Nutrition Survey. Subsequently this deeper understanding of the South Asian eating culture was used to design the probes that focused on specific food types, cultural practices including additions to foods and beverages, and recipe composition to clarify fat and dairy intakes (refer to Table 1 for examples of probes). All the participants were probed similarly on the above aspects. The additional data obtained with the in-depth probing interview were used to expand the four-day food record data collected from each subject. Changes were only made if an item was forgotten or simply not mentioned due to ignorance of the importance of a practice (e.g. standard addition of ghee (clarified butter) as a garnish on a dish) or if incomplete information was given (e.g. quantity of milk in tea or tea preparation). If the food record appeared to reflect accurate information, no changes were made.

Table 1.  Examples of interview probes and sample responses.
Probes (examples)aProbes (examples) used to encourage discussion and/or descriptionbSample responses (abbreviated)c
  1. a A probe / statement was put to the subjects during the interview; they responded and

  2. b Additional probes were used as appropriate to obtain the most

  3. c Descriptive and complementary data to the food record information as possible.

Describe how you prepare tea.Tea bags or loose leaf tea?
Think of a measuring cup (250 mL) – how many will you drink?
How much will be milk and water?
What type of milk is used?
Is any sugar added? If so, was the spoon flat or heaped? Think of sugar bags in a restaurant – how many will you add to a cup of tea
Boiled loose leaf tea, boiled with milk
Two full 250 mL cups
125 mL water & 125 mL milk
Low fat or skimmed milk
2 heaped teaspoons or 2 bags
Describe the use of deep-fat frying as a cooking method.Do you prepare deep-fat fried dishes?
What type of oil do you use when deep frying?
How often do you eat deep-fat fried foods?
What type of dishes do you usually deep fry?
Yes
Sunflower or Canola oil
Once a week, mostly on weekends
Traditional breads: puri, chapatti, poppadoms; fried foods: samosas, fish, pakora; desserts: Gulab Jamun,
Describe your use of ghee (clarified butter) or oil as a garnish.When do you use ghee?
How much do you add to food? In a dish (tablespoons or cups); in a portion (teaspoons or tablespoons)
Ghee is used on chapatti daily or for sautéing vegetables 20 g (2 teaspo ons) per chapatti or roti or in dahl's or lentils
Describe your use of coconut products.Do you use coconut products in your cooking?
What type of products do you use?
In which dishes do you use it?
How much would you add? (per dish or per person)
Sometimes substitute coconut milk/cream for butter
Freshly grated coconut or desiccated coconut
Use coconut milk/cream for chicken/shrimp curry
Desiccated coconut – 3 teaspoons per curry dish; Coconut milk/cream –¼-½ cup for chicken/shrimp curry

Under-reporting of energy intake (EI) as reported with the two dietary methods (records and records + interview) was assessed with the Goldberg cut-off method,17,18 using estimated energy requirements and a physical activity level (PAL) of 1.55 × basal metabolic rate (BMR). Schofield's standard age- and sex-specific equations23 based on measured body weight and height were used to calculate BMR. Subjects’ under-reporting was identified if EI:BMR ratio was <1.06 (cut-off value specific for a single subject and four days of mean dietary intake17).

Ethical approval was obtained from the Massey University Human Ethics Committee (Southern A), Reference No. 06/67. Subjects signed an informed consent form before participating in the study.

Statistical analyses

The nutrient composition of the diets was compared between the two dietary methods (record and record + interview) using the Wilcoxon Signed Ranks test, since most of the variables were not normally distributed. Variables were expressed as medians and 25th and 75th percentiles. Effect size was calculated for all differences to provide an objective measure of the importance of the effect by using the following formula: effect size r = Z/√n (where Z = z-score produced by Wilcoxon Signed Rank test and n = sample size). An effect size of 0.1 indicates a small effect, 0.3 a medium effect and ≥0.5 indicates a large effect.24 The data were analysed using the SPSS package version 16 (SPSS Inc., Chicago, IL, USA).

Results

The Surya participants represented approximately 0.5% of the total number of South Asian women living in Auckland, New Zealand.15 Their mean age was 42 ± 10.6 years and 76% of them had lived in New Zealand for ≤10 years. Their mean body mass index was 25.5 ± 4.39 kg/m2 and their mean waist circumference was 79.6 ± 9.7cm. The women were well-educated with 16 ± 3.8 years of education from age 5 years. This was expected from a newly migrant population.

The in-depth probing interviewed revealed the following dietary practices that were incompletely or partially reported in the four-day food records:

  • • Drinking very sweet, milky tea as opposed to westernised tea drinking habits (with a splash of milk and little or no sugar). The Indian practice is to brew tea with ½ water and ½ milk with several spoons of sugar. It is boiled on the stovetop and served from the communal pot, thus complicating the assessment of individual intakes.
  • • Skimmed milk often replaced full cream milk used with drinks and cereals.
  • • Using up to ½ cup of oil in the preparation of a dish.
  • • Plant oils like sunflower, canola, soya bean or olive oil were mostly used in cooking rather than ghee. Choice of oil was strongly driven by cost.
  • • Butter, ghee, coconut cream, and curd (yoghurt; homemade with full cream milk) were often used in curries and other dishes, sometimes as garnish and thus not considered part of the reported recipe.
  • • Traditional breads (puri, roti, naan, chapatti) were often home-made, using additional oil or ghee for frying and garnishing.
  • • Homemade cottage cheese (paneer), often deep-fried, was eaten alone or added to vegetable-based sauces or dishes.
  • • ‘Deep cooking’ is a term commonly used, that refers to a lengthy cooking process which includes the addition of fat.
  • • Honey or raw/brown sugar was often used as these were considered to be healthier than white sugar.

Subsequent re-assessment of the four-day food records plus the probing interviews showed a significant 14.2% reduction from 33.6 to 19.4% (p<0.001) of under-reporting. The average EI:BMR ratio of the participants increased from 1.21 (SD 0.35) to 1.29 (SD 0.27). Only two respondents’ diet records did not require adjustment.

The median nutrient content of the diets using the four-day food record vs using the four-day food record plus probing interview are compared in Table 2.

Table 2.  Comparative daily macronutrient and selected micronutrient intakes (average over four days between two dietary assessment measures).
NutrientsRecordaRecord + Interviewbp-valueEffect size
 Median (25P–75P)Median (25P–75P) (r)
  1. a Food record; first data set (average intake over 4 days)

  2. b Food record + in-depth probing interview; second or combined data set

  3. Differences between groups were determined by Wilcoxon Signed Ranks test.

  4. Effects r ≤0.1, small effect; r=0.3, medium effect; r ≥0.5, large effect.

Energy (MJ)6.855.80–7.697.256.51–8.14<0.0010.64
Protein (g)57.1046.90–67.7062.0053.20–74.50<0.0010.60
  %E from protein14.8012.80–17.1014.7012.80–16.900.220.11
Fat (g)57.1045.80–68.7064.8054.40–77.20<0.0010.64
  %E from fat31.2028.00–35.7032.8030.00–37.10<0.0010.32
  PUFA (g)9.657.80–12.8012.109.96–15.80<0.0010.70
  PUFA (% total FA)20.9017.00–24.4022.4018.60–25.70<0.0010.37
  MUFA (g)18.9015.50–22.5024.1019.70–27.80<0.0010.76
  MUFA (% total FA)38.6035.00–41.4041.5038.00–44.40<0.0010.72
  SFA (g)20.1015.30–25.3020.3016.20–25.900.020.19
  SFA (% total FA)39.8035.60–45.8036.1031.80–41.600.0010.69
Cholesterol (mg)128.0062.50–217.00144.0071.60–230.00<0.0010.48
Carbohydrates (g)218.00186.00–250.00222.00195.00–258.00<0.0010.41
  %E from CHO52.7048.50–57.2051.3047.50–55.30<0.0010.31
Sugar (g)79.2059.40–100.0081.1064.10–102.00<0.0010.37
Fibre (g)19.7016.10–23.8021.3017.60–25.80<0.0010.62
Calcium (mg)658.00497.00–850.00768.00588.00–933.00<0.0010.55
Vitamin E (mg)7.385.78–9.3310.207.96–12.30<0.0010.82
Riboflavin (mg)1.270.92–1.631.451.07–1.73<0.0010.56
Phosphorus (mg)995.00838.0–1242.001130.0959.0–1,352.00<0.0010.64

The addition of the probing interview resulted in significant increases (p<0.001, large effect size r≥0.5) in median energy, fat, protein, fibre and some micronutrients (calcium, vitamin E, riboflavin, phosphorus) (see table 2). Reporting of carbohydrate intake also increased, but the increase was small with a medium effect size (r=0.41). The proportion of fatty acids changed with a significant increase in polyunsaturated (PUFA) and monounsaturated fatty acids (MUFA) and decrease in saturated fatty acids (SFA).

Discussion

Implementing the in-depth probing interview technique to improve the quality of the four-day food record in this group of women resulted in more complete reporting of their dietary intakes as well as their dietary habits and practices. This resulted in a significant decrease in under-reporting, increases in most nutrients and a change in the distribution of fatty acids in the diet to reflect greater intake of unsaturated fatty acids. Dietary assessment within different cross-cultural populations require modification of the methodology including exploration of the food choices, recipes and food preparation practices,25 and adaptation of the nutrient database used, as was done in our study. However, although these may strengthen the method, it is suggested that a personal interview should be used when cultural practices and food choices are common in a study population, to prevent the detail of food choice and preparation being lost in the coding process.25 This was apparent after the self-completed records were coded and analysed, and the in-depth probing interviews were implemented to overcome this limitation. Previous studies have also shown that using only the traditional dietary assessment methods may not always provide in-depth information, especially if focused on particular dietary behaviours,19,20 as was the case in this study, where the focus was on fat and dairy intakes specifically. Qualitative dietary assessment methods are particularly successful to add data regarding particular behaviours,19,20 thus providing a complementary data set which will enrich nutrient analysis and provide a clear picture of the dietary intakes, together with explanations of practices or dietary changes.10

It was clear that dairy products were used extensively, as seen in previous research studies in Indian groups.1,4,26 These dairy products contribute to a variety of hidden sources of fat and energy within the typical practices of the cultural group. Milk was mostly used in beverages, but also used to produce curd or paneer. The women preferred to use full cream milk to make curd of adequate thickness, although skim milk was reported for other use such as in cereal or with brewing tea – this practice was only identified with the in-depth probing interview. Yoghurt, curd or paneer were also often added to dishes or eaten on the side of a hot curry dish (often legume or vegetable-based).

Rich (fat- and energy-dense) products including cream and coconut products were commonly added to protein dishes e.g. fish, legume or vegetable curries or chicken korma (traditional chicken curry). Traditional home-prepared breads (high fat content) were often served with meals. Deep-fat fried products or dishes prepared with added oil were often consumed at weekend social events, family gatherings or festivities, which are reported to occur regularly in the Indian culture4,12 and may easily be overlooked. Since some of these additions were seen as a preparation practice, garnish or condiment, the use of excess fat, butter, margarine, oils or high fat dairy products were not always mentioned when the women self-reported their food intake.

One of the key differences from the traditional South Asian diets is the shift in choice of cooking fat found in this study. Traditional ghee was mostly replaced with canola or sunflower oil for deep-fat frying and sautéing food. A dietary habits study of Asian Indians in the United States of America found similar results where long-time immigrants reported a significant reduction in consumption of ghee.12 The quantity of fat used in food preparation was high, confirming previous findings in Asian Indian immigrant populations.1,11,26 The probing interview revealed that plant oil was purchased on a regular basis in large volumes (25L) indicating probable underestimation of use in food records.

Under-reporting decreased significantly with the addition of the data from the in-depth probing interviews. The under-reporting may have been partially due to forgetfulness, lack of compliance or misinterpretation of food record requirements, subject response to food recording e.g. guilt or even age, weight status (particularly overweight) or gender (particularly women), as was found previously among women,25,27–29 and again confirmed in this study. Most of these factors may have been addressed by the in-depth probing interview, resulting in the reduction of under-reporting. Although the energy intakes of 23 subjects were adjusted down, these changes were minimal, suggesting that under-reporting, not over-reporting is the important issue in this population.

The enhancement of the food record data with the additional data from the in-depth probing interview was reflected by the differences in nutrient analysis between the two methods especially with regard to the type and amount of fats and dairy products that were added to various dishes. The standard recipes that were used to analyse the initial food record did not take these into account. The carbohydrate, fat and protein intakes increased significantly when comparing the record data to the combined data. The proportion of fatty acids also changed with an increase in PUFA and MUFA and a decrease in SFA which reflects the shift from use of fats with high SFA content e.g. ghee, butter and cream, to MUFA-rich oils e.g. canola oil. The increase in vitamin E (27.6%) with the record + interview was aligned with increases in MUFA-rich oils (21.5%). Calcium, phosphorus and riboflavin intakes increased significantly with the record plus interview data, confirming the extensive use of dairy products which might have gone unnoticed without the in-depth probing interview. Dietary calcium intake increased from a median intake of 658 mg/day to 768 mg/day (17%), which although below the current Australia/New Zealand recommended daily intake (RDI),30 is comparable to the average intake of the adult female population in New Zealand,31 and an important difference (110 mg/day) when assessing health risk. The high contribution of dairy products to calcium intake reflects that of the general New Zealand population31 and is not dissimilar to that seen in upper socio-economic groups in India.32,33

A limitation in this study was that this method was not validated and did not cover the complete diet. The in-depth probing interview should therefore be further developed and expanded beyond probes on fat and dairy intakes to address total energy intake and to improve dietary reporting (e.g. probing recipes, portion sizes, fruit and vegetable intakes and energy-dense snacks29 in future studies. It should also be validated as a method to further improve the assessment of total dietary intake in study populations from different cultural groups or in immigrant populations.

Conclusion

The addition of an in-depth probing interview to the food record resulted in an enhancement of the quality of the dietary data collected. The data suggest that in this immigrant population self-reported dietary assessments, particularly food records, require quality control to ensure accuracy. Detailed data from the in-depth probing interviews revealed valuable information about the types of fat, increased dairy product use and clarification of other culture-based food practices in this community, whether they were vegetarian or not. This was important especially with regard to their traditional practices, but also the adaptations in practices that might occur with acculturation.

Health professionals such as dietitians and nutritionists should be aware of the risk of underestimating dietary problems due to the impact of subtle misreporting. This is especially important when research is conducted in immigrant populations where language and cultural barriers exist and interviewers are unfamiliar with cultural food practices.4,9

Implications

High quality dietary data should be used to assess health outcomes and to inform public health interventions, especially in immigrant population groups like South Asian Indians who have different practices from those prevalent in their host country (NZ). Employing more than one method to assess dietary intake and combining in-depth probing interviews with quantitative methods enhanced the nutrient assessment and provided valuable information regarding culturally-based practices. Such information could be used to design effective dietary interventions to change cooking practices or food choices, resulting in improved nutrition and enhanced disease prevention. Adding a qualitative method to the quantitative method would not add extra burden, as it is brief, easy to administer and analyse and also very easy for the participant to discuss their typical behaviours and practices with an interviewer. It was apparent that many South Asian women included westernised foods (NZ specific) in their daily diets, but similarly continued to eat traditional foods and adhered to cultural practices in an effort to retain their cultural heritage. This should be acknowledged by nutritionists and the cultural diversity needs to be explored to be able to provide practical dietary recommendations. Nutrition information about specific foods and suggestions for healthy alternatives within food categories (e.g. dairy foods) should be used to improve tailoring of health messages to the daily household practices (e.g. preparation methods) of South Asian immigrants – an important strategy in the management or control of the dietary aspects related to chronic diseases of lifestyle in these populations.

Acknowledgements

The authors thank the women of the Auckland South Asian community for their participation, and Midi Tsai, Laura King, Hymavathy Danthala and Christa Riekert for their assistance in collecting dietary data. Funding for the study was provided by the NZ Lotteries Board from their Lotteries Health fund.

Ancillary