Validity and Reproducibility of a Culture-Specific Electronic Food Frequency Questionnaire: A Trinidad and Tobago Diet Assessment Study

Abstract

Nutritional epidemiologists use culture-specific food frequency questionnaires (FFQs) to assess the dietary intake of groups based on country, region or ethnic groups. This study aimed to validate a culture-specific semi-quantitative electronic Food Frequency Questionnaire (e-FFQ) to estimate food group intake in the adult population of Trinidad and Tobago. A 139-item semi-quantitative e-FFQ containing local dishes and street food was administered twice to adults 18 years and older and compared against four 1-day food records (FRs) using digital photographs, which served as the reference method. The validity and reproducibility of the e-FFQ food group intake estimates were determined using paired t-tests, bivariate correlations, and cross-classifications. Reproducibility correlations between the reported food group intakes in the first and repeat administration of the e-FFQ ranged from moderate (r = .44, P ≤ .0001) for sweetened beverages to high (r = .91 P ≤ .0001) for alcohol. The cross-classification agreements ranged from 70% (street food) to 92% (alcohol). Energy-adjusted deattenuated validity correlations between the e-FFQ and FR ranged from (r = .08) for water to (r = .81) for food from animal sources, with a mean validity correlation of .36. An average of 68% of the e-FFQ estimates was correctly classified within the ±1 quintile of the exact agreement with the FRs. Agreements between the e-FFQ and FRs ranged from 55% for street foods to 95% for water, all significant at P ≤ .0001. This study shows that the culture-specific e-FFQ is a valid tool for assessing and ranking food category intake estimates of the adult population living in Trinidad and Tobago.

Keywords

validation validity electronic food frequency questionnaire food groups Trinidad & Tobago dietary assessment

What do we already know about this topic?

Nutritional epidemiologists have developed and validated food frequency questionnaires (FFQ) to assess the dietary intake of populations and specific ethnic groups within these populations.

How does your research contribute to the field?

The dietary intake of the population can be used to investigate the association between diet and disease, given that Trinidad and Tobago have the highest prevalence of morbidity and mortality for NCDs in the Caribbean, with chronic NCDs accounting for 80% of total deaths.

What are your research’s implications toward theory, practice, or policy?

Food group intake estimates from this FFQ may be used to develop nutrition policies and programs geared toward primary prevention and intervention strategies to reduce the risk and prevent the onset of several NCDs.

Introduction

The Caribbean is a region with unique dietary patterns specific to each country due to a history of colonialism, slavery and indentured servitude, which contributed to a complex food culture within the region. The twin-island Republic of Trinidad and Tobago, with a population of approximately 1.4 million, has a diverse multi-racial population and a unique dietary pattern with contributions from several racial groups.¹ The population is comprised of 40% Afro-Trinidadians (descendants of former slaves brought to the island), 43% Indo-Trinidadians (East Indians who worked as indentured laborers) 18% mixed race individuals, 1.2% Chinese and less than 1% White.¹ Each racial group has contributed significantly to the food fusion and dietary intake which is unique to Trinidad & Tobago. In addition, street vending has over the years, also contributed to the increased accessibility and availability of popular local cuisine.

The Republic of Trinidad and Tobago, one of the most developed Caribbean countries, is considered the industrial capital of the Caribbean. The adult literacy rate is 99% with 80.1% of individuals using the internet and over 900 000 using social media.² Coupled with industrialization and changes in lifestyle and dietary habits, there has been an increase in the prevalence of risk factors associated with chronic NCDs in the population.³ As a result, Trinidad and Tobago has the highest prevalence of morbidity and mortality for NCDs in the Caribbean, with chronic NCDs accounting for 80% of total deaths.⁴

Heart disease is the leading cause of death in Trinidad and Tobago and is also the leading cause of premature death in both men and women.⁵ Trinidad and Tobago is ranked among the countries with the most prevalent and fastest-growing cases of type 2 diabetes in adults ages 20 to 79 years.⁶ The prevalence of diabetes is 14.5%, and of these, 88% to 90% have Type 2 diabetes. Diabetes accounts for 50% of deaths that occur before age 65 years.⁶ A higher prevalence of chronic disease mortality and morbidity places an economic burden on the country due to the rising cost of medical care, higher rates of absenteeism, and lower rates of productivity at work.³

Dietary intake is one of the most modifiable risk factors for reducing and preventing non-communicable diseases (NCDs). In assessing dietary intake of populations, epidemiological studies have used food frequency questionnaires (FFQs) as a reliable, inexpensive tool that can be used to assess long-term or habitual dietary intake and determine the relationship between diet and disease.⁷ FFQs have been developed and validated in other Caribbean islands including Barbados and Jamaica, and may be used to examine the associations between dietary intake and non-communicable diseases (NCDs).^8,9

Nutritional epidemiologists have developed food frequency questionnaires (FFQ) to assess the dietary intake of populations and specific ethnic groups within these populations. FFQs are reliable and inexpensive tools that can be used to assess long-term or habitual dietary intakes. FFQs are also easier to administer in large-scale studies compared to 24-hr dietary recalls.⁷ FFQs can be used in research to estimate dietary intake and its association with disease in large cohorts or populations.

A culture-specific FFQ can be a cost-effective method for investigating the role of dietary intake on the high prevalence of NCDs, at the population level. It can also be used to assess dietary intake patterns that can lead to nutrient deficiency. Food group intake estimates from this FFQ may be used to develop nutrition policies, and also programs geared toward primary prevention and intervention strategies; thus, reducing the risk and onset of several NCDs.

Development of a culture-specific FFQ that includes composite, mixed, and street foods unique to Trinidad & Tobago is necessary to accurately assess the dietary intake of the population. Since Trinidad and Tobago has high adult literacy, and internet access rates, the study used a novel approach of collecting food records using the participants’ personal mobile devices, specifically, their smartphones. Thus we designed a culture-specific semi-quantitative electronic FFQ (e-FFQ) for the adult population of Trinidad and Tobago. The aim was to evaluate its performance in assessing food group intake estimates when compared with multiple 1-day food records (FR).

Material and Methods

Study Design

A cross-sectional study was conducted for a period of 3 months to evaluate the reproducibility and validity of a culture-specific, semi-quantitative e-FFQ, for assessing food group intake in the adult population of Trinidad and Tobago.

In order to be eligible to participate in the study, individuals had to be ≥18 years of age and have access to reliable internet and a smartphone. Individuals were excluded from the study if they were out of the country during the study period; less than 18 years of age; had no access to smartphones, or the Internet; had any medical condition that required a therapeutic diet; for example Crohn’s disease, celiac disease, or end-stage renal disease.

This study was approved by the Institutional Review Board (IRB). A convenience sample of adults 18 years and older were recruited through email, social media platforms, and professional associations. Interested participants were emailed a short screening questionnaire created using Google Forms to determine their eligibility for the study.

Written informed consent was obtained from all subjects before they provided any personal or dietary data. All participants who met the inclusion criteria had equal opportunity to participate in the study. The first administration of the e-FFQ (e-FFQ1) using Google Forms was emailed to participants for completion between March and April 2019 and then re-administered (e-FFQ2) 3 months later between June and July 2019. Between the first (e-FFQ1) and second (e-FFQ2) administrations of the e-FFQ, participants provided four 1-day food records (FRs) using their smartphones over a 12-week period (Figure 1).

Figure 1.

Timeline for administration of the e-FFQs and multiple 1-day food records.

Food Frequency Questionnaire Design

The e-FFQ was designed by modifying an existing 146-item FFQ previously developed by Ramdath et al.¹⁰ Modification was done by grouping together food items that contain similar ingredients: for instance, coconut bread, sweet bread and coconut drops were grouped together. The modified e-FFQ contained 129 food items from the original FFQ, and 14 commonly consumed street foods, resulting in a total of 139 food items for the e-FFQ. Since Trinidad & Tobago has a high literacy rate and social media usage, the semiquantitative FFQ was designed, developed, and administered as an electronic survey (e-survey) using Google Forms. Instructions and examples for completion of the e-FFQ were provided as part of the e-survey.

The e-survey consists of 2 parts: a 139-item food frequency questionnaire and a section on demographics and anthropometrics. The e-FFQ groups food items similar to the Caribbean Six food groups.¹¹ Five other categories were included to capture the consumption patterns of street foods, water, non-alcoholic beverages, alcoholic beverages, and sweets. These are commonly consumed food items that do not fall into the specific Caribbean Six Food Groups. Food items with similar nutrient compositions were grouped under the following categories: (1) street foods; (2) staples, with the subcategories (a) breads, cakes, and cereals; (b) rice and pasta; (c) ground provisions; (3) vegetables; (4) legumes; (5) fruits; (6) food from animals with the subcategories (a) eggs, dairy, and dairy substitutes; (b) meat and meat substitutes; (7) fats & oils (8) water; (9) non-alcoholic beverages; (10) alcoholic beverages; and (11) sweets as shown in Table 1. The street food category was listed first to prevent participants from reporting the same information while responding to the rest of the FFQ. In addition, instructions were included to guide participants to exclude foods (such as vegetables, staples, or fruits) already reported in the street food category.

Table 1.

Food Groupings Used in the Reproducibility Analysis Assessed With the FFQ.

Food categories	Foods items	n
Street foods	Aloo pies, bake and shark/fried fish, chow (mango/pineapple), burgers (beef/chicken/vegetarian), corn soup, doubles, dhalpourie roti (beef/chicken/goat/veggie), pholourie, saheena, souse, gyros, (beef/lamb/chicken), homemade ice cream (barbadine/coconut/soursop), pies (beef/chicken/fish/cheese) punches (barbadine/peanut/channa)	14
Staples	Whole wheat bread/rolls/ hops, white bread/ rolls/ hops, sada roti/roast bake, fried bake, coconut bake, biscuit/crix/ dixie, paratha roti, banana bread, currants roll, Christmas cake/ black cake/ fruit cake, sponge cake, coo coo, corn pie, dumplings, sweet bread/coconut bread/drops, hot cereals—oats porridge, cold cereals unsweetened/high fiber - corn flakes, bran flakes; cold cereals sweetened- frosted flakes, Pasta-macaroni/ spaghetti/ noodles, macaroni pie, macaroni salad, chow mein, parboiled rice-plain, fried/vegetable rice, rice & peas/pelau, ochro/bhagi rice, English potato—baked/boiled/creamed/mashed/ potato salad, English potato - fry aloo, French fries, curried, dasheen, eddoes, yam- boiled, sweet potato- baked, boiled, cassava, green figs/banana boiled, green fig/banana salad, ripe plantain boiled	35
Vegetables	Beet root, bhagi/spinach/ dasheen bush, bodi beans/ string beans/saim, cabbage/cole slaw, callaloo, carali- fried, carrots, corn (roasted/boiled/ kernels), melongene/bhagan/eggplant, cucumbers, ochroes, patchoi, pumpkin, sweet peppers, tomatoes, tossed salad, watercress/lettuce	17
Fruits	Apple, banana, grapes, grapefruit, mango, oranges/portugals, paw paw, pineapple, pommecythere, pommerac, pears, prunes, raisins, sapodilla, watermelon	16
Legumes	Baked beans, beans, dried peas, all kinds -stewed, pigeon peas fresh, stewed/curried, channa, curried/fried, split peas/dhal, peanuts (roasted/honey roasted), cashew nuts, other nuts	8
Food from animals	Egg whole fried/boiled/scrambled, milk—low fat/skimmed (liquid), milk—full cream (liquid), milk—low fat/skimmed (powdered), milk—full cream (powdered), low fat evaporated milk, evaporated milk, condensed milk, soy/almond milk, yogurt (all types), vegetarian cheese-cheddar, regular cheese cheddar, Fish steam/stewed/ curried, fish, fried, saltfish/buljol- stewed, smoked herring, tuna/sardines in water, tuna/sardines in oil, chicken, curried/stewed, chicken, bbq/baked/grilled, chicken, fried, soup (beef, chicken, cow heel), beef, curried/stewed, beef patty, beef minced, lamb, goat, stewed/curried, pork, stewed/roasted, soya sausages/frankfurters/bologna/salami, chunks, stewed/curried, vege chick/vege fish.	29
Fats & oils	Butter (all kinds), margarine (all kinds), mayonnaise, coconut oil, soybean oil, canola oil, olive oil, avocado, salad dressing, garlic sauce	7
Water	Water	1
Non-alcoholic beverages	Diet soft drinks/soda, regular soft drinks/soda, fruit juices (no added sugar) orange/grapefruit/pineapple/etc	3
Alcoholic beverages	Beer (carib, shandy, etc.), red wine, white wine, rum/whiskey/scotch/vodka	4
Other items	Chocolate, dinner mints/cough drops, tamarind balls, sugar cake, benna ball	5

The frequency of intake section of the e-FFQ has 8 categories: never or less than once per month, 1 to 3 times per month, once per week, 2 to 4 times per week, 5 to 6 times per week, once per day, 2 to 3 times per day, and ≥4 times per day. Participants were asked to report on average how often they usually ate food relative to the given standard portion size specific to the food during the past 3 months. Familiar utensils, such as the “pot spoon” which is a common serving utensil in Trinidad and Tobago, are used for portion sizes to help participants better estimate their intake of various foods. For example, a rounded “pot spoon” is equivalent to approximately ½ cup. Since water is consumed daily, the frequency of water consumption was assessed using different categories: less than once per day, once per day, 2 to 3 times per day, 4 to 5 times per day, 6 to 7 times per day, and 8 or more times per day.

The demographic and anthropometric parts of the survey consist of questions on age, sex, height, weight, waist circumference, marital status, race, level of education, employment status, physical activity, smoking status, use of dietary supplements, and chronic disease history. Body mass index (BMI) was calculated based on self-reported weight and height using the following formula: weight (kg) divided by the square of height (m²).

Multiple One-Day Food Records With Digital Photographs

Food records provide detailed data on dietary intake with little or no reliance on memory, since food recording happens in actual time.¹² Therefore multiple 1-day FRs were used as a reference method to assess the validity of the e-FFQ because there were fewer correlated errors between the FRs and FFQs.¹ Instead of using a traditional approach (food diaries or telephone calls) to collect FR, FRs were collected electronically via email and/or WhatsApp text messages for ease of follow-up and to reduce the dropout rate from the study. This was similar to the method used in a study conducted among adolescents.¹³ The FRs collected electronically allowed for participants to submit pictures of their meals with a fiducial marker. This was a strategy used to avoid under or overestimation of portion sizes in the reporting of food intake.

Participants underwent one-on-one training via WhatsApp interviews using pictorial instructions and a checker fiducial marker (see Supplemental File 1: Fiducial Marker) that was sent to them before the training. The fiducial marker was in the form of a bookmark on card stock paper, and laminated. The front of the fiducial marker had the subject’s ID and name of the study, Trinidad & Tobago Dietary Assessment (TTADA Study). The fiducial marker served as a size reference and verification of participants’ identities.¹⁴

During the FR collection, participants were asked to record their food intake for the day using personal mobile phones to take digital photographs of meals before and after consumption with a fiducial marker placed by the food item or meal setting and to describe the foods and amounts consumed in a text message. Digital photographs and corresponding text descriptions of intake were sent via WhatsApp messaging or email to the study’s designated contact phone number or email address listed on the fiducial marker. A trained dietitian monitored the FR data collection and contacted participants whose reports were incomplete, unclear, had missing information, or to ask for clarification about food items reported to have been eaten. The digital photographs and text information were transferred and collated in PowerPoint files. The FRs collated in PowerPoint files consisted of (1) information on the time of intake, type of meal (breakfast, lunch, dinner, or snacks), and the food items eaten with their corresponding amounts and (2) digital photographs of meal settings before intake with the fiducial marker placed below the setting, as done in a previous validation study using food records with digital photographs.¹³

Data Analysis

A power analysis was conducted using G*Power software Version 3.1.9.2¹⁵ to determine an appropriate sample size for the study. A priori power analysis was calculated in order to determine the smallest amount of power needed in order to detect a difference. Pearson’s correlation coefficient was used to determine the statistical agreement between the FFQ and the multiple 1-day food records. According to Cohen’s effect sizes for Pearson’s r the power analysis was based on a medium effect size of r = .3, power = .80, and an alpha level of .05. Based on this power analysis, the study needed a sample size of 84 participants, and a 20% drop out rate was added for a total of 101 participants. Other validation studies had sample sizes that ranged from 100 to 200 participants.

Food group intake estimates

In a semiquantitative FFQ, the amount of food consumed per day is equivalent to the frequency of intake per day because the food portion sizes listed are fixed. Thus, the daily serving intake of individual foods in the FFQ was computed by converting the frequency of intake options to the frequency of intake per day. For example, 1 to 3 times per month = 2/30 = 0.067 and 1 per week = 1/7 = 0.14. Food group intake estimates were calculated by summing the values for all the food items belonging to a food group, as listed in Table 1. For the FR reference method, the four 1-day records were composed of 2 weekday and 2 weekend-day intakes. Thus, the servings of intake of the food groups were calculated by weighting the 2 weekday intakes as representing 5 days of a 7-day week and each weekend day intake as representing a day of a 7-day week. The formula below was used to calculate the weighted mean:

Weighted mean = [5 / 2 (WD1 + WD2) + (WE1 + WE2)] / 7

where WD=weekday intake; WE=weekend day intake

Reproducibility analysis

The e-FFQ was self-administered twice to 91 participants. Descriptive statistics, such as mean and standard deviation, and tests for normality were calculated for both administrations of the FFQ using SPSS version 26. As the food group intake distributions were near normal, the comparison of the estimates in the first (FFQ1) and repeated administration of the e-FFQ (FFQ2) was determined with a paired t-test. Pearson correlation coefficients were calculated to determine the agreement between the 2 administrations of e-FFQ (FFQ1 and FFQ 2). Cross-classification between FFQ1 and FFQ2 was assessed by dividing the food groups estimated intake into quintiles to determine the proportion of participants that were correctly classified under the same quintiles up to ±1 adjacent quintiles.

Validation analysis

Participants who were classified as extreme outliers in the consumption of food groups (ie, those with highly improbable intakes of less than 500 kcal and over 4000 kcal per day) were excluded from the validation analysis (n = 11). The mean food group intake estimates from the first administration of the e-FFQ were compared with the weighted mean intake of the food groups, as measured using the reference method (FR). The distribution for some food groups was near-normal; some food groups were log-transformed to achieve near-normal distributions, and a paired t-test was used to compare the means.

Pearson’s correlations were used to measure the strength of the association between e-FFQ1 estimates and FRs. The correlations were determined after energy adjustment using the residual method. Residuals were computed in SPSS using regression models with total energy as the independent variable and food group intake estimates as the dependent variable. To correct for within-person variation due to repeated measurements in the reference method, FR, 2 synthetic weeks, 1 weekday, and 1 weekend day were created. The food group intake estimates for synthetic weeks 1 and 2 were then adjusted for total energy intake and compared with the e-FFQ1.

Cross-classifications between the food group intakes in the 2 methods (e-FFQ1 and FRs) were used to determine the proportion of participants that were correctly categorized into the same quintiles (exact agreement and within ±1 adjacent quintiles). Bland-Altman plots were created for variables of interest to determine if both methods had good agreement, by plotting the difference in food group intake estimates between the e-FFQ and the FR on the y-axis, against the mean of the 2 methods on the x-axis. These food groups included staples, vegetables, fruits, street foods, and sweetened beverages.

Results

A total of 125 individuals completed a short screening form, which was used to determine their eligibility for the study. Nine individuals were ineligible because they would have been outside the country during the study period; an additional 3 individuals did not provide their name and contact information (preferred mail address, email address, and cell phone number); and one had Crohn’s disease. Of 112 eligible participants enrolled in the study, 14 did not complete the first e-FFQ. Although 98 participants completed the first e-FFQ, 7 did not complete either the second administration of the e-FFQ or the FRs. Therefore, 91 participants completed the study: 10 from Tobago and 81 from Trinidad.

Table 2 shows the baseline characteristics of the study participants. Male participants were approximately 2 years younger than their female counterparts. Both sex groups had mean BMIs bordering the obese category, with the majority of the participants classified as overweight or obese (~74%). Most participants (89%) had a university-level degree or higher, which implies that most of them were highly educated.

Table 2.

Demographic Characteristics of Participants at Baseline (n = 91).

Demographic Characteristics	Male	Female	All
n (%)	19 (21)	72 (79)	91
Age (years), mean (SD)	36.9 (9.5)	38.7 (9.6)	38.4 (9.6)
Height (cm), mean (SD)	177.5 (7.8)	165.3 (7.1)	167.9 (8.8)
Weight (kg), mean (SD)	91.8 (18.6)	84.8 (24.5)	86.3 (24.52)
Waist circ. (cm), mean (SD)	93.8 (10.6)	93.2 (16.2)	93.3 (15.2)
BMI (kg/m²), mean (SD)	29.5 (6.1)	30.5 (7.9)	30.5 (7.9)
BMI categories, %
Normal	26.3	25	25.3
Overweight	36.8	33.3	34.1
Obese	36.9	41.7	40.7
Race, %
Afro-Trinidadian	84.2	75	76.9
Indo-Trinidadian	5.3	9.7	8.8
Mixed race	10.5	15.3	14.3
Education level, %
Some university or above	84.3	90.3	89.0
Secondary school	15.7	9.7	11.0
Employment, %
Employed	89.5	84.7	85.7
Unemployed	10.5	15.3	14.3
Exercise frequency, %
≤1 per week	21.1	33.3	30.8
1-2 times per week	31.6	29.3	29.7
≥3 times per week	47.4	37.6	39.6
Chronic Disease (NCD) status (%)
One or more NCDs	21.1	26.4	25.3
None	78.9	73.6	74.7

Reproducibility of Food Frequency Questionnaire

The reported mean intake for food groups was higher in e-FFQ1 than in e-FFQ2, but not significantly different for fruits, fats, oils, street foods, sweets, sweetened beverages, and alcoholic beverages, as shown in Table 3. Correlations between the reported food group intake in the first and repeat administrations of the e-FFQ were performed to determine the reproducibility of the e-FFQ. The correlations ranged from medium (r = .44) for sweetened beverages to high (r = .91) for alcoholic beverages, with a mean correlation of .64.

Table 3.

Comparisons^a and Relationships^b Between Food Group Intake According to e- FFQ1^c and e-FFQ2^c Estimates.

Food groups	Food intake estimates, svg/d		p-value^a	r ^b	Agreement^d (%)Exact±1 quintile
	e-FFQ1^c	e-FFQ2^c
	Means SD	Means SD
Staples	4.1 (3.1)	3.6 (2.4)	0.018	0.77*	81
Legumes	1.5 (1.6)	1.1 (0.9)	0.008	0.55*	81
Vegetables	3.4 (3.0)	2.8 (2.5)	0.019	0.66*	85
Fruits	3.7 (5.9)	3.2 (4.3)	0.222	0.75*	87
Food from animals	3.7 (3.4)	3.2 (2.5)	0.037	0.74*	84
Fats and oils	1.8 (2.1)	1.55 (1.5)	0.150	0.51*	77
Street foods	0.7 (0.7)	0.7 (0.8)	0.659	0.50*	70
Sweets	0.8 (1.3)	0.7 (1.1)	0.333	0.66*	79
Water	6.5 (2.9)	5.1 (3.3)	<0.0001	0.58*	90
Sweetened beverages	0.6 (0.7)	0.5 (0.7)	0.091	0.44*	79
Alcoholic beverages	0.1 (0.5)	0.1 (0.4)	0.100	0.91*	92

Determined with paired t-tests.

Determined with Pearson’s correlations.

e-FFQ1 is the first administration, e-FFQ2 is the second administration of the food frequency questionnaire.

Overall percentage categorized in the same and adjacent quintile.

Correlation is significant at the .01 level (2-tailed).

Cross-classification into quintiles showed that 48% of participants were correctly classified into the same quintile. On average, 82% were classified into the same ±1 adjacent quintiles, whereas less than 2% were grossly misclassified. The classification within the same ±1 adjacent quintile ranged from 70% for street foods to 92% for alcohol, as shown in Table 3. Gross misclassification across categories between e-FFQ1 and e-FFQ2 ranged from 0% to 10%.

Validation of Food Frequency Questionnaire

The e-FFQ was validated against multiple 1-day FRs, which is the reference method used to assess the accuracy of the e-FFQ in estimating food group intake.¹ Table 4 shows a comparison of the mean intakes of the e-FFQ and FR. The mean food group intake estimates were higher for the e-FFQ for all food groups except sweetened beverages.

Table 4.

Comparisons^a and Relationships^d Between Food Group Intake According to e-FFQ1^b and Food Records.^c

Food groups	Food intake estimates (svg/d) Means ± SD^a		Correlations^d			Agreement (%)
Food groups	e-FFQ1 ^b	FRs ^c	Crude	En-adj	Deatt.	(Same ± 1 quintile)
Staples	4.1 ± 3.1	3.9 ± 1.7	0.22	0.32*	0.25	63
Vegetables	3.4 ± 3.0	1.4 ± 1.4	0.47	0.43**	0.61	68
Legumes	1.5 ± 1.6	0.9 ± 1.0	0.32	0.31*	0.49	70
Fruits	3.7 ± 5.9	2.1 ± 2.1	0.46	0.42**	0.53	68
Food from animals	3.7 ± 3.4	3.3 ± 1.3	0.65	0.66**	0.81	64
Fats and oils	1.8 ± 2.1	0.2 ± 0.3	0.17	0.10	0.18	57
Street foods	0.7 ± 0.7	0.5 ± 0.7	0.27	0.25	0.34	55
Sweets	0.8 ± 1.3	0.2 ± 0.4	0.06	0.29	0.09	70
Water	6.5 ± 2.9	3.3 ± 3.0	0.08	0.24*	0.08	95
Sweetened beverages	0.6 ± 0.7	1.6 ± 1.4	0.23	0.35*	0.30	56
Alcoholic beverages	0.1 ± 0.5	0.1 ± 0.2	0.35	0.26	0.25	85

Determined with paired t-tests.

Determined with Pearson correlations.

e-FFQ1 is the first administration,

FR is the weighted average of the 4 one day food records.

Correlation is significant at <.05 level (2-tailed).

Correlation is significant at <.01 level (2-tailed).

The mean of the crude validity correlations between the e-FFQ and FR was r = .34. Crude correlations between the e-FFQ food group intake estimates and multiple FRs were low (r = .06) for sweets and moderately high (r = .65) for animal food from animals, as shown in Table 4. Energy adjustment showed very little improvement in the correlations for some food groups, such as staples, food from animals, sweetened beverages, and alcohol. Deattenuation of the energy-adjusted correlation coefficients to correct for within-person variations in intake due to multiple 1-day food records increased the correlations for all food groups, except for water, sweetened beverages, alcohol, and sweets. Deattenuated energy-adjusted correlations ranged from .09 for sweets to .81 for food from animals but were not significant.

Cross-classification showed that the mean intake of the correctly ranked food group was 32%. Cross-classification for the same ±1 adjacent quintile between the FFQ and FR ranged from 55% for street foods to 95% for water after adjusting for total energy intake, as shown in Table 4. However, an average of 5.3% were grossly misclassified.

Figure 2 shows the Bland-Altman plots for staples, vegetables, fruits, street foods, and sweetened beverages. Most points were within the 95% confidence interval (95% CI) limits, with few observations outside the 95% CI limits.

Figure 2.

Bland-Altman plots of the difference between the FFQ and reference method (FR) (y-axis) versus the means of the FFQ and FR for intake of staples, vegetables, fruits, street foods and sweetened beverages.

Discussion

The 139-item semi-quantitative e-FFQ that we developed for adults in Trinidad and Tobago can estimate food group intake with acceptable reliability and validity. To the best of our knowledge, this is the first culture-specific food frequency questionnaire that has been developed specifically for Trinidad and Tobago. Trinidad and Tobago have the highest prevalence of morbidity and mortality among NCDs in the Caribbean.⁶ Studies conducted on this population often rely on anthropometric data to assess the risk factors for morbidity and mortality in the adult population.^5,16
-18 Other studies have relied on FFQ-derived food and nutrient intake estimates that have not been validated or are not culturally specific for this unique population.^19
-21

In 2011 a study was conducted to determine the dietary intake among adults in Trinidad and Tobago and to develop a culture-specific quantitative FFQ.¹⁰ However, the FFQ has not been validated, and to date, there is no known validated FFQ specific to Trinidad and Tobago. Therefore, the validation of this culture-specific semi-quantitative FFQ will fill this research gap and can be used to establish the dietary intake patterns of Trinidad and Tobago adult populations. The e-FFQ can also be used in future studies to identify associations between diet and disease, as well as to emphasize the role of diet in lifestyle interventions and the prevention and reduction of NCDs.

We evaluated the reproducibility and performance of this culture-specific e-FFQ’s ability to estimate food group intake from multiple, 1-day food records with digital photographs using personal mobile smartphones. The e-FFQs were self-administered 3 months apart to determine FFQ reproducibility. A comparison of means showed that staples, vegetables, legumes, food from animals, and water were significantly different between the 2 e-FFQ administrations. The e-FFQ showed reasonably high reproducibility with a mean correlation of r = .71. This correlation was similar to other studies with correlations between .5 and .7 or higher than reported by other studies.^22
-25 It should be noted that these studies had varying time periods for the test-retest of the FFQ, ranging from 4 months to 1 year. There was also good agreement (82%) between the 2 FFQs when participants were ranked into quintiles with the same ±1 adjacent quintiles.

To validate the FFQ, the first administration of the e-FFQ was compared to the FRs, as researchers have found that when the diet recall or food record is collected first, it has a training effect on the subjects, which increases their recall ability, which in turn can artificially increase the validity.⁷ This can be overcome by administering the FFQ first or administering the food record simultaneously with the FFQ.^1,26 Researchers have also found that the dietary habits of participants tend to change during the study period, which can decrease validity.⁷ Four1-day FRs were used as the reference method, as they are considered the gold standard method in validation studies because FRs have fewer correlated errors with FFQs.²⁷ To prevent participants from deviating from their habitual food intake, we instructed them not to change their eating patterns when recording their food intake, as the purpose of this study was to help establish dietary habits specific to Trinidad and Tobago adults.

The results of this study indicated that the FFQ is a reasonably good tool for assessing dietary intake in this population. In this study, foods were grouped based on the 6 food groups. A study conducted in Jamaica with similar food groups also had similar correlations for the respective food groups, as their study had a correlation of r = .45 for dark green leafy vegetables and other vegetables²⁴ and this study had a correlation of r = .46 for vegetables. The correlations for peas and beans were similar (r = .39 and 0.31) respectively. Fruit and staples/cereals showed similar correlations. Correlations from other culture-specific validation studies for food group intake had similar or higher correlations between r = .2 and r = .6 in these studies.^23,25,28,29 However, it should be noted that foods were grouped differently in other validation studies, and as a result, correlations would vary based on how the food items were grouped.

Study Limitations and Strengths

One of the main strengths of this study is the inclusion of local food items and street foods unique to Trinidad and Tobago. A trained registered dietitian collected food records from the participants, provided training as needed, and answered questions about the study. This helped ensure that protocols and standards were uniformly followed in collecting food records and estimating the portion sizes and different food items in each food record. The response rate for Tobago was 11% which is comparable with the population of Tobago as it makes up only 4% of the population of Trinidad and Tobago.³⁰

Another strength of the study was the electronic method for the distribution of the FFQ, which was used because of its high literacy level and wide access to the internet and electronic devices among this population.² Other studies have used similar approaches for validation studies with great success, as they reduce costs and improve the response time. It also prevents participants from submitting incomplete data because the e-FFQ allows researchers to control the quality of responses by requiring respondents to complete all items before being allowed to press the submit button.^13,31 Another strength of the study was the use of fiducial markers as a size reference to verify participant identity and authenticate photographs of the foods and beverages taken by the subject.¹⁴

As previously mentioned, Trinidad and Tobago has a multi-racial and multicultural population, which includes several religious and cultural practices. These religious and cultural practices can influence food intake patterns of the population. During the study period and the collection of food records, some participants indicated that they were observing the Lenten season and abstained from meat and/or alcohol, or limited the number of meals consumed for the day. The holy month of Ramadan also occurred during the study period, and participants who were Muslims abstained from food or drinks from sunrise to sunset for the month; thus, the FR for 3 persons was collected after the month of Ramadan. Other religious and cultural seasons or events, such as the Carnival, Divali, and Christmas, also influence the dietary habits of the population. Seasonal variations may also be caught by asking participants to report their intake during the past year and then spreading out the administration of FRs over a longer period which we were unable to do because of time constraints.

The seasonality of foods, particularly fruits, can alter food intake in this population. Traditionally during mango season in Trinidad and Tobago, persons tend to “turn down their pots” and instead replace 1 to 2 of their meals with several mangoes. Mangoes are also prepared in a variety of ways, such as preservation with sugar, spices, or salt and pepper (chow), and are used as snacks between meals. Thus, there would be an increase in fruit consumption and a decrease in the consumption of other foods, as they are replaced by mangoes. This practice was also observed during data collection from the participants’ food records.

Although 91 participants completed both the FFQ and FR, 11 were excluded because they were extreme outliers. These outliers represent an obvious improbable intake that would bias the FFQ estimates relative to the reference method. A priori power analysis showed that 84 participants were needed for the study to detect a difference. Thus, the exclusion of participants for the validation analysis, which would lead to stronger correlations, resulted in a smaller sample size of eighty (80) participants, reducing the power to establish statistical significance.

Food records were used as the reference method as it is considered a gold standard method in validation studies because FRs have less correlated errors with FFQs. However, researchers have found that when the food record is collected first, it has a training effect on the subjects and increases the recall ability, which in turn can artificially increase validity. This can be overcome by administering the FFQ first or administering the food record at the same time the FFQ is administered.²⁶ Researchers have also found that the dietary habits of the subjects tend to change during the study period, which can decrease validity.^1,26 Thus participants were advised not to change their dietary habits during the study.

Both FRs and FFQs may be prone to errors in estimating of portion sizes and the use of the same database, Nutrition Data System for Research (NDS-R), to determine the nutrient profile of foods in the FFQ and the FRs. One way to address this problem is to include a third method such as the use of biomarkers that does not have correlated errors with the other 2 methods used. However, this will be a costly venture and would require additional equipment and personnel for this type of study.

Another limitation of the study was the use of a convenience sample which resulted in a low response rate of Indo-Trinidadians (9%) when compared to the 43% making up the population of Trinidad and Tobago.

Conclusion

This study shows that the culture-specific e-FFQ is highly reproducible for assessing food group intake estimates. The e-FFQ is also a valid tool for assessing and ranking food group intake estimates in adults living in Trinidad and Tobago. Therefore, it could be used to examine the association between dietary intake and NCDs. Food group intake estimates from this e-FFQ may also be used to develop nutrition policies and programs geared toward primary prevention and intervention strategies to reduce the risk and prevent the onset of several NCDs.

Supplemental Material

sj-docx-1-inq-10.1177_00469580241273247 – Supplemental material for Validity and Reproducibility of a Culture-Specific Electronic Food Frequency Questionnaire: A Trinidad and Tobago Diet Assessment Study

Supplemental material, sj-docx-1-inq-10.1177_00469580241273247 for Validity and Reproducibility of a Culture-Specific Electronic Food Frequency Questionnaire: A Trinidad and Tobago Diet Assessment Study by Lesley Ann Foster-Nicholas, David Shavlik, Celine Heskey, Patricia Dyett and Gina Segovia-Siapco in INQUIRY: The Journal of Health Care Organization, Provision, and Financing

Footnotes

Acknowledgements

We would also like to disclose that this manuscript submission was posted as a preprint on Research Square. DOI:

Author Contribution Statement

LAFN conceptualized and designed the study; collected, analyzed, and interpreted the data; drafted the manuscript; and gave final approval of the version to be published. GS designed the study, analyzed and interpreted the data, contributed to and critically revised the manuscript, and approved the final version to be published. DS analyzed and interpreted the data, contributed to and critically revised the manuscript, and approved the final version to be published. CH contributed to and critically revised the manuscript, and approved its final version for publication. PD contributed to and critically revised the manuscript and approved the final version.

Availability of Data and Materials

Our Food Frequency Questionnaire, Databases, and analysis codes will be available with the publication by request to the corresponding author. Databases and analysis code will be available with the publication by request to the corresponding author at lfosternicholas@llu.edu

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Ethical Standards Disclosure

This study was conducted according to the guidelines laid down in the Declaration of Helsinki and all procedures involving research study participants were approved by the Institutional Review Board of Loma Linda University IRB # 5180409. Written informed consent was obtained from all subjects. before they provided any personal or dietary data. All participants who met the inclusion criteria had equal opportunity to participate in the study.

ORCID iDs

Lesley Ann Foster-Nicholas

David Shavlik

Celine Heskey

Patricia Dyett

Gina Segovia-Siapco

Supplemental Material

Supplemental material for this article is available online.

References

Nations Encyclopedia Trinidad and Tobago - Ethnic groups. 2018; Accessed October 23, 2018. https://www.nationsencyclopedia.com/Americas/Trinidad-and-Tobago-ETHNIC-GROUPS.html

Telecommunications Authority of Trinidad and Tobago. The Digital Divide Survey Trinidad and Tobago 2013. Accessed October 23, 2018. https://tatt.org.tt/wp-content/uploads/2023/02/Digital-Divide-Survey2013_FINAL-17-01-2014-reduced.pdf

Government of the Republic of Trinidad and Tobago Min. of Health. National Strategic Plan for the Prevention and Control of Non Communicable Diseases: Trinidad and Tobago 2017-2021. Accessed February 23, 2022. https://health.gov.tt/national-strategic-plan-for-the-prevention-and-control-of-non-communicable-diseases-trinidad-and

Pan American Health Organization Trinidad and Tobago - Country Profile _ Health in the Americas. 2021. Accessed February 23, 2022. https://hia.paho.org/en/countries-22/trinidad-tobago-country-profile

World Health Organization Trinidad and Tobago PANAM STEPs CNCD risk factor survey- Final Report. 2012. Accessed February 23, 2022. https://extranet.who.int/ncdccs/Data/TTO_C7_PANAM%20STEPS%20REPORT_2012%20pdf.pdf

World Health Organization Graph illustration World Health Organization – Non-communicable Diseases (NCD) Country Profiles, Trinidad and Tobago. 2014. Accessed October 23, 2018. https://www.who.int/publications/m/item/noncommunicable-diseases-tto-country-profile-2018

Willett

Nutritional Epidemiology. Oxford University Press; 2012.

Sharma

Cao

Harris

, et al.; Barbados National Cancer Study Group. Dietary intake and development of a quantitative food-frequency questionnaire for the Barbados National Cancer Study. Public Health Nutr. 2007;10(5):464-470. doi:10.1017/S1368980007220531

Jackson

Walker

Cade

, et al. Reproducibility and validity of a quantitative food-frequency questionnaire among Jamaicans of African origin. Public Health Nutr. 2001;4(5):971-980. doi:10.1079/phn2001166

10.

Ramdath

Hilaire

Cheong

Sharma

Dietary intake among adults in Trinidad and Tobago and development of a quantitative food frequency questionnaire to highlight nutritional needs for lifestyle interventions. Int J Food Sci Nutr. 2011;62(6):636-641. doi:10.3109/09637486.2011.572545

11.

Caribbean Food and Nutrition Institute CFNI. Six Food Groups for use in the Caribbean. 1998. Accessed October 23, 2018. https://www3.paho.org/english/cfni/six_food_groups.pdf

12.

Satija

Willett

FB.

Understanding nutritional epidemiology and its role in policy. Adv Nutr. 2015;6(1):5-18. doi:10.3945/an.114.007492

13.

Segovia-Siapco

Sabaté

Using personal mobile phones to assess dietary intake in free-living adolescents: Comparison of Face-to-Face versus telephone training. JMIR Mhealth Uhealth. 2016;4(3):e91. doi:10.2196/mhealth.5418

14.

Daugherty

Schap

Ettienne-Gittens

, et al. Novel technologies for assessing dietary intake: evaluating the usability of a mobile telephone food record among adults and adolescents. J Med Internet Res. 2012;14(2):e58. doi:10.2196/jmir.1967

15.

Faul

Erdfelder

Lang

Buchner

G*Power 3: a flexible statistical power analysis program for the social, behavioral, and biomedical sciences. Behav Res Methods. 2007;39(2):175-191. doi:10.3758/bf03193146

16.

Prout

Nichols

Cardio-metabolic risk and lifestyle behaviours among police officers in Trinidad and Tobago. FASEB J. 2016;30(S1). doi:10.1096/fasebj.30.1_supplement.lb396

17.

Maharaj

Teelucksingh

Chow

De Freitas

Screening for non-communicable diseases at a walk-in clinic in Trinidad, West Indies: a time-motion cross-sectional study and net present value analysis. West Indian Med J. 2018;67(2):122-130. doi:10.7727/wimj.2015.277

18.

Donkor

Farrell

Ocho

, et al. Correlates of obesity indices and cardiovascular disease risk factors among Trinidadian nurses. Int J Afr Nurs Sci. 2020;12. doi:10.1016/j.ijans.2020.100194

19.

Rocke

Nichols

Prout

Dalrympe

The cost of a healthy diet in Trinidad and Tobago. FASEB J. 2017;31(S1). doi:10.1096/fasebj.31.1_supplement.962.19

20.

Prout

Nichols

Quality-of-life, risk for malnutrition, and polypharmacy among persons on haemodialysis in Trinidad and Tobago. FASEB J. 2017;31(S1). doi:10.1096/fasebj.31.1_supplement.791.8

21.

Nichols

Prout

Haemodialysis, risk of food insecurity, and dietary behaviours in Trinidad and Tobago. FASEB J. 2017;31(S1). doi:10.1096/fasebj.31.1_supplement.791.9

22.

Vioque

Navarrete-Muñoz

Gimenez-Monzó

, et al.; INMA-Valencia Cohort Study. Reproducibility and validity of a food frequency questionnaire among pregnant women in a Mediterranean area. Nutr J. 2013;12(1):26. doi:10.1186/1475-2891-12-26

23.

Aoun

Bou Daher

El Osta

Papazian

Khabbaz

LR.

Reproducibility and relative validity of a food frequency questionnaire to assess dietary intake of adults living in a Mediterranean country. PLoS One. 2019;14(6):e0218541. doi:10.1371/journal.pone.0218541

24.

Jackson

Walker

Younger

Bennett

FI.

Use of a food frequency questionnaire to assess diets of Jamaican adults: validation and correlation with biomarkers. Nutr J. 2011;10:28. doi:10.1186/1475-2891-10-28

25.

Ahmadnezhad

Asadi

Heidarian Miri

, et al. Validation of a Short Semi-Quantitative Food Frequency Questionnaire for Adults: a Pilot study. J Nutr Sci Diet. 2017;3(2):49-55.

26.

Thompson

Kirkpatrick

Subar

, et al. The National Cancer Institute’s dietary assessment primer: a resource for diet research. J Acad Nutr Diet. 2015;115(12):1986-1995. doi:10.1016/j.jand.2015.08.016. https://dietassessmentprimer.cancer.gov/learn/software.html

27.

National Cancer Institute. Dietary assessment primer, learn more about software for dietary analysis of 24 hour dietary recalls (24HR) and food records. n.d.

28.

Matthys

Pynaert

De Keyzer

De Henauw

Validity and reproducibility of an adolescent web-based food frequency questionnaire. J Am Diet Assoc. 2007;107(4):605-610. doi:10.1016/j.jada.2007.01.005

29.

Liu

Wang

Lin

, et al. Reproducibility and validity of a food frequency questionnaire for assessing dietary consumption via the dietary pattern method in a Chinese rural population. PLoS One. 2015;10(7):e0134627. doi:10.1371/journal.pone.0134627

30.

Government of the Republic of Trinidad & Tobago Min. Foreign and Caricin Affairs. Trinidad & Tobago 2010-2022. Accessed March 7, 2024. https://foreign.gov.tt/about/trinidad-tobago/

31.

Siapco

Oda

Sabate

. Comparison of a web-based Food Frequency Questionnaire with photograph-assisted food records in assessing nutrient intake of adolescents. FASEB J. 2016;30(S1):1153-1159. doi:10.1096/fasebj.30.1_supplement.1153.9

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

7.35 MB