The Fit for Delivery study: rationale for the recommendations and test‐retest reliability of a dietary score measuring adherence to 10 specific recommendations for prevention of excessive weight gain during pregnancy

Abstract

Aiming at preventing excessive weight gain during pregnancy, 10 specific dietary recommendations are given to pregnant women in the intervention arm of the Norwegian Fit for Delivery (FFD) study. This paper presents the rationale and test‐retest reliability of the food frequency questionnaire (FFQ) and a dietary score measuring adherence to the recommendations. The study is part of the ongoing FFD study, a randomised, controlled, intervention study in nulliparous pregnant women. A 43‐item FFQ was developed for the FFD study. A dietary score was constructed from 10 subscales corresponding to the 10 dietary recommendations. Adding the subscales yielded a score from 0 to 10 with increasing score indicating healthier dietary behaviour. The score was divided into tertiles, grouping participants into low, medium and high adherence to the dietary recommendations. Pregnant women attending ultrasound screening at about week 19 of pregnancy were asked to complete the FFQ twice, 2 weeks apart. Of 154 pregnant women completing the first questionnaire, 106 (69%) completed the form on both occasions and was included in the study. The test‐retest correlations of the score and subscales were r = 0.68 and r = 0.56–0.84, respectively (both P ≤ 0.001). There was 68% test‐retest correct classification of the score and 70–87% of the subscales. In conclusion, acceptable test‐retest reliability of the FFQ and the dietary score was found. The score will be used in the FFD study to measure adherence to the dietary recommendations throughout pregnancy and in the following year post‐partum.

Introduction

Overweight and obesity have become increasingly prevalent in Norway over the past two decades, also among women of childbearing age (Henriksen 2006). Especially pre‐pregnancy body mass index (BMI), but also weight gain during pregnancy, affects gestational as well as long‐term health of both mother and child (Thorsdottir et al. 2002; Andreasen et al. 2004; Nohr et al. 2008). Excess gestational weight gain is associated with increased risk of complications both before (Thorsdottir et al. 2002) and during delivery (Callaway et al. 2006), and further with increased risk of weight retention after delivery (Linné et al. 2003; Andreasen et al. 2004). Excess gestational weight gain also increases the risk of developing diseases later in life, such as diabetes and breast cancer (Kinnunen et al. 2007; Norman & Reynolds 2011). Excess weight gain in pregnancy is clearly associated with increased birthweight and incidence of large for gestational age babies (Stamnes Koepp et al. 2012). High birthweight is further associated with increased risk of obesity and diabetes later in life for the child (Henriksen 2007; Kinnunen et al. 2007).

The American Institute of Medicine has published guidelines for weight gain during pregnancy based on a woman's pre‐pregnancy BMI (Institute of Medicine 1990). Research suggests that weight gain at or below these recommendations is associated with an optimal delivery outcome for both mother and child (Kinnunen et al. 2007). Although studies have shown that weight gain in pregnant women is influenced by health care provider recommendations, it has been reported that 30–60% do not receive weight gain advice in pregnancy (Stotland et al. 2005; Tovar et al. 2010). The few studies that have investigated specific nutritional factors related to weight gain during pregnancy indicate that diet may play a major role (Olafsdottir et al. 2006; Uusitalo et al. 2009). A recently published meta‐analysis of the effects of randomised dietary interventions in pregnancy confirms that dietary interventions are effective in limiting gestational weight gain, and significant reductions in a range of gestational complications were documented as well (Thangaratinam et al. 2012). Excess weight gain during pregnancy has been found to be related to a ‘fast food’ pattern (Uusitalo et al. 2009), ‘eating more’, drinking more milk and eating more sweets (Olafsdottir et al. 2006). There is, however, a paucity of information on what constitutes effective dietary interventions for preventing excessive gestational weight gain (Thangaratinam et al. 2012).

Apart from influencing weight gain, dietary factors in pregnancy may also impact directly on the risk of pregnancy complications (Meltzer et al. 2011).

Changing dietary habits is difficult to initiate and maintain, and psychological research indicates that only 30–40% of lifestyle change intentions are carried out (Godin & Kok 1996; Allan et al. 2008). It is, however, known that first‐time pregnant women become more aware of the health aspects of nutrition and seek this kind of information during pregnancy (Szwajcer et al. 2005). A study by Crozier et al. (2009) showed that pregnant women were able to respond to simple dietary public health messages, such as to reduce the consumption of entrails but that the overall quality of diet was more difficult to improve. Consequently, concise and specific dietary recommendations should be developed and communicated to pregnant women in a simple and comprehensible way.

The Norwegian Fit for Delivery (FFD) study is a prospective randomised, controlled, intervention study carried out among 600 nulliparous pregnant women in the area surrounding Kristiansand in the southern part of Norway. Women in the intervention arm are given dietary advice and access to biweekly exercise groups throughout pregnancy whereas the control group receives routine pregnancy health care. The purpose of the FFD study is to investigate whether a relatively low impact lifestyle intervention during pregnancy can modify gestational weight gain and influence long‐term maternal and child nutrition and health. The protocol for the FFD study will be presented elsewhere. The aim of the present paper is to describe the 10 FFD dietary recommendations and their rationale and to report test‐retest reliability of the FFQ and the dietary score measuring adherence to the recommendations.

Dietary advice in the FFD study

Both national and international dietary guidelines exist for pregnant women (Fowles 2006; The Norwegian Directorate of Health 2010a). In addition, women in the intervention arm of the ongoing FFD study receive dietary advice simplified to 10 easy‐to‐remember statements specifically targeting prevention of excess weight gain during pregnancy. The following advice is given: (1) eat regular meals; (2) drink water when thirsty; (3) in between meals, choose fruits and vegetables; (4) eat vegetables with dinner every day; (5) eat sweets and snacks only when you really appreciate it; (6) buy small portion sizes of unhealthy foods; (7) limit your intake of added sugar; (8) limit your intake of salt; (9) do not eat beyond satiety; and (10) read nutrition labels on foods before buying. The recommendations are forwarded to study participants in the intervention arm in a pamphlet describing the 10 recommendations and their (simplified) rationale. Soon after inclusion, they receive a telephone call from a skilled adviser and, 4–6 weeks later, a similar call to discuss the practical aspects of the recommendations. Participants in the intervention group are further invited to a lecture focusing on diet and physical activity in pregnancy and a separate cooking lesson aiming at inspiring the participants to cook and make use of healthy foods such as fish, barley and a multitude of vegetables in their cooking.

Rationale for the 10 dietary recommendations

The overall purpose of the FFD study is to prevent excess weight gain and its inherent consequences for mother and child through physical activity and diet modification during pregnancy. The dietary recommendations are tailored for this purpose and aim at making the pregnant women reflect on their food choices. The recommendations have a scientific foundation and are consistent with current Norwegian dietary recommendations.

Eat regular meals

Studies have shown that young people who skip breakfast are more likely to be overweight or obese (Andersen et al. 2005; Grøholt et al. 2008; Croezen et al. 2009) and have a higher BMI than those who eat breakfast regularly (Barton et al. 2005; Delva et al. 2007). It is also documented that skipping meals leads to an increased number of snacking events and a higher intake of energy‐dense foods (Larson et al. 2009). In a review of the current US Department and Agriculture (USDA) dietary guidelines for pregnant women, the author lists ‘not skipping meals’ as important during pregnancy (Fowles 2006).

Drink water when thirsty

Increased consumption of sweetened beverages is a potential contributor to higher energy intake with consequent weight gain (Malik et al. 2006). There is also evidence that the satiating effect of beverages is lower than that of solid foods, possibly resulting in poor regulation of energy intake from beverages (Malik et al. 2006; Dennis et al. 2009). Results from the National Health and Nutrition Examination Survey show that drinking plain water predicts lower energy density of the foods consumed, while drinking beverages predicts higher energy density of foods (Kant et al. 2009).

Eat vegetables with dinner every day

Vegetables are low energy‐dense foods. The systematic literature review from the World Cancer Research Fund & American Institute for Cancer Research (2007) concludes that ‘overall, the epidemiological evidence that low energy‐dense foods protect against weight gain, overweight and obesity is substantial and generally consistent’. Studies show that the general Norwegian population has difficulties in including vegetables in the diet (The Norwegian Directorate of Health 2010b). In the traditional Norwegian meal pattern, with only one cooked meal daily, the easiest way to include vegetables is with dinner. In a recent study, only 40% of adolescents and 60% of their parents reported having vegetables with dinner on a random week day (Vejrup et al. 2008).

In between meals – choose fruits and vegetables

Because fruits and vegetables are mostly low‐energy foods and therefore will not contribute to excessive weight gain, the pregnant women are recommended to choose fruits and vegetables as in‐between‐meal snacks. This is in accordance with the USDA guidelines (Fowles 2006). Several studies show that in‐between meals or snacks tend to be energy dense and contain much sugar and fat (Fowles 2006; Jaeger et al. 2009). Among subjects having many snacking events during a day, the total energy from these meals may exceed the intake from ordinary meals (Sjöberg et al. 2003).

Eat sweets and snacks only when you really appreciate it

Sweets and snacks are energy dense and a high intake is related to excessive weight gain in pregnant women (Olafsdottir et al. 2006; Uusitalo et al. 2009). Studies show that sweets and snacks are frequently eaten unplanned (Wansink et al. 2010), often while watching television (Bowman 2006; Chaput et al. 2011) or even when bored (Nelson et al. 2009). Studies show that snacking is done automatically and that breaking such routines is difficult (Tam et al. 2010). An intervention study among undergraduate students showed that both the strategy of telling students to choose healthy instead of unhealthy snacks and that of telling them to limit their intake of unhealthy snacks, actually reduced the number of unhealthy snacking events (Tam et al. 2010). The pregnant women are recommended to choose vegetables or fruits as snacks in between meals and eat sweets only when they know they will really enjoy it, aiming at reducing the unplanned and casual intake. The purpose of this recommendation is to challenge participants to reflect on why they want to eat sweets and snacks.

Buy small portion sizes of unhealthy foods

The pregnant women are recommended to consider package size and buy small portion sizes of unhealthy food. Studies in young adults show that BMI is strongly related to their selection of large portion sizes of energy‐dense, high‐carbohydrate foods (Burger et al. 2007). These studies suggest that an intervention for preventing weight gain in young adults should focus upon the importance of portion sizes and upon increasing the awareness of eating habits in response to media messages and product packaging (Burger et al. 2007). Studies by Rolls et al. (2006) show that a reduction in portion size and energy density of foods are additive and lead to sustained decrease in energy intake. Further, studies by Chandon & Wansink (2002) show that people have difficulties regulating their food intake when they are aware of leftovers still available and that the choice therefore must be taken when purchasing and choosing small sizes in order to self‐regulate their intake (Chandon & Wansink 2002).

Limit your intake of added sugar

A limitation of sugary foods and drinks is recommended for the general population, as several systematic reviews state an association between liquid sugar and overweight (Norwegian Nutrition Council 2011). Clausen et al. (2001) showed that a high intake of sucrose during pregnancy was associated with an increased risk of pre‐eclampsia and overweight. Olafsdottir et al. (2006) showed that overweight pregnant women who ate more sweets early in pregnancy increased their risk of gaining excessive weight. A recently published Finnish study among pregnant women found similar results; a dietary pattern characterised by junk food, sweets and sugar‐sweetened soft drinks was positively associated with weight gain rate (Uusitalo et al. 2009).

Limit your intake of salt

A reduced intake of salt is recommended for the general population in order to reduce the risk of hypertension (World Cancer Research Fund & American Institute for Cancer Research 2007; Norwegian Nutrition Council 2011). There is also a quantitative relationship between salt intake and fluid consumption in adults and in children (He et al. 2008). Calculations done by He et al. (2008) show that a reduction in salt intake by half would result in an average reduction in 2.3 soft drinks per week in children in the UK. A reduction in salt intake could therefore play a role in reducing excess weight gain by reducing intake of energy‐containing drinks.

Do not eat beyond satiety

Both the increase in portion size and the availability of food has made it more common to eat beyond satiety. A study among undergraduate females showed that eating beyond satiety was the strongest predictor of BMI when controlled for other eating behaviours (Yanover & Sacco 2008). Another study of African‐American and Caucasian women showed that the odds of becoming obese increased sixfold for Caucasian and 15‐fold for African‐American women who ate beyond satiation every day compared with those who rarely or never ate beyond satiation (Brewer et al. 2003).

Read nutritional labels of foods before buying

A study among American adults showed that those who read nutrient labelling had a healthier diet than those who did not read the lists. This association was specifically shown in relation to energy intake (Ollberding et al. 2010). The pregnant women in our study are taught how to understand ingredient lists and food labelling, aiming at assisting them in making healthy dietary choices.

Key messages

• Aiming at preventing excessive weight gain during pregnancy, 10 dietary recommendations are given to pregnant women in the intervention arm of the Fit for Delivery study (FFD).

• A dietary score was constructed from a 43‐item food frequency questionnaire and used to evaluate adherence to the recommendations. Acceptable test‐retest reliability of the FFQ and the dietary score was found.

• The score will be used in the ongoing FFD study to measure adherence to the dietary recommendations. If positive effects of the dietary intervention are documented, tailored advice on diet and dietary behaviour can be implemented into routine pregnancy and obstetrical care.

Materials and methods

Test‐retest of the questionnaire

A 220‐item questionnaire was developed for the FFD study to assess demographic and socioeconomic factors at inclusion and aspects of diet and physical activity prospectively and repetitively throughout pregnancy and during the first year post‐partum. In the test‐retest reliability study, pregnant women from the Kristiansand area in the southern part of Norway were invited to participate by completing the questionnaire regarding pre‐pregnant weight and height, marital status, level of education, smoking, alcohol, drugs, diet and physical activity before and during pregnancy on two occasions 2 weeks apart. Participants were recruited while attending ultrasound screening at Sorlandet Hospital HF, usually around 18–20 weeks of gestation. Written information was sent to potential participants 1 week before the scheduled ultrasound appointment. They accepted the invitation by filling in the questionnaire at the hospital (time 1) and received an identical questionnaire by mail 2 weeks later (time 2). The second questionnaire was returned to the hospital in a pre‐stamped and addressed envelope. One hundred fifty‐four pregnant women agreed to participate in the test‐retest study. Of these, 106 (69%) completed the questionnaire on both occasions and were included in the test‐retest analysis. The reliability study was carried out between October 2009 and December 2010. In the present paper, the test‐retest reliability of the food frequency part of the questionnaire (FFQ) is reported.

Instruments

The FFQ consisted of 43 items concerning dietary behaviour and the consumption of selected foods and drinks. The FFQ was constructed to yield information on the dietary aspects covered in the study and therefore did not cover the diet as a whole. Questions covered meal frequency, drink items, fruits and vegetables, sugary foods, sweets and snacks, fast food, portion sizes of unhealthy foods and drinks, enjoyment of food and familiarity with food labelling. To be able to measure the degree of adherence to the 10 dietary recommendations, a dietary score was constructed from 10 subscales, each subscale referring to one of the corresponding FFD recommendation. The subscales were single variables or sum scores constructed from relevant questions from the FFQ. Each computed subscale was then dichotomised using the median as cut‐off, and assigned a value of 0 or 1, with 1 indicating the healthier dietary behaviour. Adding the dichotomised subscales resulted in a dietary score that could take values from 0 (low adherence) to 10 (high adherence). This procedure is in accordance with the method used in assessing adherence to the Mediterranean diet (Bach et al. 2006) and adherence to healthy aspects of the Nordic diet (Olsen et al. 2011). Methods for summarising diets by means of a single index or score resulting from a function of different dietary components are increasingly being used in epidemiological studies (Bach et al. 2006). Score components may be selected by data‐driven statistical techniques like factor analysis or cluster analysis, or a priori based on knowledge or scientific evidence within the area of interest (Bach et al. 2006). In the present study, the score is constructed on the basis of a priori defined aspects of diet and dietary behaviour related to the risk of excess weight gain during pregnancy. Table 1 describes the 10 recommendations, the FFQ questions included in each subscale, resulting cut‐off points (median) and the dietary behaviour required for scoring within each subscale. Missing values in the subscales were assigned a value of 0.5 in order to neutralise influence upon the total score. To be able to investigate test‐retest agreement of the dietary score, the study population was divided into tertiles according to their computed total score, and grouped into ‘low’ (0–3.5), ‘medium’ (4.0–5.5) and ‘high’ (6.0–10) adherence to the dietary recommendations.

Table 1.

The FFD study. The components underlying the construction of 10 subscales that constitutes the FFD dietary score based on the 10 dietary recommendations for preventing excess weight gain during pregnancy and forwarded to the intervention group in the FFD study (n = 106)

Dietary recommendations	Related FFQ question(s)	Response alternatives and coding	Calculations	Possible range of each subscale	Median = cut‐off	Dietary behaviour associated with scoring
1. Eat regular meals	How often do you eat breakfast lunch dinner evening meal/supper	Never = 0 Less than once a week = 0.5 Once a week = 1 Twice a week = 2 Three times a week = 3 Four times a week = 4 Five times a week = 5 Six times a week = 6 Every day = 7	Sum of answers to the four questions	0–28 meals a week	Test: 26 Retest: 25	Test: ≤25 meals week−1 = 0 ≥26 meals week−1 = 1 Retest: ≤24 meals week−1 = 0 ≥25 meals week−1 = 1
2. Drink water when thirsty	How often do you drink tap water bottled water carbonised water whole‐fat milk low‐fat milk skimmed milk juice fruit drinks or nectar sugar‐containing beverages artificially sweetened beverages alcohol‐containing beverages	Never = 0 Less than once a week = 0.5 Once a week = 1 Twice a week = 2 Three a week = 3 Four times a week = 4 Five times a week = 5 Six times a week = 6 Every day = 7 Several times a day = 10	Sum of water intake events (frequency) divided by sum of all other drink intake events (frequency) multiplied by 100	0–100%	Test: 40% Retest: 38%	Test: <40% of drinking events is water = 0 ≥40% of drinking events is water = 1 Retest: <38% of drinking events is water = 0 ≥38% of drinking events is water = 1
3. Eat vegetables with dinner every day	How often do you eat vegetables with dinner?	Never to daily (0–7)	No calculation	0–7	Test: 5 Retest: 4	Test: <5 times week−1 = 0 ≥5 times week−1 = 1 Retest: <4 times week−1 = 0 ≥4 times week−1 = 1
4. In between meals – choose fruits or vegetables	How often do you eat fruits or vegetables as an in‐between meal? How often do you eat in‐between meals?	Never to several times a day (0–10)	Frequency of eating fruits or vegetables in between meals divided by frequency of having in‐between meals multiplied by 100	0–100%	Test: 80% Retest: 86%	Test: Fruits/vegetables <80% of in‐between meals = 0 Fruits/vegetables ≥80% of in‐between meals = 1 Retest: Fruits/vegetables <86% of in‐between meals Fruits/vegetables ≥86% of in‐between meals
5. Eat sweets and snacks only when you really appreciate it	How often do you eat sweets or unhealthy foods without really appreciating it?	Never to several times a day (0–10)	No calculation	0–10	Test: 0.5 Retest: 0.5	Test and retest: Never eating sweets without appreciation = 1 Sometimes eating sweets without appreciation = 0
6. Buy small portion sizes of unhealthy foods	Which size of the following three items do you usually buy? crisps chocolate soda	Small = 0 Big = 1	Sum of three answers	0–3	Test: 3 Retest: 3	Test and retest: Buying small portion size of at least one of the three items = 1 Buying big size of all three items = 0
7. Limit your intake of added sugar	How often do you have sugar‐sweetened fruit drinks soda cookies/crackers sweet breads cake/muffins sugar‐containing cereals sugar‐containing fruit yoghurt chocolate/sweets How often do you add sugar to your food?	Never to several times a day (0–10)	Sum of answers to the nine questions	0–90	Test: 10 Retest: 10	Test and retest: <10 (approximately equivalent to eating food with added sugar once a day or less) = 1 ≥10 (approximately equivalent to eating food with added sugar more than once a day) = 0
8. Limit your intake of salt	How often do you eat salted crackers noodles crisps or other salted snacks hot dogs from kiosk/gas station fried potato chips from fast‐food restaurants canned or freeze‐dried food How often do you add salt to your food?	Never to several times a day (0–10)	Sum of answers to the seven questions	0–70	Test: 7.5 Retest: 7.5	Test and retest: <7.5 (fast‐foods, snacks or salting of food no more than once a day) = 1 ≥7.5 (fat‐foods, snacks or salting of foods more than once a day = 0
9. Do not eat beyond satiety	How often do you eat beyond satiety?	Never to several times a day (0–10)	No calculation	0–10	Test: 1 Retest: 1	Test and retest: Less than once a week = 1 Once a week or more = 0
10. Read nutrition labels on foods before buying		Never = 1 Sometimes = 2 Usually = 3 Always = 4	No calculation	1–4	Test: 2 Retest: 2	Test and retest: Never reading labels = 0 sometimes or often reading labels = 1
Total FFD score			Sum of the 10 dichotomised subscales	0–10	Test: 5 Retest: 5	Adherence test and retest: ≤3.5 = low 4–5.5 = medium ≥6 = high

FFD, fit for delivery; FFQ, food frequency questionnaire.

Procedures and statistical analysis

The data was analysed using the computer program spss statistical software package version 19.0 (IBM Corporation, Armonk, NY, USA). Test‐retest reproducibility of the selected FFQ questions or sum scores (subscales) was assessed using bivariate correlations. As 7 of the 10 subscales were not normally distributed, rank order correlation was estimated with Spearman's correlation coefficient. The final score was normally distributed and is therefore presented with Pearson's correlation coefficient. Cross tabulations and Cohen's kappa coefficients were used to investigate the test‐retest agreement of classification into the three adherence categories. The same method was used to investigate the test‐retest classification of the subscale scoring. For the test‐retest cross tabulation of the individual dichotomised subscales, participants with the assigned missing value of 0.5 were excluded, yielding different number of participants in the various analyses (Table 2).

Table 2.

The FFD study. Test‐retest reliability of the 10 subscales that constitute the FFD dietary score and for the trichotomised score

The 10 subscales related to the corresponding FFD dietary recommendations	Spearman's rank order correlation	P‐value	Kappa measure of agreement (dichotomised subscales)	Percent agreement between test and retest (dichotomised subscales)
Meal frequency (n = 103)	0.84	≤0.001	0.71	85
Choosing water when thirsty (n = 98)	0.70	≤0.001	0.55	78
Vegetables for dinner (n = 103)	0.76	≤0.001	0.45	73
Fruits and vegetables as in‐between‐meal snack (n = 92)	0.56	≤0.001	0.39	70
Eating sweets and snacks without appreciation (n = 100)	0.68	≤0.001	0.63	82
Portion size of sweets and snacks (n = 93)	0.77	≤0.001	0.74	87
Frequency of eating foods and drinks with added sugar (n = 100)	0.78	≤0.001	0.62	81
Frequency of eating foods and snacks with added salt (n = 99)	0.77	≤0.001	0.61	81
Frequency of overeating (n = 102)	0.81	≤0.001	0.65	83
Frequency of reading food labels before buying (n = 101)	0.73	≤0.001	0.58	82
FFD dietary score (n = 106)	0.68*	≤0.001	0.51 †	68 †

FFD, fit for delivery. *Pearson correlation coefficient is used for the FFD score. ^†Trichotomised score.

Ethical considerations

The test‐retest study was included in the protocol for the FFD study that was approved by the Norwegian Regional Committee for Medical Research Ethics.

Results

A total of 106 pregnant women completed the questionnaire on both occasions and were included in the analyses. Mean age was 29 years (range 20–42), and median gestational week at the time of completing the first questionnaire was 19 (range 16–31). Table 2 presents details for the test‐retest comparison. The correlation coefficient between test and retest of the dietary score was r = 0.68, P ≤ 0.001 (Pearson) and between the test and retest subscales r = 0.56–0.84, P ≤ 0.001 (Spearman's rank correlation coefficient). Regarding the test‐retest agreement of the dietary scoring, 68% was correctly classified into low, medium or high adherence on both occasions (kappa measure of agreement 0.51). Four per cent was grossly misclassified, moving either from high to low adherence or the opposite way. For the dichotomised subscales, 70–87% were correctly classified on both occasions. The kappa measure of agreement between test and retest subscale scoring ranged from 0.39 to 0.74 (Table 2).

Discussion

Ten dietary recommendations are given to pregnant women in the intervention arm of the ongoing Norwegian FFD study. We hypothesise that adherence to the recommendations will contribute to the prevention of excess weight gain during pregnancy without sacrificing normal eating or the enjoyment of food in general. To be able to rank participants according to degree of adherence to the 10 recommendations, a dietary score was constructed based on relevant subscales constructed from the FFQ.

The test‐retest Spearman correlation coefficients from 0.59 to 0.74 for the subscales were good considering that correlation coefficients on the order of 0.5–0.7 are typical for the reproducibility of nutrient intakes and comparable with that of many biological measurements made among free‐living subjects (Willett 1998). Erkkola et al. (2001) conducted a reproducibility study of a 181‐item FFQ in 111 pregnant women who completed the FFQ twice at a 1‐month interval (Erkkola et al. 2001). The intraclass correlation coefficients for nutrients ranged from 0.44 (ice cream) to 0.91 (coffee), and the authors conclude that the FFQ has an acceptable reproducibility and represents a useful tool for categorising pregnant women according to their dietary intake. Bosco et al. (2010) investigated reproducibility of self‐reported dietary intake during an earlier pregnancy, 3 months apart, reporting mean intraclass correlations of energy‐adjusted nutrients of 0.59 (range 0.41–0.69) and good agreement of reporting multivitamin use during pregnancy (kappa = 0.66–0.85) (Bosco et al. 2010). Shu et al. (2004) investigated reproducibility of an FFQ used in the Shanghai Women's Health study with 191 participants completing the questionnaire twice, 2 years apart. They report correlation coefficients from 0.41 to 0.66 for daily intake of food groups such as rice, poultry, red meat, fish, soy foods, vegetables and fruits. Hu et al. (1999) assessed reproducibility of an FFQ completed 1 year apart in a subsample of men from the Health Professionals Follow‐Up Study in 1986, comparing two dietary patterns (so called ‘prudent’ and ‘Western’ pattern) on two occasions 1 year apart, reporting correlation coefficients ranging from 0.36 to 0.92 for individual foods, 0.70 for the prudent pattern and 0.67 for the Western pattern (Hu et al. 1999), comparable with the ones presented in our study.

In this study, a combination of methods is used to evaluate the test‐retest reliability of the FFQ and the resulting dietary score. The kappa statistic was used together with observed percent agreement as a measure of chance‐corrected proportional agreement (Altman 1991). According to Altman (1991), values of kappa above 0.80 indicate very good agreement, 0.61–0.80 good agreement, 0.41–0.60 indicates moderate agreement, 0.21–0.40 fair agreement and <0.20 poor agreement. The test‐retest subscale correlation in this study of r = 0.56–0.84 combined with 70–87% correct classification of the dichotomised subscales and kappa values ranging from 0.39 to 0.74, indicate at least acceptable test‐retest reliability of the FFQ. The test‐retest dietary score correlation of r = 0.68 combined with 68% correct classification, only 4% grossly misclassified and a kappa value of 0.51, also indicate acceptable test‐retest reliability of the trichotomised score (Masson et al. 2003).

The measuring of adherence to a dietary behaviour with a single score has its limitations. Five of the 10 subscales were based on one FFQ question only, yielding few response alternatives and a skewed distribution. Hence, for some subscales, the dichotomising by the median resulted in differently sized groups. For the purpose of this study, however, and by inspecting the resulting cut‐off points, we feel confident that the sample was acceptably divided. The identical weighting or contribution of each subscale to the final dietary score may be questioned. It is probably not possible to predict the specific impact of adherence to any one of the 10 dietary recommendations in the complicated picture of preventing excess weight gain in pregnancy.

Neither the dietary score, nor the 43‐item FFQ, has been validated. Several of the FFQ questions are inquiring dietary behaviour rather than food or drink intake, rendering validation more challenging. A 24‐h recall on selected food and drink items was included in the questionnaire, but was not sufficiently detailed for the purpose of validation. The score will be used as an instrument to rank study participants according to degree of adherence to this a priori defined dietary behaviour, hypothesised to be protective against excessive weight gain during pregnancy. This relative ranking of adherence will be the measure of a potential effect of the dietary intervention in the FFD study. The score will further be used to investigate differences in dietary behaviour related to demographic and socioeconomic factors and to explore longitudinal dietary changes during pregnancy and the following years related to aspects of mother and child health.

Conclusion

Ten specific and simplified dietary recommendations are given to pregnant women in the intervention arm of the FFD study in order to prevent excessive weight gain during pregnancy and its inherent risks and potential complications. A dietary score was constructed to evaluate adherence to the recommendations. Acceptable test‐retest reliability of the FFQ and the dietary score was found. The score will be used in the ongoing FFD study as a measure of adherence to the dietary recommendations. If positive effects of the dietary intervention are documented through the FFD study, tailored advice on diet and dietary behaviour can be implemented into routine pregnancy and obstetrical care.

Source of funding

The study was financially supported by Helse Sør Øst.

Conflicts of interest

The authors declare that they have no conflicts of interest.

Contributions

LRS, IV and EB conceived the FFD study. NCØ and EB designed the present study and developed the dietary recommendations. NCØ drafted the rationale. LRS was responsible for the data collection. ERH constructed the dietary score in cooperation with NCØ and EB, carried out statistical procedures and drafted the methods and discussion section. The other authors revised the paper critically.