In The Netherlands, there were approximately 135 food banks in 2011–2012. The inclusion criterion for food banks to participate in our study was provision of food parcels once a week. The ones providing food parcels every other week were excluded. To obtain a sample which reflects the different food banks and, therefore, recipients best, we recruited food banks that varied in the number of recipients (smaller [n < 50], medium [n = 50–100] as well as larger [n > 100] numbers of recipients), urbanization level (smaller [e.g., 30,000 inhabitants] as well as larger [e.g., > 120,000 inhabitants] cities), and region (different regions across The Netherlands). Eventually, food banks were selected based on their willingness to participate. Each participating food bank had its own way of working. They either collect foods and compose the content of the food parcels themselves, or they receive ready-to-supply food parcels, which are composed elsewhere (e.g., a distribution centre, another food bank). However, they all aim to supply food parcels of which the content is—more or less—similar.
The study was approved by the Medical Ethical Committee of the VU Medical Center in Amsterdam, The Netherlands, as well as the national board of the Dutch Food Bank. Participants were exempt from informed consent by the Medical Ethical Committee of the VU Medical Center. Prior to this study, a pilot was carried out among four food bank recipients from the food bank in Huizen, to test our materials and the feasibility of the measurements. This was done by administering the general and food security questionnaires and conducting the 24-hour (24-h) recall interview and anthropometric measurements in the four food bank recipients to find out whether the questions in the questionnaires were clear, and how much time it would take to conduct the measurements and interviews.
The target population consisted of recipients of the 12 participating food banks. We recruited recipients through information letters, active recruitment at the food banks, and promotional posters. They could sign up for the study with an application form, by telephone or e-mail within either 2 or 3 weeks after they were informed about the study through information letters, the researchers at the food banks, and promotional posters. Inclusion criteria for participation were: (1) ≥ 18 years of age; (2) adequate command of the Dutch language to participate in oral and written interviews; (3) recipient of a Dutch food bank > 1 month; (4) collect own food parcel at the food bank; and (5) possible to be contacted by phone. Only one member per household was enrolled. Of the approximately 1200 food bank recipients at the participating food banks, 284 voluntarily indicated that they were interested to participate, of which 173 (60.9%) actually participated in the study. For 66 of the 111 recipients who signed up for participation, but did not participate, we were able to ascertain the reason for non-participation: (1) not a food bank recipient anymore at the start of the study (n = 19); (2) lack of time (n = 12); (3) no longer wants to participate (n = 11); (4) no adequate command of the Dutch language (n = 8); and (5) other reasons (n = 16) such as illness or not willing to participate in anthropometric measurements. Of the 45 remaining recipients who did not participate after signing up, 15 recipients did not show up at the first measurements, 29 recipients did not respond to e-mail and/or phone calls, and 1 recipient did not fill in contact information. Data collection was scheduled between September 2011 and February 2012, and collected through a general questionnaire, anthropometric measurements, and 24-h recalls. Participants were excluded from data analyses in case of < 3 24-h recalls and/or a missing general questionnaire (n = 6). A total of 167 participants, approximately 14% of the 1200 food bank recipients at the participating food banks, were included for data analyses. Participants who completed the study received both a gift coupon of 5 Euros and a small incentive.
Participants completed a self-administered general questionnaire at the food bank, which consisted of questions regarding socio-demographics, lifestyle factors, nutrition, and the appreciation of the food parcels. Participants who had difficulties in reading or writing, were offered help filling in the questionnaire.
Socio-demographics included date of birth, sex, duration of being recipient of the Dutch food bank (0–6, 6–12 months, and > 12 months), household composition (number of children < 18 years, ≥ 18 years, and adults in household), and educational level. For the covariable household composition, we created three categories: single parent household (including one adult and at least one child), household without children (including at least one adult and no children), and multiple household with children (including at least two adults and at least one child). We created three levels of education: low (less than finished elementary school), medium (finished elementary school) and high (higher than finished elementary school).
Lifestyle factors included current smoking (yes/no) and physical activity. Physical activity was established by asking “How many days a week are you physically active at moderate intensity for at least 30 min?”. Moderately intense physical activity included sport activities, walking, cycling, gardening, and performing heavy housework. We created two categories (0–4 days a week/5–7 days a week) based on the national physical activity guidelines for adults; at least 5 days a week 30 min of physical activity at moderate intensity .
Regarding nutrition, we asked participants “How satisfied are you with your current food intake?” (very unsatisfied, unsatisfied, not unsatisfied/not satisfied, satisfied, and very satisfied), and “How healthy do you think your current food intake is?” (very unhealthy, unhealthy, not unhealthy/not healthy, healthy, and very healthy). Questions regarding the food parcels included: “How satisfied are you usually with the content of the food parcel?” (very unsatisfied, unsatisfied, not unsatisfied/not satisfied, satisfied, and very satisfied), “How healthy do you think the content of the food parcel is in general?” (very unhealthy, unhealthy, not unhealthy/not healthy, healthy, and very healthy).
Trained researchers measured participants body height, body weight, and waist circumference, according to a standardized protocol developed for this study. For the hereafter-described height and weight measurements, participants were asked to remove any items from their pockets and to take off their shoes and coat. A portable stadiometer, type Seca ® 214, was used to measure height to the nearest 0.1 cm, and a calibrated mechanical scale, type Seca ® 761, to measure weight to the nearest 0.5 kg. Waist circumference was measured in duplicate on bared skin, with a measurement tape to the nearest 0.1 cm. Body mass index (BMI) was calculated as measured weight (kg) divided by measured height squared (m2). BMI cut-off points of the WHO were used to define weight status .
Data collection of the 24-h recalls
Data on food intake and supplement use were collected by trained interviewers through three 24-h recalls in a 3-week period using the USDA five-step multiple pass method (MPM) [18,19,20]. The MPM method was developed for collecting interviewer-administered 24-h recalls and includes multiple passes through 24 h of the previous day, during which respondents receive cues to help them remember and describe foods and drinks that they consumed [18,19,20].
The first 24-h recall was conducted in-person, during which a table scale, type KERN FCE 6 K2 ®, extensive tableware, and a portion-size photo book were used to assist in portion-size estimation of consumed foods and drinks. The portion-size photo booklet was taken home by the participants to use in the second and third 24-h recalls, which were conducted by phone. We aimed to obtain dietary information on two different weekdays and one weekend day, or if not possible, on three different weekdays. Twenty-four hour recalls were conducted at dates and times unannounced to the participants.
Data processing of the 24-h recalls
All recorded foods and drinks from the three 24-h recalls were coded with the corresponding Dutch Food Composition Table code (NEVO-code) [21, 22]. Portion sizes consumed were entered in gram weights. Energy content and nutrient composition of the food and drinks consumed was determined using the 2010 NEVO-database , which provides the nutrient composition of foods and drinks commonly consumed in The Netherlands.
Data Dutch National Food Consumption Survey 2007–2010
To compare dietary intake of the Dutch food bank recipients with dietary intake of a representative sample of the general Dutch adult population, we used a selection of the food consumption data from male and female adult participants of the Dutch National Food Consumption Survey 2007–2010 (DNFCS) . This selection of the DNFCS consisted of participants aged 23–69 years, similar to the age range of the Dutch food bank recipients, and included a low-SES sample, based on the highest completed educational level; primary or lower vocational education. DNFCS-all indicates all adults aged 23–69 years (n = 1933), whereas DNFCS-low-SES indicates the subgroup (n = 312) with a low-socioeconomic status (SES). Briefly, participants were drawn from representative consumer panels. Inclusion criteria were a maximum of one person per household and an adequate command of the Dutch language. Data were collected between March 2007 and April 2010. Participants completed a general questionnaire either on paper or online on various background and lifestyle factors . Two non-consecutive 24-h recalls using the EpicSoft software  were conducted per participant by telephone, at dates and times unannounced to the participants . The 2011 NEVO database was used to determine energy content and nutrient intake. In addition, self-reported height (cm) and body weight (kg) were recorded during the telephone interviews. All interviews were carried out by trained dieticians.
Characteristics of the food bank recipients and the DNFCS participants were analyzed with IBM SPSS Statistics for Windows version 21.0 (Armonk, NY: IBM Corp, USA). Descriptive statistics were used to summarize the participant characteristics. Continuous variables were presented as mean and standard deviation (SD), whereas categorical variables were presented as frequency and relative frequency.
To make the dietary intake data more normal distribution like, a Box–Cox transformation was used. This was visually tested with Q–Q plots. Habitual dietary intake data, i.e., the long-term mean intake of energy, macronutrients, fruit, vegetables, and fish with the accompanying distributions, were estimated from the observed daily intake by correction for the intra-individual (day-to-day) variation. This was done using SPADE (Statistical Program to Assess Dietary Exposure, version 3.1, RIVM), which was implemented in R  version 3.3.1 (The R Foundation for Statistical Computing, Austria). This is a highly advanced method developed by the National Institute for Public Health and the Environment to estimate habitual dietary intake distributions on national level , which takes the within- and between-person variation into account. Uncertainty in the habitual intake distribution and the proportion below or above a cut-off value was quantified with bootstrap (1000 samples), providing 95% confidence intervals (95% CIs). Analyses for DNFCS-all and DNFCS-low-SES included age and sex as covariables to adjust for possible differences in dietary intake between males and females, and were weighed for small deviations in sociodemographic characteristics (level of education, region, and urbanization) from the Dutch population, deviations from an equal distribution of days of the week, and season of data collection based on the first interview day .
For the analyses of the habitual dietary intake of Dutch food bank recipients, an extension of SPADE was used to include non-personal covariables, i.e., household composition, and the number of days between receiving a food parcel and the day of the recall, besides age and sex. These non-personal covariables were added to the model, because they are possibly associated with dietary intake; adult caregivers may sacrifice their own diet to avoid that their children should experience hunger , and the more days between receiving a food parcel and the day of the recall, the less food is assumed to be available from the food parcel. Furthermore, a weight factor was included for sex (i.e., male: 1.34756 and female: 0.79477) to make the results more comparable with the DNFCS, where the proportion of males and females was nearly 50/50. The estimated habitual intake distribution is presented by its mean and some percentiles with the corresponding 95% CI.
To evaluate the dietary intake, energy, macronutrients, fruit, vegetables, and fish intakes were compared with DRIs, if available. The contribution of nutrients from dietary supplements was not considered. We compared dietary intake with the Dutch nutritional guidelines for a healthy diet which were in use at the time we collected our data [28,29,30]. The percentages of participants below or above these DRIs [28,29,30,31,32,33] were based on cut-off values of the recommended dietary allowance for carbohydrates, and protein, the daily adequate intake for fat, the daily tolerable upper level for polyunsaturated fat, and trans-fat, and the daily recommendation for fiber, fruit, vegetables, and fish.
We determined differences in estimated habitual intakes and percentages of participants not meeting the DRIs between the Dutch food bank recipients, the DNFCS-all and the DNFCS-low-SES by comparing the 95% CIs. When the 95% CIs did not overlap, the difference between the groups was considered to be statistically significant.