1 Dietary intake, nutritional status and mental wellbeing of homeless adults in Reading, UK1-4 Rosalind Fallaize1,2*, Josephine V. Seale1*, Charlotte Mortin1, Lisha Armstrong1, Julie A. Lovegrove1 1Hugh Sinclair Unit of Human Nutrition, Department of Food and Nutritional Sciences, University of Reading, Whiteknights, PO Box 266, Reading, RG6 6AP, UK. 2School of Life and Medical Sciences, University of Hertfordshire, College Lane, Hatfield, AL10 9AB, UK. *R Fallaize and J Seale are joint first authors Corresponding Author: Julie A. Lovegrove, Hugh Sinclair Unit of Human Nutrition, Department of Food and Nutritional Sciences, University of Reading, Whiteknights, PO Box 266, Reading, RG6 6AP, UK. Tel: +44(0)118 378 6418, Fax: +44(0)118 931 0080. E-mail: j.a.lovegrove@reading.ac.uk Running Head: Nutrition in homeless compared to housed adults Keywords: homelessness, mental health, nutrient intake, dietary methodology, nutrition intervention Abbreviations: CVD, cardiovascular disease; DRV, dietary reference value; EPIC, European Prospective Investigation into Cancer and Nutrition; FETA, FFQ EPIC tool for analysis; NSP, non- starch polysaccharides; PHQ-SADS, patient health questionnaire: somatic anxiety depressive symptoms; RNI, recommended nutrient intake; %E, percentage of total energy intake. 1 Abstract 1 Malnutrition has been reported in the homeless, yet the specific nutritional issues faced by each 2 homeless community are unclear. This is in part due to nutrient intake often being compared to 3 dietary reference values as opposed to a comparative housed population. Additionally, the 4 complex interplay between nutrient intake, reward mediated behaviour and mental illness is 5 frequently overlooked. This study aimed to compare the dietary intake, nutritional status and 6 mental wellbeing of homeless and housed adults. Homeless (n=75) and matched housed (n=75) 7 adults were recruited from Reading (UK). Nutrient intake was determined using the EPIC Norfolk 8 Food Frequency Questionnaire. The Patient Health Questionnaire: Somatic Anxiety Depressive 9 Symptoms (PHQ-SADS) assessed for signs of mental illness. Demographic, behavioural and 10 physiological information was collected using closed-ended questions and anthropometric 11 measurements. Overall, dietary intake was poorer in homeless adults who reported higher intakes 12 of salt (8.0g vs. 6.4g, P=0.017), SFA (14.6% vs. 13.0%, P=0.002) and alcohol (5.3% vs. 1.9%, 13 P<0.001) and lower intakes of fibre (13.4g vs. 16.3g, P<0.001), vitamin C (79mg vs. 109mg, 14 P<0.001) and fruit (96g vs. 260g, P<0.001) than housed. Smoking, substance misuse and PHQ-15 SADS scores were also higher in the homeless (P<0.001). Within the homeless population, street-16 homeless (n=24) had lower SFA (13.7% vs.15.0%, P=0.010), calcium (858mg vs. 1032mg, 17 P=0.027) and milk intakes (295g vs. 449g, P=0.001) than hostel residents (n=51), which may 18 reflect the issues with food storage. This study highlights the disparity between nutritional status in 19 homeless and housed populations and the need for dietary intervention in the homeless 20 community. 21 2 Introduction 22 Homelessness is a global issue. In England, over 68,500 households in 2015 were classified as 23 homeless (1) and a further 3,569 were considered rough sleepers (2). The lack of stable 24 accommodation, in conjunction with a low or absent income, raises challenges for the homeless, 25 one of which is access to food, although this is likely to differ between rough sleepers and those in 26 temporary accommodation. However despite the widespread prevalence of homelessness, 27 relatively few studies have sought to determine the nutritional status of these individuals, a 28 situation that may partly reflect the difficulty in collecting data from this transient population. 29 Despite this, several themes have emerged from studies to date, including a high SFA, low fruit 30 and vegetable diet (3-6), elevated serum cholesterol (7) and low levels of vitamin B-6, calcium and 31 iron(7; 8), although findings have not been consistent between countries. For example, lower skin-32 fold thickness and muscle mass measurements have demonstrated ‘wasting’ in homeless 33 communities in Germany and the US (1989-2001) (4; 7; 8), whereas others have found a proportion 34 of homeless that, according to their BMI, are overweight or obese in the US (2012-2013) (9; 10). 35 In the presence of physiological stresses arising from exposure to harsh environmental 36 conditions, the absence of a nutritionally balanced diet is likely to have a detrimental impact on the 37 health of a homeless individual. Elevated levels of acute and chronic disease (11; 12), increased 38 visits to emergency departments (13) and an average age of death of 47 years in the UK (14) 39 highlight the additional challenges faced by the homeless community. Cardiovascular disease 40 (CVD), for which diet is a key modifiable factor, is reported as the leading cause of mortality in 41 homeless adults between 45-65 years, (15; 16). 42 A higher prevalence of mental illness (e.g. depression, anxiety) in the homeless compared to the 43 general population has also been reported (12; 17). Reward mediated behaviors such as problematic 44 alcohol use, smoking and substance misuse are also frequently described (4; 12). The reported 45 substitution of food with alcohol by alcoholic homeless individuals (4) and the use of money for illicit 46 substances as opposed to food in homeless drug addicts (12) demonstrates the potential nutritional 47 consequences arising from reward mediated behavior. Addictive substances may also impact upon 48 the absorption, metabolism and/or requirements for nutrients (18; 19). 49 3 Habitual diets and culture limit the extent to which nutrient intake of the homeless is comparable 50 between countries or regions (20). Furthermore, studies have often failed to account for the impact 51 of physical and mental wellbeing on dietary intake; do not use a control group and have 52 widespread reliance on dietary reference values (DRV), which is potentially misleading. Comparing 53 homeless and housed individuals within the same region would overcome this issue, providing a 54 more accurate assessment of homeless intake in the specific location. However such studies are 55 currently lacking. 56 The aim of the present research was to compare nutritional status, dietary intake and mental 57 wellbeing in a group of homeless with age and gender-matched housed individuals (control group) 58 in Reading, UK. A secondary aim was to establish the marginal impact of homeless status (e.g. 59 street homeless vs. hostel residents) on these parameters. 60 61 Methods 62 Participants and study design 63 In this cross-sectional observation study homeless (n=75) and housed (n=75) men and women 64 were recruited. During recruitment, the two groups were broadly matched on the basis of age 65 range (18-29, 30-39, 40-49, >50yrs), gender and ethnicity. In order to capture homeless individuals 66 at the more extreme stages of homelessness this study only included individuals ‘living rough’ on 67 the street and those in the initial stages of housing (supported living). Street homeless were 68 recruited from a church drop-in centre that provides hot meals. Two hostels (charity and council 69 led) were used to recruit individuals residing in ‘stage 1’ accommodation. In addition to 70 accommodation for adults previously living on the street, the charity-led hostel also provides ‘in 71 house’ meals. Recruitment posters were displayed in each venue by staff, and residents/clients 72 recruited on a first-come basis. Housed volunteers were recruited at random through a volunteer 73 database at the Hugh Sinclair Unit of Human Nutrition, Reading, and via posters displayed at 74 Reading central library and around the Reading University campus. The University of Reading 75 School of Chemistry, Food and Pharmacy Research Ethics Committee (Approval number: 20/14) 76 and The Salvation Army Ethics Committee granted ethical approval for the study. All participants 77 gave informed written consent prior to participation. 78 4 Data Collection 79 Each participant completed three questionnaires during a single research session. A room was 80 provided at each venue for the purpose of the study and one of four trained researchers and a 81 volunteer from the specific venue were present during each session. Each volunteer was provided 82 with the option of either completing the questionnaires themselves or being asked the questions by 83 the researcher. For those opting to self-complete the questionnaires, responses were checked and 84 verified prior to departure. Questionnaires took between 60-90 minutes to complete. Volunteers 85 were reimbursed for their participation in the study via a £10 shopping voucher. 86 Questionnaires 87 To assess nutrient intake, the European Prospective Investigation into Cancer and Nutrition 88 (EPIC)-Norfolk FFQ was used. The FFQ is a validated semi-quantitative questionnaire consisting 89 of 130 food and drink items (21; 22). For each item the participant is required to choose one out of 9 90 possible frequency options ranging from ‘never/less than once a month’ to ‘6+ per day’. FFQs are 91 typically completed with reference to the preceding 6 months to 1 year. However to ensure 92 reported nutrient intake related to the time an individual was homeless, participants were asked to 93 complete the FFQ with reference to the last month only, as some had been homeless for one to 94 two months. FFQ EPIC tool for analysis (FETA) software was used to determine daily nutrient 95 levels from the FFQ responses, which is based upon McCance and Widdowson’s The Composition 96 of Foods (5th edition) and its supplements (23). Energy, macronutrients (including sub-classes of fats 97 and carbohydrates), key vitamins and minerals (associated with public health concerns), and 14 98 food groups (alcoholic beverages, cereal and cereal products, egg and egg dishes, fats and oils, 99 fish and fish products, fruit, meat and meat products, milk and milk products, non-alcoholic 100 beverages, nuts and seeds, potatoes, soups and sauces, sugars; preserves and snacks, and 101 vegetables) were reported in the present analysis. 102 The validated Patient Health Questionnaire: Somatic Anxiety and Depressive Symptoms (PHQ-103 SADS) was used to determine the presence of symptoms related to a mental health condition (24). It 104 combines three questionnaires to screen for the presence of depression (PHQ-9), anxiety (GAD-7) 105 and somatization (PHQ-15) disorders. Each part of the questionnaire resulted in a score from 106 5 which individuals are categorized as ‘none’, ‘mild’, ‘moderate’ or ‘severe’ with relation to the 107 presence of depressive, anxiety and/or somatic symptoms. 108 A ‘Health and Lifestyle’ questionnaire formulated specifically for the present study was used to 109 record demographic data as well as information about smoking, alcohol and substance abuse. The 110 questionnaire consisted predominantly of quantitative questions requiring a yes/no answer. 111 Individuals were also asked about their appetite, meal frequency and the amount of money spent 112 on food using questions, as per previous studies on nutritional status and homelessness (7). 113 Physiological measurements 114 A stadiometer (Seca 213, Seca medical measuring systems) and calibrated electrical scales 115 (Seca 877, Seca medical measuring systems) were used to measure height and weight 116 respectively using standard operating procedures. BMI was calculated as weight (kg)/ height (m2) 117 and classified in accordance with the WHO guidelines (25). The mean of three handgrip strength 118 measurements of the participant’s dominant hand using a hand-held dynamometer (Takei 5001, 119 Takei Scientific Instruments Co.) provided a non-invasive measure of general muscle strength (26). 120 Blood pressure measurements were taken in triplicate using an automated blood pressure monitor 121 (M10-IT, Omron healthcare Ltd.). In order to assess CVD risk, the online ‘QRISK 2-2015 Web 122 Calculator’ was used to estimate the 10-year risk of developing CVD (27). 123 Statistical analysis 124 The study was powered using previous comparison of energy intake (kJ) in homeless and 125 domiciled male youths in Toronto (28). Using G*Power (29), it was estimated that 68 participants 126 would be required in each group for a 2385kJ energy difference with s SD of 6408kJ (α level 127 P=0.05, 1-β power 0.85). Estimated samples sizes based on differences in total fat (g), protein (g), 128 vitamin B6 (mg) and calcium (mg) were lower (range n=12-32 per group). To allow for a 10% 129 dropout or incomplete data collection a total of 75 participants were recruited per group. 130 Means and standard deviations were used to describe parametric distributed data, and medians 131 and 95% Confidence Intervals (CI) for non-parametric distributed data. Counts and percentages 132 were used for categorical variables. Homeless and housed groups were broadly matched for their 133 gender, age range and ethnic category during the data collection stage. Data were checked for 134 normality of distribution, and where possible skewed variables were transformed using log10. 135 6 Parametric data were analysed using general linear models (GLM) and non-parametric using 136 Independent samples Mann-Whitney U tests for the comparison of continuous variables. For 137 categorical variables chi-squared tests were used to assess for differences between the two 138 groups. P<0.05 was classified as significant. Data were analysed using SPSS Statistics 21.0 (IBM, 139 UK). 140 141 Results 142 Participant characteristics 143 Demographic information of the homeless (n=75) and housed (n=75) groups is shown in Table 144 1. The mean age was 38 (SD 11) years (range 19-59 years) for the homeless and 38 (SD 11) 145 years for the housed participants (range 20-59 years). Ethnicity and gender distributions were 146 matched between groups. With regards to education, there was a significant difference in 147 attainment between groups (P<0.001); whilst the majority of homeless individuals had achieved 148 secondary education or lower education (O-Level’s/GCSE’s and primary education), the majority of 149 the housed group had attained above secondary level with 25% (n=19) reporting higher degrees 150 compared to 4% (n=3) in the homeless group. Homeless individuals consisted of those sleeping 151 rough on the street (n=24) or residing in Hamble Court (n=22) or Salvation Army (n=29) hostels. All 152 housed individuals lived in private sector accommodation consisting of rented, mortgaged or 153 owned property. 154 Responses of both groups to questions regarding reward mediated behaviour, meal 155 consumption and cooking facilities are shown in Table 2. Significantly more homeless compared to 156 housed individuals reported that they smoked (P<0.001) and/or had taken illicit substances within 157 the preceding month (P<0.001). There was no significant difference between the number of 158 individuals who reported consuming alcohol, although significantly more of the homeless (35%, 159 n=18) compared to housed (8%, n=5) individuals who consumed alcohol reported intakes above 160 the recommended weekly intake of 14 units (UK) for men and women respectively (P<0.001, data 161 not shown). The majority of homeless individuals reported consuming 1 or 2 meals per day in 162 contrast to the housed that predominantly reported 3 meals daily (P<0.001). In addition, 163 significantly less homeless participants reported having enough to eat, a good appetite and 164 7 cooking facilities (all comparisons, P<0.001). For the homeless individuals reporting cooking 165 facilities, the majority only had access to a microwave in a communal living space whereas all 166 housed participants reported a full kitchen in their accommodation. A greater proportion of 167 homeless individuals reported receiving less than £150 ($200) and spending less than £50 ($67) 168 on food per week compared to housed individuals (all comparisons, P<0.001). 169 Physiological and psychological characteristics 170 There was no significant difference in BMI between the two groups as shown in Table 3. 171 However, a greater number of homeless (66.6%) had a BMI<24.9kg/m2, with 5.3% classified as 172 underweight (BMI <18.5m2), whereas half (50.6%) of the housed group were classified as 173 overweight and obese (BMI>25kg/m2) and none as underweight. Despite this 4% more homeless 174 adults (21.3% vs. 17.3% for housed) were also classified as obese (BMI>30kg/m2). The homeless 175 had a significantly higher mean diastolic blood pressure (DBP) (P=0.008) and mean QRISK-2 176 score (P=0.009) compared to the housed, with no significant difference in systolic BP (SBP) or 177 handgrip strength. 178 Homeless individuals scored significantly higher than the housed group for the presence of 179 somatic (PHQ-15), anxiety (GAD-7) and depressive (PHQ-9) symptoms (all comparisons, 180 P<0.001) with a mean classification of ‘mild’ (score range 5-9) for each condition (Table 3). In total, 181 24% (n=18) of homeless adults reported mental illness diagnoses (depression, n=11; 182 schizophrenia, n=1, multiple diagnoses, n=5; undisclosed diagnosis, n=1) and 4% (n=3) of housed 183 adults (all depression). 184 When comparing street homeless (n=24) with first-stage living hostel residents (n=51), no 185 significant differences were observed for weight, BMI, SBP, DBP or PHQ-SADS scores (Table 6). 186 There was a trend for higher handgrip strength in street homeless participants compared to first-187 stage living hostel residents (P=0.058), although the difference failed to reach significance. Mean 188 duration of street homelessness was 5.4 (SD 6.8) months and hostel residency 9 months (SD 9.3). 189 Nutritional intake 190 Homeless individuals reported a significantly higher mean daily intake of total fat (P=0.049), 191 SFA (P=0.002), MUFA (P=0.026) and alcohol (P<0.001), as a percentage of energy intakes, 192 compared to the housed group (Table 4). In contrast, carbohydrate (P<0.001) and protein 193 8 (P=0.011) accounted for a significantly lower percentage of energy in the homeless group. Mean 194 daily intake of non-starch polysaccharides (NSP) was significantly lower in homeless compared to 195 homed individuals (P<0.001). Further comparison of daily NSP intake with the UK recommended 196 level of 18g (30) highlighted that the majority of homeless (n=58, 77%) and homed (n=46, 61%) 197 individuals had an intake below 18g (data not shown). Removal of over-reporters (n=2, homeless 198 adults) did not alter the statistical findings (data not shown). Whilst total energy intake did not differ 199 between street homeless and hostel residents, mean SFA intake (%TE) was significantly higher for 200 hostel residents (P=0.010). 201 Micronutrient data (Table 4) demonstrated a significantly higher mean daily intake of salt in the 202 homeless compared to housed group (P=0.014). In contrast, vitamin C intake was significantly 203 lower in the homeless compared to housed (P=<0.001). Daily intake for the majority of individuals 204 in both the homeless and homed groups was found to meet or exceed the RNI (32) for most of the 205 micronutrients measured including vitamin C. In contrast, 58 (77%) homeless and 54 (72%) homed 206 individuals had below the LRNI for selenium (Figure 1). Approximately half of the homeless (n=39, 207 52%) and homed (n=38, 51%) groups failed to meet the zinc LRNI. Although the majority of 208 homeless (n=40, 53%) and homed (n=41, 55%) individuals met the iron LRNI, these were 209 predominantly men. Consequently for women, 13 out of 15 homeless and all of the 15 women in 210 the homed group failed to reach the iron LRNI of 14.8mg (data not shown). In contrast to iron, the 211 majority of homeless (n=61, 81%) and homed individuals (n=67, 89%) reported a sodium intake 212 above the LRNI of 1600mg. Of these individuals 45 (60%) homeless and 42 (56%) homed 213 consumed above the recommended maximum salt level of 6g. 214 Division of FFQ data into food groups is shown in Table 5. The mean daily homeless diet 215 consisted of significantly higher amounts of alcoholic beverages (P<0.001), fats/oils (P=0.023), 216 meat and meat products (P=0.037) and potatoes (P=0.035). In contrast, the homeless compared to 217 homed diet was composed of a significantly lower amount of fruit and nuts and seeds (P’s<0.001), 218 and vegetables (P=0.022). Removal of individuals reporting mental health diagnoses (n=21) 219 resulted in a loss of significant difference in intake of fats/oils (P=0.18) between the groups; no 220 other findings were altered. 221 9 Calcium, iodine and riboflavin intakes were all significantly lower in street homeless compared 222 with first-stage living hostel residents (P<0.05) (Table 6). Despite this, hostel residents were found 223 to consume significantly greater quantities of milk (P=0.001) and potato (P=0.012), and less soups 224 and sauces (P=0.047). There was also a trend for greater sugary snack consumption in hostel 225 residents (P=0.052). 226 As a sensitivity analysis, data analysis was repeated in males only (n=120) and in 227 participants reporting ‘white’ ethnicity (n=122). The identified significance differences were similar 228 following removal of females, although just a tendency for a lower vitamin B6 intake in the 229 homeless was observed (P=0.078). Analysis in only white participants led to an additional 230 significant difference for PUFA (5.51% ± 1.41 homeless, 5.91% ± 1.36 housed, P=0.031). 231 232 Discussion 233 The present study compared dietary intake, nutritional status and mental wellbeing of homeless 234 and housed adults in Reading. Our findings suggest that homeless adults have a higher risk of 235 cardiovascular disease and incidence of anxiety and depressive symptoms, and poorer dietary and 236 nutrient intake than housed adults. Homeless diets were characterised by high consumption of 237 meat and meat product, fats and oils and alcoholic beverages, and significantly lower intakes of 238 fruits, vegetables, nuts and seed than housed comparators. Street homeless were at particular 239 risk of calcium and iodine deficiency, and had a significantly lower intake of milk and milk products 240 than hostel residents. 241 Whilst no significant difference in energy intake was observed between homeless and housed 242 adults, 27% of homeless reported not having ‘enough to eat’ and 38% reporting having ≤ one 243 meals per day. There was also a trend (P=0.080) for a lower BMI in the homeless group. In the 244 present study, both homeless and housed intakes of total fat and SFA exceeded the UK 245 recommended intakes (total fat, 34% total energy; SFA, 10% total energy)(30). However, homeless 246 adults reported significantly a higher intake of these fats, as observed previously in the homeless 247 community(7; 31). This may be attributed to their greater intakes of meat and meat products (e.g. 248 sausages, minced beef and processed sliced meat) and fats and oils (e.g. butter). SFA intake was 249 10 also significantly higher in hostel residents than street homeless, which supports previous data that 250 charitable meal provision is weighted towards sugar and fat energy (32). Homeless adults reported a 251 significantly lower intake of carbohydrate and protein derived energy. 252 Englyst NSP intake was below the recommended daily intake of 18g/day (30) in both housed 253 and homeless groups. However, significantly lower intakes of NSP were reported in the homeless 254 group (no difference between street homeless and first-stage living hostel residents), which may be 255 due to their lower fruit and vegetable intake. A diet low in fruit, vegetables and fibre has been 256 reported previously in the homeless community (3-5). In the present study, a greater disparity in fruit 257 intake between homeless and housed participants than vegetables was observed (170% vs.19% 258 higher in housed respectively); this may reflect the type of meals (hot meals including vegetables 259 (36)) available to the homeless population and lack of fresh fruit provided. In line with a low fruit 260 intake, a significantly lower intake of vitamin C was observed in the homeless, supporting previous 261 studies (6; 33). However the majority of homeless individuals still met or exceeded the daily vitamin C 262 RNI of 40mg. 263 Intakes of calcium, iodine and riboflavin were significantly lower in street homeless compared 264 with hostel residents, who consumed significantly more milk and milk products and potato. This 265 may be due to hostel residents having access to cold food storage facilities and regular cooked 266 meals, which has been associated with nutritional advantages in the US (34). Inadequate calcium 267 intakes have been observed previously in UK single homeless adults(35). 268 Alcohol was a significant source of energy in the homeless group, as reported previously (6). 269 Furthermore, a greater percentage of homeless had B vitamin intakes below the LRNI and, given 270 that chronic alcohol use is associated with malabsorption and reduced utilization of B vitamins (36), 271 this is likely to be underestimated. Early clinical thiamin (vitamin B1) deficiency, which causes the 272 alcohol–linked neurological disorder Wernicke–Korsakoff syndrome, has been observed previously 273 in homeless men (37) and prophylactic oral thiamine is advised for harmful or dependent drinkers at 274 risk of malnutrition (38). 275 The significantly higher salt intake in the present homeless population represents an 276 established risk factor for the development of hypertension (39) although, despite a significantly 277 higher diastolic level in the homeless group, mean blood pressure measurements were within the 278 11 normal range (40). However, the significantly higher QRISK-2 score in the homeless group indicates 279 that the homeless group are at a greater risk of developing CVD within the next 10 years. Hand 280 grip strength, a low value of which has been associated with increased mortality in adults > 50 281 years (26; 41), was significantly greater in street-homeless compared to hostel residents; although 282 this is likely to be most reflective of increased physical activity. Significantly more homeless 283 compared to housed reported smoking and substance misuse in the present study, as documented 284 previously (4; 12). Furthermore, a significantly greater number of homeless that consumed alcohol 285 reported an intake above recommended levels, which is consistent with previous data (6). 286 Homeless adults had significantly higher scores for each PHQ-SADS component compared to 287 housed group, which corresponds with the high levels of mental illness reported in the homeless 288 community versus the general population (12). Within the homeless community, street sleepers are 289 more likely to experience depression (42), as observed in the present study whereby street 290 homeless had higher scores for the depressive component (PHQ-9) of the PHQ-SADS compared 291 with hostel residents. It is currently unclear as to whether mental illness precedes homelessness or 292 homelessness induces/ exacerbates the occurrence of mental illness and the role, if any, nutrition 293 has to play in these conditions. The higher numbers of homeless compared to housed reporting a 294 poor appetite, in the presence of the increased levels of mental illness, may reflect the depressive 295 influence of mental conditions on appetite (43), which warrants further investigation. 296 The current study has a number of limitations. The high male to female ratio is consistent 297 with other studies and reflects the preponderance of males in the homeless population (12). 298 However, male dominance and potential selection bias due to reliance on services accessed by 299 the homeless to attain participants limit the generalizability of the results (44). In addition, 300 comparison with the most recent (2011) Census in Reading (74.8% white)(45), suggests that white 301 individuals may have been over-represented in this sample (81% white). The significant difference 302 between the educational status of the homeless and housed groups may represent an uncontrolled 303 confounding factor given that higher educational status has been associated with a ‘healthier’ diet 304 (46). The EPIC FFQ has been validated for the assessment of nutrient intake in different populations 305 (22; 47), is less burdensome that weighed intake dairies and was consequently considered 306 appropriate for the current research. However, due to the transient nature of the homeless 307 12 population, participants were asked to report dietary intake over the previous month (i.e. shorter-308 term intake), which may have been challenging individuals with fluid dietary patterns. Memory 309 recall may be further confounded in the homeless community whereby greater incidences of 310 reward mediated behaviour, mental illness and alcohol related brain damage are reported. 311 Objective assessment of energy expenditure, food intake and nutritional status, using biomarkers, 312 would help to confirm the observed differences. Finally, the grouping of hostels may be 313 confounding due to differences in storage facilities and the provision of food. For example, whilst 314 breakfast and dinner were provided by the charity-led hostel, residents in the council-led hostel 315 were self-catered. Further analysis regarding the impact of meal provision on nutritional status in 316 first-stage living hostels is therefore warranted. 317 The often limited and infrequent access to food by homeless individuals means that the 318 provision of nutritionally sufficient meals is of utmost importance. However, determining which 319 nutritional issues are specific to a homeless community is required in order to determine suitable 320 intervention strategies. Previous studies have aimed to address poor dietary intake in homeless 321 populations through recipe modification at food aid organisations (32) and implementation of 322 educational programs (21; 31; 48). Decreasing the total and SFA content of meals and increasing fruit 323 availability in the hostels surveyed would help to address some of the issues identified in Reading. 324 Milk supplementation in street-homeless adults could also help to address calcium, iodine and 325 riboflavin insufficiencies. Whilst beyond the scope of this study, exploration of Food Bank usage, 326 which has increased in the UK (49), may also assist in the identification of suitable interventions for 327 the local area. 328 The findings of this study highlight the vulnerability of homeless adults in Reading, who have 329 reduced mental wellbeing, a higher risk of CVD and a poorer dietary intake compared with the 330 housed population. Further objective data is warranted, but the results clearly highlight the need for 331 intervention aimed at improving mental wellbeing and nutritional status in this group. 332 333 Acknowledgments 334 The authors would like to thank Matt Farrow (Crime Reduction Initiatives), Laura Carey (the 335 Salvation Army, Willow House) and the Churches in Reading drop in centre for assistance in the 336 13 recruitment of homeless participants and Sarah Hargreaves (Hugh Sinclair Unit of Human 337 Nutrition) for assisting in the recruitment of housed participants. 338 339 Financial Support 340 This research received no specific grant from any funding agency, commercial or not-for-profit 341 sectors. 342 343 Conflict of Interest 344 None 345 346 Authorship 347 JAL and RF designed the research protocol; RF, JS, CM and LA collected homeless data; JS 348 collected housed data; JS and RF analysed data and drafted the manuscript. All authors have read 349 and approved the final manuscript. This research received no specific grant from any funding 350 agency, commercial or not-for-profit sectors. The authors have no conflicts of interest to declare.351 1 REFERENCES 1. Department for Communities and Local Government (2016) Statutory homelessness: October to December Quarter 2015 England. https://www.gov.uk/government/uploads/system/uploads/attachment_data/file/509763/Statut ory_Homelessness_Statistical_Release_October_to_December_2015.pdf (accessed June 2016) 2. Department for Communities and Local Government (2016) Rough Sleeping Statistics: Autumn 2015 England https://www.gov.uk/government/uploads/system/uploads/attachment_data/file/503015/Rough _Sleeping_Autumn_2015_statistical_release.pdf (accessed June 2016) 3. Rushton CM, Wheeler E (1993) The dietary intake of homeless males sleeping rough in Central London. J Hum Nutr Diet 6, 443-456. 4. Langnäse K, Müller MJ (2001) Nutrition and health in an adult urban homeless population in Germany. Public Health Nutr 4, 805-811. 5. Hickey C, Downey D (2003) Hungry for Change: Social Exclusion, Food Poverty and Homelessness in Dublin; a Pilot Research Study: Focus Ireland. 6. Darmon N, Coupel J, Deheeger M et al. (2001) Dietary inadequacies observed in homeless men visiting an emergency night shelter in Paris. Public Health Nutr 4, 155-161. 7. Luder E, Boey E, Buchalter B et al. (1989) Assessment of the nutritional status of urban homeless adults. Public Health Rep 104, 451. 8. Wolgemuth JC, Myers-Williams C, Johnson P et al. (1992) Wasting malnutrition and inadequate nutrient intakes identified in a multiethnic homeless population. J Am Diet Assoc 92, 834-839. 9. Koh KA, Hoy JS, O’Connell JJ et al. (2012) The hunger–obesity paradox: obesity in the homeless. J Urban Health 89, 952-964. 10. Tsai J, Rosenheck RA (2013) Obesity among chronically homeless adults: is it a problem? Public Health Rep 128, 29. 11. Hwang SW (2001) Homelessness and Health. Can Med Assoc J 164. 2 12. Homeless Link (2015) The unhealthy state of homelessness: health audit results 2014. 13. Kushel MB, Vittinghoff E, Haas JS (2001) Factors associated with the health care utilization of homeless persons. JAMA 285, 200-206. 14. CRISIS (2011) Homelessness: A silent killer https://www.crisis.org.uk/media/237321/crisis_homelessness_a_silent_killer_2011.pdf (accessed July 2017) 15. Hibbs JR, Benner L, Klugman L et al. (1994) Mortality in a cohort of homeless adults in Philadelphia. N Engl J Med 331, 304-309. 16. Hwang SW, Orav EJ, O'Connell JJ et al. (1997) Causes of death in homeless adults in Boston. Ann Intern Med 126, 625-628. 17. Krausz RM, Clarkson AF, Strehlau V et al. (2013) Mental disorder, service use, and barriers to care among 500 homeless people in 3 different urban settings. Soc Psychiatry Psychiatr Epidemiol 48, 1235-1243. 18. Lykkesfeldt J, Christen S, Wallock LM et al. (2000) Ascorbate is depleted by smoking and repleted by moderate supplementation: a study in male smokers and nonsmokers with matched dietary antioxidant intakes. Am J Clin Nutr 71, 530-536. 19. German Nutrition Society (DGE) (2015) New Reference Values for Vitamin C Intake. Ann Nutr Metab 67, 13-20. 20. Kubisová D, Adámková V, Lánská V et al. (2007) Higher prevalence of smoking and lower BMI, waist circumference, cholesterol and triacylglyceride levels in Prague's homeless compared to a majority of the Czech population. BMC Public Health 7, 1. 21. Sprake EF, Russell JM, Barker ME (2014) Food choice and nutrient intake amongst homeless people. J Hum Nutr Diet 27, 242-250. 22. Bingham SA, Welch AA, McTaggart A et al. (2001) Nutritional methods in the European prospective investigation of cancer in Norfolk. Public Health Nutr 4, 847-858. 23. Mulligan AA, Luben RN, Bhaniani A et al. (2014) A new tool for converting food frequency questionnaire data into nutrient and food group values: FETA research methods and availability. BMJ Open 4, e004503. 3 24. Kroenke K, Spitzer RL, Williams JB et al. (2010) The patient health questionnaire somatic, anxiety, and depressive symptom scales: a systematic review. Gen Hosp Psychiatry 32, 345-359. 25. World Health Organisation (WHO) expert committee (1995) Physical status: the use and interpretation of anthropometry. Geneva: World Health Organization. 26. Lauretani F, Russo CR, Bandinelli S et al. (2003) Age-associated changes in skeletal muscles and their effect on mobility: an operational diagnosis of sarcopenia. J Appl Physiol 95, 1851-1860. 27. Hippisley-Cox J, Coupland C, Vinogradova Y et al. (2008) Predicting cardiovascular risk in England and Wales: prospective derivation and validation of QRISK2. BMJ 336, 1475- 1482. 28. Tarasuk V, Dachner N, Li J (2005) Homeless youth in Toronto are nutritionally vulnerable. J Nutr 135, 1926-1933. 29. Faul F, Erdfelder E, Lang A-G et al. (2007) G* Power 3: A flexible statistical power analysis program for the social, behavioral, and biomedical sciences. Behav Res Methods 39, 175-191. 30. The Committee on Medical Aspects of Food and Nutrition Policy (COMA) (1991) Dietary reference values for food energy and nutrients: report of the panel on dietary reference values of the committee on medical aspects of food policy: HMSO, London. 31. Food Standards Agency (2006) Research into food poverty and homelessness in Northern Ireland - final report. 32. Pelham-Burn SE, Frost CJ, Russell JM et al. (2014) Improving the nutritional quality of charitable meals for homeless and vulnerable adults. A case study of food provision by a food aid organisation in the UK. Appetite 82, 131-137. 33. Malmauret L, Leblanc J, Cuvelier I et al. (2002) Dietary intakes and vitamin status of a sample of homeless people in Paris. Eur J Clin Nutr 56, 313. 4 34. Wiecha JL, Dywer JT, Jacques PF et al. (1993) Nutritional and economic advantages for homeless families in shelters providing kitchen facilities and food. J Am Diet Assoc 93, 777- 783. 35. Evans NS, Dowler EA (1999) Food, health and eating among single homeless and marginalized people in London. J Hum Nutr Diet 12, 179-199. 36. Hoyumpa A (1980) Mechanisms of thiamin deficiency in chronic alcoholism. Am J Clin Nutr 33, 2750-2761. 37. Darnton-Hill I, Truswell A (1990) Thiamin status of a sample of homeless clinic attenders in Sydney. Med J Aus 152, 5-9. 38. National Institute of Clinical Excellence (2010) Alcohol-use disorders: diagnosis and management of physical complications [CG100]. London: NICE publications. 39. Cook NR, Cutler JA, Obarzanek E et al. (2007) Long term effects of dietary sodium reduction on cardiovascular disease outcomes: observational follow-up of the trials of hypertension prevention (TOHP). BMJ 334, 885. 40. National Institute of Clinical Excellence (NICE) (2011) Hypertension: clinical management of primary hypertension in adults. NICE guidelines, CG127. London: NICE publications. 41. Stenholm S, Mehta NK, Elo IT et al. (2014) Obesity and muscle strength as long-term determinants of all-cause mortality—a 33-year follow-up of the Mini-Finland Health Examination Survey. Int J Obes 38, 1126-1132. 42. La Gory M, Ritchey FJ, Mullis J (1990) Depression among the homeless. J Health Soc Behav, 87-102. 43. Konttinen H, Männistö S, Sarlio-Lähteenkorva S et al. (2010) Emotional eating, depressive symptoms and self-reported food consumption. A population-based study. Appetite 54, 473-479. 44. Faugier J, Sargeant M (1997) Sampling hard to reach populations. J Adv Nurs 26, 790- 797. 45. Office for National Statistics (2012) 2011 Census: Key Statistics for Wales, March 2011 - Ethnic Group and Identity 5 https://www.ons.gov.uk/peoplepopulationandcommunity/populationandmigration/populatione stimates/bulletins/2011censuskeystatisticsforwales/2012-12-11#ethnic-group-and-identity (accessed June 2017) 46. Friel S, Kelleher C, Nolan G et al. (2003) Social diversity of Irish adults nutritional intake. Eur J Clin Nutr 57, 865-875. 47. McKeown NM, Day NE, Welch AA et al. (2001) Use of biological markers to validate self- reported dietary intake in a random sample of the European Prospective Investigation into Cancer United Kingdom Norfolk cohort. Am J Clin Nutr 74, 188-196. 48. Johnson LJ, Myung E, McCool AC et al. (2009) Nutrition Education for Homeless Women—Challenges and Opportunities: A Pilot Study. J Foodservice Bus Res 12, 155-169. 49. Loopstra R, Reeves A, Taylor-Robinson D et al. (2015) Austerity, sanctions, and the rise of food banks in the UK. BMJ 350, h1775. 1 Table 1: Demographic characteristics of homeless (n=75) and housed (n=75) adults Characteristics Homeless Housed n % n % Gender Males 60 80 60 80 Females 15 20 15 20 Ethnicity White 61 81 61 81 Mixed 7 9 7 9 Indian/Pakistani/Bangladeshi 2 3 2 3 Black/African/Caribbean 5 7 5 7 Education Secondary education or below 64 85 12 16 Above secondary education 11 15 63 84 Housing status Living on the street 24 32 0 0 Living in a hostel 51 68 0 0 Living in private sector accommodation 0 0 75 100 2 Table 2: Responses by homeless (n=75) and housed (n=75) groups to behavioural questions1 1Data were analysed using chi-square tests comparing homeless and housed responses on each characteristic 2Analysed using Fisher’s Exact test with Freeman-Halton extension for contingency tables greater than 2x2 Behavioural questions Subjects responding ‘yes’ P value Homeless Housed n % n % Do you smoke?2 71 95 2 3 <0.001 Do you drink alcohol? 52 69 61 81 0.070 Do you take illicit drugs?2 46 61 1 1 <0.001 Do you have enough to eat? 55 73 75 100 <0.001 Do you have a good appetite? 51 68 71 95 <0.001 Are there any cooking facilities available for use? 31 41 75 100 <0.001 How many meals do you have per day?2 <0.001 0 2 2 0 0 1 27 36 0 0 2 35 47 21 28 3 11 15 54 72 How much money do you receive per week?2 <0.001 <£50 23 31 0 0 £50-£149 45 60 12 16 >£150 7 9 63 84 How much money do you use to buy food per week?2 <0.001 <£20 53 70 4 5 £20-49 20 27 37 49 >£50 2 3 34 46 1 Table 3: Physiological characteristics, PHQ-9 SADS and QRISK-2 scores for homeless (n=75) and housed (n=75) adults1 Characteristics Homeless Housed P value Mean SD Mean SD Weight, kg 73.3 15.4 77.5 14.6 0.19 Height, m 1.74 9.0 1.73 8.8 0.63 BMI, kg/m2 24.5 5.7 25.8 4.2 0.08 Underweight (<18.49kg/m2, %) 5.3 0.0 - Healthy (18.5-24.9kg/m2, %) 61.3 48.0 - Overweight (25-29.9kg/m2, %) 12.0 33.3 - Obese (>30kg/m2, %) 21.3 17.3 - Systolic blood pressure, mm Hg 125.7 17.9 124.2 12.1 0.70 Diastolic blood pressure, mm Hg 78.2 12.0 73.6 8.2 0.008 Hand grip strength, kg 36.4 8.4 37.5 9.2 0.60 QRISK-2 Score (%)2 5.1 6.2 2.7 4.0 0.009 GAD-73 6.0 5.9, 9.1 2.5 1,7, 3.3 <0.001 PHQ-93 7.0 7.3, 10.9 2.0 1.9, 3.3 <0.001 PHQ-153 6.0 5.5, 7.7 3.0 2.7, 3.9 <0.001 1Data were analysed using independent t-tests. PHQ-15, patient health questionnaire-15 for somatic symptoms; GAD-7, general anxiety disorders-7 for anxiety symptoms; PHQ-9, patient health questionnaire-9 for depressive symptoms. 2Estimated risk of developing CVD over the next 10 years. 3Values are medians (95% CI), data analysed using Independent samples Mann-Whitney U Test. 1 Table 4: FFQ derived daily energy and nutrient intake for homeless (n=75) and housed adults (n=75)1 1Values are means ± SDs, homeless (n=75) and housed (n =75). Data were analysed using independent t-tests. NSP, non-starch polysaccharide; %TE, percentage of total energy intake. Nutrient Homeless Housed P value Mean SD Mean SD Energy, kcal 2140 1121 1848 471 0.38 Energy, kJ 8988 4700 7741 2016 0.39 Total fat, % TE 37.2 6.3 34.9 5.3 0.049 SFA, % TE 14.6 3.1 13.0 3.3 0.002 MUFA, % TE 13.9 2.6 13.0 2.2 0.026 PUFA, % TE 5.5 1.4 5.9 1.4 0.08 Cholesterol, mg 357 204 274 114 0.020 Total protein, % TE 16.7 4.0 18.2 3.5 0.011 Total carbohydrate, % TE 43.4 9.3 48.1 6.9 <0.001 Total Sugars, %TE 5.4 2.6 4.4 1.2 0.009 Englyst Fibre (NSP), g 13.4 7.9 16.3 6.4 <0.001 Alcohol, g 17.5 30.7 5.0 8.1 <0.001 Alcohol, % TE 5.3 7.7 1.9 2.6 <0.001 Calcium, mg 977 537 942 323 0.77 Iron, mg 11.0 6.1 10.9 3.0 0.18 Total folate, mcg 288 173 281 92 0.25 Iodine, mcg 148 78 146 44 0.35 Sodium, mg 3186 1974 2573 764 0.17 Salt, g 8.0 4.9 6.4 1.9 0.014 Niacin, mg 22.8 11.9 22.5 6.0 0.21 Selenium, mcg 60.7 42.0 61.2 19.2 0.083 Vitamin A, mcg 1491 2107 1122 1252 0.85 Thiamin, mg 1.5 0.8 1.5 0.4 0.24 Riboflavin, mg 2.2 1.4 2.0 0.7 0.90 Vitamin B6, mg 2.1 1.0 2.2 0.6 0.032 Vitamin B12, mcg 8.4 9.1 6.4 4.5 0.58 Vitamin C, mg 78.8 58.9 109.4 62.5 <0.001 Vitamin D, mcg 3.5 3.6 3.0 1.7 0.74 Vitamin E, mg 12.1 7.0 11.8 4.2 0.34 Zinc, mg 9.7 4.9 9.3 2.4 0.53 1 Table 5: Daily intake of the 14 food groups derived from FFQ analysis for homeless (n=75) and housed (n=75) adults1 Food group Homeless Housed P value Mean SD Mean SD Alcoholic beverages, g 363 593 93.5 185 <0.001 Cereals and cereal products, g 235 178 240 109 0.076 Eggs and egg dishes, g 20.6 21.2 19.0 17.9 0.61 Fats and oils, g 23.6 20.5 16.0 11.2 0.023 Fish and fish products, g 41.2 63.7 40.3 27.8 0.052 Fruit, g 96 107 260 224 <0.001 Meat and meat products, g 157 109 111 54 0.037 Milk and milk products, g 400 241 385 198 0.80 Non-alcoholic beverages, g 790 710 710 438 0.83 Nuts and seeds, g 3.4 7.4 9.9 15.0 <0.001 Potatoes, g 94.0 67.0 66.3 50.3 0.035 Soups and sauces, g 61.4 61.1 56.1 55.3 0.76 Sugars; preserves and snacks, g 43.3 46.4 39.6 32.0 0.96 Vegetables, g 205 156 244 149 0.022 1Data analysed using GLM. 2 Table 6: Subject characteristics and nutritional intake for street homeless (n=24) and first- stage living hostel residents (n=51)1 Characteristic Street homeless Hostel residents P-value Mean SD Mean SD Gender (m/f) 21/3 - 39/12 - - Age, years 38 11 38 11 0.99 Weight, kg 74.7 16.0 73.9 16.7 0.78 BMI, kg/m-2 23.9 4.6 25.1 6.3 0.42 SBP 130.1 17.1 123.9 17.7 0.15 DBP 81.4 9.5 76.9 12.6 0.096 Hand-grip 39.7 8.2 35.3 8.3 0.058 GAD-72 6.0 5.1, 12.6 6.0 5.3, 8.4 0.84 PHQ-92 8.0 7.2, 15.2 7.0 6.2, 10.1 0.27 PHQ-152 6.0 4.8, 9.9 6.0 5.2, 7.4 0.83 Energy, kcal 2008 1388 2202 979 0.13 Energy, kJ 8428 5814 9251 4114 0.13 Fat, %TE 36.6 7.0 37.5 6.0 0.55 SFA, %TE 13.7 3.1 15.0 3.1 0.010 MUFA, %TE 14.0 2.9 13.8 2.5 0.86 PUFA, %TE 5.7 1.6 5.4 1.3 0.56 Protein, %TE 17.1 5.0 16.6 3.5 0.89 CHO, %TE 41.5 11.2 44.5 8.2 0.19 Sugars, g 94.1 60.8 121.8 70.3 0.15 NSP, g 13.2 9.5 13.5 7.2 0.39 Alcohol, g 23.9 40.2 14.5 25.0 0.39 Calcium, mg 858 707 1032 433 0.027 Iron, mg 10.8 7.8 11.0 5.3 0.41 Total folate, mcg 226 167 304 174 0.10 Iodine, mcg 128 91 157 73 0.033 Sodium, mg 3198 2531 3180 1680 0.42 Salt, g 8.0 6.3 7.9 4.2 0.41 Niacin, mg 22.3 13.8 23.0 11.0 0.43 Selenium, mcg 62.3 49.8 59.9 38.2 0.63 Vitamin A, mcg 1252 1647 1604 1647 0.18 Thiamin, mg 1.34 0.86 1.55 0.84 0.10 Riboflavin, mg 1.79 1.22 2.42 1.45 0.012 Vitamin B6, mg 1.89 1.13 2.20 0.99 0.10 Vitamin B12, mcg 7.3 7.5 9.0 9.8 0.13 Vitamin C, mg 74.6 65.3 80.8 56.2 0.23 Vitamin D, mcg 3.2 2.6 3.7 4.0 0.22 Vitamin E, mg 11.4 7.8 12.4 6.6 0.25 Zinc, mg 9.6 6.4 9.8 4.2 0.35 Cereal and cereal products (g/day) 229 225 238 154 0.22 Egg and egg dishes (g/day) 17.4 16.4 22.1 23.1 0.36 3 Fats and oils (g/day) 22.1 23.8 24.4 19.0 0.16 Fish and fish products (g/day) 30.4 30.3 46.4 74.1 0.16 Fruit (g/day) 97 131 95 96 0.87 Meat and meat products (g/day) 164 137 153 95 0.40 Milk and milk products (g/day) 295 226 449 234 0.001 Nuts & seeds (g/day) 5.2 9.3 2.5 6.2 0.08 Potato (g/day) 66.6 43.5 107.4 72.1 0.012 Soups and sauces (g/day) 90.4 78.7 47.7 45.6 0.047 Sugars; preserves and snacks (g/day) 30.8 26.2 49.2 52.5 0.052 Vegetables (g/day) 205 162 205 155 0.71 1Values are means ± SDs, street homeless (n=24) and hostel residents (n =51). Data were analysed using GLM. NSP, non-starch polysaccharide; %TE, percentage of total energy intake. 2Data are medians (95% CI), analysed using Independent Samples Mann-Whitney U Tests. 1 Figure 1: Homeless and housed individuals with daily intake below LRNI for each micronutrient. Values are percentages (%) of individuals who did not meet the daily RNI for each micronutrient, homeless (n=75) and housed (n=75)