Questionnaire data

FFQ have been used to assess nutrient intake in the homeless. This questionnaire-based method consists of participants self-reporting their frequency and degree of consumption for a variety of food and drink items. For example, the European Prospective Investigation into Cancer and Nutrition (EPIC) Norfolk FFQ has been validated for use in individuals following a ‘traditional’ UK diet⁽ Reference Sprake, Russell and Barker ^{5 ,} Reference Bingham, Welch and McTaggart ^{18 )}. Consequently, the FFQ provides a means of attaining detailed information regarding food and drink consumed over, typically, the preceding 6 months or year⁽ Reference Cade, Burley and Warm ^{19 )}. The ease of administration and low cost makes the FFQ an ideal tool to use in an often transient population who may only be willing or able to provide short-term participation in a study. In Germany, FFQ data revealed a reduced consumption of fruit and vegetables in homeless adults (n 75) when compared with data from governmental dietary surveys of the general population⁽ Reference Langnäse and Müller ^{11 )}. Similar findings regarding low fruit and vegetable intake were found following a survey of male rough sleepers in the UK (n 79)⁽ Reference Rushton and Wheeler ^{14 )} and homeless individuals staying in temporary accommodation in Ireland (n 72)⁽ Reference Hickey and Downey ^{8 )}. Such findings demonstrate consistency in the results from FFQ between countries and highlight deviation from current recommendations of a ‘healthy diet’^{( 20 )}. With regard to micronutrient intake, the administration of the FFQ to seventy-one homeless families in Boston reported low levels of vitamin B₆, Ca and Fe when compared with the recommended daily allowances⁽ Reference Wiecha, Dwyer and Jacques ^{15 )}. However, the study consisted of families temporarily residing in hotels or shelters and so limits the extent to which such findings are reflective of the single homeless individual. Additionally, recall and social approval bias⁽ Reference Coulston, Boushey and Ferruzzi ^{21 ,} Reference Miller, Abdel-Maksoud and Crane ^{22 )} issues of under-reporting⁽ Reference Neuhouser, Tinker and Shaw ^{23 )} and a poor to moderate⁽ Reference McKeown, Day and Welch ^{24 –} Reference Kroke, Klipstein-Grobusch and Voss ^{27 )} correlation with biochemical markers are considered to reduce the accuracy of FFQ data.

As an alternative to the FFQ, advocates of 24 h dietary recall argue that this method results in a more accurate reflection of dietary intake⁽ Reference Salvador Castell, Serra-Majem and Ribas-Barba ^{28 ,} Reference Carroll, Midthune and Subar ^{29 )}. Consistency in the nutrient intake of homeless individuals from different studies using 24 h recall has been demonstrated and acts to enhance the reliability of such results. For example, Evans & Dowler⁽ Reference Evans and Dowler ^{12 )} reported mean saturated fat intakes above the recommended percentage of energy contribution in a group of London homeless following 24 h dietary recall (n 420), a situation which has also been shown in New York (n 11)⁽ Reference Luder, Boey and Buchalter ^{17 )} using this method. In terms of micronutrients, 24 h recalls have demonstrated insufficient intake of vitamin B₆, Zn and Ca for 277 adults staying in an overnight shelter in Florida⁽ Reference Wolgemuth, Myers-Williams and Johnson ^{16 )}. Despite using a broader definition of homelessness ranging from individuals accessing day centres to those residing in temporary accommodation, Luder et al.’s⁽ Reference Luder, Boey and Buchalter ^{17 )} New York study reported similar findings using 24 h recalls. In contrast, repetitive 24 h dietary recalls for twenty-four homeless individuals in Sheffield found that vitamin B₆ and niacin exceeded the recommended nutrient intake levels⁽ Reference Sprake, Russell and Barker ^{5 )}, a finding which has also been reported in a study using 48 h dietary recall for ninety-seven homeless men in Paris⁽ Reference Darmon, Coupel and Deheeger ^{10 )}. An increased intake of beer by some homeless has been postulated to account for the elevated levels of these B vitamins⁽ Reference Darmon, Coupel and Deheeger ^{10 )}. The potential for bias similar to that found for the FFQ⁽ Reference Hebert, Clemow and Pbert ^{30 ,} Reference Ma, Olendzki and Pagoto ^{31 )} in conjunction with the presence of under-reporting⁽ Reference Subar, Kipnis and Troiano ^{32 ,} Reference Hebert, Ebbeling and Matthews ^{33 )} and over-reporting⁽ Reference Montgomery, Reilly and Jackson ^{34 ,} Reference Bokhof, Buyken and Doğan ^{35 )} highlight limitations associated with the sole use of 24 h recall for assessing the nutritional status of the homeless. Additionally, unlike the FFQ, a single 24 h recall is unable to capture variations in nutrient intake over time⁽ Reference Coulston, Boushey and Ferruzzi ^{21 ,} Reference Ma, Olendzki and Pagoto ^{31 )} which may be of particular relevance in a population whose diet is likely to be inconsistent due to their transient circumstances.

In contrast, diet diaries in which individuals prospectively record all food and beverages consumed over a defined period are considered to provide an accurate reflection of an individual’s current diet⁽ Reference Coulston, Boushey and Ferruzzi ^{21 ,} Reference Day, McKeown and Wong ^{36 )}. However, limitations of this approach have been reported such as the high responder burden, need for literacy and potential reactivity to the procedure, the latter of which can result in a modification by the participant of their diet as a result of recording⁽ Reference Goris and Westerterp ^{37 ,} Reference Biró, Hulshof and Ovesen ^{38 )}. Although diet diaries have been infrequently used to assess nutritional intake in homeless individuals residing in temporary accommodation⁽ Reference Jenkins ^{39 )}, the irregularity of meals and lack of food storage facilities in the street homeless arguably prohibit its use in this group. Consequently, despite their limitations, questionnaires provide a means of attaining, at the very least, an overview of the food and nutrient intake of the homeless community.

Interview data

Interviews have been used to obtain an in-depth understanding of the issues facing the homeless, predominantly with regard to food choice. Interviews are often semi-structured to enable the participant a greater opportunity to express their views whilst maintaining a degree of control over the questioning format⁽ Reference Zohrabi ^{40 )}. A recurrent theme from such qualitative data indicates that homeless individuals aspire to consume a diet containing fruit and vegetables but their food choices are constrained by low income and a lack of storage facilities⁽ Reference Sprake, Russell and Barker ^{5 , 7 )}. For example, semi-structured interviews with twelve homeless men in Sheffield identified limited cooking and storage space as key drivers of their food choice and consequently a barrier to purchasing raw and perishable food items⁽ Reference Sprake, Russell and Barker ^{5 )}. Numerous interviews have also highlighted that, despite the desire for a more balanced diet, any food is preferable to no food⁽ Reference Sprake, Russell and Barker ^{5 , 7 )}. These data are a useful addition to questionnaire data by providing an insight into the beliefs and attitudes of homeless individuals, both of which need to be addressed in any subsequent nutritional intervention. For example, through interview data, the Irish Food Standards Agency found a lack of nutritional and cooking knowledge in their homeless group, resulting in the suggestion for educational programmes to address this gap^{( 7 )}. However, when interpreting interview-based results it is important to acknowledge the potential for interviewer bias and the ‘interviewer effect’ wherein responses from the participant are influenced by their perception of the interviewer⁽ Reference Zohrabi ^{40 )}.

Biochemical data

The biochemical measurement of various nutrients avoids issues of bias inherent in questionnaire techniques and accounts for individual differences in the absorption of nutrients. The invasive nature and practicalities associated with collecting blood from a largely transient homeless population raise issues regarding the use of biochemical methods to assess for biomarkers of nutrient intake. Additionally, although some biochemical measures, for example those used to assess lipid profile⁽ Reference Arab and Akbar ^{41 ,} Reference Lee, Wen and Xiang ^{42 )}, are well established, others, such as serum Ca, are poorly correlated with nutritional status⁽ Reference Weaver ^{43 )}. These factors, in conjunction with the effect that smoking, alcohol and/or substance abuse can have on plasma levels of certain nutrients such as vitamin C⁽ Reference van den Berg, van der Gaag and Hendriks ^{44 ,} Reference Marangon, Herbeth and Lecomte ^{45 )}, potentially limit the use of biochemical data in the homeless. Nevertheless, the use of biochemical methods can provide corroboratory evidence regarding an individual’s reported intake. For example, the finding of elevated serum cholesterol⁽ Reference Luder, Boey and Buchalter ^{17 )} is consistent with increased levels of saturated fat reported for the homeless population using both dietary recall⁽ Reference Evans and Dowler ^{12 )} and FFQ^{( 7 )} techniques. Consequently a mixed methodology can provide a robust means of assessing nutrient intake in the homeless⁽ Reference Malmauret, Leblanc and Cuvelier ^{9 ,} Reference Silliman, Yamanoha and Morrissey ^{13 ,} Reference Luder, Boey and Buchalter ^{17 )}. For example, Malmauret et al. ⁽ Reference Malmauret, Leblanc and Cuvelier ^{9 )} demonstrated low vitamin C levels following both 24 h recall and blood sample analysis.

Physical implications

Elevated levels of acute and chronic disease in the homeless community⁽ Reference Hwang ^{50 –} Reference Hwang ^{53 )}, such as HIV and hepatitis C, both of which have long-term physical and psychological sequelae⁽ Reference Beijer, Wolf and Fazel ^{51 )}, highlight a clear association between homelessness and an increased risk of medical conditions. This is further exemplified by the demonstration of increased visits to emergency departments⁽ Reference Kushel, Perry and Bangsberg ^{54 ,} Reference Kushel, Vittinghoff and Haas ^{55 )} and an average life expectancy in the UK of 42 years in the homeless population^{( 56 )}. Recent statistics demonstrating that 73% of 2590 UK homeless people reported physical health problems, 41% of whom identified a chronic condition, quantifies this situation^{( 57 )}. The importance of diet in maintaining efficient immune functioning⁽ Reference Wintergerst, Maggini and Hornig ^{58 )} and the evidence-based association between disease development and certain dietary components, such as a low salt diet and hypertension⁽ Reference Cook, Cutler and Obarzanek ^{59 ,} Reference Sacks, Svetkey and Vollmer ^{60 )}, indicate diet as a modifiable risk factor. Consequently, in the presence of physiological stresses arising from exposure to harsh environmental conditions, the absence of a nutritionally balanced diet is likely to have a detrimental impact on the health of a homeless individual⁽ Reference Luder, Boey and Buchalter ^{17 )}. It has also been demonstrated that, in the absence of food, individuals are likely to postpone seeking help for medical conditions⁽ Reference Baggett, O’Connell and Singer ^{61 ,} Reference Kushel, Gupta and Gee ^{62 )} and, if payment for prescriptions is required, will often purchase food instead of their medication⁽ Reference Biros, Hoffman and Resch ^{63 ,} Reference Sullivan, Clark and Pallin ^{64 )}. Alternatively, a pre-existing medical condition can make an impact upon nutritional needs which may not be easily met in the homeless situation. The report by Hwang & Bugeja⁽ Reference Hwang and Bugeja ^{65 )} that a limited food choice was considered a ‘barrier’ to adequate glucose control by a group of fifty homeless individuals suffering from diabetes exemplifies the extent to which diet can have an impact on disease management.

In addition to increased levels of disease, skin-fold thickness and muscle mass measurements have frequently demonstrated evidence of ‘wasting’ in the homeless community⁽ Reference Langnäse and Müller ^{11 ,} Reference Wolgemuth, Myers-Williams and Johnson ^{16 ,} Reference Luder, Boey and Buchalter ^{17 )}. In contrast, others have found a proportion of homeless who, according to their BMI, are overweight or obese⁽ Reference Koh, Hoy and O’Connell ^{66 ,} Reference Tsai and Rosenheck ^{67 )}. For example, obesity in a large group of 5632 homeless adults in Boston identified 32·3% as obese (BMI≥30 kg/m²) in comparison with 1·6% who were underweight (BMI≤18·5 kg/m²)⁽ Reference Koh, Hoy and O’Connell ^{66 )}. The similarity of these findings with the BMI distribution of the general US population demonstrates that even a community where food is considered to be restricted is not exempt from the obesity epidemic⁽ Reference Koh, Hoy and O’Connell ^{66 )}. An energy-dense, low-nutrient diet in an era characterised by low-cost convenience foods is postulated to be a reason for this paradoxical situation⁽ Reference Koh, Hoy and O’Connell ^{66 )}. The problems of obesity for the homeless are two-fold. Specifically, numerous medical complications arise from obesity such as an increased risk of CVD and diabetes⁽ Reference Tsai and Rosenheck ^{67 )}, both of which require specific dietary modifications that are often unachievable in the homeless community. Additionally there may be a misperception that an obese homeless individual is not malnourished when in reality they are deficient in micronutrients by virtue of their inadequate diet. The issue of obesity in the homeless not only reinforces the need for a nutritionally balanced diet but demonstrates the importance of assessing dietary intake as opposed to relying on anthropometric measurements when determining nutritional status.

Psychological implications

Studies regarding the homeless have consistently identified high levels of mental illness including depression, anxiety and bipolar disorders^{( 57 ,} Reference Krausz, Clarkson and Strehlau ^{68 ,} Reference Fazel, Khosla and Doll ^{69 )}. The association between depression and low plasma n-3 fatty acids⁽ Reference Féart, Peuchant and Letenneur ^{70 ,} Reference Riemer, Maes and Christophe ^{71 )}, in conjunction with the speculated beneficial effect of n-3 supplementation on mood disorder severity⁽ Reference Stoll, Severus and Freeman ^{72 –} Reference Freeman, Hibbeln and Wisner ^{74 )}, indicates a means by which diet may influence mental disorders. Although these studies involve non-homeless individuals it is likely that their effects could be replicated in the homeless community. However, the association of depression with low vitamin D⁽ Reference Kerr, Zava and Piper ^{75 )} and Se⁽ Reference Conner, Richardson and Miller ^{76 )} intake indicates the involvement of numerous nutrients and consequently makes it difficult to determine which, if any, are involved in its aetiology. Conversely, mental health may affect an individual’s diet. For example, the negative impact of depression on appetite and subsequent food intake has been reported in both housed⁽ Reference Konttinen, Männistö and Sarlio-Lähteenkorva ^{77 ,} Reference Grossniklaus, Dunbar and Tohill ^{78 )} and homeless individuals^{( 7 )}. Medication prescribed to those with psychiatric conditions also often affects appetite⁽ Reference Basson, Kinon and Taylor ^{79 ,} Reference Uher, Farmer and Henigsberg ^{80 )}, a situation which is further complicated by the frequent occurrence of alcohol and substance abuse in the homeless^{( 57 )}. Given this complex interplay and the high level of mental disorders in the homeless community^{( 57 )} the need to consider mental health issues cannot be overestimated, although numerous studies to date have failed to acknowledge this.

The high rates of addictive behaviours, such as alcoholism⁽ Reference Langnäse and Müller ^{11 ,} Reference Darnton-Hill, Mandryk and Mock ^{81 )}, in the homeless community and the implications this can have on an individual’s nutrient intake and/or absorption⁽ Reference van den Berg, van der Gaag and Hendriks ^{44 ,} Reference De Timary, Cani and Duchemin ^{82 )} make it an important factor to consider when assessing the dietary status of the homeless. The demonstration that homeless alcoholic individuals are more likely to be malnourished, predominantly due to a substitution of alcohol for food, exemplifies the impact this condition can have on their nutrient intake⁽ Reference Langnäse and Müller ^{11 )}. Alcoholism results in numerous complications including macrocytic anaemia as a result of reduced thiamin absorption⁽ Reference Hoyumpa ^{83 )}, liver disease and osteoporosis⁽ Reference González-Reimers, Santolaria-Fernández and Martín-González ^{84 )}, all of which have been reported in alcoholic homeless individuals^{( 20 )}. The association of chronic alcoholism with elevated homocysteine levels⁽ Reference Cravo, Glória and Selhub ^{85 )} and increased oxidative stress⁽ Reference Lecomte, Herbeth and Pirollet ^{86 )} demonstrates the destructive effects such behaviour can have at the molecular level and the potential benefit a diet high in antioxidant-rich foods could have for this group. Similar to alcoholism, smoking can also make an impact on the nutritional needs of the individual as evidenced by the increased vitamin C requirements in smokers⁽ Reference Marangon, Herbeth and Lecomte ^{45 ,} Reference Lykkesfeldt, Christen and Wallock ^{87 )}. This effect may, in part, explain the recurrent demonstration of low vitamin C levels in the homeless⁽ Reference Malmauret, Leblanc and Cuvelier ^{9 ,} Reference Darmon, Coupel and Deheeger ^{10 )} given the high rates of smoking in this population^{( 57 ,} Reference Lee, Hanlon and Ben-David ^{88 )}. However, the persistence of reduced vitamin C levels following adjustment for the effects of smoking⁽ Reference Tarasuk, Dachner and Li ^{89 )} indicates that such behaviour is not the only reason for the low vitamin C reported in this group. The abuse of illicit substances is also an issue for many homeless individuals^{( 57 )}. The effects on nutrient absorption vary in accordance with the type of drug⁽ Reference Mohs, Watson and Leonard-Green ^{90 )} although this, in addition to the likelihood that drug addicts will use money to feed their drug habit in preference to feeding themselves^{( 57 )}, highlights the detrimental effect of this behaviour on nutrient intake. In light of the aforementioned impact of addictive behaviour(s) on both the consumption and absorption of certain nutrients, it is clear that in some situations malnutrition in the homeless could arise from behaviour as opposed to a deficient diet.