Food insufficiency in currently or formerly homeless persons is associated with poorer health.
Article Type: Report
Subject: Prevalence studies (Epidemiology) (Research)
Homeless persons (Health aspects)
Homeless persons (Food and nutrition)
Authors: Hamelin, Anne-Marie
Hamel, Denis
Pub Date: 12/22/2009
Publication: Name: Canadian Journal of Urban Research Publisher: Institute of Urban Studies Audience: Academic Format: Magazine/Journal Subject: Social sciences Copyright: COPYRIGHT 2009 Institute of Urban Studies ISSN: 1188-3774
Issue: Date: Winter, 2009 Source Volume: 18 Source Issue: 2
Topic: Event Code: 310 Science & research
Geographic: Geographic Scope: Canada Geographic Code: 1CANA Canada
Accession Number: 254757299
Full Text: Abstract

Our objectives were to explore the relation between food insufficiency and the health of clientele who frequent services for the homeless; and to compare that clientele with the general population in Quebec and in Canada on the association between food insufficiency and health. "Ibis study is a secondary data analysis of the 1998-1999 Quebec Survey of the clientele of services for homeless people conducted in Montreal and Quebec City. A representative sample of the daily adult population frequenting shelters, soup kitchens and drop-in centres for homeless people-in these cities had been selected according to a complex survey design between December 1998 and August 1999. 458 respondents were interviewed. The questionnaire, administrated in a face-to-face interview, comprised over one hundred standardized questions covering many aspects of the respondents' demographic, social, cultural and economics, and lifestyle behaviour as well as their mental and physical health status. For the comparison with the general population in Quebec and in Canada, data file of the cycle 2 1996-1997 National Population Health Survey was used. This particular survey was chosen because it included a similar question on food insufficiency and it was reasonably close in time. The results of this study highlight that food insufficiency is statistically associated with poorer mental and physical health in currently or formerly homeless persons. The effects of food insufficiency on their health would be even greater than that observed in the general population. They also highlight that housing matters for food sufficiency. These results provide policy makers with a better understanding of the relationship between homelessness, food insufficiency and health so as to improve policies and programs directed to homeless persons. The potential role of food security in promoting better physical, mental and social health among this population should be urgently examined.

Keywords: homelessness, food insufficiency, perceived health, physical health, mental health, housing, Canada

Resume

L'etude avait pour but d'explorer la relation entre l'insuffisance alimentaire et la sante chez la clientele des ressources pour personnes itinerantes, puis de comparer cette association avec celle chez la population generale du Quebec et du Canada. Cette etude est une analyse des resultats de l'Enquete aupres de la clientele des ressources pour personnes itineranantes des regions de Montreal-Centre et de Quebec en 1998-1999. Un echantillon representatif de l'ensemble de la clientele journaliere frequentant les centres d'hebergement, les soupes populaires et les centres de jour recevant des personnes itinerantes dans ces regions ont ete selectionnes selon un plan de sondage complexe entre decembre 1998 et aout 1999. Quatre cent cinquante-huit personnes ont ete interviewees. Le questionnaire d'enquete comprenait plus de cent questions standardisees couvrant plusieurs aspects d'ordre demographique, social, culturel, economique et comportemental, de meme en lien avec la sante physique et mentale. Les donnees de l'Enquete nationale de la sante des populations 1996-1997, cycle 2, ont ete utilisees aux fins de comparaison avec la population generale du Quebec et du Canada. La periode couverte par cette enquete et l'inclusion d'une question similaire sur l'insuffisance alimentaire rendaient ce choix raisonnable. Les resultats de notre etude demontrent que l'insuffisance alimentaire est statistiquement associee avec un etat de sante mental et physique plus pauvre chez les personnes sans abri au moment de l'etude ou ayant deja ete sans abri. Les effets de l'insuffisance sur leur sante seraient meme plus grands que ceux observes chez la population generale. Les resultats montrent que le fait d'avoir un logement importe. Ces resultats offrent une lecture des relations entre l'itinerance, l'insuffisance alimentaire et la sante qui peut eclairer les decideurs politiques et leur permette d'ameliorer les politiques et programmes ayant trait aux populations itinerantes. Il est urgent que le role potentiel de la securite alimentaire dans la promotion d'un meilleur etat de sante physique, mental et social chez ces populations soit examine et pris en compte.

Mots cles : itinerance, insuffisance alimentaire, sante percue, sante physique, sante mentale, type de residence, Canada

INTRODUCTION

Data analysis of American (Siefert et al. 2004; Kaiser et al. 2007; Seligman et al. 2007; Cook and Frank 2008) and Canadian population surveys (Che and Chen 2001; Dubois et al. 2001; Vozoris and Tarasuk 2003; Kirkpatrick and Tarasuk 2008) have highlighted the relationship between food insufficiency or food insecurity and poor health. Given the exclusion of homeless people in Canadian surveys addressing food insecurity, there is a lack of data on the scope and nature of this problem among this particular segment of the Canadian population. It is recognized that the homeless, or people without a secure, adequate and affordable place to live, are rather vulnerable to food insecurity (Dachner and Tarasuk 2002; Furness et al. 2004; Booth 2006; Hamelin et al. 2006; Whitbeck et al. 2006; Lee and Greif 2008; Weiser et al. 2009) and health problems (Hwang 2001; Baggett et al. 2010). Although the magnitude of homelessness in Canada is not clearly documented, it is believed to be significant and growing (Laird 2007). Homelessness was shown to be an independent risk factor for deaths due to drug misuse, circulatory and respiratory disorders (Morrison 2009)

The importance of studying the relationship between lack of food and health conditions in homeless populations is at least twofold. First, considering that food and shelter are among the basic foundations of life-long health (Raphael 2004), such a study is necessary to fill the gap in knowledge regarding this relationship at the population level. In the pursuit of health equity, this knowledge could contribute to the implementation of preventive practices and policy interventions in favour of homeless persons and those at risk of homelessness. Second, a better understanding of the lack of food among homeless persons is essential in order to estimate the true prevalence of food insufficiency in the general population.

This study was undertaken to further contribute to the understanding of the relationship between a lack of food and health status of Canadians through an analysis of data from the 1998-1999 Quebec Survey of Clientele of Services for Homeless People (QSCSHP) in the two largest urban areas in Quebec, Montreal and Quebec City (Institut de la statistique du Quebec 2001; 2003). Our working hypothesis was that the association of food insufficiency with poorer health status was modulated by housing conditions. In line with the study by Vozoris and Tarasuk (2003), our objectives were to 1) describe the clientele of services for homeless people in Montreal and Quebec City who report food insufficiency; 2) explore the relation of food insufficiency and the physical and mental health of that clientele; and 3) compare that clientele with the general population of Quebec and Canada regarding the association between food insufficiency and health.

METHODS

Data

This study draw on data from the QSCSHP public use microdata in ASCII format health file. This cross-sectional survey is based on a representative sample of the daily adult population ([greater than or equal to] 18 years old) who frequents shelters, soup kitchens and drop-in centres for homeless people in 24 centres in Montreal and 15 centres in Quebec City. Selection happened through a complex survey design between December 1998 and August 1999. The methodology of this survey is described elsewhere (Institut de la statistique du Quebec 2001). Survey results are based on the assumption that respondents and non-respondents are similar on the measured characteristics. For each participating centre, a certain number of days, proportional to the size of the clientele, were randomly selected. For each selected day, one or more of the individuals frequenting the centre were randomly selected and then assessed for admissibility. For each day in a given center, the objective was to get at least one respondent. Overall, the initial sample included 1,168 people of whom 109 were deemed inadmissible (40 had already been interviewed; 69 identified in soup kitchens had slept in a shelter the night before rendering them inadmissible according to the survey design) leaving a total of 1,059 people. Of these, 757 accepted to be interviewed partially or totally, 509 in Montreal and 248 in Quebec City, resulting in a response rate of 76.5% in Montreal and 62.9% in Quebec City. The questionnaire, administrated in a face-to-face interview, comprised over one hundred standardized questions covering demographics, socioeconomics, lifestyle behaviors, health status as well as service needs. The interview was conducted in either English or French according to the language of the respondent. As the initial survey questionnaire was long (on average 2.5 hours), it was decided to shorten the questionnaire and to prepare two versions, each one including a common trunk and specific questions. Of the 757 eligible persons, 458 respondents were randomly selected to answer the version of the questionnaire covering lifestyle behaviors and food insufficiency. Our analysis is based on the responses of these 458 respondents.

The public use microdata health file for the household component of the Cycle 2 (1996-1997) National Population Health Survey (NPHS) was used to compare our results with those at the Canadian level. This particular survey was chosen because it included a similar question on food insufficiency and was reasonably close in time; furthermore a detailed examination of the survey data on the relationship between food insufficiency and health had been already published (Tambay and Catlin 1995; Swain et al. 1999; Vozoris and Tarasuk 2003; 2004). NPHS is an ongoing survey conducted by Statistics Canada to collect information on the health of Canadians from a panel of 17,276 individuals, re-interviewing them by telephone every two years for longitudinal purposes. Again the survey was administered either in English or French. The response rate for the individuals panel was 94% in 1996/97. In total, residents of 82,000 households in all Canadian provinces were interviewed for cross-sectional purposes with the exception of people living on Native reserves or on Canadian Forces bases, people from certain remote areas in Quebec and Ontario, and homeless persons. Sociodemographic and health variables included household type, major sources of income, home ownership, region of residence, self-rated health, chronic conditions, body mass index, distress index and food insecurity. A detailed description of the survey design and methodology appears elsewhere (Tambay and Catlin 1995; Swain et al. 1999; Vozoris and Tarasuk 2003; 2004). Both provincial (Quebec) and Canadian estimations of this survey were employed.

Food Insufficiency

The lifestyle behavior section of the QSCSHP included three questions exploring issues on the lack of food in clients who frequent services for homeless people. The first question concerns the average number of meals per day ("over the past seven days, how many meals per day have you eaten on average"). The second relates to food sources over the past seven days ("over the past seven days, have you eaten or got food from one of the following sources: a) grocery, b) food basket provided by a community organization; c) meal that you have paid for in a cafe or a restaurant, d) soup kitchen, e) friends or family members, f) leftovers found on the street, g) other source--specify). The third question specifically addresses food sufficiency. Respondents were asked a question based on the food sufficiency indicator developed for the Third National Health and Nutrition Examination Survey (Cristofar and Basiotis 1992): "Over the past seven days, was having enough food to eat a problem to you: a) often, b) sometimes, c) rarely, or d) never?" Individuals who responded "often", "sometimes" or "rarely" having enough food to eat was a problem were classified as "food insufficient" (36% of the sample); those who reported that it was "never" a problem for them in the last seven days were deemed food sufficient. In contrast to the single-item food insufficiency question, food insecurity is a comprehensive concept that addresses the many facets and dynamics of the experience related to a lack of food and a lack of control over one's food situation, as well as, the potential consequences at the individual and household levels. In an attempt to capture this complexity, the food insecurity scale is constructed using 18 questions if the household has children or 10 if it does not (Nord et al. 2008).

The identification of food sufficiency in the population is different in the NPHS for both the concept and the process. In the NPHS, the concept of food sufficiency is defined at the household level, as opposed to the individual level. Also, NPHS consists of three questions including a step question: respondents were asked if, over the past 12 months, their household had "ever run out of money to buy food." Those who responded affirmatively were then asked two additional questions: 1) did anyone in your household receive food from a food bank, soup kitchen or other charitable agency? 2) "which of the following best describes the food situation in your household? a) always enough food to eat; b) sometimes not enough food to eat; c) often not enough food to eat." In spite of these differences, the last NPHS question and the third question of the QscSHP focus both on "enough food" and their operational definition is similar to that widely used in the U.S. National Health and Nutrition Examination Surveys (Vozoris and Tarasuk 2003). The inclusion of a screening question in the NPHS means that the likelihood of classifying a respondent as food insufficient is probably lower in the NPHS compared to the QSCSHP.

Health Indicators

Two categories of health variables are considered in this paper: physical health and mental health. The physical health indicators include self-rated health scale, multiple chronic conditions (see Appendix 1 for definition), heart disease, and obesity (as indicated by body mass index (BMI) [greater than or equal to] 30 based on self-reported height and weight data), a well known risk factor for many health conditions including diabetes, heart disease and stroke. The mental health indicators include alcohol and drug disorders, depression, emotional disorders, and disorders referring to the first two axes of the DSM-IV scale (American Psychiatric Association 1996). Except for axis 2 disorders, all of these mental health problems were considered in the past 12 months only. All measures were dichotomized in order to differentiate individuals with responses indicative of poor health. (See Appendix 1 for descriptions of the health outcomes and their cut-offpoints defining poor health).

Sociodemographic variables

The following attributes were considered for their potential relationship with both health indicators and food insufficiency (Fournier 2003): region (Montreal-Centre; Quebec City), gender, age, education (high-school completed, or not), sources of income (none; only one source; more than one), and residential status (currently homeless; not currently homeless; never homeless). 'Currently homeless' which is the status of 34% of the study population refers to someone living outdoors, in a public area, in an abandoned space, or in need of finding a shelter or a hostel at the time of the study. 'Not currently homeless' (43.5%) or "formerly homeless" refers to individuals who had experienced homeless in their lifetime but had some form of housing at the time of the survey. 19.9% of the "Not currently homeless" have slept outdoors, in a public area or in an abandoned space in the past 12 months, compared to 63% of the currently homeless. In 57% of the 18-25 years, this experience of homelessness was recent, i.e., since less than a year. An individual who has never experienced homelessness was classified as 'never homeless' (22.5% of the study population); nevertheless, 47% of this group has moved at least once in the past year compared to 17% in the general population, and one in five lived in a rooming house. Although the survey did not include any question about aboriginal status, the proportion of homeless people who were Aboriginals in this study population is likely to be non-significant.

More than one in four clients of services provided to the homeless declared having experienced stressful events related to lodging within the past year. One in seven reported either a loss of lodging or eviction. Stressful events related to lodging were more prevalent in the currently homeless (35.5%) than in those who were not currently homeless (22.9%) or never homeless (20.6) at the time of the study (p< 0.05).

For the purpose of statistical analysis, "currently" and "formerly" homeless were combined in the present study on the basis that they both showed a similar risk of food insufficiency (respectively 35.9 and 33.1) compared to the "never homeless" (21.1).

Statistical methods

All analyses were performed with SAS version 8 (SAS Institute) and SUDAAN version 9.0.0 (Research Triangle Institute International 2004). SAS was used mainly for descriptive statistics. SUDAAN was used for the final analysis in order to account for unequal probabilities of selection of respondents and some clustering, especially in the same services. Contrary to classic statistical softwares, SUDAAN takes into consideration the structure of dependence of observations when it calculates standard errors for parameter estimates in statistical models. To do this, the Institut de la Statistique du Quebec provided us with information on how the survey was planned. In order to infer results obtained from the sample to the population, all estimates produced in this paper are sample-weighted.

A preliminary understanding of how food insufficiency varies according to selected individual attributes was obtained through cross-tabulations and multiple logistic regressions. Cross-tabulations were used to examine the associations between food insufficiency and socio-demographic characteristics of the homeless and the marginally-housed persons. Multiple logistic regressions were used to describe the relationship between food insufficiency and the selected characteristics.

"This approach allowed identifying the relative importance of each variable knowing that many of them were interrelated. This also helped to identify potential confounding variables on the association between food insufficiency and health indicators.

To measure whether there is an association between food insufficiency and health problem indicators, we carried out logistic regressions for the whole sample with health outcomes as dependent variables. These models enabled us to estimate the crude (unadjusted) odds ratio that a food insufficient individual would report poor health. Food sufficient individuals were considered as the category reference in models. Then individual attributes were added in all models to calculate adjusted odds ratios and to control potential confounding effects of these attributes on observed associations. We could therefore identify the net effect of food insufficiency on the health differences, not masked by, for example, the well-known relation between age and the presence of poor health. Final models included only significant and confounding variables.

The same models were carried out with NPHS data, but for a limited number of health variables. Since the mental health section in the NPHS is not as elaborated as in the QSCSHP, it was not possible to create similar mental health indicators. Only the self-rated health, obesity, and partially multiple chronic problems (the list of chronic diseases not being the same) could be compared on both populations, general and homeless. To account for the effects of complex design on variance estimate, all variability measures such as confidence intervals were calculated using bootstrap resampling techniques with a set of 500 bootstrap weights created by Statistics Canada (Yeo et al. 1999). As we did for the QSCSHP, in order to infer results from the sample to the population, all the results from NPHS presented in this paper reflect weighted estimates.

The rate for partial non-responses was relatively low (globally less than 10%) for the variables considered in our models. There was therefore no reason to believe that the non-respondents could bias our estimations.

RESULTS

The majority of the study population was male (86%) and was younger than the general population ([greater than or equal to] 45 year old: 35% vs 47% in the general population).

One third of the sample (36%; n = 155) reported food insufficiency. Amongst them, 31% mentioned that having enough food to eat was 'often' a problem and 44% reported that it was a problem 'sometimes' during a period of seven days. The majority had consumed at least one meal from a soup kitchen or a centre in a seven-day period (currently homeless: 86.6%; not currently homeless: 90.8%; never homeless: 91.3%). The proportion of men having used the service of a soup kitchen was higher than that of women (93% vs 79%, p < 0.05). Other sources of food included groceries, food baskets from community organizations, meals paid for in coffee shops or restaurants, friends or family, and leftovers found on the street. Sources of food were the same regardless of respondents' residential status except for two sources: groceries (currently homeless: 34.2%; not currently homeless: 53.5%; never homeless: 61.9; p<0.05) and use of leftovers found on streets in the last seven days (currently homeless: 14.9%; not currently homeless: 1.3%; never homeless: 3.3; coefficient of variation > 25%).

Table 1 shows the links between food insufficiency and some individual attributes. Within the sample, only the condition of being "currently or formerly homeless" was statistically associated with food insufficiency. Respondents who were homeless or had an episode of homelessness in their life were at greater risk of food insufficiency than respondents who have never been homeless (41% vs 17%). Other variables show interesting links, but are not statistically significant, mostly due to the small sample size. Male respondents and respondents between 18 and 29 years old were more likely to report food insufficiency. This result is important given that age plays an important confounding role as it is strongly linked to health problems in general.

The odds that the clientele who frequent services for the homeless and report food insufficiency also report poor self-rated health are high. Table 2 presents crude and adjusted odds ratios derived from multiple logistic regression on physical health indicators. For all of them, strong relationships are observed in the expected direction: food insufficiency is associated with poor physical health. For example, 45% of food-insufficient people declared their health as poor or fair compared to only 27% for those who are food-sufficient. All crude odds ratios are significant. Once individual attributes are controlled for, the results display stronger links for self-rated health, multiple chronic conditions and heart disease. Similar results for crude and adjusted odds ratios were seen for obesity. Apart from food insufficiency, other factors increased the probability that the respondents declared poor physical health. For self-rated health, perception of one's health decreased with increasing age. The same holds for heart disease. Education was also statistically associated with self-rated health: individuals who did not complete high school were more likely to have poor health. In the model for obesity, it was found that homeless women had a higher risk of obesity than men.

Mental health is also strongly related to food insufficiency (Table 3). All crude odds ratios are superior to 2. However, once individual attributes are added to the models, adjusted odds ratios become non-significant for alcohol and drug disorders, and axis 2 disorders. People who were food-insufficient had higher risk of suffering from depression or emotional disorder compared to those who were food-sufficient, even after individual characteristics were taken into account. Once again, apart from food insufficiency, other factors increased the probability that homeless declared poor mental health. Men, young people and those who were currently or formerly homeless were more likely to declare alcohol or drug problems. Emotional disorders occurred more often in women and in people who have been homeless at least once in their life. Finally, the risk of having axis 2 disorders was increased in young homeless respondents (18-29) compared to older respondents ([greater than or equal to] 45 year old).

Table 4 compares the above results concerning the clientele of services for homeless people with the general population in Canada and in Quebec regarding self-rated health, multiple chronic conditions and obesity. The adjusted odds ratios show that in the overall Canadian population, food insufficiency is significantly associated with poor/fair self-rated health and with multiple chronic conditions. However, unlike the clientele of services for the homeless, the link between food insufficiency and obesity is not statistically significant in the general population of Quebec or Canada.

DISCUSSION

The QSCSHP which portrays this marginalized population according to residential status, socio-economics, lifestyle behaviours, health dimensions and needs for services provided a unique opportunity to study the relationship between food insufficiency and poor health among homeless people.

Food insufficiency is associated with poorer health

The odds that the clientele who frequent services for the homeless report both food insufficiency and poor/fair self-rated health was moderately high, and when personal characteristics were controlled for, this association was even stronger, meaning that food insufficiency was associated with the perception of one's own health. Self-rated health is known to be a good predictor of overall health status. It can reflect aspects of health not captured in other measures such as physiological and psychological reserves and social and mental function (Canada 2006). The relationship is therefore not surprising since food insecurity, a concept encompassing food insufficiency, relates to the physical, mental and social aspects of the life of homeless. Qualitative work (Hamelin et al. 2002) has demonstrated that food security represents a unifying concept with the ability to link physical health (fills his or her nutritional needs), mental health (certainty of being able to eat; being able to choose; eating with dignity; feeling that one controls its basic needs) and capacity for social integration (being able to realize elementary food activities) of individuals.

The food insufficient individuals had high odds of reporting poor physical health, even after adjusting for the confounding effects of age, education, and homeless status. This was true for multiple chronic conditions such as anaemia, allergies, chronic bronchitis, diabetes, emphysema and heart disease. In this study, the adjusted odds ratio of reporting heart disease while being food insufficient was the highest. Heart disease was also the condition for which the difference between the crude and the adjusted ratios was the largest, suggesting the importance of meeting specific dietary needs for the condition. These needs are not likely to be easily fulfilled given the study population's source of food is often soup kitchens (Institut de la statistique du Quebec 2003). It is suspected that food insufficient homeless individuals do not regularly eat adequate fruits and vegetables which are important sources of antioxidant nutrients and folate associated with cardiovascular health (USDA Center for Nutrition Policy and Promotion 2000; Tarasuk et al. 2005). They would rather eat cheaper, energy dense foods which are high in fats and sugar (Drewnowski and Barratt-Fornell 2004). This is consistent with a recent study of dietary intakes of homeless women who perceived that their shelter diet contributed to chronic diseases and their symptoms (Davis et al. 2008). It was not possible to examine the relation between frequency of program use and food sufficiency status.

The likelihood of an inadequate diet is also supported by the high crude odds of being both food insufficient and obese (BMI [greater than or equal to] 30); food insufficient individuals were most susceptible to obesity. A result that is consistent with studies of homeless children in the U.S. (Wood et al. 1990; Smith and Richards 2008). Alternating periods of starvation and overeating when food is available has been suggested as a potential explanation for overweight among homeless individuals (Bouvier 2008; Smith and Richards 2008).

Individuals who reported food insufficiency also suffered from mental health problems. The crude odds ratios show a relationship between food insufficiency and emotional disorder, depression, alcohol or drug disorder, axis 1 and axis 2 disorders, in their lifetime. After adjusting for the potentially confounding factors of age, education and homeless status, the only association that remained was that food insufficient individuals were more likely to report emotional disorders, depression and axis 1 disorder. "Ibis result echoes the U.S. literature on the health of homeless children and youth (Grant et al. 2007).

Food insufficiency among homeless adults was recently shown to be associated with impaired access to medical or surgical care, prescription medications and mental health care in a national U.S. based study (Baggett et al. 2010). The researchers hypothesized that individuals experiencing food insufficiency assign lower priority to health care in favour of directing personal resources towards the fulfillment of basic needs. In the QSCSHP, it was found that "currently" homeless respondents were far more likely to report a need for financial aid (85,9%) than for improved medical services (50,5%) or for treatment of emotional problems (45,6%) (Hamelin and Fournier 2003). Kushel et al. (2006) have also shown that the competing life demands surrounding housing instability and food insecurity may lead to delays in seeking care and predispose individuals to access urgent acute care services.

Association between food insufficiency and poor health: greater than that in the general population of Quebec and Canada

The estimation of the prevalence of food insufficiency for this particular population (36%; 46% for "currently homeless") is high when compared to the proportion observed in the general population in Canada (4%), with the exclusion of the Aboriginal living on Reserves, homeless people and other vulnerable groups (Vozoris and Tarasuk 2003). The gap in the prevalence of food insufficiency between these two populations is even likely to be larger. The magnitude of food insufficiency and poor health may be underestimated for the clientele who use services for the homeless since those who were not food insufficient in the past 7 days might well have been food insufficient over the past 12 months. Also presumably the 7 day food sufficiency rate for the general population would be much lower than 4%.

Food insufficient individuals who frequent services for homeless people would also be more likely to report heart disease (adjusted ratio 5.4) than the general population (adjusted ratio 2.9), while the likelihood to report poor/fair self-rated health (adjusted ratio 2.9) would be the same in both. For obesity, there is a strong association between food insufficiency and BMI [greater than or equal to] 30 in homeless persons, while the link is not statistically significant in the general population of Quebec or Canada. Knowledge is growing about what is now called the "food insufficiency-obesity paradigm" in low income families (Casey et al. 2006; Dinour et al. 2007).

A lack of food or lack of quality food is seen as an everyday stressor for individuals suffering from mental health disorders (Campbell and Desjardins 1989; Radimer et al. 1992; Ahluwalia et al. 1998; Menke 2000; Alaimo et al. 2002; Hamelin et al. 2002; Frongillo 2003; Wu and Schimmele 2005). There is also a growing body of evidence on the relationship between food insufficiency and depression among low-income populations (Alaimo et al. 2002; Weinreb et al. 2002; Siefert et al. 2004; Vozoris and Tarasuk 2004; Heflin et al. 2005; Wu and Schimmele 2005). Findings from a 3-year panel of women welfare recipients showed that the relation between meeting the diagnosis screening criteria for major depression and food insufficiency in these women was highly significant even after controlling for known risk factors of depression (Helfin and Ziliak 2008). Similar to our findings, Wu and Schimmele (2005), who examined the NPH cycle II data, found that food insufficiency has an independent effect on depression in the Canadian population. Nonetheless, the odds ratio of this association in our study appears to be greater than that of Wu and Schimmele. Similar to our results, these authors did not find that food insufficient women were more subjected to depression than men.

Housing matters

The characteristics of clientele with food insufficiency who use services for the homeless show that those respondents who were either currently homeless or who had prior experience of homelessness in their life were at a greater risk of food insufficiency than those who have never been homeless. Statistically, no association was found between the residential status "never homeless" and food insufficiency. Interestingly, most of the "never homeless" were satisfied with their lodging. This calls for closer examination on how housing affects the social determinants of health including food sufficiency. Hwang et al. (2003), found that residents of rooming houses in Toronto aged [greater than or equal to] 35 had significantly poorer health status than their counterparts in the general population, even when compared to the lowest income quintile. Individuals reporting poorest health tended to live in rooming houses with the worst physical condition (e.g., cleanliness, noise level). Facilities such as rooming houses are important sources of housing for very low-income Canadians; more than one quarter of rooming house residents in the Hwang study had been homeless less in the last five years. In our study, this is a third of "currently homeless" persons (32.7%) and more than a third of "not currently homeless" persons (42.1%) who lived in a rooming house at the time of the study. However, it remains unclear whether the concentration of people with poor health in poor-quality housing is the result of selection process or the direct effects of housing on health. Eberle et al (2001) observed that the health, social and criminal costs for homelessness are considerable; they point out a crucial need for research comparing the cost of supporting and providing housing for vulnerable groups with the costs incurred in health and other sectors. However, Dunn et al. (2006) also underlined the need to examine whether subsidized housing, particularly for vulnerable groups, has a large enough effect on disposable income to reduce food insecurity. Mercier and Alarie, conducted a qualitative study of 30 individuals with a history of homelessness and substance abuse in Montreal, providing a glimpse into the complexity and direction of the relationship between housing and food, which could affect longer-term stability

and eventually stop the circle of drug abuse and homelessness. The researchers found that having a place to live was not enough, but to fully "live" in his home could make a difference, e.g. to invest time and energy in order to feel as if it is their own place, to do activities of everyday life such as buying and stocking food in their place. An improved relationship with their housing had a motivating effect and these individuals expressed the desire to live there for a longer period of time (Mercier and Alarie 2001). Additionally, the alleviation of food insufficiency has been suggested as a way of protecting families in danger of becoming homeless (Gundersen et al. 2003).

Limitations

Our study has a number of limitations. The small sample size limited the possibility of detecting statistical relationship between food insufficiency and health problems. For example, it was not possible to examine the association between obesity and food insufficiency among men and women separately; the latter group being far too small. In order to enlarge the sample size, the individuals who were categorized as 'rarely' food insufficient were included in the 'food insufficient' definition, potentially contributing to a modification of the true relationship examined between food insufficiency and health problems. However, because of the possibility of underreporting of a lack of food by the homeless, as shown in other studies with people having known hardship (Wolfe et al. 1998; Hamelin et al. 2006), as well as in the more general population (Derrickson and Brown 2002), and because of our classification of people as food sufficient if they reported no food problems in the past seven days, we hypothesize that this grouping had an effect in reducing the odds of an association. Also, one must take into account that the survey was conducted with clients of services provided to the homeless, which included soup kitchens. There is a possibility that our sample is biased towards the most healthy and least food insufficient group among the homeless and prior homeless in Quebec City and Montreal. In other words, our interpretation, if biased, would more conservative rather than falsely alarming.

Our analysis relied on self-reported measures of health; however, almost two thirds of the survey participants reported that at least one problem was confirmed by a doctor (Institut de la statistique du Quebec 2003). In their study of the Canadian Community Health Survey, Lyons et al. (2008) found that associations between obesity and food insecurity are more pronounced when self-reported data on height and weight are used compared with measured data on height and weight. Although under-reporting weight and over-reporting height is widely recognized, this appears less likely in our study population composed of persons who are preoccupied with daily survival and who are mainly male.

In addition, one must be careful in interpreting the data from the two surveys (QSCSHP; NPH) which differ in their design. The period of reference is not the same: the QSCSHP used a period of 7 days, versus 12 months for the NPHS. In both cases, the recall period seems appropriate considering the variability of the phenomenon of interest in the surveyed population and the intent for the use of the data. However, the impact on the results concerning an association between food insufficiency and poorer health between the general population and the homeless is still unclear. Regarding data collection, the QSCSHP provides point prevalence data on a typical day, we, therefore, do not know how many other people are homeless on a given day; this may underestimate the true situation. The concept of 'food insufficiency' was approached at a different level: in the NPHS, the questions referred to food insufficiency at the household level, whereas the QSCSHP referred to food insufficiency at the individual level.

It is also important to highlight that our study was not intended to detect potential causal relationships; this would have been impossible given the cross sectional nature of analysed data. We are aware that many other individual characteristics might be related to the health indicators and that they could affect our results in regression models. Our objective was to calculate the risk (odds ratio) of someone with food insufficiency suffering from health problems, once the selected factors were controlled.

CONCLUSION

The results of this study highlight that food insufficiency is associated with poor mental and physical health in currently or formerly homeless persons. The effects of food insufficiency on their health would be even greater than that observed in the general population. The relationship between homelessness, food insufficiency and health should prompt policy makers to improve both health and social policies and programs aimed at this uniquely important group of persons. The provision of adequate food services targeting homeless who are food insufficient would prevent further deterioration of their health. The role of food security in promoting better physical, mental and social health among the homeless and marginally housed persons should be urgently considered as one key to reduce health inequality issues in Canada.

Appendix

Descriptions of health measures selected in this paper

Self-rated health

Definition. Self-rated health is the most frequently collected variable on population's health. In this survey, it refers explicitly to physical health" "We will now talk about physical health. In general, compared to other persons in your age, would you say your health is excellent, very good, good, fair or poor?".

Dichotomized cut-point. Differentiate two groups of people, those saying their health is fair or poor and those considering it excellent, very good or good.

Multiple chronic conditions

Definition. Respondents were asked if they had chronic conditions. The conditions listed were diabetes; anaemia; skin disease; skin allergies (or coetaneous allergies); hay fever; other allergies; serious back problem; other serious bone problems or joints ache; arthritis or rheumatism; emphysema, chronic bronchitis, persistent cough or asthma; high blood pressure; heart disease; urinary incontinence; stomach ulcers; other bowel disorders; epilepsy; thyroid condition; migraine or frequent headache; paralysis.

Dichotomized cut-point. Respondents with multiple chronic conditions corresponds to those reported at least two.

Heart disease

Definition. Respondents were asked if they had a heart disease.

Dichotomized cut-point. Presence or absence.

Obese according to BMI

Definition. The body mass index is the most frequently measure to obesity. It consists to a ratio of mass in kg to the square of height in meters.

Dichotomized cut-point. Obese = BMI superior or equal to 30.

Mental health

The mental disorders were measured through the Diagnostic Interview Schedule (DIS) and the Composite International Diagnostic Interview Simplified (CI-DIS). These allow to pass in review the symptoms and criteria corresponding to these disorders according to the DSM-IV (American Psychiatric Association). The classification system is based on a multiaxial evaluation of which the two first axes were retained only. The first axis refers to clinical disorders including major mental disorders, as well as developmental and learning disorders and other situations that can be the object of a clinical investigation. These include notably those disorders that are not linked to the use of a substance : schizophrenia and other psychotic troubles, affective disorders such as depression and pathological gaming. The second axe comprehends mainly troubles linked to underlying pervasive or personality conditions, as well as mental retardation.

The term "current" refers to the past 12 months.

Current Axis 1 disorders (troubles de l'axe 1); current emotional disorders; current depression; current alcohol and drug disorders; Axis 2 disorders (troubles de l'axe 2)

Definition. As mentioned in the preceding paragraph and according to DSM-IV

Dichotomized cut-point. Presence or absence

Acknowledgments

We would like to thank Professor Valerie Tarasuk of University of Toronto for her guidance with data analysis. This research was supported in part by a grant from the Centre de sante et des services sociaux de la Vieille-Capitale, Quebec City, Canada.

References

Ahluwalia, I. B., J. M. Dodds, et al. 1998. Social support and coping behaviors of low-income families experiencing food insufficiency in North Carolina. Health Educ Behav 25(5): 599-612.

Alaimo, K., C. M. Olson, et al. 2002. Family food insufficiency, but not low family income, is positively associated with dysthymia and suicide symptoms in adolescents. J Nutr 132(4): 719-25.

American Psychiatric Association. 1996. DSM-IV Manuel diagnostic et statistique des troubles mentaux. Paris, Masson.

Baggett, T., J. O'Connell, et al. 2010. The unmet health care needs of homeless adults: a national study. American Journal of Public Health 100(7): 1326-33.

Booth, S. 2006. Eating rough: food sources and acquisition practices of homeless young people in Adelaide, South Australia. Public Health Nutr 9(2): 212-8.

Bouvier, G. 2008. Starvation then binge-eating can cause obesity in the homeless. HIV Clinician 20(3): 1-2.

Campbell, C. C. and E. Desjardins 1989. A model and research approach for studying the management of limited food resources by low income families. Journal of Nutrition Education 21: 162-171.

Canada, Statistics 2006. Health indicators: Definitions and data sources, Health status, http://www.statcan.gc.ca/pub/82-221-x/2006001/4063753-eng.htm (Retrieved January 15th, 2009).

Casey, E H., E M. Simpson, et al. 2006. The association of child and household food insecurity with childhood overweight status. Pediatrics 118(5): e1406-13.

Che, J. and J. Chen. 2001. Food insecurity in Canadian households. Health Rep 12(4): 11-22.

Cook, J. T. and D. A. Frank. 2008. Food security, poverty, and human development in the United States. Ann N Y Acad Sci 1136: 193-209.

Cristofar, S. P. and P. P. Basiotis. 1992. Dietary intakes and selected characteristics of women ages 19-50 years and their children ages 1-5 years by reported perception of food sufficiency. Journal of Nutrition Education 24: 53-58.

Dachner, N. and V. Tarasuk. 2002. Homeless "squeegee kids": food insecurity and daily survival. Soc Sci Med 54(7): 1039-49.

Davis, L. R., N. F. Weller, et al. 2008. Dietary intake of homeless women residing at a transitional living center. J Health Care Poor Underserved 19(3): 952-62.

Derrickson, J. p. and A. C. Brown. 2002. Food security stakeholders in Hawaii: perceptions of food security monitoring. J Nutr Educ Behav 34(2): 72-84.

Dinour, L. M., D. Bergen, et al. 2007. The food insecurity-obesity paradox: a review of the literature and the role food stamps may play. J Am Diet Assoc 107(11): 1952-61.

Drewnowski, A. and A. Barratt-Fornell. 2004. Do healthier diets cost more?" Nutrition Today 39: 161-168.

Dubois, L., E. Beauchesne-R., et al. 2001. Enquete sociale et de sante 1998. Chapitre 6--Alimentation: perceptions, pratiques et insecurite alimentaire. Sante et bien-etre. Quebec, Institut de la statistique du Quebec: 52pp.

Dunn, J. R., M. V. Hayes, et al. 2006. Housing as a socio-economic determinant of health: findings of a national needs, gaps and opportunities assessment. Can J Public Health 97 Suppl 3:S11-5, S12-7.

Eberle, M., D. Kraus, et al. 2001. The Relationship between Homelessness and the Health, Social Services and Criminal Justice Systems: A Review of the Literature--Volume 1. Vancouver, British Columbia Ministry of Social Development and Economic Security,

BC Housing Management Commission: 1-56.

Fournier, L. 2003. Chapitre 2--Instabilite residentielle. Enquete aupres de la clientele des ressources pour personnes itinerantes des regions de Montreal-Centre et de Quebec, 1998-1999, Volume 2. Quebec, Institut de la Statistique du Quebec: 23-46.

Frongillo, E. A. 2003. Understanding obesity and program participation in the context of poverty and food insecurity. J Nutr 133(7): 2117-8.

Furness, B. W., P. A. Simon, et al. 2004. Prevalence and predictors of food insecurity among low-income households in Los Angeles County. Public Health Nutr 7(6): 791-4.

Grant, R., A. Shapiro, et al. 2007. The health of homeless children revisited. Adv Pediatr 54: 173-87.

Gundersen, C., L. Weinreb, et al. 2003. Homelessness and food insecurity. Journal of Housing Economics 12: 250-272.

Hamelin, A. M., M. Beaudry, et al. 2002. Characterization of household food insecurity in Quebec: food and feelings. Soc Sci Med 54(1): 119-32.

Hamelin, A. M. and L. Fournier. 2003. Habitudes de vie et besoins de services. Enquete aupres de la clientele des ressources pour personnes itinerantes des regions de Montreal-Centre et de Quebec 1998-1999. Montreal, Institut de la statistique du Quebec. Volume 2, chapitre 3.

Hamelin, A. M., C. Mercier, et al. 2006. The Food Environment of Street Youth. Journal of Hunger & Environmental Nutrition 1 (3): 69-98.

Heflin, C. M., K. Siefert, et al. 2005. Food insufficiency and women's mental health: findings from a 3-year panel of welfare recipients. Soc Sci Med 61 (9): 1971-82.

Helfin, C. M. and J. P. Ziliak. 2008. Food Insufficiency, Food Stamp Participation, and Mental Health. Social Science Quarterly 89(3): 706-727.

Hwang, S. 2001. Homeless and health. Canadian Medical Association Journal 164(1): 229-33.

Hwang, S. W., R. E. Martin, et al. 2003. The relationship between housing conditions and health status of rooming house residents in Toronto. Can J Public Health 94(6): 436-40.

Institut de la statistique du Quebe. 2001. Enquete aupres de la clientele des ressources pour personnes itinerantes des regions de Montreal-Centre et de Quebec, 1998-1999, Volume 1. Sante et bien-etre. Quebec, Institut de la statistique du Quebec.

Institut de la statistique du Quebec. 2003. Enquete aupres de la clientele des ressources pour personnes itinerantes des regions de Montreal-Centre et de Quebec, 1998-1999, Volume 2. Sante et bien-etre. Quebec, Institut de la statistique du Quebec.

Kaiser, L., N. Baumrind, et al. 2007. Who is food-insecure in California? Findings from the California Women's Health Survey, 2004. Public Health Nutr 10(6): 574-81.

Kirkpatrick, S. I. and V. Tarasuk. 2008. Food insecurity is associated with nutrient inadequacies among Canadian adults and adolescents. J Nutr 138 (3): 604-12.

Kushel, M. B., R. Gupta, et al. 2006. Housing instability and food insecurity as barriers to health care among low-income Americans. Journal of Internal General Medicine 21: 71-77.

Laird, G. 2007. Shelter. Homelessness in a growth economy: Canada's 21st century paradox. Calgary, Sheldon Chumir Foundation for Ethics in Leadership: 1-98.

Lee, B. A. and M. J. Greif. 2008. Homelessness and hunger. J Health Soc Behav 49(1): 3-19.

Lyons, A. A., J. Park, et al. 2008. Food insecurity and obesity: a comparison of self-reported and measured height and weight. Am J Public Health 98(4): 751-7.

Menke, E. M. 2000. Comparison of the stressors and coping behaviors of homeless, previously homeless, and never homeless poor children. Issues Ment Health Nurs 21 (7): 691-710.

Mercier, C. and S. Alarie. 2001. Precarite residentielle et consommation de psychotropes. Alcoologie et Addictologie 23(1): 13-19.

Morrison, D. S. 2009. Homeless as an independent risk factor for mortality: results form retrospective cohort study. International Journal of Epidemiology 38: 877-883.

Nord, M., M. Andrews, et al. 2008. Household Food Security in the United States, 2007. Economic Research Report No. (ERR-66) 65. Washington, D.C., United States Department of Agriculture.

Radimer, K. L., C. M. Olson, et al. 1992. Understanding hunger and developing indicators to assess it in women and children. Journal of Nutrition Education 24(Suppl. 1): 36S-44S.

Raphael, D. 2004. Social determinants of health, Canadian Perspectives. Toronto, Canadian Scholars' Press Inc.

Research Triangle Institute International. 2004. SUDAAN. Research Triangle Park, NC. SAS Institute SAS version 8. Cary, NC.

Seligman, H. K., A. B. Bindman, et al. 2007. Food insecurity is associated with diabetes mellitus: results from the National Health Examination and Nutrition Examination Survey (NHANES) 1999-2002. J Gen Intern Med 22(7): 1018-23.

Siefert, K., C. M. Heflin, et al. 2004. Food insufficiency and physical and mental health in a longitudinal survey of welfare recipients. J Health Soc Behav 45 (2): 171-86.

Smith, C. and R. Richards. 2008. Dietary intake, overweight status, and perceptions of food insecurity among homeless Minnesotan youth. Am J Hum Biol 20(5): 550-63.

Swain, L., G. Catlin, et al. 1999. The National Population Health Survey--its longitudinal nature. Health Rep 10(4): 69-82(ENG); 73-89(FRE).

Tambay, J. L. and G. Catlin. 1995. Sample design of the national population health survey. Health Rep 7(1): 29-38, 31-42.

Tarasuk, V., N. Dachner, et al. 2005. Homeless youth in Toronto are nutritionally vulnerable. J Nutr 135(8): 1926-33.

USDA Center for Nutrition Policy and Promotion. 2000. Food insufficiency and the nutritional status of the elderly population. Nutrition Insights 18 (May 2000): 1-2.

Vozoris, N. T. and V. S. Tarasuk. 2003. Household food insufficiency is associated with poorer health. J Nutr 133(1): 120-6.

Vozoris, N. T. and V. S. Tarasuk. 2004. The health of Canadians on welfare. Can J Public Health 95(2): 115-20.

Weinreb, L., C. Wehler, et al. 2002. Hunger: its impact on children's health and mental health. Pediatrics 110(4): e41.

Weiser, S., D. Bangsberg, et al. 2009. Food insecurity among homeless and marginally housed individuals living with HIV/AIDS in San Francisco. AIDS Behavior 13(5): 841-8.

Whitbeck, L. B., X. Chen, et al. 2006. Food insecurity among homeless and runaway adolescents. Public Health Nutr 9(1): 47-52.

Wolfe, W. S., C. M. Olson, et al. 1998. Hunger and Food Insecurity in the Elderly --Its Nature and Measurement. Journal of Aging and Health 10(3): 327-350. Wood, D. L., R. B. Valdez, et al. 1990. Health of Homeless Children and Housed, Poor Children. Pediatrics 86(6): 858-866.

Wu, Z. and C. M. Schimmele. 2005. Food Insufficiency and Depression. Sociological Perspectives 48 (4): 481-504.

Yeo, D., H. Mantel, et al. 1999. Bootstrap Variance Estimation for the National Population Health Survey. 1999 Proceedings of the Survey Research Methods Section., American Statistical Association: 778-783.

Anne-Marie Hamelin

Departement des sciences des aliments et de nutrition Universite Laval

Denis Hamel

Unite Surveillance des maladies chroniques et de leurs determinants Institut national de sante publique du Quebec
Table 1 Crude and adjusted odds ratios of sociodemographic
characteristics of food insufficient individuals, QSCSHP

                                                      Food
Characteristics *                  n ([dagger])   insufficient

Gender
  Male                                 357            37.5
  (Female)                              65            26.6
Region
  Montreal                             264            37.4
  (Quebec City)                        158            33.1
Age group
  18-29                                 94            46.2
  30-44                                191            33.3
  (45+)                                132            31.3
Education
  High school not                      250            37.3
  completed
  (High school completed)              171            33.8
Number of income sources
  None or only 1 source                288            35.9
  (2 sources or more)                  133            35.8
Residential status
  Currently or formerly homeless       327            41.3
  (Never homeless)                      94            16.8

Characteristics *                  Crude OR ([double dagger])

Gender
  Male                                   1.7 (0.7-3.8)
  (Female)                                     1
Region
  Montreal                               1.2 (0.6-2.3)
  (Quebec City)                                1
Age group
  18-29                                  1.9 (0.8-4.4)
  30-44                                  1.1 (0.5-2.3)
  (45+)                                        1
Education
  High school not                        1.2 (0.6-2.2)
  completed
  (High school completed)                      1
Number of income sources
  None or only 1 source                  1.0 (0.5-1.9)
  (2 sources or more)                          1
Residential status
  Currently or formerly homeless         3.5 (1.4-8.6)
  (Never homeless)                             1

                                                OR
Characteristics *                            Adjusted
                                   ([double dagger], [section])

Gender
  Male                                     1.8 (0.8-4.4)
  (Female)                                       1
Region
  Montreal                                 1.2 (0.6-2.3)
  (Quebec City)                                  1
Age group
  18-29                                    1.8 (0.8-4.3)
  30-44                                    1.0 (0.5-2.2)
  (45+)                                          1
Education
  High school not                          1.2 (0.6-2.1)
  completed
  (High school completed)                        1
Number of income sources
  None or only 1 source                    1.2 (0.6-2.3)
  (2 sources or more)                            1
Residential status
  Currently or formerly homeless           3.3 (1.3-8.3)
  (Never homeless)                               1

* Reference category in parentheses

([dagger]) Unweighted number of respondents

([double dagger]) OR = Odds ratio; 95% confidence interval in
parentheses

([section]) Adjusted for other characteristics listed in this table
including residential status.

Table 2 Crude and adjusted odds ratios of food insufficient
individuals reporting poor/fair self rated health and physical health
problems, QSCHRP

                                   Self-rated
                                   health

                                   Poor/fair self-
                                    rated health

n *                                      140
% among, people
  food insufficient                     45.2%
  food sufficient                       27.1%

Crude Odds Ratio ([dagger])
  Food insufficient                 2.2 (1.2-4.2)
  (Food sufficient)                       1

Adjusted Odds Ratio ([dagger])
  Food insufficient                 2.9 (1.5-5.6)
  (Food sufficient)                       1
Sex
  Female                                 --
  (Male)
Age group
  (18-29)                                 1
  30-44                             3.3 (1.3-8.2)
  45+                              7.4 (2:9-19.0)
Education
  High school not completed         2.1 (1.1-4.0)
  (High school completed)                 1
Residential Status
  Currently or formerly homeless         --
  (Never homeless)

                                   Types of physical health problems

                                                Multiple
                                                chronic
                                               conditions

n *                                               187
% among, people
  food insufficient                              64.3%
  food sufficient                                42.3%

Crude Odds Ratio ([dagger])
  Food insufficient                          2.5 (1.3-4.6)
  (Food sufficient)                                1

Adjusted Odds Ratio ([dagger])
  Food insufficient                          2.8 (1.5-5.2)
  (Food sufficient)                                1
Sex
  Female                            1.8 ([double dagger]) (0.9-3.8)
  (Male)                                           1
Age group
  (18-29)                                          1
  30-44                             0.5 ([double dagger]) (0.2-1.0)
  45+                               1.7 ([double dagger]) (0.8-3.7)
Education
  High school not completed                        --
  (High school completed)
Residential Status
  Currently or formerly homeless                   --
  (Never homeless)

                                   Types of physical health problems

                                             Heart disease

n *                                                32
% among, people
  food insufficient                              16.0%
  food sufficient                                 5.9%

Crude Odds Ratio ([dagger])
  Food insufficient                          3.0 (1.0-9.6)
  (Food sufficient)                                1

Adjusted Odds Ratio ([dagger])
  Food insufficient                          5.4 (1.7-16.9)
  (Food sufficient)                                1
Sex
  Female                                           --
  (Male)
Age group
  (18-29)                                          1
  30-44                             1.5 ([double dagger]) (0.2-11.2)
  45+                              17.1 ([double dagger]) (2.7-109.8)
Education
  High school not completed                        --
  (High school completed)
Residential Status
  Currently or formerly homeless    0.6 ([double dagger]) (0.2-1.8)
  (Never homeless)                                 1

                                   Types of physical health problems

                                               Body Mass
                                                 Index:
                                   BMI [greater than or equal to] 30:
                                                 Obese

n *                                                53
% among, people
  food insufficient                              24.9%
  food sufficient                                 7.0%

Crude Odds Ratio ([dagger])
  Food insufficient                          4.4 (1.9-10.6)
  (Food sufficient)                                1

Adjusted Odds Ratio ([dagger])
  Food insufficient                          4.5 (1.8-11.5)
  (Food sufficient)                                1
Sex
  Female                                     2.8 (1.1-7.1)
  (Male)                                           1
Age group
  (18-29)
  30-44
  45+                                              --
Education
  High school not completed                        --
  (High school completed)
Residential Status
  Currently or formerly homeless                   --
  (Never homeless)

* Unweighted number of respondents

([dagger]) Reference category in parentheses; 95% Confidence interval
in parentheses beside odds ratio value

([double dagger]) Variables play a confounding role in the association
between food insufficiency and the corresponding health problem.

Table 3 Crude and adjusted odds ratios of food insufficient
individuals reportingmental health problems, QSCSHP

                                        Types of current mental
                                            health problem

                                        Alcohol
                                        or drug
Measures                                disorder         Depression

n *                                       287                78
% among people
  food insufficient                      79.2%              28.5%
  food sufficient                        62.2%              12.1%

Crude Odds Ratio ([dagger])
  Food insufficient                  2.3 (1.1-4.7)      2.9 (1.4-5.9)
  (Food sufficient)                        1                  1

Adjusted Odds Ratio ([dagger])
  Food insufficient                  1.7 (0.8-3.7)      2.9 (1.4-5.8)
  (Food sufficient)                        1                  1
Sex
  Male                               3.5 (1.7-7.4)           --
  Female                                   1
Age group
  18-29                              2.2 (0.8-6.0)           --
  30-44                              2.0 (1.0-3.9)
  (45+)                                    1
Number of income sources
  None or only 1 source                    --           2.2 (1.1-4.4)
  (2 sources or more)                                         1
Residential status
  Currently or formerly homeless      2.9 ([double           --
                                   dagger]) (1.4-5.7)
  (Never homeless)                         1

                                        Types of current mental
                                            health problem

                                               Emotional
Measures                                       disorder

n *                                               113
% among people
  food insufficient                              42.1%
  food sufficient                                17.3%

Crude Odds Ratio ([dagger])
  Food insufficient                          3.5 (1.7-7.0)
  (Food sufficient)                                1

Adjusted Odds Ratio ([dagger])
  Food insufficient                          3.3 (1.6-6.8)
  (Food sufficient)                                1
Sex
  Male                                             1
  Female                           2.2 ([double dagger]) (1 .0-5.0)
Age group
  18-29                                            1
  30-44                             0.5 ([double dagger]) (0.2-1.1)
  (45+)                             1.5 ([double dagger]) (0.6-3.7)
Number of income sources
  None or only 1 source                           --
  (2 sources or more)
Residential status
  Currently or formerly homeless    2.9 ([double dagger]) (1.1-7.2)
  (Never homeless)                                 1

                                       Types of current mental
                                            health problem

                                                Axis 1
Measures                                       Disorder

n *                                              164
% among people
  food insufficient                             50.9%
  food sufficient                               32.4%

Crude Odds Ratio ([dagger])
  Food insufficient                         2.2 (1.1-4.1)
  (Food sufficient)                               1

Adjusted Odds Ratio ([dagger])
  Food insufficient                         1.9 (1.0-3.6)
  (Food sufficient)                               1
Sex
  Male                                            1
  Female                                    1.3 (0.6-2.7)
Age group
  18-29                                           1
  30-44                                     0.6 (0.3-1.3)
  (45+)                                     1.6 (0.7-3.6)
Number of income sources
  None or only 1 source                           --
  (2 sources or more)
Residential status
  Currently or formerly homeless   3.1 ([double dagger]) (1.4-6.9)
  (Never homeless)                                1

                                   Axis 2
                                   Disorder in
Measures                           their lifetime

n *                                      83
% among people
  food insufficient                    27.3%
  food sufficient                      15.4%

Crude Odds Ratio ([dagger])
  Food insufficient                2.1 (1.0-4.4)
  (Food sufficient)                      1

Adjusted Odds Ratio ([dagger])
  Food insufficient                2.0 (0.9-4.2)
  (Food sufficient)                      1
Sex
  Male                                   --
  Female
Age group
  18-29                            2.9 (1.1-7.6)
  30-44                            2.2 (0.9-5.4)
  (45+)                                  1
Number of income sources
  None or only 1 source                  --
  (2 sources or more)
Residential status
  Currently or formerly homeless         --
  (Never homeless)

* Unweighted number of respondents

([dagger]) Reference category identified with OR= 1; 95% Confidence
interval in parentheses beside odds ratio value.

([double dagger]) Variables play a confounding role in the association
between food insufficiency and the corresponding health problem.

Table 4 Crude and adjusted odds ratios of food insufficient households
for individuals reporting poor-fair self-rated health and physical
health problems in general population, Quebec and Canada, NPHS Cycle 2

                            Self rated health and physical
                            health problems

                            Poor/fair self-rated
Measures                                health

                             Quebec       Canada

Sample size used in           2,163       61,990
  the models
of problem among
respondents living in
the household with:
  food insufficient           18.4%       24.7%
  food sufficient             7.2%         8.4%

Crude OR *
  Food insufficient           2.90         3.60
                            (1.4-6.1)   (3.2-4.0)

Adjusted OR *, ([dagger])
  Food insufficient           1.80         2.70
                            (0.8-4.1)   (2.5-2.9)

                            Self rated health and physical
                            health problems

                            Multiple chronic
Measures                    conditions

                              Quebec      Canada

Sample size used in           2,155       61,599
  the models
of problem among
respondents living in
the household with:
  food insufficient           40.4%        48.4%
  food sufficient             27.5%        29.6%

Crude OR *
  Food insufficient            1.80        2.20
                            (1.1-3.0)    (2.0-2.5)

Adjusted OR *, ([dagger])
  Food insufficient            1.40        2.20
                            (0.8-2.4)    (1.9-2.5)

                            Self rated health and physical
                            health problems

                                   Body Mass Index:
Measures                    BMI [greater than or equal to]
                                      30:Obese

                             Quebec      Canada

Sample size used in           2,086      58,991
  the models
of problem among
respondents living in
the household with:
  food insufficient           13.4%       12.6%
  food sufficient             12.0%       10.7%

Crude OR *
  Food insufficient           1.10        1.20
                            (0.5-2.3)   (1.0-1.5)

Adjusted OR *, ([dagger])
  Food insufficient           0.90        0.90
                            (0.4-1.8)   (.7-1.2)

* 95% Confidence interval in parentheses.

([dagger]) Adjusted for age group, sex, region (Quebec only, Montreal
and rest of Quebec), education (high school completed) and income
adequacy (in four categories: lowest income quartile, lower-middle,
upper-middle and higher) since the whole population is considered
here.
Gale Copyright: Copyright 2009 Gale, Cengage Learning. All rights reserved.