Objectives. We explored the effects of social determinants of health on pandemic H1N1 2009 influenza severity and the role of clinical risk factors in mediating such associations.

Methods. We used multivariate logistic regression with generalized estimating equations to examine the associations between individual- and ecological-level social determinants of health and hospitalization for pandemic H1N1 2009 illness in a case-control study in Ontario, Canada.

Results. During the first pandemic phase (April 23–July 20, 2009), hospitalization was associated with having a high school education or less and living in a neighborhood with high material or total deprivation. We also observed the association with education in the second phase (August 1–November 6, 2009). Clinical risk factors for severe pandemic H1N1 2009 illness mediated approximately 39% of the observed association.

Conclusions. The main clinical risk factors for severe pandemic H1N1 2009 illness explain only a portion of the associations observed between social determinants of health and hospitalization, suggesting that the means by which the social determinants of health affect pandemic H1N1 2009 outcomes extend beyond clinically recognized risk factors.

Similar to seasonal influenza, most cases of pandemic H1N1 2009 influenza were relatively mild; however, certain groups of individuals were at a higher risk of complication and severe disease than were others. Information from early pandemic H1N1 2009 case series indicated that risk factors typically associated with severe seasonal influenza, including underlying conditions such as pulmonary or cardiac disease, diabetes, and pregnancy, were observed among individuals very ill with pandemic H1N1 2009; however, unlike seasonal influenza, severe pandemic H1N1 2009 affected children of all ages and young adults, many of whom were previously healthy.1–6 Obesity and morbid obesity were also recognized as risk factors.2,7,8

In Ontario, Canada, a case-control study was conducted as part of a pan-Canadian approach to pandemic H1N1 2009 research. The purpose of this study was to identify risk factors for pandemic H1N1 2009 infection requiring hospitalization (L. C. R. and N. C., unpublished data, 2010).9 Many of the risk factors identified in our study and others, notably the presence of chronic conditions such as diabetes10–12and obesity13,14 and health behaviors such as smoking15,16 and accessing health care,17 are known to be influenced by the social determinants of health. The social determinants of health are social and economic conditions, such as income, education, employment, and social support, that influence the health of individuals and communities. Disparities in these conditions are reflected in a gradient of socioeconomic status (SES), which, in turn, is associated with inequalities in health. It is widely recognized that individuals at the lower end of the SES gradient experience poorer health and a reduced life expectancy compared with more advantaged groups.18–20 We were therefore interested in exploring the effects of the social determinants of health on pandemic H1N1 2009 severity in Ontario and the role of clinical risk factors in mediating any such associations.

In addition to individual characteristics, contextual socioeconomic factors, such as neighborhood conditions, influence health outcomes. A majority of studies investigating neighborhood effects on health have found significant associations between measures of area SES and health outcomes, such as self-reported health and mortality, that are independent of the effects of individual socioeconomic characteristics.21,22 The mechanisms by which neighborhood characteristics affect the health of individuals are not fully understood, but there are several potential pathways through which ecological exposures may affect the severity of influenza infection. For example, neighborhood problems such as vandalism, illegal drug use, noise, and litter are sources of psychological stress for residents,23–25 and psychological stress is known to influence immune function.26,27 Neighborhood-associated psychological stress could therefore place individuals at increased risk of severe influenza illness. Indeed, Cohen et al. demonstrated that among individuals experimentally infected with influenza, those reporting high psychological stress before inoculation experienced more severe illness than did individuals with low stress.28 Another possibility is that environmental exposures, such as air pollution, contribute to an individual’s risk for severe infection. Traffic-related air pollution is associated with incident asthma,29–32 and asthma was identified as an important risk factor for severe pandemic H1N1 2009 outcomes, including hospitalization.33 This evidence demonstrates that the social determinants of health exert influence on health at the ecological level as well as at the individual level.

We examined the effects of individual- and ecological-level social determinants of health on pandemic H1N1 2009 severity, as indicated by hospitalization, in Ontario, Canada. Furthermore, we explored the role of known clinical risk factors for severe pandemic H1N1 2009 infection in mediating any such associations.

A few studies have investigated the associations between the social determinants of health and respiratory infection outcomes, such as healthcare utilization34 and hospitalization.35–37 The outcomes measured in these studies are not specific to influenza and include outcomes from all or acute respiratory infections35 and influenza and pneumonia diagnoses combined.37 Measuring influenza-specific outcomes is challenging because of the nonspecific presentation of illness and the limited use of laboratory diagnostic testing for medical management. The prevalent use of diagnostic testing for influenza during the H1N1 2009 pandemic, therefore, provided a unique opportunity to examine the effect of the social determinants of health on influencing hospitalization specifically because of this novel strain of influenza.

Ontario, Canada residents of all ages who received nasopharyngeal swabs and tested positive for pandemic H1N1 by real-time polymerase chain reaction were eligible for inclusion in the study. We identified study participants as having confirmed pandemic H1N1 2009 influenza through testing at the Public Health Ontario Laboratory or Mount Sinai Hospital-University Health Network. The Public Health Ontario Laboratory conducted most of the testing in the province throughout the pandemic, and a small number of other laboratories in major centers, including Mount Sinai Hospital-University Health Network, performed testing in the latter portion of phase 1 and throughout phase 2.

We included individuals with a test date from April 23 to July 20, 2009, as phase 1 participants and those with a test date from August 1 to November 6, 2009, as phase 2 participants. Trained interviewers from the Institute for Social Research at York University, Toronto conducted standardized telephone interviews to obtain information on demographics, health status, symptoms, and treatment, including hospitalization. A parent or legal guardian responded as a proxy for a study participant younger than 16 years. Respondents were required to communicate in English for inclusion in the study. Interviews of phase 1 participants were conducted in August and September of 2009, and interviews of phase 2 participants were conducted between November 2009 and January 2010.

We considered individuals who self-reported being hospitalized because of pandemic H1N1 to have had severe illness, and we ascribed them case status. Nonhospitalized individuals served as controls. We excluded individuals who died.

On June 11, 2009, clinicians in Ontario were instructed to switch from submitting samples from all patients presenting with influenza-like illness to submitting samples only from patients at high risk of complications or for the clinical management of hospitalized cases of influenza-like illness.38 To avoid selecting only high-risk individuals tested after June 11 into the phase 1 control group, we excluded nonhospitalized study participants with a test date after June 11, 2009, from the analyses. Laboratory testing was restricted to high-risk and severe cases of illness throughout the second phase of the pandemic; therefore, the control group for phase 2 was representative of a higher risk population and not the population with milder disease that presented for health care but was not eligible for laboratory testing.

Socioeconomic information collected from study participants via telephone interview included ethnicity (including self-reported Aboriginal status), immigrant status (based on country of birth), household density (the ratio of the number of individuals in the household to the number of sleeping rooms), and education level (of adult participants aged 18 years or older and of parent respondents for children younger than 16 years).

We obtained ecological-level information through postal code linkage at the level of the dissemination area (the smallest census geographic unit in Canada, which includes between 400 and 700 individuals) to 2006 Canadian census data39 and the Institut national de santé publique du Québec Deprivation Index.40,41 Census variables we collected were employment rate for total population aged 15 years and older, prevalence of low income after tax among total persons in private households, and total population aged 15 years and older with a high school education or less. We ecologically classified individuals living in dissemination areas where the employment rate was in the bottom quintile of all dissemination area employment rates for the study population as having a low employment rate. We ecologically classified individuals living in dissemination areas where the prevalence of low income, or the population aged 15 years and older with a high school education or less, were in the top quintile for the study population, as high prevalence of low income or as high prevalence of individuals with a high school education or less, respectively.

We used quintiles of exposure for the ecological variables employment rate, prevalence of low income, and prevalence of high school education or less in secondary analyses to explore the effect of alternate forms of these variables.

The Institut national de santé publique du Québec Deprivation Index provides separate indices for social and material deprivation. Each is based on weighted measures of 3 variables from the 2006 Canadian census (social deprivation index: proportion of persons living alone; proportion separated, divorced, or widowed; proportion of single-parent families; material deprivation index: employment to population ratio, average income, and proportion without a high school diploma). The Deprivation Index categorized dissemination areas in the province of Ontario into quintiles of material and social deprivation, with a score of 1 indicating low deprivation and 5 indicating high deprivation. We combined material and social deprivation scores to produce a total deprivation score, which we categorized as high, middle, or low deprivation.42

We developed a conceptual model of the association between the social determinants of health and pandemic H1N1 2009 influenza severity (Figure 1). We used a hierarchical approach so that the framework could enable the examination of the overall effect of the social determinants of health on severity as well as the consideration of mediating factors.43 Previous work to identify risk factors for severe disease guided the framework’s development (L. R. and N. Crowcroft, unpublished data, 2010).9

We identified potentially mediating factors a priori and included smoking, body mass index (BMI; defined as weight in kilograms divided by the square of height in meters), chronic comorbidity, family doctor use, and antiviral treatment. We collected these variables via telephone interview and defined them as follows: smoking status (yes, no, or refused to answer), BMI (World Health Organization BMI category I, II, III, or IV, based on self-reported height and weight), chronic comorbidity (yes or no to the presence of any of the following conditions: diabetes, asthma, heart disease, chronic obstructive pulmonary disease, immune disorder, liver disease, cancer, kidney disease, splenectomy, sickle cell anemia, seizure disorder, cystic fibrosis, cerebral palsy), family doctor use (no family doctor, low use [≤ 1 visit in the past year], moderate use, high use [> 6 visits in the past year]), and antiviral use (none, antiviral treatment initiated within 48 hours of symptom onset, antiviral treatment initiated beyond 48 hours of symptom onset). We did not include immunization status as a potentially mediating factor because of the cutoff date for inclusion in the study. The pandemic H1N1 2009 vaccine campaign began on October 27, 2009, in Ontario. Only individuals with a laboratory test date up to November 6, 2009, were eligible, barring sufficient time for vaccination to take a protective effect.

We calculated descriptive statistics for all variables. We used the χ2 test or the Fisher exact test (when cell counts were < 5) to test for differences in proportions. We conducted analyses stratified by phase because of the change in the population eligible to be tested for pandemic H1N1 2009.

We calculated an age–gender adjusted odds ratio (AOR) by logistic regression for each individual- and ecological-level social determinant of health variable independently. We used generalized estimating equations to account for clustering (dependence) of the data by dissemination area in all ecological-level regressions. We considered exchangeable and unrestricted correlation structures. We assessed statistical significance using 95% confidence intervals (CIs).

When we found a significant association between a social determinants of health variable and hospitalization, we assessed the effects of potentially mediating factors. We added each potential mediator independently to the regression model of interest. We interpreted the remaining association between the social determinants of health and pandemic H1N1 2009 severity as the effect of that social determinant of health not mediated by the risk factor included in the model (i.e., the decrease in the magnitude of the association represented the effect of that social determinant of health that was mediated by the risk factor).43 We identified the 1 or 2 factors that we found to mediate the association to the greatest extent, as demonstrated by the absolute reduction of the odds ratio (OR).

Where we identified a mediating factor, we assessed its association with the social determinant of health variable mediated. As we classified mediating risk factors by either binomial or multinomial categories, we used bi- or multinomial logistic regression, as appropriate. We performed all statistical analyses using SAS version 9.2 (SAS Institute, Cary, NC).

We included 334 (150 hospitalized) individuals in phase 1 and 691 (251 hospitalized) in phase 2. Participant response rates were 63% in phase 1 and 66% in phase 2. Less than 1% of the eligible sample was deceased; the proportions excluded because of language were 3% and 2% in phases 1 and 2, respectively. The individual- and ecological-level socioeconomic characteristics of the study population are described in Table 1.

Table

TABLE 1— Individual- and Ecological-Level Socioeconomic Characteristics of the Study Population by Phase: Pandemic H1N1 Case-Control Study, Ontario, Canada, April 23–November 6, 2009

TABLE 1— Individual- and Ecological-Level Socioeconomic Characteristics of the Study Population by Phase: Pandemic H1N1 Case-Control Study, Ontario, Canada, April 23–November 6, 2009

Phase 1
Phase 2
CharacteristicHospitalized (n = 150), No. (%)Nonhospitalized (n = 184), No. (%)PeHospitalized (n = 251), No. (%)Nonhospitalized (n = 440), No. (%)Pe
Individual level
Age, y.001f≤ .001
 0–852 (34.67)34 (18.48)92 (36.65)126 (28.64)
 9–1838 (25.33)76 (41.30)49 (19.52)139 (31.59)
 19–3420 (13.33)24 (13.04)41 (16.33)98 (22.27)
 35–5018 (12.00)29 (15.76)37 (14.74)49 (11.14)
 51–6413 (8.67)18 (9.78)21 (8.37)20 (4.55)
 ≥ 659 (6.00)3 (1.63)11 (4.38)8 (1.82)
Gender: female66 (44.30)94 (51.65).183122 (48.8)246 (55.91).072
Pregnant (women aged 18–50 y)4 (16.00)2 (5.88).386f4 (8.16)41 (37.27)≤ .001f
Immigrant39 (26.00)48 (26.09).98632 (12.75)56 (12.73).993
Recent immigrant (< 5 y)10 (6.71)12 (6.56).99610 (3.98)13 (2.97).756
Ethnicity
 Whitea72 (48.32)87 (47.54).887163 (64.94)304 (69.41).227
 Aboriginal4 (2.67)1 (0.54).178f12 (4.78)18 (4.09).669
Household density > 2b4 (2.67)3 (1.63).705f3 (1.20)8 (1.82).754f
≤ high school education (among those ≥ 18 y)31 (47.69)23 (29.49).02563 (56.76)73 (40.56).007
Parental education ≤ high school (among those < 16 y)14 (18.42)17 (17.89).92934 (26.77)49 (21.59).269
Ecological level
Material deprivationc.032.296
 1 (low)20 (14.71)47 (26.86)36 (15.38)82 (19.07)
 233 (24.26)43 (24.57)45 (19.23)91 (21.16)
 325 (18.38)31 (17.71)44 (18.80)94 (21.86)
 426 (19.12)32 (18.29)57 (24.36)84 (19.53)
 5 (high)32 (23.53)22 (12.57)52 (22.22)79 (18.37)
Social deprivationc.247.073
 1 (low)39 (28.68)50 (28.57)32 (13.68)81 (18.84)
 225 (18.38)42 (24.00)53 (22.65)101 (23.49)
 327 (19.85)28 (16.00)64 (27.35)81 (18.84)
 421 (15.44)36 (20.57)45 (19.23)78 (18.14)
 5 (high)24 (17.65)19 (10.86)40 (17.09)89 (20.70)
Total deprivationc.034.066
 1–2 (low)24 (17.65)46 (26.29)37 (15.81)79 (18.37)
 3–7 (middle)87 (63.97)112 (64.00)143 (61.11)283 (65.81)
 8–9 (high)25 (18.38)17 (9.71)54 (23.08)68 (15.81)
Low employment rated35 (23.33)28 (15.22).05963 (25.10)103 (23.41).617
High prevalence of low incomed46 (30.67)45 (24.46).20544 (17.53)66 (15.00).382
High prevalence of ≤ high school educationd26 (17.33)13 (7.07).00452 (20.72)108 (24.55).251

aWhite ethnicity indicates that the respondent self-identified as belonging to the White ethnic group. The remaining participants belonged to the following ethnic groups: Aboriginal, Chinese, South Asian, Black, Filipino, Latin American, Southeast Asian, Arab, West Asian, Korean, Japanese, mixed, or other (specified).

bThe ratio of the number of individuals in the household to the number of sleeping rooms in the household.

cAs measured using the Institut national de santé publique du Québec Deprivation Index.

dBased on 2006 Canadian census data. Outcome categorized by the extreme quintile, e.g., a dissemination area was considered to have a low employment rate if it was in the bottom quintile of all dissemination area employment rates.

eWe used the χ2 test unless otherwise noted.

fWe used the Fisher exact test because of small cell count.

Individuals aged 18 years or younger accounted for nearly 60.0% of the study population in each phase. Only 3.0% of the study population was aged 65 years or older. Female participants constituted just more than half of the study population (51.5%).

More than 25.0% of the study population in phase 1 were immigrants; there were half as many immigrants (12.7%) in the phase 2 sample. Approximately half of the study population was of White ethnicity in phase 1, and this proportion increased to approximately 68.0% in phase 2. The number of Aboriginal people included in the study was small (n = 35), with 86.0% of these individuals in phase 2.

Of individuals aged 18 years or older in the first phase, 38% had a high school education or less. The proportion of individuals with a high school education or less was higher in phase 2, at 47%. The parents who responded as proxies for their ill children (younger than 16 years) were more highly educated than were the adult cases.

Individual- and Ecological-Level Social Determinants of Health

Age–gender AORs for hospitalization for each individual- and ecological-level social determinant of health variable examined are shown in Table 2.

Table

TABLE 2— Odds Ratios for Hospitalization for Each Individual- and Ecological-Level Social Determinant of Health Variable Examined, Adjusted for Age and Gender: Pandemic H1N1 Case-Control Study, Ontario, Canada, April 23–November 6, 2009

TABLE 2— Odds Ratios for Hospitalization for Each Individual- and Ecological-Level Social Determinant of Health Variable Examined, Adjusted for Age and Gender: Pandemic H1N1 Case-Control Study, Ontario, Canada, April 23–November 6, 2009

VariablePhase 1, OR (95% CI)Phase 2, OR (95% CI)
Immigrant1.01 (0.61, 1.67)0.91 (0.56, 1.46)
Recent immigrant (< 5 y)1.01 (0.42, 2.42)1.44 (0.62, 3.37)
Ethnicity
 Whitea1.08 (0.68, 1.69)0.82 (0.59, 1.15)
 Aboriginal4.96 (0.54, 45.38)1.25 (0.59, 2.65)
Household density > 2b1.70 (0.37, 7.84)0.61 (0.16, 2.34)
≤ high school education
 Adults (aged ≥ 18 y)2.28 (1.13, 4.59)1.77 (1.08, 2.89)
 Parents of children aged < 16 y0.98 (0.41, 2.35)1.32 (0.79, 2.21)
Material deprivationc
 1 (low; Ref)1.001.00
 21.15 (0.68, 1.94)1.15 (0.67, 1.98)
 31.81 (0.80, 4.07)1.05 (0.61, 1.82)
 41.98 (0.97, 4.07)1.53 (0.91, 2.58)
 5 (high)3.46 (1.65, 7.26)1.54 (0.90, 2.65)
Social deprivationc
 1 (low; Ref)1.001.00
 20.73 (0.38, 1.39)1.39 (0.81, 2.39)
 30.58 (0.38, 0.89)1.98 (1.16, 3.38)
 40.67 (0.35, 1.31)1.48 (0.85, 2.58)
 5 (high)1.66 (0.79, 3.46)1.12 (0.63, 1.98)
Total deprivationc
 1–2 (low; Ref)1.001.00
 3–7 (middle)0.97 (0.60, 1.58)1.05 (0.67, 1.64)
 8–9 (high)2.58 (1.24, 5.35)1.65 (0.97, 2.82)
Low employment rated1.60 (0.91, 2.81)0.99 (0.68, 1.45)
High prevalence of low incomed1.49 (0.94, 2.36)1.19 (0.79, 1.79)
High prevalence of ≤ high school educationd3.03 (1.45, 6.32)0.85 (0.58, 1.25)

Note. CI = confidence interval; OR = odds ratio. Each OR presented is based on a unique logistic regression model.

aWhite ethnicity indicates that the respondent self-identified as belonging to the White ethnic group. The remaining participants belonged to the following ethnic groups: Aboriginal, Chinese, South Asian, Black, Filipino, Latin American, Southeast Asian, Arab, West Asian, Korean, Japanese, mixed, or other (specified).

bThe ratio of the number of individuals in the household to the number of sleeping rooms in the household.

cAs measured using the Institut national de santé publique du Québec Deprivation Index.

dBased on 2006 Canadian census data. Outcome categorized by the extreme quintile (e.g., a dissemination area was considered to have a low employment rate if it was in the bottom quintile of all dissemination area employment rates).

At the individual level, in both pandemic phases, adults hospitalized for pandemic H1N1 2009 were more likely to have a high school education or less than were nonhospitalized individuals (phase 1: OR = 2.3; 95% CI = 1.1, 4.6; phase 2: OR = 1.8; 95% CI = 1.1, 2.9). We observed a large but nonsignificant association between Aboriginal ethnicity and hospitalization in phase 1 after controlling for age and gender (OR = 5.0; 95% CI = 0.5, 45.4).

At the ecological level in phase 1, hospitalized individuals were significantly more likely to reside in the most materially deprived neighborhoods (OR = 3.5; 95% CI = 1.7, 7.3), neighborhoods with high total deprivation (OR = 2.6; 95% CI = 1.2, 5.4), and neighborhoods with a high proportion of the population having at most a high school education (OR = 3.0; 95% CI = 1.5, 6.3). We also observed hospitalization to be associated with moderate social deprivation (3rd quintile) in both phases; however, these associations were in opposite directions (phase 1: OR = 0.6; 95% CI = 0.4, 0.9; phase 2: OR = 2.0; 95% CI = 1.2, 3.4).

In secondary analyses using quintiles of exposure for ecological variables, the association between hospitalization and employment rate became more pronounced. Hospitalized individuals were significantly more likely to reside in neighborhoods with an employment rate in the lowest quintile compared with the highest quintile (OR = 2.2; 95% CI = 1.4, 3.5). Associations with ecological measures of low income and education remained consistent (data not shown).

Mediators

The relative contributions of mediating risk factors to the observed associations between social determinants of health and pandemic H1N1 2009 severity are illustrated in Figure 2. Mediators of the associations between the social determinants of health and pandemic H1N1 2009 severity accounted for, at most, 65% of the observed association and as little as 14%.

The risk factors of smoking and antiviral treatment mediated nearly one third of the observed association between high material deprivation and hospitalization during the first phase of the pandemic. Relative to individuals from the least materially deprived neighborhoods, individuals from the most materially deprived neighborhoods were somewhat more likely to smoke (OR = 2.4; 95% CI = 0.8, 7.6), less likely to have received antiviral treatment (OR = 0.8; 95% CI = 0.3, 1.9), and—if they received antiviral treatment—more likely to have received treatment more than 48 hours after symptom onset (OR = 1.9; 95% CI = 0.3, 11.40).

BMI and antiviral treatment mediated the association between high total deprivation and hospitalization. Individuals residing in neighborhoods with high total deprivation were 4 times more likely than were individuals from neighborhoods with low total deprivation to be overweight, compared with normal weight (OR = 4.0; 95% CI = 1.1, 14.5). They were slightly less likely to receive antiviral treatment (OR = 0.8; 95% CI = 0.3, 2.2) and slightly more likely to receive treatment after 48 hours from symptom onset than within 48 hours (OR = 1.2; 95% CI = 0.1, 16.4).

Individuals hospitalized for pandemic H1N1 2009 illness in phase 1 were more likely to come from neighborhoods where a high proportion of the population had a high school education or less. This association was mediated by both BMI and chronic comorbidity. Individuals from these neighborhoods were 2.6 times more likely to be obese than normal weight (OR = 2.6; 95% CI = 1.03, 6.6). They were also approximately 40% more likely to have a chronic comorbidity (OR = 1.4; 95% CI = 0.7, 3.0).

The risk factors of smoking and antiviral treatment mediated approximately 40% of the observed association between education level (among individuals aged 18 years and older) and hospitalization during phase 1of the pandemic. Individuals with a high school education or less were 3.5 times significantly more likely to smoke than were more educated individuals (OR = 3.5; 95% CI = 1.5, 8.1). They were more likely to not receive antiviral treatment (OR = 1.3; 95% CI = 0.3, 5.0) or to receive treatment after more than 48 hours of symptom onset (OR = 1.2; 95% CI = 0.2, 6.1) than to receive treatment within 24 hours of symptom onset.

During the second phase, the association between education level (among individuals aged 18 years and older) and hospitalization was mediated by smoking and BMI. Individuals with a high school education or less were 3 times more likely to smoke (OR = 2.9; 95% CI = 1.7, 4.9) and were more likely to be obese than normal weight (OR = 1.4; 95% CI = 0.8, 2.5) than were individuals with a higher level of education.

Smoking and family doctor use mediated a small amount of the association observed between moderate social deprivation and hospitalization during the second phase. Individuals living in neighborhoods with moderate levels of social deprivation were more likely to smoke than were individuals living in the least socially deprived neighborhoods (OR = 1.5; 95% CI = 0.7, 3.3). They also appeared more likely to report high, versus moderate, family doctor use (OR = 1.9; 95% CI = 0.9, 3.9).

Hospitalization for pandemic H1N1 2009 illness was associated with several important social determinants of health. Hospitalized individuals in this study were significantly more likely than were nonhospitalized individuals to have a lower education level (high school education or less). They were also more likely to live in neighborhoods with high total or material deprivation or neighborhoods that have high proportions of individuals with a high school education or less. Risk factors for severe pandemic H1N1 2009 illness, namely smoking, antiviral treatment, and BMI, partially mediated the associations between hospitalization and the social determinants of health; however, most of the association we observed remains unexplained.

It is compelling that we observed the association between lower education level and pandemic H1N1 2009 severity for both individual and ecological measures of education. Education is a frequently used indicator of SES in epidemiological research because it is easy to measure and reflects material, intellectual, and other resources across the life course.44 Other common indicators of SES, such as income- and occupation-based measures, were unavailable at the individual level. The use of ecological-level indices of deprivation was useful in exploring the impact of complex neighborhood characteristics, which include measures of income, employment, and social support on pandemic H1N1 2009 outcomes.

Our findings are in keeping with those reported for other studies of socioeconomic factors and acute respiratory infections. In the United Kingdom, hospital admission rates for pneumonia, acute respiratory infections, and all infections combined were found to be approximately 30% to 40% higher among individuals from the most materially deprived areas than the least materially deprived areas when deprivation was assessed using the Townsend deprivation score.35 Older individuals who were socially isolated were 4.5 times more likely to be hospitalized for winter respiratory infection than were those who had regular social support (OR = 4.5; 95% CI = 1.3, 15.8).45 Conversely, Charland et al. observed a protective effect of social deprivation for emergency department and outpatient clinic visits for seasonal influenza-like illness.34 Socially isolated individuals may experience a lower risk of exposure to infection, but the infections that do occur in this population may be more severe. Alternately, as Charland et al. point out, physicians might be more likely to admit patients who lack social support if they were unlikely to receive care at home.34

Glezen et al. found that the hospitalization rate for acute respiratory conditions among individuals with underlying chronic conditions was almost 7 times higher among individuals from low-income areas than among individuals from middle-income areas.36 Low income was not found to be associated with hospitalization rates for influenza and pneumonia in an Ontario study by Crighton et al., but rates were significantly related to the proportion of the population that was Aboriginal and the proportion with less than a high school education.37 In a study of severe pandemic H1N1 2009 illness in Manitoba, Canada, First Nations ethnicity was strongly associated with being admitted to the hospital intensive care unit (vs not hospitalized; OR = 3.73; 95% CI = 1.92, 7.16).46

All participants included in this study had laboratory-confirmed pandemic H1N1 2009 infection, which is an important strength of this study, as diagnostic confirmation of the etiological agent of respiratory infection is frequently unavailable in observational studies.

The selection of cases and controls into this study depended on the laboratory testing algorithm that dictated who was tested for pandemic H1N1 2009 during the pandemic. The study population is therefore representative of the population of individuals ill with pandemic H1N1 2009 in Ontario who sought medical care and was tested for pandemic H1N1 2009 infection. Because of the change in algorithms resulting in testing high-risk and hospitalized cases of influenza-like illness throughout the second phase, the findings we have reported for the phase 2 study population (cases and controls) can only be generalized to this high-risk population. Because of the testing algorithm, the phase 2 study population was likely more homogeneous with respect to clinical risk factors for severe disease than was the phase 1 population. As these risk factors are tied, as least to some extent, to the social determinants of health, the magnitudes of the associations between the social determinants of health and hospitalization may have been obscured or apparently attenuated during the second phase.

For the majority of the pandemic, anyone in Ontario who was ill enough to be hospitalized with influenza-like illness was very likely to be tested for pandemic H1N1 2009. It is possible, however, that people who were not ill enough to be hospitalized but who were tested differed with respect to the social determinants of health from those who were equally ill but were not tested. For example, the health care–seeking behaviors, and consequently the testing patterns, may have varied by SES. If this were the case, although hospitalized cases should have been representative of the true socioeconomic gradient for disease, nonhospitalized cases may have differed in SES from the general ill population. Because of universally funded health care in Canada, individuals with lower income or education access health care services more than do individuals of higher SES.17,47 It is therefore unlikely that control selection was biased toward higher SES in a meaningful way in this study; rather, the selection of lower SES controls may have been favored, which would have biased ORs toward 1.00.

Limitations

Individual-level data were collected by self-report, which can lead to misclassification. Outcome classification (case or control) was based on self-reported hospitalization status; therefore, misclassification may have occurred because of differing individual understandings of hospital admission (vs, e.g., an emergency department visit). The potential for recall bias regarding exposures was low because the variables used in these analyses (ethnicity, education, smoking, etc.) are relatively stable over time. Additionally, ecological social determinants of health variables were not self-reported and thus not affected by self-report misclassification, but it is important to remember that an ecological measure cannot be ascribed to an individual.

Conclusions

The contribution of our study to the body of research regarding the social determinants of health and influenza and respiratory outcomes is significant in that we examined the effects of both individual- and ecological-level social determinants of health on an outcome specific to pandemic H1N1 2009 influenza and that we considered the mediating effects of recognized risk factors, which, to our knowledge, have not previously been explored. We observed that the main clinical risk factors for severe pandemic H1N1 2009 illness explain only a small portion of the associations observed between various socioeconomic indicators and hospitalization. This suggests that the means by which the social determinants of health affect pandemic H1N1 2009 outcomes extend beyond influencing these recognized risk factors. Although further research is required to determine if this is also the case for seasonal influenza, we believe that, given the large proportion of the observed associations not explained by individual risk and behavioral factors, a social determinants approach to promoting public health is an essential component of pandemic planning, and is crucial for mitigating the burden of severe influenza illness.

Acknowledgments

The Public Health Agency of Canada and the Ontario Ministry of Health and Long-Term Care provided funding for this work.

This work was previously presented in poster format at the Third North American Congress of Epidemiology, June 2011, in Montreal, Quebec, Canada.

The authors would like thank Steven Johnson, geospatial analyst at Public Health Ontario, for his assistance with the collection of ecological data and David Northrup and the staff at the Institute for Social Research for conducting the surveys required for the case-control study.

Human Participant Protection

Institutional review board approval was received from the University of Toronto Research Ethics Board.

References

1. Centers for Disease Control and Prevention. Hospitalized patients with novel influenza A (H1N1) virus infection—California, April–May, 2009. MMWR Morb Mortal Wkly Rep. 2009;58(19):536541. MedlineGoogle Scholar
2. Kumar A, Zarychanski R, Pinto R, et al. Critically ill patients with 2009 influenza A (H1N1) infection in Canada. JAMA. 2009;302(17):18721879. Crossref, MedlineGoogle Scholar
3. Novel Swine-Origin Influenza A (H1N1) Virus Investigation Team; Dawood FS, Jain S, Finelli L, et al. Emergence of a novel swine-origin influenza A (H1N1) virus in humans. N Engl J Med. 2009;360(25):2605–2615 Google Scholar
4. Chowell G, Bertozzi SM, Colchero MA, et al. Severe respiratory disease concurrent with circulation of H1N1 influenza. N Engl J Med. 2009;361(7):674679. Crossref, MedlineGoogle Scholar
5. Perez-Padilla R, de la Rosa-Zamboni D, Ponce de Leon S, et al. Pneumonia and respiratory failure from swine-origin influenza A (H1N1) in Mexico. N Engl J Med. 2009;361(7):680689. Crossref, MedlineGoogle Scholar
6. Domínguez-Cherit G, Lapinsky SE, Macias AE, et al. Critically ill patients with 2009 influenza A (H1N1) in Mexico. JAMA. 2009;302(17):18801887. Crossref, MedlineGoogle Scholar
7. Centers for Disease Control and Prevention. Intensive-care patients with severe novel influenza A (H1N1) virus infection—Michigan, June 2009. MMWR Morb Mortal Wkly Rep. 2009;58(27):749752. MedlineGoogle Scholar
8. Miller RR, Markewitz BA, Rolfs RT, et al. Clinical findings and demographic factors associated with ICU admission in Utah due to novel 2009 influenza A (H1N1) infection. Chest. 2010;137(4):752758. Crossref, MedlineGoogle Scholar
9. Skowronski DM, De Serres G, Crowcroft NS, et al. Association between the 2008–09 seasonal influenza vaccine and pandemic H1N1 illness during Spring–Summer 2009: four observational studies from Canada. PLoS Med. 2010;7(4):e1000258. Google Scholar
10. Ross NA, Gilmour H, Dasgupta K. 14-year diabetes incidence: the role of socio-economic status. Health Rep. 2010;21(3):1928. MedlineGoogle Scholar
11. Dasgupta K, Khan S, Ross NA. Type 2 diabetes in Canada: concentration of risk among most disadvantaged men but inverse social gradient across groups in women. Diabet Med. 2010;27(5):522531. Crossref, MedlineGoogle Scholar
12. Smith BT, Lynch JW, Fox CS, et al. Life-course socioeconomic position and type 2 diabetes mellitus. The Framingham Offspring Study. Am J Epidemiol. 2011;173(4):438447. Crossref, MedlineGoogle Scholar
13. McLaren L. Socioeconomic status and obesity. Epidemiol Rev. 2007;29(1):2948. Crossref, MedlineGoogle Scholar
14. Sobal J, Stunkard AJ. Socioeconomic status and obesity: a review of the literature. Psychol Bull. 1989;105(2):260275. Crossref, MedlineGoogle Scholar
15. Reid JL, Hammond D, Driezen P. Socio-economic status and smoking in Canada, 1999–2006: has there been any progress on disparities in tobacco use?Can J Public Health. 2010;101(1):7378. MedlineGoogle Scholar
16. Centers for Disease Control and Prevention. Cigarette smoking among adults—United States, 2002. MMWR Morb Mortal Wkly Rep. 2004;53(20):427431. MedlineGoogle Scholar
17. Alter DA, Stukel T, Chong A, et al. Lesson from Canada’s universal care: socially disadvantaged patients use more health services, still have poor health. Health Aff. 2011;30(2):274283. CrossrefGoogle Scholar
18. McIntosh CN, Fines P, Wilkins R, et al. Income disparities in health-adjusted life expectancy for Canadian adults, 1991 to 2001. Health Rep. 2009;20(4):5564. MedlineGoogle Scholar
19. Mackenbach JP, Stirbu I, Roskam AJR, et al. Socioeconomic inequalities in health in 22 European countries. N Engl J Med. 2008;358(23):24682481. Crossref, MedlineGoogle Scholar
20. Centers for Disease Control and Prevention. Forward: CDC health disparities and inequalities report—United States, 2011. MMWR Morb Mortal Wkly Rep. 2011;60(suppl):12. Google Scholar
21. Riva M, Gauvin L, Barnett TA. Toward the next generation of research in small area effects on health: a synthesis of multilevel investigations published since July 1998. J Epidemiol Community Health. 2007;61(10):853861. Crossref, MedlineGoogle Scholar
22. Pickett KE, Pearl M. Multilevel analyses of neighbourhood socioeconomic context and health outcomes: a critical review. J Epidemiol Community Health. 2001;55(2):111122. Crossref, MedlineGoogle Scholar
23. Ellaway A, Morris G, Curtice J, et al. Associations between health and different types of environmental incivility: a Scotland-wide study. Public Health. 2009;123(11):708713. Crossref, MedlineGoogle Scholar
24. Warr D, Feldman P, Tacticos T, Kelaher M. Sources of stress in impoverished neighbourhoods: insights into links between neighbourhood environments and health. Aust N Z J Public Health. 2009;33(1):2533. Crossref, MedlineGoogle Scholar
25. Steptoe A, Feldman PJ. Neighborhood problems as sources of chronic stress: development of a measure of neighborhood problems, and associations with socioeconomic status and health. Ann Behav Med. 2001;23(3):177185. Crossref, MedlineGoogle Scholar
26. Nakata A, Irie M, Takahashi M. Psychological distress, depressive symptoms, and cellular immunity among healthy individuals: a 1-year prospective study. Int J Psychophysiol. 2011;81(3):191197. Crossref, MedlineGoogle Scholar
27. Segerstrom SC, Miller GE. Psychological stress and the human immune system: a meta-analytic study of 30 years of inquiry. Psychol Bull. 2004;130(4):601330. Crossref, MedlineGoogle Scholar
28. Cohen S, Doyle WJ, Skoner DP. Psychological stress, cytokine production, and severity of upper respiratory illness. Psychosom Med. 1999;61(2):175180. Crossref, MedlineGoogle Scholar
29. Carlsten C, Dybuncio A, Becker A, et al. Traffic-related air pollution and incident asthma in a high-risk birth cohort. Occup Environ Med. 2011;68(4):291295. Crossref, MedlineGoogle Scholar
30. Künzli N, Bridevaux PO, Liu LJ, et al. Traffic-related air pollution correlates with adult-onset asthma among never-smokers. Thorax. 2009;64(8):664670. Crossref, MedlineGoogle Scholar
31. Jerrett M, Shankardass K, Berhane K, et al. Traffic-related air pollution and asthma onset in children: a prospective cohort study with individual exposure measurement. Environ Health Perspect. 2008;116(10):14331438. Crossref, MedlineGoogle Scholar
32. Modig L, Järvholm B, Rönnmark E, et al. Vehicle exhaust exposure in an incident case-control study of adult asthma. Eur Respir J. 2006;28(1):7581. Crossref, MedlineGoogle Scholar
33. Van Kerkhove MD, Vandemaele KA, Shinde V, et al. Risk factors for severe outcomes following 2009 influenza A (H1N1) infection: a global pooled analysis. PLoS Med. 2011;8(7):e1001053. Crossref, MedlineGoogle Scholar
34. Charland KM, Brownstein JS, Verma A, et al. Socio-economic disparities in the burden of seasonal influenza; the effect of social and material deprivation on rates of influenza infection. PLoS ONE. 2011;6(2):e17207. Crossref, MedlineGoogle Scholar
35. Hawker JI, Olowokure B, Sufi F, et al. Social deprivation and hospital admission for respiratory infection: an ecologic study. Respir Med. 2003;97(11):12191224. Crossref, MedlineGoogle Scholar
36. Glezen WP, Greenberg SB, Atmar RL, et al. Impact of respiratory virus infections on persons with chronic underlying conditions. JAMA. 2000;283(4):499505. Crossref, MedlineGoogle Scholar
37. Crighton EJ, Elliott SJ, Moineddin R, et al. A spatial analysis of the determinants of pneumonia and influenza hospitalizations in Ontario (1992–2001). Soc Sci Med. 2007;64(8):16361650. Crossref, MedlineGoogle Scholar
38. Ontario Ministry of Health and Long-Term Care. Important health notice—June 11, 2009. Available at: http://www.health.gov.on.ca/english/providers/program/emu/health_notices/ihn_20090611.pdf. Accessed May 15, 2011. Google Scholar
39. Statistics Canada. 2006 Census release topics. Available at: http://www12.statcan.gc.ca/census-recensement/2006/rt-td/index-eng.cfm. Accessed June 22, 2011. Google Scholar
40. Pampalon R, Hamel D, Gamache P, et al. A deprivation index for health planning in Canada. Chronic Dis Can. 2009;29(4):178191. MedlineGoogle Scholar
41. Institut national de santé publique du Québec. Indice de défavorisation. Canada, 2006.Available at: http://www.inspq.qc.ca/santescope/indicedefavo.asp?NoIndD=9. Accessed June 23, 2011. Google Scholar
42. Canadian Institute for Health Information. Reducing gaps in health: a focus on socio-economic status in urban Canada. Ottawa, ON; 2008. Google Scholar
43. Victora CG, Huttly SR, Fuchs SC, et al. The role of conceptual frameworks in epidemiological analysis: a hierarchical approach. Int J Epidemiol. 1997;26(1):224227. Crossref, MedlineGoogle Scholar
44. Galobardes B, Shaw M, Lawlor DA, et al. Indicators of socioeconomic position (part 1). J Epidemiol Community Health. 2006;60(1):712. Crossref, MedlineGoogle Scholar
45. Jordan RE, Hawker JI, Ayres JG, et al. Effect of social factors on winter hospital admission for respiratory disease: a case-control study of older people in the UK. Br J Gen Pract. 2008;58(551):400402. Crossref, MedlineGoogle Scholar
46. Zarychanski R, Stuart TL, Kumar A, et al. Correlates of severe disease in patients with 2009 pandemic influenza (H1N1) virus infection. CMAJ. 2010;182(3):257264. Crossref, MedlineGoogle Scholar
47. Khan Y, Glazier RH, Moineddin R, et al. A population-based study of the association between socioeconomic status and emergency department utilization in Ontario, Canada. Acad Emerg Med. 2011;18(8):836843. Crossref, MedlineGoogle Scholar

Related

No related items

TOOLS

Downloaded 914 times

SHARE

ARTICLE CITATION

Elizabeth C. Lowcock, MPH, Laura C. Rosella, PhD, MHSc, Julie Foisy, MPH, Allison McGeer, MD, MSc, and Natasha Crowcroft, MD, MSc, MAElizabeth C. Lowcock, Laura C. Rosella, and Natasha Crowcroft are with the Department of Surveillance and Epidemiology, Public Health Ontario, Toronto, Canada, and the Dalla Lana School of Public Health, University of Toronto, Toronto, Ontario. Julie Foisy is with the Department of Surveillance and Epidemiology, Public Health Ontario. Allison McGeer is with the Department of Microbiology, Mount Sinai Hospital, Toronto, Ontario, and the Dalla Lana School of Public Health. “The Social Determinants of Health and Pandemic H1N1 2009 Influenza Severity”, American Journal of Public Health 102, no. 8 (August 1, 2012): pp. e51-e58.

https://doi.org/10.2105/AJPH.2012.300814

PMID: 22698024