Questions to answer

Commentary

16 ! Public Health Reports / 2011 Supplement 3 / Volume 126

Including Gender in Public Health Research

Susan P. Phillips, MD, MSc, CCFPa

aQueen’s University, Department of Family Medicine and Department of Community Health and Epidemiology, Kingston, Ontario, Canada

Address correspondence to: Susan P. Phillips, MD, MSc, CCFP, Queen’s University, Department of Family Medicine, 220 Bagot St., Kingston, ON K7L 5E9, Canada; tel. 613-533-9303; fax 613-549-5403; e-mail <[email protected]>.

©2011 Association of Schools of Public Health

Diversity in both biological attributes and the external, lived environment gives rise to different susceptibilities, exposures, health outcomes, and longevity. Public policy can modify the effects of external differences, if groups at great- est risk are identified and pathways to excess vulnerability are understood, by rebalancing and redistributing the inputs or social determinants that work their way under the skin to ultimately cause biological disadvantage. In the past three decades, a large volume of research has identified the nature of these social determinants of health—including income, socioeconomic status (SES), income inequality, social connectedness, and social capital—and the pathways by which they undermine or reinforce innate health. Often listed among these, but rarely studied, is gender. Medical research may identify sex differences when they exist; however, the varied social roles, expectations, and constraints experienced by men and women in a given society go well beyond the individual and sex differences and are rarely examined as inputs responsible for variation in health outcomes. As a result, health-affirming policies tend to homogenize groups (e.g., assuming that all women are the same) or target individual behaviors, and do so in a gender-blind fashion rather than address- ing structural biases and inequities that undermine those behaviors. This article explores the nature of gender as a determinant of health and describes how the effects of gender inequities can be included in health outcomes research that can then shape health planning and policy.

GENDER AND GENDER EQUITY AS SOCIAL DETERMINANTS OF HEALTH

In 2008, the World Health Organization’s Commission on Social Determinants of Health reported, “It is not an unfortunate cluster of random events nor differences in individual behaviors that consistently keep the health of some countries and population groups below others. Where systematic differences in health are judged to be avoidable by reasonable action globally and within society, they are, quite simply, unjust. It is this that we label health inequity.”1

Universal among such inequities is the gender gap—that is, the different socially sanctioned options, opportunities, and realities men and women face.

Including Gender in Public Health Research ! 17

Public Health Reports / 2011 Supplement 3 / Volume 126

In this context, the term “sex” will be used to connote the biology of being female or male. Gender will refer to roles each sex assumes within a specific group, set- ting, culture, or country, and to associated hierarchies, power relations, differential access to resources, and divisions of labor.2,3 For example, bearing children is a sex role; however, taking responsibility for parenting is a gender role. There is often some overlap between the two. Gender roles may arise from biological impera- tives and, conversely, social behaviors and actions may shape biology. Animal studies have demonstrated that aggressive, risk-taking male behavior likely stems from the biological priority of mating and reproducing.4,5 In what has become a central article documenting how culture and gender determine biology, Fausto-Sterling shows that bone strength is intimately linked to gender roles and constraints.6

Despite the ubiquitous nature of gender and its associated inequities, research describing its effects on health outcomes, particularly on interactions between gender and sex, SES, and environmental antecedents of disease, is minimal.7 One of the difficulties in reviewing existing scholarship on methodologies for including gender in research is finding the appropriate studies to review. Using “gender” as a search term is problematic for several reasons. First, “gender” itself is not a U.S. National Library of Medicine Medical Subject Head- ing (or MeSH term) but maps onto “gender identity,” defined as “a person’s concept of self as being male and masculine or female and feminine, or ambivalent, based in part on physical characteristics, parental responses, and psychological and social pressures. It is the internal experience of gender role.”8 An inherently contextual or group-level characteristic thus translates into an individual trait.

Even more elusive is gender equity, defined by the World Health Organization as “fairness and justice in the distribution of benefits and responsibilities between women and men” and different from gender equality (the absence of discrimination).9 “Gender equity” does not map onto any term, while “equity” alone is interpreted to mean pay equity or is related to taxation. Secondly, conflation and confusion exist throughout the medical and scientific literature between sex and gender. Authors often stratify their findings by sex and then label these results as gender differences. Although sex-disaggregated outcomes associated with particular inputs are a component of good research, they do not identify whether gender norms or constraints are independent determinants of health.

To include any social determinants in population health research requires moving beyond the decontex- tualized study of individual risk factors and examining

in vivo outcomes where multiple interacting inputs contribute to a web of causation.7 As DeFur et al. have written, “Health outcomes are predicted by the relationships among measures of environmental condi- tions (stressors), receptor characteristics (measures of potential vulnerability), and receptor resources (abili- ties to respond or recover).”10

The concept of a single causative agent is, never- theless, appealing because embedded within it is the prospect of a resolution for the disease outcome via eradication or immobilization of that agent. The social determinants approach is more nuanced, presenting challenges for arresting illness. Such is the tension between causes and contexts, with the cause seeming to be more amenable to control and perhaps more acquiescent to change than is the context—the social milieu. However, although risk-factor approaches may identify associations between individual independent variables and likelihood of disease, only an exploration of the context in which those risk factors operate will explain why they activate disease processes in some circumstances but not in others.

Two examples of how gender interacts with biologi- cal predisposition may clarify this concept. Lupus is a condition that appears to have a non-sex-linked genetic risk factor, predisposing carriers to, but not resulting in, illness. Some as yet unidentified infectious or envi- ronmental trigger seems to “turn on” the genetic switch and lead to actual disease.11,12 Although the incidence of lupus among women is nine times that among men, research on gender differences shaping this disparity is lacking. Might there be some explanatory factor related to gender roles or equity? Could the study of interactions among sex, measures of gender, and biological antecedents shed some light on what predis- poses women to, and protects men from, this illness?

As a second example, whenever a woman in sub- Saharan Africa dies of acquired immunodeficiency syndrome (AIDS), the causative agent is human immu- nodeficiency virus. The latter is the necessary but not sufficient explanation for her demise. Of equal or greater etiological significance are social norms that accept forced sexual activity and the powerlessness of women to control circumstances surrounding sexual activity. Risk-factor analysis can identify individuals whose bodies are more able to resist infection or whether circumcision diminishes transmission of the virus, but only an analysis of social circumstances such as gender inequalities will explain why the incidence of AIDS varies so dramatically with geography. Epi- demiologist Geoffrey Rose concluded the following in his classic 1985 article, “Sick Individuals and Sick Populations”:

18 ! Commentary

Public Health Reports / 2011 Supplement 3 / Volume 126

Case-centered epidemiology identifies individual susceptibility, but it may fail to identify the underly- ing causes of incidence. The “high-risk” strategy of prevention is an interim expedient, needed to protect susceptible individuals, but only for so long as the underlying causes of incidence remain unknown or uncontrollable; if causes can be removed, susceptibility ceases to matter.13

MEASURING GENDER AND GENDER EQUITY: METHODS AND FINDINGS

Aspects of gender and gender equality can be mea- sured at the individual, group, or mixed levels, as well as by examining interactions between sex and social measures, such as income.

Individual-level measures and outcomes Describing gender roles and quantifying acceptance of or dissonance with these roles offer two options for incorporating individual-level indicators of the effects of gender on health.14,15 The Bem Sex Role Inven- tory is the most widely used and validated measure of masculinity and femininity.16 Several scales, such as the Gender Role Stress Scale, can be used to determine whether an individual’s values are at odds with the gender norms of the society in which that individual lives.17,18 Although one might intuitively expect that such measures would not be related to exposures or outcomes, the limited existing research suggests otherwise. Higher femininity scores among men, for example, are associated with a lower incidence of coronary artery disease.19 Two Swedish studies showed that when men take on female gender roles, their health appears to improve, whereas the association between masculine roles and women’s health may be less advantageous. Although freeing women from the most traditional of roles and expectations may augment health, in more egalitarian settings and cultures, female well-being may suffer when women adopt workplace behaviors traditionally seen as masculine.20,21

There are other possible but as yet unexplored individual-level indicators of gender equality specific to particular places and cultures. In many settings, gender equality parallels freedom of movement and is closely linked to control over one’s life.22 Women living in more repressive countries are often confined to the home, unable to venture out without male accompani- ment or to drive, and are generally cut off from the external world. Even in more egalitarian societies, women often monitor their surroundings for danger and threats in a manner unknown to most men.23 This gender difference could be measured by gathering information about freedom of movement and utilized

to examine whether gender has a significant impact on exposures and health outcomes. Freedom of move- ment is just one example of how to measure whether gender norms and constraints at the individual level mediate health.

Group-level measures Gender is, by definition, a group characteristic. The norms, expectations, and roles captured in this sum- mative term arise from belief systems of collections of individuals who share a commonality, such as space, religion, culture, or race/ethnicity. In reality, it is gender equity or fairness and equality (that is, equal outcomes), rather than gender itself, that can be described and quantified within and across groupings. Nevertheless, options in current use for including such measures in health research are few and limited in scope or applicability.

Several measures rank gender equality at national levels, including the United Nations Development Programme’s Gender Empowerment Measure and Gender-Related Development Index, and the World Economic Forum’s Global Gender Gap Index.24,25 Each measure summarizes proportions such as ratios of women to men in higher education, government, and the labor force, and some include male and female life expectancies across the geographic unit of countries. Their strength is that they give a national numeric ranking of global gender equality. However, because of the nature of their component measures, none can meaningfully be applied to geographic units smaller than nations. As a result, variations among groups within countries may be averaged out and disappear. If, for example, a nation included just two groups, one that discriminated against all women and the other that discriminated against all men, the overall index would rank the country at the 50th percentile, as the ratios of all measures would show that women achieved parity half of the time. In fact, such a country would represent an extreme of gender inequality for all its citizens. A second caution regarding these indices is that they should be used in conjunction with measures of SES and interactions between the two to disentangle the independent effects of gender equality and other social factors.

The health determinants literature is replete with population-based or multilevel studies showing that, in general, relative deprivation, social isolation, or being the object of discrimination somehow wound and kill.26–29 Although being worse off increases the risk of virtually all illnesses and shortens life expectancy, this effect is more pronounced in men. The direct relationship between SES and well-being interacts

Including Gender in Public Health Research ! 19

Public Health Reports / 2011 Supplement 3 / Volume 126

with gender such that, despite the lower incomes of women relative to men, they universally outlive their male counterparts. Including a group-level measure of gender equality in the research on social circumstances and health would enrich understanding of such sex differences. By measuring gender equality (that is, by quantifying the cumulative effects of gendered differ- ences in social expectations, exposures, vulnerabilities, and access to resources alone, or in combination with other social factors, such as SES), research could move beyond an understanding based solely on sex disag- gregation of data.30 To date, however, no indicator or index of gender equality for use in community (i.e., not international level) studies has been reported, studied, or validated.

There is a significant body of health outcomes research examining income inequality, defined as the gap between the richest and the poorest people across a geographic unit (often a country) and often quantified using the Gini coefficient.29 Although social programs within a country may either attenuate or eliminate the effects of this inequality, the evidence generally points to an association between greater income inequality and poorer health outcomes, independent of income itself.27,29,31 There is, however, virtually no research on the health effects of another form of income inequal- ity—the relative gap between incomes of men and women within any community or grouping—nor is there any measure of gender income inequality compa- rable to the Gini coefficient. One might not expect any association between the size of this gender income gap and population health outcomes, but it is only by ask- ing such research questions that a full understanding of which disparities enter the body to alter well-being is gained and that gender’s contribution to the web of causation becomes visible.

Multilevel or hierarchical measures The interaction between individuals and social systems is dynamic, with each shaping the other in a sort of feedback loop. Using multilevel analysis, it is possible to examine the interactions between characteristics of both the individual and the group. Such models allow us to examine how particular effects vary across groups, as well as within them, and compare the extent of outcome variability from group to group. Individuals from each group are sampled, but groups are sampled as well. If, instead, group-level data are assigned to individuals in a regression analysis (e.g., each person living in a region could be allocated the average air quality for the area), and the cluster is ignored, the methodology will assume the independence of each individual and erroneously increase the power of the

study, resulting in a type 1 error. Multilevel analysis minimizes the errors inherent in analyzing data at one level and interpreting the meaning at another level.

Some possible group-level measures of gender equal- ity that could be included in multilevel research include social indicators of freedom of movement,22 systemic support for parenting by men and women, and aver- age age at first pregnancy. In a study of sex differences in depression across 29 countries, multilevel analysis added depth to the findings that would have been unachievable by using either individual or national data alone.32 Along with individual characteristics such as sex, age, self-reported health, SES, marital and employment status, satisfaction with financial situation, freedom of choice and control, and self-efficacy, two group-level indicators of gender equality were included.

Hopcroft and Bradley replicated the findings of most individual-level studies, showing that being female, having less control, and having lower income predict poorer mental health, but they also found that in countries with lower gender equity overall, depression among men increased. Higher gender equity broadened the gap between the incidence of depression among men and women, with women suf- fering the largest disproportion of illness in the most egalitarian countries. By disentangling individual and group inputs, the authors were able to deepen the understanding of social antecedents to depression. They suggested that a sense of relative deprivation appeared to explain their findings.32

In less egalitarian societies, women may only com- pare their position with that of other women and, therefore, have less of a sense of marginalization than in more equal societies where women have broader horizons and, thus, are more aware of the limits of their own lives relative to those of men and other women. By incorporating indicators of social context into research, the variable, and perhaps the unexpected in vivo effects of external inputs such as gender equity, will begin to be understood.

MORE QUESTIONS THAN ANSWERS

Attention to gender and gender equality as variables associated with health outcomes is a prerequisite, although perhaps not the answer to many questions of etiology and health outcome. As an example, the incidence of non-sex-specific cancers (e.g., colon, blood, lymphatic, and renal), excluding those with a known cause (lung cancer), is markedly higher among men than women.33,34 No sex-specific genetic predispositions have been identified to explain the discrepancies. Somehow, as yet unrecognized aspects

20 ! Commentary

Public Health Reports / 2011 Supplement 3 / Volume 126

of behavior, exposure, and vulnerability of men and women—all components of gender—may explain this disparity.35 Is the risk-taking behavior that is so much more prevalent among men bringing them into contact with carcinogenic environmental triggers? Is there some aspect of male behavior that inhibits the body’s natural protective surveillance systems? Does sex- specific biology interact with exposures to have differ- ent health effects? Perhaps, for example, the estrogen receptors in women enable endocrine disruptors to affect them in ways that don’t affect men, conferring risk in some circumstances and protection in others. Do reproduction and lactation change the internal milieu of women’s bodies such that some environmen- tal chemicals either accumulate or deplete in women during pregnancy and postpartum in a manner that is not possible in men? These are some of what could be an extensive list of questions addressing how the individual- and group-level manifestations of gender and gender equity might translate into biochemical processes and health outcomes.

This article has built upon the increasing scholar- ship suggesting that the divide between biology and socially determined traits may, in fact, be an interac- tion rather than a division, with each one shaping the other. It has also proposed new methods and measures to address gender’s role in determining individual and public health. At present, and within health sciences research, gender remains a somewhat hidden social determinant of well-being. Gender analyses could and should extend well beyond stratification of data by sex to examine differences between men and women aris- ing from social circumstances, but triggering individual biological outcomes. Methodologies for measuring gender equity and equality, although underdeveloped and underutilized, could enrich understanding of associations and causal pathways from exposure to health outcomes and broaden our grasp of how social circumstances alter biology and shape health.

REFERENCES 1. World Health Organization, Commission on Social Determinants

of Health. Closing the gap in a generation: health equity through action on the social determinants of health. CSDH final report. Geneva: WHO; 2008.

2. Phillips SP. Defining and measuring gender: a social determinant of health whose time has come. Int J Equity Health 2005;4:11.

3. Sen G, Östlin P, George A. Unequal, unfair, ineffective and inef- ficient. Gender inequity in health: why it exists and how we can change it. Final report of the Women and Gender Equity Knowl- edge Network to the WHO Commission on Social Determinants of Health. 2007 [cited 2010 Jan 22]. Available from: URL: http:// www.who.int/social_determinants/resources/csdh_media/wgekn_ final_report_07.pdf

4. Huck UW, Banks EM. Male dominance status, female choice and mating success in the brown lemming, Lemmus Trimucronatus. Anim Behav 1982;30:665-75.

5. Gerber LR, González-Suárez M, Hernández-Camacho CJ, Young JK, Sabo JL. The cost of male aggression and polygyny in California sea lions (Zalophus californianus). PLoS One 2010;5:e12230.

6. Fausto-Sterling A. The bare bones of sex: part I, sex and gender. Signs 2005;30:1491-528.

7. Sena G, Iyerb A, Mukherjeec C. A methodology to analyse the inter- sections of social inequalities in health. J Human Dev Capabilities 2009;10:397-415.

8. National Library of Medicine (US). Medical subject headings [cited 2011 Jan 22]. Available from: URL: http://www.nlm.nih.gov/mesh/ MBrowser.html

9. World Health Organization. WHO gender mainstreaming strategy [cited 2011 Jan 22]. Available from: URL: http://www.who.int/ gender/mainstreaming/strategy/en

10. DeFur PL, Evans GW, Cohen Hubal EA, Kyle AD, Morello-Frosch RA, Williams DR. Vulnerability as a function of individual and group resources in cumulative risk assessment. Environ Health Perspect 2007;115:817-24.

11. Molokhia M, McKeigue P. Systemic lupus erythematosus: genes versus environment in high risk populations. Lupus 2006;15:827-32.

12. Sarzi-Puttini P, Atzeni F, Iaccarino L, Doria A. Environment and systemic lupus erythematosus: an overview. Autoimmunity 2005;38:465-72.

13. Rose G. Sick individuals and sick populations. Int J Epidemiol 1985;14:32-8.

14. Bekker MJ. Investigating gender within health research is more than sex disaggregation of data: a multi-facet gender and health model. Psychol Health Med 2003;8:231-43.

15. Judge TA, Livingston BA. Is the gap more than gender? A longitu- dinal analysis of gender, gender role orientation, and earnings. J Appl Psychol 2008;93:994-1012.

16. Bem SL. Bem sex-role inventory professional manual. Palo Alto (CA): Consulting Psychologists Press; 1981.

17. Eisler RM, Skidmore JR. Masculine gender role stress. Scale develop- ment and component factors in the appraisal of stressful situations. Behav Modif 1987;11:123-36.

18. Gillespie BL, Eisler RM. Development of the feminine gender role stress scale. A cognitive-behavioral measure of stress, appraisal, and coping for women. Behav Modif 1992;16:426-38.

19. Hunt K, Lewars H, Emslie C, Batty GD. Decreased risk of death from coronary heart disease amongst men with higher “femininity” scores: a general population cohort study. Int J Epidemiol 2007;36:612-20.

20. Månsdotter A, Lidholm L, Lundberg M, Winkvist A, Ohman A. Parental share in public and domestic spheres: a population study on gender equality, death, and sickness. J Epidemiol Community Health 2006;60:616-20.

21. Backhans MC, Lundberg M, Månsdotter A. Does increased gen- der equality lead to a convergence of health outcomes for men and women? A study of Swedish municipalities. Soc Sci Med 2007;64:1892-903.

22. Laws G. Women’s life courses, spatial mobility, and state policies. In: Jones JP 3rd, Nast HJ, Roberts SM, editors. Thresholds in feminist geography. Lanham (MD): Rowman and Littlefield Publishers; 1997. p. 47-64.

23. Foster C, Hillsdon M, Thorogood M. Environmental perceptions and walking in English adults. J Epidemiol Community Health 2004;58:924-8.

24. United National Development Programme, Human Development Reports. Measuring inequality: Gender-Related Development Index (GDI) and Gender Empowerment Measure (GEM) [cited 2009 Aug 29]. Available from: URL: http://hdr.undp.org/en/statistics/ indices/gdi_gem

25. Hausmann R, Tyson LD, Zahidi S. The global gender gap report 2008. Geneva: World Economic Forum; 2008. Also available from: URL: http://www.weforum.org/pdf/gendergap/report2008.pdf [cited 2009 Aug 29].

26. Berkman LF, Syme SL. Social networks, host resistance, and mortal- ity: a nine-year follow-up study of Alameda County residents. Am J Epidemiol 1979;109:186-204.

27. Kawachi I, Kennedy BP, Lochner K, Prothrow-Stith D. Social capital, income inequality, and mortality. Am J Public Health 1997;87:1491-8.

28. Lynch JW, Davey Smith G, Kaplan GA, House JS. Income inequality and mortality: importance to health of individual income, psycho- social environment, or material conditions. BMJ 2000;320:1200-4.

Including Gender in Public Health Research ! 21

Public Health Reports / 2011 Supplement 3 / Volume 126

29. Wilkinson RG. Income distribution and life expectancy. BMJ 1992;304:165-8.

30. Phillips SP. Measuring the health effects of gender. J Epidemiol Community Health 2008;62:368-71.

31. Lynch J, Smith GD, Harper S, Hillemeier M, Ross N, Kaplan GA, et al. Is income inequality a determinant of population health? Part I. A systemic review. Milbank Q 2004;82:5-99.

32. Hopcroft RL, Bradley DB. The sex difference in depression across 29 countries. Soc Forces 2007;85:1483-507.

33. Canadian Cancer Society/National Cancer Institute of Canada. Canadian cancer statistics 2008. Toronto (Canada): CCS/NCIC; 2008. Also available from: URL: http://www

.cancer.ca/~/media/CCS/Canada%20wide/Files%20List/English %20files%20heading/pdf%20not%20in%20publications%20 section/Canadian%20Cancer%20Society%20Statistics%20PDF %202008_614137951.ashx [cited 2009 Jun 19].

34. Cancer Research UK. The excess burden of cancer in men in the UK [cited 2009 Jun 15]. Available from: URL: http://info.cancer- researchuk.org/prod_consump/groups/cr_common/@nre/@sta/ documents/generalcontent/crukmig_1000ast-2748.pdf

35. Courtenay WH. Constructions of masculinity and their influence on men’s well-being: a theory of gender and health. Soc Sci Med 2000;50:1385-401.