Sex-Gender Research Sensitivity and Healthcare Disparities

Introduction

“Being male or female is an important fundamental variable that should be considered when designing and analyzing basic and clinical research.” ¹ It is obvious that there are profound differences in health and disease related to the reproductive and endocrine systems, ² but there are many more subtle differences where sex and gender and age and work and social factors and education and income and behavior interact. Large population-based studies can offer important clues even when there are many potential confounders or effect modifiers, but only if researchers address sex and gender head on. Is that happening? Linda Nieuwenhoven and Ineke Klinge examined this in “Scientific Excellence in Applying Sex- and Gender-Sensitive methods in Biomedical and Health Research” ³ and concluded that despite advances, the answer is still No, and they elaborate on the importance of sex-gender sensitivity in biomedical research.

I have long been puzzled by the realization that despite half the world being female and the other half having females as mothers, wives, and daughters, there remains a gender gap in biomedical research related to causation, diagnosis, risk factors, prevention, and treatment. It is not always possible or relevant to include males and females in a study, whether it is rodent toxicology or human epidemiology, but even if females are represented, Nieuwenhoven and Klinge aver that is no guarantee of sex-gender sensitivity in interpretation. ³

Krieger ⁴ provides a comprehensive discussion and definition of the terms “sex” and “gender.” The common distinction is that sex refers to biologically determined differences related to the XX and XY chromosomal determination (with very few deviations), whereas gender is determined by complex and continuing interactions among biological sex, personality, life experience, and culture(s)—the interaction of environment and biology. Nieuwenhoven and Klinge ³ indicate that the terms are often used interchangeably or wrongly, and as the editor of a special journal issue (16 articles) on sex and toxicology, I plead guilty, for on the vote of the authors, we used the term gender throughout, as if it were not quite tasteful for grownups to speak of sex in mixed company. Sex is a crucial variable, however, and for environmental toxicologists like me, it offers a fertile and challenging, if frustrating, field. A growing number of articles find differences between men and women in a variety of toxicological responses, ⁵ but are these caused by differential exposures (gender), differential biochemistry (sex), or a combination of the two (most likely). ⁶ Animal research can only go so far in addressing sex and gender differences.

It is generally assumed that gender is a uniquely human trait; Nieuwenhoven and Klinge say “almost exclusively,” ³ but this is certainly not true. Depending on how they are raised, male and female animals experience different social groups and dominance relationships and develop different suites of behavior appropriate to different social interactions. We should not ignore gender just because animals do not dress up or play with dolls or trucks. Roles certainly analogous to gender are obvious at least in carnivores, primates, ungulates, and elephants, with effects apparent, for example, when a senior or dominant male or female is removed from or introduced into an animal society, reflecting the “evolutionary continuity of mental experience.” ⁷ Toxicology researchers are well aware of the importance of the social combinations of rodents being tested, although unfortunately they often resort to testing only males “to keep things simple.” A good example of the weakness of single-sex studies is the discovery of the unique enzyme in the kidney cortex of male rats, an alpha-globulin, that is not expressed in females or in mice and renders the male rats vulnerable to kidney diseases caused by hydrocarbons.

Sex Insensitivity

One of my first impressions in reviewing Nieuwenhoven and Klinge ³ was that these authors correctly appreciate the importance of sex and gender in both health and disease but that they underestimated the tremendous progress in studying sex differences in morbidity, mortality, or management, not to mention toxicology and epidemiology, and overstated the lack of sensitivity in the literature. Donning gender-sensitivity goggles, however, reveals the underlying truth in their thesis. A great deal of studies include men and women (now mandated by regulations), without adequately interpreting or explaining differences or similarities.

In preparation for this editorial, I read a sample of recent medical and epidemiological studies. Searching PubMed with terms, such as gender disparities in healthcare or medical care, yielded only two pages of references. Searching on sex disparities in medical care returned 214 entries, the most recent of which were mainly about race or social class bias, revealing lack of attention to sex rather than the converse Although most of these studies included data on men and women, I was surprised at how often the data on sex differences were downplayed or even ignored or obscured by adjustment even in otherwise excellent papers.

For example, Macinko and Elo ⁸ have just published a valuable analysis of racial disparity in avoidable mortality. They compare black and white mortality for several diseases, separately for males and for females. Not surprisingly, their data show dramatic (and expected) racial differences in absolute mortality, as well a gratifying reduction in the discrepancy over a 25-year period. However, data for men and women are presented in separate parts of the table, precluding easy comparison, and there is little mention of some of the surprising sex differences, for instance, between white men and white women. Deaths avoidable by “improved medical care” showed a 6.1% improvement in white men compared with 13.4% improvement in white women, and mortality “averted by public health policy” improved 32.1% in men but only 4.7% in women. The latter gender gap was even greater for blacks 71.9% vs 9.1%. Why? This latter category includes “public health interventions and policies directed at changing behaviors (e.g., smoking, drunk driving, and excessive drinking; seatbelt use; access to firearms).” ⁸ It may simply be that men were so much more likely to die from these in the first place that they had much more room for improvement.

Burge et al. ⁹ studied age differences in the use of palliative care among 7500 cancer decedents in Nova Scotia. Old (>65 years) people were less likely to get such care, but in this age group, men were significantly less likely than women to register for care, perhaps because they are more likely to have a younger, surviving spouse as caretaker. Hanchate et al. ¹⁰ compared ethnic/racial groups with respect to cost of end of life care and found significant differences overall, which persisted within sexes, but they do not discuss any difference between sexes. Why, for example, do younger women (aged 65–75) cost more and older women cost less than men of the same age. Why is the black/white discrepancy much greater in women than men? Sex was one of the three independent variables studied by Koller and Mielck, ¹¹ investigating obesity, inadequate checkups, and inadequate vaccinations among children in Germany. The Abstract makes no mention of the sex outcome, perhaps because girls and boys were equally likely to miss vaccinations (55%) or be obese (10%), whereas girls were slightly more likely to miss examinations (24% vs. 22%, p = 0.08). Similarities must be uninteresting.

For the period 1996–2000, blacks and women on a national liver transplant list were more likely to die or become sick prior to transplant and less likely to receive a transplant within 3 years than white males. After introduction of the Model for End-Stage Liver Disease (MELD) objective score in 2002, the racial disparity declined to nonsignificant, but the sex disparity widened to a relative risk of 1.3 (p < 0.05). ¹² There is no clear explanation for the sex difference. The authors suggest that the score itself, which includes creatinine level as a component, discriminates against women who, because of smaller muscle mass, tend to have lower creatinine levels, hence, lower scores and lower priorities. ¹² This is presumably an unintended consequence of failing to pay attention to sex and gender.

The cited articles either adjust for sex (obscuring differences) or do not mention sex differences in the Abstract; sometimes, the results from men and women are combined. It is safe to say that there are probably no circumstances in which results from men and women should be lumped a priori without testing for differences, and the contribution of sex differences as confounders or effect modifiers should be elucidated in papers and referred to in the Abstract, which is what most readers see.

Nieuwenhoven and Klinge emphasize that sex differences translate into health differences and disparities in the diagnosis and treatment of diseases. ³ This has been established on numerous occasions for racial/ethnic differences but on fewer occasions for male/female differences. Ignoring, for today, the extensive literature on different incidences of various diseases among men and women, there are some pervasive treatment disparities.

Women's Health and Treatment

An oft-mentioned gender difference is that men and women differ in health-seeking behavior and in their attentiveness to symptoms. Women are more likely to seek treatment earlier than men for similar symptoms; indeed, adult women are more likely to have a physician than men. There is also a health perception difference; for example, in a study of over 22,000 men and women in Britain, women were significantly less likely to rate their health as excellent regardless of social class. ¹³ The differences, 16.1% vs. 18.4%, seem small at first but translate to a 12% difference. Even the same disease, depression, can be defined differently and carry different stigmas in males and females. ¹⁴

Nieuwenhoven and Klinge wrote “sex and gender can help explain the differences in etiology and prognosis of diseases” and can “modify the outcomes of diagnostic procedures and of preventive and treatment interventions.” ³ There is a further factor to consider, however, Physicians, regardless of their own sex and gender, treat men and women differently. Women who suffer acute myocardial infarction (AMI) are less likely to receive prompt aggressive treatment, are more likely to have symptoms labeled as psychologic or atypical, ³ and are more likely to have back pain than chest pain. ¹⁵

Discrepancies in medical treatment, now widely recognized for racial disparities as elucidated in the Institute of Medicine report on “Unequal Treatment,” ¹⁶ persist for gender. In 1991, Ayanian and Epstein, ¹⁷ noting prior reports of treatment disparity, examined 82,782 heart diseases cases including 18,759 diagnosed with AMI, and found that women in Massachusetts were 28% and 45% less likely to be referred than men for angiography and revascularization, respectively. Maryland results (15% and 27%) were less extreme, but all were statistically significant. Their report, published in the New England Journal of Medicine, ¹⁷ garnered attention, but over the ensuing years, some treatment has not changed much. “Women receive less evidence-based medical care than men and have higher rates of death after AMI.” ¹⁸ Why? Women were less likely than men to have ST-elevation myocardial infarction (STEMI), but women who were admitted with STEMI were half as likely to leave the hospital alive. Women were less likely to receive aspirin or beta-blockers or reperfusion (angioplasty), and when the latter occurred, the lag time was longer. When all MIs were considered, the sex difference was not significant, but for STEMI, the most common type of AMI in men, the results were striking and scary. The way the authors worded their conclusion is puzzling, however. They begin their conclusion with “Overall, no sex differences in in-hospital mortality rates after AMI were observed after multivariable adjustment. However, women with STEMI had higher adjusted mortality rates than men.” ¹⁸ Surely, they would vigorously deny that their work or views are insensitive to sex, yet the emphasis in the conclusion suggests otherwise, that the sex discrepancy for STEMI is just a sidelight, barely worth mentioning. Enriquez et al. ¹⁹ reported even stronger discrepancy in drug treatment rates for statins, aspirin, and beta-blockers. Nguyen et al. ¹⁵ noted in their Abstract no sex difference in drug treatment for AMI. However, even though not statistically significant at the 0.05 level, women were still 5%–10% less likely to receive the three classes of drugs. Significantly, they found a 46% lower likelihood of women being referred for invasive diagnostic procedures. ¹⁵ Perhaps the most gender-sensitive feature of their article was the conclusion that women might be undertreated or men overtreated with invasive procedures.

The American College of Cardiology and American Heart Association (ACCA-AHA) published detailed guidelines for various management approaches to different cardiac syndromes. ²⁰ Management of patients with acute coronary syndrome (unstable angina or MI) continues to be suboptimal. ²¹ Evaluating adherence to the guidelines as metrics for acute coronary care, the CRUSADE study shows that the gender gap for some treatments (aspirin and clopidogrel) has narrowed to a few percentage points, but a broad gender gap remains for some treatments. Women were less likely to be treated according to the ACCA-AHA guidelines and were less likely to receive early glycoprotein IIb/IIIa inhibitor therapy (29% vs. 39%). ²² It is scant consolation that men also are undertreated according to the new guidelines with glycoprotein IIb/IIIa inhibitors. ²³

Toxicology and Sex

In 2005, I was an organizer of a weeklong workshop on “Gender in Toxicology and Risk Assessment” for the Scientific Group on Methodologies for the Safety Evaluation of Chemicals (an affiliate of the Paris-based, international Scientific Group on Problems of the Environment ²⁴ ). Aside from confusing gender and sex, the toxicologists recognized “three major themes surrounding biologic sex differences: genetic, human health, and ecologic.” ²⁵ “From embryonic life onward males and females have very different internal milieus mediated by the powerful influence of sex hormones and their receptors” (sex) and have different external environments influenced by roles in relationships, recreation, reproduction, and employment (gender). ²⁵

Ironically, much of our knowledge of human toxicology stems from relatively highly exposed people working in chemical and other factories. Women and minority workers were almost universally excluded from such jobs and, hence, are absent in most occupational epidemiology studies or, if included in the data, were usually ignored in the analysis because of small sample sizes. ⁶ The workshop offered a framework for examining sex and gender differences by focusing on (1) exposure opportunity, (2) toxicokinetics (what the body does to chemicals), (3) toxicodynamics (what chemicals do to the body), and (4) modifiers (particularly sex hormones). ⁶ There are sex differences as early as germ cell division and mutagenesis, including differences in mutation rates at specific loci ²⁶ and in the induction of chromosomal aberrations. ²⁷ A broad range of sex differences in toxicokinetics and toxicodynamics occurs throughout the life span. ⁵

The basic paradigm for studying sex differences has centered around the sex hormones, including treatment with agonists or antagonists or castration, followed by administering the same or opposite hormones. Much has been learned from complex variations on this theme about the reproductive system, development, and effects in other organs. However, the study of the role of sex hormones and receptors modulating metabolism is in its infancy. For example, males and females exposed to arsenic in drinking water metabolize and distribute arsenic compounds differently because of different methylation rates, which become faster in older women. ²⁸ The overall workshop conclusion was: “Wherever possible, studies should use balanced gender and gender × age designs and should analyze data by sex and interactions, rather than simply adjusting for (discarding) gender.” ⁶

Drugs and Women

From toxicology, it is a short step to pharmacology. Another point articulated by Nieuwenhoven and Klinge ³ is the lack of data on drug efficacy and side effects in women. Medicine has recently emphasized the importance of individual variability in pharmacokinetics (what the body does to a drug in terms of absorption, metabolism, distribution, and elimination) and less so in pharmacodynamic variability (what a drug does to the body in terms of entering cells, binding to receptors, affecting membranes, organelles, macromolecules, cell signaling, carcinogenesis). There is a growing interest in individualizing drug treatments, tailoring prescribing practices to take into account genetic variation that may influence the metabolism of one or more drugs, requiring an adjustment in dosage or timing. Sex is likely to be a contributing factor, if only because of the commonality of metabolic pathways between xenobiotics and sex hormones, for example, the affinity of estrogen and some anti-epilepsy drugs for P450 3A4, such that coadministration results in enzyme induction and reduced efficacy of both, causing more seizures and more pregnancies. ²⁹

Calabrese, ³⁰ in 1985, assembled a long list of drugs that affect men and women differently. The list has grown since then, of course, but the mechanisms for these differences are not all apparent. Individual tailoring of drug treatment based on variations in metabolizing hormone patterns, receptor biology, circadian cycles, and monthly cycles may elucidate some of the mechanisms of male-female differences. It is an exciting challenge for biologists, pharmacologists, and toxicologists. To date, we have crude estimates that metabolic differences between sexes can be on the order of 2-fold and that xenobiotic distribution can differ greatly among individuals based on body composition, with the tendency toward women being more susceptible to drug interactions. ⁵

Some of the differences in how men and women handle and respond to drugs depend on their physical variability. Average relative organ sizes tell only a small part of the sex story but can be instructive. The International Commission on Radiation Protection (ICRP) assembled a monograph describing in detail sex differences to support health physicist calculations. ³¹ It may come as no surprise that women have (on average) a third more adipose tissue and a third less muscle mass, but other significant differences include women having 10%–15% less bone, blood, and skin, and an 8%–14% larger gastrointestinal tract, kidneys, and brain. Cartilage, skin, and liver vary <5%.

Drug Trials

A major impediment to individual tailoring is that women have always been underrepresented in drug development trials, forbidden to participate in phase I trials, and discouraged or reluctant to participate in phase II trials. The U.S. Food and Drug Administration (FDA) is sensitive to this, requiring enrollment of women, ³² and the situation has gradually improved. However, the FDA's Office of Women's Health identifies persistent obstacles while suggesting that new technologies and techniques may improve the representation of women in drug trials. ³³

The exclusion of women of childbearing age from most drug studies continued into the 1990s, and many trial protocols require women to prove they are sterile or are using effective birth control. Observational studies (rather than randomized trials) have revealed striking sex differences. For example, Gan et al. ³⁴ found that women recover from general anesthesia significantly more quickly (7 minutes) than men (11 minutes), and this persists when corrected for dose. ³⁵ The difference in the hypnotic effects of the drug propofol was partly the result of pharmacokinetics (a consistently more rapid decline in blood concentration of the drug in women) ³⁴ and partly the result of pharmacodynamics (men awoke at a higher blood concentration). ³⁷ This sex difference is more pronounced in premenopausal women, ³⁷ opening a new avenue for investigation.

Drug development studies now include more females than previously, but for many drugs that carry a potential for fetal effects or teratogenicity (almost any new drug), pregnant women are still excluded, and women of childbearing age can participate only if they adhere to a strict birth control regimen. Thus, equality in drug trials may never be achieved if a large portion of women's life cycle is excluded from studies. For example, a review of 221 trials of anti-HIV drugs averaged only 11.6% female participation. Of 24 trials that had no women, there were no outright exclusions, and nonpregnant women would have been allowed to participate (with some form of birth prevention). The reviewers concluded that other factors (not specified) precluded women from participating. ³⁸ One can infer that the trial investigators were insensitive? But how? Did information on the trials fail to reach women, or did they find the information unappealing or scary? One can image that marketing approaches advertising the availability of trials might reach one sex more than another or might appeal differently to men and women or that the warnings about reproductive consequences might discourage most women from participating. One can easily imagine that the requisite warnings on Institutional Review Board-approved informed consent forms would scare all but the most intrepid female participant.

Hormone Replacement

In my view a dramatic example of sex-gender insensitivity was the sudden, probably premature termination ³⁹ of the Women's Health Initiative (WHI) randomized, placebo-controlled, study of hormone replacement therapy (HRT) in postmenopausal women. One of the major health decisions women face is whether to start or continue HRT, and the trade-off of benefits and risks remains controversial. This arguably valuable treatment included some significant risks but was seriously compromised when the trial was terminated because of mounting adverse effects. Although the investigators reported the stopping criteria and explained that the trial involved older women and might not be applicable to perimenopausal women, there was a widespread impression that the adverse effects overwhelmed any benefits for anyone, causing confusion. Although HRT continued to be approved for menopausal symptoms, many doctors stopped prescribing HRT entirely, and many women feared to continue its use. The termination left a void. A gender-sensitive approach would have been to reconfigure the study with new warnings. Termination of the WHI study also brought about termination of other trials just getting underway, including a New York study of hormones and Alzheimer's disease ⁴⁰ and an international study (United Kingdom, Australia, New Zealand). ⁴¹ That study also found adverse cardiovascular and thromboembolic risks early in use but also found a nonstatistically significant decrease in stroke and fractures, not to mention breast cancer. ⁴¹ Moreover, in lieu of doctor-prescribed HRT, women seeking to alleviate menopausal symptoms have turned to a hodgepodge of untested, uncontrolled herbal and alternative treatments largely outside the scope of medicine and FDA control. ⁴²

Epidemiology

With so much evidence of sex-gender insensitivity, I turned to the epidemiology literature. Surely, epidemiologists appreciate sex, and consideration of sex in research studies is alive and well, or so I thought. As I was preparing this editorial, the latest issue of the American Journal of Epidemiology arrived in my email-box. I turned to it for relief from insensitivity. The Table of Contents started off balanced enough: three articles on prostate disease and three on pregnant women, but then a study of whether nitrogen dioxide exposure in early childhood affects cognitive function at age 4, which enrolled about 50% girls, makes no mention of any outcome similarity or difference by sex—it was ignored entirely. ⁴³ The next article on asthma and birth weight in same-sex twins does not tell us if male-male and female-female twins were similar or different. ⁴⁴ That sample seemed enough to demonstrate at least a lack of sensitivity.

It is not enough to include both sexes in a study. The sample size of each should be large enough to provide sufficient power for subset analyses. Power is the ability to avoid a type II error, the ability to confirm a difference that is really there, and to say that a difference has reached statistical significance. Science has been much more attentive to avoiding type I errors than type II errors, so when a difference does not reach the sacred (but entirely arbitrary) 0.05 level, it is reported as not significant or, worse, not mentioned at all. Readers should have the opportunity to judge for themselves if a difference is important and if “no difference” is real or, as is so often the case, due to lack of power.

Changes on Horizon

Gender differences evolve. Clothing changes; bare skin appears (healthful tans) then disappears (sun and cancer). Ambiguous or ambivalent gender identity is much more acceptable now than a generation ago. Expectations for life and livelihood, health and longevity have changed. Women make up about half the workforce, although they still hold somewhat different jobs (and get paid less), and by 2004, women made up half of the U.S. entering medical school class. ⁴⁵ Relationships between men and women and between patients and physicians have changed as well. As a medical house officer in the 1960s, the idea that patients might be allowed to see their medical record was anathema. Patients had to be guarded from their ignorance or innocence. Do women have different expectations from men from their doctor-patient relationships? And most importantly, will the influx of women into all these related fields mean that more female-relevant, sex-gender-sensitive research will be done? And applied? I do not think it is risky or presumptuous to answer Yes.

Conclusions

There are varying ways in which researchers are not sensitive to sex, even when data have been gathered appropriately. Doing studies that include men and women need not be difficult, although disparities in clinical trials of new drugs will be hard to overcome. Communicating results is a weak link. Like it or not, Abstracts are now among the predominant ways of communicating biomedical information. Many readers cannot or will not pay for whole articles. Therefore, Abstracts had better be right and inclusive. One of the insensitive ways of dealing with sex is to leave it out of the Abstract, for example, in studies of other risk factors, or to report data that are adjusted for sex without providing or discussing the sex differences themselves. Sex is not alone in this: age and sex interact in important ways from birth to death, through puberty, reproduction, menopause and andropause, and aging.