Revisiting Mars and Venus: understanding gender differences in critical illness

Understanding the nature and biological basis of gender-determined differences in risk of and outcome from infection might identify new therapeutic targets, allow more individualised treatment, and facilitate better risk prediction and application of healthcare resources. Gender differences in behaviours, comorbidities, access to healthcare and biology may result in differences in acquiring infection, or in response to infection once acquired. Some studies have reported higher male susceptibility to infection, and higher risk of death with sepsis, but others have found the opposite effect. The explanation for this disagreement is probably that different studies have included patients at different stages on the continuum from infectious agent exposure to death or recovery. Studying sufficient patient numbers to explore this entire continuum while accounting for heterogeneity in type of infection and comorbidity is difficult because of the number of patients required. However, if true gender effects can be identified, examination of their biological or psychosocial causes will be warranted.

Much eff ort has been expended identifying factors associated with sepsis susceptibility, prognosis and response to treatment. As with other risk factors, understanding gender-determined diff erences in response to infection might highlight new therapies or identify patients especially likely to respond to a particular treatment. Understanding the eff ect of gender in infection is therefore important. Th e paper by Nachtigall and colleagues in the previous issue of Critical Care is the latest contribution to this topic [1].
Men and women diff er in a number of respects that, overall, result in men dying earlier. Men have more highrisk behaviours such as smoking and activities leading to trauma. Men acquire chronic diseases earlier [2], in part due to lifestyle choices but also due to biological diff erences that are far from fully defi ned [3]. Men with some [4], but not other [5], actue illnesses tend to present later to medical care. Men receive more aggressive medical interventions once in hospital [6]. Sex hormones may modulate response to infectious agents [7]. Menstruation is a repeated acute infl ammatory state that might modify response to infection. Cellular mosaicism in women may infl uence immune response [8]; for example, by attenuating the eff ect of polymorphisms in X-linked infl ammatory genes such as IRAK-1 [9]. Each of these factors probably infl uences response to infection.
Many observational studies have attempted to identify gender diff erences in outcome from infection. Some studies have found that men are more susceptible to infection [2,10] and that men are more likely to die once an infection occurs [4,11,12]. Other studies, however, have found the opposite eff ect [13,14]. Much confusion is due to oversimplifi cation of the progression from infectious agent exposure to death. Rather than a single event, this progression represents a number of stages of illness, and men and women may progress through these stages diff er ently. For instance, a person exposed to an infec tious agent fi rst has a risk of that agent causing localised disease, probably related to infl ammatory and immuno logical priming. Second, once the organism has taken hold, the risk of developing a systemic infl ammatory response -sepsis -is likely to be determined by a diff er ent constellation of factors. Once sepsis occurs, the risk of progression to organ dysfunction -or severe sepsis -is probably related to still more factors, including co morbidity. Last, the risk of death is infl uenced by the ability and willingness of the patient to access organ support, and their physiological reserve. Observational studies recruit patients in varying phases of this continuum, leading to considerable confusion.
Th e paper by Nachtigall and colleagues reports that, in a largely surgical adult ICU cohort of 709 patients, ICU

Abstract
Understanding the nature and biological basis of gender-determined diff erences in risk of and outcome from infection might identify new therapeutic targets, allow more individualised treatment, and facilitate better risk prediction and application of healthcare resources. Gender diff erences in behaviours, comorbidities, access to healthcare and biology may result in diff erences in acquiring infection, or in response to infection once acquired. Some studies have reported higher male susceptibility to infection, and higher risk of death with sepsis, but others have found the opposite eff ect. The explanation for this disagreement is probably that diff erent studies have included patients at diff erent stages on the continuum from infectious agent exposure to death or recovery. Studying suffi cient patient numbers to explore this entire continuum while accounting for heterogeneity in type of infection and comorbidity is diffi cult because of the number of patients required. However, if true gender eff ects can be identifi ed, examination of their biological or psychosocial causes will be warranted. mortality was similar between men and women [1]. Importantly, all of these patients were receiving antibiotics either for surgical prophylaxis or for treatment of septic shock. Patients therefore entered the study at a mixture of points on the above continuum. In the 327 patients who had sepsis, being female nearly doubled the risk of death, independent of diff erences in age, intensity of therapeutic interventions, source of infection, organism and presence of shock. In the entire patient cohort (46% of whom were also in the sepsis group), however, gender had no infl uence on ICU mortality. Overall, then, among patients requiring antibiotics, it seems that either the benefi cial and detrimental eff ects of being female cancelled one another or gender had little eff ect. Arguing for some eff ect of gender is the convincingly higher female sepsis mortality.
Further to this observation, Nachtigall and colleagues' paper contains a striking fi nding not discussed in the manuscript. Of 400 males in the cohort, 197 (49%) developed sepsis, compared with only 130 of 309 (42%) females, a diff erence that nearly reaches signifi cance (P = 0.06) [1]. Whether this diff erence would remain or would be adjusted away in multivariable analyses remains speculative. However, it appears plausible to conclude that, if exposed to infection, men are more likely to develop sepsis -as, indeed, other studies have found [2,10]. Knowledge of the infection continuum makes the apparent contra diction between this observation and that of increased female mortality in the presence of sepsis more easily understood. Prior studies suggest that the observed gender diff erences in this study may be due to diff erences in the immune response [4], perhaps mediated by oestrogen levels rather than gender per se [15]. As the authors note, however, such explanations of the eff ect of gender in their cohort are speculative.
Th e study faced several challenges in isolating the eff ect of gender in critically ill patients. First, studies that recruit a heterogeneous population of patients must adjust for potential confounding. Th e study did adjust for factors that in univariate or backwards stepwise multivariate analysis were signifi cant predictors. However, women were more likely than men to be immunosuppressed, and this diff erence was larger among those who developed severe sepsis (2.5-fold higher risk among women). While not statistically signifi cant predictors, such diff erences may still confound the association between gender and mortality. Studies that recruit hetero geneous populations should have a suffi ciently large sample size to ensure that results are robust. Second, the authors report data regarding ICU mortality. Although men with sepsis had higher risk of developing septic shock, the length of ICU stay was similar and ICU mortality was lower among men. Many more men than women had undergone cardiac surgical procedures. If cardiac surgical patients were discharged from the ICU to a high-dependency ward earlier than other types of patient, as is true in many hospitals, their ICU mortality may be artifi cially lowered. Whether the higher mortality for women persists at 28 or 90 days remains unclear.
At fi rst glance, it is reassuring that Nachtigall and colleagues found almost no gender diff erences in quality of care. However, perhaps men should have received more resources to reduce their incidence of sepsis, or women should have been treated more aggressively to reduce their mortality once sepsis occurred? If a strategy to reduce transition from infection to sepsis was more eff ective than one to treat sepsis once established (or vice versa), an alternative strategy would be to aim for optimisation (rather than equalisation) of mortality in men and women. Ethical questions regarding resource allocation with respect to gender remain theoretical while the mechanisms underlying the observed disparities are not understood. If this changes with further work, as might be hoped, such questions of equality will need to be addressed.