Gender-related outcome difference is related to course of sepsis on mixed ICUs: a prospective, observational clinical study

The impact of gender on severe infections is in highly controversial discussion. A natural survival advantage for females [1] in data from animal experiments seems to be contradictory to human clinical data on sepsis-related mortality [2‐5]. Eachempati et al. [5] demonstrated female gender as an independent predictor of increased mortality in patients with documented infection in a surgical intensive care unit (ICU). He emphasised that different genders may need different types of therapy. Depending on the chosen subgroup, Combes et al. [3] reported an increased risk for women to die of nosocomial infections.

Different pathomechanisms were addressed to be responsible for these findings, including sex-related gene polymorphisms [2], effects of sex hormones [6], or different intensities of care, with males receiving more invasive procedures [7, 8].

One fundamental issue is to make study populations matchable for research projects (for example, for evaluation of new interventions and drugs), but as well to assess the severity of diseases in mixed populations or for benchmarking purposes. Classification systems like PIRO (predisposition, insult/infection, response, organ dysfunction) included gender with the intention of improving the comparability of studies. But although discussion is nearly a decade old, still no agreement is found as to whether female or male gender is a predisposing factor [9]. Le Gall et al. [10] created an extended version of the Simplified Acute Physiology Score-II (SAPS-II), including gender, giving male patients a higher score for predicting mortality [10]. The author excluded burned and coronary and cardiac surgery patients; the latter is a population in which women probably do worse [3]. Interestingly, it seems to be highly cohort related whether men or women are more likely to survive. This study aims to describe the impact of gender on outcome of patients on mixed ICUs with a special focus on sepsis patients.

Most important, we were able to demonstrate that a gender-related effect on mortality is limited to the specific subgroup of sepsis patients. Further, we showed that despite smaller differences in the care process, quality and quantity of key interventions in infection management are distributed equally between genders. These findings contribute to the discussion of whether different levels of care have an impact on gender-related outcome [8, 17, 18].

Concerning distribution of basic characteristics, we were able to reproduce well-described differences between genders comparing lifestyle risks, comorbidities, and age [3, 8]. Similarly, subgroup analysis for sepsis patients showed differences in basic characteristics for age, lifestyle risks, immunosuppressive status, and vascular disease. Women had lower SOFA score on admission in the main study population as well as in the sepsis subgroup. This finding is well described for other study populations [19]. However, primary assessed scores reflect only the moment of admission but not the kinetic of clinical course, for which SOFA is not validated [20]. Because this scoring system does not take gender into account [21, 22], the gender-related difference in mortality is not reflected well and remains a topic that must be elucidated.

Quantity of microbiologic diagnostics did not differ between genders, but in men, more radiologic diagnostics were performed. This is consistent with the findings of Valentin et al. [8], that men are more likely to obtain a higher intensity of care in ICU. Furthermore in our study, no differences were found for antibiotic-free days, daily antibiotic use, daily costs of antibiotics, and SOP adherence in percentage of all days--neither in the main study population nor in the sepsis subgroup--showing that in our study population, men and women get the same quality and quantity of care in means of antibiotic therapy. As time to antibiotics was found to be relevant for ICU mortality, we also included this parameter in our analysis but did not find significant differences between genders [23].

In women, pneumonia as well as overall infection rate was lower, but urinary tract infections occurred more often. This goes along well with previously described data [3, 4, 24, 25]. A higher infection rate in men, mainly based on increased pneumonia rates, may impair survival. Conversely, in our main population, outcome of gender was equal, but in the sepsis subgroup, women were more likely to die than were men. As we think that gender differences are a matter of subgroup selection, this finding is in concordance with other published studies in mixed ICUs, where several differences for gender, but not for mortality, are described [3]. Experimental data show a natural survival advantage after polymicrobial sepsis for women [1], but human studies focusing the impact of gender on sepsis-related mortality gave inconsistent results, showing a lower [7, 19], equal [26, 27] or higher [5, 28‐30] mortality rate. Vincent et al. [29] reported, in their European multicenter cohort, higher mortality for women with sepsis than for men [29]. Combes et al. [3], studying ICU patients with sepsis, reported that female gender predicted mortality, although in the univariate analysis, mortality was not significantly different. Seymour et al. [28] reported a higher mortality rate for women with Fournier gangrene, with more male patients surviving the septic phase. In contrast to this, Adrie et al. [29] found a survival advantage in women older than 50 years, which is also opposed to our findings [19]. In that study, Adrie et al. postulated women older than 50 years to be postmenopausal without measuring hormone status, which seems to be of high importance. Further on, they included only community-acquired sepsis and predominantly medical patients, which is a difference from our and other study populations. The results of the O/E mortality rate suggest that the SAPS-II score provides the most reliable predictive values for the subgroup of women with sepsis. It is an interesting finding that might indicate a limitation of intensive care scoring systems.

In our study, we also analyzed evolution of TISS-28 scoring for men and women. We found consistent differences between genders being statistically significant. As TISS-28 is a surrogate parameter for intensity of care, our results coincide with the findings that men receive more invasive procedures than do women, regardless of the reason of admission or procedures [7, 8]. Conversely, TISS-28 was thought to be used to compare predicted survival in different populations [21, 22]. Based on these findings, it might be difficult to use scoring systems not incorporating gender as an independent factor for ICU mortality.

Our study has several limitations. First, data were collected in one tertiary care center but with three ICUs with a very mixed population.

Second, we did not measure sex-hormone levels and did not stratify for age, because postmenopausal status is difficult to assess without measuring sex hormones. As experimental data showed an influence of sex hormones on outcome, this might alter our results. These data advocate further prospective trials with specific focus on this issue.

Third, we did not take into account marital status, which is meant to have an influence on mortality from sepsis [28]. Although having probably the same impact on both genders, the fact that data were collected only during winter and spring could theoretically represent another limitation.

Fourth, 81.8% of our patients are surgical, so results for medical patients might be different.

Finally, studies of gender-related differences are limited to nonrandomized designs. As in studies of other not-modifiable factors like hospital admission time, ethnicity, or genotype we cannot rule out the possibility of further confounders or interactions that were not addressable in our careful multivariate evaluation [31].

In our population, it seems as if only in a sepsis subgroup gender makes a difference and that if women develop sepsis, they do worse than men. Thus, as sepsis occurred more frequently in male patients, prevention measures for infections might be more important for men, but we should pay more attention to specific therapies for women, because of their higher attributable mortality if seriously infected.