Association of gender with outcomes in critically ill patients

The association of gender with medical care outcomes is an area of intense interest. However, the studies evaluating this relationship in critically ill patients have reached inconsistent results. Some of these studies are large population-based cohorts, and it may be difficult to apply these conclusions to critically ill patients [1]. Others have evaluated specific disease groups. There are some data to suggest that women have a delay in the diagnostic work up of coronary artery disease and do not receive appropriate and timely cardiac interventions [2‐4]. The various outcomes after coronary artery bypass graft (CABG) surgery are worse for women as well [5‐7]. The effect of gender on sepsis has varying results in different studies [8‐17] while trauma mortality is similar in both men and women [18‐21]. Men with chronic obstructive pulmonary disease (COPD) have been shown to have a higher mortality compared to women in some studies [22, 23] while others have found similar mortality rates [24]. At this time, however, there is no study on the association of gender with outcomes combining data from a large sample of critically ill patients with a variety of diagnoses from intensive care units (ICU) in the United States (US).

Studies from non-US ICUs have reached different conclusions regarding mortality based on gender, despite men receiving more aggressive care. Gender effects on outcomes in ICU patients in the US may be different from those in European and Canadian centers [25‐27], as there are differences in the health care delivery systems. Hence, there is a need to study this issue in critically ill patients in the US.

The Acute Physiology and Chronic Health Evaluation (APACHE) IV database (Cerner Corporation, Kansas City, MO, USA) provides a unique opportunity to answer this question. The data include a large number of ICUs of different sizes, location and teaching status through out the US. It has information on diagnosis, acute physiology scores, age, comorbidities, race, source of patient admission (emergency room, floor, and so on), and time spent at the source. There are data on critical therapeutic interventions and pertinent outcomes, such as length of stay, duration of mechanical ventilation and ICU and hospital mortality [28].

The objective of this study was to evaluate the APACHE IV database for outcomes data regarding gender and diagnosis for a large cohort of critically ill patients. Our hypothesis was that female gender would confer a higher risk for mortality.

We conducted a retrospective study of all adults in the APACHE IV database admitted to ICUs in the US between 1 January 2004 and 31 December 2008. Readmissions to the ICU were excluded from the analysis. The permission to access and utilize data was obtained from Cerner Corporation which supplied the data without any patient identifiers.

Our cohort consisted of 261,255 consecutive admissions, of which 144,254 (55.2%) were men and 117,001 (44.8%) were women. Other characteristics of the cohort are shown in Table 1. In bivariate analyses, there were statistically significant differences between men and women on all tested variables but most of these differences were small and most likely due to the large sample size. Of note, there were considerably more men admitted post-operatively (39.7%) than women (32.5%), and men were on average younger (60.6 years) than women (63.1 years).

In this analysis of large number of critically ill patients included in the APACHE IV database, the crude ICU mortality was higher for women compared to men. However, there was a significant interaction between gender and age. Since there were differences between men and women in severity of illness, ethnicity, comorbid conditions, pre-ICU length of stay, pre-ICU location and hospital teaching status, we controlled for these covariates. The adjusted mortality of women <50 years was lower compared to men while it was similar in patients ≥50 years of age. In the subgroup analysis, the adjusted mortality of women was higher after CABG and lower for COPD exacerbation. The adjusted mortality was similar for acute coronary syndrome, sepsis and trauma.

Our results are similar to Valentin et al. who studied a cohort of patients in Austrian ICUs [25]. They found similar illness adjusted mortality in men and women, despite men receiving an increased level of care. Fowler et al. also found a disparity in ICU care in favor of men, with an increased adjusted mortality in older women in Canadian ICUs [26]. In a Belgian retrospective study, Romo et al. concluded that women >50 years old had a higher mortality compared to men, while pre-menopausal women failed to show any survival benefit compared to men [27]. Our study is different from these studies as the women younger than 50 years had a lower adjusted ICU mortality, while it was not different in the older women. In our cohort, the APACHE IV and APS were higher in women but more men received active treatment in the ICU, including mechanical ventilation, emergent surgery, thrombolytics and CABG. They stayed longer on the ventilator and in the ICU, and had a higher readmission rate as well.

The age based difference in mortality can be explained on the basis of hormonal theory. Several laboratory studies have shown beneficial effects of estrogen. Estrogen preserves the cardiovascular and immunological function [30, 31] and tolerance to severe hypoxia, while 5 α-dihydrotestosterone (DHT) suppresses these functions. Immune responses are more vigorous in women, with greater antibody production and cell mediated immunity after immunization [32]. In an experimental model of Eschericia coli lipopolysaccharide administration, females have a greater pro-inflammatory response and norepinephrine sensitivity [33]. Hormonal manipulation in animal models reduces the sepsis related mortality in males. In a hemorrhagic shock model, female rodents have preserved splenocyte function and IL-3 secretion, along with IL-1 secretion by macrophages, in contrast to male rodents [34]. Ovariectomized rodents lose this advantage, which is later regained by administration of 17 β-estradiol [35]. Estrogens reduce the chemotaxis and activation of neutrophils and reduce pulmonary injury following hemorrhagic shock [36]. Administration of estrogen receptor- β agonist preserves gastrointestinal mucosal barrier function and improves outcome in a rodent model of sepsis [37]. 17 β-estradiol-treated males have diminished cardiomyocyte IL-6 production and improved cardiac function [38]. Despite this large accumulation of laboratory evidence on the beneficial effect of estrogens and deleterious effects of testosterone, conclusive evidence in clinical settings is not seen.

Our study showed significantly higher mortality in women after CABG compared to men while mortality in women with an acute coronary event was similar to men in multivariate analysis. Several studies have shown that the outcomes of women after CABG are worse than men, including mortality, duration of mechanical ventilation, length of stay in the ICU and hospital, and some long-term outcomes [5‐7]. However, some of the later studies have attributed this to the severity of illness and co-morbidities in women [39‐41]. In one study, increased mortality in women after CABG was attributed to infection [42]. Get with the Guidelines-Coronary Artery Database investigators reported no overall difference in mortality between men and women with acute myocardial infarction. However, ST-elevation myocardial infarction mortality was higher in women and there was an underuse of evidence-based therapy and a delay in therapeutic interventions [2]. Similar results have been reported by other researchers [3, 4, 43].

Women with sepsis had a similar adjusted mortality compared to men in our study. This is in agreement with several studies. In a large cohort of 192,980 patients with severe sepsis, Angus et al. concluded that there was no difference in mortality between men and women after adjusting for age, underlying comorbidities and the site of infection [8]. Martin et al. in a large epidemiologic study of the US population from 1979 to 2000 found that sepsis was more common in men (annual relative risk 1.28; CI: 1.24 to 1.32). However, they also did not find a difference in mortality in men versus women (22% versus 21.8%) [1]. Several other studies reported similar mortality in both genders [9‐11] while others have reported a lower mortality for women [12‐14]. In a recent prospective study, Nachtigall and colleagues found a significantly higher mortality for women admitted to the ICU with sepsis while the infection-related care was similar in both the genders [15]. In another study, female patients in the ICU with nosocomial infections had a higher mortality, after controlling for variables including age, severity score, immunocompromised status and admission source [16]. Eachempati and colleagues found that there was a higher mortality in elderly critically ill female surgical patients with sepsis [17]. The difference in outcomes in various studies can perhaps be explained on the basis of different care settings and the innate characteristics of the cohorts. A study has shown that an elevated level of 17 β-estradiol in both genders, increased progesterone in men and elevated testosterone in women is associated with worse survival [44]. However, we do not have any hormonal data available for our cohort.

We found no difference in mortality between the genders admitted with trauma. In a retrospective study by Magnotti and colleagues, there was no difference in mortality after blunt trauma between the genders, but men had a higher morbidity [18]. No difference in mortality was found in several other studies as well [19‐21].

The current study showed decreased adjusted mortality in women compared to men admitted with a diagnosis of COPD exacerbation to the ICU. This conclusion is in line with several studies [22, 23]. However, Celli and colleagues found that women who participated in the Toward a Revolution in COPD Health (TORCH) study had a lower mortality than men but this difference was not statistically significant after adjustment for important covariates including airflow obstruction and body mass index [24]. Our difference from the TORCH study could be explained on the basis that later excluded patients with significant comorbidities and long-term oxygen therapy.

The current study has several strengths. First, the study has a large sample size of critically ill patients from different demographics and with a wide array of diagnoses. Second, the data used came from a variety of hospitals with different sizes, teaching status, ICU coverage and geographic locations. Thus, the results of the study are generalizable. The weaknesses include a retrospective design that precludes from leading to any definitive conclusions. We cannot rule out other interactions and confounders such as hospital admission time, ICU staffing and so on that may play a role in the observed differences. It is also not clear why men have similar mortality compared to women, despite a trend towards getting more aggressive care. The differences in some of the ICU outcomes including mortality are statistically significant but may be of small magnitude to hold any clinical consequence. It is also possible that these results may be typical for ICUs that use the APACHE IV system and may not be applicable to other institutions. However, this study presents a snapshot of the current critical care practice in the United States.

This study shows that women younger than 50 years of age had a lower ICU mortality compared to men, after adjustment for covariates. On the other hand, women 50 years of age or older had an adjusted ICU mortality similar to men. There was a higher mortality of women admitted after CABG. However, the mortality of women admitted with COPD exacerbation was less compared to the men. There was no difference in the adjusted mortality between genders with acute coronary syndrome, sepsis and trauma.