The Impact of Gender on Outcome From Mechanical Ventilation

The Impact of Gender on Outcome From Mechanical Ventilation * Martin H.

Kollef MD, FCCP; Jeana D. O'Brien, MD; Patricia Silver, MEd

Objective: To determine the relation of gender to outcome for patients requiring mechanical Design: A prospective cohort study. Setting: Medical and surgical ICUs in two university-affiliated teaching hospitals. Patients: Three hundred fifty-seven patients requiring mechanical ventilation. Interventions: Prospective patient surveillance and data collection. Measurements and results: The primary outcome measure was hospital mortality. Secondary outcome measures included need for reintubation, hospital and ICU length of stay, duration of mechanical ventilation, and hospital charges. The hospital mortality rate for women (n=189) was significantly greater than the hospital mortality rate for men (n=168) (28.0% vs 17.3%; p=0.016). This difference in hospital mortality was observed despite similar baseline demographic characteristics, severity of illness, indications for mechanical ventilation, and number of dysfunctional organ systems in these two groups of patients. The duration of mechanical ventilation and ICU length of stay was also for female patients compared with male patients (p^0.056). In a logisticsignificantly longer female gender was found to be independently associated with hospital mortality regression analysis, 95% confidence interval [CIl=1.70 to 3.35; p=0.010). The presence (adjusted odds ratio [AOR]=2.38; of ARDS (AOR= 10.69; 95% CI=5.86 to 19.51; p<0.001), the number of dysfunctional organ systems (AOR=2.07; 95% CI=1.78 to 2.41; p<0.001), Acute Physiology and Chronic Health Evaluation (APACHE) II predicted mortality (AOR=1.15; 95% CI=1.11 to 1.19; p<0.001), and patient age (AOR=1.04; 95% CI=1.03 to 1.06; p<0.001) were also found to be independently associated with hospital mortality. The number of dysfunctional organ systems present at the start of mechanical ventilation was the major independent predictor of hospital mortality (54% of total explanatory power). Patient gender was the least important independent predictor of hospital mortality (5% of total explanatory power). Conclusions: In this patient cohort, women requiring mechanical ventilation were at greater risk for hospital mortality than men. Physicians should be aware that outcome differences according to gender can occur when evaluating or designing clinical trials involving mechanically ventilated patients. Future studies are necessary to determine the general applicability of these findings and to identify explanations for such observed gender-specific differences in outcome. ventilation.

Key words: critical care; gender; mechanical ventilation; patient outcomes

Abbreviations:

(CHEST 1997; 111:434-41)

AOR=adjusted odds ratio; APACHE=Acute Physiology and Chronic Health Evaluation; CI=confidence

interval; FIo2=fraction of inspired oxygen

ventilation represent needing mechanical Patients of all admitted subset individuals important an

an

to

ICU due to the intensity of medical services they

receive and their associated costs.1

Additionally,

*From the Department of Internal Medicine, Pulmonary and Critical Care Division (Drs. Kollef and O'Brien), the Depart¬ ment of General Medical Sciences (Dr. Kollef), Washington University School of Medicine; and the Department of Respi¬ ratory Therapy (Ms. Silver), Barnes and Jewish Hospitals of the BJC Health System, St. Louis. This work was supported in part by grants from the American Lung Association of Eastern Missouri, Merck & Co., Inc, and

in Healthcare. BJC Innovations received April 23, 1996; revision accepted August 15, Manuscript 1996. Care Divi¬ Pulmonary and Critical Reprint requests: Dr. Kollef, School Medicine, Box 8052, 660 S sion, Washington University of 63110.

Euclid Avenue, St. Louis, MO 434

approximately one third of all patients who enter an ICU do so for mechanical ventilation, making this one of the most common indications for admission.2 To our knowledge, the influence of gender on outcome from mechanical ventilation has not been well studied previously. Preliminary investigations suggest that differences in outcome do occur be¬ tween genders and that they can be explained po¬ by differences in severity of illness or pos¬ tentially sible referral biases.34 Similarly, among patients bypass surgery, who undergoing coronary artery often require mechanical ventilation postoperatively, mortality for women appears to be greater hospital than hospital mortality for men.56 Such observed in outcomes have resulted in a differences

clinical

Clinical

Investigations in Critical Care

call for well-performed investigations to better de¬ fine the relationship between gender and specific disease processes or surgical interventions that often require mechanical ventilation.78 Therefore, we per¬ formed a prospective cohort study of noncardiac surgery patients to investigate the impact of gender on outcome from mechanical ventilation. Materials

and

Methods

Study Location and Patients The study was conducted at two university-affiliated teaching hospitals: Barnes Hospital (900 beds) and Jewish Hospital (450 beds). During a 4-month period (July 1995 to October 1995), all patients receiving mechanical ventilation in the medical and surgical ICUs of these two hospitals were potentially eligible for this investigation. Patients were entered into the study if they were older than 18 years and required mechanical ventilation. Patients were excluded only if they had burns or trauma involving the head or face, if they had transferred from other hospitals and had already received mechanical ventilation, or if mechanical ventilation was required due to brain death in order to allow organ retrieval. The study was approved by the Washington University School of Medicine Human Studies Committee and the Institutional Review Board of Jewish Hospital. Data Collection A clinical research assistant made daily rounds in the partici¬ ICUs of each hospital. Relevant clinical data were re¬

pating

corded from patient medical records and bedside computerized nursing stations (EMTEK; EMTEK Health Care Systems Inc; Tempe, Ariz). Study variables recorded at the start of mechanical ventilation included the following: age, sex, race, medical insur¬ ance status, the number of dysfunctional organ systems, the ratio of the partial pressure of arterial oxygen to the inspired concen¬ tration of oxygen (Pa02/FIo2), ICU category (medical vs surgi¬ cal), hospital location (Barnes vs Jewish Hospitals), presence or absence of either COPD requiring medical treatment or ARDS, and the indication or diagnosis requiring mechanical ventilation. The main clinical outcome evaluated was hospital mortality. Secondary outcomes of interest also examined included the need for reintubation, hospital and ICU lengths of stay, duration of mechanical ventilation, and hospital charges. Patients were fol¬ lowed up prospectively until they were successfully weaned from mechanical ventilation, died, or were transferred to a long-term care facility with mechanical ventilation. All study variables were recorded in data collection books maintained at each of the

participating hospitals.

Definitions

prospectively as part of the and Chronic Health Physiology Evaluation II (APACHE II) scores and predicted mortality rates9 were calculated in a standard manner All definitions

were

original study design.

selected

Acute

using clinical data available from the first 24 h of intensive care. The organ system failure index, a measure of individual organ dysfunction, has been described previously.10 One point was given for dysfunction of each organ system using the following definitions: renal, a twofold increase in the baseline creatinine level or an absolute increase in the creatinine level by 176.8 jxmol/L (2.0 mg/dL);

hepatic,

a

rise in the

total bilirubin level

to

greater than

34.2

juimol/L (2.0 mg/dL); pulmonary, requiring mechanical ventila¬ tion for a diagnosis of pneumonia, COPD, asthma, pulmonary edema (cardiogenic or noncardiogenic), or Pa02 less than 60 mm Hg while receiving an FIo2 of 0.50 or greater, or the use of 10 cm H20 or more of positive end-expiratory pressure; bone marrow, the presence of disseminated intravascular coagulation, WBC count of less than 1.0X109/L, or a platelet count of less than 75X109/L; neurologic, new focal deficit (eg, hemiparesis follow¬ ing cerebral infarction) or new generalized process (eg, seizures or coma); GI, GI hemorrhage requiring transfusion, new ileus, or diarrhea lasting more than 24 h and unrelated to prior bowel surgery; and cardiac, acute myocardial infarction, cardiac arrest, or the new onset of congestive heart failure. Congestive heart

failure (eg, ischemic, volume overload) was defined as the presence of bilateral infiltrates on the chest radiograph resolving with the administration of diuretics. The duration of mechanical ventilation was taken from the exact time of tracheal intubation, either on the day of ICU admission or thereafter, to the time of discontinuation of me¬ chanical ventilation. The presence of ARDS was defined on the basis of the chest radiograph showing bilateral pulmonary infil¬ trates, a ratio of Pa02/FIo2 of 200 or less, regardless of thelevel of positive end-expiratory pressure, and a pulmonary artery occlusion pressure of 18 mm Hg or less or no clinical evidence of elevated left atrial pressure on the basis of the chest radiograph and other clinical data.11 All financial data were derived from the finance offices of the study hospitals. Professional fees for the services of physicians were not included in these calculations. Health insurance status was categorized as none for those individuals without any type of health insurance (uninsured, medially indigent, or patient pay), Medicare insurance, tradi¬ tional fee-for-service health insurance, and capitated health insurance obtained through either a health maintenance organi¬ zation or preferred provider organization. Patients who had primary Medicare coverage but who also possessed supplemental medical insurance were classified as having Medicare insurance. Statistical Analysis

All univariate comparisons were unpaired and all tests of significance were two tailed. Continuous variables were com¬ pared using Student's t test for normally distributed variables and Wilcoxon's Rank-Sum test for nonnormally distributed variables. The x2 statistic or Fisher's Exact Test was used to compare categorical variables. First, these data were compared between survivors and nonsurvivors. A second data analysis compared female and male patients. All values are expressed as the mean±SD (continuous variables) or as a percentage of the group from which they were derived (categorical variables), p values of <0.05 were considered to indicate statistical significance. We performed multiple logistic-regression analysis12 using a commercial statistical package.13 Logistic-regression analysis was used to estimate the adjusted odds ratio (AOR) of hospital mortality (dependent variable) for gender (independent variable) and to calculate the 95% confidence interval (CI) while control¬ ling for other significant variables. Variables with p<0.20 (Table 1) were entered into the multivariate analysis, along with gender, based on models that were judged a priori to be clinically sound. This was prospectively determined to be necessary to avoid producing spuriously significant results with multiple compari¬ son.14 To minimize the effect of collinearity, the most clinically relevant variable among related variables (eg, APACHE II predicted mortality instead of the APACHE II score) was entered into the logistic-regression model. A stepwise approach was used for new terms the

entering

limit for their acceptance

or

into model with 0.05 as the removal. This allowed us to itera-

CHEST / 111 / 2 / FEBRUARY, 1997

435

tively select only the variables that were significantly related to hospital mortality, as tested by the likelihood test. The relative importance of the independent predictor variables was examined by comparing the percentage of explanatory power (X2) each contributed to the prediction of hospital mortality. Model overfitting was examined by evaluating the ratio of outcome events to the total number of independent variables in the final model and specific testing for interactions among all significant predictor variables was included in our analyses.14

Demographics

Results

A total of 357 consecutive patients requiring me¬ chanical ventilation were evaluated. There were 189 women (52.9%) and 168 men (47.1%), with mean ages of 62.9± 16.5 years and 61.6± 17.6 years, re¬ were 162 (45.4%) post¬ spectively (p=0.480).andThere 195 (54.6%) nonoperative operative patients patients. Baseline demographic information and in¬ dications for mechanical ventilation are shown in Tables 1 and 2. The most common form of health insurance coverage was Medicare insurance (87.1%). Among the Medicare patients, 62.1% were older than 65 years of age with the remaining patients receiving Medicare for other reasons.

Hospital Mortality as an Outcome (23.0%) patients died during their hos¬ Eight-two Table 1 shows the clinical variables pitalization. associated with hospital mortality by significantly univariate analysis. Hospital nonsurvivors were more of age, were more likely to be olderhadthan 60 yearsAPACHE II scores, often women, greater more II APACHE rates, morality predicted greater more to be were likely systems, organ dysfunctional medical patients, and had a greater incidence of ARDS. The mean age of hospital nonsurvivors was greater than the mean age of hospital significantly survivors (69.0± 15.2 years vs 60.3± 17.1 years; Hospital nonsurvivors were more likely to p<0.001). have pneumonia, respiratory failure, or cardiac arrest/cardiogenic shock, and less likely to have trauma, COPD/asthma, drug overdose, pulmonary edema, or the postoperative state as their indication for mechanical ventilation compared with hospital survivors.

Comparison of Characteristics and Outcomes

According to Gender

Table 2 shows that there were no significant differences in any of the recorded patient character¬ istics between female and male patients. However, the overall hospital mortality rate for female patients (28.0%) was significantly greater than the overall 436

Table 1.Patient Characteristics

Mortality*

Stratified by Hospital

Nonsurvivors

Survivors

p Value

<50 50-60

10 (12.2) 6 (7.3) 66 (80.5)

73 (26.6) 41 (14.9) 161 (58.5)

<0.001

Female Male

53 (64.6) 29 (35.4)

136 (49.5) 139 (50.5)

0.016

46(56.1) 35 (42.7) 1 (1-2)

175 (63.6) 97 (35.3)

0.433

4(1.5)

0.454

Characteristic

Age, yr >60 Sex

Race

White Black Other Health insurance status None

Medicare Traditional

Capitated

(n=275)

(n=82)

3(1.1)

1(1.2)

75 (91.5) 1 (1-2)

236 (85.8)

16(19.5)

67 (24.4) 208 (75.6)

14(5.1) (7.6)

5(6.1)

COPD Yes No

21

66 (80.5)

Diabetes

22(26.8) 60 (73.2)

Yes No APACHE II

predicted 41.1±21.8 [41.9; 3.6 to 86.3] mortality, %

APACHE II

score

No. of dysfunctional

organ systems

Ratio of Pa02 to

FIo2

ICU

Medical

Surgical Hospital Barnes Jewish

ARDS Yes No

Indication for mechan¬ ical ventilation

Postoperative Pneumonia

COPD/Asthma Pulmonary edema

Trauma

Respiratory failuref Drug overdose Cardiac arrest/

21.7±4.9 11 to 34] 2.3±1.3 [2; 0 to 6]

[21;

0.361

0.140 53(19.3) 222 (80.7) 19.9±15.5 <0.001 [15.6; 0.2 to 78.0] <0.001 15.7±5.3 [15; 2 to 32] 1.0±1.0 <0.001 [1; 0 to 4]

254 ±134

278±125

0.156

63 (76.8) 19 (23.2)

114 161

(41.5) (58.6)

<0.001

55(67.1) 27 (32.9)

195 (70.9)

0.506

15 (18.3) 67 (81.7)

9 (3.3) 266 (96.7)

<0.001

16(19.5) 6 (7.3) 2 (2.5) 4 (4.9) 1(1.2) 43 (52.4) 0 (0.0) 10(12.2)

146(53.1) 8 (2.9) 10 (3.6) 18 (6.5) 13 (4.7) 70 (25.5) 4(1.5) 6 (2.2)

<0.001

[220; 59 to 542] [226; 60 to 552]

80(29.1)

cardiogenic shock *Means are expressed ±SD. Numbers in parentheses are percentage values. Numbers in brackets are median and range values. fMultifactorial, neurologic, upper airway protection, postendoscopic procedure, or unclear etiology for respiratory failure requiring mechanical ventilation.

Clinical

Investigations in Critical Care

Table 2.Patient Characteristics

Age, yr

<50

50-60

>60

Race

White Black Other Health insurance status

Medicare Traditional

Capitated

COPD

40(21.2) 29 (15.3) 120 (63.5)

43 (25.6) 18 (10.7) 107 (63.7)

0.333

114 (60.3) 74 (39.2) 1 (0.5)

107 (63.7) 58 (34.5)

0.369

3(1.8)

150 (89.3) 5 (3.0) 10 (5.9)

3(1.8)

0.511

Yes 47 (25.4) No 138 (74.6)

36 (23.2) 119 (76.8)

0.641

Yes No APACHE II

predicted mortality, % score

No. of dysfunctional

organ systems

ICU

Medical

ARDS

Indication for mechanical ventilation

Postoperative

Pneumonia

COPD/asthma Pulmonary edema

Respiratory failuref Drug overdose

(53.4) (46.6)

76 (45.2) 92 (54.8)

0.122

125 (66.1)

64 (33.9)

125 (74.4) 43 (25.6)

0.089

Yes 15 (7.9) No 174 (92.1)

9 (5.4) 159 (94.6)

0.331

78 (41.3)

84 (50.0) 6 (3.6) 4 (2.4) 9 (5.4) 8 (4.8)

0.613

6

Cardiac arrest/

8 (4.2) 8 (4.2) 13 (6.9)

(3.2)

63 (33.3) 2 (1.1) 11 (5.8)

50 (29.8)

2(1.2) 5 (3.0)

cardiogenic shock *Means are expressed ±SD. Numbers in parentheses are percentage values. Numbers in brackets are median and range values. Multifactorial, neurologic, upper airway protection, postendoscopic procedure, or unclear etiology for respiratory failure requiring mechanical ventilation.

hospital mortality rate for male patients (17.3%) (relative risk=1.87; 95% CI 1.25 to 2.79; p=0.016). The mortality rate for female patients was also greater than the mortality rate for male patients in the two largest subgroups of patients examined: the postoperative subgroup (female mortality rate, =

14.1%, male mortality rate, 6.0%; p=0.082) and the

respiratory failure subgroup (female mortality rate,

mortality rate, 32.0%; p 0.238). The duration of mechanical ventilation and ICU length of stay was also longer for female patients compared with male patients (p<0.056) (Table 3). Table 4 demonstrates that the hospital mortality rate was greater for female patients compared with male patients in both the medical and surgical ICUs despite similar severity of illnessA and numbers of wide range of organ system derangements. for both female II was APACHE scores observed and male patients (Fig 1). The hospital mortality rate for female patients was significantly greater than the mortality rate for male patients over the hospital most common ranges of APACHE II scores (ie, APACHE II scores ll to 30) (female mortality rate, 31.1%; male mortality rate, 18.1%; p=0.010). Figure 2 depicts the Kaplan-Meier 30-day survival curves for female and male patients. The survival curves diverge by day 3 following the start of me¬ chanical ventilation and remain divergent thereafter. In a logistic-regression analysis, female gender was found to be independently associated with hospital mortality (AOR=2.38; 95% CI 1.70 to 3.35; analysis (Ta¬ p=0.010). Multiple logistic-regression ble 5) also demonstrated that the presence of ARDS, the number of dysfunctional organ systems, the APACHE II predicted probability of hospital mor¬ tality, and patient age were independently associated with hospital mortality. The number of dysfunctional organ systems present at the start of mechanical ventilation was the most important independent predictor of hospital mortality accounting for 54% of the total explanatory power. Female gender was found to be the least important independent predic¬ tor of hospital mortality with 5% of the total explan¬ 42.9%; male

=

=

41 (24.4) 0.137 34 (18.0) 127 (75.6) 155 (82.0) 0.392 23.8±18.5 25.5±20.0 [19.2; 0.2 to 86.3] [16.8; 0.2 to 82.3] 0.166 16.6±5.8 17.4±5.8 [16; 2 to 31] [16; 3 to 34] 0.631 1.3±1.3 1.3+1.1 [1; 0 to 6] [1; 0 to 4] 0.556 268 ±125 276±130 [266; 60 to 548] [258; 59 to 552] 101 88

Surgical Hospital Barnes Jewish

Trauma

(n=168)

p Value

None 2(1.1) 161 (85.2) 10 (5.3) 16 (8.4)

Diabetes

APACHE II

(n=189)

Male

Female

Characteristic

Stratified by Gender*

=

atory power.

Analysis of Hospital Nonsurvivors

Female patients accounted for 64.6% (n=53) of the 82 hospital nonsurvivors. The APACHE II score of female nonsurvivors (n=53; APACHE II score=21.4 ±5.1) was not significantly different compared with male nonsurvivors (n=29; APACHE II score=22.1±4.6) (p=0.541). However, female nonsurvivors had significantly fewer organ system derangements compared with male nonsurvivors vs 2.6±1.5 (2.0±1.1 organs; p=0.022). The primary cause of death was attributed to pulmonary failure in 52.8% of female nonsurvivors and 41.4% of male nonsurvivors (p=0.321). For all other nonsurvivors, mortality was attributed to other organ system de¬

rangements.

Among hospital nonsurvivors, similar proportions CHEST/111 121 FEBRUARY, 1997

437

Table 3.Outcome Measures* Female (n=189) 53 (28.0)

Outcome

Hospital mortality, No. (%) Required reintubation, No. (%)

Male (n-168)

Duration of MV, h

Requiring MV >7 d, No. (%)

Hospital length of stay, d ICU length of stay, d Hospital charges, $

(95% CI)f 10.7 (1.9, 19.5) -1.9 (-8.6, 4.8) 0.6 (-30.8, 32.0)

29 (17.3) 21 (12.5) 85.3+187.1 [23; 2 to 1,320]

20 (10.6) 85.9± 104.0 [39; 2 to 532] 32 (16.9) 14.3±12.2 [10; 1 to 109] 6.8±7.1 [3; 1 to 39]

Groups

Observed Difference Between

20(11.9)

5.0 1.1

13.2 + 11.2 [9; 1 to 64] 5.8±6.6 [3; 1 to 43]

p Value* 0.016 0.570 0.001

(-2.3, 12.3) (-1.3,3.5)

0.179 0.232

1.0 (-0.4, 2.4)

0.056

0.261 55,719±48,871 51,614±47,781 4,105 (-5,939, 14,149) [37,389; 2,246 to 532,380] [33,460; 799 to 267,148] *Means are expressed ±SD. Numbers in parentheses are percentage values. Numbers in brackets are median and range values. MV, mechanical ventilation. fValues are given as observed differences in population means or observed differences in percentages. * Continuous variables compared using Wilcoxon Rank-Sum Test.

of female patients (83.0%) and male patients (86.2%) died in the ICU setting (relative risk=0.96; 95% CI=0.80 to 1.16; p=0.706). Male nonsurvivors (65.5%) were more likely to die while receiving mechanical ventilation compared with female non¬ survivors (47.2%), but this difference was not statis¬

tically significant (relative risk=1.39; 95% CI 0.94 to 2.05; p=0.111). Male nonsurvivors were also less to have mechanical ventilation actively with¬ likely drawn prior to death (20.7%) compared with female nonsurvivors (35.8%), but this difference was also not statistically significant (relative risk.0.58; 95% CI=0.26 to 1.28; p=0.154). The duration of me¬ chanical ventilation was longer for male nonsurvivors compared with female nonsurvivors (158.4±277.8 h vs 134.4± 115.1 h; p 129). Lastly, the duration of time from extubation until death was not statistically and female =

=

different between male patients (n=10) patients (n=28) who were extubated prior to death (20.5±34.1 h vs 16.8±39.8 h; p=0.788). Discussion

Our study demonstrated that women requiring mechanical ventilation were at greater risk for hos¬ Table

pital mortality compared with men. This difference in hospital mortality was observed despite similar baseline demographic characteristics, severity of ill¬ ness, indications for mechanical ventilation, and number of dysfunctional organ systems for both female and male patients. This investigation is unique in using multivariable techniques, which simultaneously take into account other variables strongly associated with hospital mortality, to dem¬ onstrate an independent association between female gender and hospital mortality. We also identified the number of dysfunctional organ systems, the APACHE II predicted probability of hospital mor¬ tality, patient age, and the presence of ARDS as additional variables independently associated with hospital mortality for this cohort of patients. Each of these variables contributed greater explanatory power to the prediction of hospital mortality com¬ pared to female gender. The specific influence of gender on clinical out¬ comes for patients requiring mechanical ventilation has not been well studied. An association between female gender and greater mortality from mechani¬ cal ventilation has been described previously. Kollef3 examined a cohort of mechanically ventilated pa-

4.Hospital Mortality According to ICU Category*

Male (n-168)

Female (n=189) No. of

Dysfunctional

Organ Systems

ICU

Category

Medical

101

Surgical

1.8±1.0 0.7±1.0

APACHE II Score

Hospital Mortality,

19.6±5.8 15.0±4.8

40 (39.6) 13 (14.8)

No.

No. of

(%)

are

between male and female patients, fComparing hospital mortalitybetween male and female

438

APACHE II Score

Hospital Mortality,

1.9±1.3 0.7±1.0

19.3±5.7 14.4±4.9

23 (30.3) 6 (6.5)

76 92

expressed ±SD. Numbers in parentheses are percentage values. p >0.2 patients. dysfunctional organ systems

*Means

Dysfunctional Organ Systems

for all

comparisons of APACHE

II

Clinical

No.

scores

(%)

p Valuef 0.199 0.072

and the number of

Investigations in Critical Care

procedures, particularly coronary artery revascular¬ ization, and among patients with various medical conditions, including asthma and coronary artery disease.6-15"19 Female patients in the present study were more likely to have medical conditions and less likely to be postoperative suggesting a potential referral bias. However, these differences were not

11 to 20

21 to 30

APACHE II Scores

Figure 1. Distribution of APACHE II scores and relationship of the APACHE II scores to hospital mortality for 189 female 168 male patients (black bars) requiring patients (white bars) and Number of patients for each range of mechanical ventilation. APACHE II scores is given by the solid lines (boxes for female patients, triangles for male patients). a military referral medical center and found that female patients had a significantly greater mortality rate compared with male patients. How¬ ever, multivariate analysis demonstrated that this mortality difference was accounted for primarily by differences in severity of illness and organ system derangements between the genders. This suggested the possibility of a referral bias, with female patients being "sicker' than male patients at the time of ICU admission. Similar referral biases have been pro¬ posed to account for gender-specific differences in outcome among patients undergoing various surgical

tients from

100 90 80

-

70

0, >

i<

6050

-

40

-

30

-

20

10

H

Male

(n 168) =

Female (n

=

189)

0 10

15

20

Survival Time (days) Figure 2.

male

25

30

Kaplan-Meier 30-day survival curves for female and

patients requiring mechanical ventilation.

In addition to referral biases, statistically significant. in differences physiologic, hormonal, and immuno¬ logic factors between female and male patients have also been postulated to influence outcome following intensive care, particularly among certain groups of solid organ transplant recipients.20'21 It is not surprising that gender-specific influences on outcome following mechanical ventilation have not been identified previously. The relatively small contribution of gender to hospital mor¬ independent current and previous as tality, suggested by our with more important fac¬ investigations,3 compared tors like severity of illness, may account for its lack of association with hospital mortality in previous stud¬ ies. Similarly, only recently have clinical studies identified age as an important independent determi¬ nant of mortality among patients requiring mechan¬ ical ventilation.22-24 However, even this relationship is controversial since other investigators have failed to identify age as an important independent deter¬ minant of mortality.2526 Such discrepant results among these investigations can be explained, in part, by differences in the patient populations examined. we have demonstrated that patient age, Additionally, like female gender, contributes less explanatory power to hospital mortality than underlying severity of illness. This certainly increased the difficulty of identifying age as a predictor of mortality. The identification of gender-specific differences in outcome may be important for several reasons. First, they may further refine our ability to predict out¬ comes for large groups of patients.727 This would be to account for in the design and especially important of clinical trials examining disease implementation with known or suspected gender differ¬ processes ences in incidence or outcome. Second, genderspecific analyses of outcomes can be used to identify differences in medical practices between female and male patients. Such observations can subsequently be used to improve medical practices for all patients irrespective of gender.16-28 Similar analyses of racespecific29 and insurance-specific30 outcomes have been used to suggest changes in critical care prac¬ tices. Our data indicate that male nonsurvivors were less likely to have mechanical ventilation withdrawn prior to death compared with female nonsurvivors. This suggests that the male nonsurvivors were more treated and as a result may have re¬ aggressively ceived needlessly prolonged mechanical ventilation CHEST 7111/2/ FEBRUARY, 1997

439

Table

5.Summary of Stepwise Logistic Regression Analysis for Predictors ofMortality From Mechanical Ventilation

Variable Presence of ARDS Female gender No. of dysfunctional organs

(1-organ APACHE II predicted mortality (10% increments)

increments)

Age, yr Intercept

P Value

10.69 2.38 2.07

5.86-19.51 1.70-3.35 1.78-2.41

<0.001 0.010 <0.001

1.15

1.11-1.19

<0.001

20.1

1.03-1.06 0.0001-0.0012

<0.001

9.9

1.04 0.0004

compared with female nonsurvivors. Third, predic¬ of hospital lengths of stay and costs can differ between genders implying that cost containment measures may need to be gender specific.31 Lastly, in societal practices, particularly among changes women (eg, increasing rates of smoking), may con¬ tribute to changes in clinical outcomes between the genders over time. Our study has several important limitations. First, it was performed at two academic referral hospitals serving similar patient populations. Therefore, our results may not be applicable to institutions with different patient populations. Second, our investiga¬ tion did not address access to medical care prior to Women have been demonstrated to hospitalization. have less access to basic health care compared with men.32 However, it is unlikely that this contributed significantly to our observed difference in hospital the similar severity of illness, demo¬ mortality given status of our female graphics, and health insurance and male patients. Third, we examined only patients requiring mechanical ventilation and not all ICU patients. Therefore, these results may not apply to the entire ICU population. Finally, we may not have examined other important variables, including body size, comorbidities, nutritional status, and degree of organ system derangements,33 which could have in tors

accounted for the observed difference outcome between female and male patients. In summary, our study suggests that genderspecific differences in outcome from mechanical ventilation can occur that are independent of sever¬ ity of illness. Additionally, we demonstrated that the process of medical care may differ between male and female patients requiring mechanical ventilation. This is supported by the greater rate of active withdrawal of mechanical ventilation among female patients compared with male patients. Future inves¬ tigations should be directed at confirming these results and identifying specific explanations, includ¬ ing possible treatment biases, which may explain observed gender-associated differences in outcomes. 440

(\2 Mortality

% of Explanatory Power

95% CI

AOR

for Predicting

10.9 5.2 53.9

Finally, more long-term follow-up studies are re¬ quired to determine whether gender-specific differ¬ ences in quality of life or functional status occur following mechanical ventilation. ACKNOWLEDGMENT: The authors wash to thank Daniel P.

Schuster, MD, for his review of this article.

1 Cohen

REFERENCES

IL, Booth FV. Cost

containment and mechanical ventilation in the United States. New Horizons 1994;

2

3 4

5 6

7

2:283-90

Groeger JS, Guntupalli KK, Strosberg M, et al. Descriptive

analysis

of critical care units in the United States: patient characteristics and intensive care unit utilization. Crit Care Med 1993; 21:279-91 Kollef MH. Do age and gender influence outcome from mechanical ventilation? Heart Lung 1993; 22:442-49 Moler FW, Palmisano JM, Green TP, et al. Predictors of outcome of severe respiratory syncytial virus-associated respi¬ ratory failure treated with extracorporeal membrane oxygen¬ ation. J Pediatr 1993; 123:46-52 Gehlot AS, Santamaria JD, White AL, et al. Current status of coronary arteiy bypass grafting in patients 70 years of age and older. Aust NZ Surg 1995; 65:177-81 He GW, Acuff TE, Ryan WH, et al. Determinants of

operative mortality in reoperative coronary artery bypass grafting. J Thorac Cardiovasc Surg 1995; 110:971-78 Bone RC, Higgins MW, Hurd SS, et al. Research needs and opportunities related to respiratory health of women. Am Rev

Respir Dis 8 Pinn VW. 9

1992; 146:528-35

Equity in biomedical research. Science 1995; Knaus WA, Draper EA, Wagner DP, et al. APACHE II: a 269:739

severity of disease classification system. Crit Care Med 1985;

13:818-29 10 Rubin DB, Wiener-Kronish JP, Murray JF, et al. Elevated Von Willebrand factor antigen is an early phase predictor of

lung injury in nonpulmonary sepsis syndrome. J Clin 1990; 86:474-80 Bernard GR, Artigas A, Brigham KL, et al. The American-

acute

11

12

13

Invest

European Consensus Conference on ARDS: definitions, mechanisms, relevant outcomes, and clinical trial coordina¬ tion. Am J Respir Crit Care Med 1994; 149:818-24 Hosmer DW, Lemeshow S. Applied logistic regression. 1st ed. New York: Wiley Interscience Publication, 1989; 25-81 SAS/STAT user's guide (vol 2). Cary, NC: SAS Institute, 1990; 1071-1126

Clinical

Investigations in Critical Care

J, Feinstein AR, Holford TR. The risk of determin¬ ing risk with multivariable models. Ann Intern Med 1993; 118:201-10 Jaglal SB, Tu JV, Naylor CD. Higher in-hospital mortality ina female patients following coronary artery bypass surgery:

IL, Lambrinos J. Investigating the impact of age

14 Concato

24 Cohen

15

107:1673-80 25 Chelluri L, Pinsky M, Grenvik A. Outcome of intensive care in the 'oldest' old critically ill patients. Crit Care Med 1992; 20:756-61 26 Pesau B, Falger S, Berger E, et al. Influence of age on outcome of mechanically ventilated patients in an intensive care unit. Crit Care Med 1992; 20:489-92 27 Hudson LD. Survival data in patients with acute and chronic lung disease requiring mechanical ventilation. Am Rev Respir Dis 1989; 140.-S19-24 28 Bergelson BA, Tommaso CL. Gender differences in clinical evaluation and triage in coronary artery disease. Chest 1995; 108:1510-13 29 Williams JF, Zimmerman JE, Wagner DP, et al. AfricanAmerican and white patients admitted to the intensive care unit: is there a difference in therapy and outcome? Crit Care Med 1995; 23:626-36 30 Kollef MH. Private attending physician status and the with¬ drawal of life sustaining interventions in a medical intensive care unit population. Crit Care Med 1996; 24:968-75 31 Paul SD, Eagle KA, Guidy U, et al. Do gender-based differences in presentation and management influence pre¬ dictors of hospital costs and length of stay after an acute myocardial infarction? Am J Cardiol 1995; 76:1122-25 32 Yuen EJ, Gonnella JS, Louis DZ, et al. Severity-adjusted differences in hospital utilization by gender. Am J Med Qual 1995; 10:76-80 33 Marshall JC, Cook DJ, Christou NV, et al. Multiple organ dysfunction score: a reliable descriptor of a complex clinical outcome. Crit Care Med 1995; 23:1638-52

study: provincial adult cardiac care network population based 18:99-107 Clin Invest Med

of Ontario. 16 Bickell NA,

1995;

Pieper KS, Lee KL, et al.

Referral patterns for coronary artery disease treatment: gender bias or good clini¬ cal judgement? Ann Intern Med 1992; 116:791-97 17 Krumholz HM, Douglas PS, Lauer MS, et al. Selection of patients for coronary angiography and coronary revasculariza¬ tion early after myocardial infarction: is there evidence for a gender bias? Ann Intern Med 1992; 116:785-90 18 Steyerberg EW, Kievit J, de Mol Van Otterloo JC, et al. Perioperative mortality of elective abdominal aortic aneurysm surgery: a clinical prediction rule based on literature and individual patient data. Arch Intern Med 1995; 155:199819

20

21

22

23

2004

Arrighi

HM. US asthma mortality. Ann Allergy Asthma Immunol 1995; 74:321-26 Kubo SH, Naftel DC, Mills RM Jr, et al. Risk factors for late recurrent rejection after heart transplantation: a multiinstitutional, multivariable analysis: Cardiac Transplant Research Database Group. J Heart Lung Transplant 1995; 14:409-18 Sarris GE, Moore KA, Schroeder JS, et al. Cardiac transplan¬ tation: the Stanford experience in the cyclosporine era. J Thorac Cardiovasc Surg 1994; 108:240-51 Cohen IL, Lambrinos J, Fein IA. Mechanical ventilation for the elderly patient in intensive care: incremental charges and benefits. JAMA 1993; 269:1025-29 Swinburne AJ, Fedullo AJ, Bixby K, et al. Respiratory failure in the elderly: analysis of outcome after treatment with mechanical ventilation. Arch Intern Med 1993; 153:1657-62

on

of mechanical ventilation using a population of 41,848 patients from a statewide database. Chest 1995; outcome

CHEST / 111 / 2 / FEBRUARY, 1997

441