1.

Introction Community-acquired pneumonia remains a common and serious illness, in spite of the availability of new antimicrobials and successful vaccines. Approximately 10% to !% of patients with community-acquiredpneumonia who need hospitali"ation require intensive care unit #$C%& treatment '1(. )ortality in these patients has been reported to ran*e from +1% to ,-% '+(. .rior investi*ations have shown that the requirement for mechanical ventilation is associated with increased mortality compared with nonventilated patients ' ,/(. 0here are numerous studies evaluatin* clinical outcomes of patients with community-acquired pneumonia '+(, but these studies included an overall population and the outcome of the sub*roup of patients requirin* mechanical ventilation has not been evaluated in detail. 0he ob1ectives of this study were to evaluate variables associated with mortality of patients with community-acquired pneumonia who require mechanical ventilation from the first day of admission to the hospital and to determine the attributable morbidity and $C% mortality of community-acquired pneumonia. 2. Materials and methods +.1 .opulation study 2e used the database of a prospective, multicenter, international cohort of ,1- adult patients who received mechanical ventilation for more than 1+ hours from )arch 1, 133-, to )arch 1, 133-, at !1 $C%s ',(.2e included in the study 1+/ patients #+% of overall population included in the $nternational 4tudy of )echanical 5entilation& who required mechanical ventilation on the first day of admission to the hospital due to acute respiratory failure secondary to severe communityacquired pneumonia. 0he followin* information was collected on each patient6 #a & demo*raphic data #sex, a*e, previous functional status, wei*ht&, date of admission to $C%, and initiatin* mechanical ventilation7 #b & variables related to mana*ement #mode of ventilation, respiratory rate, tidal volume, applied positive end-expiratory pressure, airway pressures, need for vasoactive dru*s, need for neuromuscular bloc8ers, arterial blood *as analysis&. 0he arterial blood *ases correspond to the values obtained once daily at approximately -600 am. 0he ventilator variables correspond to the time that the arterial blood *ases were obtained. 0he use of neuromuscular bloc8ers and vasoactive dru*s #*iven for at least hours in a +/-hour period& was recorded daily for a maximum of +- days7 and # c & the development of the followin* events was assessed daily durin* the course of mechanical ventilation for a maximum of +- days6 acute respiratory distress syndrome, barotrauma, sepsis, renal failure, hepatic failure, and coa*ulopathy. A patient was considered to have any of the above conditions if it was present for at least + consecutive days. 0he patients were prospectively followed up for a maximum of +- days of mechanical ventilation and9or until dischar*e from the hospital and9or death. Community-acquired pneumonia was defined as an acute infection of the pulmonary parenchyma that is associated with at least some symptoms of acute lower respiratory infection #fever or hypothermia, ri*ors, sweats, new cou*h with or without sputum production or chan*e in color of respiratory secretions in a patient with chronic cou*h, chest discomfort, or the onset of dyspnea&, accompanied by the presence of an acute infiltrate consistent with pneumonia on a chest radio*raph, in a patient not hospitali"ed within 1/ days before onset of symptoms '!(.

+.+ 4tatistical analysis :esults were expressed as mean and 4;, median with the interquartile ran*e, and proportions as appropriate. 2e used the 4tudent t test or )ann-2hitney % test to compare continuous variables and the <+ test or =isher test to compare proportions. 4i*nificant univariate predictors for mortality were entered into a forward stepwise lo*istic re*ression analysis, with criteria entry and exit at . > .0, and . > .10, respectively. 0he model?s *oodness of fit was assessed by the method of @osmer and Aemeshow. 0o determine the attributable morbidity, defined as duration of mechanical ventilation and $C% len*th of stay, and attributable $C% mortality of community-acquired pneumonia, a matched study desi*n was used, matchin* each patient with severe community-acquired pneumonia needin* mechanical ventilation to a mechanically ventilated patient but without communityacquired pneumonia from our database. 0he matched variables were variables that we have reported to be associated with mortality in mechanically ventilated patients ',(6 # a & variables present at the be*innin* of mechanical ventilation #a*e, simplified acute physiolo*ical score '4A.4 $$(, limited activity as previous functional status&7 #b & variables related to patient mana*ement #use of vasoactive dru*s, use of neuromuscular bloc8ers, plateau pressure B , cm @+C&7 and #c & complications developin* over the course of mechanical ventilation #acute respiratory distress syndrome, sepsis, shoc8, renal failure, hepatic failure, coa*ulopathy, and ratio of .ao+ to =io+&. 0he variability ran*e for matchin* a*e was D10 years and for 4A.4 $$ was D10 points. . :esults Characteristics of patients who were dia*nosed with community-acquired pneumonia needin* mechanical ventilation are shown in 0able 1. Amon* these, /0 # +%& patients died durin* the $C% stay and /! # E%& patients died durin* the hospital stay. Amon* survivin* patients #n F -/&, the destinations at hospital dischar*e were as follows6 home #E/%&, nursin* home #11%&, acute facility #E%&, and other destiny #3%&. .1 .ro*nosis factors Gonsurvivin* patients had a *reater severity of illness at $C% admission and more shoc8, acute renal failure, and coa*ulopathy over the course of mechanical ventilation #0able +&. Althou*h a trend was observed toward a better evolution in patients who survived, there were no si*nificant differences in any blood *ases parameter except for ratio of .ao+ to =io+. 4i*nificant differences were found in this parameter before mechanical ventilation #mean ratio of .ao+ to =io+, 1/1 D E1 for nonsurvivors vs 1E+ D -3 for survivors7 . F .0,&, at first day #mean ratio of .ao+ to =io+, 1// D !3 vs 1E3 D -07 . F .0 & and last day of mechanical ventilation #mean ratio of .ao+ to =io+, 1E/ D E- vs +1 D E/7 . F .01&, and in the de*ree of improvin* oxy*enation #mean difference between last and previous value of ratio of .ao+ to =io+, / D -1 in nonsurvivors vs E1 D 3! in survivors7 . F .0,&. )ultivariate analysis selected independent variables si*nificantly associated with death6 4A.4 $$ *reater than /, #odds ratio, ,., '3,% confidence interval, 1.E-1+. (&, shoc8 #odds ratio, ,.E '3,% confidence interval, 1.E-10.1(&, and acute renal failure #odds ratio, .0 '3,% confidence interval, 1.1-/.0(&. .+ $mpact on outcome Cverall, 1+/ patients with severe community-acquired pneumonia requirin* mechanical

ventilation were successfully matched to a control patient without communityacquired pneumonia. Characteristics and outcomes of cases and controls are shown in 0able . 0here was no statistically si*nificant difference in $C% mortality amon* patients with or without community-acquired pneumonia needin* mechanical ventilation # +% vs ,%7 . F .,3&. 0here were no statistically si*nificant differences in duration of mechanical ventilation #median, , days 'interquartile ran*e, -- days( vs / days 'interquartile ran*e, +-! days(7 . F .+ & nor in len*th of stay #median, - days 'interquartile ran*e, /-1, days( vs ! days 'interquartile ran*e, /-1 days(7 . F .,,& compared with patients without community-acquired pneumonia. /. ;iscussion 0he main findin*s of our study were that community-acquired pneumonia requirin* mechanical ventilation is not associated with increase $C% mortality and it did not prolon* either the duration of mechanical ventilation or the $C% len*th of stay. =actors si*nificantly associated with outcome were severity of illness at $C% admission and development of cardiovascular and renal dysfunction over the course of mechanical ventilation. $n the outpatient settin*, the mortality rate of pneumonia remains in the ran*e of 1% to ,%, but amon* patients with community-acquired pneumonia who require hospitali"ation, the mortality rate increases to 1+%, and in specific populations such as those with bacteremia or patients requirin* admission to $C%, mortality ran*es from +1% to ,-% '+,E-10(. $n our series, the mortality in the $C% of patients with community-acquiredpneumonia requirin* mechanical ventilation was +%, in the lower ran*e of those previously reported 'E,11,1+(. 4everal studies have analy"ed the factors associated with outcome of patients with communityacquiredpneumonia '1 (. 0hese factors, analy"ed in a lar*e systematic review of the literature '+( and further studies '/,E-1,(, can be divided into baseline variables #male sex, a*e, presence of coexistin* disease, livin* in a nursin* home, underlyin* immunosuppression or neoplasm&7 severity at admission at hospital #4A.4 $$ B10-1 , decreased level of consciousness, tachypnea&7 factors related with the pneumonia #radio*raphic spread ofpneumonia or bilateral pulmonary involvement, aspiration pneumonia, infection caused by aerobic *ram-ne*ative patho*ens, the extent of lun* in1ury, bacteremia, requirement of mechanical ventilation, ineffective initial antibiotic therapy&7 and factors related with extrapulmonary complications #hypothermia, leu8openia, sepsis, septic shoc8, renal failure, the number of nonpulmonary or*ans that failed&. $n our cohort of mechanically ventilated patients, the variables associated with worse outcome were severity of illness on admission at $C%, shoc8, and acute renal failure. Althou*h variables related to oxy*enation and pulmonary parameters were not associated with outcome, we found a trend toward a better evolution in patients who survived. $n a validation study of American 0horacic 4ociety *uideline severity criteria for community-acquiredpneumonia by Hwi* et al 'E(, criteria reflectin* impaired oxy*enation #respiratory rate B 09min and .ao+9=io+ >+,0 at admission& were insensitive, nonspecific, and only wea8ly associated with mortality from severepneumonia. 0o our 8nowled*e, there are no previous studies concernin* the potential pro*nostic value of oxy*enation and pulmonary parameters over the whole course of mechanical ventilation in patients with severe community-acquired pneumonia. )ost studies have evaluated hypoxemia at presentation '1+,1,( or over the first +/ hours of mechanical ventilation '11(. 0he stren*ths of our study were the use of a matchin* procedure to examine the morbidity and mortality attributable to communityacquired pneumonia needin* mechanical ventilation. $n the present study, we found a control

similar to each patient with community-acquired pneumonia requirin* mechanical ventilation usin* 1+ predefined variables associated with mortality ',(. 2e have used data from an international multicenter database addin* to the validity and *enerali"ability of the obtained results. Cur sample si"e was lar*e and we only included mechanically ventilated patients, whereas most of the prior studies of severe community-acquired pneumonia had mixed populations of ventilated and nonventilated patients. 2e must address several limitations of our study. =irst, we have only used clinical criteria to dia*nosepneumonia, and therefore, it has not been ascertained the importance of other potential determinants of patient outcome such as the specific microor*anism responsible and the adequacy of initial antimicrobial therapy. $n this sense, althou*h some authors have observed that identifyin* the etiolo*ic a*ent and ad1ustin* treatment both impact patient outcome '1+,1!(, most studies have demonstrated that causative a*ent was not related with outcome '+,1E(. .rior antibiotic administration and adequacy of initial antimicrobial therapy were not considered in this study. 4econd, in our series, we did not study, as a pro*nostic factor, the radio*raphic spread of the pneumonia at $C% admission. @owever, in the validation study by Hwi* et al 'E(, the two baseline radio*raphic criteria #multilobar or bilateral involvement in chest radio*raph& had a hi*h specificity but remained insensitive and had low positive predictive values. 0hese observations could be due to the fact that the correlation of severe pneumonia with the extension of infiltrates visible on chest radio*raph was only moderate. 0hird, we only too8 into account limited activity as prior functional status and did not account for chronic obstructive pulmonary disease, alcoholism, diabetes mellitus, and chronic heart failure, which were shown to be related with poor outcome in previous studies '/,1/(. @owever, some other authors did not find comorbidities to influence outcome '3(.