validation of the modified multisystem organ failure score as a predictor of mortality in patients...

DOI 10.1378/chest.114.1.199 1998;114;199-206Chest

Lindsay Lawson, James A. Russell and Julio S.G. MontanerDavid M. Forrest, Ognjenka Djurdjev, Carlos Zala, Joel Singer, Respiratory Failure

Pneumonia andPneumocystis cariniiMortality in Patients With AIDS-Related Organ Failure Score as a Predictor of Validation of the Modified Multisystem

http://chestjournal.chestpubs.org/content/114/1/199

and services can be found online on the World Wide Web at: The online version of this article, along with updated information

ISSN:0012-3692)http://chestjournal.chestpubs.org/site/misc/reprints.xhtml(

without the prior written permission of the copyright holder.No part of this article or PDF may be reproduced or distributed3300 Dundee Road, Northbrook, IL 60062. All rights reserved. Copyright1998by the American College of Chest Physicians,Physicians. It has been published monthly since 1935.

is the official journal of the American College of ChestChest

1998 by the American College of Chest Physicians by guest on July 21, 2011chestjournal.chestpubs.orgDownloaded from


http://chestjournal.chestpubs.org/site/misc/reprints.xhtml

http://chestjournal.chestpubs.org/

Validation of the Modified Multisystem Organ FailureScore as a Predictor of Mortality in Patients WithAIDS-Related Pneumocystis carinii Pneumonia andRespiratory Failure*David M. Forrest, MD; Ognjenka Djurdjev, Bsc; Carlos Zala, MD; Joel Singer, PhD;Lindsay Lawson, MD, FCCP; James A. Russell, MD; and Julio S.G. Montaner, MD, FCCP

Study objectives: To validate a previously developed multisystem organ failure (MSOF) score withand without the addition of the lactate dehydrogenase (LDH) level as a predictor of survival to

hospital discharge in patients with AIDS-related Pneumocystis carinii pneumonia (PCP) and acute

respiratory failure (ARF).Design: Retrospective chart review between April 1, 1991, and September 30, 1996.Setting: University-affiliated tertiary care center in downtown Vancouver, British Columbia, Canada.Patients: All patients with PCP-relatedARF admitted to the ICU of St. Paul's Hospital during the studyperiod.Interventions: As putative prognostic instruments, data were extracted regarding the APACHE II(acute physiology and chronic health evaluation II), acute lung injury (ALI), AIDS, and modifiedMSOF scores, as well as LDH levels, at entry to the ICU. Patients were stratified based on an LDHlevel of < or ^:2,000 U/L and this threshold was assessed in its predictability of outcome when addedto each of the above scores. For APACHE II, the score was categorized in six groups and evaluatedwith and without inclusion ofthe LDH. Receiver operating characteristic curves were constructed forLDH and for each score with and without the LDH level to assess accuracy of prediction. The area

under each curve was calculated and compared to estimate the statistical significance of observeddifferences.Measurements and results: There were 40 admissions to the ICU of 38 patients with 52.5% mortality.The ALI and AIDS scores were not predictive of outcome. The modified MSOF and APACHE IIscores were significant predictors of survival and the performance of both was enhanced by theaddition of LDH.Conclusions: Both the APACHE II and the modified MSOF scores were significant predictors ofoutcome in the patient population studied. These results validate the modified MSOF score as an

effective predictor of survival to hospital discharge among patients with AIDS-related PCP whodevelop ARF and the performance of the score is enhanced by the addition of the LDH level.

(CHEST 1998; 114:199-206)Key words: acquired immunodeficiency syndrome; acute respiratory failure; Pneumocystis carinii pneumonia; prognosticscores

Abbreviations: ALI=acute lung injury; APACHE=acute physiology and chronic health evaluation; ARF=acute respiratoryfailure; LDH=lactate dehydrogenase; MSOF=multisystem organ failure; VC?=Pneumocystis carinii pneumonia;ROC=receiver operating characteristic

*From the British Columbia Center for Excellence in HIV/AIDS(Drs. Forrest, Zala, and Montaner, and Mr. Djurdjev) and theDepartment of Medicine (Drs. Lawson, Russell, and Montaner),St. Paul's Hospital; the Division of Critical Care Medicine (Drs.Forrest and Russell), Division of Respiratory Medicine (Dr.Lawson), Department of Health Care and Epidemiology (Dr.Singer), and Faculty of Medicine (Drs. Lawson, Russell, andMontaner), University of British Columbia; and the CanadianHIV Trials Network (Mr. Djurdjev and Drs. Singer and Mon¬taner), Vancouver, British Columbia, Canada.Manuscript received July 11, 1997; revision accepted December26, 1997.Reprint requests: Julio S.G. Montaner, MD, FCCP, BC CenterforExcellence in HIV/AIDS, Room 667, St. Paul's Hospital, 1081Burrard St, Vancouver, BC, Canada V6Z 1Y6

TJ neumosystis carinii pneumonia (PCP) is associ-.*¦ ated with respiratory failure in 5 to 30% of cases1and has been reported to be the most common cause

of acute respiratory failure (ARF)2 and of admissionto the ICU35 among HIV-seropositive patients. ARFin the setting of AIDS-related PCP has been associ¬ated with a poor prognosis and high in-hospitalmortality.68 However, this has been variable throughthe AIDS epidemic,8 at least in part due to changesin patient selection for ICU support and increaseduse of corticosteroids as adjuvant therapy.6-8 Thus,identification of subgroups destined to have a high

CHEST/114/1/JULY, 1998 199 1998 by the American College of Chest Physicians

by guest on July 21, 2011chestjournal.chestpubs.orgDownloaded from


mortality is important in advising patients on treat¬ment options.We have developed previously a prognostic scor¬

ing instrument based on a modification of a multi¬system organ failure (MSOF) score9 and found it

highly predictive of outcome of ARF secondary toAIDS-related PCP, particularly when the lactatedehydrogenase (LDH) level was added to thescore.10 In brief, the MSOF score is calculated byassigning one point for each organ system failure.Organ systems included are respiratory, cardiovascu¬lar, renal, hepatic, and hematologic, as well as GIbleeding and encephalopathy prior to intubation. Wetherefore undertook the present study to reevaluatethe prognostic scores assessed in our original studyand to validate the modified MSOF score with andwithout the addition of the LDH level as a predictorof survival to hospital discharge at the time of ICUadmission in a subsequent set of patients with AIDS-related PCP and ARF.

Materials and Methods

Assessment of the MSOF score was undertaken originally in

patients in the period between January 1, 1985, and April 1,1991.10 For the present study, all cases of patients with AIDS-related PCP admitted to the ICU at St. Paul's Hospital withrespiratory failure between April 1, 1991, and September 30,1996, were identified through a computerized search of hospitalrecords. Patients with respiratory failure who were not admittedto the ICU were not included. All ICU cases were used in theanalysis, whether or not mechanical ventilation was used. Re¬peated admissions were considered separate episodes of PCP ifhospital admissions were separated by at least 30 days. Chartswere reviewed by hand using a standardized case report form. Asin our previous study, data were recorded regarding theAPACHE II (acute physiology and chronic health evaluationII),11 acute lung injury (ALI),12 AIDS,13 and modified MSOF10scoring instruments, as well as LDH levels, at entry to the ICU.Reviewers were not blinded as to the goals of the study. Theoutcome assessed was survival to hospital discharge.

Patients with both microbiologically proven PCP by bronchos¬copy or at autopsy and those with presumptive PCP were

included in the analysis. A presumptive diagnosis of PCP was

made if HIV-seropositive patients had a compatible clinicalhistory and physical examination, CD4+ lymphocyte count

<200/mm3, hypoxemia with alveolar-arterial oxygen gradient of>50 mmHg, a chest radiograph showing diffuse alveolar inter¬stitial infiltrates, and if no other diagnosis was established. ARFwas defined as Pa02/fraction of inspired oxygen <150.

Patients were stratified based on an LDH level of <2,000 or >

2,000 U/L to assess predictability of outcome, as previouslydescribed.10 This threshold was assessed in its predictability ofoutcome when added to the MSOF, AIDS, and ALI scores, thusincreasing the range of each score by one. This threshold cannotbe used for the APACHE II score, which requires gradation ofdegree of abnormality. Thus, as in our previous study,10 thethreshold for LDH in this instance was set at 550 U/L, 1,650 U/L,2,000 U/L, and 2,130 U/L, so increasing the range of the score byfour points. As in our previous study, the APACHE II score was

categorized in six groups.10Receiver operating characteristic (ROC) curves were con¬

structed for each score and for LDH and then repeated for eachscore plus the LDH level to assess the accuracy of prediction ofthe scores. In brief, we plotted the true-positive rate (sensitivity)on the vertical axis of a graph against the false-positive rate

(1-specificity) on the horizontal axis for each threshold. An idealtest has a high true-positive rate and a low false-positive rate andtherefore should produce a curve close to the upper left corner ofthe graph. In contrast, a test with no predictive value should haveequal numbers of true and false positive results and thus wouldproduce a curve along the identity line along the diagonal.14 Thearea under each curve so generated was calculated1415 andcompared with curves for other scores to estimate the statisticalsignificance of observed differences.16

Categorical variables were compared using Mantel-HaenszelX2. The MSOF score with and without modification by additionof the LDH level was further assessed as predictive of outcome(mortality) by logistic regression. All statistical tests were twosided and p<0.05 was considered statistically significant.

Results

Between April 1, 1991, and September 30, 1996, a

total of 40 admissions to the ICU for ARF were

identified in 38 patients. Complete data were avail¬able for all patients. Two patients were admittedtwice to the ICU; all repeat admissions were sepa¬rated by at least 30 days. A total of 37 (98%) of thesepatients were male and the median age was 38 years.Thirty (75%) of the cases represented a first episodeof PCP and the diagnosis of PCP was confirmedmicrobiologically in 33 episodes (83%). Of the seven

patients with presumptive PCP, four never hadbronchoscopy performed and three had bronchos¬copy >7 days after initiation of therapy. All patientsreceived combined therapy with anti-PCP drugs andcorticosteroids. Among the 40 patients admitted tothe ICU, 25 (63%) received mechanical ventilatorysupport. Baseline characteristics are summarized inTable 1.Twenty-one (52.5%) of the patients died during

the index hospitalization. The mortality rate was

54.6% among patients with confirmed PCP and was

42.9% among those with presumptive PCP. Thisdifference is not statistically significant (p=0.884).The ALI and AIDS scores were not predictive of

outcome (Table 2). As shown in Table 2, the MSOFand APACHE II scores were significantly differentbetween those who survived compared with thosewho did not (p=0.012 and p=0.004, respectively).Based on a threshold of 2,000 U/L, LDH alone was

not significantly different in those patients who diedcompared with those who survived (data not shown;p=0.31). As shown in Table 3, the predictability ofmortality by the MSOF and APACHE II scores was

slightly enhanced by the addition of the LDH(p=0.002 andp=0.001, respectively). The predictivevalue of the MSOF score (with and without LDH)was retained in the subgroup of patients who were

200 Clinical Investigations in Critical Care



Table 1.Baseline Characteristics of Patients Admittedto the ICU With AIDS-Related PCP and RespiratoryFailure Between April I, 1991, and September 30,

1996*

Table 3.Univariate Analysis of Mortality Accordingto the Scoring Systems, Including LDH

Characteristic Baseline Measurement

Age, yr 38 (25-64)Gender, % male 98Anti-RTV therapy, % 35PCP prophylaxis, % 30First episode, %75Symptom duration, d 14 (2-60)PCP confirmed, % 83Duration hospital to ICU admission, d 3 (0-14)APACHE II 19 (13-35)

ALI 3.25 (1.67-4)AIDS 3 (1-6)

MSOF 1 (1-4)LDH 1,564 (563-4,287)

*All values are medians with ranges shown in parentheses. Anti-RTV

therapy= antiretroviral therapy.

mechanically ventilated, as well as those in whom thediagnosis of PCP was confirmed. Addition of theLDH to the ALI and AIDS scores enhanced theirpredictive accuracy as well (Table 3).

Table 2.Univariate Analysis ofMortality Accordingto the Scoring Systems Without LDH

Scoring System No. Mortality p Value*

MSOF123

>4APACHE II<1819-2223-2425-2728-32>32

ALI0.0-2.52.6-3.03.1-3.5>3.5

AIDS1234

>5LDH

500-750750-1,0001,000-1,2501,250-2,500>2,500

27751

19134202

811129

1178

131

138

18

4157100100

3254100100

100

38277567

0535054100

10033256762

0.012

0.004

0.054

0.50

Scoring System No. Mortality p Value*

MSOF+LDH1 192 13

3 6>4 2

APACHE II+LDH<18 1119-22 1623-24 525-27 528-32 1>32 2

ALI+LDH0.0-2.5 52.6-3.0 123.1-3.5 11>3.5 12

AIDS+LDH2 113 154 10

>5 4

266983100

273880100100100

20425575

27675075

0.002

0.001

0.026

0.153

0.32

*Based on Mantel-Haenszel x2

*Based on Mantel-Haenszel x2-

As shown in Figure 1, ROC curves were generatedfor each of the scores alone and for LDH. TheMSOF and APACHE II scores were superior to theother scores and LDH alone, based on the largerarea under the ROC curve for those plots, althoughthere was no statistically significant difference be¬tween the areas under the ROC curves (Table 4). Asshown in Figure 2, ROC curves were generatedadding LDH to the MSOF, APACHE II, ALI, andAIDS scores. The predictability of the MSOF andAPACHE II scores was enhanced once again by theaddition of the LDH level (Table 4).The analysis was unchanged when restricted to

those patients who received mechanical ventilationand to those in whom the diagnosis of PCP was

confirmed (data not shown). The MSOF andAPACHE II scores remained significant predictorsof mortality among patients with confirmed PCP(p=0.016 for MSOF score; p-0.009 for APACHEII), while the other scores were not. Only the MSOFand APACHE II scores were predictive of outcomefor patients who received mechanical ventilation(p=0.042 for MSOF score; p=0.019 for APACHEII). The predictive accuracy of the scores was en¬

hanced also by the addition of LDH to the scores.

A multivariate analysis was performed to assess thepredictive ability of the MSOF score (with andwithout inclusion of the LDH) relative to the otherpredictive scores. With the MSOF score included inthe regression analysis, no other score contributed to

prediction of outcome (Table 5).

CHEST/ 114/1 /JULY, 1998 201



0.4 0.61-SPECIFICITY

1.0

Figure 1. ROC curves for scoring systems without LDH and for LDH itself. Each point on the curve

represents the true-positive rate (sensitivity) and false-positive rate (1-specificity) at different gradationsof each score (Table 2). The curves for the MSOF and APACHE II scores are the best predictors ofoutcome.

Discussion

Our results successfully validate the modifiedMSOF score as a predictor of outcome for patientsadmitted to the ICU with AIDS-related PCP andARF. However, in contrast to our earlier findings, wefound that the APACHE II score was a significant

Table 4.Areas Under the ROC Curves for ScoringSystems With and Without LDH*

Area UnderScoring System ROC Curve Scoring System

Area UnderROC Curve

MSOFAPACHE IIALIAIDSLDH

0.6820.7590.6180.6770.551

MSOF+LDH 0.776APACHE II+LDH 0.792

ALI+LDH 0.697AIDS+LDH 0.627

*The difference between areas under the curves for different scores

is not statistically significant.

predictor of outcome and was a slightly better pre¬dictor of outcome in the more recent patient groupthan the MSOF score. The LDH level alone at timeof ICU admission was not predictive of outcome, butthe combination of LDH with all the prognosticscores enhanced their predictive ability. Thus, addi¬tion of the LDH to the MSOF and APACHE IIscores significantly improved their predictive accu¬

racy. Similarly, the ALI score itself was not a signif¬icant predictor of outcome, but it was predictive ofmortality when combined with the LDH. However,the AIDS scoring system, even in combination withthe LDH level, is a poor predictor of outcome in thispatient population.

Validation of prognostic scoring systems is an

important component of their development. Giventhe intended utility of scores in AIDS-related PCPand ARF in identifying patients who are destined tohave a poor outcome and thus to assist in decision




0.0 0.2 0.4 0.61-SPECIFICITY

0.8 1.0

Figure 2. ROC curves for scoring systems, including LDH. The curves for the MSOF and APACHEII score+LDH are the best predictors of outcome, although addition of LDH enhances thepredictability of all scores.

making in this group of critically ill patients, suchvalidation seems especially crucial. Nevertheless, of16 tools developed that have been proposed to

identify prognostic variables in PCP,41317-30 only 3have been studied in order to validate them.31-34 Thismay explain, in part, the resulting reluctance ofclinicians to use these scores in clinical practice. Twoof the scoring systems were developed for use in

Table 5.Multivariate Analysis of Mortality byLogistic Regression According to MSOF Score and

LDH

Scoring SystemOdds Ratio(95% CI)* p Value

MSOF1.2826 3.606 4.4433 0.0350(1.094-11.884)

MSOF+LDH 1.4486 4.257 6.8168 0.0090(1.435-12.630)

*CI=confidence interval.

patients with PCP at time of initial presentation1729and were not specifically designed for prognostica¬tion of patients with ARF due to PCP. While bothwere found predictive of mortality,3132 neither hasbeen assessed in a critically ill population.The third score is the APACHE II system score of

Knaus and Draper.11 This tool has been validated as

predictive instrument in a variety of critically illpatients,35 but such validation was conducted prior tothe rapid escalation in the AIDS epidemic. Smithand colleagues34 and subsequently Chu33 have eval¬uated the ability of the APACHE II score to predictoutcome in patients with PCP and respiratory fail¬ure. Smith et al34 assessed the system in patients withPCP and respiratory failure specifically, while Chu33evaluated its performance in AIDS patients withrespiratory failure due to a variety of causes. Inneither case was the system found to be a goodpredictor of outcome, and observed mortality ex¬

ceeded that predicted by the model. This may be

CHEST / 114 / 1 / JULY, 1998 203



MSOF (p=0.038) AIDS (p=0.085)

a.

5?

70

60

50

40

30

20

10

0 M/a e^12 3 4

MSOF

a.o

70

60

50

40

30

20

10

0

wId Hftjfl m'&£^_

3

AIDS

>=5

ALIS (p=0.850) APACHE II (p=0.0002)

70

60

50

40

30

20

10

0 m WA Wl0.0-2.5 2.6-3.0 3.1-3.5

ALIS

>=3.5

70

60

50

40

30

20

10

0

///

Ma.

Y//A Original dataValidation data

h. I I

Figure 3. Comparison of distribution of scores in previous10 and present patient populations studied.The differences in distribution of scores (especially for the APACHE II and MSOF scores) indicate thatthe two patient groups are different.

ffin<=18 19-22 23-24 25-27 28-32 >32

APACHE II

related to the heterogeneity of the populations stud¬ied and the characteristics of the particular ICU,33 as

well as the inaccuracy of the predictive tool inmechanically ventilated patients.34 A modification ofthe instrument by addition of LDH values22 hasbeen shown to enhance the predictive accuracy ofthe APACHE II system, but this has not beenvalidated. In contrast, one group found theAPACHE II score to be a useful prognostic indicatorin AIDS patients on admission to the hospital, butthis was not specific for PCP or critically ill pa¬tients.36The APACHE II score (with or without the

addition of LDH) was a modestly better predictor ofoutcome in our patient group. While this supportsthe work of Benson and colleagues,22 it refutes thatof Smith and coworkers.34 Its effectiveness as a

predictor in this sample and not in our previousstudy and inconsistency with the results of otherinvestigators, however, suggest the sensitivity of theAPACHE II score to variability in the patient pop¬ulation studied. It may also reflect random variation,given the relatively small sample sizes in this and the

previous study and multiple comparisons, but this isunlikely, since the results are highly statisticallysignificant (p<0.005). Similarly, the ALI score whencombined with the LDH level was a significantpredictor of outcome, but was not when evaluatedalone (p=0.054). Although this may be the conse¬

quence of relatively small sample size, it also sug¬gests that the score is not highly or consistentlypredictive and that it may also be sensitive to

heterogeneity of the patient population studied. Thisagrees with the findings of Peruzzi et al,4 who foundthat markers of oxygenation and radiographic abnor¬malities on admission to the ICU were not predictiveof outcome. In contrast, the modified MSOF score

(with or without LDH) consistently predicted out¬come in two separate samples, supporting its predic¬tive utility in the population of patients with PCP-related respiratory failure and its use as the preferredpredictive physiologic score in this patient group.

Indeed, there are substantial differences betweenthe populations studied in our original analysis andthe current study. First, the era of study is different.The outcome of ARF secondary to AIDS-related




PCP has changed throughout the AIDS epidemic.8While ICU outcome in early years (1981 to 1985)was generally poor with survival to hospital dischargeof <10%, there was substantial improvement in

mortality of patients with PCP and respiratory failurebetween 1986 and 1988, with survival to hospitaldischarge of approximately 40%. However, since1989, there has been a substantial decrease in sur¬

vival to hospital discharge to <25%.6'8-10'37'38 Thesedifferences are likely due to changes in patientselection for ICU support and increased use ofcorticosteroids as adjuvant therapy68 and a dismalprognosis for patients who have failed to respond tomaximal therapy.37 The patient set used to assess thescoring instrument initially was admitted to ICUduring the period January 1, 1985, to April 1, 1991,thus spanning the three eras of ICU experiencedescribed for PCP patients with respiratory failure.In contrast, the patients included in the presentstudy were all cared for in the modern era.

There are other clinically significant differencesbetween the two groups. Those in the derivationstudy had a mortality rate of 67%, while that of thepatient group in the current study was 50%. Further,all patients in the current study received combinedcorticosteroids and anti-PCP therapy, while only85% of the earlier group received adjuvant steroids.Although 90% of patients in our first study requiredmechanical ventilation, only 63% were mechanicallyventilated in the present study. Finally, there were

important differences in the distribution of patientsacross the various score categories, particularly forthe APACHE II score (Fig 3). This suggests that themodified MSOF score is resistant to secular trends in

patient treatment and selection and may explain theimproved performance of the APACHE II score inthe present patient group. Moreover, we found thatthe MSOF score (with or without LDH) performedwell in the subgroup of patients with PCP-relatedrespiratory failure who required mechanical ventila¬tion and in those in whom the episode of PCP was

proven bacteriologically (data not shown). Thesewere the only groups studied in our original sample.The modified MSOF score reflects burden of

illness in various organs. We speculate, therefore,that survival among patients with ARF secondary toAIDS-related PCP is greatly influenced by the de¬gree of compromise and physiologic derangement ofother organ systems at the time of admission to theICU. This seems logical clinically and supports thesimilar predictability of physiologic scores in othercritically ill populations,91139 including the perfor¬mance of the APACHE II score in this study andthat of Benson and colleagues.22 Indeed, others havefound that physiologic parameters at ICU admissionfor similar patients, either assessing respiratory func¬

tion alone2324 or incorporating measures of severityof other organ dysfunction,22'3334 are valuable pre¬dictors of outcome.

In conclusion, therefore, our results successfullyvalidated the modified MSOF score as an effectivepredictor of survival to hospital discharge amongpatients with AIDS-related PCP in whom ARFdevelops. We have also confirmed that the perfor¬mance of the score is enhanced by the addition of theLDH level. The consistent performance, simplicity,and clinical relevance of the modified MSOF score

make it a useful and attractive tool for clinical andresearch purposes. Its ease of calculation shouldfacilitate use of the instrument in prognostication forpatients. This can inform a discussion about the roleand desirability of pursuing critical care supportgiven anticipated outcomes and should therefore aidpatients and their caregivers in clinical decisionmaking.

REFERENCES1 The National Institutes of Health-University of California

Expert Panel for Corticosteroids as Adjunctive Therapy forPneumocystis Pneumonia. Consensus statement on the use ofcorticosteroids as adjunctive therapy for Pneumocystis pneu¬monia in the acquired immunodeficiency syndrome. N EnglJ Med 1990; 323:1500-04

2 Torres A, El-Ebiary M, Marrades R, et al. Aetiology andprognostic factors of patients with AIDS presenting [with]life-threatening acute respiratory failure. Eur Respir } 1995;8:1922-28

3 Rogers P, Lane H, Henderson D, et al. Admission of AIDSpatients to a medical intensive care unit: causes and outcome.Crit Care Med 1989; 17:113-17

4 Peruzzi W, Skoutelis A, Shapiro B, et al. Intensive care unit

patients with acquired immunodeficiency syndrome andPneumocystis carinii pneumonia: suggested predictors ofhospital outcome. Crit Care Med 1991; 19:892-900

5 Brooks K, Ong R, Spector R, et al. Acute respiratory failuredue to Pneumocystis carinii pneumonia. Crit Care Clin 1993;9:31-48

6 Hawley P, Ronco J, Guillemi S, et al. Decreasing frequencybut worsening mortality of acute respiratory failure secondaryto AIDS-related Pneumocystis carinii pneumonia. Chest1994; 106:456-59

7 Rosen M, Palo VD. Outcome of intensive care for patientswith AIDS. Crit Care Clin 1993; 9:107-14

8 Wachter R, Luce J. Respiratory failure from Pneumocystiscarinii pneumonia. Chest 1994; 106:1313-15

9 Bihari D, Smithies M, Gimson A, et al. The effects ofvasodilation with prostacyclin on oxygen delivery and uptakein critically ill patients. N Engl J Med 1987; 317:397-403

10 Montaner J, Hawley P, Ronco J, et al. Multisystem organfailure predicts mortality of ICU patients with acute respira¬tory failure secondary to AIDS-related PCP. Chest 1992;102:1823-28

11 Knaus W, Draper E. APACHE II: a severity of diseaseclassification system. Crit Care Med 1985; 13:818-28

12 Murray J, Matthay M, Luce J, et al. An expanded definition ofthe adult respiratory distress syndrome. Am Rev Respir Dis1988; 138:720-23

CHEST/114/1 /JULY, 1998 205



13 Justice A, Feinstein A, Wells C. A new prognostic stagingsystem for the acquired immunodeficiency syndrome. N EnglJ Med 1989; 320:1388-93

14 Hanley J, McNeil B. The meaning and use of the area undera receiver operating characteristic (ROC) curve. Radiology1982; 143:29-36

15 McNeil B, Hanley J. Statistical approaches to the analysis ofreceiver operating characteristic (ROC) curves. Med DecisMaking 1984; 4:137-56

16 Hanley J, McNeil B. A method of comparing the areas underreceiver operating characteristic curves derived from thesame cases. Radiology 1983; 143:839-43

17 Nichols L, Black J, Sattler F, et al. A quantitative scoringsystem for evaluating the response to therapy of Pneumocystiscarinii pneumonia [abstract]. Fifth International Conferenceon AIDS, Montreal, Canada, 1989

18 Beck E, French P, Helbert M, et al. Improved outcome ofPneumocystis carinii pneumonia in AIDS patients: a multi-factorial treatment effect. Int J STD AIDS 1992; 3:182-87

19 Matthews W, Ferdon E, Bennett C, et al. Evaluating institu¬tional performance in AIDS-associated Pneumocystis carinii

pneumonia: a risk-adjustment approach. J Clin Epidemiol1989; 42:421-25

20 Benfield T, Vestbo J, Junge J, et al. Prognostic value ofinterleukin-8 in AIDS-associated Pneumocystis carinii pneu¬monia. Am J Respir Crit Care Med 1995; 151:1058-62

21 Bennett C, Weinstein R, Shapiro M, et al. A rapid preadmis¬sion method for predicting inpatient course of disease forpatients with HIV-related Pneumocystis carinii pneumonia.Am J Respir Crit Care Med 1994; 150:1503-07

22 Benson C, Spear J, Hines D, et al. Combined APACHE IIscore and serum lactate dehydrogenase as predictors ofin-hospital mortality caused by first episode Pneumocystiscarinii pneumonia in patients with acquired immunodefi¬ciency syndrome. Am Rev Respir Dis 1991; 144:319-23

23 Brenner M, Ognibene F, Lack E, et al. Prognostic factors andlife expectancy of patients with the acquired immunodefi¬ciency syndrome and Pneumocystis carinii pneumonia. AmRev Respir Dis 1987; 136:1199-1206

24 El-Sadr W, Simberkoff M. Survival and prognostic factors insevere Pneumocystis carinii pneumonia requiring mechanicalventilation. Am Rev Respir Dis 1988; 137:1264-67

25 Garay S, Greene J. Prognostic indicators in the initial presen¬tation of Pneumocystis carinii pneumonia. Chest 1989; 95:769-72

26 Kales C, Murren J, Torres R, et al. Early predictors of

in-hospital mortality for Pneumocystis carinii pneumonia inthe acquired immunodeficiency syndrome. Arch Intern Med1987; 147:1413-17

27 Mason G, Hashimoto C, Dickman P, et al. Prognostic impli¬cations of bronchoalveolar lavage neutrophilia in patients withPneumocystis carinii pneumonia and AIDS. Am Rev RespirDis 1989; 139:1336-42

28 Sadaghdar H, Huang Z-B, Eden E. Correlation of bronchoal¬veolar lavage findings to severity of Pneumocystis cariniipneumonia in AIDS: evidence for the development of high-permeability pulmonary edema. Chest 1992; 102:63-69

29 Speich R, Weber R, Kronauer C, et al. Prognostic score forPneumocystis carinii pneumonia in HIV-infected patients.Respiration 1990; 57:259-63

30 Zaman M, White D. Serum lactate dehydrogenase levels andPneumocystis carinii pneumonia. Am Rev Respir Dis 1988;137:796-800

31 Speich R, Opravii M, Weber R, et ai. Prospective evaluationof a prognostic score for Pneumocystis carinii pneumonia inHIV-infected patients. Chest 1992; 102:1045-48

32 Vanhems P, Toma E. Evaluation of a prognostic score:

Pneumocystis carinii pneumonia in HIV-infected patients.Chest 1995; 107:107-12

33 Chu D. Predicting survival in AIDS patients with respiratoryfailure: application of the APACHE II scoring system. CritCare Clin 1993; 9:89-105

34 Smith R, Levine S, Lewis M. Prognosis of patients with AIDSrequiring intensive care. Chest 1989; 96:857-61

35 Knaus W, Draper E, Wagner D, et al. An evaluation ofoutcome from intensive care in major medical centers. AnnIntern Med 1986; 104:410-18

36 Jiva T, Hanna N, Dubaybo B. The use of APACHE II scores

and other parameters in predicting mortality in AIDS patientsduring successive admissions [abstract]. Clin Res 1991; 39:205A

37 Staikowsky F, Lafon B, Guidet B, et al. Mechanical ventila¬tion for Pneumocystis carinii pneumonia in patients with theacquired immunodeficiency syndrome: is the prognosis reallyimproved? Chest 1993; 104:756-62

38 Wachter R, Luce J, Safrin S, et al. Cost and outcome ofintensive care for patients with AIDS, Pneumocystis cariniipneumonia, and severe respiratory failure. JAMA 1995; 273:230-35

39 Hebert P, Drummond A, Singer J, et al. A simple multiplesystem organ failure scoring system predicts mortality ofpatients who have sepsis syndrome. Chest 1993; 104:230-35




DOI 10.1378/chest.114.1.199 1998;114; 199-206Chest

Lawson, James A. Russell and Julio S.G. MontanerDavid M. Forrest, Ognjenka Djurdjev, Carlos Zala, Joel Singer, Lindsay

Pneumonia and Respiratory FailurecariniiPneumocystisPredictor of Mortality in Patients With AIDS-Related

Validation of the Modified Multisystem Organ Failure Score as a

July 21, 2011This information is current as of

http://chestjournal.chestpubs.org/content/114/1/199Updated Information and services can be found at:

Updated Information & Services

http://chestjournal.chestpubs.org/content/114/1/199#related-urlsThis article has been cited by 3 HighWire-hosted articles:

Cited Bys

http://www.chestpubs.org/site/misc/reprints.xhtmlbe found online at: Information about reproducing this article in parts (figures, tables) or in its entirety canPermissions & Licensing

http://www.chestpubs.org/site/misc/reprints.xhtmlInformation about ordering reprints can be found online:

Reprints

"Services" link to the right of the online article.Receive free e-mail alerts when new articles cite this article. To sign up, select the

Citation Alerts

PowerPoint slide format. See any online figure for directions. articles can be downloaded for teaching purposes inCHESTFigures that appear in Images in PowerPoint format



http://chestjournal.chestpubs.org/content/114/1/199#related-urls

http://www.chestpubs.org/site/misc/reprints.xhtml

http://www.chestpubs.org/site/misc/reprints.xhtml


validation of the modified multisystem organ failure score as a predictor of mortality in patients...

Documents