http://lup.sagepub.com/Lupus

http://lup.sagepub.com/content/18/14/1252The online version of this article can be found at:

 DOI: 10.1177/0961203309345720

2009 18: 1252 originally published online 22 October 2009LupusDomínguez-Cherit

SA Ñamendys-Silva, JA Baltazar-Torres, E. Rivero-Sigarroa, JA Fonseca-Lazcano, L. Montiel-López and G.Prognostic factors in patients with systemic lupus erythematosus admitted to the intensive care unit

  

Published by:

http://www.sagepublications.com

can be found at:LupusAdditional services and information for    

  http://lup.sagepub.com/cgi/alertsEmail Alerts:

 

http://lup.sagepub.com/subscriptionsSubscriptions:  

http://www.sagepub.com/journalsReprints.navReprints:  

http://www.sagepub.com/journalsPermissions.navPermissions:  

http://lup.sagepub.com/content/18/14/1252.refs.htmlCitations:  

What is This? 

- Oct 22, 2009 OnlineFirst Version of Record 

- Nov 24, 2009Version of Record >>

at The University of Melbourne Libraries on September 15, 2014lup.sagepub.comDownloaded from at The University of Melbourne Libraries on September 15, 2014lup.sagepub.comDownloaded from

Lupus (2009) 18, 1252–1258

http://lup.sagepub.com

PAPER

Prognostic factors in patients with systemic lupus erythematosus

admitted to the intensive care unit

SA Namendys-Silva1, JA Baltazar-Torres1, E Rivero-Sigarroa1, JA Fonseca-Lazcano1, L Montiel-Lopez1 andG Domınguez-Cherit2

1Department of Critical Care Medicine, Instituto Nacional de Ciencias Medicas y Nutricion Salvador Zubiran, SSA. Mexico City, Mexico; and2Division of Critical Care Medicine. Instituto Nacional de Ciencias Medicas y Nutricion Salvador Zubiran, SSA. Mexico City, Mexico

The objectives of this study were to identify risk factors associated with mortality in patientswith systemic lupus erythematosus (SLE) admitted to the intensive care unit (ICU) and toevaluate the usefulness of Acute Physiologic and Chronic Health Evaluation (APACHE) IIscore to predict outcomes in these patients, through the use of a retrospective patient recordreview from amultidisciplinary intensive care unit in a teaching hospital. One hundred and fourpatients with SLE admitted to the ICU were included in the study. The mean age of patientswas 32.44 years, 96.2% were female and 61.5% were admitted with infection. The meanAPACHE II score was 19.7, 46.2% had acute renal dysfunction, 67.3% received inotropics/vasopressors, 27.9% pulmonary artery catheter and 74% invasive mechanical ventilation. Themean length of stay in ICU was 18.5 days and mortality rate was 32.7%. In the univariatelogistic regression analysis, factors associated with mortality were high APACHE II score, useof inotropics/vasopressors, pulmonary artery catheter and invasive mechanical ventilation.High APACHE II score and use of inotropics/vasopressors remained significant in the multi-variate analysis. The area under the receiver operating characteristic curve of the APACHE IIscore to predict mortality was 0.689 (95% CI 0.586–0.791 p¼ 0.002) and the Hosmer–Lemeshow �2 was 5.094 (p¼ 0.747). We conclude that the mortality rate in patients withSLE admitted to the ICU is high. The most common cause of admission was infection. Thefactors associated with mortality were high APACHE II score and the use of inotropics/vaso-pressors. APACHE II score was unable to accurately predict mortality. Lupus (2009) 18,1252–1258.

Key words: systemic lupus erythematosus; intensive care unit; prognosis; outcome; risk factors;APACHE II score

Introduction

Systemic lupus erythematosus (SLE) is a chronicautoimmune inflammatory disease that affects mul-tiple organs. An increase in the survival rate hasbeen observed in patients with SLE in recentyears; some studies report a 5-year survival ratehigher than 90%.1–3 A number of factors haveprobably contributed to this increase in survival,including recognition of milder forms of the dis-ease, improved diagnostic techniques, and earlier

therapeutic intervention. In addition, more effica-cious use of therapeutic modalities such as immu-nosuppressive agents, antibiotics, antihypertensivedrugs, hemodialysis, and transplantation may alsobe important.4 However, because of the increasingincidence of SLE across the world,2,5 this disease isresponsible for a considerable number of deaths.

Patients with SLE may develop organs dysfunc-tion associated with disease or immunosuppressivetreatment requiring admission to the intensive careunit (ICU). Patients with SLE admitted to the ICUstill have a high mortality rate.6,7 Few studiesreported information on the prognosis, factors asso-ciated with mortality and the usefulness of prognos-tic scales and the results are discordant.8–12

The purpose of this study is to identify the riskfactors associated with death in patients with SLEadmitted to a general ICU. We also evaluated the

Correspondence to: Silvio A Namendys-Silva, MD,MS,FCCP,

Department of Critical Care Medicine, Instituto Nacional de

Ciencias Medicas y Nutricion Salvador Zubiran, SSA. Vasco de

Quiroga No.15, Col Seccion XVI, Delegacion Tlalpan, C. P. 14000,

Mexico, D. F. Email: tony75ni@msn.com

Received 8 November 2008; accepted 2 June 2009

! The Author(s), 2009. Reprints and permissions: http://www.sagepub.co.uk/journalsPermissions.nav 10.1177/0961203309345720

at The University of Melbourne Libraries on September 15, 2014lup.sagepub.comDownloaded from

usefulness of Acute Physiologic and ChronicHealth Evaluation score (APACHE II)13 to predictoutcomes in these patients.

Material and methods

A retrospective cohort study was performed in theICU of the Instituto Nacional de Ciencias Medicasy Nutricion Salvador Zubiran, SSA, Mexico City,Mexico. We reviewed the clinical records of SLEpatients older than 16 years old, admitted to theICU between January 1994 and December 2004.Patients were diagnosed according to theAmerican College of Rheumatology criteria.14

Patients whose clinical records lacked the informa-tion required for the study were excluded.

Data collection

The following demographic and clinical variableswere recorded: age, sex, admission due to infection,APACHE II score13 using the worst score for acutephysiological variables in the first 24 hours, use ofinotropics and/or vasopressors, pulmonary arterycatheter, invasive mechanical ventilation, develop-ment of acute renal dysfunction or worseningof renal failure (increase of seric creatinine of0.5mg/dL above the baseline), need of hemodialysis,length of stay in the ICU and ICU mortality. TheSystemic Lupus Erythematosus Disease ActivityIndex (SLEDAI) score15 was used to evaluate theactivity of disease. SLEDAI score of three or morewas considered evidence of active disease.16,17

Statistical analysis

Values are expressed as mean � standard deviation(SD) for continuous variables or as percentage forcategorical variables. Continuous variables werecompared using the Student’s t-test and categoricalvariables were compared using the chi-squared test.Univariate and multivariate logistic regression testwere used to identify factors associated with ICUmortality. For categorical variables with multiplelevels, the reference level was attributed to theone with the lowest probability of the dependentvariable. Variables yielding p-values < 0.05 by uni-variate analysis and those considered clinically rel-evant were entered in the multivariate analysis toestimate the independent association of each cov-ariate with the dependent variable. Results weresummarized as odds ratios (OR) and respective95% confidence intervals (CI). Possible interactionswere tested. Two-tailed p-values < 0.05 were

considered statistically significant. Goodness of fit(Hosmer–Lemeshow)18 was calculated to assess therelevance of the logistic regression model. Age,APACHE II score, SLEDAI and length of stay inthe ICU were entered as continuous variables. Theusefulness of the APACHE II score to predict out-comes in these patients was evaluated by its discri-minative capacity (area under the receiver operativecharacteristics [ROC] curve19 and its 95% CI) andcalibration (Hosmer–Lemeshow goodness of fittest18. We considered p-values of < 0.05 as statisti-cally significant. For statistical analysis, SPSS 15.0for Windows (SPSS Inc., Chicago, IL, USA) wasused. The study was observational and descriptive,the Institutional Review Board approved the studyand the need for informed consent was waived.

Results

One hundred and twenty seven patients with SLEwere admitted to the ICU from January 1994 toDecember 2004. The records of twenty threepatients were incomplete and were not included,leaving 104 patients in the study. The mean ageof patients was 32.44 years, the majority wasfemale (96.2%,), the duration of disease was 70 �91.57 months (range from 1 to 360 months) and theorgan systems involved were: immunologic, 69.2%;positive test for antinuclear antibodies, 59.7%;hematologic, 65.4%; renal, 55.8%; articular,40.4%; mucocutaneous, 37.5%; serositis, 31.7%;and neurologic, 15.4%. Ninety three (89.4%)patients were receiving immunosuppressive therapybefore ICU admission: 92 (98.9%) corticosteroids(58 [55.77%] patients received at least 0.5mg/kg/day [0.83� 0.20 mg/kg/day] and 46 [44.23%]patients received less than 0.5 mg/kg/day[0.21� 0.05 mg/kg/day]), 17 (18.3%) azathioprineand 11 (11.8%) cyclophosphamide. Sixty four(61.5%) patients were admitted to ICU due toinfection (Table 1); 67.2% with pneumonia,23.4% with peritonitis, 17.2% with urinary tractinfection and 4.7% with central nervous systeminfection. Of these patients, 62 (96.87%) had signsof lupus activity (SLEDAI� 3).16,17 The meanSLEDAI score was 11.14� 3.95.

Seventy seven (74%) patients required mechani-cal ventilation during their stay in the ICU; themedian duration was 27.1� 26.29 days. Fortyeight patients (46.2%) had acute renal dysfunctionor worsening of renal failure during their ICU stay,26 of them (54.2%) required hemodialysis for11.19� 10.41 sessions (range of 1 to 42 sessions).

Prognostic factors in patients with systemic lupus erythematosus admitted to the intensive care unitSA Namendys-Silva et al.

1253

Lupus

at The University of Melbourne Libraries on September 15, 2014lup.sagepub.comDownloaded from

Of the patients with acute renal dysfunction, 46(95.83%) had active lupus nephritis, 4.16% (2) hadfailure renal due to other causes. The ICUmortalitywas 32.7% which statistically improved over time(Figure 1). Twenty three patients died of septic com-plications and 12 of complications related to SLE(pulmonary haemorrhage, active SLE).

Table 2 lists the results of the univariate analysiswith ICU death as the outcome variable of interest.Factors associated with mortality in the ICU were:high APACHE II score, use of inotropics and/orvasopressors, use of a pulmonary artery catheterand need of invasive mechanical ventilation.

In multivariate analysis, independent prognosticfactors of in-ICU death were high APACHE IIscore and use of inotropics and/or vasopressors(Table 3).

The mean APACHE II score was 22.6� 5.99 inpatients who died versus 18.2� 6.97 in survivors(p< 0.05). The performance of APACHE II scorewas poor to predict mortality. It showed low dis-criminative capacity, with area under the ROCcurve of 0.689 (95% CI 0.586–0.791), p¼ 0.002(Figure 2), but good calibration, withHosmer–Lemeshow �2 of 5.094 (p¼ 0.747).

Discussion

The most important findings of this retrospective,cohort study of patients with SLE admitted to a

general ICU are: (1) the ICU mortality hasdecreased over time, from 47.6% for the years1994–1999 to 24.2% for 2000–2004; (2) the mostcommon cause of ICU admission was infection;(3) the factors associated with mortality for thewhole group were high APACHE II score and theuse of inotropics and/or vasopressors; and (4)APACHE II score was unable to accurately predictmortality.

There are few published studies on the outcome ofcritically ill patients with SLE admitted to ICU.8–12

Studies have shown that patients with SLEadmitted to ICU have higher mortality as com-pared with patients admitted to ICU with other

Table 1 Demographic and clinical data of patients with systemic lupus erythematosusadmitted to the intensive care unit

N 104

Age (years) 32.44� 12.95

Gender (%)

Female 96.2

Male 3.8

Reasons for admission (%)

Infection 61.5

Acute renal failure 46.2

Shock 26.9

Pulmonary hemorrhage 18.3

Seizures 11.5

APACHE II score 19.7� 6.9

Inotropics/Vasopressors (%) 67.3

Pulmonary artery catheter (%) 27.9

Invasive mechanical ventilation (%) 74

Acute renal dysfunction (%) 46.2

Hemodialysis (%) 54.2

Length of stay in ICU (days) 18.5� 24.5

ICU mortality (%) 32.7

SLEDAI 11.14� 3.95

ICU¼ intensive care unit.

SLEDAI¼ Systemic Lupus Erythematosus Disease Activity Index.

APACHE¼Acute Physiologic and Chronic Health Evaluation.

%

50 47.6

1994–1999 2000–2004

Years

24.2

p=0.013454035302520151050

Figure 1 Mortality in patients with systemic lupus erythe-matosus admitted to the intensive care unit according todecade of admission.

Prognostic factors in patients with systemic lupus erythematosus admitted to the intensive care unitSA Namendys-Silva et al.

1254

Lupus

at The University of Melbourne Libraries on September 15, 2014lup.sagepub.comDownloaded from

diseases and comparable severity of disease. In aretrospective study of 30 SLE patients admittedto the ICUs of two hospitals reported in 1996,Ansell and coworkers8 found a mortality of 47%.In 2005, Hsu and coworkers11 reported a mortalityof 47% in 51 SLE patients admitted to ICU. In ourstudy, mortality was lower, 32.7%. However, inthis study, mortality from 1994 to 1999 was similarto the value reported previously, 47.6%, while mor-tality in the current decade was 24.2%. There hasbeen a non-significant decline in the mortality inpatients admitted to our ICU over time.20,21

These trends to improved overall survival of criti-cally ill SLE patients may be a result of recentadvances in diagnostic and therapeutic strategiesin critically ill patients, which have led to bettermanagement of specific disease in the ICU.Infectious complications are the most frequentcauses of ICU admissions in patients with SLE.22

In their study, Ansell et al.8 found infection as themain cause of ICU admission. In 1997, Godeauand coworkers9 reported a retrospective study of181 patients with rheumatic diseases admitted toICU, 37 of them with SLE, and found the infectionas the most frequent cause of ICU admission as

well; while pneumonia was the main cause of ICUadmissions in the Hsu et al’s study.11 In our study,infectious complications were the cause of ICUadmission in 61.5% of patients and pneumoniawas the most frequent. This type of complicationmay be related to disease or immunosuppressivetherapy and may unleash systemic inflammatoryresponse and organ dysfunction, requiring intensivecare.23 Previous studies have also suggested thatdisease activity may constitute a risk factor of infec-tion in SLE patients.24–27 In this study, only two ofthe 64 patients with infection did not have signs ofdisease activity (SLEDAI� 3).16,17

Few studies have evaluated the risk factors asso-ciated with death in patients with SLE admitted toICU and the studies reported contradictory results.Ansell et al.8 found renal dysfunction as the onlyfactor associated with mortality. Godeau et al.9

reported previous quality of life, steroids, infectionsand high Simplified Acute Physiology Score (SAPS)as risk factors for death in these patients. Hsuet al.11 found intracranial hemorrhage, gastrointes-tinal hemorrhage and septic shock as factorsinvolved in poor outcome. In our study, only highAPACHE II score and use of inotropics and/or

Table 2 Odds ratio and 95% confidence intervals from univariate logistic regression for risk factors for death in patients withsystemic lupus erythematosus admitted to the intensive care unit

Risk factors OR 95% CI p-value

Gender male 2.03 0.27–15.05 0.489

Age 0.99 0.96–1.03 0.955

Infection 1.30 0.56–3.05 0.533

APACHE II score 1.10 1.03–1.17 0.003

Inotropics/vasopressors 31.16 4.03–240.62 0.001

Pulmonary artery catheter 3.62 1.47–8.90 0.005

Invasive mechanical ventilation 20.55 2.65–159.26 0.004

Length of mechanical ventilation 1.02 0.99–1.05 0.073

Acute renal dysfunction 0.97 0.43–2.20 0.949

Hemodialysis 1.88 0.79–4.48 0.149

Length of stay in ICU 1.01 0.99–1.02 0.227

SLEDAI 1.02 0.92–1.13 0.675

OR¼odds ratio; 95% CI¼ 95% confidence interval; ICU¼ intensive care unit; SLEDAI¼ Systemic Lupus Erythematosus Disease Activity Index;

APACHE¼Acute Physiologic and Chronic Health Evaluation.

Table 3 Odds ratio and 95% confidence intervals from multivariate logistic regression for risk factors for death in patients withsystemic lupus erythematosus admitted to the intensive care unit (goodness-of-fit, Hosmer–Lemeshow, �2¼ 3.172, p¼ 0.923)

Risk factors OR 95% CI p-value

APACHE II score 1.09 1.006–1.18 0.034

Inotropics/vasopressors 17.55 2.05–149.31 0.009

Pulmonary artery catheter 1.38 0.48–3.93 0.544

Invasive mechanical ventilation 5.62 0.57–54.94 0.137

OR¼odds ratio; 95% CI¼ 95% confidence interval; APACHE¼Acute Physiologic and Chronic Health Evaluation.

Prognostic factors in patients with systemic lupus erythematosus admitted to the intensive care unitSA Namendys-Silva et al.

1255

Lupus

at The University of Melbourne Libraries on September 15, 2014lup.sagepub.comDownloaded from

vasopressors showed statistical significance as riskfactors associated with mortality in these patients.However, the relation between APACHE II scoreand mortality should be interpreted with cautionbecause CI (1.006–1.184) indicates that a statisti-cally significant difference was found but that themagnitude of this difference was so small that itwould not likely be clinically significant. The retro-spective nature of the studies probably has an influ-ence on these results. Furthermore, there is a lackof uniformity in the risk factors analyzed betweenthe studies and this may be another reason of dis-cordant results.

Predicting mortality in these patients is anotheraspect poorly studied and few studies have ana-lyzed different prognostic scores. APACHE IIscore has been used in some studies to predict mor-tality in critically ill SLE patients. Many SLEpatients have chronic organ damage, which canbe evaluated through chronic health evaluation ofAPACHE II score. In Ansell et al.’s study,8

APACHE II score was not useful in predicting out-come in SLE patients admitted to ICU. In theirstudy of patients with rheumatic diseases admittedto ICU, Godeau et al.9 reported SAPS as a riskfactor associated with death in these patients.However, they did not analyze SAPS’s discrimina-tive capacity to predict mortality. In another retro-spective study of patients with rheumatic diseases

admitted to ICU, Moreels et al.10 reported meanAPACHE II score of 17� 5 in survivors and22� 9 in patients who died. These data are compa-rable to our results, in which survivors had a meanAPACHE II score of 18.2� 6.97 and non-survivors22.6� 5.99. Although, APACHE II score had apoor performance to predict mortality in Moreelet al.’s study;10 however, that particular studyincluded only four SLE patients. Hsu et al.11

reported that APACHE II score was not useful inpredicting outcome in SLE patients admitted toICU, however, they did not assess the score’s dis-criminative capacity and calibration.

Although several measures exist for evaluatingthe performance of prognostic models, many stu-dies use ROC curves and the area under the curve19

to evaluate discrimination and the Hosmer–Lemeshow goodness-of-fit H- or C-statistics18 toevaluate the calibration of the prognostic models.Discrimination refers to a model’s ability to distin-guish survivors from non-survivors. The area underthe curve represents the probability that a patientwho died had a higher predicted probability ofdying than a patient who survived. An area underthe curve of 0.5 indicates that the model does notpredict better than chance. The discrimination of aprognostic model is considered perfect if the areaunder the curve is 1. The area under the curve of amodel gives no indication of how close the

Sen

sibi

lity

AUROCC=0.689

p=0.002

1.0

0.8

0.6

0.4

0.2

0.0

0.0 0.2 0.4

1-Specificity

0.6 0.8 1.0

Figure 2 Discriminative capacity for death of APACHE II score in patients with systemic lupus erythematosus admitted to theintensive care unit. AUROCC¼ area under receiver operating characteristic curve. APACHE¼Acute Physiologic and ChronicHealth Evaluation.

Prognostic factors in patients with systemic lupus erythematosus admitted to the intensive care unitSA Namendys-Silva et al.

1256

Lupus

at The University of Melbourne Libraries on September 15, 2014lup.sagepub.comDownloaded from

predicted probabilities are to the observed out-come. To take this aspect of a model’s performanceinto account, we have to look at the calibration ofthe prognostic models. Calibration refers to theagreement between predicted probabilities and thetrue probabilities. Of course, the true probability ofa patient’s outcome is not known; otherwise therewould be no need to develop prognostic models.However, the true probabilities can be approxi-mated by taking the mean of the observed out-comes within predefined groups of patients.The Hosmer–Lemeshow H- or C-statistics comparethe observed mortality in a group with the pre-dicted mortality of that group. Grouping for theH-statistic is based on partitioning of the probabil-ity interval (0–1) into 10 equally sized ranges. The Cgoodness-of-fit statistic sorts observations accord-ing to their expected probability and partitions theobservations into 10 groups of equal size. A high Hor C relates to a small p-value, implying significantdifference between observed and predicted mortal-ity, and thus indicates a lack of fit of the model.Discrimination and calibration describe the overallpredictive power of a model.

In our study, we analyzed discriminative capac-ity and calibration of APACHE II score to predictdeath and its performance showed low discrimina-tion, with area under ROC curve of 0.689, but goodcalibration, with Hosmer–Lemeshow H-statistic of5.094 and p¼ 0.747.

The SLEDAI did not have prognostic value inour SLE patients admitted to ICU. These resultsare similar to those reported by Alzeer et al.12

who observed that the SLEDAI score was foundnot to be a prognostic factor for survival. Thislack of association between SLEDAI score andmortality may be related to that it only assignslow scores (score of 1) to some of the well-knownentities associated with increased SLE mortality(thrombocytopia, leukopenia).15

There are some possible reasons why our resultsare in contrast to data reported previously. First, asmall number of patients in each of the previousstudies may have limited the identification of anassociation between variables. Furthermore, asmentioned above, there is a lack of uniformity inthe risk factors analyzed between the studies.Second, the statistical methodology used to evalu-ate the association between variables was different;only in our study was the OR estimated to establishsuch an association. Third, the retrospective natureof the studies as well as the difference in the timeswhen the studies were performed might also haveinfluenced the results. With such discrepancy, it isnecessary to create prospective multicentric

databases including standardized risk factors fordeath in this population with larger sample sizesin order to obtain better results and that allowsus to identify more clearly the risk factors fordeath in SLE patients admitted to the ICU.

Conclusions

ICU mortality rate in patients with SLE admittedto ICU is high, but has decreased over time, from47.6% for 1994–1999 to 24.2% for 2000–2004. Themost common cause of ICU admission was infec-tion. The factors associated with mortality werehigh APACHE II score and the use of inotropics/vasopressors. APACHE II score was unable toaccurately predict mortality.

Acknowledgement

We are indebted to Dr. Alejandro Arroliga for crit-ical readings of this manuscript.

References

1 Ruiz-Irastorza G, Khamashta MA, Castellino G, Hughes GR.Systemic lupus erythematosus. Lancet 2001; 357: 1027–1032.

2 Uramoto KM, Michet CJJ, Thumboo J, Sunku J, O0Fallon WM,Gabriel SE. Trends in the incidence and mortality of systemic lupuserythematosus 1950–1992. Arthritis Rheum 1999; 42: 46–50.

3 Jacobson DL, Gange SJ, Rose NR, Graham NMH. Epidemiologyand estimated population burden of selected autoimmune diseasesin the United States. Clin Immunol Immunopathol 1997; 84:223–243.

4 Bresnihan B. Outcome and survival in systemic lupus erythemato-sus. Ann Rheum Dis 1989; 48: 443–445.

5 Petri M. Epidemiology of systemic lupus erythematosus. Best PractRes Clin Rheumatol 2002; 16: 847–858.

6 Raj R, Murin S, Matthay RA, Wiedemann HP. Systemic lupuserythematosus in the intensive care unit. Crit Care Clin 2002; 18:781–803.

7 Zamora MR, Warner ML, Tuder R, Schwarz MI. Diffuse alveolarhemorrhage and systemic lupus erythematosus. Clinical presenta-tion, histology, survival, and outcome.Medicine 1997; 76: 192–202.

8 Ansell SM, Bedhesi S, Ruff B, et al. Study of critically ill patientswith systemic lupus erythematosus. Crit Care Med 1996; 24:981–984.

9 Godeau B, Mortier E, Roy P, et al. Short and longterm outcomesfor patients with systemic rheumatic diseases admitted to intensivecare units: a prognostic study of 181 patients. J eumatol 1997; 24:1317–1323.

10 Moreels M, Melot Ch, Leeman M. Prognosis of patients with sys-temic rheumatic diseases admitted to the intensive care unit.Intensive Care Med 2005; 31: 591–593.

11 Hsu CL, Chen KY, Yeh PS, et al. Outcome and prognostic factorsin critically ill patients with systemic lupus erythematosus: a retro-spective study. Crit Care 2005; 9: R177–R183.

12 Alzeer AH, Al-Arfaj A, Basha SJ, et al. Outcome of patients withsystemic lupus erythematosus in intensive care unit. Lupus 2004;13: 537–542.

Prognostic factors in patients with systemic lupus erythematosus admitted to the intensive care unitSA Namendys-Silva et al.

1257

Lupus

at The University of Melbourne Libraries on September 15, 2014lup.sagepub.comDownloaded from

13 Knaus WA, Draper EA, Wagner DP, Zimmerman JE. APACHEII. A severity of disease classification system. Crit Care Med 1985;13: 818–829.

14 Tan EM, Cohen AS, Fries JF, et al. The 1982 revised criteria forthe classification of systemic lupus erithematosus. Arthritis Rheum1982; 25: 1271–1277.

15 Bombardier C, Gladman DD, Urowitz MB, et al. Derivation of theSLEDAI. A disease activity index for lupus patients. TheCommittee on Prognosis Studies in SLE. Arthritis Rheum 1992;35: 630–640.

16 Guzman J, Cardiel MH, Arce-Salinas A, Sanchez-Guerrero J,Alarcon-Segovia D. Measurement of disease activity in systemiclupus erythematosus. Prospective validation of 3 clinical indices.J Rheumatol 1992; 19: 1551–1558.

17 Arce-Salinas A, Cardiel MH, Guzman J, Alcocer-Varela J. Validityof retrospective disease activity assessment in systemic lupuserythematosus. J Rheumatol 1996; 23: 846–849.

18 Hosmer DW, Taber S, Lemeshow S. The importance of assessingthe fit of logistic regression models: a case study. Am J PublicHealth 1991; 81: 1630–1635.

19 Hanley JA, McNeil BJ. The meaning and use of the area under areceiver operating characteristic (ROC) curve. Radiology 1982; 143:29–36.

20 Rosel-Ramırez R, Loria A, Domınguez-Cherit G,Gutierrez-Sougarret B. A predictive model of mortality in

the ICU of a Mexico City hospital. Rev Invest Clin 2004; 56:591–599.

21 Loria A, Rosas-Barush A, Posadas JG, Domınguez-Cherit G,Rivero-Sigarroa E. Four years of raw mortality in an intensivecare unit. Rev Invest Clin 2008; 60: 197–204.

22 Fessler BJ. Infectious diseases in systemic lupus erythematosus:risk factors, management and prophylaxis. Best Pract Res ClinRheumatol 2002; 16: 281–291.

23 Zonana Nacach A, Barr SG, Magder LS, Petri M. Damage insystemic lupus erythematosus and its association with corticoster-oids. Arthritis Rheum 2000; 43: 1801–1808.

24 Duffy KN, Duffy CM, Gladman DD. Infection and disease activ-ity in systemic lupus erythematosus: a review of hospitalizedpatients. J Rheumatol 1991; 18: 1180–1184.

25 Petri M, Genovese. Incidence of and risk factors for hospi-talization in systemic lupus erythematosus. A prospectivestudy of the Hopkins Lupus Cohort. J Rheumatol 1992; 19:1559–1565.

26 Paton NI, Cheong IK, Kong NC, Segasothy M. Risk factors forinfection in Malaysian patients with systemic lupus erythematosus.Q J Med 1996; 89: 531–538.

27 Zonana-Nacach A, Camargo-Coronel A, Yanez P, Sanchez L,Jimenez- Balderas FJ, Fraga A. Infections in outpatients with sys-temic lupus erythematosus: a prospective study. Lupus 2001; 10:505–510.

Prognostic factors in patients with systemic lupus erythematosus admitted to the intensive care unitSA Namendys-Silva et al.

1258

Lupus

at The University of Melbourne Libraries on September 15, 2014lup.sagepub.comDownloaded from