Institution where work was performed: University of California, Los Angeles.
Journal of Critical Care (2012) 27, 739.e7739.e13 There is no financial support or conflicts of interest to disclose. Corresponding author. Tel.: +1 310 825 5316; fax: +1 310 206 8622.Intensive care unit;Pulmonary arterycatheterization
Purpose: Patients with pulmonary hypertension (PH) can decompensate to the point where they requirecare in the intensive care unit (ICU). Our objective is to examine the outcomes and characteristics ofpatients with PH admitted to the ICU.Methods: This is a retrospective study of 99 patients with PH who were admitted to the medical ICU ofa single tertiary care center. Baseline characteristics, interventions during ICU admission, and ICU and6-month outcome were documented. Univariate and multivariate logistic regressions were used toevaluate association of patient characteristics with mortality.Results: Intensive care unit mortality was 30%, and 6-month mortality was 40%. Acute Physiology andChronic Health Evaluation II score, World Health Organization Group 3 PH, and preexisting treatmentwith a prostacyclin at time of ICU admission were associated with worse outcome. Patients whoreceived cardiopulmonary resuscitation had 100% mortality. The requirement for mechanical ventilationand dialysis was also associated with increased mortality. Pulmonary artery catheter placement wasassociated with reduced mortality, specifically if it was placed early during ICU admission and ifassociated with a change in the present management.Conclusions:Mortality is high in critically ill patients with PH. The identification of prognostic baselinecharacteristics and interventions in the ICU is important and warrants further investigation. 2012 Elsevier Inc. All rights reserved.0hDepartment of BiostatistiDepartment of Psychiatryos Angeles, CA 90095-16Department of CardiologPrognostic factors and outcomes of patients withpulmonary hypertension admitted to the intensivecare unit,
Thanh N. Huynh MDa,, S. Sam Weigt MDa, Catherine A. Sugar PhDb,c,Shelley Shapiro MD, PhDd, Eric C. Kleerup MDa
aDivision of Pulmonary and Critical Care Medicine, Department of Medicine, David Geffen School of Medicine, UCLA,BOX 951690, 37-131 CHS, Los Angeles, CA 90095-1690, USAb cs, UCLA School of Public Health, Los Angeles, CA 90095-1772, USAc and Biobehaviorial Sciences, David Geffen School of Medicine, UCLA, BOX 951690, 37-131 CHS,L 90, USAd y, UCLA-VA Greater LA Healthcare Systems, Los Angeles, CA 90095-1691, USAE-mail address: firstname.lastname@example.org (T.N. Huynh).
883-9441/$ see front matter 2012 Elsevier Inc. All rights reserved.ttp://dx.doi.org/10.1016/j.jcrc.2012.08.006
hemodialysis. Few studies describe the prognostic factors,
739.e8 T.N. Huynh et al.course, and outcomes of these patients or the ICU man-agement [10,11], and none evaluate the efficacy of thevarious interventions initiated in the critical care setting.Evaluation of the impact of these interventions on mortalityis important, especially because they are often costly andinvasive. In an effort to address these questions, we retro-spectively assessed the clinical characteristics and mortalityoutcome of a group of patients with PH who requiredadmission to the ICU.
2. Patients and methods
2.1. Study design
This is a retrospective study at a single tertiary care centerto identify the characteristics and the ICU interventionsassociated with in-hospital and 6-month mortality afteradmission to the ICU. Our study was approved by the UCLAInstitutional Review Board (IRB no. 11-001695).
The first admission of all patients with PH who requiredadmission to the UCLA medical ICU between July 2004 andJune 2009 was identified via International Classification ofDiseases, Ninth Revision, codes (416.0 and 416.8). Afterexclusion of patients with PH from left heart failure (DanaPoint Group 2), 99 patients were included in the study. Morethan 900 patients with PH were seen in the outpatient settingat UCLA during the same 5-year study period.1. Introduction
Pulmonary hypertension (PH) is a progressive diseasecharacterized by sustained elevations in pulmonary arterialpressures and increased pulmonary vascular resistance [1,2].This results in right heart failure (RHF); impairment inoxygenation; exercise limitation; and, ultimately, death.Therapeutic improvements have targeted reduction inpulmonary vascular resistance in an effort to unload theright ventricle and improve cardiac output . Three classesof medications, phospodiesterase inhibitors, endothelin re-ceptor antagonists, and prostacyclin analogs increase exer-cise capacity and stabilize disease progression [3-6]. Despitethese advances, patients with PH continue to require admis-sion to the hospital and often to the intensive care unit (ICU),most commonly for decompensated RHF [7,8].
The management of a critically ill PH patient is chal-lenging. With very little physiologic reserve, they maydeteriorate rapidly into irreversible hemodynamic collapse. Treatment often involves high-dose intravenous di-uretics; vasopressor support; escalation or initiation of newPH therapy; and, possibly, mechanical ventilation and2.3. Data
Data were abstracted from electronic medical records. Age,sex, and type of PH (Dana Point Group classification) weredocumented. To estimate the severity of PH, we recorded theduration or PH diagnosis before ICU admission, the rightventricular systolic pressure estimated on their most recentechocardiogram, and prior treatment of PH. These medica-tions were endothelin receptor antagonists, phosphodiesteraseinhibitors, and prostacyclin analogues (at this center, the localpractice is to start patients on epoprostenol while they are inthe ICU and transition to treprostinol once pumps areavailable). Because PH therapies may take 3 months foronset and stabilization , the duration of PH diagnosis wasdichotomized at this time point. Biologic markers within thefirst 24 hours of ICU admission, such as brain natriureticpeptide (BNP), troponin, partial pressure of CO2, and liverfunction tests (alanine aminotransferase/aspartate aminotrans-ferase), were recorded but were not performed consistently.The indication for admission was classified into 4 groups:RHF, respiratory failure from causes other than RHF, sepsis,and other (which included arrhythmia, syncope, postoperativemonitoring, and gastrointestinal bleed). Severity of illness atthe time of admission, assessed by the Acute Physiology andChronic Health Evaluation (APACHE) II score was obtainedwithin the first 24 hours.
The interventions and therapy used during each patient'sICU stay were documented including initiation of a new PHmedication, vasopressor use, mechanical ventilation, hemo-dialysis, cardiopulmonary resuscitation (CPR), and the place-ment of a pulmonary artery catheter (PAC). The treatmentsadministered following PAC placement included diuresisand/or initiation of a PH medication.
2.4. Statistical analysis
Using 6-month mortality as the main outcome variable,patient baseline characteristics and ICU interventions werefirst individually evaluated by univariate logistic regression.Analyses were repeated for ICU mortality. Kaplan-Meiersurvival curves were generated for 6-month survival byAPACHE II quartiles. Next, block stepwise logistic regres-sion was used to model the joint relationship between thepredictors and 6-month mortality (P b .05). In the first block,significant baseline characteristics were identified. Wedecided a priori that the significant baseline characteristicswill then be used in a stepwise model to evaluate ICUinterventions. This allowed us to determine whether theinterventions provided additional information about mortal-ity beyond the baseline characteristics with which thepatients presented. To evaluate only patients with DanaPoint Group 1 PH, this analysis was then repeated afterexcluding patients with group 3 PH.
A secondary analysis was performed on the subset ofpatients who received a PAC within the first 3 days of
admission. Logistic models using 6-month and ICU mor-
physiologic measurements included in the APACHE II scoreshowed that the association with mortality is primarily driven
increased mortality (P = .01, data not shown). Admissionsfor respiratory failure and sepsis were associated with higherodds of 6-month mortality (P = .01 and 0.02, respectively)
Table 1 Baseline characteristics of patients with PH requiring
Type of PH (n)Dana Point Group 1 (PAH) 72Idiopathic PH 30Associated with connective tissue disease 20Drug/toxin induced 5Portopulmonary 10Associated with HIV 3Associated with congenital heart disease 4
Dana Point Group 3 (due to lung disease/hypoxia)
Idiopathic pulmonary fibrosis 10Obesity hypoventilation syndrome/obstructive sleep apnea
Bronchiolitis obliterans after lungtransplant
Chronic obstructive pulmonary disease 2Sarcoidosis 2Other (pulmonary alveolar proteinosis,recurrent pneumonia, radiation-inducedinterstitial lung disease,postpneumonectomy syndrome,bronchiectasis)
Time (mo) since PH diagnosed, median(IQR), n = 91
Treatment before admissionOn PH medications at admission (n) 60On prostacyclin at admission (n) 31
Reason for ICU admission (n)RHF 51Respiratory failure from causes otherthan RHF
Sepsis 8Other 17Postoperative observation 5Arrhythmia 4Syncope 2Miscellaneous (gastrointestinal bleeding,epistaxis, acute renal failure, subduralhematoma)
APACHE II, mean (SD) 17.2 (7.1)Sodium level on admission, mean (SD) 134.5 (5.1)Mean arterial pressure on admission,mean (SD)
Creatinine, mean (SD) 1.5 (1.7)BNP (pg/mL) on admission, n = 80Median (IQR) 384 (197-839)No. of patients with BNP N1000 19
Troponin (ng/mL), median (IQR), n = 76 0.14 (0.23-0.51)Most recent right ventricular systolicpressure (mm Hg) on echocardiogram,median (IQR), n = 94
739.e9Patients with PH in teh ICUby points for oxygenation (PaO2) and renal function (crea-tinine and dialysis); however, the aggregate score of theremaining 10 variables was still significantly associated withtality as the outcome variables were created to evaluatewhich treatments driven by PAC data were associated withimproved survival after adjustment for APACHE.
3.1. Baseline characteristics
After excluding elective and repeated admissions, 99patients were included in the study (Table 1). Most werefemale (64/99) and had pulmonary arterial hypertension(Dana Point Group 1) (72/99). The median length of stay inthe ICU was 10 days (interquartile range [IQR], 5-16 days).The mean APACHE II score was 17.2 (SD, 7.1). The primaryreasons for ICU admission were RHF (52%) and respiratoryfailure from causes other than RHF (23%). There were 4patients who were admitted with arrhythmias. The first patientwas admitted with a-flutter after self-discontinuing iloprost.He was cardioverted unsuccessfully. He subsequently under-went ablation and pacemaker placement but then had apulseless electrical activity cardiac arrest the next day anddied. The second patient was admitted with a-flutter withvariable block and syncope. She was diuresed, started on aprostacyclin, and converted to sinus rhythm before dis-charge. The third patient was admitted with a-flutter.Therapy with amiodarone, digoxin, sotolol, and ablationwere all unsuccessful, and she was discharged withintermittent a-fibrillation and a-flutter. The fourth patientdeveloped junctional bradycardia with hypotension during aright heart catheterization. She required atropine and adopamine drip, which then led to conversion to normal sinusrhythm. The majority of group 1 patients were admitted forRHF (57%), whereas the most group 3 patients wereadmitted for respiratory failure unrelated to RHF (63%).
3.2. Mortality associated with baselinecharacteristics
Of the 99 PH subjects, 30 died in the ICU on that ad-mission, and an additional 10 patients died within 6 months.Unadjusted odds ratios (OR) for 6-month and ICU mortalityare displayed in Table 2. Acute Physiology and ChronicHealth Evaluation II scores were positively associated withodds of death by 6 months (OR, 1.19 per point; 95% con-fidence interval [CI], 1.10-1.29; P b .0001). Six-month mor-tality significantly increases with each increase in APACHEII quartile, consistent with the logistic regression analyses(Fig. 1). Multivariable logistic regression analysis of the 12ICU admission
Sex (female/male) 64/35Age, y, mean (SD) 51.9 (13.6)
compared with RHF. Dana Point Group 3 patients trendedtoward being associated with increased 6-month and ICU
Table 2 Bivariate baseline characteristics association with mortality
Variable 6-mo mortality ICU mortality
OR (95% CI) P OR (95% CI) P
APACHE II (per point) 1.19 (1.10-1.29) b.0001 1.21 (1.11-1.31) b.0001Reason for admission a
Respiratory failure not due to RHF 3.73 (1.33-10.5) .01 5.09 (1.71-15.1) .003Sepsis 7.20 (1.30-39.8) .02 14.0 (2.42-80.9) .003
Female sex 0.34 (0.14-0.80) .013 0.33 (0.14-0.81) .015BNP N1000 (n = 80) 2.81 (0.98-8.10) .054 2.44 (1.01-5.87) .046Dana Point Group 3 vs Group 1 2.35 (0.95-5.7Age (per year) 1.03 (0.99-1.0On prior prostacyclin 1.96 (0.83-4.6Most recent right ventricular systolic pressureon echocardiogram (per point)
Time since PH diagnosed 3 months 1.55 (0.67-3.6a Reference group, patients admitted for RHF.
739.e10 T.N. Huynh et al.mortality (P = .06).Stepwise multiple regression on the baseline characteris-
tics showed APACHE score, Dana Point Group 3 PH, andpreexisting treatment with a prostacyclin were significantlyassociated with increased odds of 6-month mortality(Table 3). Other baseline characteristics were not significantonce adjusting for the other variables in this model. BNP wasnot included in multivariable analysis because the number ofmissing values made the resulting models unstable. AcutePhysiology and Chronic Health Evaluation score andpreexisting treatment with a prostacyclin were also signif-icantly associated with increased odds of 6-month mortalitywhen the analysis was performed for only patients in DanaPoint Group 1.Fig. 1 Mortality by APACHE II quartiles.3.3. Interventions in the ICU
Of the 99 patients in the study group, 20 patients requireddialysis; 44, vasopressors; and 34, mechanical ventilation.Dopamine was the initial vasopressor in 20 patients,levophed was the initial vasopressor in 12 patients,dobutamine was the initial vasopressor in 8 patients, andthe rest of the patients were started on a combination ofvasopressors that were selected by the clinician. Seventypercent of the patients who were dialyzed died in the ICU(median no. of days on dialysis, 9.5). Fifty percent of thepatients who received vasopressors died in the ICU (medianno. of days on vasopressors, 9). Seventy-one percent of thepatients who were mechanically ventilated died in the ICU(median days on mechanical ventilation, 8). Nitric oxide(NO) was used in 35 patients. A new PH medication wasinitiated in 61 patients,...