Misclassification of Pulmonary Hypertension Due to ... · 159 Hosmer-Lemeshow goodness-of-fit test.15 When conducting the goodness-of-fit test, LVEDP 160 was dichotomized as ≤ 15mmHg

DOI 10.1378/chest.08-2784; Prepublished online March 2, 2009;Chest

Scott D. Halpern and Darren B. Taichman End-Diastolic PressurePressure Rather Than Left-Ventricularto Reliance on Pulmonary Capillary Wedge Misclassification of Pulmonary Hypertension Due

http://www.chestjournal.org/content/early/2009/02/20/chest.08-2784

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

) ISSN:0012-3692http://www.chestjournal.org/site/misc/reprints.xhtml(without the prior written permission of the copyright holder.

distributedrights reserved. No part of this article or PDF may be reproduced or College of Chest Physicians, 3300 Dundee Road, Northbrook IL 60062. Allhas been published monthly since 1935. Copyright 2007 by the American CHEST is the official journal of the American College of Chest Physicians. It

initial publication. online articles must include the digital object identifier (DOIs) and date ofare indexed by PubMed from initial publication. Citations to Advance Advance online articles are citable and establish publication priority; theyversions may be posted when available prior to final publication).

typesetpublication but have not yet appeared in the paper journal (edited, Advance online articles have been peer reviewed and accepted for

Copyright Copyright © American College of Chest Physicians on May 25, 2009www.chestjournal.orgDownloaded from


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

http://www.chestjournal.org/

1

WORD COUNT: 2,858 1

ABSTRACT WORD COUNT: 250 2

3

4

5

6

MISCLASSIFICATION OF PULMONARY HYPERTENSION DUE TO RELIANCE ON PULMONARY 7

CAPILLARY WEDGE PRESSURE RATHER THAN LEFT-VENTRICULAR END-DIASTOLIC 8

PRESSURE 9

10

Scott D. Halpern, M.D., Ph.D.1,2,3

11

Darren B. Taichman, M.D., Ph.D.1 12

1Division of Pulmonary, Allergy and Critical Care Medicine;

2Center for Clinical Epidemiology 13

and Biostatistics; and 3Leonard Davis Institute of Health Economics; 14

University of Pennsylvania School of Medicine 15

16

17

18

19

ADDRESS CORRESPONDENCE AND REQUESTS FOR REPRINTS TO: 20

Scott D. Halpern 21

Center for Clinical Epidemiology and Biostatistics 22

724 Blockley Hall 23

423 Guardian Drive 24

Philadelphia, PA 19104-6021 25

Phone: (215) 898-1462 26

Fax: (215) 573-5325 27

Email: [email protected] 28

29

SUPPORT and DISCLOSURE: This work was supported by an American Thoracic Society 30

Fellows Career Development Award to Dr. Halpern. Drs. Halpern and Taichman have each 31

received support from Actelion Pharmaceuticals to conduct other research related to pulmonary 32

hypertension. The authors have no other involvement with organizations with a financial interest 33

in the subject matter.34

Page 1 of 36


mailto:[email protected]


2

Abstract 35

36

Background: Pulmonary arterial hypertension is typically distinguished from pulmonary venous 37

hypertension by documenting a pulmonary capillary wedge pressure (PCWP) ≤ 15mmHg. 38

However, PCWP has uncertain utility in establishing pulmonary venous hypertension. We 39

sought to determine the calibration, discrimination, and diagnostic accuracy of PCWP, using 40

simultaneously measured left-ventricular end-diastolic pressure (LVEDP) as the gold standard. 41

42

Methods: We examined hemodynamic data from the 11,523 unique patients undergoing 43

simultaneous right- and left-heart catheterization at a large academic center from 1998 – 2007. 44

45

Results: Among 4,320 patients (37.5%) with pulmonary hypertension (mean pulmonary artery 46

pressure ≥ 25mmHg), hemodynamic data were complete for 3,926 (90.9%). Of these, 580 47

(14.8%) met criteria for pulmonary arterial hypertension with a PCWP ≤ 15mmHg, but 310 48

(53.5%) of these patients had an LVEDP > 15mmHg. Such discrepancies remained common 49

among patients with a pulmonary vascular resistance > 3 Wood units and those being 50

catheterized specifically to evaluate pulmonary hypertension. PCWP provided moderate 51

discrimination between patients with high vs. normal LVEDP (area under the receiver-operating 52

characteristic curve = 0.84, 95% confidence interval = 0.81 – 0.86) but was poorly calibrated to 53

LVEDP (Bland-Altman limits of agreement: – 15.2mmHg to 9.5mmHg; Hosmer-Lemeshow 54

goodness-of-fit χ2

statistic: 155.4, p < 0.0001). 55

56

Page 2 of 36



3

Conclusions: Roughly half of patients presumed to have pulmonary arterial hypertension based 57

on PCWP may be found to have pulmonary venous hypertension based on LVEDP. Reliance on 58

PCWP may result in the dangerous or cost-ineffective use of pulmonary vasodilators for patients 59

with left-heart disease. Furthermore, without assessing LVEDP, investigators may include 60

patients with left-heart disease in therapeutic trials of PAH drugs, thereby limiting their ability to 61

detect beneficial drug effects. 62

Page 3 of 36



4

KEY WORDS: pulmonary hypertension, left-heart disease, pulmonary capillary wedge pressure, 63

hemodynamic assessment, cardiac catheterization 64

Page 4 of 36



5

ABBREVIATION LIST 65

66

AUROC, area under the receiver-operating characteristic curve 67

LVEDP, left-ventricular end-diastolic pressure 68

mPAP, mean pulmonary artery pressure 69

PAH, pulmonary arterial hypertension 70

PCWP, pulmonary capillary wedge pressure 71

PVH, pulmonary venous hypertension 72

PVR, pulmonary vascular resistance 73

TPG, transpulmonary gradient 74

WHO, World Health Organization 75

Page 5 of 36



6

Introduction 76

In approaching a patient with pulmonary hypertension, it is crucial to distinguish between 77

pulmonary arterial hypertension (PAH) and other causes of elevated pulmonary pressures, 78

including pulmonary venous hypertension (PVH) due to left-sided heart disease. The World 79

Health Organization (WHO) emphasizes the importance of such a differentiation in its 80

classification system that separates PAH (Group 1) from other forms of pulmonary hypertension 81

(e.g., Group 2 patients with left heart dysfunction).1-3

Patients with PAH may benefit from 82

recently approved prostacyclin analogues, endothelin receptor-antagonists, or phosphodiesterase 83

inhibitors.4 By contrast, in patients with PVH these same therapies are not indicated, may be 84

harmful, and initial management is best focused on amelioration of left-heart dysfunction.5, 6

85

86

Differentiation of PAH from PVH is most commonly accomplished by documenting a 87

pulmonary capillary wedge pressure (PCWP) of ≤ 15mmHg at the time of diagnostic right-heart 88

catheterization.7, 8

This diagnostic approach is predicated on the assumption that a normal 89

PCWP measurement adequately excludes left atrial hypertension. Indeed, rather than having 90

intrinsic value, the utility of the PCWP resides primarily in its ability to rule in or out disease 91

states characterized by an elevated left-ventricular end-diastolic pressure (LVEDP). 92

93

Although the assumption that PCWP is a useful surrogate marker for LVEDP has both strong 94

historical roots and substantial face validity,9 there is scant evidence regarding the abilty of 95

PCWP to establish the presence or absence of left-sided heart disease among patients with 96

pulmonary hypertension. Thus, when both PCWP and LVEDP are available in a patient with 97

pulmonary hypertension, the LVEDP is generally considered to be the gold standard. 98

Page 6 of 36



7

99

In a preliminary report of a study involving 131 patients with pulmonary hypertension, Soto and 100

colleagues found that PCWP has poor operating characteristics when tested against the standard 101

of LVEDP.10

Given the potential importance of such findings to the management of pulmonary 102

hypertension patients, we sought to determine the calibration, discrimination, and accuracy of 103

mean PCWP compared with the gold standard of LVEDP among a large cohort of patients with 104

pulmonary hypertension. 105

106

Methods 107

Patients 108

All patients undergoing right-heart catheterization at Penn-Presbyterian Medical Center – a large, 109

community-based, academic hospital and regional referral center for pulmonary vascular disease 110

affiliated with the University of Pennsylvania Health System – from January 1, 1998 – 111

December 31, 2007 were included. This study was deemed exempt from review by the 112

University of Pennsylvania Institutional Review Board because it used previously collected, de-113

identified data. 114

115

Patients were considered ineligible for the study if they had a diagnosis of mitral stenosis 116

(identified by an International Classification of Diseases – 9 code between 394.0 and 396.8 on 117

the catheterization record) or if tachycardia (>130 beats per minute) was present during 118

catheterization because these phenomena are known to cause discrepancies between PCWP and 119

LVEDP.12

Among the 2,763 patients who underwent multiple catheterizations during the study 120

period, only the first catheterization was included. 121

Page 7 of 36



8

122

Eligible patients were grouped according to whether they had a combined right- and left-heart 123

catheterization (the “combined catheterizations” group) or a right-heart catheterization alone. 124

Because patients in whom physicians order combined catheterizations may differ from those in 125

whom only right-heart catheterization is ordered, hemodynamic measurements were compared 126

between these groups to determine whether selection bias may have influenced the results. 127

128

In both groups, patients were considered to have pulmonary hypertension (PH) if their mean 129

pulmonary artery pressure (mPAP) (calculated as 2/3 pulmonary artery diastolic pressure + 1/3 130

pulmonary artery systolic pressure) was ≥ 25mmHg at rest.1 Patients were excluded if data were 131

missing for mPAP, PCWP, or LVEDP (among patients undergoing combined catheterization) 132

(Figure 1). 133

134

Hemodynamic Measurements 135

Catheterizations were performed by 10 interventional cardiologists, all of whom were board-136

certified and members of the University of Pennsylvania faculty. Hemodynamic parameters were 137

recorded directly into electronic spreadsheets and stored in a computerized database. 138

139

Physicians performing the catheterizations followed standard protocols for measuring 140

hemodynamic values. Hemodynamic values from both right- and left-heart catheterizations were 141

obtained prior to the injection of contrast for left ventriculography or coronary angiography. For 142

PCWP, values for the A-wave pressure, V-wave pressure, and mean pressure were recorded at 143

end-expiration. The mean PCWP was used for analyses. Among patients who underwent left-144

Page 8 of 36



9

heart catheterization, LVEDP was recorded simultaneously with PCWP using a pigtail catheter 145

placed in the left ventricle. 146

147

Pulmonary vascular resistance (PVR) was calculated as (mPAP – PCWP) / cardiac output 148

(measured using the estimated Fick method), and patients were classified as having elevated 149

PVR if the value was > 3 Wood units.1 Transpulmonary gradient (TPG) was calculated as mPAP 150

– PCWP, and patients were classified as having elevated TPG if the value was ≥ 12.13

151

152

Statistical Analysis 153

The accuracy of a mean PCWP ≤ 15mmHg vs. > 15mmHg in distinguishing between WHO 154

Groups 1 and 2 PH (i.e. PAH versus PVH) was assessed by calculating the proportion of patients 155

that would be reclassified by instead using LVEDP of ≤ 15mmHg vs. > 15mmHg. 156

157

The calibration of PCWP to LVEDP was assessed using a Bland-Altman analysis14

and the 158

Hosmer-Lemeshow goodness-of-fit test.15

When conducting the goodness-of-fit test, LVEDP 159

was dichotomized as ≤ 15mmHg vs. > 15mmHg; sensitivity analyses were performed using 160

LVEDP cut-points from 10 to 20mmHg. 161

162

The area under the receiver-operating characteristic curve (AUROC)16

was calculated to 163

determine the ability of PCWP to discriminate patients with LVEDP ≤ 15mmHg vs. > 15mmHg. 164

Wilcoxon rank-sum tests were used to compare hemodynamic values between patients who 165

underwent combined catheterizations vs. right-heart catheterization alone. Stata 9.2 (Stata Corp., 166

College Station, Texas) was used for all analyses. 167

Page 9 of 36



10

168

Results 169

There were 12,744 eligible unique patients who underwent right-heart catheterization at our 170

institution from 1998 – 2007. Of these, 11,523 had combined catheterizations, and 4,320 (37.5%) 171

of these patients had PH (Figure 1). 172

173

Disease classification 174

Among 3,926 patients (90.9%) with PH and complete data, 580 (14.8%) met criteria for PAH 175

based on a low PCWP (≤ 15mmHg). However, 310 (53.5%) of these patients would be 176

classified as having PVH if LVEDP were used instead (Table – Panel A and Figure 2). By 177

contrast, among the 3,346 patients classified as having PVH using PCWP, only 152 (4.5%) 178

would meet criteria for PAH if LVEDP were used instead. 179

180

To determine rates of misclassification among patients who might be considered to have 181

“pulmonary hypertension out of proportion to left-heart disease,5, 6

” we restricted our analyses to 182

those patients with either a PVR > 3 Wood units (1,116 patients) or a TPG > 12 (1,300 patients). 183

Among patients with an elevated PVR, 361 (32.4%) would be classified as PAH using PCWP, 184

but 148 of these (41.0%) would be reclassified as PVH based upon the LVEDP (Table – Panel 185

B). Among patients with an elevated TPG, 494 (38.0%) would be classified as PAH using 186

PCWP, but 247 of these (50.0%) would be reclassified as PVH based upon the LVEDP (Table – 187

Panel C). 188

189

Page 10 of 36



11

Compared with patients undergoing right-heart catheterization alone, the patients who underwent 190

combined catheterizations had a lower PVR (median = 2.1 Wood units, interquartile range 1.4 – 191

3.3 vs. median = 3.2 Wood units, interquartile range 1.9 – 5.9; p < 0.0001) and TPG (median = 192

9.8, interquartile range 6.7 – 14.0 vs. median = 13.3, interquartile range 8.0 – 24.3; p < 0.0001). 193

However, the two groups had similar PCWP (median = 22.0mmHg, interquartile range 14.0 – 194

30.0 vs. median = 22.0mmHg, interquartile range 18.0 – 27.0; p = 0.31). 195

196

Disease classification among patients catheterized specifically for evaluation of PH 197

To more specifically address the utility of left-heart catheterization among patients being 198

evaluated for PH, we restricted analyses to the 604 patients who were referred for catheterization 199

by PH specialists as part of their initial evaluation of PH. Of these, 340 (56.3%) had a combined 200

catheterization, and 282 (83.9%) of these patients had PH. Of the 265 patients with documented 201

PH, who had been referred for combined catheterization as part of their PH evaluation, and for 202

whom LVEDP was measured, 164 (61.9%) met criteria for PAH by virtue of having a PCWP ≤ 203

15mmHg, but 34 of these patients (20.7%) had an LVEDP > 15mmHg. 204

205

Calibration 206

In the complete sample of patients with PH and combined catheterizations, Bland-Altman 207

analysis revealed that on average, PCWP underestimated LVEDP by 2.9 mmHg (95% CI = 2.7 – 208

3.0) (Figure 3). In 39.0% of patients, the absolute difference between PCWP and LVEDP was > 209

5mmHg; in 11.3% it was > 10mmHg. The 95% limits of agreement were -15.2 mmHg to 210

9.5mmHg, indicating that even after excluding the 5% of patients with the most discrepant 211

Page 11 of 36



12

values between PCWP and LVEDP, the PCWP underestimated LVEDP by as much as 15.2 212

mmHg and overestimated LVEDP by as much as 9.5 mmHg. 213

214

Using LVEPD ≤ 15mmHg vs. > 15mmHg as a dichotomous outcome in a logistic regression 215

model, the calibration of PCWP was poor, as indicated by a Hosmer-Lemeshow χ2statistic of 216

155.4 (p < 0.0001). The goodness-of-fit test remained significant (indicating poor calibration) 217

for all cutpoints of LVEDP between 10mmHg and 20mmHg. 218

219

Because the large sample size could account for the statistical significance of the goodness-of-fit 220

test, we performed 1000 iterations of bootstrap resampling with 20% random samples of the total 221

(785 patients each). The goodness-of-fit test remained significant in 72.4% of these samples, 222

confirming the poor calibration. 223

224

Discrimination 225

The AUROC was 0.84 (95% CI = 0.81 – 0.86) (Figure 4). This indicates that among all 226

randomly selected pairs of patients in which one has an LVEDP ≤ 15mmHg and the other has an 227

LVEDP > 15mmHg, the patient with the higher LVEDP would have a higher PCWP in 84% of 228

cases. These results were similar using LVEDP cut-points of 10mmHg or 20mmHg (Figure 4). 229

230

Comparison with patients without pulmonary hypertension 231

Among 7,117 patients who underwent combined catheterizations and did not have pulmonary 232

hypertension, complete data were available in 6,551 (92.0%) patients. Misclassification was also 233

evident among these patients, as 2,253 of 5,454 patients with PCWP ≤ 15mmHg (41.3%) had 234

LVEDP > 15mmHg. A Bland-Altman analysis of calibration in this group revealed that PCWP 235

Page 12 of 36



13

underestimated LVEDP by 4.7 mmHg (95% CI = 4.6 – 4.8), with 95% limits of agreement from 236

-14.5mmHg to 5.1mmHg. Finally, the ability of PCWP to discriminate patients with high or low 237

LVEDP among patients without pulmonary hypertension, as assessed by the AUROC, was 80% 238

(95% CI = 79% – 81%). 239

240

241

Discussion 242

This study of a large number of patients undergoing sequential measurement of PCWP and 243

LVEDP suggests that PCWP frequently underestimates LVEDP, that it is poorly calibrated to 244

LVEDP, and that it has a moderate ability to discriminate between patients with normal or 245

elevated LVEDP. Perhaps most importantly, these results suggest that approximately half of all 246

patients who meet hemodynamic criteria for PAH on the basis of PCWP measurements may, in 247

fact, have elevated left-ventricular filling pressures. 248

249

This degree of misclassification was robust even when we restricted the sample to patients with 250

an elevated PVR or TPG, groups hypothesized to be more homogenous and reflective of true 251

PAH patients.5 These results emphasize the importance of avoiding the conclusion that a patient 252

has “pulmonary hypertension out of proportion to left heart disease” without evaluating the 253

LVEDP. 254

255

Although many of the patients in our study underwent cardiac catheterization for reasons other 256

than evaluation of PH, disease misclassification remained common even among patients 257

specifically referred for catheterization by PH specialists as part of their PH evaluation. Among 258

Page 13 of 36



14

such selected patients, one fifth of those who would be classified as having PAH by PCWP 259

would instead be classified as having PVH by LVEDP. 260

261

Bias resulting from the selective referral of certain patients for combined catheterization is 262

unlikely to have influenced these results. First, discrepancies between PCWP and LVEDP 263

persisted even among patients with elevations in PVR or TPG. Second, the median PCWP did 264

not differ between patients undergoing combined catheterization versus those undergoing right-265

heart catheterization. 266

267

The clinical consequences of mistakenly classifying patients as having PAH when left-heart 268

disease is present are incompletely understood. However, the potential for PAH-specific 269

therapies such as pulmonary vasodilators to precipitate the acute deterioration of patients with 270

PVH is well described.5, 6, 17

Even if frank deterioration occurs infrequently following use of 271

PAH therapies for patients with PVH, there are no high-quality data to suggest that patients with 272

PVH would benefit from these therapies. It is thus critical to make the correct diagnosis prior to 273

instituting therapies that are inappropriate, potentially harmful, and tremendously expensive. 274

275

In addition to these clinical considerations, disease misclassification due to reliance on PCWP 276

may influence the results of clinical trials. For example, the modest mean treatment effects 277

noted in most randomized trials of approved treatments for PAH may be attributable, in part, to 278

the enrollment of heterogeneous patient populations. If only some enrolled patients are afflicted 279

with diseases likely to respond to these therapies, summary treatment effect estimates would be 280

Page 14 of 36



15

biased toward the null and would not reflect the treatment benefits that true PAH patients might 281

achieve. 282

283

The implications of this study depend, in part, on the mechanisms that account for the poor 284

correspondence between PCWP and LVEDP in patients with pulmonary hypertension. One 285

possibility is that the observed measurement errors are attributable to fundamental alterations of 286

the pulmonary vascular bed among patients with pulmonary hypertension that make it difficult to 287

obtain an accurate PCWP.5 However, this explanation seems unlikely because the poor 288

calibration and moderate discrimination of PCWP were similarly evident among patients without 289

pulmonary hypertension. 290

291

Second, it is possible that PCWP systematically underestimates LVEDP in all patients. This 292

conclusion is supported by the consistent underestimation noted in our study among patients with 293

and without pulmonary hypertension, as well as by smaller studies showing that PCWP 294

underestimates LVEDP in the contexts of acute myocardial infarction18

and generalized critical 295

illness.19

However, the width of the limits of agreement in the Bland-Altman analysis and the 296

consistently poor fit of the regression slope between PCWP and LVEDP suggest that systematic 297

bias is not the only problem. Thus, clinicians cannot overcome this problem simply by adding a 298

set value to the PCWP to better estimate LVEDP or by using a different PCWP cutpoint. 299

300

Rather, the observed measurement variability suggests that PCWP is genuinely unreliable in 301

estimating left-ventricular filling pressure, that physicians err in measuring PCWP or LVEDP, or 302

that both of these explanations are true. These hypotheses have been offered previously in 303

Page 15 of 36



16

attempts to explain the consistently negative or null effects of right-heart catheterization to guide 304

therapy in many critically ill populations,20-25

including patients with left-ventricular disease.24

305

306

The present study is limited by our inability to directly review the hemodynamic tracings from 307

the catheterizations because they were not routinely stored during the study period. Thus, we 308

cannot exclude the possibility that although PCWP was recorded as a mean pressure, LVEDP 309

may have been recorded following the A wave in some patients. This could cause PCWP to 310

underestimate LVEDP. Other measurement errors, however, are unlikely to explain our results. 311

Contrast injection for ventriculography or coronary angiography might artificially elevate the 312

LVEDP, but LVEDP was measured before contrast injection in this study. Additionally, 313

although physicians did not routinely confirm proper wedge position by measuring pulmonary 314

venous saturation with the balloon inflated,5 difficulties obtaining a proper wedge position in 315

patients with pulmonary hypertension should cause PCWP to overestimate LVEDP, whereas we 316

found the opposite. Furthermore, because these “wedge saturations” are not routinely performed 317

in most settings, our results may reflect current practice more generally. 318

319

A second limitation of this study is that the use of deidentified data precluded assessment of 320

whether discrepancies between PCWP and LVEDP were particularly common when the 321

catheterizations were performed by specific physicians. However, the validity and 322

generalizability of our results are supported by the similar findings of Soto and colleagues10

at a 323

different institution. 324

325

Page 16 of 36



17

Third, we were unable to evaluate whether specific subgroups of patients were particularly likely 326

to have discrepant PCWP and LVEDP values. Patients with left ventricular diastolic dysfunction 327

(e.g., older patients with long-standing systemic hypertension), may be particularly likely to have 328

PVH despite a low PCWP measurement.17

Because the de-identified nature of our data 329

precluded confirmation of this hypothesis, future studies are needed to determine whether certain 330

patient characteristics can be used to help clinicians determine when discrepancies between 331

PCWP and LVEDP are likely to be present. 332

333

Conclusions 334

Some might conclude from our results that LVEDP should be measured routinely among all 335

patients referred for catheterization as part of an evaluation for pulmonary hypertension. 336

However, this approach carries increased risks and inconveniences for patients as well as 337

increased costs and resource utilization. We therefore suggest a more conservative approach in 338

routine practice in which clinicians obtain left-heart hemodynamic measurements whenever there 339

are reasons to suspect left-heart disease based on the patient’s history or physical exam, 340

whenever the diagnosis is uncertain following right-heart catheterization, and when patients do 341

not show favorable responses to initial therapy. If future studies identify types of patients who 342

are particularly likely to have discrepancies between PCWP and LVEDP, then combined 343

catheterization may represent a prudent initial diagnostic approach in such patients. 344

345

Ultimately, a randomized trial may be needed to determine whether treatment guided by 346

combined catheterizations leads to improved patient-centered outcomes such as quality of life, 347

symptom control, or mortality; such evidence would provide the strongest possible justification 348

Page 17 of 36



18

for routinely measuring LVEDP. Indeed, such an approach may prove to be cost-effective or 349

even cost-saving if it helps prevent the needless and potentially dangerous prescription of 350

expensive PAH therapies. 351

Page 18 of 36



19

References 352

1. Galie N, Torbicki A, Barst R, et al. Guidelines on diagnosis and treatment of pulmonary 353

arterial hypertension. The Task Force on Diagnosis and Treatment of Pulmonary Arterial 354

Hypertension of the European Society of Cardiology. European Heart Journal 2004;25:2243-78. 355

2. Rubin LJ, American College of Chest P. Diagnosis and management of pulmonary 356

arterial hypertension: ACCP evidence-based clinical practice guidelines. Chest 2004;126:7S-357

10S. 358

3. Taichman DB, Mandel J. Epidemiology of pulmonary arterial hypertension. Clinics in 359

Chest Medicine 2007;28:1-22. 360

4. Badesch DB, Abman SH, Simonneau G, Rubin LJ, McLaughlin VV. Medical therapy for 361

pulmonary arterial hypertension: updated ACCP evidence-based clinical practice guidelines. 362

Chest 2007;131:1917-28. 363

5. Benza RL, Tallaj JA. Pulmonary hypertension out of proportion to left heart disease. 364

Advances in Pulmonary Hypertension 2005;5:21-9. 365

6. Oudiz RJ. Pulmonary hypertension associated with left-sided heart disease. Clinics in 366

Chest Medicine 2007;28:233-41. 367

7. Barst RJ, McGoon M, Torbicki A, et al. Diagnosis and differential assessment of 368

pulmonary arterial hypertension. Journal of the American College of Cardiology 2004;43:40S-369

7S. 370

8. McLaughlin VV, McGoon MD. Pulmonary arterial hypertension. Circulation 371

2006;114:1417-31. 372

9. Fishman AP. A century of pulmonary hemodynamics. American Journal of Respiratory 373

& Critical Care Medicine 2004;170:109-13. 374

10. Soto FJ, Siegel R, Marks D, et al. Performance of pulmonary capillary wedge pressure 375

(PCWP) vs. left ventricular end diastolic pressure (LVEDP) in the diagnosis/classification of 376

patients with suspect pulmonary arterial hypertension. Chest 2005;128:137S [abstract]. 377

11. Halpern SD, Taichman DB. Misclassification of pulmonary arterial hypertension due to 378

use of pulmonary capillary wedge pressure (PCWP) Rather than left-ventricular end-diastolic 379

pressure (LVEDP) [abstract]. American Journal of Respiratory & Critical Care Medicine 380

2008;177:A259. 381

Page 19 of 36



20

12. Pulmonary Artery Catheter Education Project (PACEP). (Accessed June 8, 2008, at 382

www.pacep.org.) 383

13. Klotz S, Wenzelburger F, Stypmann J, et al. Reversible pulmonary hypertension in heart 384

transplant candidates: To transplant or not to transplant. The Annals of Thoracic Surgery 385

2006;82:1770-3. 386

14. Bland JM, Altman DG. Statistical methods for assessing agreement between two methods 387

of clinical measurement. Lancet 1986;i:307-10. 388

15. Hosmer DW, Lemeshow S. Model building strategies and methods for logistic 389

regression. In: Hosmer DW, Lemeshow S, eds. Applied Logistic Regression, 2nd Edition. New 390

York: John Wiley & Sons; 2000. 391

16. Hanley JA, McNeil BJ. The meaning and use of the area under a receiver operating 392

characteristic (ROC) curve. Radiology 1982;143:29-36. 393

17. Soto FJ. Pulmonary venous hypertension: A diagnostic and therapeutic dilemma. 394

Advances in Pulmonary Hypertension 2007;6:168-75. 395

18. Rahimtoola SH, Sinno MZ, Rosen KM, et al. Relationship of pulmonary-artery to left 396

ventricular diastolic pressures in acute myocardial infarction. Circulation 1972;46:283-90. 397

19. Calvin JE, Driedger AA, Sibbald WJ. Does the pulmonary capillary wedge pressure 398

predict left ventricular preload in critically ill patients? Crit Care Med 1981;9:437-43. 399

20. Harvey S, Harrison D, Singer M, et al. Assessment of the clinical effectiveness of 400

pulmonary artery catheters in management of patients in intensive care (PAC-Man): a 401

randomised controlled trial. Lancet 2005;366:472-7. 402

21. Richard C, Warszawski J, Anguel N, et al. Early Use of the Pulmonary Artery Catheter 403

and Outcomes in Patients With Shock and Acute Respiratory Distress Syndrome: A Randomized 404

Controlled Trial. JAMA 2003;290:2713-20. 405

22. Sandham JD, Hull RD, Brant RF, et al. A Randomized, Controlled Trial of the Use of 406

Pulmonary-Artery Catheters in High-Risk Surgical Patients. N Engl J Med 2003;348:5-14. 407

23. Shah MR, Hasselblad V, Stevenson LW, et al. Impact of the pulmonary artery catheter in 408

critically ill patients: Meta-analysis of randomized clinical trials. JAMA 2005;294:1664-70. 409

24. The ESCAPE Investigators and ESCAPE Study Coordinators. Evaluation study of 410

congestive heart failure and pulmonary artery catheterization effectiveness: The ESCAPE trial. 411

JAMA 2005;294:1625-33. 412

Page 20 of 36


http://www.pacep.org./


21

25. The National Heart Lung and Blood Institute Acute Respiratory Distress Syndrome 413

(ARDS) Clinical Trials Network. Pulmonary-Artery versus Central Venous Catheter to Guide 414

Treatment of Acute Lung Injury. N Engl J Med 2006;354:2213-24. 415

26. Rubenfeld GD, McNamara-Aslin E, Rubinson L. The pulmonary artery catheter, 1967-416

2007. Rest in peace? JAMA 2007;298:458-61. 417

418

Page 21 of 36



22

Legend to Figure 1 419

420 RHC, right-heart catheterization; LHC, left-heart catheterization; mPAP, mean pulmonary artery 421

pressure; PCWP, pulmonary capillary wedge pressure; LVEDP, left ventricular end-diastolic pressure. 422

Page 22 of 36



23

Legend to Table 423

Percentages reflect proportions within rows. PAH, pulmonary arterial hypertension; PVH, pulmonary 424

venous hypertension; PCWP, pulmonary capillary wedge pressure; LVEDP, left ventricular end-425

diastolic pressure; TPG, transpulmonary gradient. 426

Page 23 of 36



24


PCWP, pulmonary capillary wedge pressure; LVEDP, left ventricular end-diastolic pressure. 428

Page 24 of 36



25


*Difference represents PCWP – LVEDP, Average represents (PCWP + LVEDP)/2. Larger circles 430

represent identical observations among multiple patients. Mean bias = -2.9 mmHg (95% CI = -3.0 – -2.7); 431

Limits of agreement = -15.2 – 9.5 mmHg. PCWP, pulmonary capillary wedge pressure; LVEDP, left 432

ventricular end-diastolic pressure. 433

Page 25 of 36



26


Area under receiver-operating characteristic curve (AUROC) = 0.84 (95% CI = 0.81 – 0.86) 435

using a cutpoint of LVEDP of ≤15 mmHg to indicate PAH. If a cutpoint of LVEDP ≤ 10 mmHg 436

were used, the AUROC would be 0.86 (95% CI = 0.82 – 0.91). If a cutpoint of LVEDP ≤ 20 437

mmHg were used, the AUROC would be 0.81 (95% CI = 0.80 – 0.83). Sens, sensitivity for the 438

outcome of LVEDP > 15 mmHg; Spec, specificity for the outcome of LVEDP > 15 mmHg; 439

PCWP, pulmonary capillary wedge pressure; LVEDP, left ventricular end-diastolic pressure. 440

Page 26 of 36



27

Figure 1: Flow diagram

RHC performed: 12,823 Patients

Simultaneous LHC and RHC: 11,523

patients

RHC only: 1,211 patients

mPAP ≥ 25

mmHg: 4,320

mPAP < 25

mmHg: 7,117 mPAP ≥ 25

mmHg: 873

mPAP < 25

mmHg: 338

mPAP

missing: 86

LVEDP

missing: 346

PCWP

missing: 48

PCWP

missing: 6

3,926 evaluable

patients with

pulmonary

hypertension

867 evaluable

patients with

pulmonary

hypertension

PCWP ≤ 15:

580

PCWP > 15:

3,346

PCWP ≤ 15:

240

PCWP > 15:

627

Eligible for study: 12,744 Patients

Excluded due to:

mitral stenosis: 69

tachycardia: 25

Page 27 of 36



28

Table: Classification of PAH using PCWP or LVEDP

A. All patients with pulmonary hypertension

PAH by

LVEDP ≤ 15

PVH by LVEDP

> 15 Total

PAH by PCWP

≤ 15

270 (46.5%) 310 (53.5%)

580

PVH by PCWP

> 15

152 (4.5%) 3,194 (95.5%) 3,346

Total 422 3,504 3,926

B. Patients with pulmonary hypertension and PVR > 3

PAH by

LVEDP ≤ 15

PVH by LVEDP

> 15 Total

PAH by PCWP

≤ 15

213 (59.0%) 148 (41.0%)

361

PVH by PCWP

> 15

65 (8.6%) 690 (91.4) 755

Total 278 842 1,116

Page 28 of 36



29

C. Patients with pulmonary hypertension and TPG ≥ 12

PAH by

LVEDP ≤ 15

PVH by LVEDP

> 15 Total

PAH by PCWP

≤ 15

247 (50.0%) 247 (50.0%)

494

PVH by PCWP

> 15

61 (7.8%) 743 (92.2) 806

Total 310 990 1,300

Page 29 of 36



30

Figure 2: Scatter plot of PCWP and LVEDP among 3,926 patients with pulmonary

hypertension

0

510

15

20

25

30

35

40

45

50

LV

ED

P

0 5 10 15 20 25 30 35 40 45 50PCWP

Page 30 of 36



31

Figure 3: Bland-Altman plot of PCWP and LVEDP among 3,926 patients with pulmonary

hypertension

Diffe

ren

ce

Average0 10 20 30 40 50

-40

-30

-20

-10

0

10

20

30

40

Page 31 of 36



32

Figure 4: Receiver operating-characteristic curve of PCWP against LVEDP among 3,926

patients with pulmonary hypertension

0.0

00.2

50.5

00.7

51.0

0S

en

sitiv

ity

0.00 0.25 0.50 0.75 1.001 - Specificity

Area under ROC curve = 0.8254

PCWP = 15mmHg

Sens = 94.2%,

Spec = 60.2%

Page 32 of 36



215x282mm (600 x 600 DPI)

Page 33 of 36



215x283mm (600 x 600 DPI)

Page 34 of 36



215x283mm (600 x 600 DPI)

Page 35 of 36



215x283mm (600 x 600 DPI)

Page 36 of 36



DOI 10.1378/chest.08-2784; Prepublished online March 2, 2009;Chest

Scott D. Halpern and Darren B. TaichmanPressure

Capillary Wedge Pressure Rather Than Left-Ventricular End-Diastolic Misclassification of Pulmonary Hypertension Due to Reliance on Pulmonary

May 25, 2009This information is current as of

& ServicesUpdated Information

784http://www.chestjournal.org/content/early/2009/02/20/chest.08-2figures, can be found at:Updated Information and services, including high-resolution

Open Access Freely available online through CHEST open access option

Permissions & Licensing

http://www.chestjournal.org/site/misc/reprints.xhtmltables) or in its entirety can be found online at: Information about reproducing this article in parts (figures,

Reprints http://www.chestjournal.org/site/misc/reprints.xhtml

Information about ordering reprints can be found online:

Email alerting serviceup in the box at the top right corner of the online article.Receive free email alerts when new articles cit this article. sign

formatImages in PowerPoint

article figure for directions.teaching purposes in PowerPoint slide format. See any online Figures that appear in CHEST articles can be downloaded for

articles must include the digital object identifier (DOIs) and date of initial publication. priority; they are indexed by PubMed from initial publication. Citations to Advance online prior to final publication). Advance online articles are citable and establish publicationyet appeared in the paper journal (edited, typeset versions may be posted when available Advance online articles have been peer reviewed and accepted for publication but have not






Misclassification of Pulmonary Hypertension Due to ... · 159 Hosmer-Lemeshow goodness-of-fit test.15 When conducting the goodness-of-fit test, LVEDP 160 was dichotomized as ≤ 15mmHg

Documents