Patients with chronic pulmonary hypertension (PHTN) are an extraordinary perioperative challenge. Although it is generally accepted that the presence of PHTN increases the risk of perioperative complications and mortality, outcome data in the noncardiac surgery population are rare.1–3 Existing studies are based on small sample sizes, they used single institutional data, and they included patients undergoing a wide variety of procedures thus limiting their external validity.1–3 No studies have evaluated the incidence and risk of perioperative mortality in the setting of total joint replacement, one of the most frequently performed procedures nationwide.4 This particular patient population may be especially at risk for perioperative complications given that joint arthroplasty procedures are associated with intraprocedural pulmonary embolization (PE) of bone marrow, cement, and bone debris.5 In addition, relatively high rates of perioperative deep vein thrombosis (DVT) and subsequent PE can exacerbate or worsen preexisting right heart strain.5,6
The purpose of this study was to assess the risk of mortality for patients with PHTN undergoing total knee arthroplasty (TKA) and total hip arthroplasty (THA) using the National Inpatient Sample (NIS) administrative database. We hypothesized that patients with PHTN undergoing major joint replacement surgery are at increased risk for perioperative mortality compared with individuals without the disease.
NIS data files sponsored by the Agency for Healthcare Research and Quality were commercially obtained from the Hospital Cost and Utilization Project (HCUP) and analyzed for this study. The NIS is the largest all payer inpatient discharge database in the United States (US). In brief, the NIS contains information on inpatient discharges from approximately 8 million hospital admissions per year. Having grown since its inception in 1988 when it included data from 8 states in the US, the most recent data files used in this study are a 20% stratified sample of approximately 1000 hospitals in 38 states. The NIS provides weights that allow for nationally representative estimates. It includes >100 clinical and nonclinical data elements, such as diagnoses, procedures, admission and discharge status, patient demographics (e.g., gender, age, race), payment source, length of hospitalization, and hospital characteristics (e.g., size, location, teaching status).
Detailed information on the NIS design can be accessed electronically.a,b Because the data used in this study are sufficiently de-identified, this project was exempt from review by the IRB.
Selection of Study Sample
Our study sample consists of all data in NIS between 1998 and 2006. Discharges with a procedure code (International Classification of Diseases, ninth revision, Clinical Modification [ICD-9-CM]) for primary THA and TKA (81.51 and 81.54, respectively) were identified and included in the sample. Entries with the diagnosis of chronic PHTN were determined by the listing of an ICD-9-CM diagnosis code for chronic pulmonary heart disease (416) and included in the analysis [i.e., 416.0 Primary pulmonary hypertension (idiopathic pulmonary arteriosclerosis; pulmonary hypertension (essential) (idiopathic) (primary)), 416.8 Other chronic pulmonary heart diseases (Pulmonary hypertension, secondary), 416.9 Chronic pulmonary heart disease, unspecified (chronic cardiopulmonary disease, cor pulmonale (chronic) not otherwise specified)] as published previously.7 For the purpose of this study, we defined entries with the ICD-9-CM code 416.0 as primary PHTN and those with 416.8 and 416.9 as nonprimary PHTN.7 PE, DVT, and respiratory insufficiency after trauma or surgery/adult respiratory distress syndrome (ARDS) were assessed by using respective ICD-9-CM codes. No relevant changes in the coding as it pertains to our analysis occurred during the period of study.
To facilitate analysis of the HCUP-NIS database that uses a complex survey design and makes available sample weights, statistical analyses were performed using SURVEY procedures in SAS version 9.2 (SAS Institute, Cary, NC). In particular, SAS procedures SURVEYMEANS, SURVEYFREQ, and SURVEYLOGISTIC were used to obtain weighted means, percentages, variance, confidence interval (CI), and for fitting logistic regression taking into account the sample weights. Details about these specific procedures and incorporation of sample weights into the computation are available in the SAS online manual.c
We first describe the patient demographics and health care system–related data items stratified by patients with and without PHTN and by procedure type (THA and TKA) (Table 1). Patient demographics included age (continuous as well as categorized as 0–44, 45–64, 65–74, and >75 years), overall comorbidity burden (according to Deyo comorbidity index categories),8,9 gender, and race (Caucasian, African American, Hispanic, other). Health care system data included hospital characteristics such as size (small, medium, large), location (rural, urban), teaching status (nonteaching, teaching), and admission type (emergent, elective, urgent, and others). The Deyo comorbidity index predicts mortality and postoperative complications for a patient with a range of comorbid conditions. Each condition is assigned a score depending on the risk of dying. These scores are totaled to produce a total score.9 Absolute standardized percent difference (ASPD), defined as the difference in proportion in units of the pooled standard deviation, was computed for assessing imbalance in the covariates between patients with and patients without PHTN.10 An absolute difference >10% implies significant covariate imbalance. A matched sample was created using the “optimal” matching approach using all of the above variables.11 Given the relatively low incidence of PHTN (0.3% in TKA; 0.4% in THA) and the large number of THA and TKA patients without PHTN, we decided to match 1 patient with PHTN (case) with 3 non-PHTN (controls) patients within each distinct combination of matching variables. We decided for 1:3 matching because it has been shown that the improvement in statistical efficiency for using multiple controls is not considerable beyond 3 or 4 controls per case.12 A zero caliper width was used for matching, meaning a PHTN patient (case) with specific characteristics (age, gender, race, admission type, hospital size, hospital location, hospital teaching status, and Deyo comorbidity index) was exactly matched for each characteristic when the 3 non-PHTN patient (controls) were identified. Cases were not included in the matched sample if 3 controls with exactly matching characteristics could not be found. Patient demographics and health care system– related data items stratified by patients with and without PHTN and by procedure type (THA and TKA) were then described for the matched sample (Table 2). Reduction in ASPD from the full sample to the matched sample was examined to assess the effectiveness of the matching procedure.
In-hospital mortality was considered the primary outcome and 3 adverse events (ARDS, PE, and DVT) were identified as the secondary outcomes based on their relevance to the orthopedic patient population. The incidence of each outcome for the full and matched sample was described using bar graphs (Figs. 1 and 2). Based on the matched sample, the adjusted effect of PHTN on perioperative mortality as well as ARDS, PE, and DVT were obtained by Cochran-Mantel-Haenszel (CMH) test. Corresponding overall common odds ratio (OR) and 95% CIs are presented.13,14
Additionally, a logistic regression model on the full dataset with adjustment for the same variables used for matching procedure was fitted to check for the sensitivity of the results to different methods and to comparatively analyze the impact of individual variables.15 The area under the receiver operating characteristic curve (also referred to as the C statistic or concordance index) was used for assessing the model's discriminatory power. A large number of entries (approximately 40%) in the race category were not available and were imputed as “white.” This step was based on an approach previously described and the fact that facilities with high rates of missing data for race served populations with higher than average Caucasian/African American patient ratios.16,17 The result from this analysis did not change significantly when treating missing entries as a separate group.
We identified 670,516 entries for TKA and 360,119 for THA in the NIS. Because the NIS collects data from a weighted sample of approximately 20% of hospitalizations, these frequencies are a national estimate of 3,261,288 and 1,752,021 performed procedures, respectively, in the US between 1998 and 2006. Of the identified patients, 2184 TKA patients (weighted estimate 6651 or 0.3%) and 1359 THA patients (weighted estimate 10,619 or 0.4%) had the diagnosis of PHTN. This relates to an average of 1180 TKA (range, 507; 2073) and 739 THA (range, 467; 1054) discharges with PHTN annually in the US during the study period. The prevalence of patients with primary PHTN among all patients with PHTN undergoing TKA and THA was 17.8% and 19.9%, respectively. Table 1 details the patient and health care system–related demographics of the data entries with and without PHTN for all TKA and THA discharges for the full sample. Patients with PHTN tended to be older, were more frequently female, and they had a higher comorbidity burden compared with the group without PHTN.
Table 2 details information after the PHTN patients were matched to non-PHTN patients with the same demographics; 94.2% TKA patients (2057 of 2184) and 91.6% THA patients (1245 of 1359) with PHTN were successfully matched. Because of matching, the absolute standardized percentage difference for the covariates decreased from 15.5% (range, 0.1%–55.6%) to 0.8% (range, 0.1%–2.4%) for the THA sample and from 9.0% (range, 0.0%–37.0%) to 0.6% (range, 0.1%–1.6%) for the TKA sample, respectively.
Patients with PHTN had unfavorable outcomes compared with patients without the disease for both TKA and THA procedures. Length of hospital stay (in days) was longer for patients with PHTN after THA (6.5 vs 4.9, P = 0.0001; difference CI of 0.9, 2.1 days) and after TKA by 1.1 days (5.3 vs 4.2, P = 0.0001; difference CI of 0.9, 1.3 days). Mortality rates among patients with PHTN were increased by a factor of 3.72 (95% CI, 2.13–6.39) (incidence of 2.41% [95% CI, 1.54%–3.29%] compared with controls 0.65% [95% CI, 0.39%–0.91%]) for the cohort undergoing THA. Mortality was increased by a factor of 4.55 (95% CI, 2.16–9.39) for those with PHTN compared with those without the disease (incidence of 0.91% [CI, 0.49–1.33] vs 0.2% [CI, 0.09–0.32], respectively) for TKA recipients (Fig. 1). The CMH test– based OR for the outcome of perioperative death in patients with PHTN after THA was 3.79 (95% CI, 2.96–4.86) and 4.55 (95% CI, 3.28–6.33) after TKA compared with matched controls. The ORs from the regression analysis performed to check sensitivity were similar: 3.36 (95% CI, 2.27–4.97) for THA and 5.05 (95% CI, 3.14–8.13) for TKA. A fatal outcome was more frequent in patients with primary compared to patients with nonprimary PHTN for either procedure (5.2% vs 1.8% after THA and 2.3% vs 0.6% after TKA, P < 0.001).
The incidence of ARDS, PE, and DVT was higher in patients with a diagnosis of PHTN undergoing THA and TKA (Fig. 2). The CMH test–based ORs for the matched sample were 4.86 (95% CI, 3.60–6.57), 4.63 (95% CI, 3.32–6.45), and 2.35 (95% CI, 1.78–3.10) for ARDS, PE, and DVT after THA, respectively. Similarly, the CMH test– based ORs were 4.78 (95% CI, 3.77–6.07), 4.56 (95% CI, 3.78–5.51), and 2.35 (95% CI, 1.92–2.88) for ARDS, PE, and DVT for TKA patients, respectively.
Independent risk factors for in-hospital mortality identified in the regression analysis of the full sample included the presence of PHTN, increasing age, male gender, and emergent admission. African American race was a risk factor for perioperative death for patients undergoing THA but not TKA. Hospital location and teaching status did not affect the ORs of in-hospital mortality; however, hospital size influenced the mortality rate (P < 0.0001; Table 3). In addition, the Deyo comorbidity index was associated with increased ORs of in-hospital death (for TKA: OR 1.14 [95% CI, 1.06–1.22]; for THA: OR 1.27 [95% CI, 1.20–1.34]).
In this study, we show that patients with PHTN undergoing primary THA or TKA are at increased risk of perioperative mortality and complications. The highest rate of a fatal outcome was found among patients with a diagnosis of primary PHTN undergoing THA.
There is a paucity of studies that have examined perioperative mortality in patients with PHTN undergoing noncardiac surgery. Ramakrishna et al.1 retrospectively identified 146 patients diagnosed with PHTN who had various noncardiac surgical interventions. The authors reported a mortality rate of 7%, but no comparison was made of whether perioperative mortality differed between patients with primary versus nonprimary PHTN. Predictors of short-term morbidity after noncardiac surgery in their study included New York Heart Association functional class II or higher, history of PE, and anesthesia lasting longer than 3 hours. When risk was stratified by type of surgery, the authors found that 17% of patients undergoing low-risk surgical procedures experienced morbidity compared with 48% of those undergoing orthopedic surgery (i.e., intermediate risk), suggesting that patients with PHTN undergoing orthopedic surgery may be an especially vulnerable group.1 Lai et al.2 reported data from 62 patients with PHTN who had noncardiac surgery and found a perioperative mortality rate of 9.7%.2 The lower rate of fatalities in our analysis compared with previous studies may be explained by the shorter capture timeframe spanning only the immediate perioperative period compared with 30-days postoperatively in other reports.
Our study of 3543 patients with the diagnosis of PHTN undergoing primary THA or TKA was able to overcome some of the limitations of previous studies, and provide population-based data on the incidence of perioperative mortality for these frequently performed procedures. The mechanism leading to the significant increase in morbidity and mortality in patients with PHTN undergoing major arthroplasty is not defined by our study. Although no causal relationships can be established from our analysis, PE of intramedullary contents including fat, bone debris, and cement during arthroplasty procedures is a known phenomenon.5,6,18–22 Cardiovascular derangements manifesting in an increase in pulmonary arterial pressures, right heart strain and cardiovascular collapse, and death (in severe cases) have been described.5,18,19 Mechanisms for the various degrees of presentation of these events remain speculative but are thought to be related to the overall load and size of particles embolized to the pulmonary vasculature.5,6 Furthermore, our findings of increased rates of PE, DVT, and ARDS suggest a common pathologic pathway induced by debris embolization to the lung leading to right heart strain and venostasis. In addition, hypoventilation due to residual anesthesia and analgesia might adversely affect pulmonary hemodynamics in patients with PHTN.
Our study is limited by a number of factors inherent to secondary analysis of large administrative databases. Detailed clinical information (i.e., the severity of pulmonary vascular disease and details about the intraoperative course) in the NIS is unavailable. Only inpatient data are available from the NIS database; thus, complications and events after discharge are not captured. Furthermore, readmissions cannot be discerned from this database. Therefore, conclusions should be limited to the acute perioperative setting with the notion that mortality and complications are likely underestimated. In this respect, further study of the National Death Index may yield further information, but this was beyond the scope of this study. Coding or reporting bias with data entry cannot be excluded. This form of bias, however, is likely to affect both PHTN and non-PHTN discharges equally. It is not likely that the outcome of mortality should be subject to this bias. The NIS does not provide information about the severity of PHTN. It is feasible that more severe cases of PHTN are associated with increased risk of adverse outcomes. Ramakrishna et al.1 support this assumption with the finding that a right ventricular systolic pressure to systemic systolic blood pressure ratio ≥0.66 was a predictor for postoperative mortality. We found that patients with primary PHTN had higher mortality compared with other types of PHTN, further supporting that more severe forms of PHTN are associated with high risk for worse outcome.23,24
Surgery and perioperative complications are a substantial challenge for the treatment of patients with PHTN because modifications in specific drug treatment (e.g., converting oral or inhaled therapies to IV drugs) may be necessary.
In summary, this study showed that patients with PHTN are at increased risk for perioperative morbidity and mortality after THA and TKA. Careful consideration of the benefits of performing these procedures in individuals with PHTN seems prudent and may include a discussion about alternative, perhaps less-invasive treatments to address the orthopedic disease process and its consequences.
From the *Department of Anesthesiology, Hospital for Special Surgery; †Division of Biostatistics and Epidemiology, Public Health, Hospital for Special Surgery, Weill Medical College of Cornell University, New York, New York; ‡Department of Anesthesiology and Critical Care Medicine, University of Massachusetts Medical School, Worcester, Massachusetts; and §Department for Lung Development and Remodelling, Max-Planck-Institute for Heart and Lung Research, Bad Nauheim, Germany.
Supported by the Hospital for Special Surgery Anesthesiology Young Investigator Award provided by the Department of Anesthesiology at the Hospital for Special Surgery (SM), the Center for Education and Research in Therapeutics (AHRQ RFA-HS-05-14) (YC and MM), and the Clinical Translational Science Center (NIH UL1-RR024996) (YM and MM).
SGM helped design and conduct the study, analyze the data, write the manuscript, and secure funding. This author has seen the original study data, reviewed the analysis of the data, approved the final manuscript, and is responsible for archiving the study files. YM helped design and conduct the study and analyze the data. This author has seen the original study data, reviewed the analysis of the data, and approved the final manuscript. YLC helped analyze the data and write the manuscript. This author has seen the original study data, reviewed the analysis of the data, and approved the final manuscript. JMW helped design the study and write the manuscript. This author has seen the original study data, reviewed the analysis of the data, and approved the final manuscript. RV helped design the study and write the manuscript. This author has seen the original study data, reviewed the analysis of the data, and approved the final manuscript. MM helped design the study, analyze the data, write the manuscript, and secure funding. This author has seen the original study data, approved the final manuscript, and is responsible for archiving the study files.
a HCUP Databases. Healthcare Cost and Utilization Project (HCUP). Agency for Healthcare Research and Quality, Rockville, MD, 2008. Available at: http://www.hcup-us.ahrq.gov/nisoverview.jsp. Accessed February 15, 2010.
b Introduction to the HCUP National Inpatient Sample (NIS) 2006. Agency for Healthcare Research and Quality, Healthcare Cost and Utilization Project (HCUP), Rockville, MD, 2008. Available at: http://www.hcup-us.ahrq.gov/db/nation/nis/2006NIS_INTRODUCTION.pdf. Accessed February 15, 2010.
c The SURVEYMEANS Procedure. SAS Online Doc. SAS Institute Inc., Cary, NC, 1999. Available at: http://www.okstate.edu/sas/v8/saspdf/stat/chap61.pdf. Accessed June 15, 2010.
1. Ramakrishna G, Sprung J, Ravi BS, Chandrasekaran K, McGoon MD. Impact of pulmonary hypertension on the outcomes of noncardiac surgery: predictors of perioperative morbidity and mortality. J Am Coll Cardiol 2005;45:1691–9
2. Lai HC, Lai HC, Wang KY, Lee WL, Ting CT, Liu TJ. Severe pulmonary hypertension complicates postoperative outcome of non-cardiac surgery. Br J Anaesth 2007;99:184–90
3. Draughn D, Morris DM. Sustained pulmonary hypertension in surgical patients. Am Surg 1993;59:346–9
4. Kurtz S, Mowat F, Ong K, Chan N, Lau E, Halpern M. Prevalence of primary and revision total hip and knee arthroplasty in the United States from 1990 through 2002. J Bone Joint Surg Am 2005;87:1487–97
5. Urban MK, Sheppard R, Gordon MA, Urquhart BL. Right ventricular function during revision total hip arthroplasty. Anesth Analg 1996;82:1225–9
6. Koessler MJ, Fabiani R, Hamer H, Pitto RP. The clinical relevance of embolic events detected by transesophageal echocardiography during cemented total hip arthroplasty: a randomized clinical trial. Anesth Analg 2001;92:49–55
7. Hyduk A, Croft JB, Ayala C, Zheng K, Zheng ZJ, Mensah GA. Pulmonary hypertension surveillance—United States, 1980–2002. MMWR Surveill Summ 2005;54:1–28
8. Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis 1987;40:373–83
9. Deyo RA, Cherkin DC, Ciol MA. Adapting a clinical comorbidity index for use with ICD-9-CM administrative databases. J Clin Epidemiol 1992;45:613–9
10. Flury BK, Riedwyl H. Standard distance in univariate and multivariate analysis. Am Stat 1986;40:249–51
11. Rosenbaum PR. Optimal matching for observational studies. J Am Stat Assoc 1989;84:1024–32
12. Ury HK. Efficiency of case-control studies with multiple controls per case: continuous or dichotomous data. Biometrics 1975;31:643–9
13. Cochran WG. The comparison of percentages in matched samples. Biometrika 1950;37:256–66
14. Mantel N, Haenszel W. Statistical aspects of the analysis of data from retrospective studies of disease. J Natl Cancer Inst 1959;22:719–48
15. Hosmer DW, Lemeshow S. Applied Logistic Regression. 2nd ed. New York: Wiley Interscience, 2000
16. Memtsoudis SG, Ma Y, González Della Valle A, Mazumdar M, Gaber-Baylis LK, MacKenzie CR, Sculco TP. Perioperative outcomes after unilateral and bilateral total knee arthroplasty. Anesthesiology 2009;111:1206–16
17. Kozak LJ. Underreporting of race in the National Hospital Discharge Survey. Adv Data 1995;265:1–12
18. Lafont ND, Kostucki WM, Marchand PH, Michaux MN, Boogaerts JG. Embolism detected by transoesophageal echocardiography during hip arthroplasty. Can J Anaesth 1994;41:850–3
19. Lafont ND, Kalonji MK, Barre J, Guillaume C, Boogaerts JG. Clinical features and echocardiography of embolism during cemented hip arthroplasty. Can J Anaesth 1997;44:112–7
20. Bisignani G, Bisignani M, Pasquale GS, Greco F. Intraoperative embolism and hip arthroplasty: intraoperative transesophageal echocardiographic study. J Cardiovasc Med 2008;9:277–81
21. Duncan JA. Intra-operative collapse or death related to the use of acrylic cement in hip surgery. Anaesthesia 1989;44:149–53
22. Patterson BM, Healey JH, Cornell CN, Sharrock NE. Cardiac arrest during hip arthroplasty with a cemented long-stem component: a report of seven cases. J Bone Joint Surg Am 1991;73:271–7
23. D'Alonzo GE, Barst RJ, Ayres SM, Bergofsky EH, Brundage BH, Detre KM, Fishman AP, Goldring RM, Groves BM, Kernis JT. Survival in patients with primary pulmonary hypertension: results from a national prospective registry. Ann Intern Med 1991;115:343–9
24. ACCF/AHA 2009 expert consensus document on pulmonary hypertension: a report of the American College of Cardiology Foundation Task Force on Expert Consensus Documents and the American Heart Association developed in collaboration with the American College of Chest Physicians; American Thoracic Society Inc.; and the Pulmonary Hypertension Association. McLaughlin VV, Archer SL, Badesch DB, Barst RJ, Farber HW, Lindner JR, Mathier MA, McGoon MD, Park MH, Rosenson RS, Rubin LJ, Tapson VF, Varga J. J Am Coll Cardiol 2009;53:1573–619