Length of Hospital Stay and Complications After Percutaneous Transluminal Coronary Angioplasty
Clinical and Procedural Predictors
Background Although several studies have established that the complications of percutaneous transluminal coronary angioplasty (PTCA) are related to clinical and angiographic variables such as advanced age and lesion complexity, it is uncertain whether the use of hospital resources after PTCA also depends on the same baseline variables. The purpose of this study was to identify the factors responsible for prolonged hospital stay after PTCA.
Methods and Results The study cohort included 591 consecutive patients undergoing conventional balloon angioplasty at nine medical centers in North America. Major or minor complications occurred in 91 patients (15.4%) and were observed to be related to several baseline characteristics, including unstable angina, multivessel coronary artery disease, patient age, and lesion complexity. Compared with a median length of hospital stay of 2.0 days after PTCA (25th, 75th percentiles: 2.0, 4.0) for the entire cohort of patients, the length of stay was increased in patients with unstable angina (3.0 days [2.0, 5.0]; P=.002), multivessel coronary artery disease (3.0 [2.0, 5.5]; P=.001), age >65 years (3.0 [2.0, 5.5]; P=.02), complex lesions (3.0 [2.0, 6.0]; P=.001), and filling defects (6.0 [2.0, 11.0]; P<.001). The length of stay was more strikingly increased, however, in patients who experienced major or minor PTCA complications, such as emergency bypass surgery (9.0 days [8.0, 18.0]; P<.001), Q-wave or non–Q-wave myocardial infarction (8.0 [6.0, 15.5]; P<.001), transfusion unrelated to bypass surgery (8.0 [4.0, 12.0]; P<.001), or abrupt vessel closure (6.0 [3.0, 10.5]; P<.001). On stepwise multiple linear regression, PTCA complications appeared to be the strongest predictors of length of hospital stay (all P<.001) and overwhelmed the weaker relation between length of stay and several individual baseline variables. Inclusion of a composite clinical risk score (reflecting the presence of unstable angina, multivessel disease, advanced age, complex lesions, or filling defects) in the regression model confirmed that patients with several high-risk baseline variables had a significant increase in length of stay after PTCA (P=.003), but PTCA complications remained the strongest predictors of length of stay.
Conclusions Although PTCA complications were correlated with baseline variables such as unstable angina, multivessel disease, advanced age, complex lesions, and filling defects, excess length of stay after PTCA was most strongly influenced by the development of minor and major PTCA complications. Because patients with several baseline risk factors experienced significantly prolonged hospitalizations, improved selection of patients may contribute to reductions in length of stay after PTCA. A greater reduction in resource use after PTCA, however, would be expected from developing new treatments to decrease PTCA complications rather than limiting the access of patients with unstable angina, advanced age, or complex lesions to PTCA.
Although percutaneous transluminal coronary angioplasty (PTCA) has been an important advance in the treatment of patients with coronary artery disease, this treatment is associated with complications such as myocardial infarction or the need for emergency bypass surgery in 2% to 7% of patients.1 2 3 4 5 6 7 It has been established in several studies that patients with advanced age,8 unstable angina,9 and complex lesion morphology1 6 7 9 are at increased risk for PTCA complications.
Acute PTCA complications increase the use of hospital resources. Because resource use during PTCA has come under increased scrutiny as managed care systems and hospital partnerships emerge in an effort to reduce health care costs, it is important to identify the factors associated with increased use of resources, such as length of hospitalization after PTCA. Several studies have evaluated the relative costs of interventional cardiovascular procedures and bypass surgery,10 11 12 13 14 15 but the predictors of increased resource use after PTCA remain to be defined. The purpose of this study was to identify the factors responsible for prolonged hospital stay as a marker of resource use in patients undergoing PTCA.
Consecutive patients undergoing conventional balloon angioplasty at nine medical centers in North America between February 1, 1992, and April 30, 1992, were prospectively enrolled in this study, which was denoted the Heparin Registry and performed as a pilot study for the Hirulog Angioplasty Study to identify the outcome of patients undergoing angioplasty with heparin dosing guided by activated clotting time measurements. Patients were excluded if they were prospectively enrolled in an investigational angioplasty trial or scheduled for treatment with a second-generation angioplasty device such as laser, stent, or atherectomy. Approval for the study was obtained from the institutional review board of each participating hospital.
Balloon angioplasty was performed according to local practices at each participating institution. The protocol required intraprocedural monitoring of heparin anticoagulation with activated clotting time measurements. Angiography of the target lesion was recorded on cine film in two orthogonal or nearly orthogonal views before treatment, after each series of balloon inflations, and at the end of the procedure after the guide wire was removed from the coronary artery. If a patient developed evidence of ischemia after PTCA, repeat angiography was obligatory.
Clinical success was defined as successful dilatation of the target lesion (<50% residual stenosis as measured in the core angiographic laboratory) plus no major complication at any time during hospitalization. Major complications were defined as death, nonfatal myocardial infarction, and emergency bypass surgery (within 24 hours of PTCA). Abrupt vessel closure was classified hierarchically as either established or threatened closure. Established closure was defined by National Heart, Lung, and Blood Institute criteria5 as total or subtotal occlusion of the vessel after attempted angioplasty with corresponding Thrombolysis in Myocardial Infarction (TIMI) grade 0 to 1 flow. Threatened closure was defined angiographically in the core laboratory as an unstable lesion with >50% stenosis and decreased flow (less than Thrombolysis in Myocardial Infarction grade 3) after initial successful dilatation, requiring the use of an additional intervention such as thrombolytic therapy, a perfusion balloon catheter that had not been previously used, an intracoronary stent, or return to the cardiac catheterization laboratory for repeat coronary angioplasty.16
Unstable angina was defined as new, crescendo, or rest ischemic chest pain within 6 weeks preceding PTCA. Postinfarction angina was defined as ischemic chest pain occurring within 6 weeks of myocardial infarction. Multivessel disease was defined by the presence of >70% stenosis (by visual assessment) in at least two major coronary artery distributions (left main coronary artery; left anterior descending coronary artery or diagonal branches; left circumflex coronary artery, obtuse marginal branches, and posterior descending artery for left-dominant circulations; or right coronary artery and posterior descending artery for right-dominant circulations).17 Lesion complexity was graded according to the classification of the American College of Cardiology/American Heart Association Task Force.1 18 A filling defect was defined as the presence of a globular intracoronary shape surrounded by contrast on at least three sides, usually located immediately downstream from a stenosis.19 20 21 A myocardial infarction was defined by the presence of at least two of the three following criteria: (1) prolonged ischemic pain >30 minutes, (2) total creatine kinase elevation to greater than twice the upper limit of normal (confirmed by ≥2.0% creatine kinase-MB), and (3) ECG evidence of infarction.
All patients were followed prospectively from the time of PTCA to the time of hospital discharge by the research coordinator at each participating center. The protocol required preprocedure and postprocedure nadir measurements of hematocrit and hemoglobin, as well as measurements of creatine kinase at least once ≈12 hours after PTCA in all patients. Creatine kinase measurements were required more often (every 8 to 12 hours) if ischemic chest pain occurred. Clinical data were entered on standardized case report forms and submitted to a core clinical data center. The validity of the clinical data was documented by review of every medical record by an independent clinical monitor. Length of stay was recorded to the nearest 0.5 day of hospitalization.
Core Angiographic Laboratory
Angiograms were sent to a core angiographic laboratory to code lesion complexity and to measure minimal lumen diameter, lesion length, and stenosis severity with electronic digital calipers of optically magnified frames, with the centered, contrast-filled guide catheter used as reference. All angiograms were reviewed by experienced angiographers who were not involved in the performance of the procedures and were blinded to clinical outcome. The method of assessment used in this study has been validated against computerized edge-detection techniques.22
Fisher’s exact test and unpaired Student’s t test were used to determine baseline differences between the groups with and without complications for categorical and normally distributed continuous variables. Median values for length of stay between groups with and without complications were compared by the Mann-Whitney U test for single comparisons and by the Kruskal-Wallis test statistic for multiple comparisons.23 Normality was assessed with Kolmogorov-Smirnov testing.24
Logistic regression analysis was used to identify predictors of complications from a series of clinical (age, sex, diabetes, postinfarction angina, unstable angina, and multivessel disease), angiographic (lesion complexity, filling defects), and procedural (median value for activated clotting time) variables.25 Odds ratios (ORs) were provided to estimate the probability that patients with a given variable had increased likelihood of PTCA complication compared with all other patients without the variable.26 Of the variables evaluated, those found to have at least borderline significance on univariable analysis (P<.15) were included in the multivariable analysis.
Stepwise linear regression analysis was used to identify predictors of length of hospital stay from a series of clinical, angiographic, and procedural variables (death, myocardial infarction, abrupt vessel closure, and transfusion).27 Forward stepping was performed with a “P” to enter and a “P” to remove of .15. In both the logistic and linear regression analyses, the number of independent variables was limited to approximately 0.1 the number of complication events to reduce the likelihood of multicollinearity and “overfitting.”28 Because the length-of-stay measurements were skewed, stepwise linear regression analysis was also carried out after logarithmic transformation of the dependent variable.29 A composite clinical risk score, based on the baseline and angiographic variables from logistic regression analysis significantly related to postangioplasty complications, was constructed. The risk score was calculated as the mathematical sum of all significant baseline variables (weighted according to their coefficients from the logistic model) and included in the stepwise linear regression analysis as a replacement for the individual baseline variables.
Because the dependent variables analyzed were patient-related outcomes, all analyses were performed with a database containing one record per patient. When the independent variable was a lesion characteristic such as lesion complexity, the most complex lesion grade for each patient was entered. Missing data for length of stay (1.7%) or angiographic characteristics (3.0%) were excluded from the analyses. All analyses were performed with standard statistical software (systat 5.1, logit 2.0). All continuous data are presented as mean±SD or as median with 25th and 75th percentile (midrange) where specified.
The 591 patients in the study had a mean age of 61±11 years (Table 1⇓). The peak activated clotting time achieved during the PTCA was 363±92 seconds. Angioplasty was performed for unstable angina in 386 patients (65%), after recent myocardial infarction in 115 patients (19%), and in the presence of a filling defect in 21 patients (3.6%). The patients had a total of 756 lesions treated with PTCA (Table 2⇓). Films were available for review at the core laboratory for 733 lesions, of which 31% were graded as type A, 43% type B, and 26% type C.
Angioplasty Outcome and Complications
Forty-five patients (7.6%) experienced at least one major complication during hospitalization (Table 3⇓): 9 patients died (1.5%), 9 patients had Q-wave myocardial infarction (1.5%), 16 patients experienced non–Q-wave myocardial infarction (2.7%), and 19 patients required emergency bypass surgery (3.2%) within 24 hours of the PTCA procedure.
Abrupt vessel closure occurred in 65 of 591 patients (11%). Abrupt vessel closure was associated with a filling defect in 11 patients (17%), dissection in 15 (23%), and both angiographic findings in 16 (25%). Established closure was seen in 42 patients (7.1%) and threatened closure in 23 patients (3.9%). Strategies for managing abrupt vessel closure included repeat coronary angioplasty (18 patients), intracoronary stenting (11 patients), prolonged inflation with a perfusion balloon catheter (15 patients), or intracoronary thrombolytic therapy (18 patients). Major complications occurred in 34% of the patients with abrupt vessel closure versus 5.9% of patients without closure (P<.001). Established and threatened vessel closure both affected PTCA outcome: in the 42 patients with established closure, the major complication rate was 38% (OR=11.0 [5.30, 22.8]; P<.001), and in the 23 patients with threatened closure, the major complication rate was 17% (OR=2.9 [1.0, 8.6]; P=.05).
A total of 18 of 591 patients (3.1%) had a decrease in hemoglobin of >5 g/dL or decrease in hematocrit of >15 points. Although 51 of 591 patients (8.6%) in this study required red cell transfusions, only 29 patients (4.9%) required transfusions for bleeding unrelated to bypass surgery. These patients were treated either for a nadir hematocrit of ≤29 points (26 patients) or a decrease in hematocrit of ≥10 points (3 patients).
Predictors of Ischemic Complications
Ninety-one patients (15.4%) experienced a major ischemic complication, abrupt vessel closure, or both. Complications were found to be associated with several baseline clinical and angiographic variables on multivariable logistic regression analysis (Fig 1⇓). The presence of a filling defect before PTCA, as detected in the core angiographic laboratory, increased the likelihood of any complication (complication rate, 38.1%; OR, 3.1 [95% confidence interval, 1.2, 8.4]; P=.02). Other variables associated with an increased risk of major or minor complications included multivessel coronary artery disease (complication rate, 19.8%; OR, 1.9 [1.1, 3.3]; P=.02), unstable angina (complication rate, 18.4%; OR, 1.9 [1.1, 3.3]; P=.03), lesion complexity (OR, 1.6 [1.1, 2.2] for an additional increase in grade18 ; P=.01), and patient age (OR, 1.4 [1.1, 1.8] per additional decade; P=.002). Activated clotting time (median value, 350 seconds; range, 280 to 1032 seconds), female sex, and diabetes were not related to PTCA complications on univariable analysis.
Length of Hospital Stay
Hospital stay after PTCA for the 591 patients ranged from 1 to 49 days (median, 2.0 days [25th, 75th percentiles, 2.0, 4.0]) and was related to several baseline clinical variables (Table 1⇑). Patients >65 years old had slightly longer hospital stays after PTCA than younger patients (P=.02), women had longer stays than men (P=.01), and patients with unstable angina (P=.002) or postinfarction angina (P<.001) had longer stays than patients with stable angina. Baseline angiographic variables were also related to length of stay after PTCA. Multilesion angioplasty was associated with a longer length of stay than single-lesion procedures (P=.02), and angioplasty for complex lesions was associated with a longer length of stay than procedures for simple lesions (P=.001).
The presence of intracoronary filling defects before PTCA was associated a threefold increase in length of stay after PTCA (P<.001). The excess length of stay in this group of patients was probably due to an increased incidence of complications and not due to elective, prolonged heparinization. The total amount of heparin used was similar for patients with and without filling defects (52 600±51 900 versus 37 200±38 900 U; P=NS), whereas the number of pretreatment hospital days was lower for patients undergoing PTCA for filling defects than for patients without this finding (0.0 [0.0, 3.0] versus 1.0 [0.0, 3.0]; P<.001).
Length of hospital stay was strongly affected by PTCA outcome (Table 3⇑). Compared with a median length of stay of 2.0 days (2.0, 4.0) for 500 patients after uncomplicated PTCA (Fig 2⇓), the length of stay was 9.0 days (8.0, 18.0) for 19 patients requiring emergency bypass surgery (P<.001), 8.0 days (6.0, 15.5) for 25 patients experiencing Q-wave or non–Q-wave myocardial infarction (P<.001), and 6.0 days (3.0, 10.5) for 65 patients with established or threatened abrupt vessel closure (P<.001). The median length of stay for the 43 patients with abrupt vessel closure but without major complication was 4.5 days (3.0, 6.0), which was significantly increased over that for the patients without any complication (P<.001).
The median length of stay for 29 patients (4.9%) requiring transfusion for vascular complications or blood loss unrelated to bypass surgery was 8.0 days (4.0, 12.0). Patients requiring transfusion in the absence of bypass surgery were more likely to have been treated with additional heparin (55 800±70 200 versus 36 500±36 400 U; P=.007).
The total length of hospital stay for the 591 patients ranged from 1 to 73 days (median, 4.0 days [3.0, 8.0]) and was influenced by PTCA outcome to a degree similar to that for postprocedural length of stay. All values for length of stay were skewed and significantly different from a normal distribution (P<.001).
Predictors of Length of Stay
On stepwise multiple linear regression analysis, several baseline clinical and angiographic variables were related to length of stay (Table 4⇓). In the multivariable model, the average influence of each variable on length of stay could be estimated by its regression coefficient. Advanced age was associated with increased length of stay after PTCA by 0.6±0.2 days per decade (P=.002), unstable angina by 1.0±0.4 days (P=.02), multivessel disease by 1.0±0.4 days (P=.01), and filling defects by 2.5±1.0 days (P=.02).
On stepwise linear regression analysis, length of stay after PTCA was observed to be strongly related to major and minor PTCA complications (Table 5⇓). Major complications such as emergency bypass surgery increased the length of stay after PTCA by 7.8±1.0 days (P<.001), and myocardial infarction increased length of stay by 3.4±0.9 days (P<.001). Complications of PTCA traditionally considered to be minor also increased length of stay significantly. Transfusion unrelated to bypass surgery increased length of stay by 4.8±0.8 days (P<.001), and abrupt vessel closure increased length of stay by 3.0±0.7 days (P<.001).
Combined Models for Length of Stay
Several linear regression models were constructed to identify the strongest predictors of excess length of stay from a combination of baseline variables and PTCA complications. In these models, individual clinical variables such as age, female sex, diabetes, unstable angina, multivessel coronary artery disease, number of lesions treated, and lesion complexity did not have a significant independent influence on length of stay. No clinical or angiographic variable was associated with increased length of stay of >1.0 day, whereas each major or minor PTCA complication was associated with increases in length of stay ranging from 2.9±0.9 days for abrupt closure to 9.6±1.1 days for emergency bypass surgery. When logarithmic transformation of the dependent variable (length of stay) was used in the stepwise multiple linear regression analysis, the results were similar with the exception that death was no longer significantly associated with increased length of stay (P=.21), but multivessel disease showed borderline association with increased length of stay (P=.06). When total length of stay was used as the untransformed dependent variable in the multivariable analyses, the results were similar to those obtained for postprocedural length of stay.
The interrelations among baseline variables, complications, and length of stay were explored further with the construction of a composite risk score based on the presence or absence of unstable angina, multivessel coronary artery disease, advanced age, and complex lesions. Patients with three or four risk factors had a fivefold increased risk of major complication, fourfold greater risk of transfusion, and twofold longer hospitalization than patients with no risk factors (Table 6⇓). When the presence or absence of unstable angina, multivessel coronary artery disease, age >65 years, and complex lesions was scored with a value of 1 or 0, a composite score ranging from 0 for patients with no risk factors to a maximum value of 4 for patients with all risk factors was established. In stepwise linear regression analysis, in which the composite risk score replaced the individual baseline risk factors, the strongest correlates of excess length of stay were PTCA complications: emergency bypass surgery, 7.8±1.0 days (P<.001); death, 7.8±1.4 days (P<.001); transfusion in the absence of bypass surgery, 4.6±0.8 days (P<.001); myocardial infarction, 3.2±0.9 days (P<.001); and abrupt vessel closure, 2.9±0.6 days (P<.001). In the new model, the composite risk score was also found to influence length of stay significantly, by 0.4±0.2 day per risk factor (P=.01).
A linear regression model containing a weighted composite risk score based on the coefficients from logistic regression analysis (unstable angina, 0.64; multivessel coronary artery disease, 0.64; age, 0.34 per decade; type C lesion, 0.88; type B lesion, 0.44; and lesion with filling defect, 1.11) produced the strongest evidence of a relation between baseline variables and length of stay after PTCA (Table 7⇓). Although the strongest correlates of length of stay in the model continued to be PTCA complications (each associated with increases in length of stay of ≥3.0 days and P<.001), the weighted composite risk score was associated with an increase in length of stay of 0.7±0.2 day per risk factor (P=.003). The standardized “unitless” coefficient for the weighted composite clinical risk score (0.11) approximated that for acute myocardial infarction (0.13), suggesting that the presence of several baseline risk factors in the same patient strongly predicted a prolonged hospitalization after PTCA.
One notable finding in this study was that PTCA complications traditionally considered to be minor, such as uncomplicated abrupt vessel closure or the need for blood transfusions unrelated to bypass surgery, were associated with excess lengths of hospital stay equivalent to those for major PTCA complications. Although the PTCA complications themselves were related to several baseline clinical and angiographic variables, the individual baseline variables in this cohort of patients had limited ability to predict prolonged hospitalization. In several statistical models, the relatively weak relation between length of stay and individual baseline clinical variables was overwhelmed by the striking relation between length of stay and PTCA complications. A composite clinical risk score (reflecting the presence or absence of unstable angina, multivessel coronary disease, advanced age, complex lesions, and filling defects) established that a high-risk clinical profile was indeed related to increased length of stay after PTCA.
The relation between baseline clinical variables and complications in this study replicated the association between risk and PTCA outcome reported in several studies1 2 3 4 5 6 7 and established that the patient cohort studied here was similar to that analyzed in other reports. Patients in this study had a mean age of 61 years, which is similar to the mean ages of 54 to 62 years in other reports,4 6 12 30 a rate of unstable angina (65%) similar to that reported for patients undergoing balloon angioplasty in the Coronary Angioplasty Versus Excisional Atherectomy Trial (70%),12 and an incidence of multivessel disease (55%) slightly greater than that reported in most studies (20% to 53%).4 6 12 30
Predictors of Complications
In this study, patients with advanced age, multivessel coronary artery disease, unstable angina, lesion complexity, and filling defects were found to have increased rates of ischemic complications and abrupt vessel closure. These results are in agreement with those reported in several earlier studies, in which advanced age,8 unstable angina,9 and complex lesion morphology1 6 7 9 were identified as the strongest predictors of PTCA complications.
Predictors of Length of Hospital Stay
Patients in this study were found to have a median length of stay after PTCA of 2.0 days. When total length of hospital stay was used in the multivariable analyses, similar results were seen. For this report, we selected the length of stay after PTCA as the variable for analysis because it can be compared more easily to several economic analyses of patients admitted to the hospital on the day of their elective angioplasty.10 11 Dick and colleagues10 found that the mean length of stay for 149 elective patients admitted to the hospital on the day of treatment was 1.5 days after balloon angioplasty, 2.2 days after directional atherectomy, and 4.9 days after stenting. Although it was observed that stent placement and vascular complications increased the cost of interventional procedures, the impact of baseline variables such as unstable angina or procedural complications on costs or length of stay were not directly assessed.10 In a detailed economic analysis, Cohen and colleagues11 reported length of stay and cost and charge data for patients undergoing balloon angioplasty, directional atherectomy, stenting, and bypass surgery. In 113 patients admitted to the hospital on the day of elective balloon angioplasty, the mean length of stay was 2.6 days. The mean length of stay was similar for elective directional atherectomy (2.3 days), but increased for both elective coronary stenting (5.5 days) and elective bypass surgery (9.3 days).11 The investigators for the New Approaches to Coronary Intervention registry31 recently reported length-of-stay data for nonballoon devices in both elective and unstable patients: the total lengths of stay for directional atherectomy (3.0 days), laser angioplasty (3.0 days), rotational atherectomy (4.0 days), and stent implantation (7.0 days) were similar to those measured for balloon angioplasty in our study. These investigators also observed that median length of stay approximately doubled when devices were used in an unplanned mode (presumably as salvage therapy) compared with elective planned use.31
In this study, abrupt vessel closure increased the median length of stay after PTCA to 6.0 days. Both established and threatened closure increased the likelihood of complications and length of stay. Although 34% of patients with abrupt vessel closure required prolonged hospitalization for myocardial infarction or bypass surgery, the remaining patients with uncomplicated abrupt vessel closure also required prolonged hospitalization. The increased length of stay for these patients involved additional observation, repeat angiography, and prolonged anticoagulation.
Another important finding in this study was that patients with vascular complications or blood loss requiring transfusion therapy, even in the absence of bypass surgery, had prolonged lengths of stay after angioplasty. For patients requiring transfusion, the prolonged length of stay was associated with increased heparin dosing, prolonged observation, and the time required for ordering, processing, and administration of red cell transfusions.
Several statistical models were developed in this study to identify the predictors of excess length of stay after PTCA. All models identified PTCA complications as the strongest predictors of excess length of stay, which overwhelmed the influence of several individual baseline characteristics. The development of the composite risk score, however, added statistical power to the prediction of excess length of stay from baseline characteristics, because the assessment of several baseline variables together represented the clinical risk profile more accurately than the analysis of each variable individually.
Although the number of patients undergoing PTCA for lesions associated with filling defects in this study was relatively small, this angiographic finding emerged as the strongest baseline predictor of increased complications and excess length of stay after PTCA. Because angioplasty performed early for patients with filling defects may be associated with increased complication rates, angioplasty delayed to a more propitious time after stabilization of the ischemia-related stenosis with prolonged heparin and aspirin may be associated with lower complication rates. This strategy has been retrospectively evaluated in several studies. The procedural success rate was higher (91% versus 81% success rate, P=.02) and the abrupt closure rate lower (2% versus 8%, P<.01) for patients with unstable angina who were treated with heparin for ≥24 hours before coronary angioplasty than for those patients who did not receive heparin.32 Other investigators have observed a trend toward improved outcome if angioplasty can be deferred for about 2 weeks after the onset of unstable angina.33 In a prospective study, the clinical and angiographic outcome of 18 patients with coronary thrombus undergoing angioplasty without heparin pretreatment was compared with that of a group of 35 patients receiving preprocedural heparin therapy.34 The untreated group had a significant reduction in angiographic success (61% versus 94%, P<.05) and a significant increase in immediate postprocedural thrombotic arterial occlusion (33% versus 6%, P<.05) compared with the group pretreated with heparin. Thus, pretreatment with heparin for patients with filling defects may be a safer, more cost-effective strategy than early PTCA without pretreatment.
Limitations of the Study
Limitations of the study included the use of length of stay as a proxy for resource use. Hospital costs, hospital charges, and measures of physician effort at each participating institution were not available but would have provided a more direct index of resource use. It is possible that certain patients in this study experienced a short but intensive hospitalization associated with increased resource use, which would have been missed by using length of stay as a surrogate for resource use. On the other hand, direct measures of resource use such as hospital costs or charges may have been difficult to normalize and include in the analyses here, because such data would have been derived from disparate payer systems affiliated with the health care systems in two countries and in several different hospital settings, including academic medical centers, community hospitals, and a Veterans Administration hospital. Because a direct economic valuation of balloon angioplasty was difficult to obtain for this analysis, length of hospital stay was used as a reliable, easily measured surrogate for resource use. It is unlikely, however, that use of length of stay as a proxy for resource use exaggerated the relations reported in this study, because all patients in this study were treated with the same technology (balloon angioplasty). In addition, hospital costs and length of stay were likely to be directly related, as has been reported in several studies involving patients undergoing interventional procedures.10 11 13 Also, the intensity of resource consumption (for example, intensive care or bypass surgery) for patients with complications would have tended to exaggerate the observed relations between length of stay and cost, as they have for coronary artery bypass surgery.15 Although length of stay was found to vary among the institutions in the study, this was attributed to the intersite variability in PTCA complications, because study site was not a significant predictor of length of stay in multivariable analysis.
Another limitation of the study is the potential problem of sample-based multicollinearity in the regression analysis,28 35 which may falsely weaken the relation between baseline variables and length of stay. This problem arises because baseline clinical variables were found to be correlated with PTCA complications and both the baseline variables and PTCA complications were evaluated together as correlates of length of stay. Although it has been proposed that the problem of multicollinearity may reduce the precision of the regression coefficients presented in Tables 4⇑, 5⇑, and 7⇑, it does not affect the relative weights of the coefficients or the empirical interpretation of the overall regression equation.35 The composite risk score was developed in an effort to minimize the problem of multicollinearity affecting the individual baseline variables.
Although patients with several baseline risk factors such as unstable angina, multivessel disease, age >65 years, and complex lesions are at slightly increased risk for prolonged hospital stay after PTCA, this analysis shows that acute PTCA complications such as abrupt vessel closure, emergency surgery, myocardial infarction, and blood loss requiring transfusion are the strongest determinants of excess length of stay. These findings suggest that new approaches are needed to refine patient selection for PTCA and reduce the incidence of minor and major complications to decrease the length of hospitalization after PTCA.
This study was supported by a research grant from Biogen, Inc, Cambridge, Mass. The authors wish to acknowledge the assistance of Paula McCray, Jack Gulinello, and John Murray for data processing and analysis. The authors are indebted to Drs John Maranganore, Irving Fox, Martin B. Leon, and Eugene Braunwald for many valuable discussions. The authors wish to acknowledge the statistical assistance of Arthur McAllister, MSc, Gerry Oster, PhD, and Thomas Delea, MBA.
Reprint requests to John A. Bittl, MD, Cardiovascular Division, Brigham and Women’s Hospital, Boston, MA 02115.
of Cardiology, Loyola University, Maywood, Ill (F.L.); Montreal Heart Institute, Montreal, Quebec (R.B.); Medical Research Service, McGuire VA Medical Center, Richmond, Va (J.S.); and Biogen, Inc, Cambridge, Mass (B.A.).
Presented in part at the 66th Scientific Sessions of the American Heart Association, Atlanta, Ga, November 8-11, 1993.
- Received November 10, 1994.
- Revision received January 17, 1995.
- Accepted January 30, 1995.
- Copyright © 1995 by American Heart Association
Ellis SG, Vandormael MG, Cowley MJ, DiSciascio G, Deligonul U, Topol E, Bulle TM, Multivessel Angioplasty Prognosis Study Group. Coronary morphologic and clinical determinants of procedural outcome with angioplasty for multivessel coronary disease: implications for patient selection. Circulation. 1990;82:1193-1202.
Ellis SG, Roubin GS, King SB III, Douglas JS Jr, Weintraub WS, Thomas RG, Cox WR. Angiographic and clinical predictors of acute closure after native vessel coronary angioplasty. Circulation. 1988;77:372-379.
Detre K, Holubkov R, Kelsey S, Cowley M, Kent K, Williams D, Myler R, Faxon D, Holmes D Jr, Bourassa M, Block P, Gosselin A, Bentivoglio L, Leatherman L, Dorros G, King S III, Galichia J, Al-Bassam M, Leon M, Robertson T, Passamani E, and coinvestigators of the National Heart, Lung, and Blood Institute Percutaneous Transluminal Coronary Angioplasty Registry. Percutaneous transluminal coronary angioplasty in 1985-1986 and 1977-1981: the National Heart, Lung, and Blood Institute Registry. N Engl J Med. 1988;318:265-270.
Detre KM, Holmes DR Jr, Holubkov R, Cowley MJ, Bourassa MG, Faxon DP, Dorros GR, Bentivoglio LG, Kent KM, Myler RK. Incidence and consequences of periprocedural occlusion: the 1985-1986 National Heart, Lung, and Blood Institute Percutaneous Transluminal Coronary Angioplasty Registry. Circulation. 1990;82:739-750.
Topol EJ, Leya F, Pinkerton CA, Whitlow PL, Hofling B, Simonton CA, Masden RR, Serruys PW, Leon MB, Williams DO, King SB III, Mark DB, Isner JM, Holmes DR, Ellis SG, Lee KL, Keeler G, Berdan LG, Hinohara T, Califf RM, on behalf of the CAVEAT Study Group. A comparison of balloon angioplasty with directional atherectomy in patients with coronary artery disease. N Engl J Med. 1993;329:221-227.
Roubin GS, Cannon AD, Agrawal SK, Macander PJ, Dean LS, Baxley WA, Breland J. Intracoronary stenting for acute and threatened closure complicating percutaneous transluminal coronary angioplasty. Circulation. 1992;85:916-927.
Ryan TJ, Bauman WB, Kennedy JW, Kereiakes DJ, King SB III, McCallister BD, Smith SC, Ullyot DJ. Guidelines for percutaneous transluminal coronary angioplasty: a report of the American Heart Association/American College of Cardiology Task Force on Assessment of Diagnostic and Therapeutic Cardiovascular Procedures (Subcommittee on Percutaneous Transluminal Coronary Angioplasty). Circulation. 1993;88:2987-3007.
The TIMI IIIA Investigators. Early effects of tissue-type plasminogen activator added to conventional therapy on the culprit coronary lesion in patients presenting with ischemic cardiac pain at rest: results of the Thrombolysis in Myocardial Ischemia (TIMI IIIA) Trial. Circulation. 1993;87:38-52.
Uehata A, Matsugushi T, Bittl JA, Meredith I, Anderson T, Ryan TJ Jr, Selwyn A, Ganz P, Yeung A. The accuracy of electronic digital calipers compared with quantitative angiography in measuring arterial diameters. Circulation. 1993;88:1724-1729.
Zar JH. Two-sample hypotheses. In: Zar JH. Biostatistical Analysis. Englewood Cliffs, NJ: Prentice-Hall; 1984:122-149.
Zar JH. The normal distribution. In: Zar JH. Biostatistical Analysis. Englewood Cliffs, NJ: Prentice-Hall; 1984:79-96.
Glantz SA, Slinker BK. Regression with a qualitative dependent variable. In: Glantz SA, Slinker BK, eds. Primer of Applied Regression and Analysis of Variance. New York, NY: McGraw-Hill Inc; 1990:512-568.
Kahn HA, Sempos CT. Adjustment of data without use of multivariate models. In: Kahn HA, Sempos CT, eds. Statistical Methods in Epidemiology. New York, NY: Oxford University Press; 1989:85-167.
Glantz SA, Slinker BK. Regression with two or more independent variables. In: Glantz SA, Slinker BK, eds. Primer of Applied Regression and Analysis of Variance. New York, NY: McGraw-Hill Inc; 1990:50-109.
Concato J, Feinstein AR, Holford TR. Basic science review: the risk of determining risk with multivariable models. Ann Intern Med. 1983;118:201-210.
Zar JH. Simple linear regression. In: Zar JH, ed. Biostatistical Analysis. Englewood Cliffs, NJ: Prentice-Hall; 1984:261-291.
Baim DS, Kent KM, King SB III, Safian RD, Cowley MJ, Holmes DR, Roubin GS, Gallup D, Steenkiste AR, Detre K. Evaluating new devices: acute (in-hospital) results from the New Approaches to Coronary Intervention registry. Circulation. 1994;89:471-481.
de Feyter PJ, Serruys PW, Soward A, van der Brand M, Bos E, Hugenholtz PG. Coronary angioplasty for early postinfarction unstable angina. Circulation. 1986;74:1365-1370.
Glantz SA, Slinker BK. Multicollinearity and what to do about it. In: Glantz SA, Slinker BK, eds. Primer of Applied Regression and Analysis of Variance. New York, NY: McGraw-Hill Inc; 1990:181-238.