SEARCH

SEARCH BY CITATION

Keywords:

  • venous thromboembolism;
  • perioperative complications;
  • prostatectomy;
  • prostatic neoplasms

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. PATIENTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. ACKNOWLEDGEMENTS
  8. CONFLICT OF INTEREST
  9. REFERENCES

Study Type – Therapy (case series)

Level of Evidence 4

What's known on the subject? and What does the study add?

Several risk factors increase VTE after RP: advanced age, comorbidities such as cardiopulmonary disease, rheumatologic diseases, prior history of VTE, more advanced prostate cancer, and simultaneous pelvic lymph node dissection. To date, the effect of annual surgical caseload (ASC), an established determinant of various RP outcomes, has not been tested. A previous study showed in adjusted analyses that patients operated for colorectal cancer by very high ASC surgeons were 60% less likely to suffer a VTE than those operated by low ASC surgeons. Moreover, some authors hypothesized that laparoscopy may contribute to a higher risk of VTE, due to peritoneal insufflation, reverse Trendelenburg position and prolonged operative time. The VTE rates reported in the current population-based study closely reflect those reported in institutional series. Moreover, we validated the practice-makes-perfect concept, since ASC was linked to VTE. We could not detect statistically significantly differences between minimally invasive radical prostatectomy (MIRP) patients and others. Our results indicate that lower rates of VTE should be expected in patients treated by high ASC surgeons. Our findings suggest that VTE-specific processes of care need to be improved, with the intent of reaching the level recorded in patients treated by high ASC surgeons. Finally, MIRP seems to be no risk factor for VTE.

OBJECTIVE

  • • 
    To examine the effect of annual surgical caseload (ASC) on the likelihood of venous thromboembolism (VTE) after radical prostatectomy (RP).

PATIENTS AND METHODS

  • • 
    Between 1999 and 2008, 36 699 RPs were performed in the state of Florida. Logistic regression models predicting the likelihood of VTE were fitted.
  • • 
    Covariates included year of surgery, age, race, baseline Charlson Comorbidity Index (CCI), lymph node dissection, ASC and surgical approach.

RESULTS

  • • 
    The overall VTE rate was 0.3%. It was higher in patients operated within the low (0.4%) and intermediate (0.3%) ASC tertile than in those operated within the high-ASC tertile (0.1%, P < 0.001).
  • • 
    Mortality rate was 6.0% in patients with VTE vs 0.1% in others (P < 0.001). Median length of stay and median total hospital charges were 9 vs 3 days (P < 0.001) and $51 571 vs $24 943 (P < 0.001) in patients with VTE vs others, respectively.
  • • 
    In multivariable analyses predicting VTE, patients operated on by low-ASC surgeons were at higher risk of VTE than those operated on by high-ASC surgeons (odds ratio [OR]= 3.78, P < 0.001). Additionally, black patients were more likely to experience a VTE (OR = 1.80, P= 0.023). Patients with CCI ≥ 1 were also more likely to experience a VTE than others (OR = 1.65, P= 0.016). Conversely, patients who had undergone minimally invasive radical prostatectomy were not more likely to experience a VTE than those who had undergone open RP (OR = 1.97, P= 0.086).

CONCLUSIONS

  • • 
    RP by high-ASC surgeons exerts a protective effect on the likelihood of VTE.
  • • 
    Additionally, VTE is associated with higher mortality, prolonged length of stay and increased hospital charges.

Abbreviations
ASC

annual surgical caseload

CCI

Charlson Comorbidity Index

IQR

interquartile range

LOS

length of stay

OR

odds ratio

PLND

pelvic lymph node dissection

RP

radical prostatectomy

VTE

venous thromboembolism (VTE)

INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. PATIENTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. ACKNOWLEDGEMENTS
  8. CONFLICT OF INTEREST
  9. REFERENCES

Prostate cancer is the second leading cause of cancer mortality in the United States, and radical prostatectomy (RP) is the most common treatment option for clinically localized disease in appropriately selected individuals [1,2]. Venous thromboembolism (VTE) is a life-threatening complication of RP and a recent population-based report estimated the rates of thromboembolic events at between 2.9% and 3.9%, irrespective of surgical approach [3]. Even though VTE after RP has decreased with the routine use of VTE prophylaxis, it is still associated with a substantially higher risk of mortality than other complications [4]. Several risk factors increase VTE after RP: advanced age, comorbidities such as cardiopulmonary disease, rheumatological diseases, prior history of VTE, more advanced prostate cancer and simultaneous pelvic lymph node dissection (PLND) [5–10]. To date, the effect of annual surgical caseload (ASC), an established determinant of various RP outcomes, has not been tested [11]. To address this limitation, we examined the effect of ASC on the probability of developing a VTE within a large population-based cohort from the United States.

PATIENTS AND METHODS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. PATIENTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. ACKNOWLEDGEMENTS
  8. CONFLICT OF INTEREST
  9. REFERENCES

We relied on the 1999 to 2008 Florida Hospital Inpatient Datafile, which provides medical record information from hospital admission to discharge [12]. Exceptions are freestanding psychiatric rehabilitation, Veterans Administration, Bureau of Indian Affairs and state-operated hospitals. These exceptions resulted in the exclusion of 19.8% of all hospitals from analyses. Data collection at participating institutions is mandatory and done by trained professional coders. These data are routinely audited for errors. Elements of the 1999–2008 data file that were made available consisted of diagnostic and procedure codes as well as encrypted patient and surgeon identifiers.

Diagnostic and procedure codes were based on the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM). All patients with a primary diagnosis of prostate cancer (ICD-9-CM, code 185) were considered, and the prostatectomy procedure code (ICD-9-CM, 60.5) identified patients who underwent RP between 1999 and 2008. Using the minimally invasive modifier (ICD-9-CM, 54.21) 2046 patients who underwent minimally invasive RP (MIRP) were identified. Patients were categorized as having undergone MIRP or open RP. Patients undergoing a concomitant lymph node dissection were identified using the respective procedure codes (ICD-9-CM, 40.1, 40.29, 40.3, 40.50, 40.53, 40.59 and 40.9).

Patients identified as having VTE experienced a venous thrombosis and/or a pulmonary embolism, according to previously defined criteria [13]. Respective disease codes were ICD-9-CM 451.0, 451.1x, 451.2, 451.81, 451.9, 453.4x, 453.8, and 453.9 for venous thrombosis and ICD-9-CM 415.x for pulmonary embolism.

The ASC represents the number of procedures done by each participating surgeon during each study calendar year. Patients were divided according to ASC tertiles, including low (≤10) intermediate (11–23) and high (≥24) RPs. Other covariates consisted of year of surgery (continuously coded), age (continuously coded), race (white vs black vs Hispanic vs other, including American Indian, Asian or Pacific Islander, and patients in whom information about race was missing), baseline Charlson Comorbidity Index (CCI). CCI was derived from ICD-9 codes according to previously established criteria and was stratified into two levels: 0 and ≥1 [14].

In-hospital mortality was derived from vital status. Length of stay (LOS) represents the number of days elapsed from the admission date to the discharge date. Same-day stays, coded as 0, were excluded from current analysis (n= 168). Total hospital charges are shown as the total undiscounted amount of dollars charged to patients for all services rendered at the hospital from admission to discharge, referred to as charges.

In the statistical analysis, descriptive statistics focused on frequencies and proportions for categorical variables. Means, medians and interquartile ranges (IQRs) were reported for continuously coded variables. Chi-squared tests, independent t-tests, and Kruskal–Wallis tests were used to compare the statistical significance of differences within categorical and continuous variables.

We relied on multivariable logistic regression analyses to predict VTE according to ASC. Year of surgery, age, race, baseline CCI, lymph node dissection, ASC and surgical approach were considered as covariables. Finally, to adjust for clustering within surgeons, multivariable logistic regression models relied on generalized estimating equations [15]. All statistical analyses were performed using the R statistical package system (R Foundation for Statistical Computing, Vienna, Austria), with P < 0.05 (two-sided) considered to indicate statistical significance.

RESULTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. PATIENTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. ACKNOWLEDGEMENTS
  8. CONFLICT OF INTEREST
  9. REFERENCES

Within the Florida Inpatient Datafile, 36 699 individuals treated with RP were identified. Of those, 100 (0.3%) had VTE (Table 1). Of all patients with VTE, 61 had venous thrombosis (0.2%) and 53 had pulmonary embolism (0.2%). Fourteen patients with a venous thrombosis also developed a pulmonary embolism (23.0%). In analyses focusing on VTE rates stratified according to ASC tertiles, 50 (0.4%), 36 (0.3%) and 14 (0.1%) VTEs occurred in patients operated on by low-, intermediate- and high-ASC surgeons, respectively (P < 0.001, Fig. 1).

Table 1.  Demographic characteristics of patients with or without VTE after RP (Florida Inpatient Datafile, 1999–2008). Data are n (%) unless noted otherwise
 Entire cohortWithout VTEWith VTE P
(n= 36 699; 100.0%)(n= 36 599; 99.7%)(n= 100; 0.3%)
Age, years    
 Mean (median)61.7 (62.0)61.7 (62.0962.5 (62.0)0.26
 SD±7.2±7.2±7.2 
 IQR57–6757–6757–68 
Race    
 White26 733 (72.8)26 671 (72.9)62 (62.0)0.03
 Black4 538 (12.4)4 518 (12.3)20 (20.0) 
 Other5 428 (14.8)5 410 (14.8)18 (18.0) 
CCI    
 028 875 (78.7)28 809 (87.7)66 (66.0)0.002
 ≥17 824 (21.3)7 790 (21.3)34 (34.0) 
Pelvic lymph node dissection    
 No20 862 (56.8)20 811 (56.9)51 (51.0)0.24
 Yes15 837 (43.2)15 788 (43.1)49 (49.0) 
Surgical approach    
 Open34 117 (93.0)34 025 (93.0)92 (92.0)0.68
 Minimally invasive2 582 (7.0)2 574 (7.0)8 (8.0) 
ASC    
 Low (1–10)11 775 (32.1)11 725 (32.0)50 (50.0)<0.001
 Intermediate (11–23)12 731 (34.7)12 695 (34.7)36 (36.0) 
 High (≥24)12 193 (33.2)12 179 (33.3)14 (14.0) 
Death during hospitalization    
 No36 669 (99.9)36 575 (99.9)94 (94.0)<0.001
 Yes30 (0.1)24 (0.1)6 (6.0) 
LOS, days    
 Mean (median)2.9 (3.0)2.9 (3.0)11.7 (9.0)<0.001
 SD±2.0±1.9±9.7 
 IQR2–42–46–14.8 
Total hospital charges, US$    
 Mean (median)28 629 (24 985)28 494 (24 943)78 122 (51 571)<0.001
 SD±17 256±16 550±81 573 
 IQR18 205–34 35218 189–34 29136 989–90 756 
image

Figure 1. Rates of VTE in 36 699 patients undergoing RP, stratified according to ASC tertiles: low, ≤10 RPs; intermediate, 11–23 RPs; high, ≥24 RPs (Florida Inpatient Datafile, 1999–2008).

Download figure to PowerPoint

Patients with a VTE were more frequently black (20.0 vs 12.3%, P < 0.001) and were more often CCI ≥ 1 than others (34.0% vs 21.3%, P= 0.002, Table 1). Low-ASC surgeons more frequently operated on patients with VTE than others (50.0 vs 32.0, P < 0.001). The use of MIRP was equally distributed between the two groups (8.0 vs 7.0%, P= 0.68).

Patients operated on by high-ASC surgeons were younger (median age = 62 years) than patients operated on by intermediate-ASC (median age = 63 years) and low-ASC (median age = 62 years) surgeons (P < 0.001, Table 2). The proportion of patients with CCI = 0 was higher in patients operated on by high-ASC surgeons (82.1%) than in patients operated on by intermediate-ASC (77.4%) and low-ASC (76.6%, P < 0.001) surgeons. Patients operated on by high-ASC surgeons were more likely to be white (75.2%) than patients operated on by intermediate-ASC (71.3%) and low-ASC (72.1%) surgeons (P < 0.001). Moreover, in patients operated on by low-ASC surgeons, a PLND was more frequently omitted (77.3%) than in patients operated on by intermediate-ASC (47.9%) and low-ASC (45.3%) surgeons (P < 0.001). Finally, patients operated on by high-ASC surgeons were more likely to undergo MIRP (15%) than patients operated on by intermediate-ASC (4.3%) and low-ASC (1.7%) surgeons.

Table 2.  Demographic characteristics of patients stratified according to ASC tertiles (Florida Inpatient Datafile, 1999–2008). Data are n (%) unless noted otherwise
 Low ASCIntermediate ASCHigh ASC P
(n= 11 775; 32.1%)(n= 12 731; 34.7%)(n= 12 193; 33.2%)
  • *

    P value derived from Kruskal–Wallis test.

Age, years    
 Mean (median)61.7 (62.0)62.1 (63.0)61.4 (62.0)<0.001*
 SD±7.2±7.1±7.4 
 IQR57–6757–6756–67 
Race    
 White8 484 (72.1)9 079 (71.3)9 170 (75.2)<0.001
 Black1 664 (14.1)1 804 (14.2)1 070 (8.8) 
 Other1 627 (13.8)1 848 (14.5)1 953 (16.0) 
CCI    
 09 015 (76.6)9 853 (77.4)10 007 (82.1)<0.001
 ≥12 760 (23.4)2 878 (22.6)2 186 (17.9) 
Pelvic lymph node dissection    
 No5 338 (45.3)6 096 (47.9)9 428 (77.3)<0.001
 Yes6 437 (54.7)6 635 (52.1)2 765 (22.7) 
Surgical approach    
 Open11 579 (98.3)12 179 (95.7)10 359 (85.0)<0.001
 Minimally invasive196 (1.7)552 (4.3)1 834 (15.0) 

A total of 30 patients died during hospitalization (0.1%, Table 1). Of those, six deaths occurred in 100 patients with VTE (6%) compared with 24 in others (0.1%, P < 0.001). LOS was longer in patients with VTE than in others (median: 9 vs 3 days, P < 0.001). Finally, total hospital charges were also higher in patients with VTE than in others (median: $51 571 vs $24 943, P < 0.001).

Using multivariable analyses, we examined the predictors of VTE after RP (Table 3). Here, relative to patients operated on by high-ASC surgeons, patients operated on by low-ASC surgeons were at higher risk of VTE (odds ratio [OR]= 3.78, P < 0.001). Additionally, black patients were more likely than others to experience a VTE (OR = 1.80, P= 0.023). Patients with CCI ≥ 1 were also more likely to experience a VTE than others (OR = 1.65, P= 0.016). Conversely, patients who underwent MIRP were not more likely to experience a VTE than patients who underwent open RP (OR = 1.97, P= 0.086).

Table 3.  Multivariable logistic regression analysis after fitting with generalized estimating equation for prediction of VTE (Florida Inpatient Datafile, 1999–2008)
VariablesVTE P
OR (95% CI)
ASC  
 High (≥24)Ref. 
 Intermediate (11–23)1.52 (0.97–2.36)0.066
 Low (1–10)3.78 (1.87–7.62)<0.001
Year of surgery1.01 (0.92–1.10)0.92
Age, years1.02 (0.99–1.05)0.19
Race  
 WhiteRef. 
 Black1.80 (1.09–2.98)0.023
 Other1.41 (0.85–2.33)0.19
CCI  
 0Ref. 
 ≥11.65 (1.10–2.49)0.016
Pelvic lymph node dissection  
 NoRef. 
 Yes1.07 (0.67–1.69)0.78
Surgical approach  
 OpenRef.0.086
 Minimally invasive1.97 (0.91–4.27) 

DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. PATIENTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. ACKNOWLEDGEMENTS
  8. CONFLICT OF INTEREST
  9. REFERENCES

The objective of the current analysis was to examine contemporary VTE rates and their association with ASC, as well as the link between VTE and mortality, LOS and hospital charges. We showed that the overall VTE rate was 0.3%. The VTE rates were 0.1%, 0.3% and 0.4% for high, intermediate- and low-ASC tertiles, respectively. The effect of ASC remained statistically significant in adjusted analyses. Here, patients operated on by low-ASC surgeons were at a nearly fourfold higher risk of VTE than patients operated on by high-ASC surgeons. Finally, VTE was associated with higher in-hospital mortality, prolonged LOS and increased hospital charges.

To the best of our knowledge this is the first population-based report that examines VTE in the context of RP and identifies predisposing factors. In addition, the sample size of the examined cohort allows a more detailed examination of VTE occurrences than was achieved in other studies. The VTE rates reported in the present study closely reflect those reported by others (0.4–2.4%), which validate the concept of the current analysis [6,7,16–20]

Annual surgical caseload is known to play an important role in all RP outcomes [11]. In the present study, we validated the practice-makes-perfect concept, as ASC was closely linked to VTE. Nilsson et al.[21] showed in adjusted analyses that patients operated on for colorectal cancer by very high-ASC surgeons were 60% less likely to experience a VTE than those operated on by low-ASC surgeons (P= 0.003). While the Nilsson et al.[21] study is not directly comparable to the present study, we corroborate the practice-makes-perfect hypothesis at RP when a rare but potentially life-threatening complication is considered to be the main outcome.

We demonstrated important differences in patient characteristics according to ASC tertiles. Specifically, patients operated on by low-ASC surgeons were, on average, younger, healthier according to CCI = 0, had lower PLND rates and were more frequently treated with MIRP. These findings emphasize the need for meticulous adjustment of these important differences. Lack of adjustment for these variables could lead to confounded results. To address and control for the important differences between ASC tertiles, we relied on multivariable logistic regression models additionally fitted with generalized estimating equations for clustering within surgeons. However, despite this adjustment, ASC remained an independent predictor of VTE. Explanations for this finding include previously demonstrated VTE risk factors such as longer operation duration, use of general anaesthesia and raised perioperative blood transfusion rates in low-ASC surgeons [7,22,23]. Due to the limitations of our database, we could not examine these variables. Another explanation could be that high-ASC surgeons are more often affiliated to academic centres. It is possible that the sub-speciality practice profile at these institutions improves outcomes. It is also conceivable that certain processes of academic institutions improve outcomes. These include the peer-review process, peri-operative VTE prophylaxis, initial diagnostic tests, preoperative tests, intra-operative care protocols and postoperative care pathways. However, the observed inverse caseload–VTE relationship is even more worrying given the fact that 80% of RP surgeons in the United States perform fewer than 10 procedures per year [24].

Furthermore, we corroborated the link between VTE and mortality, as well as the association between VTE on prolonged LOS and higher hospital charges, which validates the importance of this concept. When mortality rates were examined, we recorded a 60-fold higher rate in patients with VTE than in those without (0.1% vs 6%). This finding clearly indicates that, even if the overall VTE rate is low, this complication is nonetheless a life-threatening event. The results of the present study corroborate previous findings derived from institutional data that VTE after RP is associated with higher mortality [25,26]. In one population-based report, Alibhai et al.[4] demonstrated that 19% of all deaths after RP (n= 53) were attributable to pulmonary embolisms.

Several other important findings of our study should be mentioned. First, we observed racial differences with regard to VTE rates. In particular, individuals with a VTE were more frequently of black rather than of white race (20.0% vs 12.3%, P < .001). These differences remained significant in multivariable analysis, where black men were 79% more likely to succumb to a VTE than their white counterparts (P= 0.02). A possible explanation is that the higher proportion of more advanced prostate cancers amongst black patients accounts for this observation [27]. Unfortunately, we could not adjust for disease characteristics due to limitations of our database. Moreover, genetic factors in African American men have been shown to be associated with an increased risk of VTE [28].

With regard to the effect of MIRP on VTE, previous authors hypothesized that laparoscopy contributed to an inherently higher risk of VTE, due to peritoneal insufflation, reverse Trendelenburg position and prolonged operation duration [29,30]. Nonetheless, we could not detect statistically significantly differences between patients who underwent MIRP and others. This observation was corroborated in multivariable analyses, where patients who underwent MIRP were not more likely to develop a VTE than patients operated on using an open approach (OR = 2.17, P= 0.06). This finding supports observations made by other investigators, where vascular complications rates did not differ between surgical approaches [3,31]. However, vascular complications include VTE, as well as arterial thromboembolic events and cerebrovascular complications. Since the aetiologies of some arterial vascular complications are different to those occurring in the venous part of the circulation system, the reported rates in these studies are not truly reflective of venous thrombosis and/or pulmonary embolism. A recent population-based report examined in-hospital VTE rates derived from the Nationwide Inpatient Sample in patients undergoing open or laparoscopic colorectal surgery, and demonstrated a benefit of the laparoscopic (0.8%) over the open approach (1.4%, P < 0.001) [32]. A pure laparoscopic group, inclusion of inflammatory diseases and malignant tumours of the gastrointestinal tract, as well as the lack of adjustment for surgical or hospital caseload might explain the observed differences. Further studies are needed to examine the role of surgical approach with respect to VTE rates after RP.

Despite its strengths, the present study is not devoid of limitations. The use of population-based data sets such as the Florida Hospital In-patient Datafile has several limitations. The most important of these is that no follow-up after patient discharge was available. The risk of developing a VTE persists for up to 12 weeks after surgery [8,33]. Accordingly, a VTE occurring after hospital discharge, and leading to another hospital admission or ambulant treatment, was not captured by our database.

Another limitation is that the datafile did not allow us to adjust for absence of VTE prophylaxis, prior history of VTE or other VTE risk factors such as tumour stage or grade. Specifically, less advanced prostate cancers and thereby lower VTE rates might have been more prevalent among patients operated on by high-ASC surgeons due to better patient selection. Moreover, the current guidelines for VTE prophylaxis at RP are of a wide scope, and thus the inconsistent use of heparins, sequential compression devices or early ambulation could have affected the results. That VTE was ascertained using ICD-9 codes is yet another limitation. A 10% underestimation of VTE compared with medical records is reported for administrative data [34]. Unless no regional variability in VTE instances within the United States is reported, the generalizability of the Florida Hospital Inpatient Datafile can also be questioned [35]; and although we noted that ASC is important, procedures other than RP or clinical pathways at institutions might be related to individual surgeon complication rates. Finally, since the extent of PLND is known to be associated with VTE, the lack of such information in the current study might have biased the results [36].

In conclusion, RP by high-ASC surgeons exerts a protective effect on the likelihood of VTE. Additionally, VTE increases in-hospital mortality and is associated with prolonged LOS and higher hospital charges. From a practical perspective, the results of the present study indicate that, on average, lower rates of VTE should be expected in patients treated by high-ASC surgeons. Our findings suggest that VTE-specific processes of care need to be improved, with the intention of reaching the level recorded in patients treated by high-ASC surgeons.

ACKNOWLEDGEMENTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. PATIENTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. ACKNOWLEDGEMENTS
  8. CONFLICT OF INTEREST
  9. REFERENCES

Pierre I. Karakiewicz is partially supported by the University of Montreal Health Centre Urology Specialists, Fonds de la Recherche en Santé du Quebec, the University of Montreal Department of Surgery and the University of Montreal Health Centre (CHUM) Foundation.

Jan Schmitges is supported by the German Federal Ministry of Education and Science in the framework of the program for medical genome research FKZ:01GS08189.

REFERENCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. PATIENTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. ACKNOWLEDGEMENTS
  8. CONFLICT OF INTEREST
  9. REFERENCES