SEARCH

SEARCH BY CITATION

Keywords:

  • cystectomy;
  • mortality;
  • obesity;
  • propensity score;
  • surgical outcomes

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgment
  8. Conflict of interest
  9. References

Objective

To examine the impact of body mass index, as a measure of obesity, on the surgical outcomes of cystectomy.

Methods

The American College of Surgeons National Surgical Quality Improvement Program database was used to acquire data on 1293 cystectomies carried out from 2005 to 2011. Patients were divided into two groups: body mass index <30 kg/m2 and ≥30 kg/m2. A propensity score-matched analysis of perioperative outcomes was carried out.

Results

A total of 869 patients had a body mass index <30, whereas 424 had a body mass index ≥30. Unadjusted comparisons showed higher rates of superficial surgical site infections (8.7% vs 5.3%, P = 0.04), renal insufficiency (4.0% vs 1.7%, P = 0.01) and increased operative times (365.7 min vs 338.6 min, P = 0.0004) in the obese patients, but interestingly lower rates of pneumonia (2.4% vs 4.8, P = 0.03) and cerebral vascular accidents (0.0% vs 0.9%, P = 0.05). However, the latter two observations might be explained by more tobacco use among non-obese patients (26.6 mean pack-years vs 20.0 mean pack-years, P = 0.004). Notably, no differences in 30-day mortality were noted. After adjusting for preoperative demographic and clinical data using propensity score-matching methods, there were no observed differences between the two cohorts except for operative time (P = 0.04).

Conclusions

Obesity is not independently associated with an increased risk of perioperative complications or 30-day mortality after cystectomy.


Abbreviations & Acronyms
ASA

American Society of Anesthesiologists

BMI

body mass index

CABG

coronary artery bypass grafting

COPD

chronic obstructive pulmonary disease

CVA

cerebral vascular accidents

ETOH

alcohol

HTN

hypertension

NSIQP

National Surgical Quality Improvement Program

PCI

percutaneous coronary intervention

PGY

postgraduate year

Preop.

preoperative

PS

propensity score

SSI

space surgical site infection

Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgment
  8. Conflict of interest
  9. References

The obesity epidemic has become one of the foremost public health issues in the USA. Obese patients (defined by BMI ≥30 kg/m2) have higher rates of cardiovascular, respiratory and endocrine comorbidities, which have been linked to increased perioperative morbidity in patients undergoing non-bariatric abdominal surgery.[1]

Previous studies evaluating the impact of obesity on surgical outcomes after cystectomy, in contrast, have shown mixed results. Obesity has been linked to worse oncological outcomes (inferior lymphadenectomies, higher positive margin rates, etc.), which resulted in higher tumor recurrence rates, and detrimental effects on cancer specific and overall survival.[2] Other published reports, however, showed no significant effect on the survival outcomes in obese patients.[3] Furthermore, although some studies show significant differences in perioperative morbidity,[4] others show no difference in perioperative complications, even among laparoscopic cohorts.[5-7]

These studies, however, are generally all underscored by one or more methodological limitations (retrospective design, small sample sizes, single-center experiences, etc.), which could partially explain the inconsistent findings. No population-based data evaluating the perioperative morbidity and mortality outcomes for obese patients undergoing cystectomy currently exist. The present study attempted to more clearly define the perioperative risk faced by obese patients undergoing cystectomy using a PS-matched analysis. A thorough understanding of this risk is vital to the informed consent of obese patients preparing for cystectomy.

Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgment
  8. Conflict of interest
  9. References

Data

Data from the American College of Surgeons NSQIP participant-use file were used for the present study. The general methods of NSQIP have previously been described in detail.[8-10] In brief, NSQIP collects clinical data on patients undergoing major surgical procedures at more than 200 hospitals encompassing more than 130 variables, including preoperative demographic and comorbidity data, intraoperative and perioperative complications, and mortality outcomes for 30 days after the operation.[11] Trained surgical staff reviewers record the data using standardized definitions. The accuracy of the data is ensured through vigorous quality control measures, including intensive training for data collectors, and by carrying out interrater reliability audits of participating sites.[12]

Study population

Patients were included in the study if they had undergone a cystectomy between 2005 and 2011 as determined by the following Common Procedural Terminology codes listed for the principle operative procedure: 51570, 51575, 51580, 51585, 51590, 51595 and 51596. Patients undergoing partial cystectomy were excluded. Patients were divided into two groups: those with BMI <30 and those with BMI ≥30.

Functional health status

An independent patient was defined by the ability to carry out all activities of daily living without assistance from another person, whereas a partially dependent patient requires some assistance from another person for activities of daily living. Likewise, total dependence was defined as the need for complete assistance with all activities of daily living.[12]

Complications

The 30-day outcomes analyzed included mortality, wound events (superficial, deep or organ/space SSI), development of sepsis/shock, pulmonary events (pneumonia, ventilation >48 h and/or unplanned intubation), renal failure/insufficiency rates, thromboembolic (deep venous thrombosis or pulmonary embolism) and cardiovascular events (stroke with deficit, cardiac arrest or infarct). Hospital length of stay, rates of return to operating suite, total operative time and total blood transfusions were also analyzed. A glossary of definitions for how these complications are defined have been previously published.[13]

Statistical analysis

Differences in patient demographics and clinical characteristics were evaluated using the χ2-test or the Fisher's exact test for categorical data and Student's t-test for continuous variables. All continuous variables were checked for normality using ladder of power plots. Variables that violated the normality assumption were examined using the Mann–Whitney U-test. In an effort to make causal inferences regarding the impact of obesity on the perioperative outcomes, we matched a set of patients with a BMI ≥30 to a control group of patients with a BMI <30 using PS methods with nearest neighbor matching. The PS is the probability of being obese as a function of patient demographic and clinical characteristics, which is obtained using logistic regression assuming common support.[14] All statistical calculations were carried out using Stata/SE v 11.0 (StataCorp, College Station, TX, USA).

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgment
  8. Conflict of interest
  9. References

Patient demographics and clinical characteristics

The study included a total of 1293 patients who underwent a cystectomy from 2005 to 2011 at participating NSQIP institutions. A total of 869 patients had a BMI of <30, whereas 424 had a BMI ≥30. Before PS matching, the obese population was older (P < 0.0001), had higher ASA classifications (P = 0.002), had a higher rate of hypertension requiring medication (P < 0.0001) and had higher rates of diabetes (P < 0.0001), as expected (Table 1). Tobacco pack-year history (P = 0.004) and heavy alcohol use (defined as >2 drinks per day, P = 0.019) were higher among the non-obese population. Given these marked differences in baseline demographic and clinical data between the two populations, we carried out a PS-matched analysis to account for selection bias incumbent to non-randomized comparisons. After PS matching, 457 non-obese patients were matched to 121 obese patients with no significant differences between baseline demographic and clinical characteristics except for mean BMI, as expected (34.6 kg/m2 vs 25.2 kg/m2, P < 0.0001).

Table 1. Patient demographics and clinical characteristics
 Before propensity matchingP-valueAfter propensity matchingP-value
BMI <30 (n = 869)BMI ≥30 (n = 424)BMI <30 (n = 457)BMI ≥30 (n = 121)
n%n%n%n%
  1. a

    Mean.

Sex (% male)68273.632275.90.2735878.39477.70.88
Race  0.10    0.45
White (including Hispanic origin)72084.236888.7 38283.610990.1 
Black303.5163.9 132.832.5 
Asian60.741.0 51.110.8 
Other40.500 20.400 
Unknown9511.1276.5 5512.086.6 
Diversion type    0.44    0.81
Ileal conduit (51590–95)64974.731273.6 34274.88771.9 
Neobladder (51596)17019.69321.9 9420.62823.1 
Other/unspecified (51570-85)465.2194.5 214.664.9 
Diabetes13315.311326.7<0.00017316.02722.30.10
Tobacco use          
Average pack yearsa26.620.00.00420.520.10.93
ETOH >2 drinks/day314.961.90.01993.734.50.77
Functional Status    0.58    0.51
Independent83796.540896.2 43896.111897.5 
Partially dependent263.0122.8 163.521.7 
Totally dependent40.541.0 20.410.8 
COPD748.5378.70.90357.7108.30.83
Heart disease          
Prior PCI599.4299.00.86229.123.10.11
Prior CABG497.83310.30.18156.2710.80.20
HTN requiring meds46353.329870.3<0.000125154.97461.20.22
Chronic steroid use263.0112.60.69153.354.10.65
Preop chemo within 30 days487.6288.70.562911.91218.50.17
Preop labs          
Creatininea1.21.20.571.21.10.44
Albumina3.94.00.0473.93.90.86
Resident involvement    0.98    0.20
No resident15429.27628.5 2917.5512.2 
PGY-3 or lower6311.93613.5 2816.9714.3 
PGY-46712.73512.1 1911.51428.6 
PGY-511521.85621.0 4225.3816.3 
PGY-6 or higher12924.46424.0 4828.91428.6 
ASA classification    0.002    0.41
I101.220.5 51.100.0 
II25429.28620.3 13028.53327.3 
III56064.431674.5 30065.68570.3 
IV455.2204.7 224.832.5 
Agea68.665.6<0.000168.167.10.12

Wound complications

Before PS matching, the rates of wound complications were higher in the obese patients (17.5% vs 13.1%, P = 0.04, Table 2). This included superficial infections and deep SSI, organ/space SSI (e.g. abscesses), and wound dehiscence. When stratified by wound infection type, the differences appeared to be predominantly from superficial SSI (5.3% vs 8.7%, P = 0.018). In the PS-matched cohort, however, the wound infection rates were not different (P = 0.17).

Table 2. Intraoperative and postoperative events
 Before propensity matchingP-valueAfter propensity matchingP-value
BMI <30 (n = 869)BMI ≥30 (n = 424)BMI <30 (n = 457)BMI ≥30 (n = 121)
n%n%n%n%
  1. †If patients have more than one complication per organ system, they are only included once in the total calculation. Therefore, composite complications totals are not necessarily simply a sum of the individual events. ‡Mean.

Total wound complications11413.17417.50.046714.72419.80.17
Superficial SSI465.3378.70.018286.1108.30.40
Deep SSI131.5133.10.0671.532.50.48
Organ space SSI414.7194.50.85235.086.60.49
Dehiscence283.2163.80.61163.565.00.46
Total pulmonary complications768.8276.40.14439.454.10.06
Pneumonia424.8102.40.03255.521.70.07
Pulmonary embolism273.1153.50.68163.532.50.58
Fail to wean vent within 48 h252.992.10.43122.621.70.54
Total renal complications262.9245.70.01143.186.60.07
Renal insufficiency151.7174.00.0192.054.10.17
Acute renal failure131.581.90.6051.132.50.25
Total neurological complications101.2000.0371.5000.17
Cerebral vascular accident80.9000.0551.1000.25
Peripheral nerve injury20.2000.3220.4000.47
Total cardiovascular events485.5276.40.54265.732.50.15
Cardiac arrest60.761.40.2051.110.80.8
Myocardial infarction111.361.40.8351.1000.25
Deep venous thrombosis364.1184.30.93194.221.70.19
Urinary tract infections8810.15212.30.25429.21714.10.12
Total blood transfusions30735.313431.60.1921847.75041.30.21
Sepsis episodes9010.44711.10.69408.81512.4.22
Return to operating room536.1215.00.41337.275.80.58
30-day mortality192.2112.60.65102.2000.10
Days from surgery to discharge10.110.20.8310.110.20.94
Operative time (min)338.6365.70.0004348.7373.60.04

Respiratory complications

Composite respiratory complications were similar between groups before and after PS matching; however, there was a trend toward higher respiratory complications in the non-obese population after matching (9.4% vs 4.1%, P = 0.06). Pneumonia appears to be more common in the non-obese patients before matching, but this might be explained by the higher tobacco pack-year history in this population. Nevertheless, the differences in the rates of pneumonia between the groups were non-significant after PS matching (Table 2).

Renal complications

Composite renal complications were higher in the obese population before PS matching (2.9% vs 5.7%, P = 0.01); however, this difference lost its statistical significance after matching (P = 0.07). Renal insufficiency, which is defined as a rise in the serum creatinine of >2 mg/dL from the preoperative value without requirement for dialysis, was higher in the obese population before PS matching (Table 2). The rate of hypertension requiring medications was also much higher in the obese population before PS matching (70.3% vs 53.3%), which could partially account for the higher rates of renal complications in the obese cohort before PS matching.

Neurological complications

Rates of CVA, albeit low for the entire cohort, were higher in the non-obese cohort (0.9% vs 0%, P = 0.05); however, this difference was non-significant after PS matching. Notably, rates of preoperative CVA were not statistically significant between groups either before or after PS matching (before matching: 1.4% [non-obese] vs 0.9% [obese], P = 0.31).

Other complications and operative time

Rates of cardiac events, urinary tract infections, total blood transfusions within 72 h after surgery, sepsis episodes, rates of returning to the operating room within 30 days, length of hospitalization, and 30-day mortality were similar between obese and non-obese patients before and after PS matching. Operative time was significantly higher in the obese groups before and after PS matching (Table 2).

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgment
  8. Conflict of interest
  9. References

The effect of obesity on surgical outcomes after radical cystectomy has not been clearly defined in the literature. Obesity would seem to be an obvious risk factor for perioperative morbidity and mortality. However, only a few studies have shown the detrimental effects of obesity on complication rates and surgical outcomes.[2, 4] These studies are countered by reports showing no differences in surgical outcomes when stratified by obesity.[3, 5-7] These studies, however, are all limited by one or more major methodological limitation (retrospective design, small sample sizes, single-center experiences, etc.), which hinders one's ability to make causal inferences about the effect of obesity on surgical outcomes without significant statistical error. Given that randomization is not a practical method of reducing selection bias in this setting, we explored this question of whether obesity itself is directly related to surgical outcomes through PS-matching methods. PS matching estimates each individual's propensity to be obese as a function of preoperative demographic and clinical data, and matches individuals with similar propensities.[14, 15] In this way, we matched two cohorts that were equal in every way with regard to known preoperative clinical and demographic variables, except BMI status, and compared the 30-day complication and mortality rates.

Using a large population-based database (NSQIP), the present study showed significant increases in the rates of wound complications, renal complications and operative times among the obese cohort before PS matching. Aside from slightly higher rates of pneumonia and CVA in the non-obese cohort (which could be partially explained by the higher rates of tobacco use in this group), these findings would, at first glance, implicate obesity as a potential predictor of poor perioperative and postoperative outcomes. However, after accounting for the selection bias that is incumbent to non-randomized observational data by matching a set of obese patients to a set of non-obese patients of similar propensity scores, we observe that many of these differences become non-significant. There was still a non-significant trend toward increased pneumonia rates in the non-obese patients, and a non-significant trend toward increased rates of renal insufficiency in the obese cohort after matching. These data argue that obesity itself might not actually be a predictor of surgical complications after cystectomy, as previously suggested, but rather a marker for preoperative medical comorbidity that predisposes to surgical complications. This assertion is supported by the lack of statistically significant differences in postoperative complication rates after the differences in preoperative clinical and demographic data are accounted for.

These data should be considered in the context of certain limitations. First, the matched cohort (n = 578) was relatively small by the standards of population-based studies. This could introduce statistical error into the comparisons of infrequent surgical complications (e.g. CVA). Second, the short timeframe (30 days) provided by NSQIP made it impossible to ascertain whether these results are durable over longer-term follow up. Third, because of the American College of Surgeons policy to maintain confidentiality of data-reporting institutions, we cannot account for differences in surgeon and hospital volume, which could possibly influence surgical outcomes. Fourth, no data on stoma-related complications or urinary continence after orthotopic diversion are available in the NSQIP dataset. These would be potentially informative datapoints in a study of obese patients undergoing cystectomy. Finally, because self-selected institutions contribute patient data, the surgical population captured by NSQIP might not be fully representative of the USA cystectomy population. However, the data were collected prospectively from several different types of hospital systems, which provided a large and diverse sample size that roughly parallels the race and sex distributions of the USA.[16] Furthermore, the data are collected in a standardized manner at each site with strict variable definitions with annual quality checks. The database has been validated for accuracy and reproducibility, and achieves a >95% 30-day outcome follow-up rate.[16]

Despite these limitations, we believe that our claim that obesity is not associated with perioperative complications or 30-day mortality after cystectomy is valid. Obese patients do have longer operative times, which might reflect increased surgical difficulty. These data are vital to the informed consent of obese patients preparing for cystectomy.

Acknowledgment

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgment
  8. Conflict of interest
  9. References

Mark Tyson had full access to all of the data in the study, and takes responsibility for the integrity of the data and the accuracy of the data analysis.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgment
  8. Conflict of interest
  9. References