Estimation of Risk in Cancer Patients Undergoing Palliative Procedures by the American College of Surgeons Risk Calculator

Abstract

Background:

Surgical palliation is defined as the use of a procedure to relieve symptoms. The American College of Surgeons Risk Calculator (ACSRC) was created based on data from the National Surgical Quality Improvement Program to predict the risk of surgical complications on a patient-specific level. Whether the ACSRC can accurately predict the risk of postoperative complications following palliative procedures in cancer patients is unknown. The purpose of this study was to determine if the ACSRC accurately predicted postoperative complication rates in this setting.

Study Design:

A prospectively maintained, surgical oncology database of patient outcomes from 2011 to 2013 was queried. Data extracted included the following: demographics, comorbidities, site and stage of cancer, type of procedure, and post-operative complication rate and type. Risk assessment was performed for each patient using the ACSRC. Predicted outcomes were compared to actual outcomes for length of stay (LOS), complications, and death. Main outcome measures were differences in actual versus predicted outcomes.

Results:

Thirty-two patients were included. Occurrence of any complication was significantly lower than predicted (31% vs. 59%, p < 0.05). The predicted LOS, however, was 2.9 days; significantly lower than 5.4 days (p < 0.05).

Conclusion:

The ACSRC is a powerful tool for aid in surgical decision-making; however, in the case of palliative procedures for cancer patients, it overestimated the risk of postoperative complications and underestimated the LOS. Overestimation of post-operative complications could result in fewer patients being offered potentially beneficial palliative procedures.

Introduction

For many surgeons, the decision of when to operate in a setting of advanced cancer without a curative intent is often challenging. It is not surprising that we are seeing an increase in the need for discussions regarding the utility of surgical palliation.¹ Surgical palliation is the deliberate use of a procedure in a patient with incurable disease with the intention of relieving symptoms, minimizing patient distress, and improving quality of life.²

As physicians and patients work toward a more shared decision-making (SDM) model, both parties struggle to reliably assess the risk of possible procedures, including those with palliative intent. To make optimal decisions regarding whether or not to undergo a palliative procedure, both patients and physicians require reliable predictors of treatment-related complications. The American College of Surgeons (ACS) developed a risk calculator (ACSRC) based upon data from the National Surgical Quality Improvement Program database in an effort to create an accessible predictive tool for post-operative complications. The ACSRC estimates the risk for post-operative complications, including nine specific complications such as pneumonia, return to the operating room, or death within the 30-day post-operative period. It also estimates the length of stay (LOS) required to recover from the planned procedure.³

To our knowledge, the utility of the ACSRC as a resource for patients with cancer has not been validated for aid in the decision of whether to have a palliative procedure.^4,5 The purpose of this study was to determine if the ACSRC accurately predicted the risk of complications and LOS for patients undergoing palliative surgical interventions for cancer in a tertiary care cancer center.

Materials and Methods

After obtaining IRB approval, a prospectively maintained database of patients treated by the Surgical Oncology Division of the University of New Mexico between 2011 and 2013 was queried. A subset of patients who underwent surgical procedures with palliative intent was identified. All patients who received a palliative operation were included regardless of type of tumor or palliative procedure performed. Exclusion criteria included the following: patients lost to follow-up and those with incomplete clinical data. A total of 32 patients met inclusion criteria.

Data extracted included the following: demographics, date of admission, comorbidities, type of malignancy, cancer-directed therapies received before the procedure, palliative procedure performed, presence and type of any postoperative complication within 30 days of surgery, and length of hospital stay following the procedure.

Risk calculation was performed by inputting each patient's variables into the online ACSRC with the “Surgeon Adjustment of Risks” set at “no adjustment necessary.”⁶ All ACSRC outcomes were recorded. We then compared results from the ACSRC to outcomes observed complications in our cohort.

Descriptive statistics were used to calculate frequencies, means and standard deviations of study variables. Continuous variables were analyzed using Student's t-tests. Data analysis was performed with SPSS™ Version 21.0 (IBM Corp., Armonk, NY).

Results

Demographics

The demographic and clinical characteristics of the study population are summarized in Tables 1 and 2. Over 70% of the patients were women. Non-Hispanic Whites accounted for the majority of patients in this series. Median follow-up for this study was four months.

Table 1.

Demographic Characteristics

	N (%)	Age, years, median (range)
Total	32	61 (30–81)
Sex
Female	23 (72)	61 (32–81)
Male	9 (18)	61 (30–81)
Ethnicity
Non-Hispanic White	14 (44)	63 (30–79)
Hispanic	12 (38)	61 (43–81)
Native American	4 (12)	49 (32–59)
Other	2 (6)	63 (63)

Table 2.

Clinical Characteristics

Clinical variable	N (%)
Comorbidities^a
Hypertension	10 (31)
CAD	3 (9)
Diabetes Mellitus	7 (22)
Smoker	6 (18)
Pulmonary^b	4 (12)
Primary tumor site
Colorectal	13 (41)
Pancreas	7 (22)
Skin/soft tissue	3 (9)
Breast	2 (6)
Other	7 (21)
Tumor Stage at diagnosis
I	1 (3)
II	1 (3)
III	9 (30)
IV	19 (63)
Unknown	2 (6)
Therapy before surgery
Chemotherapy alone	2 (6)
Chemotherapy+radiation	9 (28)
None	21 (66)
Surgery performed
Hepaticojejunostomy	5 (16)
Colostomy	5 (16)
Lesion/mass excision	5 (16)
Lysis of adhesions	4 (12)
Other^c	13 (40)

Percentages provided are not cumulative since some patients presented with 2 or more comorbidities.

Includes the following: 2 COPD, 2 dyspnea.

Others include the following: 2 hernia repairs, 2 tumor debulking, 2 splenectomies, 1 complete axillary lymph node dissection, 1 cholecystectomy, 1 ileostomy, 1 hemicolectomy, 1 hepatectomy, 1 mastectomy, and 1 stent placement.

The most frequent site of primary cancer was colorectal followed by pancreas. The majority of patients had Stage IV disease. Approximately one-third of patients received chemotherapy or chemoradiation within a median of 90 days (range 12–592 days) before their palliative surgical procedure. Sixty-nine percent had one or more significant noncancer-related comorbidity. All procedures were performed under general anesthesia. Among the reviewed cases, 75% were performed within 24 hours of admission to the hospital due to severe symptoms affecting quality of life.

Post-operative and risk calculator outcomes

Risk calculator outcomes are summarized in Table 3. Approximately one-third of patients developed a post-operative complication within 30 days from surgery. The most frequent complications were cardiac events and urinary tract infections. The ACSRC predicted a post-operative complication rate almost double of what was observed (59% vs. 31%, p < 0.05). Furthermore, the risk calculator predicted that nearly all patients (94%) had an above-average risk of at least one post-operative complication. According to the ACSRC, the three most commonly predicted complications besides death were (in order of predicted frequency) the following: urinary tract infections, cardiac events, and pneumonia. In contrast to the 0% observed 30-day post-operative mortality, the predicted post-operative mortality was 69%.

Table 3.

Risk Calculator Results

Complication	Below-average risk N (%)	Average risk N (%)	Above-average risk N (%)	Observed N (%)	p^a
Any complication	13	0	19 (59)	10 (31)^b	<0.05
Pneumonia	14	3	15 (47)	0 (0)	<0.001
Cardiac event	13	0	19 (59)	7 (21)	<0.05
Wound infection	15	8	9 (28)	0 (0)	<0.05
Urinary tract infection	10	0	22 (69)	3 (9)	<0.001
Deep venous thrombosis	10	6	16 (50)	0 (0)	<0.001
Renal failure	15	0	17 (53)	0 (0)	<0.001
Reoperation	12	10	10 (31)	0 (0)	<0.001
Death	10	0	22 (69)	0 (0)	<0.001

Comparison between observed outcomes and above-average risk.

Two patients presented with two complications.

In addition to postoperative complication risk, the ACSRC also predicts the average length of hospital stay for a given procedure. The average length of in-hospital post-operative stay estimated by the ACSRC was significantly shorter compared to the observed average LOS (2.9 days vs. 5.4 days, respectively, p < 0.05).

Discussion

Advanced stages of cancer are frequently associated with multiple symptoms, which decrease quality of life. It is in this setting that patients and their families often face the decision to undergo palliative surgical procedures to alleviate symptoms and improve their quality of life. These decisions are challenging for surgeons, patients, and families. A tool able to predict the risk of complications after surgical palliation would aid in informed and SDM for both patients and healthcare providers.

The main finding of this study was that the ACSRC overestimated the risk of post-operative complications in patients undergoing surgical palliation for cancer by nearly twofold. The ACSRC predicted that roughly 60% of this cohort would experience some form of complication, whereas less than one-third of patients actually developed a post-operative complication. The majority of these patients (69%) were also predicted to be at an increased risk of death within 30 days post-surgery, while the observed 30-day mortality was 0. Secondarily, we found that the ACSRC underestimated the length of patient in-hospital stay after surgical palliation for our patients, with the average LOS predicted by the ACSRC being approximately half of the actual LOS.

SDM is a key factor in the improvement of quality of care and a top-priority for cancer care.^7,8 In this setting, however, there is currently no single tool to aid in the decision to pursue a palliative procedure. Roses et al. derived a scoring system, which they applied to 143 cancer patients undergoing emergent surgery.⁹ The scoring system showed prognostic value for overall survival, but limited accuracy in predicting post-operative morbidity. Another tool was developed by Tseng and associates to predict risk of 30-day morbidity and mortality in patients with disseminated cancer undergoing surgical intervention. The nomogram proved to be a good predictor of both morbidity and mortality.¹⁰ These tools, however, have limited utility in clinical practice, secondary to not being easily available to clinicians or patients and not being specific for surgical intent.^9,10

The results of our study suggest that the currently available risk calculators may not be sufficiently accurate for patients undergoing surgical palliation. We believe the reasons for discrepancy are most likely related to the absence of information regarding intent of the surgery (curative vs. palliative) in the patient series used to create the ACSRC. In addition, risk calculators based on large aggregate data can be less accurate when applied on the smaller scale of a single institution or physician.¹¹ The ACSRC also does not control for cancer patient-specific comorbidities such as chemotherapy or radiation, known to increase the risk of postoperative morbidity and mortality. These limitations, however, are currently being addressed to improve the ACSRC.⁴

This study has several limitations. The retrospective nature may introduce miscoding bias, such as identifying procedures that were initially intended to be curative and changed to palliative during the course of the procedure, although detailed chart review makes this less likely. In addition, the small number of subjects in our single-institution cohort limits the power of our statistical analysis and generalizability of our results, but also speaks to the fact that the ACSRC may not be a reasonable measure to compare actual outcomes from individual physicians due to the larger effect of outliers in smaller series as was described by Cologne et al.¹¹

The fact that the procedures were deemed palliative could have also influenced the surgeon's attitude during the surgical procedure, as to not taking a more aggressive approach, thus reducing procedures that could have increased post-operative morbidity and mortality. It is this potential confounder that speaks to the need for surgeons to classify their procedures as curative or palliative intent prospectively so as to better measure our true morbidity and mortality for palliative procedures. Furthermore, increased clarity on the part of the surgeon and patient regarding the goals of any surgical procedure and, if palliative, the symptoms we are seeking to improve upon, is likely to lead to improved SDM.

In conclusion, we found that the ACSRC overestimated the risk of postoperative morbidity and mortality following palliative surgical procedures in patients with cancer. Although the use of ACSRC may aid in SDM for multiple procedures, its overestimation of post-operative complications in a palliative setting could result in fewer patients being offered potentially beneficial procedures. The observed 30-day morbidity and mortality of 31% and 0%, respectively, indicates that patients with advanced malignancy can undergo palliative procedures with acceptable risk when carefully selected. These results require further validation in a larger cohort of advanced, symptomatic cancer patients.

Footnotes

Author Disclosure Statement

No competing financial interests exist.

References

Dunn

: Surgical palliative care: recent trends and developments. Anesthesiol Clin, 2012; 30:13–28.

Cady

, Easson

, Aboulafia

, Ferson

: Part 1: Surgical palliation of advanced illness—what's new, what's helpful. J Am Coll Surg, 2005; 200:115–127.

Bilimoria

, Liu

, Paruch

, et al.: Development and evaluation of the universal ACS NSQIP surgical risk calculator: A decision aid and informed consent tool for patients and surgeons. J Am Coll Surg, 2013; 217:833–842.e1–e3.

Paruch

, Ko

, Bilimoria

: An opportunity to improve informed consent and shared decision making: The role of the ACS NSQIP Surgical Risk Calculator in oncology. Ann Surg Oncol, 2014; 21:5–7.

Mohanty

, Liu

, Paruch

, et al.: Risk of discharge to postacute care: A patient-centered outcome for the American college of surgeons national surgical quality improvement program surgical risk calculator. JAMA Surg, 2015; 150:480–484.

ACS Risk Calculator 2015: http://riskcalculator.facs.org/ (Last accessed December 1, 2015).

Godolphin

: Shared decision-making. Healthc Q, 2009; 12:e186–e190.

Levit

, Balogh

, Nass

, Ganz

: Delivering High-Quality Cancer Care: Charting a New Course for a System in Crisis. National Academies Press, Washington, DC, 2013.

Roses

, Tzeng

, Ross

, et al.: The palliative index: Predicting outcomes of emergent surgery in patients with cancer. J Palliat Med, 2014; 17:37–42.

10.

Tseng

, Yang

, Wang

, et al.: Nomogram to predict risk of 30-day morbidity and mortality for patients with disseminated malignancy undergoing surgical intervention. Ann Surg, 2011; 254:333–338.

11.

Cologne

, Keller

, Liwanag

, et al.: Use of the American College of Surgeons NSQIP Surgical Risk Calculator for Laparoscopic Colectomy: How good is it and how can we improve it?. J Am Coll Surg, 2015; 220:281–286.