Header bg
  • Users Online: 262
  • Print this page
  • Email this page
Header bg

Table of Contents
Year : 2020  |  Volume : 4  |  Issue : 2  |  Page : 53-58

Evaluation of POSSUM scoring systems in predicting postoperative morbidity and mortality in indian patients operated for esophageal cancer

1 Department of Anaesthesiology, Himalayan Institute of Medical Sciences, Swami Rama Himalayan University, Dehradun, Uttarakhand, India
2 Department of Surgical Oncology, Himalayan Institute of Medical Sciences, Swami Rama Himalayan University, Dehradun, Uttarakhand, India
3 Department of BioStatistics, Himalayan Institute of Medical Sciences, Swami Rama Himalayan University, Dehradun, Uttarakhand, India

Date of Submission20-Feb-2025
Date of Decision05-Mar-2020
Date of Acceptance12-Mar-2020
Date of Web Publication11-May-2020

Correspondence Address:
Dr. Sunil Saini
Department Surgical Oncology, Cancer Research Institute, Himalayan Institute of Medical Sciences, Swami Rama Himalayan University, Dehradun - 248 140, Uttarakhand,
Login to access the Email id

Source of Support: None, Conflict of Interest: None

DOI: 10.4103/BJOA.BJOA_13_20

Rights and Permissions

Background: Surgical treatment for esophageal cancer is a high-risk procedure. Prediction of postoperative adverse events could aid in the stratification of patients, thus improving outcomes as well as achieving optimal use of resources. The Physiological and Operative Severity Score for the Enumeration of Mortality and Morbidity (POSSUM) is a prediction model that utilizes both physiological and surgical parameters to assess risk. This study evaluates the effectiveness of POSSUM, Portsmouth-POSSUM (P-POSSUM), and esophagogastric-POSSUM (O-POSSUM) scoring systems in predicting postoperative morbidity and mortality in Indian patients operated for esophageal cancers. Patients and Methods: It is a retrospective study conducted in a tertiary care teaching hospital with data collected from esophagectomies performed from January 2015 to January 2019. The calibration and discriminative abilities of the scores to predict 30-day morbidity and mortality were analyzed using the Hosmer–Lemeshow test, observed to predicted ratios (observed/expected [O/E]), and the receiver operating characteristic curve tests. Results: A total of sixty patients were included. The 30-day mortality and morbidity were 6.67% (4/60) and 46.66% (28/60), respectively. POSSUM morbidity showed proper calibration and discrimination (O/E: 0.86) with a modest predictive ability (area under the curve [AUC]: 0.701). While analyzing mortality, though all scores displayed good calibration, O-POSSUM displayed superior predictive ability (O/E: 1.02). The POSSUM score overpredicted mortality by nearly twice (O/E: 0.52), whereas P-POSSUM underpredicted it (O/E: 1.71). All scores showed moderate discrimination with P-POSSUM outperforming other tests (AUC: 0.825). Conclusions: The POSSUM scoring system was useful in predicting morbidity risk following esophageal resection for cancer, with O-POSSUM more accurate for mortality prediction in this group of patients.

Keywords: Esophageal cancer, mortality, morbidity, POSSUM score, receiver operating characteristic curve, scoring systems

How to cite this article:
Ramakrishnan P, Pattanayak M, Arora A, Singh A, Asthana V, Saini S. Evaluation of POSSUM scoring systems in predicting postoperative morbidity and mortality in indian patients operated for esophageal cancer. Bali J Anaesthesiol 2020;4:53-8

How to cite this URL:
Ramakrishnan P, Pattanayak M, Arora A, Singh A, Asthana V, Saini S. Evaluation of POSSUM scoring systems in predicting postoperative morbidity and mortality in indian patients operated for esophageal cancer. Bali J Anaesthesiol [serial online] 2020 [cited 2023 Mar 23];4:53-8. Available from: https://www.bjoaonline.com/text.asp?2020/4/2/53/284173

  Introduction Top

Cancer of the esophagus has traditionally been a difficult disease to treat, with surgery being the mainstay of curative treatment. Esophagectomies, which have been performed from as far back as the early 19th century, are extensive procedures associated with high mortality and morbidity.[1],[2] An audit by the National Health Service (NHS) of 2200 esophagectomies in 2010 reported inhospital mortality of 5%, with 3 in 10 cases having some postoperative complications.[3] Other studies have reported 30-day mortalities ranging from 4% to 14% and morbidities around 30%–65%.[4],[5] Complications have been attributed to surgeon's experience, patient turnover at the center, presence of comorbidities, and type of surgical procedure among others.[2],[5],[6],[7],[8]

Esophageal cancer is the sixth most common cause of cancer-related death in India, with about 47,000 cases reported annually.[9] In the Indian subcontinent, late presentation, preexisting malnutrition, and economic constraints impact the management of complex cancer surgeries like that of the esophagus. Complications not just lead to patient morbidity but also have severe financial implications for the family. A case-mix model recently observed that just decreasing complications reduced costs by one-third.[10] Hence, it is imperative to make a thorough risk assessment before surgery.

A scoring system that could predict the postoperative mortality and morbidity could help in stratifying patients at-risk. This would aid in the treatment planning and preoperative optimization of high-risk patients as well as in the counseling of patients and family. Furthermore, risk stratification would help in classifying patients needing intensive care from those who could make do with simple postoperative care, achieving optimal use of both human and monetary resources. The standard American Society of Anaesthesiologists Score (ASA-PS) followed by anesthesiologists worldwide, does not account for age, body mass index, or type of surgical procedure. Furthermore, the interobserver variability of this scoring system reduces its objectivity.[11]

The Physiological and Operative Severity Score for the Enumeration of Mortality and Morbidity (POSSUM) developed by Copeland in 1991 has been used in a number of surgical procedures to predict risk-adjusted mortality and morbidity using both physiological and operative variables.[12] Modifications of this score to provide better predictive accuracy were done later (Portsmouth-POSSUM [P-POSSUM]), and it was adapted for specific surgeries as well.[13] Tekkis et al.[14] devised a score specific to esophageal-gastric surgeries called the esophagogastric-POSSUM (O-POSSUM). Mohil et al.[15] previously examined POSSUM and P-POSSUM scores for patients undergoing emergency laparotomy and reported good suitabLEity in the Indian population. This study is the first to apply these scores for esophageal surgeries in the Indian population and evaluates their suitability in the prediction of 30-day mortality and morbidity.

  Patients and Methods Top

This was a retrospective observational study that included all patients with biopsy-proven cancer of the esophagus who underwent surgery for the same at our institution between January 2015 and January 2019. Patients with inoperable or unresectable primary were excluded. Furthermore, those patients whose records were incomplete or not available were excluded from the study. Ethical clearance was taken from the institutional review board. Both electronic hospital information system records and manually recorded case sheets were used for data collection. All relevant records until postoperative day 30 were retrieved and recorded.

The preoperative staging was done by cross-sectional imaging and clinical examination. The same team of Oncosurgeons performed surgery over the study period. Any complication or death that occurred within 30 days postsurgery was identified as postoperative morbidity and mortality, respectively. Respiratory complications recorded included pneumonia, aspiration, atelectasis, collapse, and respiratory failure, whereas cardiovascular complications included myocardial infarction, severe arrhythmia, heart failure, cardiogenic or pulmonary edema, and pulmonary embolism. Surgical complications noted were an anastomotic leak, surgical site infections, hemorrhage, recurrent laryngeal nerve injury, and chylothorax. The total length of stay and length of stay more than 30 days were also noted.

We calculated the operative and the physiological scores using the retrospectively collected data. For missing data, a score of 1 was assigned to the corresponding variable. POSSUM morbidity and mortality, as well as P-POSSUM, were calculated using regression equations based on the original description, as outlined in [Table 1].[12],[13]
Table 1: Variables of Physiological and Operative Severity Score for enumeration of mortality and morbidity (POSSOM) score with the logistic equations

Click here to view

Statistical analysis of the data was performed using the R Project for Statistical Computing.[17] Descriptive analysis of continuous variables was reported as mean and standard deviation, whereas that of categorical variables as proportion. The ability of different scores to correctly predict the outcomes was assessed by both tests of calibration and discrimination. Discrimination assesses the capability of a model to predict the probability of event occurrence for patients at varying risk.[18]

In order to assess the discriminative ability of the various systems, receiver operating characteristic (ROC) curves were generated for individual scoring systems, with sensitivity plotted on the Y-axis and specificity plotted on the X-axis. Area under the curve (AUC) values of >0.9 were interpreted as good discrimination, 0.7–0.9 as moderate, and <0.7 as poor.[19] The mortality and morbidity rates predicted by the scoring systems were compared with observed rates using linear analysis. An observed/expected (O/E) value of 1 indicates that the scoring system is able to predict the events accurately. An O/E ratio <1 indicates a lower number of events than expected, whereas an O/E ratio >1 indicates a greater number of events than expected. In addition, to assess the calibration of the various systems, the Hosmer–Lemeshow (H-L) goodness-to-fit test was applied.[20] Calibration shows how a model can predict absolute risk as well as predict risk for different groups.[18] Assessed using the X2 test, larger X[2] values and corresponding P ≤ 0.05 indicate that the model is poorly calibrated.

  Results Top

Sixty patients were included in the final analysis. The patient demographics are described in [Table 2]. In the majority of patients, the tumor was in the middle third (41.67%), with adenocarcinoma being the most frequent histology finding (78.34%). Neoadjuvant chemotherapy was received by 63.34% of patients preoperatively.
Table 2: Patient demographics (n=60)

Click here to view

Four out of 60 patients died within 30 days (6.67%), whereas 28 of them developed complications in the same period (46.66%). In 6 patients, the length of stay exceeded 30 days, with 41 days being the maximum hospital stay. The surgical outcomes and complications are shown in [Table 3]. Pulmonary complications (most common were consolidation and pneumonia) were observed in 15/60 patients, whereas 18/60 patients had cardiac complications (atrial fibrillation in 2/60, paroxysmal supraventricular tachycardia in 1/60, shock requiring inotropic support in 4/60, and postoperative hypotension in 7/60 [11.67%] patients). Recurrent laryngeal nerve injury and chylothorax were not found in any patient
Table 3: Surgical outcome in terms of 30.day mortality and morbidity (n=60)

Click here to view

The predicted mortality rates were calculated for the various POSSUM scores [Table 4], along with the O/E ratio. The POSSUM mortality overpredicted the mortality by nearly twice the expected number (O/E: 0.52), though the H-L analysis showed no significant difference (χ2 = 3.37, P > 0.05). Analysis of expected mortality as predicted by the P-POSSUM score showed a rate of 3.8% as opposed to observed of 6.6% (O/E of 1.71), indicating underprediction. The H-L test of P-POSSUM scores did not show much difference between values (χ2 = −1.502, P > 0.05). When the calibration of the O-POSSUM scoring system was analyzed, it revealed both good correlations with the expected rate of 3.92 (O/E: 1.02) and a good fit (χ2 = 0.066, P > 0.05).
Table 4: Predicted compared with observed mortality (n=60)

Click here to view

The result of 30-day POSSUM morbidity stratified by risk groups is displayed in [Table 5]. POSSUM score groups were stratified as low risk (0–49), moderate risk (50–69), and high risk (≥70). The O/E ratio for all risk groups was 0.862, indicating overprediction of morbidity by the score, though for the high-risk group, there was slight underprediction (O/E: 1.11). However, there was no significant difference between observed and predicted values (χ2 = 2.30, P > 0.2).
Table 5: Outcome of POSSOM stratified by risk groups for morbidity (30 day)

Click here to view

Discrimination of the various scoring systems to predict mortality was assessed using the receiver operator characteristic curve (the C-statistic). ROC curve [Figure 1] analysis applied to O-POSSOM scores showed moderate discriminatory capability for mortality, as shown in [Table 6], although significantly better than chance (AUC: 0.738, 95% confidence interval [CI] = 0.53–0.94). The ROC curve analysis of total scores, P-POSSOM, and POSSOM scores also revealed better discriminatory power for mortality (AUC: 0.81, 95% CI = 0.56–1, and 0.82, 95% CI = 0.58–1.00), respectively. The standalone physiological and surgical scores showed poor discriminatory capability for mortality. For morbidity, the ROC curve [Figure 2] analysis of POSSOM, physiological, and total score revealed moderate discriminatory power (AUC <0.8) but better than chance, whereas for the surgical score, the discrimination was poor.
Figure 1: Receiver operating characteristic curve for mortality

Click here to view
Table 6: Area under the curve for predicting postoperative mortality and morbidity

Click here to view
Figure 2: Receiver operating characteristic curve for morbidity

Click here to view

  Discussion Top

Preoperative assessment of patients at risk is essential for defining and modifying postoperative care. Scoring systems such as POSSUM, which are risk adjusted could provide an objective evaluation of true risk. These systems are easy to calculate and require only small amounts of data. Incorporation of intraoperative events and operative severity into the variables studied is an added advantage. However, since they were developed for a specific population, their performance in varied cohorts, like that of the Indian subcontinent, needs to be evaluated

When analyzed for esophageal surgeries, Filip et al. and Bollschweiler et al. found a good correlation of POSSUM mortality scores with actual mortality, but others have reported overprediction.[21],[22] In our study, we found that while the POSSUM mortality system overestimated mortality (O/E: 0.52), it exhibited moderate discrimination and good calibration. This apparent discrepancy, also observed by Lai et al. and Bosch et al., has been attributed to the difference between predictive accuracy and discriminatory power of a model.[23],[24]

The overprediction of mortality could be because the POSSUM score was designed for general surgery, and specialized surgery like esophagectomies might require other predictors. Furthermore, as elucidated by Warnell et al., patients with poor Glasgow Coma Scale and extremes of biochemical and hematological values are not usually taken up electively for such major procedures.[4] The absence of those values could have affected the score. Our data of only elective surgeries could have also have impacted the prediction.

POSSUM morbidity, on the other hand, showed good performance as the observed morbidity rate approximated the expected mortality rate (O/E ratio of 0.862) and exhibited reasonable discriminatory power as well (AUC: 0.701). Hong et al. reported good predictive rates (O/E: 0.93) with moderate discrimination (AUC: 0.787) while analyzing morbidity in patients with gastric cancer using the POSSUM morbidity score.[25] Other studies have also indicated reasonable accuracy, meriting the use of the POSSUM scale for the prediction of postoperative morbidity.[26],[27] Using the stand-alone physiological and operative scores without the logistic equations was adequate for morbidity prediction not for mortality. Similar findings for the physiological score were also noted by other authors.[21],[22]

P-POSSUM was developed chiefly to address the overprediction issues reported with POSSUM by developing better regression analysis.[13] Lai et al. while comparing O-POSSUM and P-POSSUM in esophageal surgeries reported better prediction with P-POSSUM, a result that is also reported by Bosch et al. and Mutiso et al.[23],[24],[28] In our study, however, P-POSSUM exhibited reasonable discrimination and good calibration, and it underpredicted the mortality (O/E: 1.71), displaying poor predictive accuracy.

The O-POSSUM was developed by Tekkis et al. to provide a scoring system, especially for esophagogastric surgeries. They reported both good calibration and discrimination for O-POSSUM when compared to P-POSSUM scoring system, a result also validated by Gocmen et al.[14],[29] Conversely, Lagarde et al.[30] reported poor performance of O-POSSUM scores in both discrimination and calibration tests.

In our study, we found that the O-POSSUM score exhibited good calibration with excellent predictive accuracy (O/E: 1.01). This could be because the score was developed specifically for esophagogastric surgeries. Yet, its predictive ability was only moderate (AUC: 0.738), indicating that though it had better overall performance, it could not be used as a benchmark. Nagabhushan et al. had commented that in elective oncologic surgeries, none of the available scoring systems display sufficient predictive accuracy.[31]

Our study had a few limitations. Being a retrospective study from a single center, we only could examine secondary data, and we had some missing data to contend with. Furthermore, we had a small sample size and event rate, which could have had some impact on the outcome. Further extensive studies taking into account the diversity of the Indian patient populace needs to be carried out to confirm the findings. Maybe indigenous scoring systems pertinent to the geographic milieu could be developed to improve accuracy in risk prediction.

  Conclusions Top

This study showed that the POSSUM scoring system was useful in predicting morbidity risk following esophageal resection for cancer. Furthermore, the best overall performance taking into account both discrimination and calibration tests was that of O-POSSUM, though we must note that its performance in the discrimination test was at best middling.


We wish to acknowledge the expertise and help rendered by Dr. Gurjeet Khurana, MD (Anaesthesia), Professor and Ex Head, Department of Anaesthesiology, throughout this study.

Financial support and sponsorship


Conflicts of interest

There are no conflicts of interest.

  References Top

Brewer LA 3rd. History of surgery of the esophagus. Am J Surg 1980;139:730-43.  Back to cited text no. 1
D'Amico TA. Outcomes after surgery for esophageal cancer. Gastrointest Cancer Res 2007;1:188-96.  Back to cited text no. 2
Cromwell D, Palser T, Meulen J, Hardwick R, Riley S, Greenway K, et al. National Oesophago Gastric Cancer Audit 2010. Available from: https://digital.nhs.uk/datainformation/publications/statistical/national-oesophago-gastric-cancer-audit/national-oesophago-gastric-cancer-audit-2010-annual-report. [Last accessed on 2020 Feb 24].  Back to cited text no. 3
Warnell I, Chincholkar M, Eccles M. Predicting perioperative mortality after oesophagectomy: A systematic review of performance and methods of multivariate models. Br J Anaesth 2015;114:32-43.  Back to cited text no. 4
Gockel I, Exner C, Junginger T. Morbidity and mortality after esophagectomy for esophageal carcinoma: A risk analysis. World J Surg Oncol 2005;3:37.  Back to cited text no. 5
Wright CD, Kucharczuk JC, O'Brien SM, Grab JD, Allen MS; Society of Thoracic Surgeons General Thoracic Surgery Database. Predictors of major morbidity and mortality after esophagectomy for esophageal cancer: A society of thoracic surgeons general thoracic surgery database risk adjustment model. J Thorac Cardiovasc Surg 2009;137:587-95.  Back to cited text no. 6
Mousavi SR, Akbari ME. Comparison of early and late complications in three esophagectomy techniques. Int J Cancer Manag 2017;10:e7644.  Back to cited text no. 7
Bhayani NH, Gupta A, Dunst CM, Kurian AA, Reavis KM, Swanström LL. Esophagectomies with thoracic incisions carry increased pulmonary morbidity. JAMA Surg 2013;148:733-8.  Back to cited text no. 8
Samarasam I. Esophageal cancer in India: Current status and future perspectives. Int J Adv Med Health Res 2017;4:5-10.  Back to cited text no. 9
  [Full text]  
Basavaiah G, Rent PD, Rent EG, Sullivan R, Towne M, Bak M, et al. Financial impact of complex cancer surgery in India: A study of pancreatic cancer. J Glob Oncol 2018;4:1-9.  Back to cited text no. 10
Daabiss M. American Society of Anaesthesiologists physical status classification. Indian J Anaesth 2011;55:111-5.  Back to cited text no. 11
[PUBMED]  [Full text]  
Copeland GP, Jones D, Walters M. POSSUM: A scoring system for surgical audit. Br J Surg 1991;78:355-60.  Back to cited text no. 12
Whiteley MS, Prytherch DR, Higgins B, Weaver PC, Prout WG. An evaluation of the POSSUM surgical scoring system. Br J Surg 1996;83:812-5.  Back to cited text no. 13
Tekkis PP, McCulloch P, Poloniecki JD, Prytherch DR, Kessaris N, Steger AC. Risk-adjusted prediction of operative mortality in oesophagogastric surgery with O-POSSUM. Br J Surg 2004;91:288-95.  Back to cited text no. 14
Mohil RS, Bhatnagar D, Bahadur L, Rajneesh, Dev DK, Magan M. POSSUM and P-POSSUM for risk-adjusted audit of patients undergoing emergency laparotomy. Br J Surg 2004;91:500-3.  Back to cited text no. 15
O-POSSUM Scoring Risk Prediction in Surgery. Available from: http://www.riskprediction.org.uk/index-op.php. [Last accessed on 2019 Apr25].  Back to cited text no. 16
Statistical Software version 3.6.2 (2019). Available from: https://www.r-project.org. [Last accessed 2019 Sep 02].  Back to cited text no. 17
Alba AC, Agoritsas T, Walsh M, Hanna S, Iorio A, Devereaux PJ, et al. Discrimination and calibration of clinical prediction models: Users' guides to the medical literature. JAMA 2017;318:1377-84.  Back to cited text no. 18
Swets JA. Measuring the accuracy of diagnostic systems. Sci 1988;240:1285-93.  Back to cited text no. 19
Hosmer DW, Hosmer T, Le Cessie S, Lemeshow S. A comparison of goodness-of-fit tests for the logistic regression model. Stat Med 1997;16:965-80.  Back to cited text no. 20
Filip B, Hutanu I, Radu I, Anitei MG, Scripcariu V. Assessment of different prognostic scores for early postoperative outcomes after esophagectomy. Chirurgia (Bucur) 2014;109:480-5.  Back to cited text no. 21
Bollschweiler E, Lubke T, Monig SP, Holscher AH. Evaluation of POSSUM scoring system in patients with gastric cancer undergoing D2-gastrectomy. BMC Surg 2005;5:8.  Back to cited text no. 22
Lai F, Kwan TL, Yuen WC, Wai A, Siu YC, Shung E. Evaluation of various POSSUM models for predicting mortality in patients undergoing elective oesophagectomy for carcinoma. Br J Surg 2007;94:1172-8.  Back to cited text no. 23
Bosch DJ, Pultrum BB, de Bock GH, Oosterhuis JK, Rodgers MG, Plukker JT. Comparison of different risk-adjustment models in assessing short-term surgical outcome after transthoracic esophagectomy in patients with esophageal cancer. Am J Surg 2011;202:303-9.  Back to cited text no. 24
Hong S, Wang S, Xu G, Liu J. Evaluation of the POSSUM, p-POSSUM, o-POSSUM, and APACHE II scoring systems in predicting postoperative mortality and morbidity in gastric cancer patients. Asian J Surg 2017;40:89-94.  Back to cited text no. 25
Yan J, Wang YX, Li ZP. Predictive value of the POSSUM, p-POSSUM, cr-POSSUM, APACHE II and ACPGBI scoring systems in colorectal cancer resection. J Int Med Res 2011;39:1464-73.  Back to cited text no. 26
Ugolini G, Rosati G, Montroni I, Zanotti S, Manaresi A, Giampaolo L, et al. Can elderly patients with colorectal cancer tolerate planned surgical treatment? A practical approach to a common dilemma. Colorectal Dis 2009;11:750-5.  Back to cited text no. 27
Mutiso FM. Evaluation of P-POSSUM and O-POSSUM scores in prediction of 30 day mortality rate in patients undergoing resection for oesophageal cancer. University of Nairobi; 2011.  Back to cited text no. 28
Gocmen E, Koc M, Tez M, Keskek M, Kilic M, Ertan T. Evaluation of P-POSSUM and O-POSSUM scores in patients with gastric cancer undergoing resection. Hepatogastroenterology 2004;51:1864-6.  Back to cited text no. 29
Lagarde SM, Maris AK, de Castro SM, Busch OR, Obertop H, van Lanschot JJ. Evaluation of O-POSSUM in predicting in-hospital mortality after resection for oesophageal cancer. Br J Surg 2007;94:1521-6.  Back to cited text no. 30
Nagabhushan JS, Srinath S, Weir F, Angerson WJ, Sugden BA, Morran CG. Comparison of P-POSSUM and O-POSSUM in predicting mortality after oesophagogastric resections. Postgrad Med J 2007;83:355-8.  Back to cited text no. 31


  [Figure 1], [Figure 2]

  [Table 1], [Table 2], [Table 3], [Table 4], [Table 5], [Table 6]


    Similar in PUBMED
   Search Pubmed for
   Search in Google Scholar for
 Related articles
    Access Statistics
    Email Alert *
    Add to My List *
* Registration required (free)  

  In this article
Patients and Methods
Article Figures
Article Tables

 Article Access Statistics
    PDF Downloaded148    
    Comments [Add]    

Recommend this journal