Comparison of P-POSSUM and Cr-POSSUM Scores

ORIGINAL ARTICLE

Comparison of P-POSSUM and Cr-POSSUM Scoresin Patients Undergoing Colorectal Cancer ResectionMatija Horzic, MD, PhD; Mario Kopljar, MD, MSc; Kristijan Cupurdija, MD, MSc;Djana Vanjak Bielen, MD; Domagoj Vergles, MD; Zeljko Lackovic, MD

Objectives: To compare the Portsmouth (P) Physi-ological and Operative Severity Score for the Enumera-tion of Mortality and Morbidity (POSSUM) and spe-cialized colorectal (Cr) POSSUM scoring systems inthe prediction of mortality after resection of colorectalcancer.

Design: Retrospective study of patients after resectionof colorectal cancer.

Setting: University hospital.

Patients: One hundred twenty patients with completemedical records who underwent resection of colorectalcancer between January 1, 1996, and December 31, 2004,at our institution were enrolled in the study.

Main Outcome Measures: P-POSSUM and Cr-POSSUM scores were calculated for each patient. In-hospital mortality rate and number of deaths within 30

days after surgery were recorded. The ratio of observedto expected deaths was calculated for each analysis.

Results: The P-POSSUM system underpredicted mor-tality by 25%, with no significant difference between thepredicted and observed values (P=.96). The observed toexpected ratio for Cr-POSSUM was 1.11, with no sig-nificant difference between the observed and predictedvalues (P=.19). Area under the receiver operating curvefor P-POSSUM was 0.70 and for Cr-POSSUM was 0.59.

Conclusions: Both P-POSSUM and Cr-POSSUM per-form well in predicting mortality after colorectal cancersurgery, but the Cr-POSSUM is more accurate. There isa constant need for reevaluation of existing and any newscoring systems outside original development and vali-dation populations. The Cr-POSSUM score is a promis-ing specialized tool for monitoring surgical outcomes incolorectal cancer surgery.

Arch Surg. 2007;142(11):1043-1048

O PERATIVE MORTALITY RATE

is a common measure ofoutcome and can be usedto compare quality ofhealth care.1 However,

when comparing quality of care, mortal-ity and morbidity rates have obvious limi-tations and may give misleading results be-cause they do not consider the physiologiccondition of the patient at the time of sur-gery, the severity of the surgery, and the

age and general health of the patient.2,3 Togive a more objective comparison for qual-ity of care, various scoring systems havebeen introduced.

One of the first scoring systems for pre-dicting outcome in surgery was the Physi-ological andOperativeSeverityScore for the

Enumeration of Mortality and Morbidity(POSSUM), which was designed for gen-eral surgery.4 Since the original POSSUMsystem was introduced, several modifica-tions have been suggested for the specificrequirements of certain surgical subspeci-alities.1,5,6 Also, there is concern about theapplicability of POSSUM scores in healthcare domains other than the one it wasoriginally designed for.7 Therefore, modi-fications of the original POSSUM scorewere created. The Portsmouth POSSUM(P-POSSUM) system was designed to over-come the problem of overpredicting mor-tality in patients at low risk by using theoriginal POSSUM score.5,8 P-POSSUM sys-tem was found to be more accurate in pre-dicting mortality in general surgery.5

Colorectal surgery is a specific surgi-cal subspeciality. The colorectal POSSUM(Cr-POSSUM) system was created in 2004specifically for this field of surgery.1 Within

See Invited Critiqueat end of article

Author Affiliations:Department of Surgery,University Hospital Dubrava,Zagreb, Croatia.

(REPRINTED) ARCH SURG/ VOL 142 (NO. 11), NOV 2007 WWW.ARCHSURG.COM1043

©2007 American Medical Association. All rights reserved.Downloaded From: http://archsurg.jamanetwork.com/ on 05/31/2012

colorectal surgery, oncologic colorectal surgery is par-ticularly demanding. Patients with colorectal cancer areoften at increased risk of complications owing to spe-cific features of colorectal cancer such as malnutrition,anemia, and compromised immune systems.9-11 The ob-jective of this study was to assess the accuracy of P-POSSUM and Cr-POSSUM systems in predicting post-operative mortality in patients with colorectal cancer.

Table 1. Parameters for Calculating P-POSSUM Scorea

Parameter Points

Physiologic scoreAge, y

�60 161-70 2�71 4

Cardiac signs and symptomsNo failure 1Diuretic, digoxin, antianginal, or

antihypertensive therapy2

Peripheral edema, warfarin therapy,borderline cardiomegaly on chestradiograph

4

Elevated jugular venous pressure,cardiomegaly on chest radiograph

8

Respiratory findingsNo dyspnea 1Dyspnea on exertion; mild evidence of COPD

on chest radiograph2

Limiting dyspnea after walking up 1 flight ofstairs; moderate COPD on chest radiograph

4

Dyspnea at rest; respiratory rate �30breaths/min; fibrosis or consolidation onchest radiograph

8

Systolic blood pressure, mm Hg�89 890-99 4100-109 2110-130 1131-170 2�171 4

Pulse rate, per minute�39 840-49 250-80 181-100 2101-120 4�121 8

Glasgow Coma Scale score15 112-14 29-11 4�8 8

Hemoglobin, g/dL�9.9 810.0-11.4 411.5-12.9 213.0-16.0 116.1-17.0 217.1-18.0 4�18.1 8

White blood cell count, No./µL�3000 43100-3999 24000-10 000 110 100-20 000 2�20 100 4

Serum urea nitrogen, mg/dL�21.0 121.3-28.0 228.3-42.0 4�43.0 8

Sodium, mEq/L�125 8125-126 4131-135 2�136 1

(continued)

Table 1. Parameters for Calculating P-POSSUM Score(cont)a

Parameter Points

Potassium, mEq/L�2.8 82.9-3.1 43.2-3.4 23.5-5.0 15.1-5.3 25.4-5.9 4�6.0 8

Electrocardiographic findingsNormal 1Atrial fibrillation with heart rate 60-90 beats/min 4Any other abnormal rhythm 8�5 Ectopic beats per minute 8Q wave or ST-T wave changes 8

Operative severity scoreOperative severity

Minor 1Moderate 2Very major 4Complex major 8

No. of procedures1 12 4�2 8

Total blood loss, mL�100 1101-500 2501-999 4�1000 8

Peritoneal soilingNone 1Minor, serous fluid without pus 2Local pus 4Free bowel content, pus or blood 8

Presence of cancerNone 1Primary lesion only 2Nodal metastases 4Distant metastases 8

Mode of surgeryElective 1Emergency with resuscitation �2 h 4Surgery �24 h after admission 4Immediate surgery 8

Abbreviations: COPD, chronic obstructive pulmonary disease; P-POSSUM,Portsmouth Physiologic and Operative Severity Score for the Enumeration ofMortality and Morbidity.

SI conversion factors: To convert hemoglobin to grams per liter, multiplyby 10.0; white blood cell count to �109/L, multiply by 0.001; serum ureanitrogen to millimoles per liter, multiply by 0.357; and sodium and potassiumto millimoles per liter, multiply by 1.0.

a ln [R/(1−R)] = −9.065 � (0.1692 � Physiologic Score) �(0.1550 � Operative Severity Score), where R is the predicted risk ofmortality.



METHODS

Patients who underwent resection of colorectal cancer betweenJanuary 1, 1996, and December 31, 2004, at our institution wereretrospectively included in the study. Those patients for whomP-POSSUM and Cr-POSSUM scores could not be calculated be-cause of lack of data were excluded. Parameters for calculatingP-POSSUM and Cr-POSSUM are given in Table1 and Table 2.The remaining 120 patients were included in the study. Physi-ologic scores for both P-POSSUM and Cr-POSSUM were calcu-lated for each patient from their medical records. Operative se-verity scores were calculated based on findings recorded by theoperating surgeon. In-hospital mortality and death within 30 daysafter colorectal surgery were recorded. Both scores were calcu-lated as previously described.1,5

Data were analyzed using the linear method of analysis de-scribed by Wijesinghe et al.6 In this type of analysis, patientsare stratified into groups according to the predicted risk of death.Expected number of deaths is then calculated for each risk groupby multiplying the number of patients in a given group withaverage risk of death in that group. The ratio of observed toexpected deaths (O:E ratio) was calculated for each analysis.The �2 test of Lemeshaw and Hosmer12 was used to assess anydifferences between predicted and observed morbidity and mor-tality rates. Furthermore, 3 separate subgroups were analyzedaccording to the type of operation, including right-sided hemi-colectomy or transverse colon resection; left-sided hemicolec-tomy, the Hartmann procedure, anterior resection of the rec-tum, or resection of the sigmoid colon; and abdominoperinealresection. Discrimination ability, that is, the ability of the modelto assign higher probabilities of death to those patients whodied, was measured using receiver operating characteristiccurves, which were analyzed for both scores. P� .05 was con-sidered statistically significant.

RESULTS

The study included 69 men and 51 women. Ten patients(8.3%) died either in hospital or within 30 days after co-lorectal surgery and 23 (19.2%) developed complications.Two patients (1.7%) underwent 2 repeated laparotomiesand 15 patients (12.5%) underwent 1 repeated operation.

Potentially curative surgery was performed in 101 pa-tients (84.2%) and included right-sided hemicolectomyin 19 patients, left-sided hemicolectomy in 7, resectionof the transverse colon in 5, resection of the sigmoid co-lon in 21, anterior resection of the rectum in 18, abdomi-noperineal resection in 22, and the Hartmann proce-dure in 9. In the remaining 19 patients, palliative surgerywas performed that always included laparotomy, as fol-lows: bypass surgery in 5 patients, local excision of thetumor in 2, creation of a palliative stoma in 8, and sur-gical exploration in 4.

In 24 patients in whom right-sided hemicolectomy ortransverse colon resection was performed, the O:E ratiofor all risk groups was 1.00, indicating that the Cr-POSSUM system correctly predicted mortality (1 pre-dicted vs 1 observed). There was no significant differ-ence between the observed and predicted values (�2

8=0.35;P=1.00). P-POSSUM also correctly predicted mortality(O:E ratio 1.00). There was no significant difference be-tween the predicted and observed values (�2

8=8.93; P=.35).In 55 patients in whom left-sided hemicolectomy, the

Hartmann procedure, or anterior or sigmoid resection was

performed, the O:E ratio for all risk groups was 0.80, in-dicating that Cr-POSSUM overpredicted mortality in thisstudy by 20% (5 predicted vs 4 observed). There was nosignificant difference between the observed and pre-dicted values (�2

7=5.55; P=.593). P-POSSUM system cor-rectly predicted mortality (O:E ratio 1.00). There was nosignificant difference between the predicted and ob-served values (�2

7=7.24; P=.41).

Table 2. Parameters for Calculating Cr-POSSUM Scorea

Parameter Points

Physiologic scoreAge, y

�60 161-70 371-80 4�81 8

Cardiac failureNo failure 1Moderate 2Severe 3

Systolic blood pressure, mm Hg100-170 1�170 or 90-99 2�90 3

Pulse, per minute40-100 1101-120 2�120 or �40 3

Hemoglobin, g/dL13.0-16.0 110.0-12.9 or 16.1-18.0 2�10 or �18.0 3

Serum urea nitrogen, mg/dL�10 110.1-15.0 2�15.0 3

Operative severity scoreOperative severity

Minor 1Intermediate 3Major 4Complex major 8

Peritoneal soilingNone or serous fluid 1Local pus 2Free pus or feces 3

Cancer stageb

No cancer or Duke A or B 1Duke C 2Duke D 3Mode of surgery

Elective 1Urgent 3Emergency 8

Abbreviation: Cr-POSSUM, colorectal Physiologic and Operative SeverityScore for the Enumeration of Mortality and Morbidity.

SI conversion factors: To convert hemoglobin to grams per liter, multiplyby 10.0; serum urea nitrogen to millimoles per liter, multiply by 0.357.

a ln [R/(1−R)] = −9.167 � (0.338 � Physiologic Score) �(0.308 � Operative Severity Score), where R is the predicted risk ofmortality.

bDuke system of cancer staging: A, lesion confined to the mucosa; B,varies by system; C, lymph nodes positive for cancer; and D, presence ofdistant metastases.



In 22 patients in whom abdominoperineal resectionwas performed, the O:E ratio for all risk groups was 1.00,indicating that Cr-POSSUM correctly predicted mortal-ity (2 predicted vs 2 observed). There was no significantdifference between the observed and predicted values(�2

6=0.95; P=.99). P-POSSUM underpredicted mortal-ity with an overall O:E ratio of 2.00. There was no sig-nificant difference between the predicted and observedvalues (�2

8=8.40; P=.40).Table 3 gives the number of deaths predicted by

Cr-POSSUM with linear analysis when all patients wereanalyzed, including those who underwent palliative pro-cedures. The O:E ratio for all risk groups was 1.11, in-dicating that the Cr-POSSUM system underpredicted

mortality in this study by 11%. However, there was nosignificant difference between the observed and pre-dicted values (�2

7=10.05; P=.19). P-POSSUM systemalso underpredicted mortality by 25%, with an overallO:E ratio of 1.25. There was no significant differencebetween the predicted and observed values (�2

8=2.54;P=.96) (Table 3).

Discriminatory power of P-POSSUM and Cr-POSSUM scores in predicting death as an outcome mea-sure was analyzed using receiver operating characteris-tic curves. Area under the receiver operating characteristiccurve (AUC) for Cr-POSSUM was 0.59 (95% confi-dence interval, 0.36-0.82) (Figure). For P-POSSUM, theAUC was 0.70 (95% confidence interval, 0.52-0.88), in-dicating satisfactory discriminatory power (Figure).

COMMENT

In this study, validity of P-POSSUM and Cr-POSSUM scoresin patients who underwent resection of colorectal cancerwas analyzed by assessing calibration and discrimination.Calibration is the ability of the model to assign the correctprobabilities of outcome to individual patients. In this analy-sis, patients were stratified into risk groups on the basis ofpredicted mortality. The predicted number of deaths in eachrisk group was compared with the observed number ofdeaths using the Hosmer-Lemeshaw goodness-of-fit test(Tables 1-3). Both scores demonstrated good calibrationability, with no statistically significant differences in ob-served to expected number of deaths. P-POSSUM systemunderpredicted mortality by 25%. More accurate predic-tion of mortality with P-POSSUM was observed in pa-tients at high risk compared with patients at low risk. Simi-larly, the Cr-POSSUM system also underpredicted mortalityin patients at low risk, but overall accuracy was greater, withan O:E ratio of 1.11.

Discriminatory power of P-POSSUM and Cr-POSSUM, that is, the ability of the model to assign higher

Table 3. Comparison of Observed and Predicted Mortality Rates by P-POSSUM and Cr-POSSUM Using Linear Analysis

Predicted Mortality Rate, % No. of Patients Predicted No. of Deathsa Observed No. of Deaths O:E Ratio

P-POSSUM�10 105 5 6 1.20�10 to �20 8 1 1 1.00�20 to �30 3 1 0 0�30 to �40 1 0 1�40 to �50 2 1 1 1.00�50 to �100 1 1 1 1.00

Total 120 8b 10 1.25Cr-POSSUM

�10 101 5 5 1.00�10 to �20 11 2 4 2.00�20 to �30 3 1 0 0�30 to �80 4 1 1 1.00�80 to �100 1 1 0 0

Total 120 9b 10 1.11

Abbreviations: Cr, colorectal; P, Portsmouth; POSSUM, Physiological and Operative Severity Score for the Enumeration of Mortality and Morbidity; O:E,observed to expected.

aRounded to nearest whole number.bThe rounded sum of nonrounded results (predicted number of deaths) to avoid cumulative mistake.

100

50

75

25

0 25

ReferenceP-POSSUMCr-POSSUM

50 75 100Specificity, %

Sens

itivi

ty, %

Figure. Receiver operator characteristic curves for performance ofPortsmouth (P) Physiologic and Operative Severity Score for theEnumeration of Mortality and Morbidity (POSSUM) and colorectal (Cr)POSSUM scores in patients with colorectal cancer.



probabilities of outcome to patients who die comparedwith those who live, was analyzed using the AUC. In gen-eral, the AUC ranges from 0.5 for chance performanceto 1.0 for perfect prediction.13 In this study, the AUC forP-POSSUM was 0.70, representing good discriminationpower of this score. However, Cr-POSSUM did not per-form as well, and the AUC was only 0.59. These resultsindicate that although P-POSSUM and Cr-POSSUM maybe used to calculate predicted mortality rates in givenpopulations, they are less accurate for predicting the riskof death for individual patients.

The results obtained in this study are somewhat dif-ferent from those previously published. Some valida-tion studies of P-POSSUM report slight overpredictionof mortality, especially in populations at low risk.7,14 Thisoverprediction has been explained in part by the math-ematical characteristics of the scoring system; that is, thelowest probability for each patient with the P-POSSUMscoring system is 0.2%, which is substantially more thanobserved in young, fit patients undergoing elective mi-nor surgery.

Substantial differences in prediction of mortality basedon P-POSSUM have been described when applying thisscore in different populations and health care systems.7

Bennett-Guerrero et al7 compared P-POSSUM mortalityrates after surgery between patients in the United Statesand the United Kingdom and found overprediction of mor-tality by a factor of 4 to 6 in the United States.7 Possiblereasons for such overprediction may include differencesin the organization of intensive care units. Another pos-sible explanation may be the difference in population char-acteristics. For example, patients may have more ad-vanced disease, which may have profound implications inthe development of a scoring system.7 Patients with ad-vanced gastrointestinal cancer often have pronounced nu-tritional deficit.11 Advanced protein-calorie malnutritioncaused by decreased nutritional intake, dysfunctional meta-bolic processes, and hormonal- and cytokine-related ab-normalities are major causes of morbidity and mortalityin patients with cancer.11 According to official cancer reg-istry data, age at onset of colorectal cancer in Croatia15 is5 to 10 years later than in the United Kingdom.16

Specific scoring systems may be required to evaluatesurgical outcomes in different specialties. The Cr-POSSUM system was created as a modification of an origi-nal POSSUM score to suit the specific needs of colorectalsurgery.1 The results of our study demonstrate better ac-curacy of Cr-POSSUM compared with P-POSSUM in pre-dicting mortality after surgery for colorectal cancer, whichis in agreement with the results of another publishedstudy.14 However, all scoring systems tend to optimize thefit of the data to the original population. Although duringdevelopment, Cr-POSSUM fitted the data well in both thedevelopment and validation sets, it is important to cross-validate the scoring system externally by applying the modelto a different population to assess its predictive power.1

Practical value of the scores can be noted at differentlevels. Scores can indicate patients at high risk who re-quire additional postoperative care in intensive care unitsor surgical wards, although their vital functions could besatisfactory. Also, scores could indicate patients at highrisk who could benefit from postponing surgical treat-

ment and receive preoperative treatment to improve theircondition and decrease operative risk. Scores might aid indecision making about the extent of a surgical procedurein patients at high risk. In addition, they can offer an ob-jective parameter of risk that could help the patients indeciding to consent to a surgical procedure. In compar-ing mortality rates between institutions or individuals,scores can give an objective measure of patient preopera-tive condition and operative risk and, thus, provide a ba-sis for comparison of quality of health care and surgicalprocedures.

The results of this study demonstrate that both P-POSSUM and Cr-POSSUM perform well in prediction ofmortality after surgery for colorectal cancer. Specializedscoring systems more accurately predict mortality. Al-though both scoring systems are based on universallyavailable and clearly defined variables, there are differ-ences in observed and expected mortality in various geo-graphic settings. Therefore, there is a constant need forreevaluation of existing and new scoring systems out-side of original development and validation popula-tions. Cr-POSSUM is a promising tool for monitoring sur-gical outcomes in colorectal cancer surgery.

Accepted for Publication: April 23, 2006.Correspondence: Kristijan Cupurdija, MD, MSc, Depart-ment of Surgery, University Hospital Dubrava, AvenijaGojka Suska 6, HR-10000 Zagreb, Croatia ([email protected]).Author Contributions: Study concept and design: Hor-zic, Kopljar, and Cupurdija. Acquisition of data: Kopljar,Vanjak Bielen, Vergles, and Lackovic. Analysis and in-terpretation of data: Horzic, Kopljar, Cupurdija, VanjakBielen, Vergles, and Lackovic. Drafting of the manu-script: Horzic, Kopljar, and Cupurdija. Critical revisionof the manuscript for important intellectual content: Hor-zic, Kopljar, Cupurdija, Vanjak Bielen, Vergles, and Lack-ovic. Statistical analysis: Horzic and Kopljar. Obtained fund-ing: Horzic. Administrative, technical, and material support:Kopljar, Cupurdija, Vanjak Bielen, Vergles, and Lack-ovic. Study supervision: Horzic.Financial Disclosure: None reported.Funding/Support: This study was supported by the Min-istry of Science, Education, and Sports of the Republicof Croatia (project 0198020).

REFERENCES

1. Tekkis PP, Prytherch DR, Kocher HM, et al. Development of a dedicated risk-adjustment scoring system for colorectal surgery (colorectal POSSUM). Br J Surg.2004;91(9):1174-1182.

2. Mohil RS, Bhatnagar D, Bahadur L, Rajneesh , Dev DK, Magan M. POSSUM andP-POSSUM for risk-adjusted audit of patients undergoing emergency laparotomy.Br J Surg. 2004;91(4):500-503.

3. Neary WD, Heather BP, Earnshaw JJ. The physiological and operative severityscore for the enumeration of mortality and morbidity (POSSUM). Br J Surg. 2003;90(2):157-165.

4. Copeland GP, Jones D, Walters M. POSSUM: a scoring system for surgical audit.Br J Surg. 1991;78(3):355-360.

5. Prytherch DR, Whiteley MS, Higgins B, Weaver PC, Prout WG, Powell SJ. POSSUMand Portsmouth POSSUM for predicting mortality: Physiological and Operative Se-verity Score for the enUmeration of Mortality and morbidity. Br J Surg. 1998;85(9):1217-1220.

6. Wijesinghe LD, Mahmood T, Scott DJ, Berridge DC, Kent PJ, Kester RC. Com-



parison of POSSUM and the Portsmouth predictor equation for predicting deathfollowing vascular surgery. Br J Surg. 1998;85(2):209-212.

7. Bennett-Guerrero E, Hyam JA, Shaefi S, et al. Comparison of P-POSSUM risk-adjusted mortality rates after surgery between patients in the USA and the UK.Br J Surg. 2003;90(12):1593-1598.

8. Whiteley MS, Prytherch DR, Higgins B, Weaver PC, Prout WG. An evaluation ofthe POSSUM surgical scoring system. Br J Surg. 1996;83(6):812-815.

9. Inoue Y, Miki C, Kusunoki M. Nutritional status and cytokine-related protein break-down in elderly patients with gastrointestinal malignancies. J Surg Oncol. 2004;86(2):91-98.

10. Hatada T, Miki C. Nutritional status and postoperative cytokine response in co-lorectal cancer patients. Cytokine. 2000;12(9):1331-1336.

11. Palesty JA, Dudrick SJ. What we have learned about cachexia in gastrointestinalcancer. Dig Dis. 2003;21(3):198-213.

12. Lemeshow S, Hosmer DW Jr. A review of goodness of fit statistics for use in thedevelopment of logistic regression models. Am J Epidemiol. 1982;115(1):92-106.

13. Gocmen E, Koc M, Tez M, Keskek M, Kilic M, Ertan T. Evaluation of P-POSSUMand O-POSSUM scores in patients with gastric cancer undergoing resection.Hepatogastroenterology. 2004;51(60):1864-1866.

14. Senagore AJ, Warmuth AJ, Delaney CP, Tekkis PP, Fazio VW. POSSUM, p-POSSUM, and Cr-POSSUM: implementation issues in a United States health caresystem for prediction of outcome for colon cancer resection [published onlineahead of print July 15, 2004]. Dis Colon Rectum. 2004;47(9):1435-1441.

15. Cancer Incidence in Croatia. Zagreb, Croatia; Cancer Registry, Croatian NationalInstitute of Public Health; 2002.

16. Large Bowel (Colorectal) Cancer Factsheet. Cancer Research UK; April 2005. http://info.cancerresearchuk.org. Accessed October 2, 2005..

INVITED CRITIQUE

A ssessment of outcomes for surgical services maybecome an important way for the governmentand insurance companies to grade health sys-

tems, hospitals, and individual practitioners. Any timeoutcomes are evaluated or compared, a method of as-sessing the influence of medical comorbidity on the se-lected outcome of a selected procedure for a selected dis-ease must be provided to make the evaluation meaningful.POSSUM, P-POSSUM, and Cr-POSSUM scores are meth-ods for assessing the severity of comorbid and operativefactors that might influence surgical outcomes using com-plicated formulas associating numerous data points thatinclude medical comorbidity and operative severity fac-tors. Senagore and colleagues1 showed that the 3 POSSUMscores need to be calibrated by each system and suggestthat comparison between 2 systems should be under-taken carefully. The Cr-POSSUM score predicts mortal-ity closely even though missing data from the medicalrecords of patients causes little variation in the ability topredict outcome for colon cancer. This suggests that thescore for an individual patient may not be reliable. There-fore, we must be careful when using the scores to pre-dict individual patient outcomes, influence selection ofa complicated procedure, and determine futility of op-erative management of advanced disease on the basis ofindividual POSSUM scores. In the article by Horzic et althis point is emphasized by calculating the AUC. The valueof 0.59 for Cr-POSSUM indicates almost random pre-diction of mortality for an individual patient, even thoughit is an accurate predictor for a population of patients withcolorectal cancer. It is unlikely that the Cr-POSSUM scorewill ever be calculated for use in an individual patient aspart of the decision making preoperative process.

Several questions remain that can only be answeredby more experience with the Cr-POSSUM score. Whatinfluence will surgeon judgment and experience, vol-ume of cases, or hospital capability and resources haveon the predictability of the scores? How and where doesthe Cr-POSSUM score interface with the definition of fu-tility? Can Cr-POSSUM indexed outcomes reflect qual-ity better than process improvement currently man-dated by the Surgical Clinical Improvement Projectprograms? How will the POSSUM scores interact withthe outcomes assessment in the National Surgical Qual-ity Improvement Program project? How can we preventhealth care systems, hospitals, and surgeons from “gam-ing” the system to improve reported outcomes on an un-fair basis? Only time and more experience with these scor-ing systems will provide the answers to these and otherquestions.

Correspondence: Dr Fleshman, Department of Sur-gery, Section of Colon and Rectal Surgery, WashingtonUniversity, 660 S Euclid, Campus Box 8109, St Louis,MO 63110 ([email protected]).Financial Disclosure: None reported.

REFERENCE

1. Senagore AJ, Warmuth AJ, Delany CP, Tekkis PP, Fazio VW. POSSUM, p-POSSUM, and Cr-POSSUM: implementation issues in a United States healthcare system for prediction of outcome for colon cancer resection [publishedonline ahead of print July 15, 2004]. Dis Colon Rectum. 2004;47(9):1435-1441. doi:10.1007/s10350-004-0604-1.

James W. Fleshman Jr, MD



Documents

Comparison of P-POSSUM and Cr-POSSUM Scores