Elderly POSSUM, a dedicated score for ... - Wiley Online Library

8 downloads 5605 Views 135KB Size Report
Jan 28, 2010 - Elderly POSSUM, a dedicated score for prediction of mortality and morbidity after major colorectal surgery in older patients. P. Tran Ba Loc1, ...
Original article

Elderly POSSUM, a dedicated score for prediction of mortality and morbidity after major colorectal surgery in older patients P. Tran Ba Loc1 , S. Tezenas du Montcel1,3 , J. J. Duron2 , H. Levard4 , B. Suc5 , B. Descottes6 , B. Desrousseaux7 and J. M. Hay8 1

Biostatistics and Medical Information Unit and 2 Digestive Surgery and Transplantation Unit, Groupe Hospitalier Piti´e –Salpˆetri`ere, Assistance ˆ Publique–Hopitaux de Paris, Paris, 3 Pierre and Marie Curie Paris 6 University, Er 4 (former EA3974), Modelling in Clinical Research, Pierre and Marie Curie Paris 6 University, Paris, 4 General and Digestive Surgery Unit, Institut Monsouris, Paris, 5 Digestive Surgery Unit, Centre Hospitalier Universitaire de Rangueil, Rangueil, 6 Visceral Surgery and Transplantation Unit, Centre Hospitalier Universitaire de Limoges, Limoges, 7 General and ˆ Digestive Surgery Unit, Centre Hospitalier General de Lomme, Lomme, 8 General and Digestive Surgery Unit, Hopital Louis Mourier, Assistance ˆ Publique–Hopitaux de Paris, Paris, France Correspondence to: Dr S. Tezenas du Montcel, Unit´e de Biostatistiques et Information Medicale, D´epartement de Sant´e Publique, Groupe Hospitalier ˆ Piti´e –Salpˆetri`ere, 47–83 Boulevard de l’Hopital, 75651 Paris Cedex 13, France (e-mail: [email protected])

Background: Several scores have been developed to evaluate surgical unit mortality and morbidity. The

Physiological and Operative Severity Score for the enUmeration of Mortality and morbidity (POSSUM) and derivatives use preoperative and intraoperative factors, whereas the Surgical Risk Scale (SRS) and Association Franc¸ aise de Chirurgie (AFC) score use four simple factors. To allow for advanced age in patients undergoing colorectal surgery, a dedicated score – the Elderly (E) POSSUM – has been developed and its accuracy compared with these scores. Methods: From 2002 to 2004, 1186 elderly patients, at least 65 years old, undergoing major colorectal surgery in France were enrolled. Accuracy was assessed by calculating the area under the receiver operating characteristic curve (AUC) (discrimination) and calibration. Results: The mortality and morbidity rates were 9 and 41 per cent respectively. The E-POSSUM had both a good discrimination (AUC = 0·86) and good calibration (P = 0·178) in predicting mortality and a reasonable discrimination (AUC = 0·77) and good calibration (P = 0·166) in predicting morbidity. The E-POSSUM was significantly better at predicting mortality and morbidity than the AFC score (Pc = 0·014 and Pc < 0·001 respectively). Conclusion: The E-POSSUM is a good tool for predicting mortality, and the only efficient scoring system for predicting morbidity after major colorectal surgery in the elderly. Paper accepted 8 October 2009 Published online 28 January 2010 in Wiley InterScience (www.bjs.co.uk). DOI: 10.1002/bjs.6903

Introduction

In developed countries, the elderly population is defined by the World Health Organization (WHO) as people over 65 years old1 . Assessing the risk of mortality and morbidity after surgery in this age group is essential, as these patients represent a large proportion of the population, particularly in France, where they account for 32 per cent of the population2 . The mortality and morbidity of these patients, who are more likely to be admitted to hospital3 , are higher than in younger patients4,5 . This is particularly noticeable in major colorectal surgery, which is the primary Copyright  2010 British Journal of Surgery Society Ltd Published by John Wiley & Sons Ltd

type of digestive surgery, mostly because of colorectal cancers and diverticular disease6 . In this setting, mortality and morbidity rates are still high (6·4 and 19·4 per cent respectively) and depend on the patient’s age and whether the surgery is emergency or elective7 – 9 . Thus, a scoring system for elderly patients undergoing major colorectal surgery would clearly be of great use. Comparing the performance of hospitals, surgical units and surgeons is a delicate issue, as a simple comparison of postoperative mortality or morbidity rates does not take into account the severity of disease. In practice, to evaluate the performance of hospital surgical units and surgeons as well as the postoperative morbidity and mortality of British Journal of Surgery 2010; 97: 396–403

Score for predicting mortality and morbidity after colorectal surgery in the elderly

patients, a multivariable analysis highlighting independent risk factors is performed, followed by the development of a scoring system. The first such score, developed by Copeland and coworkers10 for purposes of surgical audit and based on 12 preoperative and six intraoperative items, was the Physiological and Operative Severity Score for the enUmeration of Mortality and morbidity (POSSUM). Following comments in the literature concerning the performance of POSSUM11,12 , Prytherch and colleagues13 , using the same collection of information but a different risk-predicting equation, designed the Portsmouth (P) POSSUM. The P-POSSUM is used to assess the risk of mortality, whereas the POSSUM is used to assess the risk of morbidity. In 2004, a modified POSSUM score called the Colorectal (Cr) POSSUM was constructed14 and externally tested15 to predict mortality in patients undergoing colorectal surgery. The Surgical Risk Scale (SRS) is a simple, easy-to-use alternative score that was designed for general surgery and has been internally validated favourably16 . In 2007 another simple, four-item score, called the Association Fran¸caise de Chirurgie (AFC) score, was developed for patients undergoing surgery for colorectal cancer and diverticulitis17 . The AFC score was externally validated18 . It was also validated for assessing the risk of mortality and includes only a score, not a risk prediction equation. However, it is also possible to assess its ability to predict morbidity by modelling the risk of morbidity as a function of this score. Despite numerous studies reporting the use of the POSSUM or its derived scores and the development and use of other scores19 specifically dedicated to cancer20 – 22 , no study to date has assessed a score dedicated for use with elderly patients undergoing major colorectal surgery. The purpose of this study was to develop a new score, the Elderly (E) POSSUM, to allow for advanced age in patients undergoing major colorectal operations, and to compare its discrimination and calibration (goodness-of-fit) with those of other available scores (the P-POSSUM, Cr-POSSUM, AFC score and SRS for mortality, and the POSSUM and AFC score for morbidity) in order to identify the most accurate test for this population. Methods

Between 1 January 2002 and 31 December 2004, all consecutive patients aged at least 18 years undergoing a first colorectal operation or early colorectal reoperation (if not yet enrolled) were prospectively recruited in 41 hospitals (teaching hospitals, general hospitals and private hospitals) covering all but two French regions. The Copyright  2010 British Journal of Surgery Society Ltd Published by John Wiley & Sons Ltd

397

eligible operations were elective (81·8 per cent), urgent (12·6 per cent) and emergency (5·6 per cent) major colorectal procedures corresponding to major and majorplus classes in the classification described by Copeland and co-workers10 or to complexity grades 3, 4 and 5 in the classification of Aust and colleagues23 . Of the 2214 patients of the initial cohort, 1245 were at least 65 years old. One was excluded because neither POSSUM scoring items nor mortality/morbidity status was available. When a single data point was missing for computing the published scores (POSSUM/PPOSSUM/Cr-POSSUM or SRS score and AFC score), the score was assigned the lowest value for the given item. If more than one data item was missing, the patient was excluded (56 patients). Two other patients were excluded because no dated information was available beyond the operation date. Thus 1186 patients were available for analysis.

Outcome and score definition The two main outcomes were death within 30 postoperative days (patients alive with a follow-up of less than 30 days were considered to be alive at 30 days) and morbidity. Morbidity consisted of at least one in-hospital complication, listed according to the definitions used by Copeland and colleagues10 and Alves and co-workers24 (Table 1). Table 1

In-hospital complications after major colorectal surgery No. of patients (n = 1186)

Pulmonary complications* Urinary complications† Parietal complications‡ Cardiac complications§ Deaths Deep abdominal complications¶ Hypotension/collapse Fever of unknown origin Septicaemia Neurological complications# Catheter complication** Deep vein thrombosis Hepatocellular failure††

168 (14·2) 152 (12·8) 117 (9·9) 109 (9·2) 108 (9·1) 104 (8·8) 60 (5·1) 50 (4·2) 45 (3·8) 41 (3·5) 38 (3·2) 19 (1·6) 16 (1·3)

Values in parentheses are percentages. Some patients had more than one complication, leading to a total greater than the global 41·0 per cent morbidity rate. *Respiratory failure, bronchopulmonary infection, atelectasis, pleural effusion, glottis oedema, bronchospasm and Mendelson’s syndrome. †Acute renal failure and urinary infection. ‡Abscess, wound dehiscence, wound cellulitis and wound haemorrhage. §Cardiac failure, myocardial infarction, pulmonary embolism, cardiac arrhythmia and hypertensive crisis. ¶Intra-abdominal collection, peritonitis, anastomotic leak and haemorrhage. #Cerebrovascular accident and encephalopathy. **Infection and thrombosis. ††Jaundice and ascites.

www.bjs.co.uk

British Journal of Surgery 2010; 97: 396–403

398

P. Tran Ba Loc, S. Tezenas du Montcel, J. J. Duron, H. Levard, B. Suc, B. Descottes et al.

The POSSUM and P-POSSUM scores are the sum of a physiological score (12 variables) and an operative severity score (six variables)10,13 . The Cr-POSSUM score, the sum of the physiological score (six variables) and an operative severity score (four variables), was calculated as defined by Tekkis and colleagues14 . Malignancy status was classified as follows: local cancer, nodal metastatic cancer and distant metastatic cancer were considered as Dukes’ A/B, C and D tumours respectively. The AFC score uses four simple dichotomous variables: age no greater than or greater than 70 years, weight loss no greater than or greater than 10 per cent within 6 months, presence or absence of neurological co-morbidity, and emergency or elective surgery. This score ranges from 0 to 4. The SRS score uses three simple variables: mode of presentation, as described by the Confidential Enquiry into Perioperative Deaths (CEPOD) classification; operative severity, as described by the British United Provident Association (BUPA) classification; and American Society of Anesthesiologists (ASA) score. For the mode of presentation and operative severity, information collected for the POSSUM score was used as follows: elective, urgent and emergency procedures were considered as CEPOD 2, 3 and 4, and major and major-plus operative severities were considered as BUPA 4 and 5 (major plus and major complex respectively).

Development of E-POSSUM To design and validate the E-POSSUM score, the sample was randomly divided into two groups: two-thirds of the initial population (791 patients) was used as an analysis sample, and the remaining 395 patients were used as a validation sample. A POSSUM score without age was then calculated for each patient in the analysis sample. The age classes of the POSSUM score were then redefined to fit to people aged 65 years or older, according to the WHO age classification (65–74, 75–84 and 85 years or older)1 , and dummy variables were consequently created. To define each class coefficient according to the POSSUM scoring grid, logistic regression analysis modelling the risk of death was carried out on the analysis sample. This used the POSSUM score without age plus the WHO classifications of age (as dummy variables) as covariables, leading to the following coefficients: 3·2 for the 75–84year-old group and 10·4 for the group aged 85 years or more, with the 65–74-year-old group as the reference. For morbidity, a similar analysis was performed, leading to the following coefficients: 1·3 and 2·2 for the 75–84-year-old group and the group aged 85 years or more respectively. Copyright  2010 British Journal of Surgery Society Ltd Published by John Wiley & Sons Ltd

In order to keep the same weights as in the original POSSUM, the coefficients for the 75–84-year-old group both for mortality and morbidity were rounded to 4 and the coefficients for the group aged 85 years or more were rounded to 8. Finally, using these weighted age classes, the physiological score (PS) and operative severity score (OS) were recalculated and logistic regression analysis was performed on the analysis sample to obtain new equations to predict mortality and morbidity risks (R). The resulting equation of the E-POSSUM for mortality was ln(R/1 − R) = −7·6942 + (0·1399 × PS) + (0·1126 × OS), and the equation for morbidity was ln(R/1 − R) = −3·3526 + (0·0779 × PS) + (0·0949 × OS).

Assessment of the validity of the scores and statistical analysis The E-POSSUM, P-POSSUM, Cr-POSSUM, AFC and SRS scores for mortality, and the E-POSSUM, POSSUM and AFC scores for morbidity were compared. The validity of the scores was assessed using the area under the receiver operating characteristic (ROC) curve (AUC) (discrimination)25 and calibration26 . The AUC reflects the capacity of the score to be higher for dead patients or those with complications than for alive patients or those without complications. The AUC was calculated with 95 per cent confidence intervals for all scores and compared using the non-parametric pairwise test described by DeLong and co-workers27 . Pairwise comparisons were adjusted for multiple testing using the Bonferroni correction (Pc ). Calibration reflects the ability of the score to predict the risk of death or complications correctly for an individual patient or a small group of patients. Concerning the calibration assessment, the null hypothesis is that observed and expected outcomes match (the higher the P value, the better the calibration). For calibration validation, the Hosmer–Lemeshow test was used. This test compares, by deciles of risk, the number of observed and expected events (death or morbidity). Because there were not enough different classes of risk predicted by the SRS, only four classes of risk were used to assess its calibration. Given that the AFC score provides no risk prediction equation, logistic regression analysis was carried out to compute a risk formula, which was then used to assess calibration. In accordance with the method of Wijesinghe and colleagues28 , linear analysis using decile ranges of risk was used to assess mortality for the E-POSSUM, P-POSSUM and Cr-POSSUM, and an exponential analysis using increasing ranges of risk was used to assess morbidity for the POSSUM. For all scores, AUC and calibration were assessed on the validation sample. The results are given as mean(s.d.). For all tests, statistical www.bjs.co.uk

British Journal of Surgery 2010; 97: 396–403

Score for predicting mortality and morbidity after colorectal surgery in the elderly

significance was set at P < 0·050. Statistical analyses were performed using SAS software version 9.1 (SAS Institute, Cary, North Carolina, USA).

399

1·0

0·8

Demographics The cohort comprised 1186 patients with a mean(s.d.) age of 75·7(7·0) (median 75·0, range 65–98) years. The 30-day postoperative mortality rate was 9·1 per cent (108 patients), and the in-hospital mortality rate was 9·0 per cent (106 patients). The mean(s.d.) time between operation and death was 10·5(7·8) (range 0–29) days. The in-hospital morbidity rate was 41·0 per cent (486 patients), primarily

Sensitivity

Results 0·6

0·4

0·2

0

0·2

Scores in the analysis and validation groups Analysis sample (n = 791)

E-POSSUM P-POSSUM Cr-POSSUM AFC score SRS score

Validation sample (n = 395)

35·4(10·3) (21–95) 35·8(9·6) (22–91) 19·9(3·8) (15–40) 1·1(0·8) (0–3) 8·7(1·1) (7–13)

0·4

0·6

0·8

1·0

1 − Specificity

a

Values are mean(s.d.) (range). E/P/Cr-POSSUM, Elderly/Portsmouth/Colorectal Physiological and Operative Severity Score for the enUmeration of Mortality and morbidity; AFC, Association Fran¸caise de Chirurgie; SRS, Surgical Risk Scale.

Mortality 1·0

35·4(9·4) (21–75) 35·8(8·8) (22–71) 19·8(3·6) (15–35) 1·0(0·7) (0–3) 8·7(1·0) (7–13)

0·8

Sensitivity

Table 2

E-POSSUM P-POSSUM Cr-POSSUM SRS score AFC score

0·6 E-POSSUM POSSUM AFC score

0·4

Table 3 Discrimination and calibration of the studied scores for predicting mortality and morbidity (validation sample)

0·2 Discrimination AUC Mortality E-POSSUM P-POSSUM Cr-POSSUM AFC score SRS score Morbidity E-POSSUM POSSUM AFC score

Calibration P*

O/E ratio

P†

1·35 1·23 1·06 1·00 1·08

0·178 0·584 0·655 0·135 0·300

0·97 1·22 1·00

0·166 < 0·001 0·016

0

0·017 0·86 (0·81, 0·92) 0·86 (0·81, 0·92) 0·81 (0·73, 0·88) 0·76 (0·68, 0·84) 0·78 (0·70, 0·86) < 0·001 0·77 (0·72, 0·82) 0·75 (0·70, 0·80) 0·63 (0·58, 0·69)

Values in parentheses are 95 per cent confidence intervals. AUC, area under the receiver operating characteristic curve; O/E, observed/expected; E/P/Cr-POSSUM, Elderly/Portsmouth/Colorectal Physiological and Operative Severity Score for the enUmeration of Mortality and morbidity; AFC, Association Fran¸caise de Chirurgie; SRS, Surgical Risk Scale. *P of the test of differences between scores for the AUC as described by DeLong and co-workers27 . Pairwise significant differences between scores: mortality: E-POSSUM > AFC (Pc = 0·014); P-POSSUM > AFC (Pc = 0·015); morbidity: E-POSSUM > AFC (Pc < 0·001); POSSUM > AFC (Pc < 0·001). †Hosmer–Lemeshow test.

Copyright  2010 British Journal of Surgery Society Ltd Published by John Wiley & Sons Ltd

0·2

0·4

0·6

0·8

1·0

1 − Specificity

b

Morbidity

Fig. 1 Receiver operating characteristic (ROC) curves for scores predicting a mortality and b morbidity in the validation sample. E/P/Cr-POSSUM, Elderly/Portsmouth/Colorectal Physiological and Operative Severity Score for the emUmeration of Mortality and morbidity; SRS, Surgical Risk Scale; AFC, Association Fran¸caise de Chirurgie

due to pulmonary (14·2 per cent) or urinary (12·8 per cent) complications (Table 1). Fourteen patients (1·2 per cent) died without obvious complications. For the 133 patients (11.2 per cent) lost to follow-up before 30 days following surgery, who were considered alive at 30 days, the mean time between operation and the date of lost to follow-up was 19.4 days. For these patients, the expected number of www.bjs.co.uk

British Journal of Surgery 2010; 97: 396–403

400

Table 4

P. Tran Ba Loc, S. Tezenas du Montcel, J. J. Duron, H. Levard, B. Suc, B. Descottes et al.

Calibration of the E-POSSUM for predicting mortality and morbidity (validation sample) Mortality*

Decile range of risk 1 2 3 4 5 6 7 8 9 10

Morbidity†

No. of patients

Mean risk

Observed deaths

Expected deaths

No. of patients

Mean risk

Observed complications

Expected complications

40 38 39 41 39 40 40 39 40 39

0·01 0·01 0·02 0·02 0·03 0·04 0·06 0·08 0·14 0·38

0 0 0 2 2 2 2 3 12 19

0 0 1 1 1 2 2 3 6 15

39 40 41 36 41 40 40 40 39 39

0·20 0·24 0·28 0·32 0·36 0·41 0·47 0·53 0·61 0·77

6 6 8 10 9 20 15 24 29 32

8 10 11 11 15 16 19 21 24 30

42

31

159

164

Total

*χ2 = 13·90, 10 d.f., P = 0·178; †χ2 = 14·16, 10 d.f., P = 0·166.

deaths at 30 days (computed with the survival probability of the patients not lost to follow-up) was 2.8 deaths (2.1 per cent of the patients lost to follow-up). The observed values for the different scores are shown in Table 2. For predicting mortality, the E-POSSUM, P-POSSUM and Cr-POSSUM showed a good discrimination (AUC at least 0·80), whereas the AFC and SRS scores showed a reasonable discrimination (AUC at least 0·70) (Table 3, Fig. 1a). These five AUCs were significantly different (P = 0·017). The E-POSSUM and P-POSSUM had the highest AUC (both 0·86). They were both significantly higher than that of the AFC score (Pc = 0·014 and 0·015 respectively), but did not differ from the other scores. The hypothesis of good calibration was not rejected for any score (Table 3). Nevertheless, the E-POSSUM and P-POSSUM both underestimated mortality, the EPOSSUM showing the greatest underestimation. Table 4 shows the calibration of the E-POSSUM in detail. For predicting morbidity, the AUCs were globally lower than those calculated for predicting mortality, the highest AUC being that of the E-POSSUM (0·77) (Table 3, Fig. 1b). The E-POSSUM and POSSUM AUCs were both significantly higher than that of the AFC score (both Pc < 0·001). The E-POSSUM had a good calibration, with a fairly well balanced morbidity estimation across risk classes (Table 4), whereas the hypothesis of good calibration of the POSSUM and AFC score was rejected (P < 0·001 and P = 0·016 respectively). For POSSUM, no underestimation or overestimation could be stated, probably because of the exponential analysis of the data. The AFC score had an overall observed/expected (O/E) ratio of 1·00, although it underestimated the morbidity in

low-risk (first quartile O/E = 1·31) and high-risk (last quartile O/E = 1·09) patients and overestimated it in middle-risk patients (second and third quartiles O/E = 0·87).

Copyright  2010 British Journal of Surgery Society Ltd Published by John Wiley & Sons Ltd

www.bjs.co.uk

Discussion

Various scoring systems have been introduced over the past few decades to allow objective comparisons of quality of care; however, to date, the E-POSSUM is the only score dedicated to major colorectal surgery in the elderly. The POSSUM and its derived scores (P-POSSUM and the SRS score) were developed and validated for general surgery13,16 . The Cr-POSSUM14 and the AFC score18 fit to subspecialties of colorectal surgery. Furthermore, most of the scores (the POSSUM, P-POSSUM, SRS and the AFC score) were designed and tested for predicting mortality, whereas only the POSSUM is able to predict morbidity10 . Nevertheless, none of these scores is useful in predicting postoperative outcome, as some of the items included in these scores are monitored during the perioperative period. A comparison of other scores with the results obtained with the E-POSSUM showed that, for predicting mortality, the E-POSSUM had, along with the P-POSSUM, the best discrimination, and that it also showed a good calibration. According to the literature, the POSSUM has been found to overestimate mortality in low-risk patients12,29 . The P-POSSUM13 , developed to improve this drawback, was reported to predict mortality better than the POSSUM but to underestimate the risk in the elderly and in patients undergoing emergency surgery. Conversely, it also overestimates the risk in an elective British Journal of Surgery 2010; 97: 396–403

Score for predicting mortality and morbidity after colorectal surgery in the elderly

401

surgery setting and in young people14,17 . Regarding the Cr-POSSUM, many workers have reported that it predicts mortality better than the POSSUM and P-POSSUM30 – 32 . Nevertheless, Horzic and co-workers33 reported that the Cr-POSSUM slightly underestimated mortality in lowrisk patients. The validation of the AFC score showed a good discrimination (AUC = 0·89) and a good and well-balanced calibration in colonic surgery18 , but the low number of classes of risk, owing to the low number of items used to calculate this score, makes the relevance of the calibration uncertain. The SRS score has been shown to have a good discrimination (AUC = 0·95) for predicting mortality, and its good calibration indicates that the model fits the data well16,34 . Other scoring systems for mortality prediction in surgery for colorectal cancer20,21 or in elderly patients19 were not compared with the E-POSSUM owing to their specificity and the difference in the collected variables. Regarding morbidity, the E-POSSUM showed the best results both for discrimination and calibration. Nevertheless, the globally low discriminating power of the tested scores in predicting morbidity may be due to the lack of a common definition for morbidity. Indeed, the present authors used a definition of morbidity close to that used by Copeland and colleagues10 for the POSSUM (with two additional items); however, their item ‘other or any other complication’ makes any comparison difficult. The following factors should be taken into account. First, the study was based on recent multicentre and 3-year enrolment (2002–2004), such that the mortality and morbidity rates should be representative of the current workup of major colorectal surgery in elderly patients. Second, the elderly population is increasing2 , and colorectal surgery is a major issue. Thus, in this setting, the E-POSSUM may be useful for comparisons among hospitals, surgical teams and/or a surgeon’s personal performance35 . Third, postoperative mortality was defined as death occurring within 30 days of surgery, which is a classical outcome measure for such studies12,28 . This endpoint varies from one study to another (in-hospital death13,14,16 and mortality 6 weeks10 or 4 months18 after surgery have been used as endpoints). In-hospital mortality is frequently used13,14,16 for practical reasons, but is not always a good reflection of the quality of care because it can be artificially reduced by discharging patients shortly after operation. This could potentially lead auditors to miss the deaths that occurred shortly thereafter. Fourth, even though other scores rarely deal with morbidity, it is natural to assume that morbidity is associated with an increase in immediate postoperative mortality rate. Interestingly, however, it also increases the long-term mortality rate,

particularly for elderly people36 . Finally, the E-POSSUM, based on a 3-year cohort encompassing all categories of major digestive surgery, was shown to be more accurate than the AFC score in predicting mortality and morbidity. This could be a reflection of the patients used to construct the AFC score, as only patients with colonic cancers and diverticulitis were included, eliminating benign tumours and specific colitis and taking into account both highand low-risk patients. Furthermore, the construction and validation cohorts for the AFC score were recruited during a 4-month summer period, which may have led to a seasonal bias37 . Some limitations to the present study should be noted. First, like many other scores10,13,14,16 , the E-POSSUM score includes intraoperative factors and cannot be classified as a strict preoperative predictive tool, unlike the otherwise less accurate score of Alves and colleagues18 . Second, patients with less than a 30-day follow-up were assumed to have been alive at 30 days, and this may have led to an underestimation of the actual mortality rate. The authors assume that the actual mortality rate was not significantly underestimated as the expected number of deaths at 30 days for these patients was very small. Third, the score did not include geriatric-specific items such as malnutrition or depression, which may affect morbidity and consequently justify a systematic geriatric preoperative assessment in elderly patients. Fourth, although the E-POSSUM score was validated on a sample different from that used to create the model, these two samples came from the same initial population. An external validation is needed to confirm the present findings. Lastly, like many scores, the E-POSSUM is useful for comparing performance and not in decision-making, whereas the decision to operate or not is a major issue in the elderly. Among the different tested scores, the E-POSSUM showed both the greatest discrimination and the greatest calibration in predicting mortality and morbidity in elderly patients undergoing major colorectal surgery. This tool appears to be effective for comparing the performance of both healthcare institutions and individual surgeons, but these results should be confirmed by external validation.

Copyright  2010 British Journal of Surgery Society Ltd Published by John Wiley & Sons Ltd

www.bjs.co.uk

Acknowledgements

Contributing surgeons and participating units: D. Brassier, (Aulnay/Bois); D. Collet (Pessac/Bordeaux); L. De Calan (Tours); G. Decker (Luxembourg); C. Bakoto (Chateauroux); B. Descottes (Limoges); F. Demaizieres (Paray-le-Monial); B. Desrousseaux British Journal of Surgery 2010; 97: 396–403

402

P. Tran Ba Loc, S. Tezenas du Montcel, J. J. Duron, H. Levard, B. Suc, B. Descottes et al.

(Lomme); G. Desvignes (Montargis Amilly); C. Dillin (Thonon les Bains); C. Ducerf (Lyon); S. Evrard (Bordeaux); X. Fabre (Cholet); Y. Flamant (Colombes); A. Fingerhut (Poissy); G. Fourtanier (Toulouse); P. Gabelle (Grenoble); A. Gaignant (Limoges); B. Gayet (Paris); H. Hennet (Romorantin); P. Herbiere (Albi); M. Herjean (Valenciennes); A. Ianelli (Menton); D. Jaeck (Strasbourg); G. Kohlmann (Corbeil-Essonne); Y. Laborde (Pau); P.A. Lehur (Nantes); O. Langlois-Zantain (Montlu¸con); G. Leynaud (Montlu¸con); F. Merad (Eaubonne Alger); F. Michot (Rouen); P. Oberlin (Villeneuve St Georges); E. Pellissier (Besan¸con); P. Pessaux (Angers); P. Peyrard (Compiegne); O. Philippe (Orange); J. Pujol (Bergerac); J.-M. Regimbeau (Amiens); C. Rey (Vernon); M. Sage (Auxerre); P. Segol (Caen); E. Tarla (Cannes); M. Tison (Dunkerque); J.-P. Triboulet (Lille); K. Troalen (Gonesse); B. Vacher (Argenteuil); M. Veyri`eres (Pontoise). This study was supported by three grants: Haute Autorit´e de Sant´e (HAS) GJ/SF/109-05 PR 01-010; Programme Hospitalier de Recherche Clinique (PHRC) Appel d’Offre R´egional (AOR) 03-01; and Ligue Nationale contre le Cancer (LNCC) PRC-2003-LNCC/JMH1. The authors declare no conflict of interest.

9

10 11

12

13

14

15

References 1 World Health Organization. Centre for Health Development. 2001 Ageing and Health Technical Report/WHO Kobe Centre for Health Development. http://www.who.or.jp/ [accessed 29 June 2009]. 2 Institut National de la Statistique et des Etudes Economiques. R´esultats du Recensement de la Population–2006. http://www.recensement.insee.fr/home.action [accessed 29 June 2009]. 3 Etzioni DA, Liu JH, O’Connell JB, Maggard MA, Ko CY. Elderly patients in surgical workloads: a population-based analysis. Am Surg 2003; 69: 961–965. 4 McNicol L, Story DA, Leslie K, Myles PS, Fink M, Shelton AC et al. Postoperative complications and mortality in older patients having non-cardiac surgery at three Melbourne teaching hospitals. Med J Aust 2007; 186: 447–452. 5 Polanczyk CA, Marcantonio E, Goldman L, Rohde LE, Orav J, Mangione CM et al. Impact of age on perioperative complications and length of stay in patients undergoing noncardiac surgery. Ann Intern Med 2001; 134: 637–643. 6 Pofahl WE, Pories WJ. Current status and future directions of geriatric general surgery. J Am Geriatr Soc 2003; 51(Suppl): S351–S354. 7 Colorectal Cancer Collaborative Group. Surgery for colorectal cancer in elderly patients: a systematic review. Lancet 2000; 356: 968–974. 8 Alves A, Panis Y, Mathieu P, Kwiatkowski F, Slim K, Mantion G. Mortality and morbidity after surgery of mid and

Copyright  2010 British Journal of Surgery Society Ltd Published by John Wiley & Sons Ltd

16

17

18

19

20

21

22

low rectal cancer. Results of a French prospective multicentric study. Gastroenterol Clin Biol 2005; 29: 509–514. Finlayson EV, Birkmeyer JD. Operative mortality with elective surgery in older adults. Eff Clin Pract 2001; 4: 172–177. Copeland GP, Jones D, Walters M. POSSUM: a scoring system for surgical audit. Br J Surg 1991; 78: 355–360. Whiteley MS, Prytherch D, Higgins B, Weaver PC, Prout WG. Comparative audit of colorectal resection with the POSSUM scoring system. Br J Surg 1995; 82: 425–426. Whiteley MS, Prytherch DR, Higgins B, Weaver PC, Prout WG. An evaluation of the POSSUM surgical scoring system. Br J Surg 1996; 83: 812–815. Prytherch DR, Whiteley MS, Higgins B, Weaver PC, Prout WG, Powell SJ. POSSUM and Portsmouth POSSUM for predicting mortality. Physiological and Operative Severity Score for the enUmeration of Mortality and morbidity. Br J Surg 1998; 85: 1217–1220. Tekkis PP, Prytherch DR, Kocher HM, Senapati A, Poloniecki JD, Stamatakis JD et al. Development of a dedicated risk-adjustment scoring system for colorectal surgery (colorectal POSSUM). Br J Surg 2004; 91: 1174–1182. Bromage SJ, Cunliffe WJ. Validation of the CR-POSSUM risk-adjusted scoring system for major colorectal cancer surgery in a single center. Dis Colon Rectum 2007; 50: 192–196. Sutton R, Bann S, Brooks M, Sarin S. The Surgical Risk Scale as an improved tool for risk-adjusted analysis in comparative surgical audit. Br J Surg 2002; 89: 763–768. Slim K, Panis Y, Alves A, Kwiatkowski F, Mathieu P, Mantion G. Predicting postoperative mortality in patients undergoing colorectal surgery. World J Surg 2006; 30: 100–106. Alves A, Panis Y, Mantion G, Slim K, Kwiatkowski F, Vicaut E. The AFC score: validation of a 4-item predicting score of postoperative mortality after colorectal resection for cancer or diverticulitis: results of a prospective multicenter study in 1049 patients. Ann Surg 2007; 246: 91–96. Lloyd H, Ahmed I, Taylor S, Blake JR. Index for predicting mortality in elderly surgical patients. Br J Surg 2005; 92: 487–492. Fazio VW, Tekkis PP, Remzi F, Lavery IC. Assessment of operative risk in colorectal cancer surgery: the Cleveland Clinic Foundation colorectal cancer model. Dis Colon Rectum 2004; 47: 2015–2024. Heriot AG, Tekkis PP, Smith JJ, Cohen CR, Montgomery A, Audisio RA et al. Prediction of postoperative mortality in elderly patients with colorectal cancer. Dis Colon Rectum 2006; 49: 816–824. Ferjani AM, Griffin D, Stallard N, Wong LS. A newly devised scoring system for prediction of mortality in patients with colorectal cancer: a prospective study. Lancet Oncol 2007; 8: 317–322.

www.bjs.co.uk

British Journal of Surgery 2010; 97: 396–403

Score for predicting mortality and morbidity after colorectal surgery in the elderly

23 Aust JB, Henderson W, Khuri S, Page CP. The impact of operative complexity on patient risk factors. Ann Surg 2005; 241: 1024–1027. 24 Alves A, Panis Y, Mathieu P, Mantion G, Kwiatkowski F, Slim K. Postoperative mortality and morbidity in French patients undergoing colorectal surgery: results of a prospective multicenter study. Arch Surg 2005; 140: 278–283. 25 Hanley JA, McNeil BJ. The meaning and use of the area under a receiver operating characteristic (ROC) curve. Radiology 1982; 143: 29–36. 26 Lemeshow S, Hosmer DW Jr. A review of goodness of fit statistics for use in the development of logistic regression models. Am J Epidemiol 1982; 115: 92–106. 27 DeLong ER, DeLong DM, Clarke-Pearson DL. Comparing the areas under two or more correlated receiver operating characteristic curves: a nonparametric approach. Biometrics 1988; 44: 837–845. 28 Wijesinghe LD, Mahmood T, Scott DJ, Berridge DC, Kent PJ, Kester RC. Comparison of POSSUM and the Portsmouth predictor equation for predicting death following vascular surgery. Br J Surg 1998; 85: 209–212. 29 Chandra A, Mangam S, Marzouk D. A review of risk scoring systems utilised in patients undergoing gastrointestinal surgery. J Gastrointest Surg 2009; 13: 1529–1538. 30 Senagore AJ, Warmuth AJ, Delaney CP, Tekkis PP, Fazio VW. POSSUM, p-POSSUM, and Cr-POSSUM: implementation issues in a United States health care system

Copyright  2010 British Journal of Surgery Society Ltd Published by John Wiley & Sons Ltd

403

for prediction of outcome for colon cancer resection. Dis Colon Rectum 2004; 47: 1435–1441. ¨ 31 Tez M, Yolda¸s O, Gocmen E, Kulah B, Koc M. Evaluation of P-POSSUM and CR-POSSUM scores in patients with colorectal cancer undergoing resection. World J Surg 2006; 30: 2266–2269. 32 Constantinides VA, Tekkis PP, Senapati A; Association of Coloproctology of Great Britain and Ireland. Comparison of POSSUM scoring systems and the surgical risk scale in patients undergoing surgery for complicated diverticular disease. Dis Colon Rectum 2006; 49: 1322–1331. 33 Horzic M, Kopljar M, Cupurdija K, Bielen DV, Vergles D, Lackovic Z. Comparison of P-POSSUM and Cr-POSSUM scores in patients undergoing colorectal cancer resection. Arch Surg 2007; 142: 1043–1048. 34 Brooks MJ, Sutton R, Sarin S. Comparison of Surgical Risk Score, POSSUM and p-POSSUM in higher-risk surgical patients. Br J Surg 2005; 92: 1288–1292. 35 Copeland GP. The POSSUM system of surgical audit. Arch Surg 2002; 137: 15–19. 36 Khuri SF, Henderson WG, DePalma RG, Mosca C, Healey NA, Kumbhani DJ et al. Determinants of long-term survival after major surgery and the adverse effect of postoperative complications. Ann Surg 2005; 242: 326–341. 37 Englesbe MJ, Pelletier SJ, Magee JC, Gauger P, Schifftner T, Henderson WG et al. Seasonal variation in surgical outcomes as measured by the American College of Surgeons–National Surgical Quality Improvement Program (ACS-NSQIP). Ann Surg 2007; 246: 456–462.

www.bjs.co.uk

British Journal of Surgery 2010; 97: 396–403