An assessment of the POSSUM system in
orthopaedic surgery
K. Mohamed, G. P. Copeland, D. A. Boot, H. C. Casserley,
I. M. Shackleford, P. G. Sherry, G. J. Stewart
From Warrington Hospital, England
e describe the development and validation of a
W scoring system for auditing orthopaedic surgery.
It is a minor modication of the POSSUM scoring
system widely used in general surgery. The
orthopaedic POSSUM system which we have
developed gives predictions for mortality and
morbidity which correlate well with the observed rates
in a sample of 2326 orthopaedic operations over a
period of 12 months.
J Bone Joint Surg [Br] 2002;84-B:735-9.
Received 11 July 2001; Accepted after revision 17 December 2001
The assessment of outcome after surgical intervention is
not a new science. As early as 1750 BC King Hammurabi
of Babylon issued a number of decrees relating to surgeons
and their surgery. The most infamous of these codices was
that if a surgeon operated on a free man and the patient
became blind or worse still died, the surgeon should have
his operating hand cut off. While, to some a modication of
this codex may still seem to be in operation, many have
attempted to devise more reliable and robust methods for
1-3
assessing the outcome of surgical intervention.
The use of raw data on mortality and morbidity from
different units produces disparities in outcome which may
be explained by variations in the case mix and in tech-
4,5
niques of surgery. Methods have been devised to allow
comparison between units by taking the physiological sta-
tus of the patients and the operative complexity into
1,6,7
account.
In general surgery the POSSUM and P-POSSUM sys-
tems have proved to be the most reliable and widely
K. Mohamed, FRCS, Orthopaedic Registrar
G. P. Copeland, MCh, Consultant General Surgeon
D. A. Boot, FRCS, Consultant Orthopaedic Surgeon
H. C. Casserley, FRCS, Consultant Orthopaedic Surgeon
I. M. Shackleford, FRCS, Consultant Orthopaedic Surgeon
G. J. Stewart, FRCS, Consultant Orthopaedic Surgeon
P. G. Sherry, FRCS, Consultant Orthopaedic Surgeon
Warrington Hospital, Lovely Lane, Warrington, Cheshire WA5 1QG,
UK.
Correspondence should be sent to Mr G. P. Copeland.
2002 British Editorial Society of Bone and Joint Surgery
0301-620X/02/512626 $2.00
VOL. 84-B, NO. 5, JULY 2002
1,8
applicable scoring methods so far devised. Both are been
found to be equally applicable to subspecialties including
vascular surgery, surgical gastroenterology and urology and
9-11
are used by many health-care organisations.
A number of orthopaedic studies have drawn attention to
the variability in outcome after surgery, in particular for
fracture of the neck of the femur, and it has been argued
that this is more likely to be related to the case mix than to
12,13
the hospital facilities or the surgeons. A recent report
has cautioned that predicting the outcome of orthopaedic
intervention in the elderly injured patient, merely on the
basis of the injury severity score, can be hazardous since
14
host factors appear to be of greater importance.
A system which is weighted towards physiological status
would appear to be of potential benet in assessing the
outcome after orthopaedic surgery. Such a system would
allow comparison based on the patients physiological sta-
tus and an assessment of the magnitude of surgery and its
timing.
Between 1996 and 1998 we used methods similar to
those which we have described previously in the develop-
1
ment of the POSSUM system, in order to design a severity
score which would allow the logistic regression equation
used in general surgery to be applied to orthopaedics. We
assessed 22 surgical severity factors which were later
reduced by multivariate analysis to the minimum number
necessary to produce an accurate estimate of mortality and
morbidity. The resulting operative severity score was sim-
ilar to the system for general surgery, although individual
factors and weightings differed. In this study we have now
attempted to validate the application of this new method.
Patients and Methods
Every patient admitted to Warrington Hospital over a
period of 12 months on whom orthopaedic surgery had
been performed, was assessed using the new orthopaedic
POSSUM system. Day-care patients or those treated on an
outpatient basis were excluded.
The system includes a physiological assessment and an
analysis of operative severity (Table I). The former has
12 variables each of which is divided into four grades
with an exponentially increasing score value (1, 2, 4 and
8). Almost all the score variables were available for every
735
736 K. MOHAMED, G. P. COPELAND, D. A. BOOT, H. C. CASSERLEY, I. M. SHACKLEFORD, P. G. SHERRY, G. J. STEWART

Table I. Physiological and operative severity assessment in the orthopaedic POSSUM system
Physiological score Operative severity score
1 2 4 8 1 2 4 8
Age (yrs) <60 61 to 70 >71 Magnitude Minor Inter Major Major+
Cardiac signs Normal On cardiac drugs Oedema Raised JVP* Number of operative 1 2 >2
or steroid Warfarin variables within 30 days
Chest radiograph Normal Borderline Cardiomegaly Blood loss per <100 101 to 500 501 to 999 >1000
cardiomegaly operation (ml)
Resp signs Normal SOB exertion SOB stairs SOB rest Contamination None Incised wound, Minor contamination Gross contamination
i.e. stab or necrotic tissue or necrotic tissue
Chest radiograph Normal Mild COAD Mod COAD Any other Presence of None 10 Node metastases Distant
change malignancy metastases
Systolic BP 110 to 130 131 to 170 >171 <89 Timing of Elective Emergency. Emergency.
(mmHg) 100 to 109 90 to 99 operation Resuscitation Immediate
possible <48 hrs <6 hrs
Pulse 50 to 80 81 to 100 101 to 120 >121
(/min) 40 to 49 <39
Coma score 15 12 to 14 9 to 11 <8
Blood urea <7.5 7.6 to 10 10.1 to 15 >15.1
(mmol/l)
Blood Na >136 131 to 135 126 to 130 <125
(mmol/l)
Blood K 3.5 to 5 3.2 to 3.4 2.9 to 3.1 <2.8
(mmol/l) 5.1 to 5.3 5.4 to 5.9 >6
Hb (g/100ml) 13 to 16 11.5 to 12.9 10 to 11.4 <9.9
16.1 to 17 17.1 to 18 >18.1
White cell count 4 to 10 10.1 to 20 >20.1
(1012/l) 3.1 to 3.9 <3
ECG Normal AF (60 to 90) Any other
change
*jugular venous pressure
shortness of breath
chronic obstructive airways disease
atrial brillation

patient, but when a figure was missing, a score of 1 was
allocated. The operative severity score includes six vari-
ables, each divided into four grades, with an exponen-
tially increasing value (Table I). Definitions of operative
complexity are given in Table II. When an operation was
not listed the most nearly similar group was chosen. The
number of operative variables occurring within 30 days
of the surgery demonstrates the chronology of proce-
dures. The operative severity score was available for all
patients, although histological confirmation was required
to complete the scores for some.
The outcome was assessed as 30-day morbidity and
mortality, which allowed comparability with the system for
general surgery and could be checked against the 30-day
mortality statistics of the Department of Health. The pres-
ence of the following complications were recorded as
morbidity.
1) Infection: chest, urinary, wound, joint, bony, septicaemia
or pyrexia of unknown origin.
2) Haemorrhage: deep or supercial.
3) Other wound problems: seroma requiring drainage,
dehiscence.
4) Thromboembolic complications: deep-venous throm-
bosis, pulmonary embolus, cerebrovascular accident.
5) Cardiac: failure, myocardial infarction, arrhythmia.
6) Respiratory: failure, atelectasis, pneumonia.
7) Renal: failure, retention of urine.
8) Unanticipated displacement of an implant.
1
Exact denitions have been described previously. Pre-
dictions of mortality and morbidity for individual patients
were estimated using the following equations, in which R1
relates to mortality and R2, morbidity.
Loge R1/(1-R1) = -7.04 + (0.13 physiological score) +
(0.16 operative severity score)
Loge R2/(1-R2) = -5.91 + (0.16 physiological score) +
(0.19 operative severity score)
Results
Over a period of 12 months there were 2326 orthopaedic
operations of which 44% were elective and 56% urgent or
emergency procedures. The overall rate of mortality was
2.2%, of which 0.2% were in the elective group and 3.8%
in the urgent or emergency group. The overall rate of
morbidity was 10.8%, 4.4% for elective and 15.9% for
urgent or emergency cases. Table III gives the prole of
complications.
The POSSUM logistic regression equations yield an
overall predicted mortality of 53 patients (versus 51
observed) and a predicted morbidity in 254 patients (versus
252 observed). The risk spectra for both mortality and
morbidity are shown in Table IV, and follow the expected
exponential distribution.
The number of operations performed by each of the six
surgeons contributing to the study is shown in Table V. As
can be seen there are differences in each surgeons overall
THE JOURNAL OF BONE AND JOINT SURGERY
 AN ASSESSMENT OF THE POSSUM SYSTEM IN ORTHOPAEDIC SURGERY 737

Table II. Operative complexity for the operative severity score Table III. The number of patients with complica-
Minor Fasciotomy tions after operation. Some patients had multiple
Ganglion/bursa complications
Tenotomy/tendon repair Complication Number
Arthroscopic surgery Haemorrhage 14
Carpal tunnel/nerve release
Removal of metal Infection
Closed reduction of fracture Chest 59
Urinary 34
Intermediate Excision/osteotomy small bone Wound 73
Minor joint replacement Septicaemia 7
Amputation digit/digits Pyrexia of unknown origin 5
Closed reduction with external xation Respiratory failure 17
Open reduction of fracture of small bone
Cardiac
Major Osteotomy long bone Hypotension 29
Ligamentous reconstruction + prosthesis Cardiac failure 23
Arthrodesis large joint Myocardial infarction 17
Major joint replacement Arrhythmia 6
Amputation limb Thrombotic
Disc surgery Deep-venous thrombosis 12
Open reduction of fracture of a long bone Pulmonary embolus 9
Major + Radical tumourectomy Cerebrovascular infarction 7
Major spinal reconstruction Limb occlusion 2
Revision prosthetic replacement, major joint Other vascular complications 4
Hindquarter/forequarter amputation Renal failure 15
Urinary retention 32
Other wound problems 10
Prosthetic problems 7
Miscellaneous 26
Table IV. Risk spectra for mortality and morbidity by number of
patients
Risk band (%) Mortality Morbidity Table V. The variability in surgeon workload, and the rates of mortality
<10 2086 1042 and morbidity (%) for each surgeon
10 to 20 140 600 Surgeon Number of procedures Mortality Morbidity
20 to 30 51 230
30 to 40 23 124 1 495 3.0 12.9
40 to 50 11 99 2 474 1.1 10.5
50 to 60 3 72 3 372 2.4 11.0
60 to 70 6 48 4 216 1.8 4.6
70 to 80 3 49 5 426 2.1 9.6
80 to 90 1 37 6 343 2.6 13.4
>90 2 25

Table VI. Comparisons between observed and predicted rates of mortal- ROC curves are illustrated in Figures 1 and 2, which show
ity and morbidity (%) for individual surgeons during the study period.
good correlation across the range.
Mortality Morbidity
Surgeon Observed Predicted Observed Predicted
1 15 15 62 64 Discussion
2 5 6 52 50
3 9 10 41 43 Worldwide public and political interest has been turned on
4 4 4 10 10 the assessment of quality of care and surgical outcome.
5 9 9 41 40
6 9 9 46 47 This is perhaps easier within the surgical specialties since
death after surgery is an obvious adverse outcome. This has
led many non-surgical clinicians to suggest that rates of
mortality are a suitable indicator of surgical prowess. There
rates for mortality and morbidity. When corrected for case is, however, increasing awareness that raw data may be
mix using the POSSUM system, however, there is little both erroneous and deceptive, while morbidity is often
difference between the observed rates and those predicted ignored. Death after orthopaedic surgery is rare and usually
by POSSUM (Table VI). follows a number of antecedent complications.
The predictive accuracy of these equations was assessed The overall rates of mortality and morbidity for our
by determining the receiver-operating-characteristic curves period of study appear to be in keeping with previously
(ROC curves) and the classication matrices for different published rates for inpatient surgery in district general
1,6
levels of predicted mortality and morbidity. The resultant hospitals and the risk prole shows similarities to those
VOL. 84-B, NO. 5, JULY 2002
738 K. MOHAMED, G. P. COPELAND, D. A. BOOT, H. C. CASSERLEY, I. M. SHACKLEFORD, P. G. SHERRY, G. J. STEWART
1.0
Sensitivity (true-positive rate)
0.8
0.6
0.4
0.2
0.2 0.4 0.6 0.8
1- Specificity (false-positive rate)
Fig. 1
ROC curve for mortality. A curve approaching the linear line
indicates no predictive ability for the assessing system. The
further from the linear line the better is the predictive ability.
1,6
of general surgery suggesting that our study group was
comparable with the average orthopaedic case mix for a
district general hospital. In general surgery, however, there
is a tendency to have a greater number of patients whose
1,4,6,7
risk of death is greater than 70%.
Our study shows a close correlation between the overall
observed rates for mortality and morbidity and the predic-
tions derived from the POSSUM logistic regression equa-
tions. The physiological variables assessed were those
shown by logistic regression analysis to be the most impor-
tant in predicting mortality and morbidity. Other additional
variables were not found independently to improve the
predictive ability of the logistic regression equation. We
have not compared the POSSUM predictions with P-POS-
SUM. It has been shown previously that if the correct
mathematical model is applied there are no signicant
15
differences in the predictive ability of either method.
There would appear at first sight to be significant differ-
ences in outcome between the six surgeons studied. Rates of
mortality varied between 1.1% and 3.0% and of morbidity
between 4.6% and 13.4%. Similar variations have been
shown for general surgery and can be explained on the basis
of the case mix and operative complexity. The orthopaedic
POSSUM predictions in our analysis suggest that a similar
explanation may account for apparently marked differences in
surgical outcome in orthopaedic surgery. This is shown by the
close correlation between observed and predicted rates for
both mortality and morbidity for individual surgeons.
The ROC curves suggest that the orthopaedic POSSUM
score is equally applicable across the spectrum of surgical
1.0
0.8
Sensitivity (true-positive rate)
0.6
0.4
0.2
1.0 0.2 0.4 0.6 0.8 1.0
1- Specificity (false-positive rate)
Fig. 2
ROC curve for morbidity.
risk and that the predictions for an individual patient may
be of benet when assessing a patient who has died or
suffered a complication. We have previously drawn atten-
tion to the benets of audit in patients who survive, but for
whom a predicted risk of death exceeds 50%, since data on
these patients are often of more value than those on patients
7 16
who die. Recently, Wilson et al have shown that poten-
tially signicant improvements in overall care are possible
by optimisation in this particular group.
Clearly, no regression equation for risk assessment should
remain static indefinitely. We have found no need for change
over the past ten years, but should dramatic changes occur in
the future the equation can be easily updated without the
need to alter the score variables. Should this occur, patients
already scored would be assessed with the present equation
and new patients with an updated format.
Our study demonstrates that POSSUM can be used as an
audit aid to assess the quality of orthopaedic care. Thus a
quality measurement similar to that used in general surgery,
namely the ratio of observed adverse events to predicted
adverse events, may be used in orthopaedic surgery and be
more sensitive than simple rates of mortality and
morbidity.
No benets in any form have been received or will be received from a
commercial party related directly or indirectly to the subject of this
article.
References
1. Copeland GP, Jones D, Walters M. POSSUM: a scoring system for
surgical audit. Br J Surg 1991;78:355-60.
THE JOURNAL OF BONE AND JOINT SURGERY
 AN ASSESSMENT OF THE POSSUM SYSTEM IN ORTHOPAEDIC SURGERY
2. Jones HJ, de Cossart L. Risk scoring in surgical patients. Br J Surg
1999:86:149-57.
3. Pillai SB, van Rij AM, Williams S, et al. Complexity and risk-adjusted
model for measuring surgical outcome. Br J Surg 1999;86:1567-72.
4. Copeland GP, Jones D, Wilcox A, Harris PL. Comparative vascular
audit using the POSSUM scoring system. Ann R Coll Surg Engl
1993;75:175-7.
5. Sagar PM, Hartley MN, MacFie J, Taylor BA, Copeland GP.
Comparison of individual surgeons performance, risk adjusted analysis
with POSSUM scoring system. Dis Colon Rectum 1996;39:654-8.
6. Copeland GP, Sagar P, Brennan J, et al. Risk adjusted analysis of
surgeon performance: a one year study. Br J Surg 1995;82:408-11.
7. Copeland GP. Assessing the surgeon: 10 years experience with the
POSSUM system. J Clin Excell 2000;2:187-90.
8. Prytherch D, Whiteley MS, Higgins B, et al. POSSUM and Ports-
mouth POSSUM for predicting mortality: physiological and operative
severity score for the enumeration of mortality and morbidity. Br J
Surg 1998;85:1217-20.
9. Brunelli A, Fianchini A, Xiume F, et al. Evaluation of the POSSUM
scoring system in lung surgery: physiological and operative severity
score for the enumeration of mortality and morbidity. Thorac Cardio-
vasc Surg 1998;46:141-6.
VOL. 84-B, NO. 5, JULY 2002
739
10. Gotohda N, Iwagaki H, Itano S, et al. Can POSSUM, a scoring
system for perioperative surgical risk, predict postoperative clinical
course? Acta Med Okayama 1998;52:325-9.
11. Tekkis PP, Kocher HM, Bentley AJ, et al. Operative mortality rates
among surgeons: comparison of POSSUM and p-POSSUM scoring
systems in gastrointestinal surgery. Dis Colon Rectum
2000;43:1528-32.
12. Withey C, Morris R, Beech R, Backhouse A. Outcome following
fractured neck of femur: variation in acute hospital care or case mix?
J Public Health Med 1995;17:429-37.
13. Thomas M, Eastwood H. Re-evaluation of two simple prognostic
scores of outcome after proximal femoral fractures. Injury
1996;27:111-5.
14. Van der Sluis CK, Timmer HW, Eisma WH, ten Duis HJ. Outcome
in elderly injured patients: injury severity versus host factors. Injury
1997;28:588-92.
15. Wijesinghe LD, Mahmood T, Scott DJ, et al. Comparison of
POSSUM and the Portsmouth predictor equation for predicting death
following vascular surgery. Br J Surg 1998;85:209-12.
16. Wilson J, Woods I, Fawcett J, et al. Reducing the risk of major
elective surgery: randomised controlled trial of preoperative optimisa-
tion of oxygen delivery. BMJ 1999;318:1099-103.