Read Core Topics in General & Emergency Surgery: Companion to Specialist Surgical Practice Online
Authors: Simon Paterson-Brown MBBS MPhil MS FRCS
Core Topics in General & Emergency Surgery: Companion to Specialist Surgical Practice (73 page)
67.
Ten Cate-Hoek, A.J., Prins, M.H., Management studies using a combination of D-dimer test result and clinical probability to rule out venous thromboembolism: a systematic review.
J Thromb Haemost
. 2005;3(11):2465–2470.
16150049
68.
Kearon, C., Julian, A.J., Newman, T.E., et al, Noninvasive diagnosis of deep vein thrombosis.
Ann Intern Med
1998;128:663–677.
9537941
69.
Stevens, S.M., Elliott, C.G., Chan, K.J., et al, Withholding anticoagulation after a negative result on duplex ultrasonography for suspected symptomatic deep venous thrombosis.
Ann Intern Med
. 2004;140(12):985–991.
15197015
70.
Cogo, A., Lensing, A.W., Prandoni, P., et al, Distribution of thrombosis in patients with symptomatic deep vein thrombosis. Implications for simplifying the diagnostic process with compression ultrasound.
Arch Intern Med
. 1993;153(24):2777–2780.
8257253
71.
Goodacre, S., Sampson, F., Stevenson, M., et al. Measurement of the clinical and cost-effectiveness of non-invasive diagnostic testing strategies for deep vein thrombosis.
Health Technol Assess
. 2006;10(15):1–168. [iii–iv].
72.
Torbicki, A., Perrier, A., Konstantinides, S., et al, Guidelines on the diagnosis and management of acute pulmonary embolism: the Task Force for the Diagnosis and Management of Acute Pulmonary Embolism of the European Society of Cardiology (ESC).
Eur Heart J
. 2008;29(18):2276–2315.
18757870
73.
Wells, P.S., Anderson, D.R., Rodger, M., et al, Derivation of a simple clinical model to categorize patients probability of pulmonary embolism: increasing the model's utility with the SimpliRED D-dimer.
Thromb Haemost
2000;83:416–420.
10744147
74.
Anderson, D.R., Kahn, S.R., Rodger, M.A., et al, Computed tomographic pulmonary angiography vs ventilation perfusion lung scanning in patients with suspected pulmonary embolism.
JAMA
. 2007;298(23):2743–2753.
18165667
75.
Van Beek, E.J.R., Brouwers, E.M.J., Song, B., et al, Lung scintigraphy and helical computed tomography for the diagnosis of pulmonary embolism: a meta-analysis.
Clin Appl Thromb Hemost
. 2001;7(2):87–92.
11292198
76.
Garcia, D., Ageno, W., Libby, E., et al, Update on the diagnosis and management of pulmonary embolism.
Br J Haematol
2005;131:301–312.
16225649
77.
Clemens, S., Leeper, K.V. Newer modalities for detection of pulmonary emboli.
Am J Med
. 2007;120(10, Suppl. 2):S2–12.
78.
Royal College of Obstetricians and Gynaecologists. Green Top Guideline No. 28. Thromboembolic disease in pregnancy and the puerperium: acute management.
79.
Büller, H., Agnelli, G., Hull, R.D., et al. Antithrombotic therapy for venous thromboembolic disease: the Seventh ACCP Conference on Antithrombotic and Thrombolytic Therapy.
Chest
. 2004;126(3, Suppl.):401S–428S.
Evidence-based guideline for the use of anticoagulants for the treatment of VTE recommending that outpatient LMWH is as safe as UFH for the management of DVT (grade I recommendation).
80.
van Dongen, C.J., van den Belt, A.G., Prins, M.H., et al. Fixed dose subcutaneous low molecular weight heparins versus adjusted dose unfractionated heparin for venous thromboembolism.
Cochrane Database Syst Rev
. (4):2004. [CD001100].
81.
Sanchez, O., Trinquart, L., Colombet, I., et al, Prognostic value of right ventricular dysfunction in patients with haemodynamically stable pulmonary embolism: a systematic review.
Eur Heart J
. 2008;29(12):1569–1577.
18495689
82.
Lee, A.Y., Levine, M.N., Baker, R.I., et al, Low-molecular-weight heparin versus a coumarin for the prevention of recurrent venous thromboembolism in patients with cancer.
N Engl J Med
. 2003;349(2):146–153.
12853587
The first study of extended LMWH anticoagulation for VTE. It demonstrated a clear reduction in recurrent thromboses in patients with VTE and cancer compared to adjusted-dose coumarin with no increase in bleeding.
83.
Shorr, A.F., Jackson, W.L., Moores, L.K., et al, Minimizing costs for treating deep vein thrombosis: the role for fondaparinux.
J Thromb Thrombolysis
. 2007;23(3):229–236.
17131172
84.
Makris, M., Watson, H.G., The management of coumarin-induced over-anticoagulation.
Br J Haematol
. 2001;114(2):271–280.
11529844
85.
Baglin, T.P., Keeling, D.M., Watson, H.G., Guidelines on oral anticoagulation (warfarin): third edition 2005 update.
Br J Haematol
2006;132:277–285.
16409292
86.
The Einstein Investigators, Oral rivaroxaban for symptomatic venous thromboembolism.
N Engl J Med
2010;363:2499–2510.
21128814
87.
The Einstein-PE investigators, Oral rivaroxaban for the treatment of symptomatic pulmonary embolism.
N Engl J Med
2012;366:1287–1297.
22449293
88.
Schulman, S., Crowther, M.A., How I treat with anticoagulants in 2012: new and old anticoagulants, and when and how to switch.
Blood
. 2012;119(13):3016–3023.
22302737
89.
Watson, L.I., Armon, M.P. Thrombolysis for acute deep vein thrombosis.
Cochrane Database Syst Rev
. (4):2004. [CD002783].
90.
Alesh, I., Kayali, F., Stein, P.D., Catheter-directed thrombolysis (intrathrombus injection) in treatment of deep venous thrombosis: a systematic review.
Catheter Cardiovasc Interv
. 2007;70(1):143–148.
17377992
91.
Segal, J.B., Streiff, M.B., Hofmann, L.V., et al, Management of venous thromboembolism: a systematic review for a practice guideline.
Ann Intern Med
. 2007;146(3):211–222.
17261856
92.
The PREPIC Study Group, Eight-year follow-up of patients with permanent vena cava filters in the prevention of pulmonary embolism: the PREPIC (Prevention du Risque d'Embolie Pulmonaire par Interruption Cave) randomized study.
Circulation
2005;112:416–422.
16009794
93.
Hutten, B.A., Prins, M.H. Duration of treatment with vitamin K antagonists in symptomatic venous thromboembolism.
Cochrane Database Syst Rev
. (1):2006. [CD001367].
94.
Schulman, S. The effect of the duration of anticoagulation and other risk factors on the recurrence of venous thromboembolisms. Duration of Anticoagulation Study Group.
Wien Med Wochenschr
. 1999;149(2–4):66–69.
95.
Baglin, T., Luddington, R., Brown, K., et al, Incidence of recurrent venous thromboembolism in relation to clinical and thrombophilic risk factors: prospective cohort study.
Lancet
2003;362:523–526.
12932383
Chris Deans
Surgical risk
is an estimation of the likelihood of an adverse event occurring as a consequence of a patient undergoing a particular surgical procedure or intervention.
Patient assessment
is a process that attempts to quantify this risk for an individual patient. The ability to undertake patient assessment to determine surgical risk for a patient is fundamental to modern surgical practice. Appropriate patient assessment and estimation of risk will inform surgical decision-making, assist patient decision-making and facilitate informed consent. It is important to remember that risk is not only associated with undertaking surgical procedures, but may also include other treatments or investigations that may pose a particular risk to the patient, for example performing a colonoscopy or interventional radiological procedure. The risks of not performing a particular procedure or intervention should also be considered and the possible implications for the patient of not undertaking a procedure should be part of any informed consent process (
Box 15.1
).
1
Box 15.1
GMC definition of risk of investigation/treatment
Side-effects
Complications
Failure of an intervention to achieve the desired aim
The potential outcome of taking no action
Estimation of surgical risk is important for several reasons. Firstly, as already stated, determining a patient's risk will influence surgical decision-making and, in turn, facilitate informed consent. This process will ultimately influence choice of treatment options for individual patients. Secondly, identifying higher risk patients will also allow appropriate pre-emptive measures to be undertaken and target particular areas of concern to optimise the patient in the perioperative period (see also
Chapter 16
). This process may also help anticipate potential adverse events. A further positive effect of this process is to aid case mix adjustment. There is increasing public release of activity/outcome figures (or ‘league tables’) in surgery, which may be crude mortality or complication rates. Case mix adjustment allows units with a greater proportion of high-risk patients to compensate for any differences in their figures with respect to national outcomes and allow for meaningful comparison of data against national audits. This will ensure quality assurance for the future (
Box 15.2
).
Box 15.2
Why assess surgical risk?
Allow informed consent
Facilitate surgical decision-making
To anticipate adverse events
To minimise risk to patients, staff and healthcare system
To allow for meaningful comparison of outcomes
Determination of surgical risk is complex and is influenced by many variables. However, the process may be simplified by thinking of the assessment according to two main factors – the patient, and those related to the surgical procedure itself.
Procedural-related risks
are generally easier to quantify where national, local and even individual complication rates may be known. The introduction of national and regional audit programmes in some specialities, as well as the improved quality of data collection in local departments, have enabled a better understanding of the more common risks and complications that are associated with many surgical procedures. However, procedural-related risk will also depend on additional factors, such as the urgency and duration of the procedure, volume of blood loss and type of surgery undertaken. The National Institute for Clinical Excellence (NICE) has attempted to stratify surgical procedures into different grades of severity in an effort to provide guidance on the use of preoperative investigations and estimate the likelihood of perioperative risk (
Table 15.1
).
2
Table 15.1
Examples of surgical procedures by severity grading (NICE)
Patient-related risk
factors are less easy to quantify. They may be broadly divided into either subjective or objective factors.
Subjective risk
assessment includes patient history, clinical examination, pattern recognition, accumulated clinical experience and ‘the end of the bed test’.
Objective risk
assessment includes formal laboratory results and assessment of comorbidity and physiological function through further investigation. Patient factors will be influenced by the ‘fitness’ of the patient (functional/performance status), age, comorbid illness, the underlying disease process and nutritional status, as well as many other inter-related variables.
Several
risk prediction models and scoring systems
have been developed in an attempt to help measure surgical risk. In addition, techniques to formally quantify levels of patient fitness (
functional assessment
), such as exercise testing, and measurements of
serum biomarkers
have also been introduced specifically to predict perioperative risk, with variable success. This chapter will discuss some of these scoring systems and functional assessment tools that may be used for patient assessment and estimation of surgical risk. These may be broadly classified into risk prediction models (general and specific), functional assessment tools and novel biomarkers.
Clinical assessment
It is the role of the surgeon as a clinician to undertake a thorough clinical assessment of every patient in order to carefully identify individual characteristics of the patient's comorbidity and underlying disease process that may influence surgical risk. Only then may a fully informed decision be made regarding treatment choices for individual patients. Some patient factors may be clearly identifiable, such as the presence of ischaemic heart disease or obesity, whereas others may not yet have been diagnosed. A thorough history and examination should be undertaken and targeted investigations requested, based on the clinical findings. This process may be difficult and more challenging situations should involve consultation with colleagues and other clinicians as part of the wider multidisciplinary team – for example, obtaining a cardiology review or asking for an anaesthetic opinion. In difficult cases a second opinion may be sought from an independent source.
There are in fact data to support the concept that ‘gut instinct’ or ‘surgical intuition’ can be more effective than formal risk prediction models in identifying the patient with a poor prognosis in the context of a particular procedure,
3
especially in the elective setting.
4
However, there is also epidemiological evidence suggesting that clinicians often fail to identify patients at high risk of complications and as a result do not allocate them to the appropriate level of perioperative care.
5
Attempts to improve the accuracy of estimation of risk and to provide a more robust quantitative assessment have led to the development of several scoring systems and risk prediction models. Many of these are freely available online.
6
These tools have been developed to objectively estimate morbidity and mortality rates for individual patients prior to the proposed intervention. It should be noted that there is no perfect tool, that the models available must not be used to guide decision-making in isolation and that there is no substitute for the combination of objective markers with surgical experience and intuition. Furthermore, the models available pertain to populations rather than individual patients, and therefore have limitations that need to be recognised before using them to inform surgical risk assessment. For example, the mortality rate for a surgical intervention in a particular population may be 5%, but for the individual patient it can only be 0% or 100%.
The scoring systems available usually incorporate physiological and comorbidity data that have been selected using logistical regression techniques in a large database of patients, which may not be similar to the local population. A
coefficient
may be assigned in order to weight the variables and the resulting equation provides a
numerical indication of risk
for the patient, although it is frequently more meaningful to the operating surgeon.
Ideally the patients in the database on which the scoring system is developed and validated should be similar to the individual patient in question. However, publication bias may mean that only the best data are published. While this has inherent flaws, it may also be viewed as an opportunity to benchmark one's own results against those from a centre of excellence.
Finally, the accuracy of predictive models is dynamic and they should be periodically retested against an evolving surgical patient population. When accuracy deteriorates they should be revised and updated.
The Physiological and Operative Severity Score for the enUmeration of Mortality and morbidity (POSSUM) was first described in 1991.
7
It was designed as a scoring system to estimate morbidity and mortality following a surgical procedure and, by including data on the patient's physiological condition, it provides a risk-adjusted prediction of outcome. This facilitates more accurate comparison of hospital or surgeon performance and it can be used as an audit and clinical governance tool. However, the model involves inclusion of operative variables, such as volume of blood loss and the presence of peritoneal soiling, which precludes its use in the preoperative setting to inform the consent process. Despite this, POSSUM is the most widely applied, validated, surgical risk scoring system in the UK and has been modified by several authors to provide speciality-specific information.
The original POSSUM score was developed after initially subjecting 62 parameters to multivariate analysis in order to determine the most powerful outcome predictors. Twelve physiological and six operative parameters were identified, and each of these factors were weighted to a value of 1, 2, 4 or 8 to simplify the calculation. It was re-evaluated in 1998 in Portsmouth, UK by Whiteley et al., who reported concerns that it overestimated mortality in their patients, particularly in the lowest risk group.
8
They modified the POSSUM formula and they constructed the Portsmouth predictor equation for mortality (P-POSSUM;
Box 15.3
). The modified formula fitted well with the observed mortality rate; however, it still overestimates mortality in low-risk groups, the elderly and in certain surgical subspecialities.
9
The latter finding has prompted the development of speciality-specific POSSUM for major elective surgery.
Box 15.3
Variables used in the calculation of the P-POSSUM score
Physiological variables
Age
Cardiac disease
Respiratory disease
Electrocardiogram (ECG)
Systolic blood pressure
Pulse rate
Haemoglobin concentration
White cell count
Serum urea concentration
Serum sodium concentration
Serum potassium concentration
Glasgow Coma Scale (GCS)
Operative variables
Operation severity class
Number of procedures
Blood loss
Peritoneal contamination
Malignancy status
Urgency
P-POSSUM formula
Ln
R/
1 −
R
= − 9.065 + (0.1692 × physiological score) + (0.1550 × operative severity score)
The value of POSSUM and P-POSSUM in predicting in-hospital mortality was examined in patients undergoing colorectal surgery in France. Both POSSUM and P-POSSUM performed well but overestimated postoperative death in elective surgery and the authors concluded that it had not been validated in France in the field of colorectal surgery.
10
The original POSSUM score and P-POSSUM were derived from a heterogeneous population of general surgical patients. Subgroup analysis of high-risk colorectal surgery patients found that the models under-predicted death in the emergency patients.
11
There was also a lack of calibration at the extremes of age in both emergency and elective work. This resulted in remodelling of the POSSUM score for colorectal surgery patients and led to the development of the Colorectal-POSSUM model (CR-POSSUM). The CR-POSSUM model was superior to the P-POSSUM model in predicting operative mortality in a study involving almost 7000 patients undergoing emergency and elective colorectal surgery in the UK.
12
External validation of CR-POSSUM was derived from three multicentre, UK studies involving a total of 16 006 patients: the original CR-POSSUM study population (
n
= 6883), the Association of Coloproctology of Great Britain and Ireland (ACPGBI) Colorectal Cancer (CRC) Database (
n
= 8077) and the ACPGBI Malignant Bowel Obstruction (MBO) Study (
n
= 1046). Of the different risk models that were tested, CR-POSSUM was superior in predicting postoperative death.
13
This conflicted to a certain extent with a later study in New Zealand involving 308 patients undergoing major colorectal surgery. In this cohort POSSUM, P-POSSUM and CR-POSSUM were all satisfactory predictive tools for postoperative mortality but the latter tended to be relatively less accurate. However, the authors noted that CR-POSSUM requires fewer individual patient parameters to be calculated and is therefore simpler to use.
14
A more recent systematic review pooled data from 18 studies to compare the accuracy of POSSUM, P-POSSUM and CR-POSSUM in predicting postoperative mortality for patients undergoing colorectal cancer surgery.
15
This study also reported greater predictive accuracy for the P-POSSUM model compared with CR-POSSUM.
CR-POSSUM is not an accurate predictor of disease-specific colorectal mortality. A study from the Netherlands on patients undergoing elective sigmoid resection for either carcinoma or diverticular disease demonstrated that CR-POSSUM over-predicted mortality in the patients with malignant disease and under-predicted it in patients with benign disease. However, in the whole group CR-POSSUM predicted postoperative mortality accurately.
16
A possible modification to the CR-POSSUM system was suggested following a single-centre UK study in 304 patients. CR-POSSUM proved to be a more accurate predictive model than POSSUM and P-POSSUM and, interestingly, logistic regression demonstrated a significant correlation between albumin and mortality. It may therefore be possible to improve the accuracy of CR-POSSUM further by modifying the equation to include serum albumin.
17
A UK study of 204 patients demonstrated that POSSUM did not accurately predict morbidity and mortality in patients undergoing oesophagectomy.
18
A dedicated oesophagogastric model (O-POSSUM) developed in a study population of 1042 patients was described in 2004. The O-POSSUM model used the following independent factors: age, physiological status, mode of surgery, type of surgery and histological stage. It provided a more accurate risk-adjusted prediction of death from oesophageal and gastric surgery for individual patients than P-POSSUM.
19
However, to date there have been several conflicting studies examining the predictive value and accuracy of O-POSSUM. A Dutch study of 663 patients undergoing potentially curative oesophagectomy in a tertiary referral centre (in-hospital mortality 3.6%) demonstrated that O-POSSUM over-predicted in-hospital mortality threefold and could not identify those patients with an increased risk of death.
20
This was supported in similar studies from both the UK and Hong Kong, which found that P-POSSUM provided the most accurate prediction of in-hospital mortality and O-POSSUM again over-predicted mortality in patients, particularly with low physiological scores and in older patients.
21
,
22
A recent systematic review comparing P-POSSUM with O-POSSUM, which included data from 10 studies, concluded that P-POSSUM was the most accurate predictor of postoperative mortality and O-POSSUM consistently overestimated postoperative mortality in gastro-oesophageal cancer patients.
23
In summary, the data reported in the original study to construct O-POSSUM have not been validated in other centres. It may be that individual units have to modify the O-POSSUM model to take account of local factors.
V-POSSUM was devised for use specifically in patients undergoing arterial surgery. One study examined the records of 1313 patients and added ‘extra items’ to the original POSSUM dataset, although this did not appear to significantly improve the accuracy of prediction.
24
The model has, however, been used in further studies and has been modified further to only take account of the physiology component of the score, with improved prediction accuracy (V-POSSUM physiology only). However, a current UK study involving almost 11 000 patients undergoing elective abdominal aortic aneurysm repair evaluated the accuracy of five risk prediction models, including V-POSSUM.
25
V-POSSUM performed poorly, with the Medicare and Vascular Governance North West (VGNW) models demonstrating the best discrimination, leaving the authors to conclude that V-POSSUM should not be used for risk prediction for these patients. Neither the V-POSSUM nor P-POSSUM models appear to be accurate in predicting mortality in the context of ruptured aortic aneurysms.
24
The finding that V-POSSUM may be of limited value in the context of emergency arterial surgery was confirmed by a larger and more recent evaluation of the appropriate POSSUM models in the context of ruptured abdominal aortic aneurysms (RAAAs).
26
When the P-POSSUM, RAAA-POSSUM, RAAA-POSSUM (physiology only), V-POSSUM and V-POSSUM (physiology only) models were all compared in 223 patients with RAAA (in-hospital mortality was 32.4%), all except V-POSSUM and P-POSSUM (physiology only) demonstrated no significant lack of fit.
As one may expect, the various POSSUM models are not accurate predictive tools in the context of elective carotid surgery. Both POSSUM and V-POSSUM over-predicted mortality in a large single-centre study (
n
= 499) of patients undergoing carotid endarterectomy.
27
This is not surprising given the nature of the surgery and the reduced surgical insult compared to body cavity procedures.
A recent evaluation of V-POSSUM in New Zealand has indicated that it is a useful tool not only in the assessment of outcome, but of longitudinal surgical performance in major vascular surgery. Major vascular procedures (
n
= 454) were prospectively scored for V-POSSUM over a 10-year period. There was a trend towards improved surgical performance over time, with a drop in the observed to predicted ratios of deaths. This novel role has not yet been tested in the other POSSUM models, but given these data there may be the potential to use them to evaluate surgical training and performance in other surgical subspecialities.
28
POSSUM is the most widely applied, validated, surgical risk scoring system currently used in the UK. The original POSSUM equation has been modified in an effort to increase its accuracy as a risk prediction tool for in-hospital surgical mortality. Of these, P-POSSUM is the most widely used and validated general modification. Speciality-specific POSSUM has also been developed for use in colorectal, oesophagogastric and vascular patients, with some variable improvement in risk stratification. The available data, however, suggest that the various POSSUM models have a tendency to overestimate mortality rates. Within these limitations, the POSSUM models provide a useful tool for risk assessment, audit, and comparing outcomes between different units and within the same unit over a period of time. However, due to the variables included in the calculation, POSSUM cannot be used in the preoperative setting to inform risk.