Predicting postoperative morbidity in adult elective surgical patients using the Surgical Outcome Risk Tool (SORT). Wong DJN, Oliver CM, Moonesinghe SR.

British Journal of Anaesthesia 2017;119(1):95-105 doi: 10.1093/bja/aex117

Presented by: Dr Alex Cormack

Background

  • Perioperative risk assessment is a key part of the consent process
  • Risk stratification tools also allow comparison between outcomes of different institutions
  • Morbidity following surgery can have a significant impact on quality of life and needs to be a consideration when considering surgical options
  • Morbidity is more common than mortality following surgery and potentially provides a more sensitive measure of comparison between different healthcare providers
  • P-POSSUM and POSSUM are currently the most frequently used tools for perioperative risk prediction

POSSUM:

  • Physiological and Operative Severity Score for the enumeration of Mortality and morbidity
  • Developed in the 1990s
  • For use in elective and emergency general surgical procedures
  • Does not apply to trauma patients
  • Calculated at the time the decision to operate is made
  • Variants include CR-POSSUM, Vascular-POSSUM and O-POSSUM
  • Requires 12 physiological and 6 operative parameters to calculate

P-POSSUM:

  • Portsmouth modification of the Physiological and Operative Severity Score for the enumeration of Mortality and morbidity.
  • A variation of the POSSUM tool

SORT:

  • Surgical Outcome Risk Tool
  • Developed after the 2011 NECEPOD report
  • Uses six parameters collected preoperatively
  • Designed to predict probability of 30 day mortality following surgery
  • The authors state that it ‘compared favourably with other previously validated risk stratification tools’ and ‘has been externally validated recently in a cohort of patients undergoing hip fracture surgery’.
  • Predictor variables: ASA grade (III, IV or V), surgical urgency (expedited, urgent or immediate), high risk specialities (GI, thoracic or vascular surgery), surgical severity (major or complex), malignancy, age (65-79 or >80)

 Design and Setting

  • Single centre prospective study at University College London Hospital to:

“develop and validate a new model to predict the likelihood of postoperative                       morbidity using predictor variables found in SORT, and then compare its      performance against POSSUM.”

  • 3 year period (June 2009 – May 2012)
  • Data collection carried out by trained research staff independent of the clinical teams responsible for the patient

Subjects

Inclusion criteria:

  • Patients undergoing elective major inpatient operations
  • 1934 patients included

Exclusion criteria:

  • Patients with duplicated or missing data
  • Patients who did not have POMS (Post Operative Morbidity Survey) scores recorded on Day 7

Intervention

  • Data collected:
    • 1934 patients identified, 1583 patients included
    • 58% female
    • 45% orthopaedic and 39% abdominal procedures
    • 6 deaths within 30 days of surgery
  • Data excluded:
    • 351 patients excluded
    • Missing predictor variables: DOB, ASA status, surgical speciality, malignancy status
    • Missing POMS outcomes: duplicated or missing entries
    • Clear summary of reasons for exclusion, no patients unaccounted for
  • POMS administered prospectively to patients at several time points postoperatively (day 3,5,7 or 8, 14 or 15 and 21)
  • Morbidity outcome measure selected was POMS-defined morbidity recorded after postoperative day 7 or 8
  • Data randomly split into two groups
    • 1/3 validation group (n=527)
    • 2/3 derivation group to define new model (n=1056)

Outcomes

  • Predictor variables from the original SORT variables were adjusted to generate SORT-morbidity models
  • Outcome variable was set as the presence of POMS defined morbidity on postoperative day 7 or 8
  • SORT-morbidity models then tested in the validation group using statistical analysis
  • Final model then tested against POSSUM

Results

  • No statistically significant difference between new SORT-morbidity model and POSSUM at discrimination of morbidity at 7 days post surgery
  • Linear shrinkage factors estimated to improve prediction of morbidity at later time points

Conclusions

  • New SORT-morbidity model is comparable to POSSUM at prediction of morbidity 1 week post operatively
  • Linear shrinkage factors can be applied to improve morbidity prediction further in the postoperative course

Strengths

  • Morbidity is an important consideration within the surgical consent process
  • Data collection carried out by research staff independent of clinical teams
  • Clear documentation of reasons for exclusion
  • Good number of data sets
  • New SORT-morbidity model found to compare favourably to POSSUM
  • POMS is a validated measure of morbidity to use in data collection

Weaknesses

  • Morbidity defined as POMS defined morbidity after day 7
    • Patients discharged prior to this time period excluded despite possible morbidity
  • Only looked at elective patients
  • Single centre study
  • Unequal representation of surgical specialities
  • Comparatively low mortality rate (0.31%) documented – the authors comment that rates of 0.37-0.67% have been documented elsewhere in the literature
  • POMS domains include some relatively minor measures of morbidity that may influence the results (for example urinary catheter following elective urology cases)
  • Required ‘linear shrinkage factors’ to enable morbidity to be predicted later than 7 days postoperatively

Implications and Potential for Impact

  • Possible development of a new tool to use alongside existing risk assessment tools
  • SORT-morbidity was only used in elective cases and therefore could not be used in a CEPOD setting without further studies
  • Potential for further studies to develop the SORT-morbidity tool for more widespread use
  • P-POSSUM currently universally understood amongst the theatre MDT whereas SORT is less widely understood
  • SORT and SORT-morbidity require fewer variables to calculate, however this is less relevant as SORT-morbidity has only been developed using elective cases

The use of SORT-morbidity as an alternative to P-POSSUM does not yet seem a realistic prospect. P-POSSUM is understood amongst surgical and anaesthetic professionals and allows management decisions to be made appropriately. It is used for both emergency and elective patients, and arguably its most important use is in planning the management of emergency patients. This is an area that the SORT-morbidity tool has not been developed for. Further studies and multi-centre validation would be required before it could reliably be used in clinical practice.