Medical versus surgical causes of death following colorectal resection: a Queensland Audit of Surgical Mortality (QASM) study

The causes of death following colorectal resection remain poorly explored. Few studies have addressed whether early post‐operative mortality is predominantly caused by a patient's medical co‐morbidities, or from factors pertaining to the presenting surgical disease process itself. This study analyses data from the Queensland audit of surgical mortality (QASM) to report the causes of in‐hospital death following colorectal resection, identifies whether these were due to either medical or surgical factors, and determines the patient characteristics associated with a medical cause of death.


Introduction
][4][5] While these risk factors are well defined, they do not explain the underlying causes of death.This is defined by the World Health Organisation (WHO) as 'the disease or injury that initiated the train of morbid events leading directly to death', 6 which can only be determined critically through an analysis of all the factors involved.As a result, few studies have investigated the underlying causes of death, and it remains uncertain whether these are predominantly related to a patient's medical co-morbidities, or from factors pertaining to the surgery or presenting surgical disease process itself.Knowledge of the patients at risk of medically related deaths is particularly important, as these may be more readily preventable through models of care that focus on the peri-operative optimisation of co-morbidities. 7,8he Queensland audit of surgical mortality (QASM) is an independent and peer-reviewed audit of post-operative mortality, in which all surgical centres in the state participate in. 9Following an in-patient post-operative death, operating surgeons are required to complete a 26-item Surgical Case Form (Appendix S1), which includes a written narrative of the patient's circumstances and course to death (item nine).The combination of qualitative information describing the causal pathway in item nine, supplemented by objective data about the patient's clinical state allows for reviewers to determine an accurate underlying cause of death.
The aims of this study were twofold.Firstly, we used QASM data to determine the underlying causes of death following colorectal resection and attributed these to a specific medical or surgical cause.Secondly, we compared the characteristics between patients who died due to medical causes and patients who died due to surgical causes, to determine the factors independently associated with a medical cause of death.

Inclusion criteria
We included all patients in Queensland who died in-hospital following colorectal resection over the 10 years between January 2010 and December 2020.Patients who died following colorectal procedures not involving resection were excluded.Patients undergoing appendicectomy were only included if they underwent surgery that involved additional colonic resection.

Determining the underlying cause of death
We observed the WHO definition of the 'underlying cause of death', and in each case determined this critically through a qualitative analysis of item nine of the Surgical Case Form.Each underlying cause of death was attributed to a specific surgical or medical aetiology.Surgical causes of death included those related to postoperative complications due to either an advanced surgical disease process noted on presentation, or a technical complication of surgery.Medical causes of death were defined as those occurring due to post-operative medical complications strictly in the absence of complicating surgical factors as described above.Both item nine and objective data about the patient in the form were critically evaluated by medically trained individuals DM and TC to classify the cause of death as medical or surgical.Disagreements were resolved by consensus or arbitration by a third reviewer.

Patient characteristics and co-morbidities
Patient characteristics documented included age, sex, hospital type and hospital geographical location as determined by the Rural, Remote and Metropolitan (RRMA) Classification. 10In addition to the American Society of Anaesthesiologists (ASA) score, several medical co-morbidities were recorded, including cardiovascular disease, respiratory disease, renal disease, hepatic disease, neurological disease, diabetes mellitus, advanced malignancy and advanced age.

Clinical features
Clinical features recorded included admission type, requirement for pre-operative inter-hospital transfer, requirement for critical care unit, colorectal pathology, colorectal resection type, and time from surgery to death.The surgeon's subjective determination on the overall risk of death before any surgery was also included.

Statistical analysis
Differences in patient and clinical characteristics between the two groups were analysed using t-tests and Chi-square tests as appropriate.Categorical variables were presented as frequencies with proportions, and continuous variables were presented as mean values with standard deviations.The level of statistical significance was defined at the level of P ≤ 0.05.The association between dichotomous variables and a medical cause of death was analysed using univariate regression to obtain odds ratios.To determine the factors independently associated with a medical cause of death, only significant co-morbidities as determined by the univariate analysis were assessed using a multivariate model.Data were analysed using SPSS ® version 28.0 for Windows (IBM Corporation, New York, USA).

Ethics approval
Ethics approval was not required as the QASM is a quality assurance activity under the governance of the Australian and

Deaths following colorectal resection: A QASM study
New Zealand Audits of Surgical Mortality (ANZASM).Consent from operating surgeons to participate was implied through voluntary return of the Surgical Case Form to the QASM.As previously described, data is de-identified.

Clinical features
A comparison of clinical features between groups is shown in Table 3.Both groups had a similar percentage of patients treated as an emergency (79.5% vs. 75.6%,P = 0.195), with an equal proportion in both groups transferred pre-operatively (20.9% vs. 21.4%,P = 0.870) or requiring critical care unit (CCU) admission (84.2% vs. 85.6%,P = 0.589).Patients dying from medical causes were more likely to receive their index procedure due to malignancy (54.0% vs. 28.6%,P < 0.001), and were less likely to undergo total or subtotal colorectal  resection (7.9% vs. 17.2%,P < 0.001).Surgeons similarly determined a large proportion of patients in both groups to have a 'considerable' or 'expected' risk of death (61.8% vs. 68.3%,P = 0.062), and there were no differences in the mean number of days between surgery and death (18.3 vs. 17.7,P = 0.645).

Multivariate analysis of factors associated with a medical cause of death
Multivariate logistic regression was undertaken to allow for adjustment of relevant factors.Only statistically significant factors determined in the univariate analysis were subsequently analysed with multivariate analysis.These results are presented in Table 4.

Discussion
Through critical analysis, this study examines the underlying causes of death following colorectal resection.We determined that medical and surgical causes of death both contributed approximately equally to the mortality burden, and that respiratory co-morbidity, neurological co-morbidity, advanced malignancy and advanced age were independent predictors for patients to experience a medicallyrelated death.
Ultimately, the two most common causes of post-operative death in our study were ischaemic colitis and bowel perforation, both surgical issues.This is in keeping with findings from larger studies.Mamidanna et al. 11 identified that in patients dying within 30 days of non-elective colorectal resection, 42.3% of deaths were caused by 'gastrointestinal events' such as perforation, obstruction, ischaemia or colitis.Despite including over 170 000 patients, the administrative nature of this data means that the prevalence of specific gastrointestinal events remains unknown.In contrast, our study has quantified the proportion of deaths from specific gastrointestinal pathologies, and highlighted whether these occurred pre-operatively or manifested post-operatively as a surgical complication.
3][14] It should be noted however, that none of these studies have explicitly stated whether the medical complications observed were the true underlying causes of death, or merely the mechanisms that led to the death.The latter would not have otherwise occurred without a precipitating primary insult and is therefore not aetiologically specific.Patients who die following colorectal resection often follow a complex clinical course, where superimposing acute surgical illness may not only exacerbate preexisting co-morbidity, but may also ultimately manifest as medical events such as multiple organ failure. 11,15Therefore, the significance of medical causes of death previously described may be overstated.By excluding cases with significant confounding surgical issues, we have allowed for a highly accurate report on postoperative deaths due to underlying medical causes.
Patients in the QASM database were predominantly of older age and died following emergency procedures.The older surgical Deaths following colorectal resection: A QASM study patient is well known to be at an increased risk of post-operative mortality due to medical factors such as multi-morbidity, frailty and reduced physiologic reserve. 16These patients should be offered peri-operative models of care that focus on both identifying and optimizing their underlying co-morbidities.Comprehensive Geriatric Assessment is an evolving collaborative model of care between medical and surgical teams in the peri-operative setting.It encompasses not only the evaluation of co-morbidities, but also other domains such as nutrition, cognition and functional status to guide patient-centred and multi-disciplinary targeted interventions in the post-operative period. 17Although often described in the elective setting, CGA has been validated to improve outcomes in the emergency setting 18,19 , with the National Emergency Laparotomy Audit (NELA) demonstrating that geriatrician intervention was associated with a significant reduction in post-operative mortality (OR 0.38, 95% CI: 0.35-0.42). 18urthermore, it has been found that the driving factor determining the relationship between high-volume centres and improved patient outcomes may not be their ability to provide procedural interventions to treat surgical complications, but rather their capacity to offer a full spectrum of clinical services from cardiac, critical care and medical sub-specialty services. 8,20This further strengthens the argument that patients with significant comorbidities should be managed in collaborative multi-disciplinary units to facilitate the prompt recognition and treatment of medical complications if they occur.With fluid therapy a cornerstone of peri-operative medicine, strict attention should be drawn to this area.This is particularly pertinent in emergency contexts, where patients are often elderly with impaired organ function, yet frequently have intercurrent pathology associated with large intravascular losses.In these high-risk situations, a judicious strategy maintaining fluid balance between 0 and 2 L has been found to reduce the rate of cardio-respiratory complications. 21ur finding of respiratory and neurologic co-morbidity being independent risk factors for a medical cause of death is evidenced by the significant number of our patients who died due to aspiration pneumonia.Although this may occur in the context of surgical issues such as post-operative ileus, it should be recognized that aspiration does not usually result in mortality especially in the absence of complicating surgical factors.Failure to recover after medical treatment for aspiration pneumonia implies that there are underlying respiratory, neurologic and cognitive co-morbidities impeding recovery. 22t remains controversial whether enhanced recovery protocols such as immediate nasogastric tube (NGT) removal and early feeding can be applied in the emergency setting.Venara et al. 23 found that immediate NGT removal following emergency colectomy was not associated with the need for NGT re-insertion, however only eight out of 79 patients were actually NGT free post-operatively.Existing studies on ERAS protocols in emergency colectomy remain similarly low powered.If undertaken, we advocate that any early or subtle signs of ileus should act as a warning to re-insert a NGT and slow feeding.
We have also determined that advanced malignancy and advanced age were strong independent predictors for medical causes of death following colorectal resection.These patients underwent non-emergent operations for colorectal cancer.Even in the absence of post-operative complications in these patients, the physiologic insult of surgery initiated a chain of events resulting in medical decompensation, failure to thrive and ultimately death.
Although we only analysed inpatient deaths, we also recognize that the true impact of post-operative mortality in cancer patients may extend far beyond the in-hospital period.Dekker et al. 24 found that in elderly patients with colorectal cancer, excess deaths occurring up to 1-year post-operative was the greatest contributor towards age-related differences in cancer survival.While there is consensus that death within this extended period is commonly due to the sequalae of underlying malignancy rather than surgical complications, [25][26][27] our study is one of few to determine that this is also the case in the immediate post-operative period.Given that the predominant causes of death in the early and late post-operative period similarly reflect the gravity of advanced oncologic illness, surgeons should consider rationalizing the selection of such patients for surgery.In this context, the use of risk scoring systems may be of value.The revised Association of Coloproctology of Great Britain and Ireland (ACPGBI) model is specific to oncologic resection and has been validated to closely resemble the true mortality rate in both elective and emergency settings, with an observed to expected (O:E) ratio of 1.06 and 0.91, respectively. 28When prognosis appears poor, transition to a formal palliative care pathway may be a better alternative than insisting on surgery, where the latter may be futile and only serve to prolong suffering without medical benefit. 29Furthermore, there should be greater consideration in the timing at which palliative care input is sought for surgical patients.Rather than post-operative referral following a complicated course, surgeons should recognize that pre-operative palliative care referral can reduce operative-related mortality. 30This demonstrates that palliative care referral does not imply cessation of acute operative management, but though goaldirected treatment of anticipated surgical complications can actually improve post-operative outcomes.
Although analysing both quantitative and qualitative mortality information directly from the primary source allowed for a highly specific underlying cause of death to be determined, we recognize that information from these retrospectively written reports may be influenced by confirmation bias.It should be noted however, that QASM data has been shown to be credible and high quality, with an overall concordance of 98.2% and median concordance of 100% compared to medical records. 31

Conclusion
In conclusion, in the absence of complicating surgical factors, a significant number of patients die in hospital following colorectal resection as a result of their underlying co-morbidities.Respiratory co-morbidity, neurologic co-morbidity, advanced malignancy and advanced age are independent predictors associated with a medical cause of death.Multi-disciplinary models of care which allow for the early recognition and treatment of medical complications may reduce post-operative mortality in these patients.

Fig. 1 .
Fig. 1.Comparison in the distribution of the number of co-morbidities between patients dying from medical causes versus surgical causes.

Table 2
patient demographics and co-morbidities Abbreviation: ASA, American Society of Anaesthesiologists.

Table 4
Factors associated with a medical cause of death following colorectal resection with multi-regression analysis ANZ Journal of Surgery published by John Wiley & Sons Australia, Ltd on behalf of Royal Australasian College of Surgeons.