ASA classification and surgical severity grading used to identify a high‐risk population, a multicenter prospective cohort study in Swedish tertiary hospitals

Identification of surgical populations at high risk for negative outcomes is needed for clinical and research purposes. We hypothesized that combining two classification systems, ASA (American Society of Anesthesiology physical status) and surgical severity, we could identify a high‐risk population before surgery. We aimed to describe postoperative outcomes in a population selected by these two classifications system.


| INTRODUC TI ON
As surgical techniques and perioperative medicine have developed, more elderly patients with significant co-morbidities are offered curative and complex surgical procedures in Sweden. Postoperative morbidity, that is, adverse events, is relevant from a patient perspective as it reduces the health-related quality of life up to 1 year 1 and increases mortality during the first three years after surgery. 2 During the last decades, complex surgical procedures have been centralized into specialized hospitals which has led to a concentration of patients classified as ASA 3 (American Society of Anesthesiology physical status classification system) or more undergoing major to complex major surgery on tertiary hospitals. After this centralization process, it is reasonable to generate data on postoperative morbidity and mortality for clinical planning of resources, audit of performance and for research purposes. 3 We hypothesize that a population classified as ASA 3 or more undergoing major to complex/major (AXA PPP/Specialist Procedure Codes; AXA-insurance company with Public-Private Partnership) gastrointestinal, orthopedic, or urogenital surgery in Swedish academic hospitals are at high risk with a mortality exceeding 5 per cent. 3 The primary aim of this study is to describe the frequency and types of postoperative morbidities and mortality in a study cohort identified by surgical severity (AXA PPP) and ASA.

| MATERIAL S AND ME THODS
The method section is written in accordance to the STROBE statement for transparent reporting of observational studies. 4 The cohort study was conducted per Handbook for Good Clinical Research Practice (GCRP/WHO). Compliance with GCRP provides public assurance that the rights and safety of study subjects are protected, consistent with the principles of the Declaration of Helsinki, the Swedish Personal Data Act and the Personal Data Ordinance.
Processing of the personal data was authorized by the Swedish Data Protection Agency. The study was approved by the Regional Ethical Committee, Stockholm (ID: 2015/1128-31/4). The study was classified as a clinical follow-up study and written consent was waived by the Ethical committee. This was a multicentre observational closed cohort consecutive study (Predictors, Risk factors, and Outcome following major surgery, PROFS NCT02626546). Exclusion criteria were the following surgical procedure groups as they were considered to already being reasonably well described (immediate trauma, neurosurgery, cardiac, vascular, pulmonary, transplantation), or not providing a major part of the high-risk population at tertiary hospitals (breast, eye, ear, nose and throat, endocrine, face, mouth, salivary, skin, subcutaneous tissue, endoscopic gastrointestinal).
As uniform classification of surgical severity is not used in Sweden, inclusion to the prospective study cohort needed to be done manually by the research team. The research team at each study site was instructed how to use the surgical procedure coding tool. The assessment of eligibility criteria of major to major/complex procedure was based on the AXA/PPP classification of severity. In practice, we used the procedure list of Surgical outcome risk tool calculator. The background population, from which the study cohort was selected, consisted of patients scheduled for any type of surgical procedure except for day-case or thoracic surgery and classified as ASA class 3-4. The study participants were identified by daily screening of the scheduled surgical procedures (elective and emergency) coded as ASA ≥3 generated by the surgical schedulingsoftware used at the respective study site. The length of follow up was 1 year.

| Baseline characteristics in prospective study cohort
Individuals were characterised by age, co-morbidities, types and urgency of surgery. All characteristics were accessed through electronic health records and operation planning system. The ASA classification 5 as documented by attending anaesthesiologist during the pre-anaesthesia visit was used.

Editorial Comment
At hospitals more elderly patients undergo surgeries of increased complexity. Identification of patients with increased risk of postoperative complications is needed.
This study finds that preoperative severity grading using a combination of ASA scoring and a surgical procedure coding identified patients at increased risk of postoperative morbidity and mortality following major gastrointestinal and orthopaedic surgeries at tertiary hospitals in Sweden.

| Mortality in prospective study cohort
The postoperative mortality was assessed at 30, 180, and 360 days following surgery.

| Data sources in prospective study cohort
Patient characteristics and postoperative morbidity were extracted manually from the medical notes by the research team. For mortality, the study cohort was matched with Swedish Personal Address Registry (SPAR).

| Background and eligible population
The cohort where selected manually, consequently there was a risk that not all eligible patients were included. To test whether the cohort was representative we derived the eligible population after the closure of the trial. This was done as follows; from the operation databases at each study site a background population (all patients undergoing any surgery but day-care and thoracic during the study period) and a subpopulation with ASA ≥3 was extracted. In the latter subset, the performed surgery was classified by the UK severity grading system (AXA PPP/Specialist Procedure Codes, United Kingdom) 6 by one member of the research team. At last codes were constructed for the inclusion criteria (surgical subspecialties and severity) and a dataset of eligible patients was derived.

| Statistical methods
The number of eligible patients during the study period determined the sample size.
Descriptive patient data are presented by number and percent.

| Prospective study cohort characteristics
The study cohort consisted of 1089 participants. Thirteen were excluded due to violation of inclusion criteria (misclassification of surgical severity), three were lost to follow-up at 30 days and ten were lost at one year. Consequently, 1063 patients entered the final analysis (Appendix Box 1, Consort diagram).
Demographic characteristics, co-morbidities and types of surgical procedures are displayed in Table 1. The number of females/ males was about equal. None of the patients was classified as ASA 5.
The most common co-morbidity was renal failure (mild to end stage), followed by ischemic heart disease and chronic obstructive pulmonary disease. The most common surgical procedure was orthopedic surgery, followed by gastrointestinal and urogenital surgery.

| Postoperative morbidity and mortality
The percentage of patients with one or more postoperative morbidities defined by POMS was 78/48/47 per cent on day 3/7/10 ( Table 2). The point prevalence of postoperative morbidity given per POMS domain on day 3, 7 and 10 in elective/emergency surgery as well as categorized according to surgery; gastrointestinal, orthopaedic or urogenital surgery respectively, are displayed in Figure 1 and Appendix Tables A1-A4. Considering the whole cohort, the main source of morbidity on day 3 was mobilisation, renal and or pain ( Table 2). This pattern changed over time, as the incidence  Table A1). In orthopaedic patients undergoing emergency surgery, morbidities were 85 per cent on day 3 compared to 72 per cent after elective procedures (Appendix Table A2). On day 3 the most common sources of morbidity for both emergency and elective orthopaedic surgery were mobilisation, renal, followed by infectious and pain (Appendix Table A2).
In gastrointestinal emergency/elective procedures the postoperative morbidities were almost similar 87/86 per cent on day 3 (Appendix Table A3). On day 3, the most common source was pain, renal and mobilisation in elective group with the addition of infectious morbidity in the emergency group. This pattern changed over time and gastrointestinal and infectious morbidities increased by day 7 and 10 in elective group and decreased in emergency group (Appendix Table A3).
In urogenital surgical procedures, the pattern of morbidity was different. Emergency procedures had lower risk for morbidity compared to elective procedures on day 3 (42 vs. 62 per cent, Appendix Table A4).
The severity of postoperative morbidity was assessed by the Clavien-Dindo grading system. 8 There was a deviation from normal course in 50 per cent of patients, 47 per cent within the ASA III group and 72 per cent within ASA IV group ( Table 3). Majority of morbidities (67.5 per cent) were graded >1 by the Clavien-Dindo system (Table 3). Postoperative morbidity, classified as Clavien Dindo >1, and emergency surgery was associated with higher risk of death ( Figures 2 and 3 Table A5).

| Background and eligible population
The background population, that is, those who underwent any type of surgery (with exception of thoracic and outpatient procedures) during the study period at the five study sites, consisted of 18 382 individuals (ASA 1-6). The mortality was 1.7/4.9/7.4 per cent at 30/180/360 days (Appendix Table A6).  Table A6).   which is a clinical routine for both elective and emergency major     The prospective study cohort have several limitations. One is that not all eligible patients entered the cohort. The difference between the manually obtained prospective Study cohort and the electronically obtained eligible population, was approximately 30%

| D ISCUSS I ON
(1089 included vs 1595 eligible). The case mix and postoperative mortality of the eligible population was comparable with the prospective Study cohort which suggest that the cohort could be described as representative.
Another possible limitation of the study could be any misclassification in either the ASA categories and/or in the UK surgical severity coding system. The ASA classification is subjective but has been found to be robust in large populations. 13 To minimize researcher induced selection bias the ASA classification was done by attending clinicians and was not revised by the research team. In Sweden there is no uniform grading system of size and severity of surgery. As a uniform grading is required if fair comparisons are aimed for, we applied UK surgical severity grading for the selection of the population.
The Swedish and UK surgical procedure codes, ie the name of the surgical procedures, are not identical. So, major surgery could erroneously have been classified as major complex or vice versa as in any categorization/translation. A possible flaw in using the UK surgical severity grading system, in this study not attended to, is that the UK coding system of surgery was developed to establish reimbursement in independent UK hospitals. Besides the magnitude and severity of surgery, reimbursement also must consider technical requirements for the surgery. This is illustrated by the low morbidity and mortality in the emergency urologic group that in the system is classified having a high severity.
We suggest that the results are generalizable to other academic hospitals in Sweden, but also to countries with similar health care system as the Scandinavian countries as this is a multicentre study including 4 tertiary hospital on 5 sites. It would be highly interesting to perform a new study in a multinational context. Such an initiative F I G U R E 3 The severity of any recorded morbidity and the cumulated risk of death in the prospective Study cohort described by a Kaplan-Meier plot. The severity of morbidity was assessed by the Clavien Dindo  This study adds basic information about the morbidity and mortality in patients who are offered major to complex major surgery and presenting with an ASA classification of 3 or more in the context of academic hospitals in Sweden.

| CON CLUS IONS
Preoperative severity classification as major to complex major according to AXA PPP/Specialist Procedure Codes and ASA≥3 could be used to identify a high-risk surgical population concerning postoperative morbidity and mortality before gastrointestinal and orthopaedic but not urological surgery. Combing the two-classification systems provides a future possibility for electronic data extraction of a high-risk population in tertiary hospitals.