Achieving equity: patient demographics and outcomes after surgical and non‐surgical procedures in South Australia, 2022

Although modern Australian healthcare systems provide patient‐centred care, the ability to predict and prevent suboptimal post‐procedural outcomes based on patient demographics at admission may improve health equity. This study aimed to identify patient demographic characteristics that might predict disparities in mortality, readmission, and discharge outcomes after either an operative or non‐operative procedural hospital admission.


Introduction
Health equity is a state and ethical human rights principle which describes the absence of systematic disparities to promote equal health outcomes between groups with different levels of underlying social advantage and disadvantage. 1Within this principle, the patient's social determinants of health, psychosocial environment and support structures play important roles. 2,3In striving for health equity, demographic characteristics that predict disparities in healthcare outcomes should be identified 4 so that they can be integrated within pathways of, and approaches to, clinical care. 5Within Australia, one of the most prominent examples of health inequity is the disparity in healthcare outcomes between Indigenous Aboriginal and Torres Strait Islanders (ATSI) and the general population. 6or all patient populations, but particularly those that are disadvantaged at baseline, undergoing a formal interventional procedure, be it surgical or non-surgical, can present a traumatic biopsychosocial disruption to the patient's life especially if the post-procedure outcome is poor. 7When evaluating surgical recovery, patient, surgical, system, and societal factors must be considered. 8Although modern procedural healthcare systems provide individualized, patientcentred care, the ability to predict and prevent suboptimal postprocedural outcomes based on patient demographics at admission may potentially improve health equity.Accordingly, to inform systemic modifications to clinical pathways, this study aimed to identify patient demographic characteristics that might predict disparities in mortality, readmission and discharge outcomes after either an operative or non-operative procedural hospital admission.

Study design, setting and population
This study followed Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guidelines. 9A retrospective cohort study was conducted including consecutive elective and emergency surgical and non-surgical procedural admissions across all services at the Queen Elizabeth Hospital, Royal Adelaide Hospital, and Flinders Medical Centre in South Australia between 1 January and 31 December 2022.This timeframe was selected so that the resultant study findings would represent a recent cohort for increased relevancy to modern-day practice.Within the study cohort, a range of protocols for clinical care were employed given that all healthcare services were included and the participating centres spanned two metropolitan healthcare networks.Confirmation of the operative or non-operative procedure was from the 'operation note' within the electronic medical record.'Operation notes' are used to document surgical and non-surgical procedures (e.g., interventional, diagnostic, catheter-based, endoscopic, interventional radiology, biopsies).Within the participating centres, operation notes in the electronic medical record are entered for any surgical or non-surgical procedure that is conducted in an operating room setting.Patients who did not have an operation note entered in the electronic medical record were excluded from the present study.Institutional ethical approval for this study was obtained from the Central Adelaide Local Health Network Human Research Ethics Committee (reference number: 16860), with a waiver of individual consent.

Data collection
Data were retrospectively collected from institutional electronic medical record by clinical staff.These data were originally recorded throughout each patient admission, by healthcare and administrative staff.The electronic medical record from which data were extracted from is standardized, and was used in the same fashion across all participating centres during the study period.Demographic data were collected from the Patient Administrative Software database.

Statistical analysis
Descriptive statistics were used to describe the cohort characteristics.Length of hospital stay was calculated in days as the difference between the date of hospital admission and the date of hospital discharge.For patients who died during the admission, time from the procedure to death was calculated as the difference in days between the date of the procedure (according to time of operation note entry in the electronic medical record) and the date of death listed in the electronic medical record.Within the electronic medical record, mortality data are regularly collected, matched, and cross-checked with the South Australian Births, Deaths and Marriages registry at monthly intervals. 10Accordingly, deaths are captured within the electronic medical record, regardless of whether they occur in a hospital or in the community within South Australia.Multivariable logistic regression analyses were separately conducted to investigate patient demographic factors that were independently associated with the primary outcome of 30-day post-procedural all-cause mortality rate, and secondary outcomes of in-hospital all-cause mortality rate, 90-day post-procedural all-cause mortality rate, 30-day post-discharge all-cause readmission rate, and hospital discharge directly to home (as compared with another facility or location post-hospital discharge), respectively.Predictive or explanatory variables for the multivariable logistic regression models were: patient sex (male vs. female), patient age, comorbidity burden (Charlson comorbidity index 11 score), admission urgency (emergency admission versus non-emergency admission), identification as indigenous Aboriginal or Torres Strait Islander (ATSI) ethnicity (ATSI vs. not ATSI), English as a primary language (English listed as a primary language vs. language other than English listed as a primary language), the presence of religion (no religion stated vs. religion stated), birthplace in Australia (birthplace listed as Australia vs. birthplace listed as a country other than Australia), and marital status (married vs. not married).Only binary outcomes were used, and all predictor variables were either binary or continuous.The dispersion parameter for binomial family was taken to be 1.Socioeconomic status, although known to impact health outcomes and particularly mortality, 12 was not included as a variable within the present study given it is not readily measurable by hospital staff upon admission.Furthermore, this avoided concerns regarding the translatability of subsequent findings outside of South Australia and outside of Australia.Similarly, residential address was not evaluated as each healthcare network in South Australia is organized by geographical proximity, and accordingly would likely creating bias as analyses were conducted at the system (and not individual hospital) level.Median imputation was used to replace missing Charlson comorbidity index data (21 292 of 40 882 admissions) only.For P-values generated within the analysis, the threshold for statistical significance was set at P = 0.05.After the multivariable logistic regression models were generated, backwards selection was performed by iteratively removing predictor variables from the models based on their P-values until no more variables could be removed without significantly reducing the model fit.Statistical analysis was conducted using Python and R (version 1.1.456).

Patient characteristics
This study included 40 882 individual patient admissions with either operative or nonoperative procedures conducted, for 36 719 unique patients.Of the admissions, 19 960 (48.8%) of the patients were of female sex.Median age was 59.8 years (IQR: 39.6-73.2).Regarding admission urgency, 13 194 (32.3%) were emergency admissions.The most common admission diagnoses are outlined in coded format, according to the 10th Revision of the International Classification of Diseases, 13 within Data S1.The median length of hospital stay was 1 day (IQR: 0-4).For the 205 (0.5%) patients who had a readmission after hospital discharge, the median number of days after discharge for readmission was 12 (IQR: 5-19).A total of 203 (0.5%) patients had a readmission within 30 days after hospital discharge.A total of 301 (0.7%) patients experienced inhospital mortality.A total of 537 (1.3%) patients experienced 30-day post-procedural all-cause mortality.A total of 1023 (2.5%) patients experienced 90-day post-procedural all-cause mortality.For patients who experienced mortality following their procedure, the median time to death from the procedure was 69 days (IQR: 23-153).Characteristics of the study cohort are presented in Table 1.
Results and details of all multivariable logistic regression models prior to backwards selection, and details of the backwards selection conducted, can be found in Data S1.

30-day post-procedural all-cause mortality
After backwards selection, the variables of patient sex, patient age, Charlson comorbidity index, admission urgency and the presence of religion remained in the logistic regression model for 30-day post-procedural all-cause mortality (Table 2).Of these, older age (P < 0.001), high Charlson comorbidity index score (P < 0.001), having an emergency admission (P < 0.001), and being of male sex (P = 0.03) were significantly associated with increased likelihood of 30-day post-procedural all-cause mortality.The presence of religion was not significantly associated with 30-day post-procedural all-cause mortality (P = 0.1062).

In-hospital all-cause mortality
After backwards selection, the variables of patient sex, patient age, Charlson comorbidity index and admission urgency remained in the logistic regression model for in-hospital all-cause mortality and all had statistically significant association with the outcome (Table 3).Specifically, older age (P < 0.001), higher Charlson comorbidity index score (P < 0.001), having an emergency admission (P < 0.001), and being of male sex (P = 0.046) were significantly associated with increased likelihood of in-hospital all-cause mortality.

90-day post-procedural all-cause mortality
After backwards selection, the variables of patient sex, patient age, Charlson comorbidity index, admission urgency, and Indigenous ethnicity status remained in the logistic regression model for 90-day post-procedural all-cause mortality (Table 4).Of these, being of male sex (P < 0.001), older age (P < 0.001), higher Charlson comorbidity index score (P < 0.001), and having an emergency admission (P < 0.001) were significantly associated with increased likelihood of 90-day post-procedural all-cause mortality.Indigenous ethnicity status was not significantly associated with 90-day post-procedural all-cause mortality (P = 0.09).

30-day post-discharge all-cause readmission
Backwards selection was not possible on the multivariable logistic regression model for 30-day post-discharge all-cause readmission due to missing data (Data S1).Accordingly, results of the model prior to backwards selection are presented in Table 5.Out of the variables, higher Charlson comorbidity index score (P < 0.001), having an emergency admission (P < 0.001), identifying as Indigenous ethnicity (P < 0.001), and being born in Australia (P = 0.03) were significantly associated with increased likelihood of 30-day post-discharge all-cause readmission.Patient sex (P = 0.12), patient age (P = 0.68), primary language (P = 0.06), presence of religion (P = 0.77), and marital status (P = 0.54) were not significantly associated with 30-day post-discharge all-cause readmission.

Hospital discharge directly to home
After backwards selection, the variables of patient sex, patient age, Charlson comorbidity index score, admission urgency, Indigenous ethnicity status, presence of religion, birthplace, and marital status remained in the logistic regression model for hospital discharge to home (Table 6).Of these, being married (P < 0.001) and of male sex (P = 0.003) were significantly associated with increased likelihood of hospital discharge directly to home as opposed to requiring discharge to another facility or location.Of the variables, older age (P < 0.001), higher Charlson comorbidity index score (P < 0.001), having an emergency admission (P < 0.001), identifying as Indigenous ethnicity (P < 0.001), and being born in Australia (P = 0.001) were significantly associated with a decreased likelihood of hospital  discharge directly to home, indicating that discharge to another facility, or to home with formal supports, was required.The presence of religion was not significantly associated with a likelihood of hospital discharge directly to home (P = 0.09).

Discussion
This study characterized several associations between patient demographic factors present on admission and outcomes after surgical and non-surgical procedures, that can be integrated within patient flow pathways through the Australian healthcare system to improve healthcare equity.An increased likelihood of all-cause, postprocedure mortality in-hospital, at 30 days, and at 90 days, were all significantly associated with increased age, increased comorbidity burden, having an emergency admission, and being of male sex.Ethnicity characteristics showed more association with measures relating to safe hospital discharge, with identification as ATSI and being born in Australia being associated with an increased likelihood of hospital readmission within 30 days after discharge and decreased likelihood of being safe to discharge directly to home, as was increased comorbidity burden and having an emergency admission.Being married and of male sex were predictive of an increased likelihood of being able to discharge safely and directly to home.This is in contrast to increased age which was predictive of a decreased likelihood of this occurring.Given all the patient demographic characteristics that were predictive of post-procedural outcomes are readily identifiable upon admission, hospital staff involved in the admission process should seek to integrate consideration of these factors within the patient's clinical pathway.Increased age, increased comorbidity burden and being an emergency admission were all predictive of postprocedural mortality from in-hospital up to 90 days, and already confer peri-procedural clinical modifications in the form of expedited intervention and greater clinical supports within the current healthcare system.Identification of these factors early during an admission may facilitate more appropriate discharge, or increase focus on that patient to receive more appropriate resources.Interestingly, being of male sex was also predictive of increased likelihood of being able to safely discharge from hospital back to home, suggesting that the association with mortality may be reflective of known differences in this outcome between the sexes 14 rather than underlying inequity in the social determinants of health. 15This has been shown previously where being of female sex was predictive of a hampered recovery experience in other measures. 16he increased likelihood of married patients being able to discharge directly to home could be representative of increased stability in community and social supports. 17It is however important to note that the administrative data used do not distinguish between transgender and non-transgender males and females, where sex and gender identity may be incongruent.Patient identification as indigenous ATSI being predictive of increased likelihood 30-day post-discharge hospital readmission and decreased likelihood of being able to discharge directly to home may be indicative of underlying lack of social supports and longstanding inequity in the social determinants of health. 6However, it may also be reflective of appropriate cultural safety, with these patients being discharged prior to being medically suitable for discharge in order to permit culturally-appropriate community reinforcement and preserve their cultural and psychological safety, with the appropriate safety-netting measures being taken to ensure re-presentation and readmission if required.For these patients, involvement of additional services, such as Aboriginal Liaison Officers early on during procedural admissions may be helpful to ensure swift post-procedural recovery and safe hospital discharge.However, this was outside the scope of the present study.
This study has limitations.The timeframe was purposely limited to the year of 2022 to ensure temporal relevance of resultant findings, however, a longer study period or comparison to a pre-COVID-19 historical cohort 18 may have provided more historically representative results.This study analysed the patient population based on admissions, rather than unique individual patients, and while patients with multiple admissions may have been included multiple times within analyses, this approach was taken to be more representative of how healthcare services function.Similarly, the analysis of major and minor procedures of both a surgical and nonsurgical (e.g., interventional, diagnostic, catheter-based, endoscopic) nature all together, while being more representative of system-level associations, may have missed differences between these cohorts.Within the dataset, a significant proportion of Charlson comorbidity index values were missing, however, given the relatively narrow interquartile range of the values, the replacement of these missing data via median imputation was likely to provide a reasonable representation of the real-life patient cohort.Further, this study was conducted in three English-speaking metropolitan centres, and given the importance of patient demographic characteristics that include language, further research in other hospitals, particularly those in rural settings, is required.This study characterized several associations between patient demographic characteristics present on admission, and outcomes after surgical and non-surgical procedures in South Australian patients in 2022.It is important that Australian healthcare systems seek to consider these characteristics and associations within peri-procedural clinical patient flow pathways, so as to improve healthcare equity.Of particular note, additional support must be given for ATSI patients early on during admission, and also during the process of determination of fitness for safe hospital discharge and male sex and their potential for re-admission.In addition to evaluating healthcare systems that have integrated these findings, future studies should also look to investigate predictive technologies that can utilize these data for improved healthcare equity, such as artificial intelligence.

Dedication
This study is dedicated to the memory of Dr. Esther Hui-Lin Chong for her dedicated service and faithful commitment to support different communities for better health care, which was a true expression of her Christian faith.

Author contributions
Joshua G. Kovoor: Conception and design; acquisition of data; analysis and interpretation of data; drafting the article; revising the article critically for important intellectual content, and final approval of the version to be published.Aashray K. Gupta: Conception and design; acquisition of data; analysis and interpretation of data; revising the article critically for important intellectual content, and final approval of the version to be published.Stephen Bacchi: Conception and design; acquisition of data; analysis and interpretation of data; revising the article critically for important intellectual content, and final approval of the version to be published.Brandon Stretton: Conception and design; analysis and interpretation of data; revising the article critically for important intellectual content, and final approval of the version to be published.Patrick G. O'Callaghan: Conception and design; analysis and interpretation of data; revising the article critically for important intellectual content; final approval of the version to be published.Elizabeth Murphy: Conception and design; analysis and interpretation of data; revising the article critically for important intellectual content, and final approval of the version to be published.Thomas J. Hugh: Conception and design; analysis and interpretation of data; revising the article critically for important intellectual content, and final approval of the version to be published.Robert T. Padbury: Conception and design, analysis and interpretation of data; revising the article critically for important intellectual content, and final approval of the version to be published.Markus I. Trochsler: Conception and design, analysis and interpretation of data, revising the article critically for important intellectual content, and final approval of the version to be published.Guy J. Maddern: Conception and design; analysis and interpretation of data; revising the article critically for important intellectual content, and final approval of the version to be published.

Table 1
Characteristics of the study cohort

Table 2
Variables remaining in logistic regression model for 30-day post-procedural all-cause mortality after backwards selection

Table 3
Variables remaining in logistic regression model for in-hospital all-cause mortality after backwards selection

Table 5
Results of multivariable logistic regression model for 30-day post-discharge all-cause readmission before backwards selection *P-value <0.05; **P-value <0.01; ***P-value <0.001.© 2024 The Authors.ANZ Journal of Surgery published by John Wiley & Sons Australia, Ltd on behalf of Royal Australasian College of Surgeons.

Table 6
Variables remaining in logistic regression model for hospital discharge to home after backwards selection