In this case-control study of adults who presented to the ED at a tertiary academic hospital, arrival by ambulance, absence of a family physician, a higher HEWS, and presence of dyslipidemia, were significantly associated with sepsis. Following adjustment, higher HEWS remained significantly associated with sepsis, confirming previous findings on the importance of vital signs at ED presentation in patients with an infection.11 In contrast, the SDoH available in health records were not significantly associated with sepsis. Despite this lack of association between SDoH and sepsis, our study highlighted that there were twice as many patients presenting to the ED with sepsis, and deaths from sepsis, among patients living in the Code Red Zone than among patients living outside of the Code Red Zone. Limited and inconsistent SDoH data available in health records may explain the lack of association between SDoH and the development of sepsis.

In a scoping review of SDoH and sepsis, studies evaluating SES, gender, smoking status, alcohol disorders, social support, age, and frailty all showed an association between the SDoH studied and sepsis.9 Four studies evaluating race showed a higher risk of sepsis among Black patients than among White patients, and two studies showed an increased risk among Black patients compared with White patients.9 Contrary to our hypothesis and the results of previous studies, SDoH, smoking status, social support, and alcohol disorders, were not significantly associated with sepsis. Individual or household income, race, and gender, as distinct from sex, were not available in health records and could not be studied.

These results can be explained by missing or unreported data in health records, a well-documented challenge when using electronic health records,16,17 and even more pronounced when integrating individual-level SDoH.18 For example, among critically ill patients, smoking and alcohol use are frequently underrecognized and underreported.19 Although studies have shown the contribution of individual-level SDoH to the prediction of 30-day readmission,20 HIV risk assessment,21 suicide attempts,22,23 and hospitalization, they have also highlighted the challenges of accurately collecting SDoH data. These challenges include the limited amount of SDoH data included in electronic health records, the difficulty in capturing accurate and complete data, and the lack of common measures or a minimum set of SDoH measures that should be recorded in a patient’s record.18 This emphasizes the need to collect and better integrate individual-level SDoH into health records to help guide and inform equitable care.

In addition to individual-level SDoH, neighbourhood-level SDoH including median household income were collected. The associations between median household income, residence in the Code Red Zone, and the incidence of sepsis were not significant. Other neighbourhood SES studies have shown an increased risk of microbiologically verified Staphylococcus aureus bacteremia,24 bloodstream infection,25 and higher rates of infection.26 Nevertheless, in line with our results, Donnelly et al. did not find an association between neighbourhood SES and the risk of sepsis at presentation among patients hospitalized with an infection.26 In addition to SES, residence in medically underserved areas (defined according to the ratio of primary care physicians per 1,000 people), infant mortality rate, the proportion of individuals with an income below the poverty line, and the percentage of the population over 65 yr of age, are also associated with a higher incidence of sepsis and higher rates of mortality due to sepsis.19 Although residence in the Code Red Zone was not significantly associated with sepsis, a disproportionate number of patients from the Code Red Zone were admitted from the ED to the hospital with sepsis, relative to the proportion of Hamilton residents living in the Code Red Zone. Similar to medically underserved areas, the Code Red project showed glaring disparities in social and economic factors, including ED visits, rates of cardiovascular incidents, respiratory emergencies, and, perhaps most concerning, a 21-yr difference in life expectancy between those living in and outside of the Code Red Zone.8,27 The increased number of patients presenting with sepsis from the Code Red Zone suggests that the incidence of sepsis may be affected by neighbourhood-level determinants, including access to care. These results may be transferable to other communities with similar structural challenges.

Arrival by ambulance and the absence of a family physician were also significantly associated with sepsis. Although the data in this study do not establish a causal mechanism between SDoH and sepsis, access to primary care and early detection and treatment of infections can limit sepsis cases. In this study, the median CCI value was the same for patients with sepsis and patients without sepsis, albeit CCI is only a measure of the number of comorbidities and does not account for how the conditions are managed. Access to primary care and a designated family physician can improve the management of chronic comorbidities and decrease the risk of developing sepsis.28 Therefore, access to care, as shown in our study and corroborated by the Code Red project8 and existing literature,29 is an important determinant for further studies. This includes understanding how structural determinants impact health literacy, for example, and delays in accessing care.

This study has limitations. First, this was a retrospective chart review of existing data, not originally collected for research purposes. Although efforts were made to ensure the quality and rigour of the data extracted, including extracting data independently and in duplicate, considerable challenges remained in collecting accurate and complete SDoH data. Second, in addition to quality issues and missing data, the limited amount of SDoH data available in health records prevented the analysis of other determinants, including individual household income and race. Lastly, this was a single-centre study of patients admitted to the ED of a large academic tertiary care hospital in Hamilton and was limited by sample size. It is not clear whether the results of this study can be generalized to other health care settings. Future prospective studies are needed to address data quality issues and the limited SDoH available in health records. This includes going beyond SDoH data inputted by health care workers and including self-reported measures such as race and ethnicity, among other self-reported SDoH.

Nevertheless, this study highlights the importance of adding of SDoH data into electronic health records and its implications for the future of sepsis research. Importantly, this work exposed the current challenges of integrating SDoH into health records, a crucial element to ensure equitable care and to inform health care policy.