Rapid systematic review on risks and outcomes of sepsis: the influence of risk factors associated with health inequalities

Abstract

Background and aims

Sepsis is a serious and life-threatening condition caused by a dysregulated immune response to an infection. Recent guidance issued in the UK gave recommendations around recognition and antibiotic treatment of sepsis, but did not consider factors relating to health inequalities. The aim of this study was to summarise the literature investigating associations between health inequalities and sepsis.

Methods

Searches were conducted in Embase for peer-reviewed articles published since 2010 that included sepsis in combination with one of the following five areas: socioeconomic status, race/ethnicity, community factors, medical needs and pregnancy/maternity.

Results

Five searches identified 1,402 studies, with 50 unique studies included in the review after screening (13 sociodemographic, 14 race/ethnicity, 3 community, 3 care/medical needs and 20 pregnancy/maternity; 3 papers examined multiple health inequalities). Most of the studies were conducted in the USA (31/50), with only four studies using UK data (all pregnancy related). Socioeconomic factors associated with increased sepsis incidence included lower socioeconomic status, unemployment and lower education level, although findings were not consistent across studies. For ethnicity, mixed results were reported. Living in a medically underserved area or being resident in a nursing home increased risk of sepsis. Mortality rates after sepsis were found to be higher in people living in rural areas or in those discharged to skilled nursing facilities while associations with ethnicity were mixed. Complications during delivery, caesarean-section delivery, increased deprivation and black and other ethnic minority race were associated with post-partum sepsis.

Conclusion

There are clear correlations between sepsis morbidity and mortality and the presence of factors associated with health inequalities. To inform local guidance and drive public health measures, there is a need for studies conducted across more diverse setting and countries.

Introduction

Sepsis is “life-threatening organ dysfunction caused by a dysregulated host immune response to an infection” [1]. A 2015 study estimated the in-hospital mortality rate for sepsis in UK hospitals to be around 30% [2]. As well as high mortality rates, sepsis survivors often experience longer-term mental and physical health problems and are at high risk of post-discharge hospital readmission or death [3,4,5]. Risk factors for developing sepsis include frailty, immunocompromised status, recent surgical procedures, and comorbidities such as cancer, kidney disease, lung disease and diabetes [6,7,8]. The risk of contracting sepsis increases with age, with many sepsis cases occurring in people over the age of 65 [9]. Additionally, there is a higher risk of sepsis in neonates and women who are pregnant or have recently given birth.

Bacterial infections are the most common cause of sepsis and therefore antibiotics are widely used for treatment. A 2022 report published by the United Kingdom (UK) Academy of Medical Royal Colleges (AMRC) outlined recommendations for the recognition and early management of sepsis [10]. An aspect the report did not address, however, was the impact of health inequalities on sepsis recognition, management and outcomes. Inequalities can impact life expectancy, access to healthcare and general health status. Factors that are associated with these disparities include level of deprivation, ethnicity and belonging to more vulnerable groups within society, for example people experiencing homelessness [11]. In 2021, the National Healthcare Inequalities Programme was set up and developed the Core20PLUS5 approach, with the aim of supporting the National Health Service and local authorities in reducing health inequalities [11]. Core20 refers to populations living in the most deprived 20% of areas according to the Index of Multiple Deprivation (IMD). PLUS refers to population groups identified at local level that could include ethnic minority groups, coastal communities, populations defined as having a protected characteristic under the Equality Act 2010 or belonging to an inclusion health group, amongst others. ‘5’ refers to five clinical areas of importance, which are maternity, severe mental illness, chronic respiratory disease, early cancer diagnosis, and hypertension case-finding.

Variations in rates of antimicrobial resistant infections and microorganisms (associated with higher mortality rates in sepsis [12]) have been reported in the UK amongst different ethnic groups and levels of deprivation [13]. A recent study reported increased odds of non-COVID 19 related sepsis and increased mortality in more socioeconomically deprived people during the pandemic [14]. In the face of increasing resistance to antimicrobials globally, knowing who is at greatest risk of developing sepsis may not only improve patient outcomes but help target the use of antimicrobials more effectively.

A 2019 systematic review assessed the link between race and socioeconomic status and sepsis outcomes. However, they only included studies conducted in the USA [15]. The purpose of this review, therefore, was to identify studies from all high- income countries that have assessed additional factors associated with health inequality. The aims of this rapid review were (i) to summarise the literature that investigated health inequalities and sepsis incidence and mortality outcome and (ii) to provide an evidence base for public health advice to reduce the impact of health inequalities with sepsis.

Methods

Eligibility criteria

Peer-reviewed journal articles published between 01/01/2010 and 31/01/2023, written in English, were eligible for inclusion. Included studies had to be observational in design where the main outcome was either incidence or risk of sepsis (in the general population or hospital admissions) or sepsis-associated mortality. We included studies where the aim was assessing the impact of one of the following health inequality factors: socioeconomic, race/ethnicity, community, medical vulnerability, or pregnancy. Studies were excluded if they were conducted in a low- or middle-income country (LMIC) (according to the World Bank, to minimise differences in healthcare systems), were not observational in design (intervention studies or qualitative studies), full text was not available, or abstract was published in conference proceedings.

Study selection

The database Embase (accessed through Ovid, last searched 25/03/3023) was used to search for relevant articles. Separate searches were carried out using the following terms in the titles of articles: sepsis OR septic in combination with one of the following groups of terms:

• Socioeconomic factors – depriv* or socioeconomic or socio-economic or socio or social or SES or IMD or income or occupation or education.

• Race/ethnicity factors – race or racial or ethnic* or minorit*.

• Community factors – urban* or rural or coast*.

• Medical vulnerability factors – residen* or care home or nursing home or care facility or living or social care or drug* or alcohol or disabil* or vulnerab*.

• Pregnancy – pregnan* or matern* or “post-partum” or “postpartum”.

Duplicated articles were removed. All articles identified in the search went through a title and abstract screening to exclude ineligible articles. A full article review was then performed on the remaining papers, with any ineligible articles identified during the full paper review being excluded. In accordance with the PRISMA guidelines, the reasons for exclusion at this stage were recorded (see Fig. 1). Any further duplicates (studies that appeared in multiple searches) were also removed. The searches were performed in Ovid and the results were downloaded to Mendeley Reference Manager to apply the inclusion/exclusion criteria. Data extracted from the eligible articles were stored in Microsoft Excel. The following information was extracted from each included paper: title, authors, year published, study design, country where study was conducted, data source, sepsis identification method, number of patients in sepsis cohort, factors associated with inequality used in study and how they are measured, outcome(s) assessed in the study and key findings of associations between the factors and outcomes. For reporting we referred to the PRISMA guidelines [16] for systematic reviews, however, as this is a rapid review not all items are relevant. Further details of the search strategy can be found in the supplementary information.

Results

Selection of sources of evidence

The five searches returned a total of 1,402 results (185 socioeconomic, 92 race/ethnicity, 126 community, 494 medical/care needs, 505 pregnancy). After deleting duplicates, 1,338 papers were screened on title and abstract, with 1,254 excluded. 108 papers underwent full article screening, after which 53 were eligible. Of these, there were 13 articles assessing socioeconomic factors, 14 race or ethnicity, 3 assessing community, 3 care/medical needs, and 20 assessing pregnancy and post-partum factors. As the searches and selection were conducted separately there were 3 papers duplicated between the searches, resulting in 50 unique papers to include. Flowcharts showing the selection process for each search are shown in Fig. 1.

Fig. 1

Flowchart showing search results and screening of studies. Five separate searches were conducted. After screening 53 papers were included

Characteristics of sources of evidence

Table 1 displays the characteristics of studies included in the review. The majority of the included studies (31 out of 50) used data collected in the United States of America (USA) [17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47], with four in the UK [48,49,50,51], three in Israel [41, 52, 53], two in Canada [54, 55], one in Australia [56] and the others in Europe [57,58,59,60,61,62,63] or not specified [64, 65].Most of the studies used hospital or ICU discharge databases [17, 21, 23,24,25, 28,29,30,31, 36,37,38,39, 41,42,43,44,45, 47, 50, 60], other sources included national birth or obstetric registries [48, 49, 51, 62, 63], death registry datasets [27, 35], secondary analysis on data collected for other cohort studies [18, 26, 61], and a US cities public health dataset [22]. Six identified sepsis in neonatal patients or infants [21, 47, 53, 55, 56, 63] and another only included children aged between 0 and 20 years [28]. The rest either specified adults only or did not specify any age restrictions for the cohorts.

Table 1 Characteristics of all included studies in the review

The most common method of identifying sepsis in the studies was based on ICD codes. ICD-10 codes were used in twelve studies [17, 20, 27, 28, 32, 40, 46, 54, 60,61,62,63], ICD-9 codes were used in twenty-one studies [20, 21, 24, 25, 29, 31,32,33, 36,37,38,39, 41,42,43,44,45,46, 52, 61, 65] and one study did not specify which ICD version [34]. Of these studies, some used specific codes for sepsis, severe sepsis or septic shock whilst others used more comprehensive sets of codes including the Angus criteria [66] or Martin criteria [67]. Both the Angus and Martin methods include the sepsis specific codes and non-specific ICD codes for infection in combination with a code for organ dysfunction. Five studies [18, 19, 22, 23, 30] used the 2016 International Consensus definition for sepsis, otherwise known as the Sepsis-3 criteria [1]. Other studies used Systemic Inflammatory Response Syndrome with [50] and without criteria for organ dysfunction [26, 64], medical chart review [48, 49, 51, 53, 55, 56, 59] or did not specify [22, 35, 47, 57, 58]. The size of the sepsis patient cohorts varied from 14 [56] to 16,779,820 [[33] with a median cohort size of 2,913.

Regarding outcomes, 15 studies assessed the incidence or risk of sepsis [18, 26, 29, 31, 37, 41, 42, 48,49,50, 54, 57, 59,60,61,62], 23 studies looked at in-hospital (or short-term) mortality [17, 19,20,21, 24, 25, 28, 31, 33, 34, 36, 37, 39, 43,44,45, 50,51,52, 54, 58, 59, 65], four studies assessed mortality after hospital discharge [30, 32, 38, 58], five studies assessed hospital readmission rates after discharge [23, 30, 32, 40, 58] and four studies calculated population-level sepsis mortality rates [22, 27, 35, 61]. Seven of the studies relating to pregnancy assessed adverse perinatal outcomes and incidence of sepsis in neonates [46, 47, 53, 55, 56, 63, 64].

Results of individual sources of evidence

Socioeconomic factors

The most common socioeconomic factors were income, level of education, employment status, unemployment rate and poverty rate. Others included were insurance status, occupation, cohabitation status and access to healthcare. There was variation in whether these were recorded at an individual level or matched to local data based on small area geographic identifiers (ZIP/postcode), and whether they were summarised into an overall score or included as individual covariates.

Five studies assessed the impact of socioeconomic factors on sepsis incidence or risk of developing sepsis. Factors found to be associated with increased risk of sepsis included low income [54, 60], low education level [57, 60, 61], lower socioeconomic status [18], marital/living status [54, 57] not being in work [54, 57], lower class of occupation and those who receive social benefits [61]. Three studies assessed 30-day or in-hospital mortality, which all found lower income was associated with increased risk of mortality when compared to the highest income groups. Another study reported decreased odds for highest household income quartile compared to the lowest quartile [20]. Hidalgo et al. [19] reported that unemployment and a neighbourhood poverty rate > 10% were all predictive of greater 30-day mortality. One study calculated population sepsis mortality rates per 10,000 persons and compared between income and poverty levels. Low-income neighbourhoods had a death rate of 3.65 (inter-quartile range (IQR) 2.78–4.40) versus high income neighbourhoods 2.80 (IQR 2.05–3.55) and high poverty neighbourhoods 4.20 (IQR 2.90–5.30) versus low poverty neighbourhoods 2.90 (IQR 2.00-3.60) [22]. For longer-term outcomes, two studies assessed the impact of socioeconomic factors on 30-day readmission after discharge. Lower income, lack of health insurance [23] and being more socioeconomically disadvantaged [23] were found to be associated with increased risk.

Race & ethnicity

Of the 14 studies assessing the impact of race or ethnicity on sepsis, 13 were based in the USA. Five of these studies only included race categorised as white or black/African-American [25,26,29,31,32], whilst other studies included categories for Hispanic [21, 24, 27, 28, 33, 34], Asian-American [24, 30], Asian/Pacific Islander (API) or Native American [21, 28, 33] and other/unknown. The study by Rush et al. [34] used a different approach by classifying hospitals as non-minority or minority, if the patient population of the hospital was more than twice the geographical census division mean.

The studies by Chaudhary et al. [31], Mayr et al. [29] and Moore et al. [26] compared rates of sepsis amongst either black or white populations only. Chaudhary et al. reported a higher sepsis rate for white patients compared to black patients with 109.4 cases (95% CI 109.2-109.6) per 1,000 hospitalisations versus 106.7 cases (95% CI 106.3-107.1). Moore et al. also reported a higher incidence of sepsis in white patients (9.10 per 1,000 person years) compared to black patients (6.93 per 1,000 person years). Contrary to these, Mayr et al. found a 67% higher severe sepsis hospitalisation rate in black patients (9.4 per 1,000 population) than white (5.6 per 1,000 population). All three studies covered hospital admissions in multiple US states, but they did differ in the age of patients included and severity of sepsis.

Eight studies considered the impact of race on in-hospital mortality in sepsis patients, with mixed results. Three studies reported higher mortality rates or increased risk of mortality in black or African-American patients than white patients [25, 28, 33] whilst Sandoval et al. [24] reported higher case fatality rates in white patients (15.1%) compared to black (14.0%), Hispanic (13.8%) or Asian patients (16.2%). One study [33] reported increased mortality rates in Hispanic patients compared to white patients, but two other studies did not find significant differences [21, 28]. Rush et al. reported unadjusted mortality rates at non-minority hospitals of 11.1%, compared with 12.3% (p < 0.001) at minority black hospitals and 12.7% (p < 0.001) at minority Hispanic hospitals. The only non-USA based study was based in Israel. Karp et al. [52] found that differences in risk of in-hospital mortality between Bedouin Arabs and the Jewish population could be explained by differences in age and Charlson comorbidity score.

For longer-term outcomes, one study [30] reported small differences in 90-day mortality rates between African American (18%), Asian-American (19%) and white (22%) patients. Lizza et al. [32] reported black patients had significantly higher rates of all-cause readmission (71.1% vs. 60.8%, p < 0.001) and sepsis readmission (19.8% vs. 14.0%, p < 0.001) than white patients. However, rates of post-discharge death were similar (white patients 36.5% vs. black patients 36.7%, p = 0.876). Ogundipe et al. [27] calculated age-adjusted sepsis death rates in non-Hispanic black, non-Hispanic white and Hispanic populations and reported lower death rates in Hispanic populations than non-Hispanic populations.

Community factors

The three papers included in the review that looked at community factors were conducted in the USA and used different ways of measuring urbanicity or rurality. Oud et al. [35] compared age-adjusted sepsis mortality rates between rural and urban communities from 2010 to 2019. The study reported in 2019 the overall rural rate was 57.9 deaths per 100,000, but in urban areas it was 48.3 deaths per 100,000 population. This was not a consistent pattern when adjusting for race. For example, in non-Hispanic blacks the urban mortality rates were higher than the rural rates. Ogundipe et al. [27] found the highest age-adjusted sepsis death rates were in non-metropolitan areas for both non-Hispanic black (micropolitan area 120.4 per 100,000 population, non-core area 109.4 per 100,000) and non-Hispanic white populations (micropolitan area 67.6 per 100,000, non-core area 66.4 per 100,000). Mohr et al. [36] assessed whether there were differences in patients in rural areas who attended their local emergency department or who bypassed their local hospital and travelled further to present to a hospital of top-decile inpatient sepsis volume. Sepsis patients who bypassed their local hospital had increased odds of mortality, with an OR of 1.26 (95% CI 1.03–1.53).

Medical needs

The three studies that considered factors relating to additional medical needs each used different measures. Goodwin et al. [37] identified patients living in medically underserved areas (MUA’s) based on the ratio of primary care physicians per 1,000 population, infant mortality rate, the proportion of the population with income below the poverty level and the proportion of the population over 65 years of age. The study reported higher incidence of sepsis (8.6 vs. 6.8 admissions per 1,000 people, p < 0.01) and mortality rates (15.5 versus 11.9 deaths per 10,000, p < 0.01) in MUA residents compared to non-MUA. Ginde et al. [39] included residence in a nursing home prior to an emergency department visit for sepsis, and reported increased risk of mortality for nursing home residents (OR 3.1, 95% CI 1.2–7.8). The study by Ehlenbach et al. [38] found that sepsis patients not discharged to a skilled nursing facility (SNF) had a mortality rate of 35.6%, while those discharged to a SNF but whom had not been resident in an SNF prior to sepsis had a mortality rate of 43.2% and patients who had been in a SNF before and after sepsis had a mortality rate of 52.8%.

Pregnancy/maternity

Studies assessing incidence of maternal sepsis reported rates of severe sepsis of 1.00 per 100,000 [48] maternities, 4.7 per 10,000 [49] maternities and 4.9 per 10,000 live births [42]. Estimates of non-severe sepsis included 198.69 per 100,000 [59] maternities, 2.4 per 10,000 women [62] and 10 per 10,000 live births [42]. Acosta et al. [50] estimated the absolute risk of maternal critical care unit admission with severe sepsis was 4.1 per 10,000 maternities (95% CI 2.9–5.6). Factors including increased BMI [41, 50, 62], older age [42, 50, 62], black and other ethnic minority race [49], increased levels of deprivation [50], African American race [41], pre-existing medical conditions [41, 49], complications of delivery and delivery via caesarean Sects [41, 49, 50, 59, 62] were found to be associated with an increased risk of developing maternal sepsis.

Two studies [43, 44] assessing mortality in maternity patients reported lower case-fatality rates in pregnancy associated severe sepsis (PASS) compared to non-pregnancy associated severe sepsis (NPSS). Maternal mortality rates in other studies varied, with reported rates of 10% [65], 10.7% [51], 1.8/100,000 maternities [50] and no deaths in one study [59]. Increased BMI [50], being in the most deprived two IMD quintiles [50], pre-existing medical conditions [51] and being multi-parous [51] was found to be associated with increased maternal mortality. Antepartum sepsis was found to be associated with increased risk of placental dysfunction and maternal ICU admission during delivery hospitalization [46]. Five studies considered outcomes relating to early onset neonatal sepsis (EONS), with reported rates of 1.03 cases per 1,000 live births [53] and 1.48 per 1,000 live births [63]. Risk factors associated with EONS were maternal exposure to antibiotics [47, 53, 55], maternal BMI [63], caesarean section delivery [63] and gestational age [47].

Supplementary Tables 1 and 2 show the findings from all included studies and can be found in additional file 1.

Discussion

Summary of evidence

Socioeconomic factors associated with increased incidence of sepsis included lower socioeconomic status, unemployment, and lower education level, although findings were not consistent across studies. Studies assessing the association between ethnicity and sepsis rates reported mixed results, with two studies finding increased sepsis rates in white populations compared to black populations and another showing higher rates in black populations than white. Living in a medically underserved area or being resident in a nursing home was also shown to increase risk of sepsis. In terms of mortality, lower income, unemployment, and poverty levels were all associated with increased in-hospital mortality. In studies considering effects of ethnicity on in-hospital and longer-term mortality the results were mixed, with some studies finding no significant associations, some reporting increased odds of mortality in non-white populations and others reporting increased mortality in white populations. Sepsis mortality rates were also found to be higher in people living in rural areas and those who were resident in a skilled nursing facility.

It is notable that the literature is dominated by research conducted in the USA and none of the studies identified under the non-pregnancy related searches used UK data. This is an important consideration for healthcare and public health professionals outside of the USA as differences in structural inequalities between the USA and other high-income countries may make the results less generalisable. The majority of studies focused on in-hospital mortality as the primary outcome, so there also needs to be more focus on the risks of developing sepsis and longer-term outcomes such as healthcare utilisation.

The sources of data varied between the studies, as did the methods of identifying sepsis. Differences in sepsis definitions leads to different reported prevalence/cohort sizes [68]. Some of the studies were based in single centres and only included a few hundred patients, whilst others represented national populations and included millions of patients. Many of the studies used data from secondary care only and none used primary care data, even though the majority of cases of sepsis develop in the community rather than the hospital. Additionally, there was a lot of variation in measures used in the analyses, particularly in the studies assessing socioeconomic factors, where there was no standardised definition of socioeconomic status and therefore results varied. There were some studies who assessed a combination of socioeconomic, community and race factors, however, some only focused on one area related to health inequality. This is important as there is overlap between the different areas. The paper by Vazquez Guillamet et al. [20] concluded that race did not have a significant effect on sepsis mortality when accounting for socioeconomic variables. A commentary piece published in 2018 by Shankar-Hari and Rubenfeld [69] titled “Race, ethnicity and sepsis: beyond adjusted odds ratios” suggested that there needs to be more research into the underlying causes of race/ethnicity disparities not just in sepsis but in wider health areas. Future studies should take into account not just socioeconomic status and population demographics, which will likely vary between ethnic groups, but also consider the intersectionality between these and other factors such as comorbidity levels and health behaviours e.g. smoking, alcohol use or exercise.

Limitations

Due to the rapid nature of the review the scope was limited and the search strategy not as comprehensive as for systematic reviews. We searched for the key terms in the titles only, searched a single database (Embase) and only included studies published from 2010 onwards. We also acknowledge that pathogen specific publications which do not specifically include the word sepsis may have been screened out. Examples include those that report on invasive group A and B streptococcal disease [67]. Studies conducted in LMICs were excluded as the results will be less generalisable to the UK population. Whilst the burden of sepsis is highest in LMICs there is a lack of good quality data from these countries [70]. The challenges in recognising and managing sepsis within LMICs, such as lack of access to healthcare, malnutrition and infrastructure [71], are not as applicable in higher- income countries. Some aspects of the Core20PLUS5 approach to addressing health inequalities were not included in this rapid review. These mainly related to inclusion health groups, including people with multi-morbidities, vulnerable migrants, Gypsy, Roma and Traveller communities, sex workers, people in contact with the justice system and victims of modern slavery. Additionally, the 4 comorbidities/conditions within the ‘5’ component other than maternity were not areas of focus (severe mental illness, COPD, cancer & hypertension). The five included areas were chosen as they are the factors that cover the largest groups in the population and were identified as the most important. Although we did not include the other areas in our review it is still vital that future studies consider these aspects in order to address all potential influences on sepsis risks and outcomes. The bias of the included studies was not assessed, nor did we critically appraise them.

Future work

From the studies identified, there are clear correlations between sepsis morbidity and mortality and the presence of factors associated with health inequalities. There is a need for UK based studies, using nationally representative data, to better understand how factors associated with health inequalities affect sepsis incidence and mortality in the UK population. With the availability of electronic health record data for research there are increasing opportunities to disaggregate the data and stratify risk by patient demographic. For example, in the UK the Clinical Practice Research Datalink (CPRD) and Hospital Episode Statistics (HES) data provide nationally representative primary and secondary care records with linkage available to deprivation and socioeconomic scores. Once this is better understood, healthcare and public health professionals can be empowered to close the health gap and reduce inequalities through targeted recommendations for the recognition and early management of sepsis. Recent guidance in the UK highlighted the importance of early intervention in sepsis whilst balancing that with the need to use antibiotics more appropriately. Understanding which patients are at greater risk of sepsis mortality and morbidity, in terms of the factors associated with inequalities discussed in this review and other known risk factors, may help clinicians target antibiotic use more effectively. Given the lack of evidence from outside the USA, there is not sufficient information available to inform policy, at either a global level or an individual country level (except the USA). Although some of the findings from USA studies may be generalisable to other settings there needs to be further exploration of the similarities and differences in inequality factors in different populations. Critical to the above is improved coding in electronic health records alongside appropriate data linkage.

Conclusion

Factors relating to health inequalities such as deprivation and ethnicity have been shown to be associated with poorer outcomes in COVID-19 and increased rates of antimicrobial resistance. In order to inform local guidance and drive public health measures, there is a need for studies conducted across more diverse setting and countries.