1 Introduction

Addressing health inequities requires improved measurement, monitoring, and reporting of population health status and its association with social determinants of health [1,2,3]. Avoidable mortality, a subset of premature mortality, is widely used by public health researchers to measure population health, and many related methodologies have been proposed for doing so. This scoping review presents a comprehensive view of global peer-reviewed and grey literature exploring the association between socioeconomic status (SES) and avoidable mortality. The studies tend to vary not only in how they define avoidable mortality but also in how they measure SES and the association between them. We also report on how these studies do or do not standardize their results for age and sex and encapsulate their main findings with respect to SES inequalities. Our aim is to facilitate a more widespread and in-depth analysis of inequities in avoidable mortality by consolidating these resources in one location.

Avoidable mortality is a relatively new concept, with corresponding methodologies still under refinement. In 1976, the Working Group on Preventable and Manageable Diseases, chaired by Rutstein, introduced the new concept of “unnecessary untimely death” as a health outcome indicator designed to evaluate the quality of medical care [4]. Subsequently, an American research group empirically applied this concept to “demonstrate the usefulness of this approach” as a measure of healthcare quality [5, 6]. The term “avoidable mortality” came into use following a study by Charlton et al. investigating geographical variation in mortality across England and Wales [7]. While premature death is defined as mortality that happens before a certain age, for example, before 75 years of age [8], “avoidable mortality” refers to premature deaths that could have been avoided in the presence of timely and effective health and social policies and public health interventions aimed at addressing the social determinants of health or reducing risk factors contributing to ill health [8]. Nevertheless, researchers do not adopt a uniform approach to conceptualizing “avoidable mortality,” and it is not always clear whether the terms “avoidable,” “amenable,” “preventable,” and “unnecessary” causes of death are being used interchangeably in the literature [9, 10]. Over time, researchers have refined and updated the list of avoidable causes of death used to classify mortality events as avoidable to suit the situational context of their research [8, 11,12,13,14,15,16]. Recently, there have been efforts to propose standardized global definitions of avoidable mortality, as seen in the work of the Organization for Economic Co-operation and Development (OECD) and the Office for National Statistics (ONS) [17,18,19]. These standardized definitions are pivotal for reliable comparison of health indicators within or between countries [20].

Avoidable mortality is an important summary measure of overall population health status [8], and one of its many applications has been as an indicator of health equity [5, 21]. Researchers have used avoidable mortality to investigate health inequity with the goal of informing political decisions trying to address health inequity [11, 22,23,24,25,26]. Socioeconomic status (SES) is one of the most important “fundamental causes” of disparities in morbidity and mortality [27, 28]. People with lower SES have access to a limited range of resources, such as money, knowledge, prestige, social capital, and power, which could affect their health through many pathways; lack of these resources exposes individuals to a higher risk of mortality [27]. These deaths could be avoided by appropriate means such as timely and effective treatment or the application of health and social policies.

2 Methods

We used Arksey and O’Malley’s framework [29] to conduct this scoping review in five stages:

2.1 Identification of studies

Our central question in this review was “How have avoidable mortality and inequalities therein been defined and operationalized in studies investigating SES inequalities in avoidable mortality?” To answer this question, we developed a search strategy (see Additional file: Appendix A) to find relevant articles by searching Ovid Medline, Scopus, and Web of Science, narrowing down the results to English language and publication dates from 2000 to 2020. We omitted PubMed in our search since the Ovid Medline interface allows a more focused search, excluding citations such as “in process” and “ahead of print” articles [30], which were not of interest to the current review. All the searches were conducted on June 18th, 2020. For grey literature, first, we searched several leading international websites, including Google, Google Scholar, The World Health Organization (WHO) [31], United Nations [32], and Organization for Economic Co-operation and Development (OECD) [33]. We expanded our search by choosing Canada as a sample for a more thorough grey-literature search since our research team is located in Canada. We searched leading Canadian health information websites, including Public Health Agency of Canada (PHAC) [34] restricting our review to the first 5 pages, with 10 results on each page. The following websites were also searched with no restriction on page numbers of the search results: Canadian Institute for Health Information (CIHI) [35]; National Collaborating Center for Determinants of Health [36]; and Ontario Public Health Libraries Association (OPHLA)-Custom Search Engine for Canadian Public Health information [37]. For this study we also partnered with the Urban Public Health Network (UPHN), an association of urban local public health units in Canada; in order to ensure coverage of their membership, we also included their websites in our grey literature search (see Additional file: Appendix B) with no restriction on the page numbers of the search results. The keywords searched to capture relevant grey literature included “avoidable mortality,” “avoidable death,” “preventable mortality,” “preventable death,” “amenable mortality,” and “amenable death.” We did not include the keyword “premature death” as we were focused on avoidable mortality.

2.2 Screening and identifying relevant studies

All the articles were imported into Rayyan® for deduplication and title and abstract screening. The first author reviewed the articles by title and abstract to find relevant studies using the inclusion–exclusion criteria in Table 1. We were only interested in articles investigating socioeconomic status indicators as they are amongst those most frequently implicated as a contributor to inequities in health [38]. Although there is debate about how and whether SES can be measured [39], researchers often consider factors such as income, education, employment, and occupation to assess SES, analyzing them either individually or combining them into a single indicator. These indicators are usually measured at the household or area level so that they can relate to children’s health outcomes as well. We excluded articles investigating a specific cause of death or premature mortality since they are different from indicators of avoidable mortality. As we were interested in studies investigating the entire avoidable mortality at-risk population (aged 0 to 65 or 75), we excluded articles that limited their analysis to only a subset of this population, such as one sex, race, or children. The strict inclusion–exclusion criteria and minimal risk of misclassifying articles led us to use a single reviewer for screening and selecting the articles.

Table 1 Inclusion and exclusion criteria for selecting the articles and grey literature
Full size table

2.3 Selecting eligible studies

After screening the titles and abstracts, the selected articles were transferred from Rayyan® to Paperpile® for in-depth, full-text review. At this stage, the first author evaluated the complete texts using the inclusion–exclusion criteria, determining the final article selection. Once all articles from the searched databases were assessed, the first author applied the criteria outlined in Table 1 to evaluate results from the grey literature search and decide on their inclusion in the review. Finally, the reference lists of all chosen articles were examined to identify any additional pertinent studies.

2.4 Charting the data

We created a spreadsheet to catalog characteristics and data from the chosen articles and grey literature. This form captured details such as the study's timeframe, design, region examined, unit of analysis, population, SES indicators, definition of avoidable mortality, upper age boundaries, avoidable mortality measurement methods, and techniques used to analyze the relationship between avoidable mortality and SES. Additionally, we documented findings regarding the association between SES and avoidable mortality, the International Classification of Diseases (ICD) codes employed to identify avoidable deaths, and the classification of Ischaemic Heart Disease (IHD).

2.5 Collating, summarizing, and reporting the results

Finally, we compiled, summarized, and presented the gathered data. We highlighted the definitions of avoidable mortality used across the articles and determined whether Ischaemic Heart Disease was considered an avoidable cause of death. We summarized the insights from the articles regarding SES disparities in avoidable mortality and identified the SES indicators employed. Moreover, we compared the list of avoidable mortalities of a sample of five articles that used different definitions of avoidable mortality to identify avoidable deaths.

3 Results

3.1 Search results

In total, we identified 2457 articles. Of these, 2436 articles were obtained by searching the three predetermined databases using our tailored syntax (see Additional file: Appendix A). From grey literature, we found 16 articles, and 5 more articles were identified by examining the reference lists of the included studies. The database search yielded 810 articles from Ovid Medline, 669 from Web of Science, and 957 from Scopus. After removing duplicates, 1393 potential records underwent title and abstract screening. Of these, 1301 were deemed irrelevant to our research and consequently excluded. This left 92 articles for a full-text review. From these, 58 were further excluded with the reasons for exclusion documented. Figure 1 offers a PRISMA flowchart detailing the selection process of this scoping review.

Fig. 1
figure 1

PRISMA flow diagram

Full size image

The grey literature search results were as follows: the World Health Organization (0 records); Public Health Agency of Canada (0 records); United Nations (1 record found and excluded after screening); OECD (1 record found and excluded after screening); CIHI (2 records found, 1 was included after screening); National Collaborating Center for Determinants of Health (1 record found and excluded after screening); OPHLA-Custom Search Engine for Canadian Public Health information (6 records found and 3 were included after screening). Additionally, we identified 5 records from the UPHN members' websites, of which one report was considered suitable for inclusion after the screening process.

In conclusion, 34 articles were selected for detailed review and data extraction, comprising 29 scientific papers and 5 grey literature documents.

3.2 Study designs and distribution by region and timeline

Studies of avoidable mortality heavily rely on having access to previously collected death data. Analyzing and disseminating this data demands time, so the study coverage years might not always match their publication dates. Figure 2 contrasts these trends, showcasing the years studied in blue and publication years in orange. A noteworthy observation is that a significant portion (7 out of 34 articles) were unveiled in 2019, with 4 classified as grey literature. Moreover, there's been a rising trend in the frequency of data collection annually from 1971 to 2016. Regarding the region of study, our findings indicate that the majority of the chosen articles originated from European nations, specifically England, Finland, Spain, France, Switzerland, and Hungary.

Fig. 2
figure 2

Annual data collection frequency in selected studies (blue line) and publication dates (orange line)

Full size image

Most of the selected studies adopted quantitative observational approaches. To explore disparities in avoidable mortality, researchers employed a myriad of designs such as population-based, ecological, longitudinal, cross-sectional, prospective cohort surveys, exploratory spatial analysis, and official reports. The majority established an upper age threshold of 75 years. However, three distinct studies concentrated solely on deaths before age 65, covering the spans of 1971–2008, 1997–2001, and the 1990s, respectively [40,41,42].

3.3 Unit of analysis

The “unit of analysis” pertains to the level of grouping at which researchers measured avoidable mortality, subsequently stratifying these measurements based on SES to examine disparities. Most studies used small areas as their unit of analysis although the size of these areas varies between concepts and countries. These areas include the Lower Super Output Area (LSOA) in England [43, 44], Dissemination Area (DA) [8, 21, 44,45,46,47], health region [48], and Census Tract (CT) in Canada [49] and Spain [23, 24], commune and canton in France [41, 42], Local Government Area (LGA) [26, 50] and Statistical Local Area (SLA) [51, 52] in Australia, meshblock in New Zealand [53], districts in Brazil [54], neighborhoods [25, 55,56,57], and other small areas [11, 22, 58].

Only two studies used individual-level data for their analysis [59, 60]. A study carried out in Taiwan examined inequality between townships (city districts) [40], and three articles investigated SES inequality in avoidable mortality at the municipal level [61, 62]. Surenjav et al. examined both provincial and municipal (capital) data [63], and Neethling et al. compared provinces and population groups in their analysis [64]. Table 2 presents a summary of the selected articles' characteristics.

Table 2 Summary of the selected articles
Full size table

3.4 Avoidable mortality definitions and terminology

Researchers have used various definitions of avoidable mortality as the foundation for their studies and referred to it using different terminology. Table 3 displays the definitions adopted within the selected articles. While numerous studies adopted definitions proposed by other researchers or established organizations, several articles introduced their own distinct definitions of avoidable mortality, drawing from pertinent literature and consultations with expert panels.

Table 3 Avoidable mortality definition across the selected articles
Full size table

There is a difference in how researchers phrase and conceptualize the type of mortality deemed avoidable. The term “amenable mortality” is mainly used to describe deaths that could have been averted through medical interventions. This term is often used interchangeably with “treatable mortality” in the literature [12, 65, 66]. A definition by Nolte and Mckee, which is widely used in the selected articles (cited in 11 studies when identifying avoidable deaths,) characterizes “amenable mortality” as premature deaths that could have been avoided through timely and effective health care (i.e., secondary and tertiary prevention) [12]. Consequently, their list of amenable causes does not include deaths preventable via public health policy interventions [12] which are referred to as “preventable mortality” and assessed by other researchers [46, 47, 67].

Conversely, certain definitions, such as the one tailored for the Canadian setting, do not use the term “amenable mortality” [8]. The Canadian Institute for Health Information (CIHI) categorizes “potentially avoidable mortality” into preventable and treatable mortality. Here, “preventable mortality” refers to deaths before the age of 75 that could have been potentially avoided through public health initiatives (i.e., primary prevention.) Meanwhile, “treatable mortality” pertains to premature deaths that could have been potentially avoided through health care interventions and treatments (i.e., secondary and tertiary prevention) [8].

Another important definition, which was published after our search, is that offered by the OECD [17]. In 2022, the OECD and Eurostat partnered to create joint lists of “avoidable” causes of death [17]. This definition builds upon the lists developed by Nolte and McKee [12], Eurostat [68], and CIHI [8], categorizing avoidable mortality into “preventable” and “treatable” mortalities. These categories are aligned with CIHI's definitions [17]. Within this framework, the term “amenable” from the prior Eurostat list has been rephrased as “treatable.” Similarly, the NHS Outcome Framework [69] along with the Australian and New Zealand Atlas of Avoidable Mortality [66] classify avoidable deaths into “amenable” and “preventable” categories, resonating with CIHI's definition of “treatable” and “preventable” mortality.

Simonato et al. [13] and Tobias and Jackson [11] use unique terminology to define avoidable mortality. They partition avoidable mortality into three categories, each attributed to a set of mortalities that could be averted through one of the levels of prevention (i.e., primary, secondary, and tertiary levels). A detailed breakdown of the commonly used definitions in the selected articles is provided in Table 4.

Table 4 Description of frequently used definitions of avoidable mortality
Full size table

3.5 Operationalization of avoidable mortality measures

Measures of avoidable mortality vary with respect to their operationalization using ICD codes (e.g., whether IHD is classified as avoidable). We evaluated the list of avoidable mortalities from five articles chosen at random and found significant variances. Table 5 presents these five articles, each tagged with a unique article code. These article codes serve as references when comparing their respective lists of avoidable mortality by ICD codes, as detailed in Additional file: Appendix C. These differences largely arise from the adoption of varied definitions of avoidable mortality. Another contributing factor is the timeframe in which these lists were created; perceptions of avoidability evolve over time due to advancements in healthcare services and public health.

Table 5 Articles chosen for comparison of their lists of avoidable causes of death
Full size table

Despite the diverse views on which causes of death are deemed avoidable across the selected articles (see Additional file: Appendix C) Ischemic Heart Disease (IHD) remains a focal point in discussions about avoidability and its measurement. Among the chosen articles, there is no consensus on whether death due to IHD should be labelled as an avoidable cause. Numerous studies subdivide IHD into two [8, 21, 56, 57, 70] or even three [11, 50] subsets of avoidable mortality. Several studies included only half of the IHD mortality in their calculation of amenable mortality rates as proposed by Nolte and McKee [12, 25, 53, 55, 58]. Additionally, some articles opted out of classifying IHD as an avoidable cause, excluding it from their calculations of avoidable mortality rates [22, 24, 42]. (See Additional file: Appendix D for the number of studies that followed each approach). Taking note of these various approaches is particularly important in health inequality research as IHD is known to be one of the leading conditions in health inequality [71, 72].

3.6 Socioeconomic status indicator

We classified the articles based on the type of indicator used for socioeconomic status (SES) into four categories, namely index, income, education, and employment. The index category includes any article using a predefined index for SES, such as the Ontario Marginalization Material Deprivation Index (ON-Marg) [44, 56], Index of Relative Socioeconomic Disadvantage (IRSD) [26, 50, 51], and Index of Multiple Deprivation (IMD) [43, 44]. Some authors constructed a variable representing SES in a population using indicators such as income and education [23, 24]. We placed these studies in the index group as well. To decide whether to include or exclude an article that used an SES index, we carefully examined the index to ensure that at least one of our indicators of interest (i.e., education, income, and employment/occupation) was taken into account in its construction. One study in South Africa used apartheid classifications (Africans, Whites, Asians, and Coloureds) as a proxy for socioeconomic status, arguing that income and education disparities continued to persist in the population even after apartheid was ended [64].

Most of the articles (n = 20) used an index to investigate SES inequality in avoidable mortality. This number was 13 for income indicators, 4 for education, and 2 for employment. If a study examined more than one category in its analysis, we counted each category separately. For example, if an article examined income and education inequality in avoidable mortality separately, we counted this article once for the income category and once for the education category.

3.7 Standardization for age and sex

Most of the included articles used directly age-standardized mortality rates to assess avoidable mortality. In some cases, mortality rates were also standardized by sex. However, some studies applied unique methodologies. For example, one study estimated smoothed standardized mortality rates using the Bayesian model proposed by Besag et al. [73], to address the problem of age standardization in small areas [22]. A Hungarian study calculated mortality amenable to health care ratios using full hierarchical Bayesian methods. In doing so, the authors calculated smoothed indirectly standardized mortality ratios using sex- and age-specific rates for the Hungarian population [62]. A Canadian study calculated the age-standardized expected years of life lost (SEYLL) rate instead of the mortality rate, using the life expectancies of the richest income quintile as the standard [49]. Another methodology examined the contribution of different groups of mortality, including death amenable to health care and death amenable to health policy, to life expectancy at age 35 and partial life expectancy between 35 and 75 [59]. Doing so, Manderbacka et al. were able to assess the impact of health policy and care on income disparities in life expectancy in Finland [59]. In a study conducted by Masters et al. the authors tested the central claims of Fundamental Cause Theory (FCT) and then performed a retrospective cohort study to examine the association between preventable mortality and educational attainment [60]. An ecological study conducted in France calculated the standardized mortality ratio by dividing the observed mortality in a spatial unit by the corresponding expected mortality [41].

3.8 Association between SES and avoidable mortality

The findings of the selected articles consistently indicate a negative relationship between SES and avoidable mortality rates, where the avoidable mortality rate is higher among those with lower socioeconomic status. This aligns with the substantial body of knowledge indicating the role of SES in population health [2, 3, 27, 74, 75]. Khan et al. found a similar trend among immigrants and long-term residents, with a downward gradient in age-adjusted avoidable mortality rates as income quintiles increase [45]. However, a study conducted in Mongolia reported no significant relationship between avoidable mortality rate and the percentage of poor households [63]. This could be due to the lack of individual or neighborhood-level data in this study, which focused on the capital and provincial levels [76]. Nevertheless, the study revealed higher amenable mortality rates in remote western provinces in Mongolia, characterized by harsh weather conditions, high poverty rates, lack of human resources for health, and poor infrastructure [63]. Details on the findings of the selected articles are provided in Table 2.

While some studies adopted a descriptive approach and compared the avoidable mortality rates between different SES groups, many articles examined SES inequality in avoidable mortality using various types of analysis, including Poisson regression models, calculating Slope Index of Inequality (SII) and Relative Index of Inequality (RII), disparity rate ratio, incidence rate difference, random coefficient growth curve modeling approach, ecological regression, among other analytical methods.

The majority of findings from the captured studies indicate a decline in avoidable mortality rates over time while socioeconomic disparities persisted [43, 54, 56]. A study conducted in South Africa found that the socioeconomic disparity widened from 2000 to 2005 and narrowed thereafter until 2012 [64]. Similar patterns were observed in France, with a 40% higher avoidable mortality rate for the fifth deprivation quintile communes compared to the first quintile in 1988–92, and 78% higher in 1997–2001 [42]. A study conducted in Taiwan reported a faster decline in avoidable mortality from 1971 to 2008 in affluent townships compared to less affluent townships [40]. Similarly, an Australian study found a larger decline in avoidable mortality rates in higher socioeconomic groups, with increasing relative inequality and decreasing absolute inequality between 1986 and 2002 [52].

4 Discussion and conclusion

The findings of this review highlight the variability in terminology and definition for the concept of avoidable mortality used in health inequality research. Using common indicator definitions are essential for reliable comparison of health indicators within or between countries [20]. Recently, an OECD working group created a harmonized avoidable mortality definition to be used internationally [17, 77]. This definition can potentially address the gap we identified in the literature concerning the lack of consensus among researchers regarding the measurement of avoidable mortality. However, while there's a pressing need for standardized terminology and definition for avoidable mortality, it is important to recognize that a single, universal list of avoidable causes of death applicable in all countries and contexts is not likely feasible. The avoidability of specific causes of death may vary across countries with different levels of advancement in medical sciences. Therefore, it is crucial to consider the contextual factors and available healthcare technologies when developing lists of avoidable causes of death, acknowledging the variations across low-, middle-, and high-income countries. Nonetheless, this should not deter the establishment of a standardized definition, which is pivotal for supporting global comparability in this area of research.

The majority of studies captured in our scoping review were conducted in high-income countries, with Europe having the highest number of publications, followed by Australia and New Zealand. This trend may be attributed to the presence of national or regional standard definitions for avoidable mortality in these countries, such as the well-established list of avoidable causes proposed for use in the Australian [66] and European [78] contexts. Nonetheless, it is worth noting that the availability of the list of avoidable causes of death primarily designed for high-income countries may hinder the conduction of avoidable mortality research in lower-income countries.

The absence of an established list of avoidable causes of death can be one reason for the low number of research conducted in low-income countries as compared to high-income countries. This difference in the number of studies can also be explained by numerous other factors, including the unequal distribution of money, power, and other resources between low- and high-income countries. Nevertheless, the impact of the absence of an established list of avoidable causes of death on the number of studies can also be observed in high-income countries like Canada. The introduction of Canada’s first list of potentially avoidable causes of death in 2012 by CIHI [8] has led to a significant increase in inequality research on avoidable mortality in the Canadian literature [8, 21, 56, 57, 70]. Four out of five Canadian governmental reports included in this review were published in 2019. These findings underscore the crucial role played by established definitions and lists of avoidable causes of death in promoting research on SES-related inequalities in avoidable mortality.

Furthermore, it should be noted that our data collection took place during the early stages of the COVID-19 pandemic. Consequently, the literature captured and discussed in this scoping review does not include the subsequent debates regarding the consideration of COVID-19 as an avoidable mortality and its association with SES. Since then, organizations such as CIHI [79] and OECD/Eurostat [17], have revised and updated their lists of avoidable causes of death to incorporate COVID-19.

4.1 Study limitations

A potential limitation of this scoping review is that only one reviewer screened the articles. However, given the rigorous inclusion–exclusion criteria we used, we are confident that the risk of misclassification of articles remains minimal. In addition, this scoping review focused on studies of avoidable mortality that included analysis of SES-related inequalities. However, there exists a body of literature exploring inequality in avoidable mortality in terms of social determinants of health other than socioeconomic status, such as ethnicity. For instance, in their study of avoidable mortality in South Africa, Debbie Bradshaw and colleagues used apartheid classifications as a proxy for socioeconomic status. While we decided to include this particular study in our scoping review, we acknowledge that studies not considering SES indicators in assessing inequalities and those that did not investigate disparities may have used additional conceptualizations and definitions of avoidable mortality.