Since late 2019, the Severe Acute Respiratory Syndrome Corona Virus type 2 (SARS-CoV-2) and the corresponding coronavirus disease (COVID-19) have rapidly spread worldwide, leading to the declaration of COVID-19 as a pandemic in March 2020 by the World Health Organization (WHO) [1]. To date, the pandemic has led to approximately 397 million cumulative cases and 5.7 million cumulative deaths globally and remains a significant challenge for societies worldwide [2, 3].

Knowledge of the social epidemiological patterns in the distribution of SARS-CoV-2 infections and COVID-19 outcomes was limited at the beginning of the pandemic [4]. However, as soon as the first phases of the pandemic, several studies found socioeconomic inequalities in the risk of infection with SARS-CoV-2 and COVID-19 outcomes [4–7]. These early findings were confirmed by several international studies over the further course of the pandemic that presented additional evidence that socioeconomic inequalities in COVID-19 outcomes were observable in a variety of different national settings and at different time points during the pandemic [8].

But as with health inequalities in general [9], socioeconomic inequalities in COVID-19 outcomes may change or reproduce over time, leading to specific social epidemiological patterns of disease distribution during different phases of the pandemic. Studies from, e.g., Germany [10, 11] Hong Kong [12], and the United States [13] have described temporal dynamics from initially higher infection rates in more affluent populations and a later crossover to higher rates in socioeconomically disadvantaged populations. However, knowledge about specific temporal patterns of socioeconomic inequalities in COVID-19 remains limited and to date has not been systematically reviewed. Scientific evidence of temporal dynamics in the social epidemiological patterns of COVID-19 outcomes across populations will be vital to developing more targeted interventions and to guide future pandemic preparedness.

Using the example of the H1N1 pandemic in 2009–2010, Quinn and Kumar [14] emphasized the necessity of considering socioeconomic inequalities in general in pandemic preparedness plans. Because socioeconomically disadvantaged populations are generally at higher risk to get ill and have fewer resources to prevent infections or a severe course of the disease, an effective pandemic preparedness plan that addresses inequalities in exposure, susceptibility, and healthcare access is crucial to prevent or reduce an ongoing infectious trajectory, especially in today’s globalized world. Understanding the time-dependent patterns of socioeconomic inequalities in COVID-19 may therefore be helpful in identifying high-risk groups at different phases of the pandemic and inform targeted and timely public health interventions to reduce health inequalities and the overall burden of disease.

We therefore conducted a scoping review to map and synthesize the available evidence on temporal dynamics of socioeconomic inequalities in COVID-19 incidence, hospitalizations, and mortality.

This scoping review follows the methodological steps described by Arksey and O’Malley [15], i.e., 1) identify the research question; 2) identify relevant literature; 3) study selection; 4) charting the data; and 5) collating, summarizing and reporting the results. We used the PCC mnemonic (Population, Concept, Context) to formulate the research question aiming to achieve comprehensive coverage of the available scientific literature [16]: What is known about the temporal dynamics of socioeconomic differences in COVID-19 incidence, hospitalizations, and mortality (Concept) during the pandemic (Context) in the context of high-income countries (Population)?

A study protocol was published in the Open Science Framework [17]. The present article follows the Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for Scoping Reviews (PRISMA-ScR, Supplementary Material S1) [18].

We identified 46 full-text articles that met all eligibility criteria. Figure 1 shows a PRISMA flow chart with detailed information on the study selection process. We achieved an interrater percent agreement of 99.6% and a Cohen’s Kappa coefficient of 0.80 in the title and abstract screening. In the full-text screening, we achieved an interrater percent agreement of 88.9% and a Cohen’s Kappa coefficient of 0.71. The extracted data are presented for each study in a summary table (Table 2).

Most of the included studies were conducted in the United States (US, n = 23). The remaining studies were conducted in European countries (n = 20), such as Germany (n = 6), Italy (n = 4), and the United Kingdom (UK, n = 4), with three exceptions that were conducted in Hong Kong (n = 2) and Israel (n = 1). All of the included studies were observational; nine were published in 2020, while 37 were published in 2021. Most of the COVID-19-related data used in the included studies were derived from surveillance data (n = 44) and referred to notification data collected by public health institutions. One study used data from a seroepidemiological cohort study [28]. Another study used registry data on mortality [29].

This study shows that patterns of initially higher COVID-19 outcome rates in more affluent populations and subsequent crossover dynamics to higher COVID-19 outcome rates in the more deprived populations are internationally present in high-income countries. Moreover, several study results show that crossover dynamics frequently occur over time leading to opposing sinusoidal patterns. The results indicate that well-described inequalities in health appear to have persisted or even increased throughout the pandemic, predominantly to the disadvantage of more deprived populations. However, analyses to estimate the effect of socioeconomic factors on the overall burden of COVID-19 outcomes, such as the population attributable fraction, are still lacking but would be highly desirable for future research.

The included studies presented different possible explanations for the observed temporal patterns. Particularly for COVID-19 incidence and infection rates, the studies seem to confirm the important impact of mobility and the possibility of reducing mobility when advised by public health authorities. Doblhammer et al. discussed that mobility was higher in low infection periods among the more affluent populations than among less affluent populations [43]. This might have economic and occupational reasons: more affluent populations have a higher proportion of individuals working in jobs involving business travel, which may have introduced the virus early in the pandemic [11, 12, 42, 55, 62]. Higher incomes also provide the financial resources to travel during holidays, increasing the risk of introducing the virus into defined and socioeconomically more homogeneous local populations [62]. This would be in line with findings such as those from Germany, where no socioeconomic inequalities in infections were found in well-contained early hotspots [70]. Furthermore, this would be in line with the conceptualization of temporal dynamics in disease distribution proposed by Clouston et al. [71] using stages of disease theory. In the first stage, individuals have biologically the same risk of getting infected. In this stage, socioeconomic differences in disease distribution may occur at random. However, as Bambra conceptualized paths of socioeconomic inequalities in emerging infectious diseases, unequal exposures due to different sets of resources and conditions may contribute to socioeconomic inequalities in health and disease [72, 73]. In the case of the early COVID-19 pandemic, better-off populations may have been at higher risk due to their higher mobility.

However, when mobility restrictions were implemented, populations in lower-paid occupations, precarious jobs, and those living in higher-density housing or using public transport may have been less able to maintain physical distancing, while better-off populations tend to have more resources to follow infection protection restrictions, such as the ability to work remotely [13, 36, 42–44, 49, 60, 66, 74]. Furthermore, populations of lower education may have had insufficient information about the pandemic situation. Relevant information or preventive guidance may not have met the requirements of certain living environments of lower-educated or socioeconomically deprived populations, therefore leading to reduced compliance with recommended mitigation strategies [38, 44]. This would be in line with Bambra’s conceptualization of unequal transmission across different socioeconomic groups as it implies “inequality in the passing of a pathogen between community members” [72]. According to Clouston et al., socioeconomic inequalities arise during the second stage of disease [71]. Some populations will fare better with new prevention strategies and treatments, while others will face difficulties in prevention and treatment of emerging diseases due to strained resources or their living environment.

In addition, these inequalities might also be linked to social identity processes. Pre-existing social identities of individuals (such as living in deprived areas, working in essential or precarious jobs) and the belief of collectively being more affected due to the pandemic situation might lead to the emergence of new social group identities that might impact individuals’ behavior [75, 76]. For instance, in the US and Europe, right-wing populists instrumentalized public health measures such as mask-wearing to declare new social identities, promoting the social distinction between “us” and “them,” “encourage citizens to risk their health” [77] and reducing the compliance of certain communities with mitigation strategies. This might as well have influenced the time depending socioeconomic inequalities in the distribution of COVID-19, provided that these identities were associated with the individuals’ socioeconomic positions.

The majority of the included studies recommended public health measures that are relevant for future pandemic preparedness, such as improved low-threshold public health communication [39, 45, 51], coordination of community-based social services that meets the needs of local communities [51], promoting equity in testing capabilities [46, 52] and vaccine distribution [13, 32, 38, 45], providing personal protective equipment [66] or generally speaking, the implementation of specifically tailored interventions for socially disadvantaged populations by investing more resources [10, 11, 28, 30, 31, 35–39, 41, 44, 49, 50, 53, 54, 60, 61, 66]. Furthermore, some studies found, that the identification of vulnerable groups and data on specific subgroups were needed to enable targeted support [30, 31, 35, 40, 44, 48, 51, 60, 67].

Similar patterns of socioeconomic inequalities were observed for COVID-19 mortality as well across the included studies. A possible explanation for higher mortality rates in socioeconomically disadvantaged populations could be the well-described higher prevalence of chronic health conditions that are risk factors for severe and fatal COVID-19 in populations with a lower socioeconomic level. This leads to unequal susceptibility as described by Bambra [72]. Furthermore, populations with a lower socioeconomic level usually face greater barriers in accessing healthcare or even have geographical barriers, especially in rural regions [13, 32, 33, 35, 37, 49, 51]. This unequal treatment as conceptualized by Bambra might lead to a higher risk of facing a severe COVID-19 trajectory or even death for socioeconomically disadvantaged populations [72]. Low-threshold accessibility to diagnosis, treatment and healthcare should be provided to face this issue, as recommended by two of the included studies [30, 51].

This is the first systematically conducted review on the temporal dynamics of socioeconomic inequalities in COVID-19 outcomes. The systematic literature search contributes to a high comprehensiveness and assures reproducibility of the results, and the Embase and Scopus databases provide broad coverage of literature in the biomedical and social science fields. However, we cannot completely rule out that searches in additional databases could have yielded additional relevant records. We attempted to minimize this possibility by hand-searching relevant journals and the reference lists of all included studies in addition to the database searches.

Eligible studies were restricted to those conducted in high-income countries. Thus, the generalizability of the presented results to the contexts of low- and middle-income countries is limited. However, we also identified two studies from Brazil that ecologically analyzed social determinants of COVID-19 outcomes. Both studies concluded similar temporal patterns as most of the included studies, indicating that outcome rates initially were higher in more affluent populations but shifted to a higher burden in poorer populations over the course of the pandemic [78, 79].

Due to the pronounced heterogeneity of the included studies, no structured quality-of-evidence assessment could be conducted. This could potentially have led to biased results in the narrative synthesis as methodical limitations of the studies were not systematically assessed and considered [80]. This approach, however, seemed to be most appropriate given the aims of the review, i.e., to systematically generate a comprehensive overview of the international evidence on temporal dynamics in COVID-19 outcomes across different socioeconomic populations. Furthermore, the included studies were very heterogenous regarding socioeconomic indicators, observation periods and reported effect estimates. We therefore were not able to conduct any further quantitative analyses or systematical comparisons of effect estimates but instead summarized them narratively. Whereas this is often the case in scoping reviews as a result of the relatively broad research question, it limits our results to a qualitative interpretation. Further research is needed to quantitively describe the magnitude of the association between socioeconomic indicators and COVID-19 outcomes over time. To answer this research question, future reviews will have to focus on smaller sets of socioeconomic indicators and outcomes.

Publication bias is a potential issue because this review focused exclusively on peer-reviewed publications [81]. This bias could possibly have been minimized by including non-peer-reviewed literature and so-called grey literature. However, because no structured quality-of-evidence assessment could be conducted, eligibility was restricted to peer-reviewed articles to increase the validity of the included study results.

Our review may help to inform pandemic preparedness and emphasizes that social determinants of health, such as living and working conditions, provide relevant entry points for infection protection and control during epidemics with novel respiratory pathogens. As socioeconomic inequalities can be seen as unrealized opportunities to increase population health and health equity, targeted and timely prevention and intervention programs considering a population’s socioeconomic heterogeneity and the changing patterns of socioeconomic inequalities should be part of future pandemic preparedness plans. Considering the temporal dynamics may help to minimize the detrimental effects on specific groups during certain pandemic phases and should help to inform more specific containment measures. Public health interventions such as increased testing capabilities, vaccination programs, information campaigns, and healthcare access should meet the requirements of socioeconomically diverse populations—during the current COVID-19 pandemic and beyond—to strengthen the overall resilience of societies against new emerging infectious diseases and to tackle health inequities.