Table 2 Inclusion and exclusion criteria

Study design

All cohort studies using multilevel data that consider income inequality and all-cause mortality or SRH We will also include all natural experiment studies that consider income inequality and all-cause mortality. Multilevel studies will be included if they report at least two levels and have conditioned upon individual income or another measure of individual socioeconomic position.

1. Individual-level studies (i.e. those that evaluate the relationship between individual income and individual mortality).

2. Studies that do not condition upon individual-level measure of income or socioeconomic position.

Population

We will include studies with an adult population. We will not limit studies based on relative income (e.g. restrict studies to only high-income countries) but will note differences in effect estimates for studies that do condition upon area-level income (see Table 4)

3. Majority (i.e. >/50%) of the study population is under eighteen years old.

Intervention/exposure

All measures of income inequality, including the Gini coefficient, the ratio of incomes between high-income individuals/earners to low-income individuals/earners, the Theil index or the share of income that is earned by high-income individuals [55]. We will not exclude based on measures of income inequality but will discuss the potential impact of different income inequality measures on statistical heterogeneity. We will also not exclude based on the quality of studies but will note differences according to our risk of bias assessment (see Table 4).

4. Studies that do not measure income inequality.

Comparator

All comparator types will be included.

None.

Outcome

All-cause mortality (including mortality rates and life expectancy) and SRH will be included as common indicators of health [56]. SRH is typically a response to a single-item question about general health which has been shown to be a reliable predictor of mortality [57]. Response categories are usually based on a Likert scale, with four or five categories most commonly used and the variable analysed as either binary or continuous [58].

5. Specific causes of mortality or specific health outcomes, such as mental or physical health only OR wellbeing or happiness (rather than SRH).

Setting

All area-types (e.g. municipalities, states, provinces, regions, or countries) will be included and differences due to type of area-level will be accounted for in the synthesis.

None

Year

We will search for and include all studies from 1992, based on the first Wilkinson study to suggest a relationship between income inequality and mortality based on ecological data [17], to present. However, we do not expect to identify multi-level studies published before 1996, as noted in the most recent meta-analysis of multi-level studies in income inequality and mortality and SRH [13].

6. Studies prior to 1992.

Follow-up

We will not limit studies based on follow-up time between income inequality intervention/exposure and outcome.

None.