Background Lower incomes are associated with poorer mental health (MH) and wellbeing, but the extent to which income has a causal effect (rather than reflecting reverse causation or confounding factors such as education) is debated. We synthesised evidence from studies measuring the impact of changes in individual or household incomes on MH and wellbeing for working-age adults (16–64 years).

Methods We searched MEDLINE, Embase, Web of Science, PsycINFO, ASSIA, EconLit and RePEc for randomised controlled trials (RCTs) and quantitative non-randomised studies (NRSs) – PROSPERO registration CRD42020168379. We included studies measuring effects of income change on any MH or wellbeing outcome. Screening and risk of bias (RoB) assessment were completed independently by two reviewers, using ROBINS-I for NRSs and RoB-2 for RCTs. As per Cochrane guidance, we conducted narrative synthesis based on direction of effects (benefit vs harm) for relevant datapoint(s) within each included study, and compared findings by RoB. Meta-analysis is in progress.

Results Of 16,521 hits screened, 17 RCTs and 118 NRSs (67.0% longitudinal) were included. Most studies were from high-income settings (71.9%), with 26.7% from USA. RoB was high: 102 studies (75.6%) were rated serious/critical, with confounding being the highest-rated RoB domain in 81.1% of these studies. Where known (53.7% of studies), the most common income sources studied were cash transfers (20.2%), natural disasters/welfare policy changes (7.5%) and lottery wins (5.2%). For mental health, 80.4% of 112 datapoints reported beneficial effects of income (95%CI 73.0–87.7%, sign test p<0.001). For wellbeing outcomes 86.5% of 74 datapoints reported a beneficial effect (95%CI 78.7–94.3%, p<0.001). However, effect sizes were small: for low/moderate RoB datapoints, binary exposures (i.e. increases/decreases in income of any amount) were associated with median 0.089 standard deviation (SD) improvement/worsening in outcome respectively (IQR 0.162, n=24); a 1SD increase/decrease in continuous exposures was associated with median 0.008 SD improvement/worsening (IQR 0.276, n=4). Studies with low/moderate RoB were slightly less likely to report beneficial effects compared with serious/critical RoB studies (77.6% vs 84.7%, p=0.257).

Conclusion Income increases are linked to improved MH and wellbeing, but on preliminary synthesis effect sizes appear small in the most robust studies. Income provision alone may not be adequate to improve mental health – given that more generous welfare policies are known to be linked with better population MH, it is possible other elements such as conditionality or financial security are also important.