Psychological morbidity and quality of life of ethnic minority patients with cancer: a systematic review and meta-analysis

Abstract

Background: Ethnic minority is associated with higher cancer incidence and poorer survival than is being in the majority group. We did a systematic review and meta-analysis to assess whether psychological morbidity and health-related quality of life (HRQoL) were affected by minority status.

Methods: We searched Medline, AMED, PsycINFO, Embase, CENTRAL, CINAHL, PubMed, Sociological Abstracts, and Web of Science for English-language articles published between Jan 1, 1995, and October, 2009. Articles were eligible if they reported original data on anxiety, depression, distress (for psychological morbidity), or HRQoL in minority and majority cancer patients or survivors. Minority status was defined as being an immigrant or having an ethnic, linguistic, or religious background different to the majority of the population in the country where the research was done. We excluded African Americans and indigenous groups. Eligible articles were rated for quality of reporting, external validity, internal validity, sample size, and power. Each quality criterion was rated independently by two reviewers until inter-rater reliability was achieved. In a meta-analysis we compared mean scores adjusted for socioeconomic status and other sociodemographic and clinical variables, where available. Effect sizes greater than 0·5 and 95% CI that included 0·5 or -0·5 were deemed clinically important, with negative values indicating worse outcomes in minority patients. We assessed publication bias by estimating the number of potential unpublished studies and the number of non-signficant studies with p=0·05 required to produce a non-significant overall result.

Findings: We identified 21 eligible articles that included 18 datasets collected in the USA and one in each of Canada, Romania, and the UK. Ethnic minority groups were Hispanic, Asian or Pacific Islander, or Hungarian (one dataset). Overall, we found minority versus majority groups to have significantly worse distress (mean difference -0·37, 95% CI -0·46 to -0·28; p<0·0001), depression (-0·23, -0·36 to -0·11; p=0·0003), and overall HRQoL (-0·33, -0·58 to -0·07; p=0·013). Further analyses found disparities to be specific to Hispanic patients in the USA, in whom poorer outcomes were consistent with potentially clinically important differences for distress (effect size -0·37, 95% CI -0·54 to -0·20; p<0·0001), social HRQoL (-0·45, -0·87 to -0·03; p=0·035), and overall HRQoL (-0·49, -0·78 to -0.20; p=0·0008). Results were significantly heterogeneous for overall HRQoL and all domains. Tests for interaction, for adjusted versus unadjusted and comparisons of high-quality, medium-quality, and low-quality articles, were generally non-significant, which suggests no bias. We found no evidence of any substantive publication bias.

Interpretation: Hispanic cancer patients in the USA, but not other ethnic minority groups, report significantly worse distress, depression, social HRQoL, and overall HRQoL than do majority patients, of which all but depression might be clinically important. Heterogeneous results might, however, have limited the interpretation. Data for other minority groups and for anxiety are scarce. More studies are needed from outside the USA. Future reports should more clearly describe their minority group samples and analyses should control for clinical and sociodemographic variables known to predict outcomes. Understanding of why outcomes are poor in US Hispanic patients is needed to inform the targeting of interventions.

Funding: Prince of Wales Hospital, Sydney, Australia.