Using aggregated ethnicity categories masks inequalities in smoking prevalence in England

BACKGROUND AND AIMS: Smoking prevalence in England is usually reported using aggregated ethnicity categories, which may obscure important differences. This study aimed to: [1] estimate smoking prevalence in England for six aggregated categories; [2] examine differences between the 18 detailed constituent groups within these categories; and [3] assess whether patterns varied by gender. DESIGN/SETTING: Data were collected between 2013 and 2025 in a series of monthly cross-sectional surveys of representative samples of the adult population in England. PARTICIPANTS: 229 979 adults aged 18 +. MEASUREMENTS: Weighted smoking prevalence was estimated for each of the 18 detailed and the six aggregated Office for National Statistics ethnicity categories. Logistic regression, adjusted for age, gender, socioeconomic status, region, survey year and survey mode examined the association between ethnicity categories and current smoking. Analyses were repeated including interactions for ethnicity by gender. FINDINGS: Smoking prevalence in England differed substantially by ethnicity. Aggregated estimates were: Asian 12.1%, Black 11.1%, Mixed or Multiple 23.9%, White 18.0%, Arab 21.8%, and Other ethnicities 18.4%; however, these aggregates obscured important variation between detailed constituent groups. Among people from Asian ethnic groups, smoking prevalence was higher for people who were Pakistani (13.5%; adjusted odds ratio [AOR] = 1.22 [95% confidence interval (CI) = 1.07-1.39]), Bangladeshi (15.9%; AOR = 1.44 [1.25-1.71]), Chinese (11.1%; AOR = 1.25 [0.99-1.57]), and Other Asian (12.6%; AOR = 1.29 [1.07-1.55]) adults compared with people who were Indian (9.3%; reference). Among people from Black ethnic groups, smoking prevalence was higher for Black Caribbean (17.2%; AOR = 3.34, [2.87-3.89]) and Other Black (17.2%; AOR = 2.92 [2.32-3.69]) adults compared with Black African adults (7.3%; reference). Among people from Mixed or Multiple ethnicity groups, smoking prevalence was lower for White and Black African (20.5%; AOR = 0.64 [0.50-0.81]), White and Asian (20.7%; AOR = 0.70 [0.57-0.87]), and Other Mixed or Multiple ethnicity (23.4%; AOR = 0.86 [0.70-1.06]) adults compared with White and Black Caribbean adults (29.1%; reference). Among people from White ethnic groups, smoking prevalence was higher for White Irish (19.9%; AOR = 1.32 [1.18-1.48]), White Gypsy/Traveller (39.1%; AOR = 1.99 [1.43-2.76]), and Other White (25.2%; AOR = 1.19 [1.13-1.25]) adults compared with White British adults (17.5%; reference). There was little difference in interpretation between fully adjusted models and those adjusted for just survey year and survey mode. CONCLUSIONS: Smoking rates differ greatly among ethnic groups in England that are often aggregated together in research and national statistics. Such aggregation hides important inequalities, including very high smoking prevalence in certain communities.