Causal Inference
Causal effects of plant-based diet indices on incident obstructive sleep apnoea: TMLE and AIPW analyses in UK Biobank Yohannes Adama Melaku* Yohannes Adama Melaku Melaku Melaku Melaku Melaku Melaku Flinders University
Background: Plant-based dietary patterns may reduce obstructive sleep apnoea (OSA) risk through pathways related to adiposity, inflammation, and cardiometabolic health, but causal evidence remains limited. We estimated the causal effects of healthy, overall, and unhealthy plant-based diet indices on incident OSA in UK Biobank.
Methods: We analysed 114,990 UK Biobank participants, with incident OSA identified via linked hospital admissions and primary care records (1,438 cases). Exposures were three plant-based diet indices—healthy plant-based diet (hPDI), overall plant-based diet (PDI), and unhealthy plant-based diet (uPDI)—evaluated using percentile-based contrasts (>25% vs ≤25%, >50% vs ≤50%, and >75% vs ≤75%). We estimated causal risk ratios (RRs) using targeted maximum likelihood estimation (TMLE) and augmented inverse probability weighting (AIPW), adjusting for baseline sociodemographic and behavioural confounders.
Results: Higher adherence to healthier plant-based diets was associated with lower incident OSA risk. For hPDI, high versus low adherence was associated with reduced risk at >25% (AIPW RR 0.77, 95% CI 0.68–0.86; TMLE RR 0.76, 0.68–0.86) and >50% (AIPW RR 0.88, 0.79–0.99; TMLE RR 0.78, 0.80–0.98), with attenuation at >75% (AIPW RR 0.88, 0.76–1.03; TMLE RR 0.88, 0.75–1.02). For PDI, high adherence was consistently protective at >25% (both methods RR 0.82, 0.73–0.92) and >50% (AIPW RR 0.86, 0.78–0.96; TMLE RR 0.86, 0.77–0.95), with weaker evidence at >75% (AIPW RR 0.89, 0.78–1.02; TMLE RR 0.88, 0.77–1.01). In contrast, uPDI showed no evidence of benefit and suggested a trend toward increased risk at higher thresholds (e.g., >75%: AIPW RR 1.11, 0.98–1.26; TMLE RR 1.13, 0.99–1.27).
Conclusions: Using two doubly robust causal estimators, greater adherence to healthy and overall plant-based diets was associated with lower incident OSA risk, whereas adherence to an unhealthy plant-based diet was not protective and may increase risk at higher levels. These findings highlight the importance of diet quality within plant-based eating patterns for OSA prevention.