Relationship between Recovery from COVID-19-Induced Smell Loss and General and Oral Health Factors.

Background and Objectives : Loss of smell is one of the strongest predictors of coronavirus disease 2019 (COVID-19) and can persist long after other symptoms have resolved. "Long" cases (>28 days) of smell dysfunction present future challenges to medical and dental professionals, as there is a lack of evidence on the causes and any exacerbating or relieving factors. This study aimed to explore the persistence of COVID-19-induced smell loss and association with physical, lifestyle and oral health factors. Materials and Methods : This study was a cross-sectional survey of 235 participants. Recovery of smell was explored, comparing rapid recovery (≤28 days) with prolonged recovery (>28 days). Associative factors included age, sex, illness severity, diet, BMI, vitamin D supplementation, antidepressants, alcohol use, smoking, brushing frequency, flossing, missing teeth, appliances and number of dental restorations. Results : Smell loss showed 87% resolution within 30 days. Prolonged smell loss was significantly associated with older age (mean ± 95%, CI = 31.53 ± 1.36 years for rapid recovery vs. mean ± 95%, CI = 36.0 ± 3 years for prolonged recovery, p = 0.003) and increased self-reported illness severity (mean ± 95%, CI = 4.39 ± 0.27 for rapid recovery vs. 5.01 ± 0.54 for prolonged recovery, p = 0.016). Fisher's exact test revealed flossing was associated with rapid recovery, with flossers comprising 75% of the rapid-recovery group, compared to 56% in the prolonged-recovery group (odds ratio ± 95%, CI = 2.26 (1.23-4.15), p = 0.01). All other factors were not significantly associated ( p > 0.05). Conclusions : Increased age and illness severity were associated with prolonged smell recovery. Use of floss was the only modifiable factor associated with rapid recovery of smell loss. As 87% of cases resolve within 30 days, future studies may benefit from targeted recruitment of individuals experiencing prolonged sense loss. This would increase statistical confidence when declaring no association with the other factors assessed, avoiding type II errors.