Cognitive decline is responsible for significant loss of independence and quality of life among older adults. It is often associated with dementia, a leading worldwide cause of morbidity and mortality (International, 2015). While more prevalent in older age, cognitive decline may commence earlier, and impede higher level functions such as effective participation in the workforce (Anstey et al., 2013). The societal and economic impacts of cognitive decline are increasing with population ageing, and prevention remains a key strategy. Vitamin D deficiency has been linked with poorer cognitive performance and cognitive decline in observational studies (Balion et al., 2012, Etgen et al., 2012, Goodwill and Szoeke, 2017), but results from supplementation trials have been mixed (Bischoff-Ferrari et al., 2020, Goodwill and Szoeke, 2017, Kang et al., 2021). The nature of the relationship between vitamin D and cognitive health appears complex but warrants further investigation.

Vitamin D becomes active in the body as a hormone after undergoing hydroxylation in the liver and kidneys. Levels of circulating vitamin D hormone (VDH) are tightly controlled by parathyroid hormone such that the precursor metabolite, 25-hydroxyvitamin D (25OHD), is most commonly used to measure vitamin D status (Deluca, 2014). While the role of VDH in calcium and bone metabolism is long established, there is ongoing debate concerning optimal 25OHD levels for skeletal health. The US Institute of Medicine suggests that levels of 50 nM/L are adequate for most of the population and there is likely little further skeletal benefit to achieving higher levels (Ross et al., 2011). However, the Endocrine Society recommends targeting higher levels of 75 nM/L (Holick et al., 2011). There is consensus that levels below 25-30 nM/L are deficient and can lead to rickets in children and osteomalacia in adults (Holick et al., 2011).

The mapping of VDH receptors throughout the body has triggered interest in functions beyond skeletal health. VDH is known to modulate general immune, inflammatory, and oxidative pathways, some of which are associated with cellular ageing (Landel, Annweiler, Millet, Morello, & Feron, 2016). In the brain, VDH receptors are expressed throughout the frontal, temporal and parietal cortices as well as the thalamus, hypothalamus, basal ganglia, hippocampus, amygdala, cerebellum, brainstem nuclei, and ventricular system (Eyles, Smith, Kinobe, Hewison, & McGrath, 2005), areas associated with most aspects of cognition. Biological research supports a neuroprotective role for VDH through numerous mechanisms such as maintaining intraneuronal calcium levels and regulating the release of neurotransmitters such as acetylcholine, dopamine, and serotonin (Garcion, Wion-Barbot, Montero-Menei, Berger, & Wion, 2002). Disruption to cholinergic transmission has been specifically implicated in Alzheimer’s disease (Cédric Annweiler et al., 2010). VDH also stimulates the production of nerve growth factor, neurotrophin 3, and glial cell line-derived neurotrophic growth factor (GDNF), which are required for neuronal growth and differentiation (Cédric Annweiler et al., 2010). In rat models, the upregulation of GDNF by VDH has been correlated with reduced infarct size after cerebral artery ligation (Garcion et al., 2002). In vitro, VDH has been demonstrated to promote macrophage clearance of amyloid beta material, a hallmark of Alzheimer’s disease (AD) (Masoumi et al., 2009). Furthermore, supplementation in animal models has been shown to improve cognitive function, reduce AD biomarkers (Landel et al., 2016) and to protect neurones from excitotoxic insults in a nonlinear, dose-dependent manner (Brewer et al., 2001).

In observational research on adults, low levels of 25OHD are associated with poorer global cognition: A recent meta-analysis that pooled outcomes from 26 observational studies, summarised the risk of poorer cognitive performance as 1.24 (95% CL:1.14, 1.34) times higher for those with low, compared to high, levels of 25OHD (Goodwill & Szoeke, 2017). While the cut-off levels used to categorise 25OHD varied across individual studies, the lowest category fell below 50 nM/L in all but one study (Schneider et al., 2014) while the highest category was above 50 nM/L in all. The relationship between low 25OHD levels and poorer performance is most often noted in relation to global cognition but has also been reported for performance in specific cognitive domains, particularly executive function (Cedric Annweiler et al., 2013). The relationship also appears stronger in cross-sectional than in longitudinal analyses (Goodwill & Szoeke, 2017), suggesting reverse causation may be a factor. Additionally, while the relationship between low levels and poorer cognition is consistent across studies, the relationship at high levels is not. Some suggest cognitive performance improves throughout the range (Ahn & Kang, 2015; C. Annweiler et al., 2016, Chei et al., 2014, Llewellyn et al., 2011, Sakuma et al., 2019), while others suggest there is little improvement above mid-range (Bartali et al., 2014, Breitling et al., 2012, van Schoor et al., 2016). Poorer outcomes have also been reported at high levels (Granic et al., 2015, Maddock et al., 2015).

In most observational studies, circulating 25OHD levels are categorised or treated as a continuous variable, with few attempting to fully characterise the dose-response association. The use of polynomial terms or splines in some studies has suggested nonlinear relationships (Bartali et al., 2014, Breitling et al., 2012; D. M. Lee et al., 2009, Pettersen, 2016, van Schoor et al., 2016), although inflection or threshold levels have varied from 35 nM/L (D. M. Lee et al., 2009) to 120 nM/L (Pettersen, 2016). A nonlinear relationship between vitamin D and cognitive performance is feasible and would help explain the lack of findings in supplementation trials of healthy adults (Bischoff-Ferrari et al., 2020, Kang et al., 2021, Rossom et al., 2012), as those with adequate baseline levels may not derive any benefit.

Over the last decade, dose-response meta-analyses have highlighted nonlinear associations between vitamin D status and other health outcomes such as diabetes (Hou et al., 2021), heart disease and stroke (Sofianopoulou et al., 2021), and mortality (Durup et al., 2012, Sofianopoulou et al., 2021). The relationship between vitamin D status and dementia has also been examined, with the risk of incident dementia reported to decrease as levels of 25OHD increase up to at least 70 nM/L (Jayedi, Rashidy-Pour, & Shab-Bidar, 2019). However, to our knowledge, no meta-analysis has examined the dose-response association between 25OHD levels and cognitive performance. We have conducted a systematic review and dose-response meta-analysis to examine this relationship in community-dwelling adults.