Vitamin D levels and status amongst asthmatic and non-asthmatic adolescents in Cyprus: a comparative cross-sectional study

Table 3 Differences in mean vitamin D levels and prevalence of inadequate vitamin D status between asthmatic and non-asthmatic participants

		Mean (SD)	Unadjusted Model ^†	Adjusted Model*
		Mean (SD)	b coefficient (95% CI)	b coefficient (95% CI)
Vitamin D levels	NWNA	22.90 (6.41)	Ref	Ref
Vitamin D levels	AA	21.15 (5.59)	−1.76 (−3.34 , −0.17); p = 0.03	−1.67 (−3.20, −0.14); p = 0.03
		Prevalence (%)	Unadjusted Model ^‡	Adjusted Model*
		Prevalence (%)	OR (95% CI)	OR (95% CI)
Severe vitamin D deficiency	NWNA	4.0%	Ref	Ref
(levels < 12 ng/mL)	AA	8.7%	2.32 (0.92, 5.84); p = 0.08	2.54 (0.97, 6.62); p = 0.06
Vitamin D deficiency	NWNA	34.7%	Ref	Ref
(levels <20 ng/mL)	AA	40.6%	1.26 (0.76, 2.10); p = 0.37	1.21 (0.72, 2.04); p = 0.47
Moderate vitamin D deficiency	NWNA	61.6%	Ref	Ref
(levels <25 ng/mL)	AA	75.4%	1.88 (1.05, 3.37); p = 0.03	1.91 (1.05, 3.49); p = 0.04
Vitamin D insufficiency	NWNA	84.0%	Ref	Ref
(levels < 30 ng/mL)	AA	94.2%	3.45 (1.06, 11.22); p = 0.04	3.23 (0.98, 10.65); p = 0.05
Odds of belonging to a lower category of vitamin D levels ^#	NWNA		Ref	Ref
	AA		1.60 (1.01, 2.51); p = 0.04	1.60 (1.01, 2.53); p = 0.04
p value for proportionality assumption in the ordinal model			p = 0.509	p = 0.656

Notes: ^†b coefficients corresponds to difference in mean vitamin D levels between Active Asthmatics (AA) and the Non asthmatic group (NWNA) as estimated in linear regression models before and after adjusting for potential confounders ^‡ Odds ratios of inadequate vitamin D status between AA and NWNA as estimated in logistic regression models before and after adjusting for potential confounders *Multiple linear and logistic regression models adjusted for gender, body fat percent, season of assessment, self-reported sun exposure in winter, skin type. ^#Estimated in ordinal logistic regression models along with p-value of test for parallel lines; a non-significant value provides evidence that the proportionality assumption of the ordinal model holds.

ISSN: 1471-2458