Introduction

Vitamin D has been shown to benefit bone health and to reduce the burden of various diseases [1]. To mediate the health benefits, the Institute of Medicine and Health Canada recommend a vitamin D intake of 600 IU per day for adults and 800 IU per day for those above the age of 70 years [2,3]. These intakes are assumed to be sufficient to achieve an adequate serum 25-hydroxyvitamin D (25(OH)D) concentration, the established measure of vitamin D status. However, recent studies have suggested that the intake needed to achieve an adequate vitamin D status varies by weight status [46].

In Canada, diets contribute an estimated 232 IU of vitamin D per day [7] and, given its Northern latitude, cutaneous synthesis of vitamin D by sun exposure is limited [8]. Despite the Health Canada recommendations, vitamin D deficiency and insufficiency continue to be prevalent [9]. Specifically, the Canadian Health Measures Survey had shown that 4.1 and 10.6 % of Canadians aged 6–79 years had serum 25(OH)D concentrations of less than 27.5 and 37.5 nmol/L, respectively [9].

For older populations, several vitamin D supplementation studies have suggested, though not consistently, less functional limitation [10], prevention of falls [1113], reductions in fractures [1416], and benefits to mental health [17, 18]. Based on a meta-analysis, Bischoff-Ferrari et al. [12] had recommended vitamin D supplementation to achieve a reduction in falls. The above studies suggest a relationship between vitamin D and objectively measured health conditions. Health-related quality of life (HRQOL) aims to quantify the subjective experiences of the consequences of these health conditions. Huang et al. attributed improvements in quality of life along with improvements in pain and sleep to vitamin D supplementation in a case series of patients with chronic pain [19]. For a cross-sectional sample of older healthy volunteers, we recently showed that vitamin D status was positively associated with HRQOL [20]. In the latter study, we acknowledged the issue of reverse causality and recommended that intervention studies are needed to show whether vitamin D supplementation leads to improvements in HRQOL [20].

In the present study, we examine the effect of a preventive health program that encourages vitamin D supplementation on HRQOL among older residents of Canada. This seems particularly important to Canadians given the relatively high latitude and consequent reliance on vitamin D from diet and supplements.

Methods

This is a real-world evaluation of information gathered at baseline and follow-up visits from volunteer participants of a preventive health program by the Pure North S’Energy Foundation (PN). PN, a charitable, not-for-profit organization, provides lifestyle counseling and encourages vitamin D and multivitamin supplementation as described in more detail elsewhere [2123]. As of August of 2012, PN started the recruitment of older residents of the city of Calgary, Alberta, Canada. PN advertised their program for seniors through local newsletters and through the distribution of flyers in senior homes and community centers. PN organized weekly information meetings after which attendees could elect to sign up to enroll in the program. At their baseline and follow-up visits, participants completed a survey, had their body height and weight measured, and had their blood drawn for the assessment of serum 25-hydroxyvitamin D (25(OH)D), a measure of vitamin D status. Until April 2013, 25(OH)D was measured with the Liaison (chemiluminescent reaction) and, after that with a liquid chromatography, tandem mass spectrometry (LC/MS–MS) method. The correlation for 25(OH)D between methods was 0.801 (n = 3015, p < 0.001).

The survey included the five-level EQ-5D (EQ-5D-5L) to measure health-related quality of life (HRQOL) [24]. The EQ-5D consists of a five-dimensional descriptive system asking whether participants have (1) no problems; (2) slight problems; (3) moderate problems; (4) severe problems; or (5) extreme conditions or are unable to perform, or extreme conditions are fully constrained or restricted, with each of the following: (1) mobility; (2) self-care; (3) usual activities; (4) pain or discomfort; and (5) anxiety or depression [24]. HRQOL scores are based on responses to each of the five dimensions and were derived from the US value sets [25]. The EQ-5D is an established and validated instrument [24] with the major advantages of being short and easy to complete [26].

The survey further included questions on age, gender, perceived health, and income. Individuals and couples with an annual income of less than $25,000 and of less than $41,000, respectively [27], were considered as low income. Body mass index (BMI) was calculated on the basis of the measured heights and weights (weights in kilograms divided by the square of height in meters). Individuals with a BMI of 25 or more were considered overweight, and those with a BMI of 30 or more were considered obese [28].

Serum 25(OH)D is the established proxy for vitamin D status. Some have suggested that individuals with serum 25(OH)D concentrations below 50 nmol/L should be considered vitamin D deficiency, and those with serum 25(OH)D concentrations of 50 or more and less than 75 nmol/L be considered vitamin D insufficiency [29]. As recent studies have suggested extra-skeletal benefits, such as reduction in colorectal cancer and cardiovascular disease risk, for individuals with serum concentrations of 75 nmol/L or more [1, 30, 31], and where sufficient number of observations with serum concentrations above 75 nmol/L are observed, we further categorized serum concentrations into ≥75 and <100 nmol/L, ≥100 and <125 nmol/L, and ≥125 nmol/L. Where the above categorizations of vitamin D status provide some ease in interpretations, we did confirm our analysis while considering vitamin D status as a linear covariate.

As descriptive statistics we present the changes in the reporting of problems in the five EQ-5D dimensions and in the HRQOL score between baseline and 6 months of follow-up and between baseline and 1 year of follow-up (also referred to as temporal changes). The changes over 6 months of follow-up have the advantage of a larger sample size, and the changes over 1 year of follow-up have the advantage of longer intervention. We used a Chi-square test to test for the significance of temporal changes in the prevalence of problems and a one sample t test to test whether the mean temporal changes in HRQOL score were significantly different from zero. We applied univariate and multivariable regression to identify determinants of the temporal changes in HRQOL scores at 6 months and at 1 year. In addition to quantify the temporal changes at 6 months and at 1 year, we fitted an overall linear mixed-effect model for changes in HRQOL over time in the program using all observations regardless of when the observation was made. In all multivariable regression models, we adjusted for age, gender, body weight status, and baseline HRQOL. Furthermore, we examined the importance of vitamin D status for temporal changes in HRQOL scores by considering (1) the baseline serum 25(OH)D concentrations and (2) the difference between the follow-up and baseline serum 25(OH)D concentrations (also referred to as the temporal increase in serum 25(OH)D) as determinants of temporal changes in HRQOL scores.

Quality-adjusted life years (QALY’s) were estimated by the area under the HRQOL curve method [32].

PN anonymized their data prior to forwarding it to the University of Alberta for statistical analysis. The analysis was conducted using STATA version 12 (College Station, Texas). The Human Research Ethics Board of the University of Alberta approved the data access and analysis for this study.

Results

A total of 2119 participants had both baseline and follow-up assessments of HRQOL and serum 25(OH)D. Baseline characteristics of the 2119 participants are presented in Table 1. Of the 2119 participants, 1795 (84.7 %) and 761 (35.9 %) had follow-up assessments at 6 months (4–8 months) and 1 year (9–15 months), respectively.

Table 1 Baseline characteristics of 2119 participants with a baseline and follow-up assessments of HRQOL

Temporal changes in the prevalence of problems in the five EQ-5D dimensions and the HRQOL score are presented in Table 2. At the follow-up visits, participants reported fewer problems with mobility, usual activities, pain or discomfort, and depression or anxiety compared with what they reported at baseline (Table 2). For the 1795 participants with follow-up assessments at 6 months, the mean HRQOL score was 0.816 at baseline and 0.833 at 6-month follow-up, representing a temporal increase of 0.018 units in 6 months (Table 2). As a result, at 6 months of follow-up, fewer participants would retain in the baseline lower tertile and instead moved to the higher HRQOL tertiles (Table 2). For the 761 participants with follow-up assessments at 1 year, the mean HRQOL score was 0.808 at baseline and 0.832 at 1-year follow-up, representing a temporal increase of 0.025 units in 1 year (Table 2). For participants with follow-up at 1 year, the mean serum 25(OH)D concentration increased from 95 nmol/L at baseline to 129 nmol/L at 1 year of follow-up. At baseline, participants reportedly supplemented on average 2427 IU of vitamin D per day. At one year, this was 7510 IU per day.

Table 2 Changes in problems in EQ-5D dimensions, HRQOL score, serum 25(OH)D concentrations from baseline to 6 months and 1 year

Figure 1 and Table 3 show that temporal increases in HRQOL scores during follow-up were more pronounced among participants in the lowest tertile of baseline HRQOL score compared to those in the higher tertiles. Table 3 further shows that gender, body weight status, and changes in serum 25(OH)D were important and statistically significant determinants of improvement in HRQOL at 1 year of follow-up. The adjusted difference in mean temporal change was estimated to be 0.060 units higher among participants in the lowest relative to the highest tertile (Table 3). Those participants who realized increases in serum 25(OH)D of more than 50 nmol 25(OH)D/L had mean-adjusted temporal change that were 0.024 units higher as compared to those whose serum concentrations did not increase (Table 3). As per 25 nmol/L increase in 25(OH)D, HRQOL increased by 0.005 units, after adjusting for baseline HRQOL, baseline serum 25(OH)D, age, and gender (Table 3). Further, male participants experienced higher improvements in HRQOL compared to female participants and participants with obesity experienced lower improvements in HRQOL compared to normal weight participants. At 6 month of follow-up, changes in HRQOL were not statistically significantly different for weight status and increases in 25(OH)D but were different for baseline HRQOL tertile and gender (Table 3).

Fig. 1
figure 1

Temporal changes in HRQOL by baseline HRQOL tertile. Note: Bubbles represent the mean HRQOL level at baseline, at 6 month, and at 1 year of follow-up. The size of the bubbles is proportional to the number of assessments at each reporting time

Table 3 Determinants of temporal changes in HRQOL of program participants at 6 months (N = 1795) and 1 year (N = 761)

When considering all determinants simultaneously, the HRQOL score increased by 0.023 (95 % CI 0.018, 0.028) per year (Table 4). When further considering vitamin D supplementation doses, the estimates in Table 4 did not substantially change. In addition, vitamin D supplementation dose had a minimal and not statistical significant effect on HRQOL because of collinearity of vitamin D supplementation doses and 25(OH)D concentrations and because vitamin D status is on the causal pathway. The observed increase of 0.023 in HRQOL per year translates into 0.012 QALYs gained per person per year. When change in 25(OH)D was added to the model, the effect of time in the program reduced slightly from 0.023 to 0.020 units per year and each 25 nmol/L increase in serum 25(OH)D was associated with a 0.002 increase in HRQOL (Table 4). This translates into 0.001 QALYs gained per 25 nmol/L increase in serum 25(OH)D per person per year.

Table 4 Determinants of temporal changes in HRQOL among 2119 program participants

Discussion

This study revealed benefits of a real-world preventive health program that encourages vitamin D supplementation. Problems with mobility, usual activities, pain or discomfort, and depression or anxiety were all reported less at follow-up as compared to baseline. Overall HRQOL and QALYs gained during follow-up and paralleled improvements in vitamin D status.

The present study showed improvements in mobility and usual activities. This seems consistent with findings from observational studies and clinical trials that had reported benefits to mobility and functional status in terms of prevention of falls [12] and improvement in physical function [33]. A recent meta-analysis of randomized controlled trials concluded that vitamin D supplementation among unselected community-dwelling individuals may reduce the risk of skeletal or non-skeletal outcomes to up to 15 % [34]. In the present real-world evaluation, we observed reductions of 5.2 and 8.3 % in the reporting of problems with mobility and with usual activities, respectively. The very low prevalence of problems with self-care among program participants may be causing a ceiling effect whereby further improvements are difficult to achieve. Vitamin D has also been shown to be associated with less depression [17, 35] which seems consistent with our observation of a reduction in the reporting of problems with depression and anxiety. Of all dimensions, the largest reduction was reported for problems with “pain and discomfort” that dropped from 81.7 to 69 % in the first year. Bias by indication [36], whereby individuals with depression and anxiety and with pain and discomfort are more likely to participate and comply with program recommendations, may have contributed to a selective enrollment. At enrollment, we observed higher proportions of program participants reported problems with depression and anxiety (48 vs 32 %) and pain and discomfort (77 vs 70 %) compared to the general population of Alberta in the same age range [37].

As the present study is an evaluation of a real-world program rather than a randomized controlled trial with a strict protocol and blinded administration of the vitamin, it is not possible to disentangle the benefits of vitamin D supplementation from benefits of other program components such as promotion of healthy eating, active living, multivitamin use, and sun exposure. Therefore, and in light of the health promotion focus of this program evaluation, we had studied the importance of improving vitamin D status as a distinct achievement of the program. The regarding analyses revealed independent benefits to HRQOL arising from improvements in vitamin D status and from other program components.

The program effect translated into 0.012 QALYs gained per person per year, and the vitamin D status effect into 0.001 QALYs gained per 25 nmol/L increase in serum 25(OH)D per person per year. Assuming costs of CA$30 for an annual supply of vitamin D supplements, the cost-effectiveness ratio for someone who increases their vitamin D status with 50 nmol/L would be CA$15,000 per QALY ($30 divided by 2 times 0.001 QALY). The costs of the program in its pilot stage were estimated to range from CA $900 to 1200 per participant per year and for a sustained delivery of the scaled-up program CA $500 per participant per year. The cost-effectiveness ratio for the program is estimated to be CA$41,667 per QALY ($500/0.012 QALY). Both these ratios are considered cost-effective according to Rocchi et al. [38] who considered ratios below $50,000 per QALY as cost-effectiveness. In Canadian dollars in 2008, this is CA$114,000 per QALY. Canada and its provinces do not have an established cutoff threshold levels for cost per QALY and have a practice of considering other factors as well. A review of Canadian Common Drug Review concluded that new drugs were frequently accepted up to CA$80,000 per QALY though also noted that this threshold was not consistently applied [38]. Vitamin D supplementation is paid for by the individual, and the PN program is charitable. But even if vitamin D and the PN program were publicly funded, both ratios (program and vitamin D) would be cost-effective as the cost per QALY is in a range where most drugs are accepted. Moreover, it is important to note that HRQOL declines with age and the magnitude of this decline increases with age [39, 40]. Hazall et al. had estimated this natural (aging) decline in HRQOL to approximate 0.005 which translate in a reduction of 0.0025 QALY’s per person per year [39]. If we had contrasted the observed increases among program participants against the natural decline in HRQOL and QALY’s resulting from aging, the program and vitamin D benefits would have appeared even more favorable and more cost-effective.

The average serum 25(OH)D concentration of program participants was 93.4 nmol/L which is substantially higher than the average among Canadians aged 6–79 (67.7 nmol/L) and among Canadians 60–79 years of age (72.0 nmol/L) [9]. Self-selection of health-aware individuals interested to participate in the preventive health program may have contributed to this. As vitamin D status is positively related to HRQOL [20], and the present study revealed substantial larger program benefits for participants in the lower HRQOL tertile, one may expect larger benefits to similar interventions in populations that are better representative of Canadians. Where Health Canada recommendations apply to the general Canadian population, no earlier study reported on the benefits of improvements in vitamin D status for the quality of life in a general population. Though we conducted this in a community sample, caution is warranted when it comes from generalizing our findings as this sample may be self-selected toward a health-aware subpopulation and with a relatively high prevalence of problems with pain and discomfort [20]. Despite the large sample size, a second study limitation relates to the small number of participants that reported problems with self-care, which hampered some of the analyses. Further limitations relate to the limited duration of follow-up and the study design. The present study evaluated a real-world program. These results may therefore be particularly relevant for public health decision makers, but we acknowledged that stronger biological evidence comes from randomized controlled trials. We therefore recommend randomized controlled trials include pre- and post-intervention assessments of HRQOL to strengthen the biological evidence. A final limitation relates to the use of self-reported information which is prone to error, though the EQ-5D is an established and validated instrument [24] whereby the reporting error is expected to be limited given that it is short and easy to complete [26].