The finding that half of the residents had B6 deficiency defined as p-PLP < 20 nmol/L was discouraging. The cut-off level was in accordance with the literature, and the local laboratory has shown that the 2.5 percentile in a healthy adult population with their method was 23 nmol/L with a decline with age (personal communication)
]. High prevalence rates of B6 deficiency (51-75%) has also been reported in other studies in elderly people in institutions and hospitals
]. On the other hand, Gonzalez-Gross et al. reported p-PLP < 20 nmol/L in only one out of 218 Spanish institutionalised elderly people
]. B6 deficiency has been shown to be common also in the adult population with prevalence rates of 11-24%
]. In all, B6 deficiency seems to be a common problem in elderly people in nursing homes and occurs also in subjects without malnutrition. No symptoms and signs are specific for B6 deficiency. In this study, as in other reports, malnutrition, inactivity, old age, low s-albumin and s-ALAT values, and high homocysteine values were associated with B6 deficiency and seem reasonable to have in mind
Mean dietary intake of B6 was in large in accordance with Norwegian recommendations, but the prevalence of B6 deficiency was nevertheless high. Also in subjects with B6 deficiency the intake was only marginally below the recommendations. This supports the assumption that the need of B6 is increased in elderly people, and indicates that the recommendations need adjustment for age
]. Vitamin supplements were not included in the calculation of dietary intake of B6 and could explain the lack of significant associations between dietary intake of B6 and p-PLP. An important finding was that vitamin supplements efficiently protected against B6 deficiency. None of the users of vitamin supplements had B6 deficiency. Recommendation of vitamin supplement to all elderly people in institutions is an easy and safe prophylaxis.
The nutritional state was unsatisfactory, malnutrition was present in 11.5% and only 27% were at no risk of malnutrition. Even more unsatisfactory findings have been reported in other studies in nursing homes with malnutrition prevalence as high as 71%
]. Malnutrition should be unnecessary and has been associated with high costs
]. Poor nutritional status, old age, inactivity and low s-albumin were all related to the overall poor health of subjects with B6 deficiency.
The study excluded a strong and clinically significant association between B6 and FGID in this group of elderly people. One explanation of the discrepancy between this study and our findings in a previous study in younger patients with IBS is that the pathogenesis and pathophysiology of FGID in elderly people differs from that in younger subjects with IBS
]. An exact diagnosis of IBS is based on precise information about the symptom’s duration and relation to defecation, which was impossible to obtain in this group. The diagnosis of IBS was therefore omitted in this study and replaced by the unspecific diagnosis of FGID. P2X receptor antagonists like PLP attenuate gut sensitivity after infections
]. This might be a cause of symptoms in subjects with IBS, but perhaps not in elderly people with FGID. Nor were associations with other common disorders such as dementia, cognitive impairment, psychiatric disorders and cardiovascular diseases shown. The discrepancy between our findings and the reported associations between B6 deficiency and somatic and psychiatric diseases in other studies could be due to a high degree of morbidity in this study in elderly people or a type II error. The association between B6 and hypertension was judged as clinically insignificant and might have occurred by chance. The lack of associations between B6 deficiency and diseases/disorders in this study is no excuse for not preventing B6 deficiency in elderly people. For unknown reasons, subjects with normal/ high p-PLP used more sedatives/ hypnotics, SSRI, and anti-dementia drugs.
Strengths and limitations
The population characteristics, such as gender, age, BMI and nutritional status, and the prevalence of dementia, cardiovascular diseases etc. were, with some variation, comparable with most other studies in nursing homes in Norway and other countries
]. Compared with our main study, participants in this study might have had somewhat better nutritional status and been less dependent on support in ADL. Data from non-participants in the main study were not available, but a biased selection seems unlikely. Exceptionally poor health status did not explain the high prevalence of B6 deficiency.
Data quality might have been variable. Demographics and background variables were classified according to a national classification system, drugs according to ATC classification, and nutrition and activity of daily living with validated questionnaires (MNA and Katz). These data probably have high quality. The nursing staff collected data from the participants, their medical records and their next of kin. The high proportion of participants with cognitive impairment and next of kin’s often insufficient knowledge might have reduced the data quality. The nursing staff’s knowledge of the residents was, however, good, since all participants had stayed in the nursing home for more than eight weeks. Because resources in the nursing homes are limited, exact registration of intake of food was difficult and might have been inadequate and explains in part the poor correlation between intake of B6 and p-PLP.
When planning the study, a clinically significant correlation between p-PLP and GI symptoms was considered to be below minus 0.3. Although the number of subjects in the study was smaller than planned, the study excluded a clinically significant association between B6 deficiency and GI symptoms. The lower limit of any of the 95% CI of the correlations between p-PLP and GI symptoms was minus 0.27.
Multivariable analyses were performed but turned out to be unreliable. The limited number of subjects in the study and the high number of variables made a reduction in the number of variables necessary, and the results varied depending on the method used for reduction of the variables. Multivariable analyses were therefore omitted. The low number of subjects might have influenced the conclusions and the generalisability of the results.