  |
|
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| Journal of the New Zealand Medical Association,
21-September-2007, Vol 120 No 1262
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Seasonal variation in vitamin D levels in the
Canterbury, New Zealand population in relation to available UV radiation
John Livesey, Peter Elder, M Jane Ellis, Richard McKenzie,
Ben Liley, Chris Florkowski
It is recommended by some authorities that the plasma
concentration of 25-hydroxyvitamin D3 [25(OH)D]
should be at least 75 nmol/L.1–3 This
figure is based both on observational evidence relating 25(OH)D levels to the
risks of fracture, periodontal disease, colorectal cancer, and lower-extremity
muscle weakness, and on the 25(OH)D levels found in those, such as farmers and
lifeguards, with sun exposure typical of conditions in which modern skin tones
evolved.
Further, a randomised controlled 4-year study of
cholecalciferol plus calcium supplementation in postmenopausal women, which
raised plasma 25(OH)D from 72 to 96 nmol/L, reduced by 77% the likelihood of
being diagnosed with cancer between 1 and 4 years after the initiation of the
trial.4
In contrast, surveys, especially of older people, even in
apparently sunny countries such as Spain, Italy and
Greece,5
Brazil,6 and
Australia,7 show high proportions of people
with 25(OH)D levels less than 75 nmol/L, particularly during winter months.
Within New Zealand, a survey of the Auckland workforce found
that a large proportion of the workers, if not the majority, had serum 25(OH)D
concentrations below 75 nmol/L,8 as did
Auckland elderly,9 pregnant women in
Wellington,10 Dunedin
elderly,11 and New Zealand children throughout
the country.12
In a survey of New Zealanders aged 15 years and older, 3%
were considered to have frank deficiency (<18 nmol/L) and 48% insufficiency,
based on a cut-off of 50 nmol/L,13 and with
differences apparent due to age, gender, latitude, and season.
Only a few studies at a limited number of latitudes
(23oS,6
37oS,9 and
68oN14) have
measured the intensity of ultraviolet radiation (UV) at ground level at the same
time as the 25(OH)D measurements were made. The studies at
23oS and 37oS
were limited to older subjects (>40 years and >65 years respectively) and
the study at 68oN had only 15 participants and
covered a time period of only 60 days.
We now report the relationship between plasma 25(OH)D levels
and solar UV in the general adult population in a southerly New Zealand
location. We also develop a model of vitamin D metabolism to assist in the
effective remediation of poor vitamin D status.
MethodsSubjectsVolunteer group—The subjects
were residents of Christchurch, New Zealand
(44oS), who volunteered to participate in a
study to establish reference intervals for endocrine and metabolic test methods.
The study was approved by the Upper South B Regional Ethics Committee.
Recruitment was performed by contacting individuals selected randomly from
Christchurch electoral rolls (241 responses, a 14% response rate) or by
advertising (76 responses).
Volunteers completed a health questionnaire and were
accepted if aged 18 or over, considered themselves healthy, and did not meet
exclusion criteria that included diabetes and endocrine conditions, relevant
cancers, steroid medication and recent hospitalisation. Of the initial
volunteers, 33% were excluded either to meet study criteria or to avoid a more
severe gender imbalance.
A single morning blood sample was collected within the
period of February to August 2004 from 209 individuals of whom 105 were fasting.
After further excluding those who were taking vitamin D supplements or cod-liver
oil, 25(OH)D measurements were available for 201 volunteers (119 females, 82
males).
The mean (±SD) age was 46±14 years with a
median of 45 years (range 18 to 83) and the mean (±SD) body mass index was
26.3±4.7.
Patient group—This group
consisted of patients within the Christchurch region from whom samples were
submitted to Canterbury Health Laboratories for measurement of plasma 25(OH)D
between 1 July 2003 and 31 December 2004.
3702 samples were from females and 1138 from males,
with a mean (±SD) age of the whole group of 59±23 years and a median
age of 63 years (range 0.1 to 101). Samples were submitted from hospital wards,
outpatient clinics and private practices. It is not known how many patients were
taking vitamin D supplements.
Biochemical analyses25(OH)D was measured using the DiaSorin
radioimmunoassay kit (Stillwater, MN, USA). The antibody cross-reacts with
vitamin D2 and D3
equally and results are the mean of duplicate determinations with internal and
external QC samples in each batch. The low, medium, and high QC values with
coefficients of variation are 16.5 (16.2%), 35.9 (7.8%), and 132 (7.8%) nmol/L
respectively. The laboratory is a participant in the DEQUAS (Charing Cross
Hospital, London, UK) vitamin D external quality control programme.
Plasma calcium was measured on the Abbott Aeroset
analyser (Abbott Laboratories, Abbott park, IL, USA) by colorimetry using the
Arsenazo-III dye method with correction for albumin (+ 0.02*[40 –
albumin(g/L)]). The intra-assay CV was 0.8% at 3 mmol/L.
Serum ionised calcium was measured on the Corning C865
blood gas analyser (Ciba Corning Diagnostic; Medfield, MA, USA) by calcium
ion-selective electrode (between-batch CV 1% at 1.22 mmol/L). Assay stability
was assured, as a matter of routine, by collation of daily patient means for all
major analytes on the Abbott Aeroset analyser and by collation of monthly means
and SDs for all internal QC samples. No significant assay drift was evident for
either calcium or albumin over the time period of the study.
Parathyroid hormone (PTH) and C-telopeptide (CTX) were
measured using the Roche Elecsys 2010 system. The low, medium, and high QC
values with coefficients of variation are 2.2 (6.9%), 8.2 (5.3%), and 31.5
(4.7%) pmol/L for PTH—and 0.46 (7.5%), 0.59 (10.4%), and 1.64 (5.2%)
µg/L for CTX.
Bone-specific alkaline phosphatase (BALP) was measured
using the Beckman ACCESS system and the low and high QC values with coefficients
of variation are 10.9 (7.5%) and 64.2 (6.8%) µg/L.
UV radiationDaily UV irradiances
(W/m2) at 1-hour intervals were taken from the
National Institute of Water and Atmospheric Research (NIWA) UV Atlas software
package (http://www.niwascience.co.nz/services/uvozone/atlas).
The irradiances for Christchurch were summed to give
the mean for each month of total erythemally-weighted UV (eUV) and vitamin
D-weighted UV (dUV) per day. For erythema, the weighting is according to
McKinlay and Diffey.15 For vitamin D, the
weighting is from Maclauglin Anderson and
Holick,16 normalised to unity at 315 nm and
truncated at 315 nm as suggested by MF Holick (personal communication,
2006).
The data product is derived from a combination of
short-wave pyranometer data to estimate cloud effects, satellite-derived
estimates of ozone ,and a radiative transfer
model.17 The pyranometer data are from LICOR
LI-200 sensors. The accuracy is approximately ±10%.
The UV data in the current study is the environmentally
available UV. This study did not attempt to measure or estimate personal UV
exposure, which depends on an individual’s lifestyle, and may be only 3%
or less of that available.18
Statistical analysisThe NCSS statistical package (Kaysville, Utah, USA) was
used for analyses. Confidence intervals (CI) for estimates are 95% intervals.
Correlations were calculated by the non-parametric Spearman procedure and a
significance level of 0.01 was used in view of the number of correlations
examined.
ModellingWe used the three-compartment model of vitamin D
metabolism shown in Figure 1 to represent the production and losses of vitamin
D3 and 25(OH)D in humans.
Compartments 1 and 2 are, respectively, faster and
slower turnover vitamin D3 compartments while
compartment 3 contains 25(OH)D. Parameter
v3, the volume of the 25(OH)D
compartment, was taken to be 8 litres,19 and
the half-life of 25(OH)D was taken to be about 10
days,20 giving a value for
k34 of 0.5. The other parameters in the
model are: v1,
v2, the volumes of compartments 1 and 2;
rf, the dietary intake of vitamin D;
kuv, the rate constant for cutaneous
vitamin D production by solar dUV; rate constants
k12 and
k21 representing the interchange of
vitamin D between compartments 1 and 2;
k23, the rate of conversion of vitamin D to
25(OH)D; k24, the rate of metabolism of
vitamin D by other pathways; and k34,
the rate of conversion of 25(OH)D to other metabolites. The parameters
α and β define the non-linear feedback control of
k23 by the concentration of
25(OH)D.
Figure 1. Model of vitamin D
metabolism
 Compartments 1 and 2 contain vitamin
D3 and compartment 3 contains 25(OH)D; 4 denotes
further (undefined) metabolites. Rates (nmol/day) are denoted r and
rate constants k. Parameters α and β define the feedback of
25(OH)D on its production rate.
rf=dietary vitamin
D3,
ruv=kuvEuv,
Euv=dUV energy,
r12=k12c1,
r21=k21c2,
r24=k24c2,
r23=k23c2=c2α/(1+βc3),
r34=k34c3.
To obtain values for the parameters in the model, the
best-fit values for v1,
v2,
k12,
k21, α, β,
k24, and
k34 were estimated from the data taken
from the literature given in Table 1 using the downhill simplex optimisation
procedure.21 Then, to estimate the quantitative
relationship between UVB radiation in Christchurch and the plasma 25(OH)D levels
in the volunteer group, best-fit estimates of
rf and
kuv were obtained by fitting the model
to the mean monthly 25(OH)D values in the volunteer group and the daily dUV
energy intensity measurements during 2003 and 2004. The goodness-of-fit
criterion was the minimum absolute deviation.
Table 1. Data from the literature used to
estimate the parameters of the model in Figure 1
Values predicted by best-fit model parameter values
also shown.
ResultsThe results for plasma 25(OH)D are summarised in the upper
two panels of Figure 2. Plasma 25(OH)D tended to rise as UVB energy rose in
spring and to fall as UVB energy fell in autumn. For the individual values in
the volunteer group, the amplitude of a sine function fitted to 25(OH)D was 17.3
nmol/L (CI=12.1 to 22.6, n=201), and for the monthly mean 25(OH)D in the patient
group was 8.4 nmol/L (CI=4.5–12.2, n=17).
The 25(OH)D levels tended to lag behind UVB and for both
subject groups the Spearman correlation between dUV and 25(OH)D was at a maximum
when the lag was 2 months, being 0.89 (n=6, p=0.02) for the volunteer group and
0.79 (n=16, p<0.001) for the patient group.
There was no significant difference in the monthly mean
25(OH)D levels between females and males in the volunteer group (mean
difference=0.5 nmol/L, CI =-4.4–5.3 nmol/L, n=6)—but in the patient
group, monthly mean 25(OH)D levels were higher for females (mean difference=2.9
nmol/L, CI=0.5–5.3 nmol/L, n=18).
Table 2 gives the proportions of subjects in the volunteer
group with 25(OH)D levels below commonly
used9,10 cut-off levels. The large majority had
below optimal levels (<75 nmol/L) regardless of the time of year and the
majority showed insufficiency (<50 nmol/L) in June, July, and August.
Deficiency (<25 nmol/L) was evident in at least a few
individuals in each month studied and rose to 35% of the volunteer group in
July-August. Only 1.5% of the volunteers had 25(OH)D below 12.5nmol/L (1 person
in May and 2 in July).
Table 2. Monthly proportions of the volunteer
subject group with plasma 25(OH)D concentrations below specified
levels
The correlations between the time of year and the measured
analytes, and between the analytes, are given in Table 3 for the volunteer
group. As the year progressed 25(OH)D levels fell (p<0.001), total calcium
(Catot) rose (p<0.001), and ionised calcium
(Ca++) fell (p<0.01). PTH levels were
neither significantly correlated with time of year nor with 25(OH)D levels but
for the non-linear regression function, PTH=a + b/25(OH)D, the
estimates of a and b (with CI) were 2.83 (2.51–3.15) and
24.4 (12.7–36.1).
The mean±SEM February (n=40) and July/August (n=25)
values for total calcium were 2.20±0.02 and 2.28±0.02 mmol/L
respectively. The corresponding values for ionised calcium were 1.19±0.01
mmol/L and 1.14±0.01 mmol/L.
The monthly ratios of mean daily dUV and eUV energies in
Christchurch are shown in Figure 3.
Figure 2. Monthly plasma 25-hydroxy vitamin D
concentrations in male and female Christchurch residents (mean±SEM) for (a)
the volunteer group and (b) the patient group; (c) represents the
corresponding monthly means of the UVB energy received daily at ground level and
the maximum daily temperature. The first month is July 2003 and the last month
is December 2004.
 In (a) and (b), ▲——▲
male, ▼——▼ female. The smooth dashed line in
(a) is the prediction of the best-fit model (dietary vitamin D 350 IU/d), and
the dotted line shows the best fit if dietary vitamin D is assumed to be 200
IU/d. In (c), ●——● dUV, ○ - - - ○ eUV,
■ ∙ ∙ ∙ ■ maximum temperature. The one volunteer
studied in August (with a 25(OH)D value of 22 nmol/L) is not included.
Table 3. Correlations within the volunteer
group
* p<0.01, **p<0.001; PTH=Parathyroid hormone;
BALP=bone-specific alkaline phosphatase;
CTX=C-telopeptide. Figure 3. Ratio of mean daily vitamin
D-weighted UV energy (dUV) to mean daily erythemally-weighted UV energy (eUV) in
Christchurch monthly from July 2003 to December 2004
 The best-fit values of the parameters for the model of
vitamin D metabolism are given in Table 4. The estimated dietary intake of
vitamin D (rf) is 23 nmoles/day (9
µg/day or 350 IU/day) and kuv is
estimated to be 8
nmol∙m2∙kJ-1.
The predicted values for the literature-derived data using the best-fit
parameters are given in Table 1.
The plasma 25(OH)D concentrations predicted for the
volunteer group throughout 2004 are shown in Figure 2a, both for the best-fit
dietary vitamin D intake of 350 IU/d, and for an assumed dietary intake of 200
IU/d, in which case the best-fit value of
kuv is 10
nmol∙m2∙kJ-1.
Table 4. Best-fit estimates of the parameters
of the model of vitamin D metabolism
†Values obtained directly from literature.
The quantity of supplemental vitamin D needed to raise the
modelled annual minimum mean plasma 25(OH)D in the volunteer group to 75 nmol/L
is predicted to be 1450 IU/d (36µg/d), or to raise the annual minimum mean
to 100 nmol/L, 2600 IU/d (64 µg/d). In the latter case, the annual maximum
mean plasma 25(OH)D is predicted to be 114 nmol/L. For both simulated supplement
doses the annual maximum mean plasma 25(OH)D occurred in mid February and the
annual minimum mean in early September.
In the absence of sunlight, the corresponding
supplementation required is predicted to be 1600 IU/d (41 µg/d) and 2700
IU/d (68 µg/d). On the other hand, if year-round sunlight exposure were to
be doubled, in the absence of supplementation, the annual maximum plasma 25(OH)D
is predicted to rise from 56 nmol/L to 80 nmol/L and the annual minimum to rise
from 29 nmol/L to 37 nmol/L.
DiscussionThe two principal findings of this study are firstly that
most, if not all, of the apparently healthy general population in Christchurch
do have not adequate circulating levels of 25(OH)D at some time during the year,
and secondly that relatively high levels of supplementation with vitamin D would
be required to achieve healthy concentrations of 25(OH)D year round.
Considering first the high prevalence of vitamin D
deficiency, Table 2 shows that even in summer, in February, only 12% of the
volunteer group had optimal (>75 nmol/L) plasma 25(OH)D levels. This
percentage fell during the autumn and winter until in June, July, and August, no
one achieved this level.
Of even greater concern is that in every month studied, at
least some of these self-declared “healthy” people were vitamin D
deficient (<25 nmol/L). By July-August this proportion reached 35%.
The situation appears to be little better in the patient
group where the average monthly levels of plasma 25(OH)D ranged from 40–62
nmol/L. This is despite it being likely that many of this group were on vitamin
D supplementation and were having their 25(OH)D measured to check the dose level
(Simon Wynn-Thomas, personal communication, 2007). Evidently few of this group
would have maintained a level of 25(OH)D of at least 75 nmol/L year round.
These findings raise the question of what could be done
about the near universal state of vitamin D deficiency in Christchurch,
particularly in winter. Supplementation with vitamin
D3 is probably the most practical possible remedy
on account of its ease of administration,
safety,25 and cost effectiveness. The novel
modelling done in this paper is designed to give a guide to the necessary dosage
levels and is the first description of a model of vitamin D metabolism that
includes solar radiation.
The model suggests that taking 1450 IU of vitamin
D3 per day, in addition to normal sun exposure
and dietary intake, would maintain the average 25(OH)D levels in healthy
Christchurch people at 75 nmol/L or above all year. However many would still
fall below 75 nmol/L at some time, so an annual minimum average of 100 nmol/L is
probably more desirable and this is predicted to require a supplement of 2600 IU
per day.
In New Zealand, prescription vitamin
D3 is only available as 50000 IU (1.25 mg)
tablets (Cal.D.Forte®), hence 2600 IU/d is
approximately equivalent to 1 tablet every 19 days, or approximately 2 per
month. This is considerably greater than the dose rate recommended in the
Medsafe datasheet (http://www.medsafe.govt.nz/Profs/Datasheet/c/CalDFortetab.htm)
of 1600 IU/d (1 tablet per month), but well below the 10000 IU per day which has
recently been suggested as no-observed-adverse-effect
level.25 However safety data from large studies
and beyond 5 years is lacking.
Our finding that supplemental vitamin
D3 of about 1450 IU/d would be required to raise
minimum average levels of 25(OH)D to 75 nmol/L is comparable to the 1700 IU/d
that Vieth and coworkers1 estimate to be
required to raise levels from 50 nmol/L to 80 nmol/L, and so suggests our model
is plausible.
Another strategy for raising 25(OH)D levels might be greater
personal exposure to solar dUV, although this is not without risk as excessive
UVB exposure can result in skin and eye damage.
Could the risk be minimised by choosing conditions where the
ratio of vitamin D production to erythema is maximal? As can be seen from Figure
3, the vitamin D produced from a given erythemal exposure is greater in summer
than at other times of year. This is because the sun is higher in the sky and
hence the shorter wavelength dUV is less attenuated by the atmosphere compared
to the attenuation of the longer wavelength eUV. Similarly, sun exposure at
midday is preferable to exposure earlier or later in the day if the aim is to
maximise vitamin D production while minimising burning.
However, we would not recommend a vitamin maintenance
strategy based on midday summer sun for three reasons. Firstly, in summer it can
take as little as 15 minutes to cause sunburn in sensitive individuals.
Secondly, low levels of 25(OH)D are primarily a wintertime problem—but, as
vitamin D in the body has a relatively short half-life of about 90
days,22 summer production would be of limited
effectiveness by late winter. Thirdly, because the skin has some ability to
repair UV damage, the same total dose received over a longer period, say weeks,
results in less (or no) burning, compared to the same dose received in a shorter
period, say 30 minutes.
Our estimate of 8
nmol∙m2∙kJ-1
for the rate constant for production of vitamin D by dUV
(kuv, Table 4) for a presumably typical
person in Christchurch, allows the estimate that on an average day in midsummer
when dUV energy is say 10 kJ/m2, that person
gains 1200 IU/d of vitamin D from their solar exposure. In contrast, in
midwinter, when average daily dUV is probably about 0.5
kJ/m2, only about 60 IU would be made.
Since the amount of solar exposure our subjects had was not
measured, we cannot estimate the absolute amount of additional solar exposure
that would be needed to raise 25(OH)D levels to at least 75 nmol/L year round.
However our modelling prediction that doubling average solar exposure year-round
would only raise the average minimum 25(OH)D by 8 nmol/L suggests that it would
probably be impossible to raise the annual minimum to 75 nmol/L by increasing
sun exposure in Christchurch.
Hence this leaves vitamin D supplementation as the only
practical measure for usefully raising the annual minimum plasma 25(OH)D level.
Note that doubling solar exposure does not double 25(OH)D levels, and the
non-linearity between dUV exposure and plasma 25(OH)D is also evident between
summer and winter when dUV varies by about ten-fold but 25(OH)D by about
two-fold.
A similar study to the present one was conducted in
Auckland, New Zealand, at about the same time (January 2004 to May
2005).9 There, the monthly mean plasma 25(OH)D
levels in women were a little higher (maximum 63 nmol/L in March, minimum 40
nmol/L in August) than in Christchurch, but the mean levels found in men were
markedly higher (maximum 102 nmol/L in March and minimum 59 nmol/L in
September).
The generally higher 25(OH)D levels in Auckland no doubt
reflect its lower latitude (37º) but it is not clear why the men were found
to have much higher levels than the women, particularly as we did not find a
significant gender difference in Christchurch in the volunteer group and, if
anything, a reverse gender difference in the patient group (Figure 2).
Our observation of no significant negative correlation
between PTH levels and 25(OH)D may be because very few (1.5%) of the volunteer
group had severe vitamin D deficiency (25(OH)D<12.5 nmol/L). However when
parametric statistics rather than non-parametric were used, a non-linear
hyperbolic relationship was evident.
Our three compartment model of vitamin D metabolism (Figure
1) is based on the two compartment model proposed and used by Heaney and
coworkers,24 but with two additions. One
addition is a rapid turnover vitamin D compartment (compartment 1) in order to
account for plasma measurements of vitamin
D3,23 and the
other addition is that the rate of conversion of vitamin D to 25(OH)D,
r23, is under feedback control by
25(OH)D. This is to account for observations suggesting that the increase in
plasma 25(OH)D is not linear with increasing dose of vitamin D, but tends to
lessen with larger doses.26
The assumption that this non-linearity is due to feedback by
25(OH)D is largely speculative, since another possibility is that the
25-hydroxylase enzymes in the liver tend to become saturated by larger
quantities of vitamin D. However preliminary manipulation of our model suggested
that the former assumption provided a better fit to the data in Table 1 than did
the latter possibility (data not shown). Note that our model is a pragmatic
mathematical construct designed to describe the plasma measurements in a way
that allows predictions to be made; it is not intended that the compartments
correspond to definite physiological or anatomical entities.
There is no straightforward way to estimate the
uncertainties in the parameter estimates for the model (Table 4) because
different assays for 25(OH)D can give differing values, the uncertainty of some
of the data in Table 1 is unknown, and the parameter optimisation algorithm does
not lend itself to error estimation. In addition, different experimenters can
obtain quite different estimates for ostensibly the same parameter, for example
the slope of the relationship between vitamin D3
dose and plasma 25(OH)D increment.26
The limited number of months for which we measured 25(OH)D
levels in the volunteer group limits the precision of the estimates of the
values of the model parameters kuv and
rf. As the estimate of 350 IU/d for
vitamin D intake from food (and from visits to sunnier places),
rf, might be considered high,
kuv was also estimated on the assumption
that rf was 200 IU/d. This gave a value
of kuv that was only 25% higher,
suggesting that the best-fit estimate of
kuv is reasonably robust. However, we
have not been able to allow for the probably greater exposure of skin to the sun
in the warmer months.
Studies are underway to improve the seasonal range of
measurements, to directly measure personal exposure to solar UV and to quantify
the relationship between personal UV exposure and vitamin D status. The
measurement of actual personal exposure to UV is critical and will be novel and
challenging.
It should be noted that our model, and its predictions
regarding dosages to raise 25(OH)D levels, apply only to vitamin
D3 since Heaney’s
group23 have shown that vitamin
D2 is metabolised much more rapidly and does not
raise or maintain plasma 25(OH)D levels as effectively.
In summary, our study finds firstly, as many
others1 have, that residents in temperate
regions have unhealthily poor vitamin D status, and secondly, that modelling
suggests that vitamin D supplementation in greater than the usually recommended
dosage is required to address the problem.
Competing interests: Drs Florkowski
and Elder are employees of Canterbury Health Laboratories, which might benefit
commercially if there is an increased interest in testing vitamin D levels in
patients upon this paper’s publication (although this paper does not
explicitly advocate testing vitamin D status).
Author information: John Livesey,
Scientific Officer, Endolab, Christchurch Hospital, Christchurch;
Peter Elder, Scientific Officer, Canterbury Health Laboratories;
Christchurch Hospital, Christchurch; M Jane Ellis, Scientific Officer,
Endolab, Christchurch Hospital, Christchurch; Richard McKenzie, Research
Scientist, National Institute of Water & Atmospheric Research (NIWA),
Lauder, Central Otago; Ben Liley, Research Scientist, NIWA, Lauder, Central
Otago; Chris Florkowski, Chemical Pathologist, Canterbury Health
Laboratories, Christchurch Hospital, Christchurch
Acknowledgements: We gratefully acknowledge
the support of the Canterbury District Health Board and NIWA. The UV Atlas
software package was developed at NIWA Lauder by Dr Greg Bodeker.
Correspondence: Dr John Livesey, Endolab,
Christchurch Hospital, Private Bag 4710, Christchurch. Email: john.livesey@cdhb.govt.nz
References:
|
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| | Current
issue | Search journal |
Archived issues | Classifieds
| Hotline (free ads) Subscribe | Contribute | Advertise | Contact Us | Copyright | Other Journals |
|