| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Institute of Bone and Joint Research, Department of Public Health, ANZAC Research Institute, University of Sydney, Prince of Wales Medical Research Institute, Sydney, New South Wales 2065, Australia
Address all correspondence and requests for reprints to: Professor Philip Sambrook, Institute of Bone and Joint Research, Royal North Shore Hospital, St. Leonards, New South Wales 2065, Australia. E-mail: sambrook{at}med.usyd.edu.au.
 |
Abstract |
|---|
 Top Abstract IntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
|
Introduction |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
There have been few prospective studies of the relationship among vitamin D status, PTH, and mortality. Trivedi et al. (14) reported a trend for better survival with vitamin D supplementation in a randomized trial of 2686 community-dwelling subjects aged 65–85 yr, but serum 25-hydroxy vitamin D (25OHD) and PTH were only measured in a small subsample (235 subjects) of the total cohort. We hypothesized that serum 25OHD and/or bone ultrasound would predict mortality in this frail very old population, and we have measured calcaneal ultrasound, serum 25OHD, and PTH in a prospective study of frail, very old, institutionalized subjects to examine their relationship to deaths.
|
Subjects and Methods |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Older people residing in hostels and nursing homes in Northern Sydney were invited to participate. Individuals who were bedbound, bilateral amputees, non-English speaking, or under the age of 65 yr were not included in the study. Eligible residents gave informed consent, or, if unable, their next of kin gave proxy consent. Ethics approval was given by the institutional Human Research Ethics Committee. Institutions were visited in random order, and the study commenced in March 1999. The overall participation rate of eligible residents was 54.5% (hostel, 65.4%; and nursing home, 45.3%). This paper reports data on the first 865 subjects who had completed at least 16 months follow-up, during which time baseline serum 25OHD and PTH were available. The study is referred to as the Fracture Risk Epidemiology in the Elderly study (15). Clinical risk factors were assessed in all subjects by interview including age, sex, weight, height, and medications including use of vitamin D supplements. Comorbidities were assessed in a number of ways, including as illness severity using the Implicit Illness Severity Scale (16). In this assessment, residents were classified on a four-point scale where 1 indicated no symptoms, 2 indicated mild symptoms or conditions, 3 indicated moderate symptoms or conditions, and 4 indicated seriously ill.
Direct assessment of dietary intake is difficult in this population. However, although there are no mandated dietary guidelines for residential care facilities in Australia, all facilities require accreditation, and meals are relatively standardized. In a typical day, nursing home residents are offered three meals and three snacks, whereas hostel residents are offered three meals and two snacks but have access to refreshments at any time. Breakfast usually consists of cereal with milk, prunes, scrambled or boiled eggs, or baked beans with toast and tea or coffee. Lunch usually consists of soup, a main course of either meat or fish with vegetables and a dessert. Dinner usually consists of a light meal such as soup with sandwiches or pies or eggs with a salad and dessert. Snacks usually consist of biscuits or cake and tea or coffee. Each resident’s weight is monitored, and for those losing weight, protein energy supplements are provided. Daily dietary calcium intakes are on average 480 mg in hostels and 360 mg in nursing homes in Australia (17).
Clinical biochemistry and calcaneal ultrasound
At the baseline assessment, a blood sample for measurement of clinical biochemistry was collected. The samples were drawn in the morning, although the patients were not fasting. Serum 25OH vitamin D was measured by RIA (DiaSorin Inc., Stillwater, MN). The assay has a sensitivity of 4 nmol/liter with an intraassay precision of 7.6% and an interassay precision of 9.0%. The laboratory reference range is 39–140 nmol/liter.
Serum levels of intact PTH were determined by a two-site ELISA. The assay has an intraassay precision of 5.5% and an interassay precision of 7.9%. The laboratory reference range is 23–66 pg/ml.
Serum calcium was measured by colorimetric assay (Roche Diagnostics, Indianapolis, IN) using p-cresolphthalein and adjusted for circulating albumin levels, with a normal range of 2.15–2.55 mmol/liter. A modified Jaffé (picric acid) kinetic colorimetric assay was used to measure serum creatinine with a normal range in males of 70–110 µmol/liter and in females 50–90 µmol/liter. Inorganic phosphorus levels were measured by an endpoint method with sample blanking, based on the formation of ammonium phosphomolybdate complex. The normal range in this laboratory is 0.6–1.3 mmol/liter. Serum albumin was measured by BCG colorimetric assay (Roche Diagnostics), with a normal range of 40–50 g/liter (determined in a young adult population). Creatinine clearance was calculated using the Cockcroft-Gault formula (18).
Bone ultrasound attenuation (BUA) was measured at the calcaneus in all participants using the same McCue CUBA Mark II ultrasound machine (15).
Endpoint ascertainment
We ascertained mortality and cause of death from death certificates and chart review during regular visits to institutions.
Statistical analysis
Serum PTH was transformed to a normal distribution by using the natural logarithm. Cox proportional hazards models were used to calculate and determine age and gender-adjusted hazard ratios (HRs) and 95% confidence intervals (CIs) for relations among PTH, 25OHD, BUA, and mortality. Gender, type of residence, and season were included as variables in the multivariate analysis.
|
Results |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Eleven patients with primary hyperparathyroidism and 12 patients with severe renal impairment (serum creatinine > 300 µmol/liter) were excluded from the analysis. The diagnosis of primary hyperparathyroidism was based upon the presence of an elevated serum calcium and PTH level. None were surgically proven in this frail very old cohort. After excluding these 23 subjects, the study sample comprised 842 subjects (182 men and 660 women). Vitamin D supplements in a dose range of 100-1000 IU daily were used by 6.6% of men and 10.0% of women.
Over an average of 31 months (range 16–53 months), 345 subjects died (47.8% of men and 39.1% of women). Causes of death were cardiac, 126; cerebrovascular, 54; infection (mainly pneumonia), 97; dementia, 20; cancer, 17; renal, 9; other, 27; and undetermined, 5 (moved interstate). Vitamin D deficiency (defined as 25OHD < 39 nmol/liter) was present in 75% of men and 83% of women. Secondary hyperparathyroidism (defined as PTH > 66pg/ml) were present in 33% of men and 42% of women. Population characteristics by survival status are shown in Table 1
. Subjects who survived were significantly younger, heavier, stronger, had lower serum creatinine levels, higher 25OHD levels, lower PTH levels, and higher BUA. Although there was also a significant difference for serum calcium, it was a small absolute difference within the normal range and unlikely to be of biological significance. There was a trend for survivors to be less likely to be receiving treatment for hypertension, but this difference was not significant. There was no significant difference in the use of antidiabetic medication by survivor status.
|
After adjustment for age and gender, lnPTH (HR 1.39; 95% CI, 1.17–1.65; P < 0.001), 25OHD (HR, 0.990; 95% CI, 0.982–0.998; P = 0.01), and BUA (HR, 0.991; 95% CI, 0.985–0.997; P = 0.003) were all significant predictors of time to death. The relationship for PTH was seen both using log-transformed and untransformed data. In multivariate analysis that included serum PTH, serum 25OHD, and BUA, only PTH (HR, 1.46; 95% CI, 1.15–1.86; P = 0.002), creatinine clearance, serum albumin, gender, and illness severity remained significant predictors (Table 2). In the final model, serum phosphate and calcium were included as both variables were considered to be important factors that could influence the effect of PTH on mortality. Figure 1 shows the survival curve by tertiles of PTH after adjusting for all other variables in the final model. The effect of PTH on mortality did not change after correcting for variables such as the use of antidiabetic medication, antihypertensive medication, or use of vitamin D supplements. During follow-up, 33 of 345 subjects who died sustained a hip fracture, but there was no significant difference in hip fracture rates by tertiles of PTH.
|
|
To determine whether elevated PTH had an effect on mortality through cardiovascular diseases, cardiovascular mortality was treated as an outcome in survival analyses. The five deaths in which the cause could not be definitely determined were excluded from the analyses. In multivariate analysis, serum PTH was associated with an increased risk of cardiovascular death after stratifying for gender (HR, 1.88; P = 0.003).
|
Discussion |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Vitamin D supplements are not routinely administered to residents of care facilities in Australia, and food is not fortified with vitamin D. Moreover, there is considerable concern about sun exposure due to public health campaigns about skin cancer that severely limits sun exposure. This probably explains the high prevalence of vitamin D deficiency in this population. Although we estimated a maximum calcium intake of up to 640 mg daily, most residents are unlikely to achieve this. Our findings are unlikely to be explained by poor nutritional status because the effect of PTH was seen in the multivariate model when serum albumin, a measure of nutrition and independent predictor of mortality (20, 21), was included. Moreover nutritional status, as assessed by serum albumin, was surprisingly good for frail very old cohort.
Primary hyperparathyroidism has been associated with increased mortality in Europe, at least partly because of increased cardiovascular disease (7, 8, 9, 10, 11), but not in North America (12, 13). However although epidemiological data from referrals to the Mayo Clinic (12) and residents of Rochester, MN (13) did not reveal any overall increase in mortality in primary hyperparathyroidism compared with euparathyroid patients, those subjects whose serum calcium was in the highest quartile had an increased mortality (13). The prevalence of primary hyperparathyroidism in our population was 1.3%, which is higher than that reported in North America (22) but lower than that reported in Europe (23).
PTH and vitamin D have been shown to influence cardiac and vascular function, and in patients with end stage chronic renal failure, hyperparathyroidism and altered vitamin D status are thought to contribute to an increased prevalence of cardiovascular disease and subsequent mortality (24, 25, 26). However, it is unknown whether mortality is increased in patients with mild secondary hyperparathyroidism without end stage renal disease. Carlstedt et al. (26) studied the relationship between serum PTH and mortality and severity of illness in critically ill patients in an emergency or intensive care department. Mortality was 6.4% in all 140 patients and 17.8% in those with elevated PTH, independent of serum calcium levels. Trivedi et al. (14) also reported a slightly higher cumulative survival in vitamin D-supplemented community-living individuals, although the difference between groups was not significant. Renal function was moderately impaired in our population, but PTH remained an independent predictor, after correcting for renal function and vitamin D status. This suggests PTH may have effects on mortality independent of vitamin D status and renal function, the two factors usually thought to be the main drivers of elevated PTH in secondary hyperparathyroidism. In regard to premature cardiovascular mortality, noncalcemic effects of PTH may also be important. PTH has been shown to have direct effects on vascular smooth muscle cells and ventricular myocytes (27) and can impair cardiac energy production, transfer, and utilization by enhanced entry and the accumulation of calcium in the myocardium, either directly and/or indirectly (28). PTH excess is associated with impaired fatty acid oxidation and in animals with secondary hyperparathyroidism due to chronic renal failure have impaired oxidation of fatty acid oxidation that can affect myocardial metabolism (29). In patients with untreated primary hyperparathyroidism, left ventricular diastolic filling dynamics are strongly correlated with PTH levels but not with calcium levels (30). Hemodialysis patients with increased bone resorption and secondary hyperparathyroidism have an increased heart rate, a higher systolic blood pressure, increased velocity of fiber shortening, and shorter left ventricular ejection time (31). Persistent elevated serum concentrations of PTH in the face of normocalcemia after operation for parathyroid adenoma have been associated with development of cardiovascular disease (32).
Other than low serum 25OHD levels and impaired renal function, what other causes of secondary hyperparathyroidism exist in this cohort? Population studies show a gradual decrease in glomerular filtration rate with age from about 125 ml/min at age 20 yr to about 60 ml/min at age 80 yr (33). We have included estimated glomerular filtration rate in our multivariate analysis. A low calcium intake also increases PTH secretion. Mean dietary calcium intakes in residential care in Australia are 401 mg per day, but for many residents it may be substantially lower, with 95% CIs ranging from 333–517 mg per day (17).
It is known that aging is associated with decreased responsiveness of the renal 1-alphahydroxylase to PTH (34). It has recently been reported in a renal failure population that survival is enhanced by treatment with paricalcitol (35). Because we did not measure serum 1,25 dihydroxy vitamin D levels in our cohort, we are unable to determine whether altered responsiveness to PTH induction of 1,25 dihydroxy vitamin D is a mechanism associated with increased mortality.
Mortality rates have also been associated with low bone mass (3, 4, 5, 6, 16). Browner et al. (3) measured bone mineral density (BMD) by single photon absorptiometry at three peripheral sites, including the calcaneus, in 9704 community-dwelling women aged greater than 65 yr of whom 299 died during a mean of 2.8-yr follow up. Each 1 SD decrease in calcaneal BMD was associated with a 20% increase in mortality after adjusting for age. Most deaths were unrelated to occurrence of fractures. The association was no longer significant after adjustment for measures that included general health and muscle strength. The association was strongest for deaths from stroke, and it was speculated low bone mass and stroke may be influenced by low 25OHD, which may affect vascular reactivity. Johansson et al. (4) similarly reported an increased risk associated with BMD (relative risk, 1.23 in men and 1.19 in women for each 1 SD decrease in calcaneal BMD, after adjustments for age and follow up) in a 7-yr study of 653 men and 873 women aged 70 yr and above. Von der Recke et al. (5) found this relationship was strongest for cardiovascular mortality in healthy postmenopausal women, and Trivedi and Khaw (6) observed that hip BMD was a strong and independent predictor of both all cause and cardiovascular mortality in men aged 65–76 yr. Recently, Bauer et al. (36) reported calcaneal BUA predicted mortality in the same population as the original description of an association with BMD (3), and the relationship was independent of age and health status. In our population, PTH remained in independent predictor, after correcting for calcaneal BUA and health status.
This study has certain strengths and limitations. The inclusion of both nursing home and intermediate-care residents allowed examination of mortality in a cohort with a broad range of medical conditions and functional limitations. However, we found that a moderate proportion of subjects, particularly those in nursing home care, declined participation; therefore, the data collected relates only to the more able residents of these facilities. This more frail group would have had a higher mortality, but it is unclear whether this would act as either a positive or negative bias in finding a relationship between mortality and PTH. Approximately 52% of deaths in our population were definitely cardiovascular related (cardiac or cerebrovascular). Inaccuracies in death certification and the resultant assignment of misleading International Classification of Diseases codes, particularly for cardiovascular conditions, can have important effects on cardiovascular mortality rates and trends (37). Also, only a single baseline measurement of the clinical chemistry was obtained. It is generally considered that the estimation of creatinine clearance in the elderly from the serum creatinine using the Cockcroft Gault equation should be viewed with caution, although there have been both positive and negative publications in this regard (38, 39). Most of the concern about the Cockcroft Gault equation relates to those subjects with abnormal BMI (40). However when we have repeated the analysis excluding those with abnormally low or high BMI (i.e. those <19 or 
30), it made no difference to the analysis. Moreover, serum PTH remained a predictor of mortality when either measure of renal function (creatinine or estimated creatine clearance) was included in the model. It was not possible to perform a formal creatinine clearance in this population due to difficulties in collecting 24-h urine samples (e.g. incontinence). A majority of subjects were vitamin D deficient or insufficient. Because serum 25OHD and PTH are closely related, it may be difficult to completely separate the effects of elevated serum PTH from low serum 25OHD on mortality, even with multivariate analyses. It would be important to replicate the study findings in a cohort that included more subjects with normal 25OHD levels.
Our findings support the hypothesis that like some reports in primary hyperparathyroidism, secondary hyperparathyroidism is associated with increased mortality, especially cardiovascular mortality. Because this effect appears independent of vitamin D status, renal function, bone mass, and comorbidities, its mechanisms require further investigation.
|
Footnotes |
|---|
Received February 19, 2004.
Accepted July 22, 2004.
|
References |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
 |
|
 |
|
  H. Dobnig, S. Pilz, H. Scharnagl, W. Renner, U. Seelhorst, B. Wellnitz, J. Kinkeldei, B. O. Boehm, G. Weihrauch, and W. Maerz Independent Association of Low Serum 25-Hydroxyvitamin D and 1,25-Dihydroxyvitamin D Levels With All-Cause and Cardiovascular Mortality Arch Intern Med, June 23, 2008; 168(12): 1340 - 1349. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. P Bjorkman, A. J Sorva, and R. S Tilvis Elevated serum parathyroid hormone predicts impaired survival prognosis in a general aged population Eur. J. Endocrinol., May 1, 2008; 158(5): 749 - 753. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. A. Fisher, E. K. Southcott, W. Srikusalanukul, M. W. Davis, P. E. Hickman, J. M. Potter, and P. N. Smith Relationships between Myocardial Injury, All-cause Mortality, Vitamin D, PTH, and Biochemical Bone Turnover Markers in Older Patients with Hip Fractures Ann. Clin. Lab. Sci., January 1, 2007; 37(3): 222 - 232. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Endocrinology | Endocrine Reviews | J. Clin. End. & Metab. |
| Molecular Endocrinology | Recent Prog. Horm. Res. | All Endocrine Journals |
