Supplemental calcium for the prevention of hip fracture: Potential health-economic benefits

CLINICAL

THERAPEUTICSVVOL.

21, NO. 6,1999

Supplemental Calcium for the Prevention of Hip Fracture: Potential Health-Economic Benefits Adriunne Ben&h;

PhD, Shelah Leader, PhD,* and Pradip Muhuri, PhD*

‘SmithKline Beecham Consumer Healthcare, Parsippany, New Jersey

ABSTRACT We assessedthe cost-effectiveness of daily calcium supplementation for the prevention of primary osteoporotic hip fractures. The assessment was based on our meta-analysis of the published relative-risk estimates from 3 double-masked, placebo-controlled, clinical trials and our analysis of raw data from the National Health and Nutrition Examination Survey 1988-1994 on the daily intake of calcium supplements by adults in the United States. These data were then used to estimate the preventable proportion of hip fractures. The 1995 National Hospital Discharge Survey database provided the number and demographic characteristics of patients discharged with a primary diagnosis of hip fracture, as well as their discharge destination. The 1990 itemized costs of hip fractures, as estimated by the US Congress Office of Technology Assessment, were inflated to 1995 dollars using the medical ‘Aftiliated Virginia,

with Pmcon, Excerpta Medica, Inc., Reston, at the time this research was conducted.

Accepted

for publication

Printed in the USA. Reproduction in whole

1058

April 9, 1999. or part is not permitted.

care component of the Consumer Price Index. Using these inflated itemized costs, we then estimated the weighted average expenditures, reflecting both the types of services associated with specific hospitaldischarge destinations and the demographic characteristics of discharged patients. The cost of supplements containing 1200 mg/d of elemental calcium for the mean duration (34 months) of the 3 clinical trials was calculated on the basis of 1998 unit-price and market-share data for 6 representative products. For 1995, the data indicate that 290,327 patients aged ~50 years were discharged from US hospitals with a primary diagnosis of hip fracture, at our estimated direct cost of $5.6 billion. Based on the risk reductions seen in the 3 trials, we estimated that 134,764 hip fractures and $2.6 billion in direct medical costs could have been avoided if individuals aged 250 years consumed approximately 1200 mg/d of supplemental calcium. Additional savings could be expected, because this intervention is also associated with significant reductions in the risk for all nonvertebral fractures. Comparing the cost of calcium with the expected medical savings from hip fractures avoided, it is cost-effective to give 34 0149.2918/99/$19.00

A. BENDICH

ET AL.

months of calcium supplementation to women aged 275 years in the United States. If, as the published studies suggest, shorter periods of supplementation result in an equivalent reduction in the risk of hip fractures, calcium supplementation becomes cost-effective for all adults aged 265 years in the United States. The data support encouraging older adults to increase their intake of dietary calcium and to consider taking a daily calcium supplement. Even small increases in the usage rate of supplementation are predicted to yield significant savings and to reduce the morbidity and mortality associated with hip fracture at an advanced age. Key woti: hip fracture, relative risk, calcium supplementation. INTRODUCTION Six million or more adults in the United States have osteoporosis,’ a disease in which the bones become more porous. Osteoporosis, defined as a bone mineral density (BMD) measuring >2.5 standard deviations below that in young adults, is a known risk factor for hip fracture.2-5 Osteoporotic hip fractures entail enormous human and economic costs. Estimates of the direct medical costs for the nearly 300,000 annual hip fractures in the United States have varied widely depending on study year, methods, assumptions, and data sources. One government estimate put 1990 costs for hip fracture patients aged 250 years at $5.4 billion,(j whereas a recent study using 1985 data arrived at an estimated 1995 cost of $8.7 billion for osteoporotic hip fractures in Americans aged 245 years? Furthermore, patients with hip fractures are at increased risk for institutionalization and death.8 There is mounting scientific evidence that supplemental calcium with or without

vitamin D can reduce the relative risk of hip fractures.s11 The National Institutes of Health and the National Academy of Sciences respectively recommend 1500 and 1200 mg/d of elemental calcium12*13 for postmenopausal women who are not taking estrogen replacement therapy. The benefit of estrogen replacement therapy for osteoporosis prevention is predicated on the daily consumption of 1000 mg of calcium.14 However, data from a telephone survey of a representative sample of US households revealed that the average daily intake of dietary calcium was much less than the minimum recommended daily amount of 1000 mg; only half of adults aged 60 to 94 years drank 1 glass of milk every day, which provides only 300 mg of calcium.15 Using risk-reduction estimates from the calcium intervention studies”” and current data on the population at risk for osteoporotic hip fractures, we estimated the number of potentially avoidable hip fractures in the United States for 1995, and the attendant savings, with optimal supplementation of calcium with or without vitamin D, as defined by the intervention studies. We compared the projected 1-year savings in direct health costs with the cost of supplements providing 1200 mg/d of elemental calcium over the mean length of the studies to estimate the cost-effectiveness of calcium supplementation in the prevention of osteoporotic hip fractures. MATERIALS

AND METHODS

Data used in this study came from primary and secondary sources.The primary sources included raw data from the third National Health and Nutrition Examination Survey 1988-199416 (NHANES III) and the 1995 National Hospital Discharge Survey” from the National Center for Health Statistics, 1059

CLINICAL

Centers for Disease Control and Prevention. We also used proprietary market data on the 1998 unit cost and market share for 1200 mg of 6 branded and generic elemental calcium supplements with and without vitamin D to develop a weighted average price. Secondary sources include 3 published articles based on clinical trials of the effects of calcium supplementatiot? 1 and data from the US Congress Office of Technology Assessment (OTA),6 which produced cost estimates for in-hospital and posthospital care for patients with hip fractures. The US Bureau of the Census was the source of data on the size of the 1995 resident population, by demographic subgroup.i8 Number of Hospital Discharges and Patient Demographics We analyzed raw data from the 1995 National Hospital Discharge Survey17 to identify the number and demographic characteristics of patients with a primary diagnosis of hip fracture and their discharge destination. This database captures information about all discharges from nonfederal, short-stay hospitals in the United States. Discharges are described by diagnostic codes based on the International Classijkation of Diseases, 9th Revision, Clinical Modification (ZCD-9-CM).19 The primary diagnosis of interest was fracture of the neck of the femur (ZCD-9-CM 820), although we also examined all nonvertebra1 fractures combined (ZCD-PCM codes 800-829, excepting codes 805 and 806). Medical Costs of Hip Fractures To estimate the costs associated with hip fractures occurring nationwide in 1995, as well as the magnitude of potential savings from avoided hip fractures, we inflated the 1060

THERAPEUTICS’

OTA itemized estimates of 1990 average expenditures (the most current available estimates from a government source) for in-hospital and posthospital care of patients with hip fracture! Between 1990 and 1995, the medical care component of the Consumer Price Index increased 35.4%. This inflation factor was used to adjust the 1990 OTA itemized expenditure estimates to 1995 dollars (Table I). Based on the inflated 1995 OTA itemized cost estimates shown in Table I, total expenditures per patient were projected to reflect the types of services associated with each discharge destination category (Table II). We assumed that patients discharged to their home would incur all of the OTA itemized expenses except those associated with care in a nursing home or a rehabilitation/short-term facility. Patients discharged to a rehabilitation/shortterm facility would incur all but the cost of nursing-home care. Patients discharged to a long-term care facility were not assigned the costs of a rehabilitationlshortterm facility, home health care, or nonmedical home care. Patients who died or had other discharge destinations were assigned only inpatient hospital costs. Because the percentage of patients discharged to different destinations varied by age, sex, and race, we computed the weighted average cost for subgroups of men and women of various ages in 4 racial categories (white, black, other, and race unknown): men and women aged 50 to 64 years; men and women aged ~65 years; women aged ~75 years; and women aged ~85 years. For each of the 20 subgroups, the computation involved a 2-step process. First, the 4 discharge destination-specific expenditures reported in Table II were averaged using destination weights (the proportion of patients in each of the 4 dis-

A. BENDICH ET AL.

Table I. Average 1995 expenditures for in-hospital and posthospital cam per hip-fracture patient.* 1995 ($) In-hospital care for personsaged 50 to 64 years Hospital services Physicianservices Anesthesiaservices Radiology services Physicaltherapy Total In-hospital care for personsaged 265 years Hospital services Physicianservices Anesthesia services Radiology services Physicaltherapy Total Posthospitalcare for personsof all ages Nursing home care Care in a rehabilitation/short-stay facility Readmissionto a short-stayfacility Home health care Nonmedical home care Outpatient physician services Emergencydepartment and ambulance services Total

10,469 2635 780 403 1063 15,320 10,322 1674 432 157 38 12,623 955 1 1005 596 613 445 745 385 13,340

Source: US Congress Office of Technology Assessment (OTA) Background Paper OTA-BP-H-1205 ‘1995 expenditures were derived by inflating OTA estimates for 1990 by the changes from 1990 to 1995 in the medical care component of the Consumer Price Index (35.4%). Due to rounding, totals do not exactly eqti the sums of their components.

charge-destination categories). Second, these weighted estimates were further averaged using race weights reflecting the proportion of discharged patients by race to generate weighted average estimates for each of the 5 age groups. Pooled Relative Risk A MEDLINE@’ search was conducted to identify studies published in English that reported the direct relationship between supplementation with calcium with or without vitamin D and the observed risk of

hip fracture. Although this database indexes only -70% of the world’s literature, we are confident that our search strategy was reliable, because we also searched the references in each of the 3 intervention trials %11as well as those in the literature revie’w in the report from the National Academy of Sciences. l3 No other trials were cited by these expert sources. Studies that reported only the association between calcium supplementation and BMD were not used, because BMD is an intervening end point in terms of the primary outcome of interest-hip fractures. 1061

CLINICAL THERAPEUTICS”

Table II. Estimated average 1995 costs ($) of in-hospital and posthospital care per patient with a primary diagnosis of hip fracture, by discharge status, age, and sex. Discharge Status Routine/discharged home (1, 4, 5, 6, 7, 8)* Discharged/transferredto rehabilitation/ short-term facility (1, 3,4, 5, 6, 7, 8)* Discharged/transferredto long-term care institution (1, 2, 4, 7, 8)* Dead or other (l)t

Men and Women Aged 50-64 y

Men and Women Aged 265 y

18,134

15,406

19,139

16,411

26,627 15,350

23,899 12,622

Source: Inflated US Congress Office of Technology Assessment itemized costs of care (see Table I) and National Hospital Discharge Survey I7 data for 1995 (see Table VI). *Costs included in the calculation: 1 = in-hospital care; 2 = nursing home care; 3 = rehabilitation/short-term facility; 4 = hospital readmission; 5 = home health care; 6 = nonmedical home care; 7 = outpatient physician; 8 = emergency department. tIncludes the following: left against medical advice; alive, disposition not stated; and disposition not stated or not reported.

Three intervention studies found a positive effect of calcium supplementation on the relative risk of hip fracture (Table III).9-11 Because these studies yielded different estimates of relative-risk reductions, we conducted a meta-analysis using STATA’” software (STATA Corporation, College Station, Texas) to arrive at a Mantel-Haenszel combined relative-risk estimate of 0.53 (95% confidence interval [CI], 0.31 to 0.90) for hip fractures. The pooled relative risk for all nonvertebral fractures, including fractures of the hip, was 0.62 (95% CI, 0.47 to 0.82). We did not include the brief note about results from an 18-month follow-up to the study by Chapuy et aL9 because there was little change in the relative risk of hip fracture after 36 months compared with the previous 1%month interventionzO Use of Calcium Supplements We used a previously published method of converting relative-risk estimates to an 1062

estimate of the preventable fraction (PF) of events by first establishing the prevalence of calcium-associated risk (PCAR), which reflects the percentage of the atrisk population not currently receiving 1200 mg/d of supplemental calcium.*l PF is calculated as follows: PF =

PCAR (IRR - 1) 1 +PCAR(IRR1)

where IRR is the inverse of the relative risk. Clearly, if all at-risk older Americans were currently consuming the amounts of calcium with or without vitamin D associated with reduced risk of hip fracture, we would not expect the current number of hip fractures to decrease further. To ascertain the extent of calcium supplementation and the residual PCAR, we analyzed newly released, unpublished raw data on vitamin and mineral use from the Dietary Supplement to the NHANES III.16 The survey used a random, representative, stratified, multistaged proba-

s 8

Once-daily 500 mg calcium (citrate malate) + 700 IU vitamin Ds vs placebo for 3 years

Men and women (mean [r SD] age, 71+ 5 y for men and 71 + 4 y for women [Cl, 70 + 4 y for men and 72 * 5 y for women [PI); dietary calcium, 673 f 349 (p) and 748 f 391 (C) for men, 798 f 366 (P) and 689 i 286 (C) for women

Dawson-Hughes et al” (United States)

Enrolled 187 (C) 202 (P) Completed 148 (C) 170 (P)

3g (Cl 40 (P)

Enrolled 1634 (C) 1636 (P) Followed up 877 (C) 888 (P)

No. of Subjects

RR = relative risk*; C = calcium group; P = placebo group. ‘Ratio of incidence in the calcium group to incidence in the placebo group. The denominator calculated as the number of subjects times the number of years for which they were followed pertains to subjects enrolled, not subjects completing the study.

Daily 1 g elemental calcium (lactate-glutonate and carbonate twice daily) vs placebo for 4 years

Postmenopausal women (mean [? SD] age, 581t4y[C],59*6y[PJ); median (* SD) dietary calcium (mg/d), 760 i 290 (C) and 710*300(P)

Reid et allo (New Zealand)

Once-daily 1.2 g elemental calcium (tricalcium phosphate) + 800 ID vitamin D, vs placebo for 18 months

Intervention and Duration

Women (mean [i SD] age, 84 i 6 y); mean (i SD) dietary calcium (mg/d), 511+ 172 (C)and 514 + 158 (P)

Site)

Subject Characteristics

Chapuy et aI9 (France)

Author (Study

Table III. Calcium intervention trials.

0s;

O2+

= 0.58

= 0.0

152 =o,o 160

2;7+*1f;;;5

RR of Hip Fracture

= 0.26

= 0.69

11+561 = 0.46 26 i 606

‘s

“;;+l;;:;

RR of Nonvertebral Fracture

in the calculation of incidence includes person-years, which are up. In Dawson-Hughes et al, I1 the number of fractures reported

Hip 0 CC) 1 P) Nonvertebral 11 (Cl 26 (0

0 (C) 2m Nonvertebral 2 ((3 8 0’)

Hip

37 m Nonvertebral 66 0 97 m

21 (C)

Hip

No. of Fractures

F

$

3 R X

? Kl

CLINICAL THERAPEUTICS*

bility sample of the noninstitutionalized civilian US population interviewed in their homes to assess the health and nutrition of Americans from 1988 to 1994. All adult respondents aged ~17 years were first asked, “Have you taken any vitamins or minerals in the past month?’ Those who answered affirmatively (41.2% of the weighted respondents) were then asked to show the products

they had taken, and the trained interviewer recorded detailed information about these products. Approximately 2% of the adults aged ~50 years had taken some form of calcium supplement (with or without vitamin D) during the previous month (Table IV). This led to a PCAR estimate of nearly 98% for all older Americans. We also tested the sensitivity of this estimate using published

Table IV. Use of calcium supplements by age, sex, and race, 1988-1994. Unweighted % Total Sample

Weighted %

(N = 20,050)

(N = 187,647,206)

36.8

41.2

2.4 2.2

2.5 2.3

2.4 2.7 3.4

2.5 2.7 3.0

2.9 2.4

2.8 2.4

2.6 2.8 3.5

2.7 2.8 3.1

1.4 1.4

1.4 1.4

1.4 1.9 2.2

1.5 2.3 1.7

1.1 2.1

0.3 0.5

Took any vitamin/mineral

supplement during past month, aged .Z17 years Took calcium supplement during past month 50-64 y ~65 y Women 265 y 27.5 y 285 y White 50-64 y 265 y

White women 265 y 275 y 285 y Black 50-64 y 265 y Black women 265 y 275 y 885 y Other races 50-64Y 265 y Source: Dietary

1064

Unpublished Supplement.

data from

the third

National

Health

and Nutrition

Examination

Survey

1988-1994

A. BENDICH ET AL.

results from the 1986 National Health Interview Survey?2 which indicated that 21.4% of adults aged 265 years had used a calcium supplement in the previous 2 weeks.

years were discharged to their homes and thus incurred no posthospital institutional charges. As expected, Medicare was the predominant payment source for hip fracture patients and accounted for >80% of discharges.

RESULTS

Number of Hip Fractures and Patient Demographics Our analysis showed that in 1995, 290,327 patients aged 250 years with a primary diagnosis of hip fracture were discharged from nonfederal US hospitals; total discharges for all nonvertebral fractures in that age group were 553,689 (Table V). The mean length of stay for hip fracture patients was 8.4 days. Among discharged patients with a primary diagnosis of hip fracture, the majority were aged 265 years, female, and white. More than one third of these patients were aged 285 years. The number of patients who were discharged to specific destinations and their demographic characteristics are shown in Table VI. About half of the women aged 265 years were sent to a long-term care facility, whereas -57% of women aged 285 were sent to a long-term care facility. In contrast, 75% of patients aged 50 to 64

Estimated Weighted Average and Total Expenditures for Care The average expenditure was highest for patients discharged to a long-term care facility, less for those discharged to a shortterm care facility, and lowest for patients discharged to their homes. Estimated 1995 weighted average per-patient expenditures were $18,906 for men and women aged 50 to 64 years; $19,364 for men and women aged 265 years; $19,45 1 for women aged 265 years; $19,744 for women aged 275 years; and $19,943 for women aged 285 years. The highest average costs were incurred by women aged 285 years, largely because of their disproportionate use of long-term care facilities after hospitalization. However, costs were less for black women in the same age group because of lower use of posthospital institutional care. The estimated total cost of inpatient and posthospital care of hip fracture patients aged 250 years was $5.6 billion in 1995.

Table V. Number of fracture patients discharged from US nonfederal hospitals by age, 1995. Hip Fracture Primary or secondary diagnosis of fracture-all ages Primary or secondary diagnosis of fracture-aged

250 years Primary diagnosis of fracture-aged

2% years

Nonvertebral (Including

327,853

1,123,559

314,482 290,327

671,096 553,689

Fracture Hip)

Source: Unpublished data from the National Hospital Discharge Survey.17

1065

CLINICAL

Preventable Hospital Discharges and Costs The number of potentially avoidable hospital discharges was calculated as the number of hip fracture discharges times the pre-

THERAPEUTICS’

ventable fraction of hip fractures associated with calcium supplementation. The preventable expenditures for hip fractures were estimated by multiplying the number of avoidable hospital discharges times the weighted average cost per patient.

Table VI. Number of patients with primary diagnosis of hip fracture in 1995, by discharge status, age, sex, and race. Discharge Status Routine/discharged home White Black Other Race not known Total Discharged/transferred to short-term facility White Black Other Race not known Total Discharged/transferred to long-term care institution White Black Other Race not known Total Dead and other* White Black Other Race not known Total Discharge status (all) White Black Other Race not known Total

Men and Women Aged 50-64 y

Women Men and Women Women Aged 265 y Aged 265 y Aged 275 y

Women Aged 285 y

9209 1127 14 1832 12,182

42,747 3885 2178 6344 55,154

31,088 3179 1585 4863 40,715

24,638 2756 1277 3281 3 1,952

8630 1682 229 463 11,004

516 0 0 406 922

29,617 1233 1467 3085 35,402

24,059 865 1327 2600 28,851

21,806 573 1277 2206 25,862

12,341 152 90 1012 13595

1031 51 0 705 1787

100,375 5766 1509 30,730 138,380

79240 4479 1156 23,210 108,085

71,285 4393 1070 21,440 98,188

35,905 88 433 13,638 50,064

801 13 0 482 1296

34,085 1194 1137 8788 45204

25,810 875 685 6071 33,441

20,021 572 658 4462 25,713

11,114 410 221 1926 13,671

11557 1191 14 3425 16,187

206,824 12,078 6291 48,947 274,140

160,197 9398 4753 36,744 211,092

137,750 8294 4282 31,389 181,715

67,990 2332 973 17,039 88,334

Source: Unpublished data from the National Hospital Discharge Survey.t7 *Includes the following: left against medical advice; alive, disposition not stated; and disposition not stated or not reported.

1066

A. BENDICH

ET AL.

Potential savings associated with avoided hip fractures had to be balanced against the added cost of optimal calcium supplementation for 34 months (the mean duration of the 3 trials). The weighted average of the 1998 per-person cost for 1200 mg/d of calcium supplementation, using the costs of 6 representative products, was $125.68 for 34 months. The costs of the representative supplements with or without vitamin D were similar. The net expected total and per-capita benefits from calcium supplementation were then predicted based on the 3 trials. Assuming that only 2% of adults took calcium supplements, the prevalence of calcium-associated risk was nearly 98%. The pooled relative risk associated with calcium supplementation was 0.53, with an IRR of 1.89. This led to an estimated PF of 0.46 for hip-fracture discharges. We applied this PF to the total number of hospital discharges and total charges by patient demographic characteristics to cal-

culate the number of potentially avoidable hip-fracture discharges and the magnitude of potential savings in total and per-capita medical expenditures based on the 1995 US population. Calcium supplementation was found to be cost-effective for women aged 275 years. The potential savings were greatest among older women, particularly those aged 285 years (Table VII). Calcium supplements remained cost-effective when we applied the 21.4% calcium use estimate from the National Health Interview Survey.22 We also examined the intermediate end point of a prevention program involving daily supplementation with 1200 mg of calcium in adults not already taking supplemental calcium by calculating the direct input costs per avoided hip fracture. The cost of avoiding a hip fracture in the entire US population aged 265 years was $33,118. In comparison, the cost per avoided hip fracture decreased to $14,075

Table VII. Estimated potential benefits of calcium supplementation. Women Aged 275 y Estimated 1995 population (thousands)* Individuals using calcium (%) PF of hospital discharges No. of preventable hospital discharges’

84,168

Preventable total expenditures ($ millions)z Cost of calcium for 2.83 years ($ millions)5 Net benefit ($ millions) Net per-capita benefit

1662 1185 477 50.62

PF = preventable fraction. *Source: US Department of Commerce, Bureau of the tPF x number of hip-fracture discharges. *PF x 1995 weighted average total cost per discharged §$125.68 per person multiplied by the 1995 number of The costs are based on 1200 mg/d elemental calcium

Women Aged 285 y

9426 2.7 0.46

Census,

Population

2612 3.0 0.46 40,847 815 328 486 186.20

Division.‘*

patient. women residents of the United taken for 34 months.

States in the age group.

1067

CLINICAL THERAPEUTICS”

when the calculation was limited to women aged 275 years. Assuming that calcium supplementation was targeted to women aged 285 years, the direct cost of each avoided hip fracture was $8037, far less than the estimated direct medical costs associated with treating a fracture. Because hip-fracture patients are disproportionately white, reflecting the higher prevalence of osteoporosis in this group reported by NHANES III,’ we examined the potential savings from supplementation targeted at this at-risk group. Using the same range of assumptions about the cost of supplements and total costs of care, the trends were similar for white women aged 275 years (results not shown). A recently published study’ based on NHANES III data found that the prevalence of osteoporosis was so small among black and Mexican American men that the data were not statistically reliable. It also found that the prevalence of osteoporosis among Mexican American women was slightly less than that among white women and more than that among black women. Unfortunately, the National Hospital Discharge Survey data do not identify Mexican American patients as such. Therefore, we could not estimate the potential benefits of calcium supplementation in this ethnic group. However, another recently published study 23 based on Medicare claims data for elderly enrollees with Spanish surnames estimated that the hip fracture rate per thousand from 1992 to 1993 was less than that for whites but more than that for blacks, regardless of sex. This finding suggests that calcium supplementation may also be cost-effective for older Mexican American women. Because Reid et allo and DawsonHughes et al” reported that no hip fractures occurred among recipients of cal1068

cium supplementation and that most improvements in BMD occurred after -1 year, and because the study by Chapuy et al9 lasted 18 months, we also tested an alternative set of assumptions based on a mean trial duration of 14 months of calcium supplementation. With this shorter period of calcium supplementation, the intervention became cost-effective for all adults aged 265 years in the United States. The estimated total savings were nearly $372 million, and the per capita net benefit of calcium supplementation for 14 months was $11.08. Although it is beyond the scope of this study to assess the cost-effectiveness of 34 months of calcium supplementation for the prevention of all nonvertebral fractures, using a pooled relative-risk estimate of 0.62, we estimate that 208,430 nonvertebral fractures in adults aged 250 years could have been avoided in 1995 with optimal calcium supplementation.

DISCUSSION Osteoporosis is defined as a BMD that is >2.5 standard deviations below the peak bone mass of young individuals. Those with osteoporosis have an increased risk of hip fracture; however, there is currently no direct association between a particular BMD and the occurrence of hip fracture.13 Thus to determine the cost-effectiveness of calcium supplementation, our analysis concentrated on studies that directly examined the effects of supplementation on the occurrence of hip fracture. We identified 3 intervention studies showing that calcium supplementation reduced the risk for hip fracture.%l’ These were well-controlled clinical intervention trials of sufficient size and/or duration to result in significant effects. Each trial was

A. BENDICH

ET AL.

carried out in a different geographic area (Francep New Zealand,‘O and the United States”); 1 study included 700 III/d” and another 800 IU/d9 of vitamin D. Subjects’ mean (k SD) ages ranged from 58 f 4 to 84 f 6 years, and the duration of the studies ranged from 1.5 to 4 years. One study” included men as well as women. Calcium supplement levels ranged from 500 to 1200 mg/d. Total calcium intake (dietary and supplemental) ranged from 1200 to 1700 mg/d. No serious adverse effects were reported at this level of supplementation. Even with these variations, the studies consistently found significant reductions in hip fracture risk in the supplemented versus placebo groups. All 3 studies also found a significant reduction in the risk for nonvertebral fractures among the supplemented groups. Calcium supplementation for 34 months was found to be cost-effective for the prevention of hip fractures in women aged 275 years; supplementation was particularly economical when targeted at women aged 285 years. Most of the savings would be realized by the Medicare program. The expected cost savings associated with supplemental calcium depend on the choice of baseline estimates of the direct medical cost of hip fractures and the choice of assumptions about the number of months of supplementation, cost of supplements, and preexisting use of calcium supplements alone or in conjunction with estrogen or bisphosphonates . Available estimates of the direct medical cost of hip fractures vary, reflecting a number of methods and a variety of assumptions and proxy measures. Our estimation method is sensitive to cost variations resulting from differences in services associated with specific posthospital discharge destinations. The actual total cost

has not been reported by the dominant insurer for this population (Medicare). Because the OTA data used for our itemized cost estimates did not include prescriptions or assistive devices, we may have underestimated total direct costs. The 1995 data on calcium product prices are comparable to those for 1998, but market-share data were not available for 1995. Therefore, to realistically predict potential savings, we used 1998 market data to determine the average cost of calcium supplements. Published estimates of calcium supplement use derived from national survey data yield widely varying results. Our 2% estimate of calcium supplement use was derived from the raw data recently made available from the NHANES III survey and is consistent with the findings from NHANES I for 197 1 to 1974, which found that 1.2% of all surveyed adults regularly used calcium supplements.24 The 1986 National Health Interview Survey found a much higher estimate of calcium supplementation--20%-among adults.22 The differences in these estimates may reflect both measurement error and variations in survey designs.25 For example, in the NHANES III study, data were not given that could be used to identify the types of antacids used; therefore, the use of calcium carbonatexontaining antacids could not be included in our calculations. However, even when we chose the higher estimate of supplement use22 for our calculations, the main findings were confirmed. It is important to note that estrogen or bisphosphonate use was an exclusion criterion in 1 clinical trial.” Chapuy et al9 did not exclude estrogen users, who accounted for ~1% of patients; Reid et allo did not provide data about estrogen use. Many American women have taken estrogen, and 1069

CLINICAL THERAPEUTICS@

bisphosphonate use is growing. However, data on the combined use of these drugs from representative surveys are not currently available, so it is not possible to determine the extent to which women at risk for osteoporosis are taking either type of medication with or without supplemental calcium (as recommended by experts).8 Although this study focused on the potential for calcium supplementation to reduce the risk of hip fractures, evidence from the intervention trials suggests that there is a similar potential for reduction in the risk of all nonvertebral fractures (Table III). Because there are more nonvertebral fractures, the cost-effectiveness of calcium supplementation may go beyond the reductions associated with avoiding hip fractures alone. It is difficult to estimate total costs for nonvertebral fractures, however, since patients are often treated on an outpatient basis, and available databases do not record these costs per person. CONCLUSIONS Based on pooled risk-reduction estimates from the 3 published placebo-controlled clinical trials in which calcium supplementation was the common denominator, our analysis predicts that this intervention could have reduced the 1995 incidence of hip fractures in the United States from 290,327 to 155,563 (a decrease of 134,764 hip fractures), with consequent savings of $2.6 billion. Further savings could be expetted, because the same intervention is also associated with significant reductions in the risk of nonvertebral fractures. Daily consumption of 1200 mg of supplemental calcium by all women in the United States aged 275 years was found to be costeffective, based on 34 months of supplementation. Supplementation is particularly 1070

economical when targeted at women aged 285 years. If shorter periods of supplementation can yield this benefit, as suggested by several studies,lo*ll daily calcium supplementation becomes cost-effective for all adults aged 265 in the United States. The evidence presented here strongly supports encouraging people aged 265 years to not only increase their intake of dietary calcium but also consider daily calcium supplementation. Even small increases in the rate of supplementation in the group at highest risk (women) is predicted to yield savings for the Medicare program and to reduce the morbidity and mortality associated with hip fracture at an advanced age. ACKNOWLEDGMENTS This study was sponsored by SmithKline Beecham Consumer Healthcare, Parsippany, New Jersey. The authors thank Amy Frankel and Erika Haas for the market data on calcium supplements and Dr. Bruce Daggy for his constructive comments. Address correspondence to: Adrianne Bendich, PhD, SmithKline Beecham Consumer Healthcare, 1500 Littleton Road, Parsippany, NJ 07054-3884. REFERENCES 1. Looker AC, Orwell ES, Johnston CC Jr., et al. Prevalenceof low femoral bone density in older US adults from NHANES III. J Bone Miner Res. 1997;12:1761-1768. 2. MarshallD, Johnell0, WedelH. Meta-analysis of how well measuresof bone mineral density predict occurrenceof osteoporotic fractures.BMJ. 1996;312:1254-1259.

A. BENDICH

ET AL.

3. Cummings SR, Nevitt MC, Browner WS, et al. Risk factors for hip fracture in white women. Study of Osteoporotic Fractures Research Group. NEJM. 1995;332:767-773.

11. Dawson-Hughes B , Harris S , Km11 E, Dalla1 G. Effect of calcium and vitamin D supplementation on bone density in men and women 65 years of age or older. NEJM. 1997;337:67@676.

4. Augat P, Fan B, Lane NE, et al. Assessment of bone mineral at appendicular sites in females with fractures of the proximal femur. Bone. 1998;22:395-402.

12. NIH Consensus Development Panel. Optimal calcium intake. JAMA. 1994;272: 1942-1948.

5. Huang ZP, Himes JH. Nutrition, bone mass, and subsequent risk of hip fracture in white women. Am J Hum Biol. 1998; 10:661667.

13. Institute of Medicine, Food and Nutrition Board. Dietary Reference Intakes for Calcium, Phosphorus, Magnesium, Vitamin D, and Fluoride. Washington, DC: National Academy Press; 1997:S9.

6. US Congress OflIce of Technology Assessment. Hip Fracture Outcomes in People Age Fifty and Older-Background Paper OTABP-H-120. Washington, DC: US Govemment Printing Office; July 1994. Table J-4.

14. Nieves J, Komar L, Cosman F, Lindsay R. Calcium potentiates the effects of estrogen and calcitonin on bone mass: Review and analysis. Am J Clin Nutr. 1998;67: 18-24.

7. Ray N, Chan J, Thamer M, Milton LJ. Medical expenditures for the treatment of osteoporotic fractures in the United States in 1995: Report for the National Osteoporosis Foundation. J Bone Miner Res. 1997; 12:24-35. 8. Kleerekoper M, Avioli L. Evaluation and treatment of postmenopausal osteoporosis. In: Favus M, ed. Primer on the Metabolic Bone Diseases and Disorders of Mineral Metabolism. Philadelphia: Lippincott-Raven; 1996:264-27 1. 9. Chapuy M, Arlot M, Duboeuf F, et al. Vitamin D, and calcium to prevent hip fractures in elderly women. NEJM. 1992;327: 1637-1642. 10. Reid I, Ames R, Evans M, et al. Longterm effects of calcium supplementation on bone loss and fractures in postmenopausal women: A randomized controlled trial. Am /Med. 1995;98:331-335.

15. Elbon S, Johnson M, Fischer J. Milk consumption in older Americans. Am J Public Health. 1998;88: 1221-1224. 16. National Health and Nutrition Examination Survey 1988-1994. Public use data. Hyattsville, Md: US Department of Health and Human Services; 1998. 17. National Hospital Discharge Survey. Public use data. Hyattsville, Md: US Department of Health and Human Services; 1995. 18. US Department of Commerce, Bureau of the Census, Population Division. Current Population Reports, P251095 and Population Paper Listing 41. 19. International Classification of Diseases, 9th Revision, Clinical Modification. Hyattsville, Md: US Department of Health and Human Services, Public Health Service; 1998. 20. Chapuy M, Arlot M, Delmas P, Meunier P. Effect of calcium and cholecalciferol treat-

1071

CLINICAL

ment for three years on hip fractures in elderly women. BMJ. 1994;308:1081-1082. 21. Bendich A, Mallick R, Leader S. Potential health economic benefits of vitamin supplementation. West J Med. 1997;166:306-312. 22. Moss A, Levy A, Kim I, Park Y. Use of vitamin and mineral supplements in the United States: Current users, types of products and nutrients. US Advance Data from Vital and Health Statistics. Hyattsville, Md: National Center for Health Statistics. No. 174, vol. 8: July 18, 1989.

1072

THERAPEUTICS”

23. Lauderdale D, Jacobsen S, Furner S, et al. Hip fracture incidence among elderly Hispanics. Am J Public Health. 1998;88: 124551247. 24. Block G, Cox C, Madans J, et al. Vitamin supplement use by demographic characteristics. Am J Epidemiol. 1998;127:297-309. 25. Patterson R, Neuhouser M, White E, et al. Measurement error from assessing use of vitamin supplements at one point in time. Epidemiology. 1998;9:567-569.