Calcif Tissue Int (1992) 51:100-104

Calcified Tissue International 9 1992 Springer-Verlag New York Inc.

Secular Trends in the Incidence of Postmenopausal Vertebral Fractures Cyrus Cooper, 1 Elizabeth J. Atkinson, 1 Mark Kotowicz, 2 W. Michael O'Fallon, 1 and L. Joseph Melton, III 1 1Department of Health Sciences Research, and ZDivisionof Endocrinology/Metabolism and Internal Medicine, Mayo Clinic and Foundation, 200 First Street SW, Rochester, Minnesota 55905 USA Received January 2, 1992

Summary. Several studies suggest secular increases in hip fracture incidence through this century, but little is known about such trends for vertebral fracture. We have examined changes in the incidence of clinically ascertained vertebral fractures among Rochester, Minnesota residents aged 35-69 years, that were first diagnosed between 1950 and 1989. Our results indicate no overall increase in incidence over the 40-year period. Categorization of fractures according to the level of preceding trauma, however, revealed a significant increase in the incidence of fractures following moderate trauma among women aged 60-69 years. This increase occurred between 1950 and 1964, and leveled off thereafter. Rates for severe trauma fractures among postmenopausal women, and for vertebral fractures from any cause among younger men and women, remained stable. The rise in moderate trauma fractures in postmenopausal women paralleled that for hip fractures in Rochester and began to plateau at around the same time. It might have resulted from increased diagnosis of vertebral fractures, but the increase in hip fracture incidence is inconsistent with this explanation. An increase in the prevalence of osteoporosis, however, might account for the trend in both types of fractures.

Key words: Epidemiology - Vertebral fracture - Hip fracture - Osteoporosis - Time trend.

Osteoporosis constitutes a major public health problem through its association with age-related fractures, most notably those of the vertebrae, hip, and distal forearm. Over a million such fractures occur in the United States each year, about half of which involve the vertebral column [1]. Vertebral fractures alone are estimated to account for 52,000 hospital admissions and 161,000 outpatient visits annually [2]. The costs associated with these fractures can only rise in the future, commensurate with the increasing number of elderly in the population. It remains uncertain, however, whether the age-specific incidence of vertebral fracture is also rising, and will further inflate the public health burden. Studies of time trends in the incidence of hip fractures within the United States have revealed steep secular increases in the earlier decades of this century, with a dramatic slowing in the rate of increase from the 1950s onward [3, 4]. In the only two investigations to specifically investigate secular changes in vertebral fracture incidence [5, 6], rates were compared at

Offprint requests to: L. J. Melton, III

only two points in time. Thus, the pattern of any secular trend could not be delineated. The availability of a community-wide diagnostic indexing system and comprehensive unit medical records spanning many decades permitted the investigation of secular trends in the incidence of vertebral fractures among Rochester, Minnesota residents over the 40-year period from 1950 through 1989.

Patients and Methods Population-based epidemiologic research is possible in Rochester, Minnesota because medical care is virtually self-contained within the community and is delivered by a small number of providers [7]. Most care is provided by the Mayo Clinic, which has maintained a common medical record system with its two large affiliated hospitals for over 80 years. The Mayo Clinic dossier-type record, therefore, contains both inpatient and outpatient data. The diagnoses reported in these records are indexed. The index includes diagnoses made for outpatients seen in office or clinic consultations, emergency room visits, or nursing home care, as well as diagnoses recorded for hospital inpatients, on death certificates, and at autopsy examination. Medical records of the other providers who serve the local population are indexed in the same way and are also retrievable. Thus, it is possible to review the details of the medical care provided to the residents of Rochester by the Mayo Clinic, the Olmsted Medical Group, the Olmsted Community Hospital, and hospitals in surrounding areas, including the University of Minnesota and Veterans Administration hospitals in Minneapolis. Only a few private practitioners are located in Rochester, and their records are included in this system as well. Using this data resource (the Rochester Epidemiology Project), we identified all Rochester residents aged 35--69years who were first diagnosed as having one or more vertebral fractures in the 40-year period, 1950 through 1989. Inpatient and outpatient medical records were screened of all patients with any diagnosis relating to fracture, osteoporosis, or demineralization of the spine. Vertebral fractures were documented by radiologist's report. Only compression fractures (including wedge, crush, or endplate fractures) of a vertebral body between T1 and L5 were included; fractures of the posterior elements and transverse processes of these vertebrae were excluded. The date of diagnosis was the date of radiological diagnosis. The occurrence of symptoms, their duration prior to diagnosis, and the level of trauma were recorded. Trauma was categorized as severe (traffic accidents and falls from greater than standing height) or moderate (less than or equal to a fall from standing height). Fractures through areas of bone affected by primary or metastatic cancer or localized bone disease were classified as pathological. In calculating incidence rates, the entire population of Rochester aged 35-69 years was considered to be at risk. The total population of Rochester rose from 29,885 in 1950 to 70,745 in 1990. The age- and sex-specific person-years (p-y) used as denominators in the incidence rates were estimated from decennial census data for Rochester, with finear interpolation between census years [8, 9]. Incidence

C. Cooper et al.: Vertebral Fracture Incidence Trends 200

101

m

>,

~. 175 0 -- ~ o 150 0

Female

d

0 125 |

75

"0

|

5o

_~

25

0

0

-

B

I

I

I

I

1

I

I

I

1950-541955-591960-641965-691970-741975-791980-841985-89 Calendar year

Fig. l. Age-adjusted incidence rates for vertebral fractures in Rochester m e n and w o m e n b e t w e e n 1950 and 1954 and 1985-89. *Adj u s t e d for age to U.S. white population, 1990.

rates were directly age- and sex-adjusted, or age-adjusted for comparisons of men and women, to the population structure of United States whites aged 35-69 in 1990. Ninety-five percent confidence intervals (95% CI) for Rochester rates were estimated from the cumulative Poisson distribution. Changes in age- and sex-specific incidence over time were assessed using generalized linear modeling, where the error structure fitted a Poisson distribution. The dependent variable was the number of fractures that occurred during a given time interval, gender, and age-group. The independent variables were log (person-years), gender, age-group, and time interval.

Results Six hundred and ninety-two Rochester residents (190 men, 502 women) aged 35-69 years were newly diagnosed as having one or more vertebral fractures from 1950 through 1989. Twenty-one patients (9 men, 12 women) sustained pathological fractures. Fracture was related to severe trauma in 159 patients (23%; 77 men, 82 women). The remaining 512 (74%; 104 men, 408 women) had fractures due to moderate or minimal trauma. In 396 (77%) of these moderate trauma fractures, the vertebral fracture was associated with back pain. In the remaining 116 patients, the diagnosis was made incidentally while the patient was undergoing radiography for nonskeletal indications or to exclude vertebral osteoporosis. The age-adjusted incidence rates for all vertebral fractures are shown separately for men and women by 5-year time periods in Figure 1. Rates in women increased during the first 20 years of the study such that the age-adjusted incidence in 1970-74 was 80% greater than that in 1950-54. Rates in men remained stable over the 40-year study period. Different patterns were seen, however, by trauma level and gender. Rates for severe trauma fractures did not change significantly over the 40-year period in either sex (Fig. 2a). As shown in Table 1, overall age-adjusted rates for severe trauma fractures were actually greater in young men (20/ 100,000 p-y; 95% CI, 13-26) than young women (14/100,000 p-y; 95% CI, 8-19). Fractures following moderate or minimal trauma, in contrast, showed a markedly higher overall incidence in women 113/100,000 p-y; 95% CI, 102-124) than in men (36/100,000 p-y; 95% CI, 29-43), as well as a greater age-related increase in incidence among postmenopausal women (Table 1). As shown in Figure 2B, the incidence of moderate trauma fractures in women increased significantly (P = 0.03) between 1950-54 (73/100,000 p-y; 95% CI, 44103) and 1960--64 (114/100,000 p-y; 95% CI, 81-148). Rates

thereafter appeared to plateau, with an age-adjusted rate in 1985-89 of 124/100,000 p-y (95% CI, 94-153). The incidence of moderate trauma fractures in men did not change significantly over the period 1950 through 1989. The trends in moderate trauma fractures among women are explored in more depth in Figure 3. Rates among women aged 35-49 and 50-59 years showed no significant change between 1950 and 1989. In the oldest age category (60-69 years), however, there was an increase in incidence between 1950-54 and 1960-64 which was marginally significant (P -0.05). Even in this age group, rates remained stable between 1965 and 1989. To see if a cohort effect could explain the observed secular trends for moderate trauma vertebral fractures among women, we studied the available rates for 10-year birth cohorts between 1895-1904 and 1935-44. Figure 4 reveals no convincing evidence for a cohort effect. Similar analyses for severe trauma fractures among women, and for both fracture types in men, also provided no evidence for a birth cohort effect but were limited by the small number of cases. An alternative explanation lies in changing case ascertainment. Table 2 shows the distribution of vertebral fracture patients according to trauma level and gender. The proportion of moderate trauma fractures that were asymptomatic was similar in men aged 35-49 years (16%), men aged 50-69 years (11%), and premenopausal women (17%). Among postmenopausal women, however, this proportion was 26%. When classified according to year of diagnosis, the proportion of moderate trauma fractures that were asymptomatic rose from 22% in 1950-69 to 29% in 1970-89. The greater frequency of these incidentally diagnosed moderate trauma fractures among postmenopausal women suggests that an increasing tendency to diagnose vertebral fractures might have contributed to the observed secular trend among this group of women.

Discussion We have examined secular trends in the incidence of clinically diagnosed vertebral fractures among Rochester men and women aged 35--69 years at diagnosis, during the 40-year period 1950 through 1989. Our results suggest that, in postmenopausal women, there has been a statistically significant increase in the incidence of fractures following moderate or minimal trauma, even after age adjustment. Most of this increase appeared to have occurred between 1950 and 1964. The incidence of severe trauma fractures among postmenopausal women, and of both severe and moderate trauma fractures among younger men and women, remained stable between 1950 and 1989. Differences in the time trends we observed for vertebral fractures following severe and moderate trauma are striking. There are several potential explanations for the increase in moderate trauma fractures in postmenopausal women. It might stem, for example, from increased ascertainment of vertebral fractures by physicians. We have p r e v i o u s l y shown that the incidence of clinically diagnosed vertebral fractures in Rochester is substantially lower than the incidence derived from age-specific prevalence estimates, and that the majority of clinically diagnosed cases reveal significant degrees of vertebral deformity [10, 11]. A rise in the proportion of milder cases diagnosed might have accompanied the surge of interest in osteoporosis, but our study was restricted to men and women under the age of 70 years, where such changes in ascertainment are less likely to exert an influence on diagnosis than among the very elderly. Nev-

102

C. Cooper et al.: Vertebral Fracture Incidence Trends 200

200

~.. 175

~.

175

150

8

150

125

0d

125

8 ~'

75 so

8 ~'

50

2 ~_

25

"2 ~_

25

:

:

~

o

I

I

I

g

,oo

"-

o

I

T

I

I

I

I

I

,oo

ale

o

75

I

1950-54 1955-59 1960-64 1965-69 1970-74 1975-79 1980-84 1985-89

I

I

I

I

I

1950-54 1955-59 1960-64 1965-69 1970-74 1975-79 1980-84 1985-89

a Calendar year b Calendar year Fig. 2. Age-adjusted incidence rates for vertebral fractures following (a) severe and (b) moderate trauma, among Rochester men and women between 1950-54 and 1985-89. *Adjusted for age to U.S. white population, 1990.

Table 1. Age- and sex-specific incidence rates for vertebral fractures categorized by trauma level Age group (years) 35-49

50-69

All

Gender

Trauma level

No.

Rate ~

95% CI

No.

Rate a

95% CI

No.

Rate b

95% CI

Men

Severe Moderate Pathological Total Severe Moderate Pathological Total

31 31 0 62 24 48 2 74

20 20 -39 14 28 1 43

13-26 13-26 -29-49 8-19 20-36 0.5-3 33-52

46 73 9 128 58 360 10 428

35 56 7 98 35 215 6 255

25-45 43-69 2-11 81-115 26-43 192-237 2-10 231-279

77 104 9 190 82 408 12 502

27 36 3 66 23 113 3 140

21-33 29-43 1-5 57-75 18-28 102-124 2-5 128-152

Women

CI = confidence interval a Incidence rate per 100,000 person-years u Incidence rate age-adjusted (directly to 1990 U.S. whites)

500 4OO

6.

60-69 yr

00 300 0

,>" o.. 00 400 O0

--

o

300

-

-~

200

-

"(3

100 --

-~ -__ ......

1895-1904 1905-1914 1915-1924 1925-1934 1935-1944

, ~ ~.,

~

7

S

200 II) 0t -

100

0 t-

I

t

I

I

1950-54 1955-59 1960-64 1965-69 1970-74 1975-79 1980-84 1985-89

0 35 44

45-54

I

I

55-64

65-69

Calendar year Fig. 3. Incidence rates for vertebral fractures following moderate trauma, by age group, among postmenopausal Rochester women between 1950-54 and 1985289.

Age group Fig. 4. Incidence of vertebral fractures following moderate trauma in Rochester women, according to decade of birth.

ertheless, asymptomatic moderate trauma fractures were m o r e f r e q u e n t a m o n g p o s t m e n o p a u s a l w o m e n (26%) t h a n a m o n g y o u n g e r w o m e n (17%), or m e n (t3%). A s the frequency of these fractures provides some indication of c h a n g e s in a s c e r t a i n m e n t o v e r the s t u d y period, it a p p e a r s likely t h a t s o m e p a r t o f t h e s e c u l a r i n c r e a s e in v e r t e b r a l f r a c t u r e i n c i d e n c e a m o n g w o m e n r e s u l t s f r o m i n c r e a s e d di-

agnosis of t h e disorder. H o w e v e r , i n c r e a s e d a s c e r t a i n m e n t w o u l d n o t e x p l a i n t h e p l a t e a u in v e r t e b r a l f r a c t u r e i n c i d e n c e o b s e r v e d since the mid-1970s. A n a l t e r n a t i v e e x p l a n a t i o n is t h a t t h e s e c u l a r t r e n d s reflect c h a n g e s in the f r e q u e n c y o f o s t e o p o r o s i s in t h e p o p u lation. T h e r e is a m p l e e v i d e n c e t h a t r e d u c e d b o n e d e n s i t y is a m a j o r d e t e r m i n a n t o f v e r t e b r a l f r a c t u r e risk [12-15]. O u r

C. Cooper et al.: Vertebral Fracture Incidence Trends

103

Table 2. Clinical characteristics of 692 Rochester, MN residents first diagnosed with a vertebral fracture, 1950-89 Men 35-49 years Severe trauma Moderate trauma Symptomatic Asymptomatic Pathological Total

Women 50-69

35-49

50-69

31

46

24

58

26 5 0 62

65 8 9 128

40 8 2 74

265 95 10 428

trend for vertebral fractures following moderate trauma among postmenopausal women resembles that observed for hip fractures in the United States, Sweden, and Britain [3, 4]. Hip fracture incidence rose steeply during the earlier part of this century but began to plateau thereafter. Intriguingly, the plateau in hip fracture incidence in these three Western countries seems to have begun at different time points in this century. It occurred first in the United States during the 1950s, and was followed in Europe some 20 years later [16]. Several determinants of bone density have been proposed to explain this pattern, including the frequency of oophorectomy, increased use of cigarettes and alcohol, dietary changes, and physical inactivity. Oophorectomy prior to natural menopause is associated with accelerated bone loss and vertebral fractures, The age-adjusted incidence of bilateral oophorectomy among women in Rochester more than doubled, from 94/100,000 p-y in 1950-59 to 193/100,000 p-y in 1980-89 [17]. However, the extent to which any effect of oophorectomy might have been offset by postmenopausal estrogen use is unclear. Use of postmenopausal estrogens by U.S. women has risen from 0.6% in 1964 to 7.4% in 1986 [18, 19]. The pattern of use has changed dramatically over this period, however, a consequence of the concern over endometrial carcinoma during the late 1970s. Nevertheless, estrogen exposure might have contributed to the plateau in vertebral fracture rates since 1975. Cigarette smoking and alcohol use are known risk factors for osteoporosis. However, these associations have been detected in men as well as women [20-22], and our data point to a factor that would exert a disproportionate influence on the risk of moderate trauma fractures in postmenopausal women. Moreover, usage patterns of tobacco and alcohol seem out of phase with the secular changes observed in Rochester [3]. The evidence that nutrition is a major determinant of osteoporosis is discordant, and changes in the nutritional status of U.S. women are unlikely to account for the observed increases in fracture incidence [23]. Physical inactivity represents a more plausible contributory factor. High levels of weight-bearing activity are known to protect against osteoporotic fractures [24, 25], and the amount of labor performed by women in developing countries may have declined in the first half of this century, as has walking in lieu of transportation [26]. Perhaps even more noteworthy is the increase in hip fracture incidence among Oriental populations over the last 30 years, commensurate with documented declines in physical activity [27]. Finally, the increase in leisure-time physical activity observed in the United States between 1957 and 1987 might have contributed to the plateau in incidence of osteoporotic fractures [28]. Two other studies, both Scandinavian, have specifically investigated secular trends in the incidence of vertebral fractures. Bengnrr et al. [5] studied men and women aged 60

years and over presenting with thoracic and lumbar vertebral fractures between 1950-52 and 1982-83 in Malmr, Sweden. Among women, incidence rates in 1982-83 were higher than those in 1950-52 at all age groups over 60 years. The rate increase appeared steepest (fourfold) in those aged over 80 years and was least marked in those aged 60-69 years. Vertebral fracture incidence in men only increased substantially among those aged 80 years and over. In contrast, Hansen et al. [6] found the prevalence of vertebral fractures in two samples of 70-year-old Danish women studied in 1979 and 1989 to be virtually identical. Their study focused on healthy subgroups of the general population, however, excluding any impact of secondary osteoporosis on vertebral fracture incidence [29]. The secular trend that we observed for moderate trauma vertebral fractures in postmenopausal women, with a rise in incidence between 1950 and 1964 followed by a plateau, is consistent with both of these reports. In conclusion, this study suggests little overall change in the incidence of clinically diagnosed vertebral fractures among Rochester residents aged 35-69 years, between 1950 and 1989. A statistically significant increase for moderate trauma fractures in postmenopausal women occurred between 1950 and 1964, but rates stabilized thereafter. Although this increase may have resulted from an increased prevalence of osteoporosis, it could also have resulted from increased ascertainment. Further studies of the pattern of incidence of vertebral fractures in different populations are urgently needed. Such studies will enable more effective comparisons with the prevalence of potential risk factors, as well as generating testable etiological hypotheses.

Acknowledgments. The authors would like to thank Ms. Nancy Houar for assistance with data analysis and Mrs. Mary Roberts for help in preparing the manuscript. This work was supported in part by research Grants AG-04875 and AR-30582 from the National Institutes of Health, United States Public Health Service. Cyrus Cooper was supported by a Travelling Fellowship from the Medical Research Council of Great Britain.

References

1. Melton LJ III (1990) Osteoporosis. In: Berg RL, Cassells JS (eds) The second fifty years. Promoting health and preventing disability. National Academy Press, Washington, DC, p 76 2. Grazier KL, Holbrook TL, Kelsey J, Stauffer RN (1984) The frequency of occurrence, impact and cost of musculoskeletal conditions in the United States. American Academy of Orthopaedic Surgeons, Chicago 3. Melton LJ III, O'Fallon WM, Riggs BL (1987) Secular trends in the incidence of hip fractures. Calcif Tissue Int 41:57-64 4. Melton LJ III (1990) Epidemiology of fractures in North America. In: Christiansen C, Overgaard K (eds) Osteoporosis 1990. Proc 3rd Intl Syrup on Osteoporosis. Osteopress ApS, Copenhagen, vol. 1, p 36 5. Bengnrr U, Johnell O, Redlund-Johnell I (1988) Changes in incidence and prevalence of vertebral fractures during 30 years. Calcif Tissue Int 42:293-296 6. Hansen MA, Overgaard K, Gotfredson A, Christiansen C (1990) Does the prevalence of vertebral fractures increase? In: Christiansen C, Overgaard K (eds) Osteoporosis I990. Osteopress ApS, Copenhagen, p 95 7. Kurland LT, Molgaard CA (1981) The patient record in epidemiology. Sci Am 245:54-63 8. Schroeder DJ, Offord KP (1982) A SAS macro which utilizes local and reference population counts appropriate for incidence, prevalence, and mortality rate calculations in Rochester and

C. Cooper et al.: Vertebral Fracture Incidence Trends

104

9.

10. 11.

12. 13.

14. 15. 16. 17. 18.

Olmsted County, Minnesota. Technical Report Series, No. 20, Section of Medical Research Statistics, Mayo Clinic, Rochester, Minnesota Bergstralh EJ, Offord KP (1988) Projected Rochester and Olmsted County populations for 1981-1995. Technical Report Series, No. 38, Section of Medical Research Statistics, Mayo Clinic, Rochester, Minnesota Melton LJ III, Kan SH, Frye MA, Wahner HW, O'Fallon WM, Riggs BL (1989) Epidemiology of vertebral fractures in women. Am J Epidemiol 129:1000-1011 Cooper C, Atkinson EJ, O'Fallon WM, Melton LJ (1992) Incidence of clinically diagnosed vertebral fractures: a populationbased study in Rochester, Minnesota, 1985-1989. J Bone Miner Res 7:221-227 Krr B, Pors-Nielsen S (1982) Bone mineral content of the lumbar spine in normal and osteoporotic women: crosssectional and longitudinal studies. Clin Sci 62:329-336 Firooznia H, Golimbu C, Rafii M, Schwartz MS, Alterman ER (1984) Quantitative computed tomography assessment of spinal trabecular bone. II. Osteoporotic women with and without vertebral fractures. CT: J Comput Tomogr 8:99-103 Cann CE, Genant HK, Kolb FO, Ettinger B (1985) Quantitative computed tomography for prediction of vertebral fracture risk. Bone 6:1-7 Hayes WC, Piazza SJ, Zysset PK (1991) Biomechanics of fracture risk prediction of the hip and spine by quantitative computed tomography. Radiol Clin North Am 29:1-18 Spector TD, Cooper C, Lewis AF (1990) Trends in admissions for hip fracture incidence in England and Wales, 1968-85. Br Med J 300:1173-1174 Melton LJ III, Bergstralh EJ, Malkasian GD, O'Fallon WM (1991) Bilateral oophorectomy trends in Olmsted County, Minnesota, 1950-1987. Epidemiology 2:149-152 Rosenberg L, Shapiro S, Kaufman DW, Slone S, Miettinen OS,

19. 20.

21. 22. 23. 24. 25. 26. 27. 28.

29.

Stolley PD (1979) Patterns and determinants of conjugated estrogen use. Am J Epidemiol 109:676-686 Hemminki E, Kennedy DL, Baum C, McKinlay SM (1988) Prescribing of non-contraceptive estrogens and progestins in the United States, 1974-86. Am J Publ Hlth 78:147%1481 Cooper C, Wickham C (1990) Cigarette smoking and the risk of age-related fractures. In: Wald N, Froggatt P (eds) Smoking and hormone-related disorders. Oxford University Press, Oxford, p 93 Spencer H, Rubio N, Rubio E, Indreika M, Seitam A (1986) Chronic alcoholism: frequently overlooked cause of osteoporosis in man. Am J Med 80:393-397 Seeman E, Melton LJ III, O'Fallon WM, Riggs BL (1983) Risk factors for spinal osteoporosis in men. Am J Med 75:977-983 Schaafsma G, van Beresteyn ECH, Raymakers JA, Duursma SA (1987) Nutritional aspects of osteoporosis. World Rev Nutr Diet 49:121-159 Cooper C, Barker DJP, Wickman C (1988) Physical activity, muscle strength and calcium intake in fracture of the proximal femur in Britain. Br Med J 279:1443-1446 Lau E, Donnan S, Barker DJP, Cooper C (1988) Physical activity and calcium intake in fracture of the proximal femur in Hong Kong. Br Med J 297:1441-1443 Zain Elabdien BSZ, Olernd S, Karlstr6m G, Smedby B (1984) Rising incidence of hip fracture in Uppsala, 1965-1980. Acta Orthop Scand 55:284-289 Lau EMC, Cooper C, Wickham C, Donnan S, Barker DJP (1990) Hip fracture in Hong Kong and Britain. Int J Epidemiol 19:111%1121 Jacobs DR, Hahn LP, Folsom AR, Hannan PJ, Sprafka JM, Burke GL (1991) Time-trends in leisure-time physical activity in the Upper Midwest 1957-1987: University of Minnesota studies. Epidemiology 2:8-15 Riggs BL, Melton LJ Ill (1986) Involutional osteoporosis. N Engl J Med 314:1676-1686