Document 6bGNX0jgYJaOvMeRognG0G483

> SEP IS '92 14:49 FROn CLEU-HLTH-SCI-LIB PAGE.002 C, Hooper A. ct al. Blood interaction in Tokelauan mi'and. J Chronic Oil 1977; E. Saimond C. et ai. Blood ,i children in two contrasting Epidemiol l97S:lOS:-2S3-8. SL. Acculturation and coro' Japanese-Araericans. Am J 125-47. J. The health of Hispanic* in Itcd States: an epidemiologic ;h Rep 1986:101:253-65. American JeumJ of Eptiamniogy Cepyngrit 1992 &y tt jonrw HopHn* twveroty School Hygiene and Pubic wealth Ail ngtit* reserved Vd. 135, No. 12 Printedin U.S.A. Increased Mortality from Brain Tumors: A Combined Outcome of Diagnostic Technology and Change of Attitude toward the Elderly Baruch Modan,1 * Diane K. Wegener,' Jacob J. Feldman,' Harry M. Rosenberg,1 and Manning Fetnleib' United States national data were used to assess factors responsible for the increase of brain tumor mortaBty. Between 1068 and 1088, death rates increased 50% among those aged 65-74 years, 200% among those aged 75-84 years, and 800% m the oldest old. Rate of increase and maximum death rate have changed over time. Death rate among the population aged 65-74 years peaked in the mid-1980s, while among those aged 85 years and older it is projected to continue increasing throughout the 1990s. The patterns of rate increases were almost identical in the two sexes, as weO as among whites and nonwhites. There was a strong correlation over time of death rates with head diagnostic procedures (r - 0.96) and with the pace of computerized axial tomography mstaHation if * 0.91). The authors conclude that the reported increase in brain tumor mortality is not genune, but represents a combination of three factors: availability of more sophisticated, noninvasive diagnostic technology; change in the attitude toward care of the elderty; and introduction of support programs such as Medicare that facilitate diagnostic procedures in the elderty. Am j Epidemiol 1992:135:1349-57. aging; brain neoplasms; incidence; mortality; patient discharges; technology Several observations made over the past few years have demonstrated an apparent increase of both mortality and incidence rates of brain tumors in the United States and elsewhere (1-7). There are at least six possible explanations for this phenomenon: i) It is a true increase, due to an exposure Raetttvad tar publication Sptmbr 4, 1991. and n Inal form January 21.1992- Abbrawation: CO. International OettiHeedoIt of OHaasas. ' Nation* Center for Health Statistic*, Centers tor Olaeata Control. HyattsvHa. MO * Department of Clinical Epidemiology, Own Sheba Medical Canter, Tel Aviv University Medical School. Tel Aviv, Israel. Reonm requests to Dr. Saroch Modan. National Cantor for Health Statistics. Room 1070, 6525 Betcraat Road. Hyattsvie. MO 20782. The authors appreciate the contributions of Patricia Knapp- Office of Analysis and Epidemiology; and Jeffrey Maurer, Elizabeth Godfrey, and Joan Ksemtr. Mortality Statistics Branch. Division of Vital Statiatica. They also acknowledge the gracious effort of Homy Knox of the Canter ter Device* and Control. Food and Drog Adminis tration. to a newly emerging carcinogen or to decreased exposure to a protective factor, 2) It is a delayed reaction to an increased exposure during early age to a carcinogen with an extremely prolonged latency (e.g., the insecticide dicWorodiphenyltrichloroethane (DDT); 3) There is a change in disease classification (i.e., a shift from the International Clas sification of Diseases (ICD) version 8 to version 9); 4) It i$ the result of improved diagnostic workup (e.g.. use of computerized axial tomography or magnetic resonance im aging, or a more precise histopathologic diagnosis); 5) There is a higher probability for brain tumor to be coded as an underlying, rather than a contributory, cause of death; or 6) there is a change in health care delivery _ -reflecting, in part, a change in social val4 1348 CMA113925 SEP IS '32 14:50 FROM CLEU-HLTH-SCI-LIB PAGE.003 1350 Modan et al. The differentiation between these six al ternatives has strong etioiogic and public health implications. The following report presents an in-depth assessment of brain tumor mortality patterns in the United States over a 20-ycar period in an attempt to delineate the main factors responsible for the reported increase. The distinct morpho logic categories of brain tumor were taken into consideration, and health care utiliza tion data were incorporated. MATERIALS AND METHODS United States mortality statistics, based on death certificate information available from the National Center for Health Statistics through the Compressed Mortality File (8, 9). were used for the period 1968-1988. Data from all 50 states and the District of Columbia were analyzed. Age, sex, race, and ICD-specifk rates were derived using the underlying cause of death. For verification purposes, similar data were obtained for se lected years using multiple cause-of-death files (10). These data were used to identify deaths for which brain neoplasm was present on the death certificate, irrespective of whether brain neoplasm was selected as the underlying cause of death. Because of the shift from International Classification of Diseases, Adapted, version 8 (1CDA-8) to version 9 (ICDA-9) in the midst of the study period, different ICD categories had to be selected. For the earlier period (1968-1978X using the 1CDA codes selected included the following: malignant brain tumors (ICDA-8 codes 191, 192.0. 192.1. and 192.9): benign brain tumors (ICDA-8 codes 225.0. 225.1. 225.5, and 225.9) ; and unspecified brain tumors (ICDA-8 codes 238.1. 238-2, 238.3. and 238.9) . For the later period (1979-1988), the codes selected were: malignant brain tumors (ICDa-9 codes 191, 192.0. 192.1. and 192.9) : benign brain tumors (ICDA-9 codes 225.0, 225.1. 225.2, and 225.9); and. un specified brain tumors (ICDA-9 codes 237.5. 237.6. 237.9. 239.6, 239.7, and 239.9). For the multiple cause-of-death data, the addi tional benign brain tumor codes of 198.3 and 198.4 (ICDA-8) and 198.3 (ICDA-9) were included. Information on the number ofyearly neu rosurgical and neurodiagrtostic procedures was obtained from the National Center for Health Statistics National Hospital Dis charge Survey (ll). Data from the survey were analyzed for the yean 1979-1989. Rec ords were selected that included any men tion of operations of the head (including International Classification of Diseases, Ninth Revision. Clinical Modification (ICD9-CM) classification ofprocedures 01.02. or 04) or mention of diagnostic procedures in volving the head, including ICD-9 CM clas sifications for tomography (ICD-9-CM codes 87.03 and 87.04). magnetic resonance imaging (1CD-9-CM code 88.91), and radio isotope scans (ICD-9-CM codes 92.11 and 92.12). Information on the installation of major imaging equipment was provided by the Center for Devices and Radiological Health, Food and Drug Administration. Depart ment of Health and Human Services. The information was provided separately for to tal body and head only equipment and in cluded the date and state of installation of each unit, irrespectively of whether it was new or old. Therefore, for some equipment, there would be more than one installation noted. However, no major bias could be ascribed to this multiple installation factor since the data are used here to reflect the increment in the number of operating units. A greater bias would occur when new equip ment replaces old equipment that is retired and a new installation is counted. Regression analyses were performed using least square procedures for the equation ln(rate) = bn + b\T + bfT1, where hi are estimated coefficients, and T is the time in years. Correlations were estimated using the Pearson procedure. RESULTS There were 250,597 deaths with an un derlying cause of brain tumor in the United States from 1968 to 1988. Approximately, three quarters of the tumors leading to death were classified as re ure 1 for selected y years, there was a coi during this period in for those over age no increase at young death rates was 50 p ulation aged 65-74) population aged 75centin those aged 8: Figure 1 and table that the rate of inert the point of maxir changed over time, specific death rate i: population aged 55* the maximum ctpang In 1980. the maxin rate again changed years. We regressed the brain-tumor death r quadratic functions curves are shown ii evident that the me population aged 65mid-1980s, while older it is projecte throughout the li years present an i the mortality inert The characterist: Rat* fe*rr< as Lot* 5 CMA 113926 figure 1. Aqaspecffie ndisaa. 1 SEP IS '92 4:51 FROM CLEU-HLTH-SCI-LIB ) PAGE.004 Brain Tumor Mortality 1351 } and 198.3 (1CDA-9) number of yearly neu<diagnostic procedures e National Center for tional Hospital Dis9ata from the survey -cars 1979-1989. Recit included any men- the head (including /ration of Diseases. al Modification (ICD- procedures 01.02. or znostic procedures in king 1CD-9 CM clasography (ICD-9-CM ). magnetic resonance ide 88.91), and radioCM codes 92.11 and installation of major vas provided by the : Radiological Health, nlnistration. Depart* 'tuman Services. The ded separately for to,y equipment and in* ate of installation of ly of whether it w for some equipment, than one installation ntyor bias could be >ie installation factor 'd here to reflea the r of operating units, cur when new equippment that is retired i$ counted, vere performed using es for the equation biT\ where b, are and T is the time in i estimated using the were classified as malignant. As seen in fig ure l for selected years and table 1 for all years, there was a consistent gradual increase during this period in age-specific death rates, for those over age 65 years, but practically no increase at younger ages. The increase in death rates was 50 percent among the pop ulation aged 65-74 years, 200 percent in the population aged 75-84 years, and 800 per cent in those aged 85 years and older. Figure 1 and table i also illustrate vividly that the rate of increase and the location of the point of maximum death rates have changed over time. The maximum agespecific death rate in 1968 occurred in the population aged 55-64 years, but in 1972 the maximum changed to those aged 65-74. In 1980, the maximum age-specific death rate again changed to those aged 75-84 years. We regressed the age-specific malignant brain-tumor death rates on time by using a quadratic functional form. The resulting curves are shown in figure 2. It becomes evident that the mortality rate among the population aged 65-74 years peaked in the mid-1980s, while among those age 85 and older it is projected to continue increasing throughout the 1990s. Those aged 75-84 years present an intermediate pattern, but the mortality increase appears to be slowing. The characteristic patterns of increases in age-specific death rates only among the pop ulation aged 65 years and older, changes in the age of maximum death rates, and rates of increase in death rates within age groups over the time period are almost identical in the two sexes, as well as among whites and nonwhites. Similarly, no consistent pattern was observed in change of death rates by state, except that those states where faster change was noted in the early period (19681978) largely had slower rates of increase in the later one (1979-1988). These characteristic changes in mortality patterns have occurred in both the malig nant and benign brain-tumor categories. Only the unspecified brain-tumor mortality, i.e.. those tumors not classified as either malignant or benign, which are primarily without a histopathologic classification, continued to increase with age through the oldest-old age category in recent years (figure 3). Moreover, the proportion of all brain tumors that were unspecified increased with age. For the period 1979-1988. the propor tion of the total brain tumors that were unspecified was 14 percent in the age group 45-64 yean, 18 percent in the group aged 65-74 years. 27 percent in those aged 75-84 years, and 41 percent in those aged 85 years and older. These proportions have been rel atively steady over the 20-year study period. Figure 4 highlights the changes that should deaths with an untumor in the United ->83. Approximately, nore leading to death Laws *14 15-44 a*34 33-44 4*84 S*4 5*44 7**4 aSPIu* An* <h ywil "1OM'-- --1>4&72- - -1O47#-......1.4*S.O.........184|*.4.........1A48..8... FIQUAE 1. Ag*tp8eMlo death cm** *i the United States from aft brafri tumors for selected years between 1968 and 1988. GMA113927 SEP IS '92 14:51 FROM CLEU-HLTH-SCI-LIB PAGE.005 1352 Modan et ai. TABLE 1. Ags-apscific mortality* from all brain tumors in th# United Staffs, by year of dsaih, 1968-1985 1 par 100,000 papulation Age group (year*) Year 65-74 75-84 &$ MO. R4M No. Rat* No- Rat* il: *1r 1968 1969 1.800 1,916 14.8 15.8 552 9.3 38 3.1 556 9.3 53 4.8 1970 1,990 16.0 642 105 58 3.8 1* 1971 2,161 17.0 756 11.8 75 5.0 1972 2.294 17.8 804 12.3 78 5.1 Jr 1973 17.6 867 13.0 99 6.2 1974 2,463 18.1 971 14.3 118 6.9 ! 1975 1976 2.617 2.870 18.8 20.2 1.028 1,197 14.8 * 16.8 159 185 8-7 9.8 1977 2.984 20.4 1.351 18.8 254 128 1978 3.197 21.3 1.590 215 271 12.9 1979 3.176 20.7 1.646 21.7 323 14.7 1980 3.378 21.7 1,820 23.6 377 16.8 ;f 1981 3.368 21.2 1.957 24.5 443 18.8 1982 3,506 21.7 2,178 26.4 484 19.9 1983 3.401 20.6 2.179 25.9 9R0fl0fl 22.1 i 1984 3,638 21.7 2.358 27.4 609 23.2 1988 3.719 21.9 2,435 27.6 674 24.9 1986 3,750 21.6 2.509 28.6 672 24.2 1987 3,886 22.0 2.648 283 793 27.7 ,, 1988 3,817 213 2.701 28.4 806 27.3 1: * Undertymg cause of deatti onty. fiat* (per 100,000) Rat* (peril 20 .. . . 10 5_ a ------- L***4 FIGURE 3. Age-tpsdfc ' between 1968 and 1988. Rats (pat so id ,.H- 5 --'-- 3 2 1-- 1935 CMA 113928 FIGURE*. Ag*adjiisttd otdsr. 1968-1988,6y rmy tor IndMdutts aged 66 ye Am6*74 A0K73-O4 Age* *5P*Ji FIGURE 2. Expected agfrapadfte megn*nt trairvtumor death rat**, based on regression on time u*w a Quadratic function form (based on United States death rates. 1968-1988). be ascribed to ICD~9, which was adopted in 1979. There was an apparent temporary de crease in malignant brain tumor death rates, hut a slight increase among the unspecified brain-tumor death rates. The benign braintumor mortality was unchanged. The ratio of deaths with any mention of brain tumors to those deaths for which the underlying cause of death was determined to be brain tumor has remained unchanged throughout the period f study. For malig nant brain turnon, there was an excess of :% j up to 9 percent in rec cause of brain tum< unspecified brain ha amounted to appr These differences w were similar in male indicate tltai brain brain tumor spccifi lected as the underl) figures 5-7 disp changing medical study period. These enures of the head SEP IS '92 14:52 FROM CLEU-HLTH-SCI-LlB PAGE . BBS trotdactn. 1968-19M RaW (pW 100,000) Brain Tumor Mortality 1353 *85 >. Kata < 3.1 3 4.8 $ 3.8 5 SO 3 5.1 9 6.2 8 6.9 9 8.7 IS 9.8 4 12.8 1 12.9 3 14.7 7 18.8 3 18.8 4 19.9 3 22.1 9 23.2 4 24.9 2 24.2 3 27.7 i 27.3 FIGURE 3. Age-specWe death raws from unspecified 6rar> tumor mortality in the united Stales tor selected years between 1968 end 1988. Rato (per 100,000) vswon on Dm* using a ith any mention of oaths for which the >.th was determined .`mained unchanged >f study. For malige was an excess of Year Mtfgwr* Barter UrwpgsMad .FIGURE 4 AgeedjusMddsath races m the united States from brain tumors among mdMduate aged 66yews and Older, 1968-1988, by morphologic dawfflcMfcn. Reset are age adjusted to the 1980 United States age distribution for individuals aged 65 years and Older, using dlredt adjustment up to 9 percent in records with an underlying cause of brain tumor, while in benign and unspecified brain neoplasms the discrepancy amounted to approximately 30 percent. These differences were not age related and were similar in males and females. The data indicate that brain tumor, and malignant brain tumor specifically, were usually so* lected as the underlying cause of death. Figures 5-7 display several aspects of changing medical technology during the study period. These include diagnostic pro cedures of the head, surgical procedures of the head, and the rate of computerized axial tomography installation. Figures 5 and 6 summarize the steep increase in use of di agnostic procedures of the head, especially among patients aged 55 years or older, in contrast to the plateau in brain surgery. The correlation between the rates of diagnostic and surgical procedures to the head among patients aged 5 years or older was only 0.27. However, if the diagnostic procedure rate is compared with mortality rates, the correla tion was 0.96. The correlation of surgical procedures with mortality was 0.29. v SEP '92 14:52 FROM CLEU-HLTH-5CI-LIB PAGE.007 1354 Modanetai. Procedures (in thousands) Population (in millions) so F- )l*d Surreal Pnudun RGVM 5. Number of hwd surgical and diagnostic procedure* among individuals aged 6S years and older and popUatfon size In the United State*. 1979-1889. T W RGUAS7. Number numtMrdntaaddEagr 1974-1989. Number (In thousands) FKiutEft. niw--.?-j wr--gR -a_i iTo. w-i-M--fi *f HoaddaonoatePrnceduras A Year Ages88*74 ...... AgMT$+ -- agaepecWc number* of head surgical and diagnostic procedure* m the United States. 1979-1989. In figure 7, comparison is made between the number ofdiagnostic procedures and the pace of computerized axial tomography in* stallation in the United States (r = 0.97). The figure shows a very fast incline starting in 1974. The correlation between comput erized axial tomography installations and brain-tumor mortality after age 64 years is 0.91. The number of neurologists in the United States increased from 1,192 in 1970 to 5,087 in 1987 (12), a 3.6-fold increase in the ratio of neurologists per capita. DISCUSSION At least 50 percent of cancer has been estimated to be environmentally determined (13). Therefore, a sudden change in mortal ity patterns, especially in one specific tumor type, might indicate an introduction of a new carcinogen, an increased exposure to a prevailing carcinogen, or a decreased expo sure to a protective factor. The apparent increase in the morbidity and mortality from brain tumors has been construed as a posa ble failure of public health officials to inter- cept the rise in cs been an increase mental exposures an exposure 'to stance to which exposed (e.g.. aii the. increase is I females, in wh out a specific another withir In contrast increases amo population str has been limite and older. Furtl* increase is direct! elderly populatic crease among th an 800 percent ir years or older. By the same tc ual impediment occurs fust in the only subsequent!: be more consist However, no. inc noted among the the period 1968-- if a cohort efTe -V reached a maxi '.'A*"h aged 65-74 years CMA113930 SEP IS '32 14:53 FROM CLEU-HLTH-SCI-LIB PAGE.008 Brain Tumor Mortality 1S5S llions) I vooo ! 500 300 ZOO too so 30 30 to years and oWsr and HoaOonty.CT --A-- Dooms --*-- FIGURE 7. Nurtibor of computerized axial tomography installations for total body and hoad only oqulpmont. number of hood diagnostic procedures. and deem* among indMduals aged 65 yaars or oldor In tne twtod SUM. 1974-1989. 1090 ataa. 1979-1989 oncer baa been ally determined <ange in mortal* i specific tumor traduction of a d exposure to a lecreased expo The apparent 1 mortality from `rued as a possiifficials to inter cept the rise in cancer (14). Yet, if there has been an increase due to changes in environ mental exposures, one would have to assume an exposure to a widely distributed sub stance to which all population strata are exposed (c.g.. air or drinking water) since the increase is found in both males and females, in whites and nonwhites, and with out a specific predilection to one state or another within the United States. In contrast to the equity in death rate increases among sex, racial and geographic population strata, the increase in mortality has been limited to individuals aged 65 years and older. Further, the magnitude of the increase is directly related to age within the elderly population, with a 50 percent in-' crease among those aged 65-74 years and an 800 percent increase among those age 85 years or older. By the same token, the fact that the grad ual impediment to the increasing mortality occurs first in the age group 65-74 years and only subsequently in those aged 75-84 might be more consistent with a cohort effect However, no increases in death rates were noted among the younger age groups during the period 1968-1988. as would be expected if a cohort effect were present and had reached a maximum effect among those aged 65-74 years in 1980. Moreover, for the population as a whole, the maximum age* specific death rate shifted from the popula tion aged 55-64 yean in 1968 to the those aged 75-84 years by 1980. That is, the tran sition occurred faster than the population was aging. This indicates that a process sep arate from aging was increasing the braintumor mortality in the elderly population. It is also apparent that the factor affecting the change is beginning to subside, since the rate of change in mortality is decreasing in all age groups and is now unchanged in the past decade for the population aged 65-74. years. Given the fact that the risk of cancer does not continually increase with increasing age either for specific years or within birth cohorts, it is unlikely that a long-latency: carcinogenic effect is present This is in line: with the observations of Schwara (15). With regard to the impact of changes in the ICD versions, there was a one-time, slight decrease in the death rates from ma lignant brain tumor when the ICD-9 version was introduced. This decrease, in addition to the observation that the increases in death rates have occurred throughout the 19681988 period, is contradictory to the hypoth esis that ICD changes may explain the in crease in either malignant or total braintumor mortality. Another possible explanation, that a CMA 113931 SEP IS '92 14:53 FROM CLEU-HLTH-SCI-LIB PAGE.009 1356 Modanetai. change has occurred in the way physicians complete the death certificate, also appears to be unlikely. The proportions of braintumor deaths for which the brain tumor is listed as the underlying cause has been un changed for all of the morphologic types throughout the period. The alternative ex planations for the change in brain-tumor mortality. i.e.. that we arc dealing with a methodological and/or social phenomenon, seem more likely and are further supported by the strong correlation of mortality with technologic advances. The increase in mor tality parallels the introduction of sophisti cated diagnostic technology, first with com puterized axial tomography and later with magnetic resonance imaging. The fact that a better correlation of mor tality is obtained with diagnostic techniques than with brain surgery supports the hypoth esis that the increase is due to more intensive diagnosis that may have identified a greater proportion of existing brain tumors rather than a true incidence increase that would have necessitated more surgical as well as diagnostic efforts. Indeed, the number of surgical procedures is largely unchanged in recent yean. By the same token, the fact that in recent years the ill-defined category of brain tumor of unspecified origin has con stituted the only brain tumor diagnostic group that has had no age-related deflection in mortality rates indicates that only limited surgical action is undertaken after diagnosis, perhaps because of the physicians' reluc tance to operate on the oldest old. In fact. Surveillance, Epidemiology, and End Re sults data show that in 1983-1987 the per cent of diagnosed brain tumors that were microscopically confirmed was 88.2 among individuals aged 65-69 years, but slightly less than 40 percent among individuals aged 85 years or older (3). In spite of the above deliberation, im proved diagnostic technology alone can hardly provide the full answer to the phe nomenon under consideration. Computer ized axial tomography was instituted only in the mid-1970s. Although the use of this im aging ton! propagated quickly, it did not gain an Overall comprehensive population cov erage until the late 1970s, i.e., approximately 10 years after the increase in death rates started. We believe, therefore, that in addition to the improved diagnosis per se, the apparent increase in brain tumor mortality can be explained by a drastic change in the overall approach of our society to the elderly and their consequent increase of health utiliza tion (16). These changes reflect a new set of values toward the elderly, particularly the oldest old. who when afflicted might have been previously diagnosed as stroke victims with only minimal diagnostic workup, and the introduction of support programs such as Medicare that cover the bulk ofdiagnostic expenditures. Those factors might also explain why the introduction of new imaging techniques hardly affected the death rates from brain tumors in younger subjects. Among the younger subjects, the previous use of less sophisticated, but more invasive diagnostic aids, such as angiography or myelography, was replaced by the new diagnostic technol ogy. Among the older subjects, there was a reluctance to - use invasive procedures. Hence, with the introduction ofless invasive diagnostic procedures, there was an increase in the number of tumors diagnosed. Indeed, the Surveillance. Epidemiology, and End Results data indicate that among the oldest old the percent of cases with surgical confir mation dropped from 54 percent in 19731977 to 40 percent in 1983-1987 (3). while the incidence increased, indicating that, al though more tumors are diagnosed, there is a reluctance to perform invasive procedures among this age group. It might be pointed out that historically, at least as far hack as the 1945-1954 data, the incidence and mortality rates of brain tumor for the elderly in Rochester and Olmsted County. Minnesota, populations (17-19) have been much higher than those reported nationally. The combination of ex cellent health care in Olmsted County, in cluding neurology and neurosurgical exper tise. and an overall high autopsy rate that assured accurate diagnosis of brain tumor even before computerized axial tomography and magnetic : available are in are brought out In conclusion bined effect b more sophistic madeavailablei a change m tht toward elderly; ingness to diagr procedures: am Medicare progr medical cart. T cess with each decade and is s oldest old. REFERENCES 1. Bahanuka M. t national mortal neoplasm: (fistr 1988:17:33-8, 2. Davis DU Hod In cancer-mortal Japan, England 1990:336:474-8 3. Davis DL, Hod increasint in in Hod D, edl " trial countrie 191-204. 4. Greig NH. R annual indde mots in the 811621-4. 5. Boyle P. Mai increased itxr. turnon in theca 81*1394-6. 6: Ahlbom A. Son oWfir. Anti N Y 7. DidU R. PfOgrtt CMA113932 SEP IS '92 14:54 FROM CLEU-HLTH-SCI-LIB PAGE.010 Os, i-c,, approximately crease in death rates i, that in addition to ; per se. the apparent or mortality can be change in the overall y to the elderly and tie of health utilizas reflect a new set of `riy, particularly the afflicted might have sed as stroke victims gnostic workup, and ipon programs such he bulk ofdiagnostic also explain why the imaging techniques th rates from brain ibjects. Among the nevious use of less invasive diagnostic hy or myelography, ' diagnostic tcchnolsubjects, there was a vasive procedures, ction ofless invasive here was an increase s diagnosed. Indeed, miology. and End it among the oldest with surgical confir* t percent in 197.383-1987 (3). while indicating that, oldiagnosed, there is ivasive procedures ut that historically, te 1945-1954 data, alitv rates of brain in Rochester and esota. populations i higher than those combination of cxmsted County, in* curosurgical exper- aut psy rate that >is of brain tumor 4 axial tomography Brain Tumor Mortalty 1357 and magnetic resonance imaging became available are in line with the reasons which are brought out in our paper. In conclusion, we are witnessing a com bined effect based on three components: more sophisticated diagnostic techniques made available to more recent birth cohorts; a change in the definition of and attitude toward elderly persons with a greater will ingness to diagnose them with less invasive procedures; and the introduction of the Medicare program that facilitates access to medical care. This has been a gradual pro cess with each advancing age group and decade and is still taking place among the oldest old. REFERENCES 1. Babcmuka M. Massey EW, Scttocnbcrg BS. Inter national mortality tram primary nervous system neoplasms: distribution and trends. Int J Epdemiot 1988:17:33-8. 2. Dsvis DL, Hoe! D. Fox J. et aL Imernarionii trends in cancer mortality in France, WestGermany. Italy, Japan. England and Wales, and the USA. Lancet 1990036:474-81. 3. Devix DL, Hoel D. Percy C. et at Is brain mortality increasing in industrial countries? In: Davis DL, Hoel D. eds. Trends in cancer mortality in indus trial countries. Aim N Y Acad Sri 1990^09: 191-204. 4. Greig NH. Rio LG, Yancik R, et al. Increasing' annual incidence of primary malignant brain tu mors in the elderly. J Nail Cancer Inst 1990: 82:1621-4. 5. Boyle P. Maisonneuvc P, Saracci R, ct al. Is the increased incidence of primary malignant brain tumors in the cklcriv real? J Nat) Cancer Inst 1990: 82:1594-6. 6. xhlbom A. Some notes on brain tumor epidfeiniology. Ann N Y And Sri 1990:609:179-85. 7. Doll. R. Progress against cancer an.epidcmiofogic assessment The 1991 John C. Casscl Memorial lecture. Am J Epidemiol 1991:134:675-88. National Center for Health Statistics. Vital Statis tics of the United States. 1987. Vol. II. Mortality, Pan a. Section 7. Technical Appendix. Washing ton DC; US Public Health Service. 1990. National Center for Health Statistics. Public use data tape documentation: compressed mortality file. 1968-85. Hyattsville, MD: National Center for Health Statistics, 1988. National Center for Health Statistics. Public use data tape documentation: multiple cause of death for ICD-9. Hvattsvilk, MD: National Center for Health Statistics. 1968-1988. National Center for Health Statistics. Detailed di agnoses and procedures for patients discharged from short-stay hospitals. United States. 1986. Washington DC: US GPO. 1988. (Vital and Health Statistics. Series 13: Data from the National Health Survey, no. 95) (DHHS publication no. (PHS) 881756). National Center for Health Statistics. Health. United States. >990. Hyattsville. MD: US GPO. 1991. (DHHS publication no, (PHS) 91-1232). Dolt R. Peto R. The causes ofcancer quantitative' estimates ofavoidable risks ofcancer in the United States. J Nail Cancer Inst 1981,66:1191-308. 14. Marshall E. Experts dash over cancer data. Soeoce 1990250:900-2. 15. Schwartz J. Multinational trends in cancer mortal tty rates. Methodological Issues and results. In: Davis DL Hoel D, eds. Trends in cancer mortality in industrial countries. Ann N Y Acad Set 1990; 609:136-45. 16. National Research Council .The aging population in the twenty-first century. Statistics for health pol icy. Washington DC National Academy Press. 199117. Kurland. LT. The frequency of intracranial and inuttspinal neoplasms in the resident population of Rochester. Minnesota. 3 Ncvrosurg 1958:15: 627-41. 18- Percy AK. Elvcback LR, Okazaki H. ct al. Neo plasms of the central nervous system: epidemio logic considerations. Neurology 197222:40-8. 19. Kurland LT. Schoenberg BS. Annepers JT\ aL The incidence of primary intracranial neoplasms in Rochester. Minnesota. 1935-1977. In: Selikoff IJ. Hammond EC. eds. Brain turnon in the chent: ical industry. Ann N Y Acad Sri 1982:38 i:6-l 6. CMA113933