- Email: [email protected]
Decreased incidence of cervical cancer in medicare-eligible California women1
Decreased Incidence of Cervical Cancer in Medicare-Eligible California Women Terri L. Cornelison, MD, PhD, Fredrick J. Montz, MD, KM, Robert E. Bristow, MD, Betty Chou, MD, Alessandro Bovicelli, MD, and Scott L. Zeger, PhD OBJECTIVE: To determine if the incidence of invasive cervical cancer relative to carcinoma in situ decreased in Medicare-eligible women. METHODS: A retrospective cohort was amassed from the California Cancer Registry database. The hypothesis was prospectively specified. Mean ratio of invasive (International Federation of Gynecology and Obstetrics Stages I–IV) to in situ cervical carcinoma in 1988 –1990 versus 1991–1995 was stratified by age (24 or younger, 25– 44, 45– 64, 65 or older) and race (all races, whites, blacks, Hispanics, Asian/Pacific Islanders). RESULTS: The mean ratio of invasive to in situ cervical cancer incidence for women at least 65 years old was lower in 1991–1995 compared with 1988 –1990 (P < .001, 95% confidence interval 0.893, 0.954); and had decreased more than observed for women aged 45– 64 and 25– 44, for all races combined, and for white women. The decreased ratio of invasive to in situ cancer for blacks, Hispanics, and Asian/Pacific Islanders at least 65 years old was no different than the decreased ratio in younger women. CONCLUSION: In California, in the 5 years after the 1990 change in Medicare funding statutes for cervical cytology screening, the ratio of invasive cervical cancer to in situ disease decreased more in Medicare-eligible patients than in younger women. (Obstet Gynecol 2002;100:79 – 86. © 2002 by The American College of Obstetricians and Gynecologists.)
The goal of cervical cytology screening is to identify precancerous cervical lesions that can be treated, and thus prevent development of invasive disease, and decrease disease-related mortality. From the time of its From The Kelly Gynecologic Oncology Service, Departments of Gynecology and Obstetrics, and Department of Biostatistics, The Johns Hopkins Medical Institutions, Baltimore, Maryland; and Division of Cancer Prevention, National Cancer Institutes, Bethesda, Maryland. This work was supported by the institutional research funds of The Johns Hopkins Medical Institutions. The authors acknowledge the efforts of the California Cancer Registry participants and staff and are extremely grateful for their sustained high-quality registry operations.
establishment in 1965 until July 1, 1990, Medicare did not pay for screening cervical cytology. In the United States, Medicare (Hospital Insurance Part A, or Supplementary Medical Insurance Part B) is the sole or dominant source of medical insurance coverage for approximately 39 million individuals.1 Because the elderly continue to have cervical cancer incidence rates that are statistically higher than the population as a whole,2 a change in Medicare funding to provide coverage for screening cervical cytology would potentially improve cervical cancer screening rate and secondarily decrease cervical cancer-related morbidity and mortality for many high-risk women. Medicare statutes were modified in 1990 such that, effective July 1 of that year, Medicare would help pay for cervical cytology performed to screen for cervical cancer. It was the goal of our investigation to determine if there was a decrease in the incidence of invasive cervical cancer relative to carcinoma in situ in Medicare-eligible women in the years after Medicare policy change. MATERIALS AND METHODS To test the hypothesis that the ratio of invasive to in situ cervical cancer incidence in elderly California women 1) decreased after 1990 and 2) decreased more than in younger women, data were reviewed from the State of California Cancer Registry. This registry was instituted in 1975, is fully funded by State of California tax dollars, and is administered by the California Department of Health Services, Cancer Surveillance Section. By mandate, all newly diagnosed cancer cases in the state of California are reported to the Central Registry. Reporting facilities include hospitals, physicians’ offices, nursing homes, pathology laboratories, ambulatory care facilities, and radiation and chemotherapy treatment centers within the geographic borders of the state.3 California Cancer Registry data from 1988 through 1995 were accessed. Data before these years were excluded because of revised abstraction guidelines in 1988. Data were stratified by patient age (24 or younger,
VOL. 100, NO. 1, JULY 2002 © 2002 by The American College of Obstetricians and Gynecologists. Published by Elsevier Science Inc.
0029-7844/02/$22.00 PII S0029-7844(02)02025-2
79
Table 1. Cervical Cancer Rates for Women of All Ages Year
All
White
1988 1989 1990 1991 1992 1993 1994 1995
11.2 11.2 11.8 10.9 11.5 10.9 10.8 10.2
9.8 9.7 10.5 9.3 9.8 8.9 8.7 8.6
Year
All
White
1988 1989 1990 1991 1992 1993 1994 1995
44.9 44.3 50.3 45.7 47.4 46.5 45.8 47.0
43.4 42.8 48.4 44.6 44.6 41.6 40.6 39.6
Year
All
1988 1989 1990 1991 1992 1993 1994 1995
0.25 0.25 0.24 0.24 0.24 0.23 0.24 0.22
A. Invasive cervical cancer Black 14.6 12.1 12.6 11.2 12.1 11.3 10.3 11.2 B. In situ cervical cancer Black 37.5 35.8 41.7 35.9 43.5 47.2 42.7 39.9
Hispanic
Asian/Pacific Islander
12.6 13.1 14.1 13.9 14.5 13.4 13.6 12.4
12.5 12.5 11.5 11.2 12.4 12.9 14.5 11.6
Hispanic
Asian/Pacific Islander
42.1 44.4 50.5 43.4 46.5 46.6 47.2 49.8
19.7 21.6 24.1 26.5 27.2 30.1 28.0 28.1
C. Invasive to in situ cervical cancer White Black Hispanic 0.23 0.23 0.22 0.21 0.22 0.22 0.22 0.22
0.39 0.34 0.30 0.31 0.28 0.24 0.24 0.28
0.30 0.30 0.28 0.32 0.31 0.29 0.29 0.25
Asian/Pacific Islander 0.64 0.58 0.48 0.42 0.46 0.43 0.52 0.41
Observed cervical cancer incidence rates (number of cases per 100,000) in California women from 1988 to 1995 remained relatively stable for all ethnic groups combined and within each ethnic group throughout the study interval, although there were remarkable differences between the different ethnic groups.
25– 44, 45– 64, 65 or older), ethnic group (all races, whites, blacks, Hispanics, Asian/Pacific Islanders), and stage of disease (carcinoma in situ, International Federation of Gynecology and Obstetrics Stages I–IV).4 This information was abstracted in the standard registry reporting form. Observed and fitted incidence rates of in situ and invasive (Stages I–IV) cervical cancer per year were assessed separately for each age group and ethnic group, and the ratio of invasive disease to in situ cervical cancer per year was calculated by age group and ethnic group to identify trends in the relative incidence rates. The cumulative mean ratio of invasive to in situ disease for the years before and including the enactment of the change in Medicare funding (1988 –1990) was compared with the cumulative mean ratio of invasive to in situ disease for the first 5 complete calendar years after the policy change (1991–1995). Ratio of the 1991–1995 invasive to in situ mean ratio versus the 1988 –1990 invasive to in situ mean ratio (ie, “ratio of the ratios”) for each of the
80
Cornelison et al
Medicare Cervical Screening
above identified subgroups was also calculated. Poisson regression5 was used to describe the expected rate of cervical cancer as a function of stage (Stages I–IV or in situ), year, race, and age, and to calculate the relative risk among subgroups. RESULTS When all age groups are combined, observed incidence rates of invasive cervical cancer (number of cases per 100,000) in California remained relatively stable for all ethnic groups combined and within each ethnic group throughout the study interval, although there were remarkable differences between the different ethnic groups (Table 1A). Invasive cervical cancer incidence decreased minimally for whites (range 8.6 –10.5 per 100,000) and blacks (range 10.3–14.6), and was more stable for Hispanics (range 12.4 –14.5) and Asian/Pacific Islanders (range 11.2–14.5). Similarly, rates of in situ cervical cancer by ethnic group did not fluctuate remarkably
OBSTETRICS & GYNECOLOGY
Table 2. Cervical Cancer Rates for Women Aged 65 or Older Year
All
White
1988 1989 1990 1991 1992 1993 1994 1995
20.8 20.0 22.1 16.7 21.3 18.0 17.9 17.2
15.6 15.1 17.1 11.3 15.6 12.8 11.7 11.7
Year
All
White
1988 1989 1990 1991 1992 1993 1994 1995
12.4 14.0 14.8 13.8 16.2 19.0 17.2 16.4
9.7 11.4 11.0 12.4 12.3 13.3 12.5 11.7
Year
All
1988 1989 1990 1991 1992 1993 1994 1995
1.68 1.43 1.50 1.21 1.32 0.95 1.04 1.05
A. Invasive cervical cancer Black 43.2 40.8 30.4 37.9 30.9 25.3 23.0 31.5 B. In situ cervical cancer Black 14.0 23.8 26.0 17.9 26.8 32.4 23.0 16.7
Hispanic
Asian/Pacific Islander
42.0 33.0 38.3 35.9 42.0 36.1 35.6 32.5
37.5 43.3 43.3 31.7 44.2 36.6 45.3 35.5
Hispanic
Asian/Pacific Islander
23.2 24.8 27.9 22.6 26.1 36.5 29.9 30.9
21.1 14.8 26.3 19.7 29.2 30.8 33.0 25.2
C. Invasive to in situ cervical cancer White Black Hispanic 1.61 1.33 1.56 0.91 1.27 0.96 0.94 0.99
3.08 1.71 1.58 2.12 1.15 0.78 1.00 1.88
Asian/Pacific Islander
1.81 1.33 1.37 1.59 1.61 0.99 1.19 1.05
1.78 2.93 1.64 1.61 1.51 1.19 1.38 1.41
For women aged 65 or older, observed cervical cancer incidence rates (number of cases per 100,000) decreased from 1988 to 1995 in California, with a concomitant increase in the rates of in situ disease.
during the period evaluated with a minimal rise for all races (range 44.3–50.3 per 100,000), blacks (range 35.8 – 47.2), and Hispanics (range 42.1–50.5), more of a rise in Asian/Pacific Islanders (range 19.7–30.1), and a slight decrease in whites (range 39.6 – 48.4) (Table 1B). Accordingly, the invasive to in situ ratio remained constant for all races combined (range 0.22– 0.25), whites (range 0.21– 0.23), blacks (range 0.24 – 0.39), Hispanics (range 0.25– 0.31), and fell minimally for Asian/Pacific Islanders (range 0.41– 0.64) (Table 1C). In contradistinction, for those women aged 65 and older there was a decrease in the rates of cervical cancer (range 22.1–16.7 per 100,000, Table 2A) and a concomitant increase in the rates of in situ disease (range 12.4 – 19.0, Table 2B). The ratio of invasive to in situ disease decreased (range 1.68 – 0.95, Table 2C). Restated, from 1988 to 1990, of every 100,000 women in California over the age of 65, on average, 21 were found to have invasive cervical cancer, and 14 women were identified as having in situ disease (Table 3). In the years 1991– 1995, more women (17) were found to have carcinoma
VOL. 100, NO. 1, JULY 2002
in situ, and fewer (18) invasive cancers were detected. The ratio of invasive to in situ disease decreased from an average of 1.5 to 1.1 (Figure 1). This decrease in the observed incidence of invasive to in situ disease for all women aged 65 and older was verified by more stringent calculations based on incidence rates fitted to the Poisson regression model. The cumulative mean of these ratios of invasive to in situ Table 3. The Mean Cervical Cancer Rate All ages Invasive In situ Invasive to in situ ⱖ65 y Invasive In situ Invasive to in situ
1988–1990
1991–1995
11 47 0.25
11 47 0.23
21 14 1.5
18 17 1.1
For 1988 –1990 and for 1991–1995 in California women of all ages and ages 65 or older. These calculations are based on observed incidence rates.
Cornelison et al
Medicare Cervical Screening
81
Figure 1. The ratio of observed invasive to in situ cervical cancer rate for 1988 –1995 in California women of all races for all ages and for ages 65 and older. The ratio of invasive to in situ disease decreased from an average of 1.5 in years 1988 –1990 to 1.1 in years 1991–1995. Cornelison. Medicare Cervical Screening. Obstet Gynecol 2002.
disease for years 1988 –1990 was compared with the cumulative mean ratio for 1991–1995. The ratio of mean ratios for 1991–1995 versus 1988 –1990 was less than one (0.93, P ⬍ .001, 95% confidence interval [CI] 0.89,
0.95), which indicated improvement in the detection of cervical cancer in its in situ stage after 1990 (Table 4). When women were stratified by race, this ratio of mean ratios remained less than one for whites and Hispanics as
Table 4. Ratio of the Mean Ratios: For All Age Groups Relative risk (95% CI) (P) All races White Black Hispanic Asian/Pacific Islander
Age ⱕ24 y
25–44 y
45–64 y
ⱖ65 y (Medicare eligible)
0.928 (0.859, 1.002) (.056) 1.078 (0.968, 1.202) (.172) 0.663 (0.482, 0.912) (.012) 0.791 (0.695, 0.900) (⬍.001) 1.084 (0.716, 1.643) (.702)
1.004 (0.923, 1.025) (.735) 1.024 (0.993, 1.055) (.128) 0.982 (0.908, 1.063) (.660) 0.955 (0.922, 0.990) (.012) 1.003 (0.934, 1.078) (.935)
0.964 (0.943, 0.986) (.001) 0.972 (0.938, 1.007) (.366) 1.044 (0.951, 1.145) (.366) 0.914 (0.873, 0.956) (⬍.001) 0.958 (0.892, 1.029) (.238)
0.923 (0.893, 0.954) (⬍.001) 0.926 (0.885, 0.968) (.001) 0.936 (0.830, 1.055) (.280) 0.936 (0.840, 0.966) (.003) 0.962 (0.874, 1.058) (.420)
CI ⫽ confidence interval. The ratio of the mean invasive to in situ cervical cancer rate for 1991–1995 vs the mean invasive to in situ cervical cancer rate for 1988 –1990 in California women for all age groups, based on fitted incidence rates. 95% CIs and P values are given. Ratios of statistical significance are in bold type. A ratio of less than one indicates improved detection of cervical cancer at the in situ stage.
82
Cornelison et al
Medicare Cervical Screening
OBSTETRICS & GYNECOLOGY
Table 5. Ratio of the Mean Ratios: Age 65 or Older vs Age 64 or Younger Relative risk (95% CI) (P) All races White Black Hispanic Asian/Pacific Islander
Age 65 or older vs age 24 or younger
Age 65 or older vs age 25–44
Age 65 or older vs age 45–64
0.995 (0.918, 1.080) (.912) 0.858 (0.766, 0.962) (.008) 1.412 (1.016, 1.963) (.040) 1.139 (0.988, 1.314) (.074) 0.887 (0.581, 1.353) (.577)
0.920 (0.889, 0.951) (⬍.001) 0.904 (0.864, 0.946) (⬍.001) 0.953 (0.884, 1.075) (.434) 0.942 (0.879, 1.012) (.102) 0.959 (0.865, 1.062) (.418)
0.958 (0.923, 0.993) (.020) 0.952 (0.906, 1.001) (.053) 0.867 (0.786, 1.023) (.105) 0.986 (0.914, 1.064) (.714) 1.004 (0.905, 1.113) (.942)
Abbreviation as in Table 4. The ratio of the mean invasive to in situ cervical cancer rate for 1991–1995 vs the mean invasive to in situ cervical cancer rate for 1988 –1990, comparing women aged 65 or older with women aged 64 or younger. These calculations are based on fitted incidence rates. 95% confidence intervals and P values are given. Ratios of statistical significance are in bold type. A ratio of less than one means that there was improved detection of cervical cancer at the in situ stage in women of Medicare-eligible age compared with women aged 64 or younger.
individual subgroups, 0.93 (P ⬍ .001, 95% CI 0.89, 0.96) and 0.94 (P ⬍ .003, 95% CI 0.84, 0.97), respectively. Improved detection of in situ disease was also seen for women aged 45– 64, black women aged less than 25, and Hispanic women of all age groups. To determine whether the decreased rate of invasive to in situ disease was greater in the Medicare-eligible patients than in younger patients, the ratio of the mean invasive to in situ cervical cancer rates for 1991–1995 versus the mean invasive to in situ cervical cancer rates for 1988 –1990 for women aged 65 and older was compared with the corresponding ratio of the mean ratios for women aged 64 and younger (Table 5). A resultant ratio of less than one meant that there was greater improvement (decrease) in the invasive to in situ cancer ratio in women of Medicare age than in women aged 64 or younger. When all ethnic groups were combined, Medicare-eligible women had greater decrease in invasive to in situ cancer rates than women aged 45– 64 (0.96, P ⬍ .02, 95% CI 0.93, 0.99) and women aged 25– 44 (0.92, P ⬍ .001, 95% CI 0.89, 0.95). When groups were stratified by ethnic group, this improvement was maintained for white women. Medicare-eligible blacks, Hispanics, and Asian/Pacific Islanders had no more improvement in the ratio of invasive to in situ disease than younger women. DISCUSSION Medicare is the federal government’s health insurance program for the elderly and disabled.6 Devised in 1965, the program was intended to insure the elderly against
VOL. 100, NO. 1, JULY 2002
debilitating hospital and physician expenses, and it was explicitly limited in the law to coverage of diagnostic and therapeutic short-term medical care. Medicare paid for diagnostic cervical cytology only if the patient was being treated for an existing gynecologic cancer or other disorder or showed signs of some abnormality.7 Extending Medicare coverage to preventive services required a congressional act.6 Cervical cytology screening was expected to cost Medicare 15 million dollars in fiscal year 1990 and another 30 million dollars in fiscal year 1991 for the anticipated 2.9 and 4.1 million claims to be filed under the new benefit in 1990 and 1991, respectively.7 Cervical cancer screening in elderly women has been controversial not only because of its potential cost to the Medicare program, but to its questioned effectiveness in preventing morbidity and mortality in this age group.6 Although cervical cytology screening effectiveness in general is undisputed, few studies have included any significant number of elderly women. In England, where screening decreased cervical cancer mortality by over 60%, the largest reduction was found in younger women with the least reduction found in women over 70 years.8 However, a screening program in Iceland showed that when 65% of women over age 60 had been screened at least once, the mortality fell by 60% in this age group.9,10 In addition, two case-control studies found that the risks associated with the lack of screening were higher for elderly than for younger women.6,11,12 One-quarter of new cases of invasive cervical cancer occur in women aged 65 and older, yet only 52% of women aged 65 and older have had cervical cytology within the past 3
Cornelison et al
Medicare Cervical Screening
83
years, and approximately 25% of elderly women have never been screened.7,13 Conversely, 91% of women aged 20 –39 and 73% of women aged 40 – 64 reported having cervical cytology during the past 3 years.6 Older women have a higher incidence of invasive cervical cancer, are more likely to have advanced disease at the time of diagnosis, and have slower declines in mortality rates.6 The relatively high incidence of invasive cancer in older women strengthens reasoning to continue screening at older ages, despite some reports of a decrease in the positive predictive value because the prevalence of cancer in situ drops rapidly after a peak at about age 35.14 More recently, in a selected part of a population not submitted to organized mass screening, the increased risk of cervical intraepithelial neoplasia (CIN) III in first cervical cytology smears was most pronounced in postmenopausal women (older than 50 years).15 Also, CIN III was most frequent in smears from women older than 70 years (0.4%), and all cases with malignancy were in smears from women older than 50 years. In this study, the risk of cytologically diagnosed CIN III increased with time since the previous smear. Postmenopausal women without previous smears had the highest risk of CIN III and cancer. When analyzing risk factors for cervical cancer, it becomes apparent that the elderly are at higher risk for the disease notably because they do not receive routine screening.16 The cost-effectiveness Markovian model used to support Medicare screening provisions predicted that screening elderly women would not necessarily save overall health care costs, but the costs per potential years of life saved from this intervention would be relatively low.6 Screening all elderly women every 3 years was estimated to save about 43,200 life-years per one million women screened, and to cost 2254 dollars for each year of life saved.6,7,13 This would reduce mortality from cervical cancer among the elderly by 74%.17 Annual screening would save an additional 6800 life-years per one million women screened with an incremental cost of 39,700 dollars for each additional year of life saved,13 increasing overall costs to 7345 dollars per year of life saved. The cost-effectiveness of every 3-year screening is comparable with that of vaccination against pneumococcal pneumonia in elderly women, whereas the costeffectiveness of annual cervical cytology is comparable with annual mammogram screening for breast cancer in this age group.13 As of July 1, 1990, all Medicare-eligible women were covered for cervical cytology screening tests up to every 3 years. The goal of our study was to determine if the incidence of invasive cervical cancer relative to in situ disease in Medicare-eligible women decreased subsequent
84
Cornelison et al
Medicare Cervical Screening
to Medicare statute changes providing coverage for cervical cytology screening. Because cytology detects cervical disease in preinvasive stages, more effective screening was expected to decrease the ratio of the number of invasive cancers versus in situ lesions. The annual ratio of invasive to in situ cervical cancer incidence decreased for women 65 years and older in the 5 years after enactment of the Medicare funding changes when compared with the preceding 3 years. Furthermore, this ratio decreased more for these women than for women aged 45– 64 and 25– 44. When our study population was stratified by race, white women were shown to have a greater decrease in invasive to in situ cancer rates than younger women. White women appeared to have benefited more than women in other ethnic groups. Medicare-eligible blacks, Hispanics, and Asian/Pacific Islanders had no more decrease in the ratio of invasive to in situ disease than younger women. There are several limitations of the present study. First, the study does not determine if Medicare coverage actually caused increased screening in elderly women. It is unknown if statute changes prompted more Medicare patients to request screening or more health care providers to perform cervical cytology when Medicare patients presented for routine or acute care visits. Although both patient- and clinician-initiated screening may have resulted from better coverage, direct causality has not been established. Moreover, better distinction could identify whether the patient or the health care provider is the better target for further improvement of screening efforts. Second, it is not known what percentage of screening Papanicolaou smears in Medicare-eligible women were in fact paid for by Medicare. This study does not address what percentage of screened women had Medicare coverage, or used Medicare as their payment method if they had more than one medical coverage source. It may be possible to get these data from linkage analysis with the Medicare database. Third, the number of women screened is unknown, and therefore it cannot be determined from these data whether more Medicareeligible women were screened in 1991–1995 than in 1988 –1990, or if there were proportionately more elderly women screened than younger women during the same periods. Fourth, this study groups women 65 years and older homogeneously, when other data have demonstrated the heterogeneity of the older adult cohort. There may be a need to report data by smaller age intervals than the 65 years and over category that is commonly used in health services research.18 Sociodemographic, health status, and health belief variables are associated with older women’s use of cancer screening procedures, and at older ages, women display a reduced
OBSTETRICS & GYNECOLOGY
use of these techniques. Fifth, it is not known if cervical cytology accuracy in this study was different between elderly and younger women or among elderly women. This information may help tailor screening more effectively to specific ages within the Medicare-eligible cohort. Sixth, our analysis does not capture information on other high-grade squamous intraepithelial lesions, namely CIN II and CIN III, which would have been ascertained with better access to screening. Capturing these data may be reflected in a further decrease in invasive disease over a greater time. And seventh, this study does not give survival information. The purpose of a screening test is to detect cancer at an earlier stage and improve survival. An extended time delay is expected to observe actual survival; however, projected survival benefit may be calculated. Decreased incidence of invasive cancer may not necessarily translate into improved survival. Screening tests may produce a shift to earlier stages at diagnosis without any improvement in survival because of the length and lead time bias effects.19,20 There were generalized inconsistencies in cervical cytology readings from 1988 to 1995 in California. These inconsistencies did not follow a generalized trend, were not related to patient age or ethnicity, occurred broadly across the state, and were largely attributable to the 1988 Bethesda System reclassification of Papanicolaou smears. First, after the reclassification, there was confusion converting from the old nomenclature to the new nomenclature. Second, it became recognized that cytologists and pathologists had difficulty reproducibly distinguishing moderate dysplasia from severe dysplasia on cervical cytology and CIN II from CIN III from in situ disease on pathology specimens, respectively. Cytologists and pathologists began to modify their slide interpretation to improve and standardize their readings. Compounded upon these inconsistencies, the California Cancer Registry had just begun to collect cervical cytology data in 1987, and there was a 1–2-year learning curve for consistent reporting and data quality. In response to these inconsistencies, the California Cancer Registry discontinued collection of CIN III and in situ data from all physician reportable cases (ie, from physicians’ offices and ambulatory care facilities) in 1992, and in 1996 discontinued collection of CIN III and in situ data from all reportable sources. Therefore, the actual ratios of mean invasive to in situ cervical cancer rates for 1991–1995 in California were probably lower than the ratios reported in this manuscript. In Figure 1, the slope of decrease in the ratio of invasive to in situ disease in women aged 65 and older does appear to be the same from 1988 through 1993, and there may be other effects to cause this relative decrease. It is possible that before 1990 elderly women were al-
VOL. 100, NO. 1, JULY 2002
ready seeking better screening opportunities for their health care in general with breast cancer and other health initiatives. It is also possible that other behavior modifications were already in effect such as decreased smoking or healthier sexual behaviors. However, this graph does not take into account the physician-reportable in situ cases no longer collected by the California Cancer Registry from 1992 to 1995, which would have decreased the slope for these years. Also, this graph only represents a limited time frame. Having the data to extend the graph before 1988 and after 1995 would enable better interpretation of the significance of changes seen in the immediate years after 1990. However, despite limitations of this study, there is indication that cervical cancer incidence decreased more in Medicare-eligible patients in the 5 years after the 1990 change in Medicare funding statutes for cervical cytology screening, and that cervical cancer incidence decreased more in Medicare-eligible patients than in younger women during this period. These observations support the ideology of successful cervical cancer screening in elderly women. Additional efforts are needed to ensure that older nonwhite women are better screened. REFERENCES 1. Health Care Financing Administration, Medicare. Available at http://www.hcfa.gov/medicare/medicare.htm. Accessed 2001 Feb 6. 2. Herrero R. Epidemiology of cervical cancer. J Natl Cancer Inst Monogr 1996;21:1– 6. 3. Swan J, Wingo P, Clive R, West D, Miller D, Hutchison C, et al. Cancer surveillance in the U.S. Cancer 1998;83: 1282–91. 4. Benedet JL, Bender H, Jones H III, Ngan HY, Pecorelli S. FIGO staging classifications and clinical practice guidelines in the management of gynecologic cancers. FIGO Committee on Gynecologic Oncology. Int J Gynaecol Obstet 2000;70:209 – 62. 5. Clayton D, Hills D. Statistical models in epidemiology. Oxford, England: Oxford University Press, 1993. 6. Power EJ. Pap smears, elderly women, and Medicare. Cancer Invest 1993;11:164 – 8. 7. Medicare to help pay for cervical cancer screening every 3 years. JNCI 1990;82:1242–3. 8. Sasieni P, Adams J. Effect of screening on cervical cancer mortality in England and Wales: Analysis of trends with an age period cohort model. BMJ 1999;318:1244 –5. 9. Fletcher A. Screening for cancer of the cervix in elderly women. Lancet 1990;335:97–9. 10. Johannesson G, Geirsson G, Day N, Tulinius H. Screening for cancer of the uterine cervix in Iceland, 1965–1978. Acta Obstet Gynecol Scand 1982;61:199 –203. 11. Aristizabal N, Cuello C, Correa P, Collazos T, Haenszel
Cornelison et al
Medicare Cervical Screening
85
12.
13.
14.
15.
16.
86
W. The impact of vaginal cytology on cervical cancer risks in Cali, Colombia. Int J Cancer 1984;34:5–9. Clarke EA, Anderson TW. Does screening by “Pap” smears help prevent cervical cancer? A case-control study. Lancet 1979;2:1– 4. Muller C, Mandelblatt J, Schechter CB, Power EJ, Duffy BM, Wagner JL. U.S. Congress, Office of Technology Assessment. Costs and effectiveness of cervical cancer screening in elderly women. Feb. background paper no. OTA-BP-H-65. Washington, DC: US Government Printing Office, 1990. Gustafsson L, Spare´n P, Gustafsson M, Pettersson B, Wilander E, Bergstro¨m R. Low efficiency of cytologic screening for cancer in situ of the cervix in older women. Int J Cancer 1995;63:804 –9. Forsmo S, Jacobsen BK, Stalsberg H. Cervical neoplasia in Pap smears: Risk of cervical intra-epithelial neoplasia (CIN) after negative or no prior smears in a population without a mass screening programme. Int J Epidemiol 1996;25:53–7. Celentano DD, Klassen AC, Weisman CS, Rosenshein NB. Cervical cancer screening practices among older
Cornelison et al
Medicare Cervical Screening
17.
18.
19.
20.
women: Results from the Maryland cervical cancer casecontrol study. J Clin Epidemiol 1988;41:531– 41. Fahs MC, Mandelblatt J, Schechter C, Muller C. Cost effectiveness of cervical cancer screening for the elderly. Ann Intern Med 1992;117:520 –7. Ruchlin HS. Prevalence and correlates of breast and cervical cancer screening among older women. Obstet Gynecol 1997;90:16 –21. Riley GF, Potosky AL, Lubitz JD, Brown ML. Stage of cancer at diagnosis for Medicare HMO and fee-for-service enrollees. Am J Public Health 1994;84:1598 – 604. Prorok PC, Hankey BF, Bundy BN. Concepts and problems in the evaluation of screening programs. J Chronic Dis 1981;34:159 –71.
Address reprint requests to: Fredrick J. Montz, MD, KM, The Johns Hopkins Medical Institutions, Departments of Gynecology and Obstetrics, 600 North Wolfe Street, Houck 248, Baltimore, MD 21210; E-mail: [email protected]. Received August 28, 2001. Received in revised form December 11, 2001. Accepted January 10, 2002.
OBSTETRICS & GYNECOLOGY
The goal of cervical cytology screening is to identify precancerous cervical lesions that can be treated, and thus prevent development of invasive disease, and decrease disease-related mortality. From the time of its From The Kelly Gynecologic Oncology Service, Departments of Gynecology and Obstetrics, and Department of Biostatistics, The Johns Hopkins Medical Institutions, Baltimore, Maryland; and Division of Cancer Prevention, National Cancer Institutes, Bethesda, Maryland. This work was supported by the institutional research funds of The Johns Hopkins Medical Institutions. The authors acknowledge the efforts of the California Cancer Registry participants and staff and are extremely grateful for their sustained high-quality registry operations.
establishment in 1965 until July 1, 1990, Medicare did not pay for screening cervical cytology. In the United States, Medicare (Hospital Insurance Part A, or Supplementary Medical Insurance Part B) is the sole or dominant source of medical insurance coverage for approximately 39 million individuals.1 Because the elderly continue to have cervical cancer incidence rates that are statistically higher than the population as a whole,2 a change in Medicare funding to provide coverage for screening cervical cytology would potentially improve cervical cancer screening rate and secondarily decrease cervical cancer-related morbidity and mortality for many high-risk women. Medicare statutes were modified in 1990 such that, effective July 1 of that year, Medicare would help pay for cervical cytology performed to screen for cervical cancer. It was the goal of our investigation to determine if there was a decrease in the incidence of invasive cervical cancer relative to carcinoma in situ in Medicare-eligible women in the years after Medicare policy change. MATERIALS AND METHODS To test the hypothesis that the ratio of invasive to in situ cervical cancer incidence in elderly California women 1) decreased after 1990 and 2) decreased more than in younger women, data were reviewed from the State of California Cancer Registry. This registry was instituted in 1975, is fully funded by State of California tax dollars, and is administered by the California Department of Health Services, Cancer Surveillance Section. By mandate, all newly diagnosed cancer cases in the state of California are reported to the Central Registry. Reporting facilities include hospitals, physicians’ offices, nursing homes, pathology laboratories, ambulatory care facilities, and radiation and chemotherapy treatment centers within the geographic borders of the state.3 California Cancer Registry data from 1988 through 1995 were accessed. Data before these years were excluded because of revised abstraction guidelines in 1988. Data were stratified by patient age (24 or younger,
VOL. 100, NO. 1, JULY 2002 © 2002 by The American College of Obstetricians and Gynecologists. Published by Elsevier Science Inc.
0029-7844/02/$22.00 PII S0029-7844(02)02025-2
79
Table 1. Cervical Cancer Rates for Women of All Ages Year
All
White
1988 1989 1990 1991 1992 1993 1994 1995
11.2 11.2 11.8 10.9 11.5 10.9 10.8 10.2
9.8 9.7 10.5 9.3 9.8 8.9 8.7 8.6
Year
All
White
1988 1989 1990 1991 1992 1993 1994 1995
44.9 44.3 50.3 45.7 47.4 46.5 45.8 47.0
43.4 42.8 48.4 44.6 44.6 41.6 40.6 39.6
Year
All
1988 1989 1990 1991 1992 1993 1994 1995
0.25 0.25 0.24 0.24 0.24 0.23 0.24 0.22
A. Invasive cervical cancer Black 14.6 12.1 12.6 11.2 12.1 11.3 10.3 11.2 B. In situ cervical cancer Black 37.5 35.8 41.7 35.9 43.5 47.2 42.7 39.9
Hispanic
Asian/Pacific Islander
12.6 13.1 14.1 13.9 14.5 13.4 13.6 12.4
12.5 12.5 11.5 11.2 12.4 12.9 14.5 11.6
Hispanic
Asian/Pacific Islander
42.1 44.4 50.5 43.4 46.5 46.6 47.2 49.8
19.7 21.6 24.1 26.5 27.2 30.1 28.0 28.1
C. Invasive to in situ cervical cancer White Black Hispanic 0.23 0.23 0.22 0.21 0.22 0.22 0.22 0.22
0.39 0.34 0.30 0.31 0.28 0.24 0.24 0.28
0.30 0.30 0.28 0.32 0.31 0.29 0.29 0.25
Asian/Pacific Islander 0.64 0.58 0.48 0.42 0.46 0.43 0.52 0.41
Observed cervical cancer incidence rates (number of cases per 100,000) in California women from 1988 to 1995 remained relatively stable for all ethnic groups combined and within each ethnic group throughout the study interval, although there were remarkable differences between the different ethnic groups.
25– 44, 45– 64, 65 or older), ethnic group (all races, whites, blacks, Hispanics, Asian/Pacific Islanders), and stage of disease (carcinoma in situ, International Federation of Gynecology and Obstetrics Stages I–IV).4 This information was abstracted in the standard registry reporting form. Observed and fitted incidence rates of in situ and invasive (Stages I–IV) cervical cancer per year were assessed separately for each age group and ethnic group, and the ratio of invasive disease to in situ cervical cancer per year was calculated by age group and ethnic group to identify trends in the relative incidence rates. The cumulative mean ratio of invasive to in situ disease for the years before and including the enactment of the change in Medicare funding (1988 –1990) was compared with the cumulative mean ratio of invasive to in situ disease for the first 5 complete calendar years after the policy change (1991–1995). Ratio of the 1991–1995 invasive to in situ mean ratio versus the 1988 –1990 invasive to in situ mean ratio (ie, “ratio of the ratios”) for each of the
80
Cornelison et al
Medicare Cervical Screening
above identified subgroups was also calculated. Poisson regression5 was used to describe the expected rate of cervical cancer as a function of stage (Stages I–IV or in situ), year, race, and age, and to calculate the relative risk among subgroups. RESULTS When all age groups are combined, observed incidence rates of invasive cervical cancer (number of cases per 100,000) in California remained relatively stable for all ethnic groups combined and within each ethnic group throughout the study interval, although there were remarkable differences between the different ethnic groups (Table 1A). Invasive cervical cancer incidence decreased minimally for whites (range 8.6 –10.5 per 100,000) and blacks (range 10.3–14.6), and was more stable for Hispanics (range 12.4 –14.5) and Asian/Pacific Islanders (range 11.2–14.5). Similarly, rates of in situ cervical cancer by ethnic group did not fluctuate remarkably
OBSTETRICS & GYNECOLOGY
Table 2. Cervical Cancer Rates for Women Aged 65 or Older Year
All
White
1988 1989 1990 1991 1992 1993 1994 1995
20.8 20.0 22.1 16.7 21.3 18.0 17.9 17.2
15.6 15.1 17.1 11.3 15.6 12.8 11.7 11.7
Year
All
White
1988 1989 1990 1991 1992 1993 1994 1995
12.4 14.0 14.8 13.8 16.2 19.0 17.2 16.4
9.7 11.4 11.0 12.4 12.3 13.3 12.5 11.7
Year
All
1988 1989 1990 1991 1992 1993 1994 1995
1.68 1.43 1.50 1.21 1.32 0.95 1.04 1.05
A. Invasive cervical cancer Black 43.2 40.8 30.4 37.9 30.9 25.3 23.0 31.5 B. In situ cervical cancer Black 14.0 23.8 26.0 17.9 26.8 32.4 23.0 16.7
Hispanic
Asian/Pacific Islander
42.0 33.0 38.3 35.9 42.0 36.1 35.6 32.5
37.5 43.3 43.3 31.7 44.2 36.6 45.3 35.5
Hispanic
Asian/Pacific Islander
23.2 24.8 27.9 22.6 26.1 36.5 29.9 30.9
21.1 14.8 26.3 19.7 29.2 30.8 33.0 25.2
C. Invasive to in situ cervical cancer White Black Hispanic 1.61 1.33 1.56 0.91 1.27 0.96 0.94 0.99
3.08 1.71 1.58 2.12 1.15 0.78 1.00 1.88
Asian/Pacific Islander
1.81 1.33 1.37 1.59 1.61 0.99 1.19 1.05
1.78 2.93 1.64 1.61 1.51 1.19 1.38 1.41
For women aged 65 or older, observed cervical cancer incidence rates (number of cases per 100,000) decreased from 1988 to 1995 in California, with a concomitant increase in the rates of in situ disease.
during the period evaluated with a minimal rise for all races (range 44.3–50.3 per 100,000), blacks (range 35.8 – 47.2), and Hispanics (range 42.1–50.5), more of a rise in Asian/Pacific Islanders (range 19.7–30.1), and a slight decrease in whites (range 39.6 – 48.4) (Table 1B). Accordingly, the invasive to in situ ratio remained constant for all races combined (range 0.22– 0.25), whites (range 0.21– 0.23), blacks (range 0.24 – 0.39), Hispanics (range 0.25– 0.31), and fell minimally for Asian/Pacific Islanders (range 0.41– 0.64) (Table 1C). In contradistinction, for those women aged 65 and older there was a decrease in the rates of cervical cancer (range 22.1–16.7 per 100,000, Table 2A) and a concomitant increase in the rates of in situ disease (range 12.4 – 19.0, Table 2B). The ratio of invasive to in situ disease decreased (range 1.68 – 0.95, Table 2C). Restated, from 1988 to 1990, of every 100,000 women in California over the age of 65, on average, 21 were found to have invasive cervical cancer, and 14 women were identified as having in situ disease (Table 3). In the years 1991– 1995, more women (17) were found to have carcinoma
VOL. 100, NO. 1, JULY 2002
in situ, and fewer (18) invasive cancers were detected. The ratio of invasive to in situ disease decreased from an average of 1.5 to 1.1 (Figure 1). This decrease in the observed incidence of invasive to in situ disease for all women aged 65 and older was verified by more stringent calculations based on incidence rates fitted to the Poisson regression model. The cumulative mean of these ratios of invasive to in situ Table 3. The Mean Cervical Cancer Rate All ages Invasive In situ Invasive to in situ ⱖ65 y Invasive In situ Invasive to in situ
1988–1990
1991–1995
11 47 0.25
11 47 0.23
21 14 1.5
18 17 1.1
For 1988 –1990 and for 1991–1995 in California women of all ages and ages 65 or older. These calculations are based on observed incidence rates.
Cornelison et al
Medicare Cervical Screening
81
Figure 1. The ratio of observed invasive to in situ cervical cancer rate for 1988 –1995 in California women of all races for all ages and for ages 65 and older. The ratio of invasive to in situ disease decreased from an average of 1.5 in years 1988 –1990 to 1.1 in years 1991–1995. Cornelison. Medicare Cervical Screening. Obstet Gynecol 2002.
disease for years 1988 –1990 was compared with the cumulative mean ratio for 1991–1995. The ratio of mean ratios for 1991–1995 versus 1988 –1990 was less than one (0.93, P ⬍ .001, 95% confidence interval [CI] 0.89,
0.95), which indicated improvement in the detection of cervical cancer in its in situ stage after 1990 (Table 4). When women were stratified by race, this ratio of mean ratios remained less than one for whites and Hispanics as
Table 4. Ratio of the Mean Ratios: For All Age Groups Relative risk (95% CI) (P) All races White Black Hispanic Asian/Pacific Islander
Age ⱕ24 y
25–44 y
45–64 y
ⱖ65 y (Medicare eligible)
0.928 (0.859, 1.002) (.056) 1.078 (0.968, 1.202) (.172) 0.663 (0.482, 0.912) (.012) 0.791 (0.695, 0.900) (⬍.001) 1.084 (0.716, 1.643) (.702)
1.004 (0.923, 1.025) (.735) 1.024 (0.993, 1.055) (.128) 0.982 (0.908, 1.063) (.660) 0.955 (0.922, 0.990) (.012) 1.003 (0.934, 1.078) (.935)
0.964 (0.943, 0.986) (.001) 0.972 (0.938, 1.007) (.366) 1.044 (0.951, 1.145) (.366) 0.914 (0.873, 0.956) (⬍.001) 0.958 (0.892, 1.029) (.238)
0.923 (0.893, 0.954) (⬍.001) 0.926 (0.885, 0.968) (.001) 0.936 (0.830, 1.055) (.280) 0.936 (0.840, 0.966) (.003) 0.962 (0.874, 1.058) (.420)
CI ⫽ confidence interval. The ratio of the mean invasive to in situ cervical cancer rate for 1991–1995 vs the mean invasive to in situ cervical cancer rate for 1988 –1990 in California women for all age groups, based on fitted incidence rates. 95% CIs and P values are given. Ratios of statistical significance are in bold type. A ratio of less than one indicates improved detection of cervical cancer at the in situ stage.
82
Cornelison et al
Medicare Cervical Screening
OBSTETRICS & GYNECOLOGY
Table 5. Ratio of the Mean Ratios: Age 65 or Older vs Age 64 or Younger Relative risk (95% CI) (P) All races White Black Hispanic Asian/Pacific Islander
Age 65 or older vs age 24 or younger
Age 65 or older vs age 25–44
Age 65 or older vs age 45–64
0.995 (0.918, 1.080) (.912) 0.858 (0.766, 0.962) (.008) 1.412 (1.016, 1.963) (.040) 1.139 (0.988, 1.314) (.074) 0.887 (0.581, 1.353) (.577)
0.920 (0.889, 0.951) (⬍.001) 0.904 (0.864, 0.946) (⬍.001) 0.953 (0.884, 1.075) (.434) 0.942 (0.879, 1.012) (.102) 0.959 (0.865, 1.062) (.418)
0.958 (0.923, 0.993) (.020) 0.952 (0.906, 1.001) (.053) 0.867 (0.786, 1.023) (.105) 0.986 (0.914, 1.064) (.714) 1.004 (0.905, 1.113) (.942)
Abbreviation as in Table 4. The ratio of the mean invasive to in situ cervical cancer rate for 1991–1995 vs the mean invasive to in situ cervical cancer rate for 1988 –1990, comparing women aged 65 or older with women aged 64 or younger. These calculations are based on fitted incidence rates. 95% confidence intervals and P values are given. Ratios of statistical significance are in bold type. A ratio of less than one means that there was improved detection of cervical cancer at the in situ stage in women of Medicare-eligible age compared with women aged 64 or younger.
individual subgroups, 0.93 (P ⬍ .001, 95% CI 0.89, 0.96) and 0.94 (P ⬍ .003, 95% CI 0.84, 0.97), respectively. Improved detection of in situ disease was also seen for women aged 45– 64, black women aged less than 25, and Hispanic women of all age groups. To determine whether the decreased rate of invasive to in situ disease was greater in the Medicare-eligible patients than in younger patients, the ratio of the mean invasive to in situ cervical cancer rates for 1991–1995 versus the mean invasive to in situ cervical cancer rates for 1988 –1990 for women aged 65 and older was compared with the corresponding ratio of the mean ratios for women aged 64 and younger (Table 5). A resultant ratio of less than one meant that there was greater improvement (decrease) in the invasive to in situ cancer ratio in women of Medicare age than in women aged 64 or younger. When all ethnic groups were combined, Medicare-eligible women had greater decrease in invasive to in situ cancer rates than women aged 45– 64 (0.96, P ⬍ .02, 95% CI 0.93, 0.99) and women aged 25– 44 (0.92, P ⬍ .001, 95% CI 0.89, 0.95). When groups were stratified by ethnic group, this improvement was maintained for white women. Medicare-eligible blacks, Hispanics, and Asian/Pacific Islanders had no more improvement in the ratio of invasive to in situ disease than younger women. DISCUSSION Medicare is the federal government’s health insurance program for the elderly and disabled.6 Devised in 1965, the program was intended to insure the elderly against
VOL. 100, NO. 1, JULY 2002
debilitating hospital and physician expenses, and it was explicitly limited in the law to coverage of diagnostic and therapeutic short-term medical care. Medicare paid for diagnostic cervical cytology only if the patient was being treated for an existing gynecologic cancer or other disorder or showed signs of some abnormality.7 Extending Medicare coverage to preventive services required a congressional act.6 Cervical cytology screening was expected to cost Medicare 15 million dollars in fiscal year 1990 and another 30 million dollars in fiscal year 1991 for the anticipated 2.9 and 4.1 million claims to be filed under the new benefit in 1990 and 1991, respectively.7 Cervical cancer screening in elderly women has been controversial not only because of its potential cost to the Medicare program, but to its questioned effectiveness in preventing morbidity and mortality in this age group.6 Although cervical cytology screening effectiveness in general is undisputed, few studies have included any significant number of elderly women. In England, where screening decreased cervical cancer mortality by over 60%, the largest reduction was found in younger women with the least reduction found in women over 70 years.8 However, a screening program in Iceland showed that when 65% of women over age 60 had been screened at least once, the mortality fell by 60% in this age group.9,10 In addition, two case-control studies found that the risks associated with the lack of screening were higher for elderly than for younger women.6,11,12 One-quarter of new cases of invasive cervical cancer occur in women aged 65 and older, yet only 52% of women aged 65 and older have had cervical cytology within the past 3
Cornelison et al
Medicare Cervical Screening
83
years, and approximately 25% of elderly women have never been screened.7,13 Conversely, 91% of women aged 20 –39 and 73% of women aged 40 – 64 reported having cervical cytology during the past 3 years.6 Older women have a higher incidence of invasive cervical cancer, are more likely to have advanced disease at the time of diagnosis, and have slower declines in mortality rates.6 The relatively high incidence of invasive cancer in older women strengthens reasoning to continue screening at older ages, despite some reports of a decrease in the positive predictive value because the prevalence of cancer in situ drops rapidly after a peak at about age 35.14 More recently, in a selected part of a population not submitted to organized mass screening, the increased risk of cervical intraepithelial neoplasia (CIN) III in first cervical cytology smears was most pronounced in postmenopausal women (older than 50 years).15 Also, CIN III was most frequent in smears from women older than 70 years (0.4%), and all cases with malignancy were in smears from women older than 50 years. In this study, the risk of cytologically diagnosed CIN III increased with time since the previous smear. Postmenopausal women without previous smears had the highest risk of CIN III and cancer. When analyzing risk factors for cervical cancer, it becomes apparent that the elderly are at higher risk for the disease notably because they do not receive routine screening.16 The cost-effectiveness Markovian model used to support Medicare screening provisions predicted that screening elderly women would not necessarily save overall health care costs, but the costs per potential years of life saved from this intervention would be relatively low.6 Screening all elderly women every 3 years was estimated to save about 43,200 life-years per one million women screened, and to cost 2254 dollars for each year of life saved.6,7,13 This would reduce mortality from cervical cancer among the elderly by 74%.17 Annual screening would save an additional 6800 life-years per one million women screened with an incremental cost of 39,700 dollars for each additional year of life saved,13 increasing overall costs to 7345 dollars per year of life saved. The cost-effectiveness of every 3-year screening is comparable with that of vaccination against pneumococcal pneumonia in elderly women, whereas the costeffectiveness of annual cervical cytology is comparable with annual mammogram screening for breast cancer in this age group.13 As of July 1, 1990, all Medicare-eligible women were covered for cervical cytology screening tests up to every 3 years. The goal of our study was to determine if the incidence of invasive cervical cancer relative to in situ disease in Medicare-eligible women decreased subsequent
84
Cornelison et al
Medicare Cervical Screening
to Medicare statute changes providing coverage for cervical cytology screening. Because cytology detects cervical disease in preinvasive stages, more effective screening was expected to decrease the ratio of the number of invasive cancers versus in situ lesions. The annual ratio of invasive to in situ cervical cancer incidence decreased for women 65 years and older in the 5 years after enactment of the Medicare funding changes when compared with the preceding 3 years. Furthermore, this ratio decreased more for these women than for women aged 45– 64 and 25– 44. When our study population was stratified by race, white women were shown to have a greater decrease in invasive to in situ cancer rates than younger women. White women appeared to have benefited more than women in other ethnic groups. Medicare-eligible blacks, Hispanics, and Asian/Pacific Islanders had no more decrease in the ratio of invasive to in situ disease than younger women. There are several limitations of the present study. First, the study does not determine if Medicare coverage actually caused increased screening in elderly women. It is unknown if statute changes prompted more Medicare patients to request screening or more health care providers to perform cervical cytology when Medicare patients presented for routine or acute care visits. Although both patient- and clinician-initiated screening may have resulted from better coverage, direct causality has not been established. Moreover, better distinction could identify whether the patient or the health care provider is the better target for further improvement of screening efforts. Second, it is not known what percentage of screening Papanicolaou smears in Medicare-eligible women were in fact paid for by Medicare. This study does not address what percentage of screened women had Medicare coverage, or used Medicare as their payment method if they had more than one medical coverage source. It may be possible to get these data from linkage analysis with the Medicare database. Third, the number of women screened is unknown, and therefore it cannot be determined from these data whether more Medicareeligible women were screened in 1991–1995 than in 1988 –1990, or if there were proportionately more elderly women screened than younger women during the same periods. Fourth, this study groups women 65 years and older homogeneously, when other data have demonstrated the heterogeneity of the older adult cohort. There may be a need to report data by smaller age intervals than the 65 years and over category that is commonly used in health services research.18 Sociodemographic, health status, and health belief variables are associated with older women’s use of cancer screening procedures, and at older ages, women display a reduced
OBSTETRICS & GYNECOLOGY
use of these techniques. Fifth, it is not known if cervical cytology accuracy in this study was different between elderly and younger women or among elderly women. This information may help tailor screening more effectively to specific ages within the Medicare-eligible cohort. Sixth, our analysis does not capture information on other high-grade squamous intraepithelial lesions, namely CIN II and CIN III, which would have been ascertained with better access to screening. Capturing these data may be reflected in a further decrease in invasive disease over a greater time. And seventh, this study does not give survival information. The purpose of a screening test is to detect cancer at an earlier stage and improve survival. An extended time delay is expected to observe actual survival; however, projected survival benefit may be calculated. Decreased incidence of invasive cancer may not necessarily translate into improved survival. Screening tests may produce a shift to earlier stages at diagnosis without any improvement in survival because of the length and lead time bias effects.19,20 There were generalized inconsistencies in cervical cytology readings from 1988 to 1995 in California. These inconsistencies did not follow a generalized trend, were not related to patient age or ethnicity, occurred broadly across the state, and were largely attributable to the 1988 Bethesda System reclassification of Papanicolaou smears. First, after the reclassification, there was confusion converting from the old nomenclature to the new nomenclature. Second, it became recognized that cytologists and pathologists had difficulty reproducibly distinguishing moderate dysplasia from severe dysplasia on cervical cytology and CIN II from CIN III from in situ disease on pathology specimens, respectively. Cytologists and pathologists began to modify their slide interpretation to improve and standardize their readings. Compounded upon these inconsistencies, the California Cancer Registry had just begun to collect cervical cytology data in 1987, and there was a 1–2-year learning curve for consistent reporting and data quality. In response to these inconsistencies, the California Cancer Registry discontinued collection of CIN III and in situ data from all physician reportable cases (ie, from physicians’ offices and ambulatory care facilities) in 1992, and in 1996 discontinued collection of CIN III and in situ data from all reportable sources. Therefore, the actual ratios of mean invasive to in situ cervical cancer rates for 1991–1995 in California were probably lower than the ratios reported in this manuscript. In Figure 1, the slope of decrease in the ratio of invasive to in situ disease in women aged 65 and older does appear to be the same from 1988 through 1993, and there may be other effects to cause this relative decrease. It is possible that before 1990 elderly women were al-
VOL. 100, NO. 1, JULY 2002
ready seeking better screening opportunities for their health care in general with breast cancer and other health initiatives. It is also possible that other behavior modifications were already in effect such as decreased smoking or healthier sexual behaviors. However, this graph does not take into account the physician-reportable in situ cases no longer collected by the California Cancer Registry from 1992 to 1995, which would have decreased the slope for these years. Also, this graph only represents a limited time frame. Having the data to extend the graph before 1988 and after 1995 would enable better interpretation of the significance of changes seen in the immediate years after 1990. However, despite limitations of this study, there is indication that cervical cancer incidence decreased more in Medicare-eligible patients in the 5 years after the 1990 change in Medicare funding statutes for cervical cytology screening, and that cervical cancer incidence decreased more in Medicare-eligible patients than in younger women during this period. These observations support the ideology of successful cervical cancer screening in elderly women. Additional efforts are needed to ensure that older nonwhite women are better screened. REFERENCES 1. Health Care Financing Administration, Medicare. Available at http://www.hcfa.gov/medicare/medicare.htm. Accessed 2001 Feb 6. 2. Herrero R. Epidemiology of cervical cancer. J Natl Cancer Inst Monogr 1996;21:1– 6. 3. Swan J, Wingo P, Clive R, West D, Miller D, Hutchison C, et al. Cancer surveillance in the U.S. Cancer 1998;83: 1282–91. 4. Benedet JL, Bender H, Jones H III, Ngan HY, Pecorelli S. FIGO staging classifications and clinical practice guidelines in the management of gynecologic cancers. FIGO Committee on Gynecologic Oncology. Int J Gynaecol Obstet 2000;70:209 – 62. 5. Clayton D, Hills D. Statistical models in epidemiology. Oxford, England: Oxford University Press, 1993. 6. Power EJ. Pap smears, elderly women, and Medicare. Cancer Invest 1993;11:164 – 8. 7. Medicare to help pay for cervical cancer screening every 3 years. JNCI 1990;82:1242–3. 8. Sasieni P, Adams J. Effect of screening on cervical cancer mortality in England and Wales: Analysis of trends with an age period cohort model. BMJ 1999;318:1244 –5. 9. Fletcher A. Screening for cancer of the cervix in elderly women. Lancet 1990;335:97–9. 10. Johannesson G, Geirsson G, Day N, Tulinius H. Screening for cancer of the uterine cervix in Iceland, 1965–1978. Acta Obstet Gynecol Scand 1982;61:199 –203. 11. Aristizabal N, Cuello C, Correa P, Collazos T, Haenszel
Cornelison et al
Medicare Cervical Screening
85
12.
13.
14.
15.
16.
86
W. The impact of vaginal cytology on cervical cancer risks in Cali, Colombia. Int J Cancer 1984;34:5–9. Clarke EA, Anderson TW. Does screening by “Pap” smears help prevent cervical cancer? A case-control study. Lancet 1979;2:1– 4. Muller C, Mandelblatt J, Schechter CB, Power EJ, Duffy BM, Wagner JL. U.S. Congress, Office of Technology Assessment. Costs and effectiveness of cervical cancer screening in elderly women. Feb. background paper no. OTA-BP-H-65. Washington, DC: US Government Printing Office, 1990. Gustafsson L, Spare´n P, Gustafsson M, Pettersson B, Wilander E, Bergstro¨m R. Low efficiency of cytologic screening for cancer in situ of the cervix in older women. Int J Cancer 1995;63:804 –9. Forsmo S, Jacobsen BK, Stalsberg H. Cervical neoplasia in Pap smears: Risk of cervical intra-epithelial neoplasia (CIN) after negative or no prior smears in a population without a mass screening programme. Int J Epidemiol 1996;25:53–7. Celentano DD, Klassen AC, Weisman CS, Rosenshein NB. Cervical cancer screening practices among older
Cornelison et al
Medicare Cervical Screening
17.
18.
19.
20.
women: Results from the Maryland cervical cancer casecontrol study. J Clin Epidemiol 1988;41:531– 41. Fahs MC, Mandelblatt J, Schechter C, Muller C. Cost effectiveness of cervical cancer screening for the elderly. Ann Intern Med 1992;117:520 –7. Ruchlin HS. Prevalence and correlates of breast and cervical cancer screening among older women. Obstet Gynecol 1997;90:16 –21. Riley GF, Potosky AL, Lubitz JD, Brown ML. Stage of cancer at diagnosis for Medicare HMO and fee-for-service enrollees. Am J Public Health 1994;84:1598 – 604. Prorok PC, Hankey BF, Bundy BN. Concepts and problems in the evaluation of screening programs. J Chronic Dis 1981;34:159 –71.
Address reprint requests to: Fredrick J. Montz, MD, KM, The Johns Hopkins Medical Institutions, Departments of Gynecology and Obstetrics, 600 North Wolfe Street, Houck 248, Baltimore, MD 21210; E-mail: [email protected]. Received August 28, 2001. Received in revised form December 11, 2001. Accepted January 10, 2002.
OBSTETRICS & GYNECOLOGY