SEARCH

SEARCH BY CITATION

Keywords:

  • breast carcinoma screening;
  • primary care physicians;
  • Medicare;
  • aging/older women

Abstract

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES

BACKGROUND

Mammography screening rates are below national recommendations for older women. Understanding the relation between the characteristics of primary care physicians (PCPs) and mammography rates for older women can help to target screening improvement efforts.

METHODS

Subjects were 2527 PCPs practicing in Michigan between 1997 and 1998. A cross-sectional design used Medicare data to identify women age 68 years or older in 1998 whom PCPs treated in 1997–1998 and to determine whether these women had a mammogram between 1996 and1998. Eligible women were Medicare beneficiaries age 65 years or older by 1996, residing in Michigan from 1996 to 1998, without specified comorbidities likely to affect decisions regarding mammography. Correlations and multiple regressions examined the relation between this score and characteristics of both PCPs and their practice populations of older women.

RESULTS

Mammography rates across physicians' practices ranged from 3–100% (mean = 59%, standard deviation = 17%). Five predictors accounted for 55% of the variance in mammography rates across practices. Higher mammography rates were found to be independently related to physicians who have: a lower mean age for female Medicare patients, a higher mean number of physicians billing for patients' care, a lower mean number of inpatient admissions, obstetrics/gynecology practices, and a higher mean education level in patient's zip code (beta weights ≥ 0.25, P < 0.0001).

CONCLUSIONS

PCPs vary substantially with regard to mammography rates for older women. Mammography rates vary more with the population of patients in physicians' practices than with commonly measured personal characteristics of physicians. Mammography rates should be adjusted for patient population to target individual PCPs with low mammography rates for interventions. Cancer 2003. © 2003 American Cancer Society.

Studies have repeatedly shown that lack of a specific recommendation from a physician is one of the most important reasons that older women cite for not undergoing mammography screening for breast carcinoma.1–7 Lack of communication by physicians is also related to other important reasons cited by women for not having screening mammography, such as not believing that general recommendations apply to them and believing that symptoms should be present.

One approach to improving screening mammography is to identify those physicians less likely to recommend screening and target them for specific interventions. Studies have identified various physicians' characteristics associated with the likelihood that their patients will receive screening mammography. For example, obstetrician/gynecologists were more likely to offer screening mammography to women older than 75 than were other primary care physicians (PCPs).8, 9 When self-reports of screening mammography referrals were compared, younger physicians were more likely than older physicians to refer older women.10 Patients of more recently graduated PCPs had higher mammography rates than patients of less recent graduates.11 Female physicians are often found to have higher mammography screening rates than male physicians.12–15

Many previous studies of physicians' characteristics related to breast carcinoma screening of older women confound physicians' characteristics with the type of patients in the physicians' practices. For example, although obstetrician/gynecologists report higher screening rates, they typically treat somewhat younger patients than do other adult PCPs. Mammography screening rates decrease appreciably with age.9, 16 Obstetrician/gynecologists may, indeed, have higher screening rates across all types of women or they may have the same screening rates as other physicians for women of the same age, but see a younger age group on average. Differences in physicians' characteristics could also result from underlying differences in physicians' practices on other patient characteristics associated with lower mammography rates, including poorer health,17 lower level of education,2, 17, 18 lower incomes,2, 18 being African-American,16, 19 and living in rural areas.2, 3

Another problem is that many previous studies of physicians' characteristics and mammography screening were based on self-reports by physicians. Comparisons of physician self-report and chart audit found that physician self-reports of the screening rates for patients in their practices were unreliable.13, 20

The current study sought to identify characteristics of PCPs and older women patients in their practices that can be used to target efforts to improve mammography screening. The goals of the current study were to 1) use Medicare claims to score PCPs on the percent of their women patients age 68 years or older who have had a recent mammogram (within a 3-year period), 2) identify the characteristics of these physicians that may be related to their screening practice, 3) identify the characteristics of older women in physicians' practice populations that are likely to be independent influences on the observed screening rates, and 4) determine the extent to which screening rates are a function of physicians' characteristics or the characteristics of the population of older women in their practices.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES

Design and Sample

The study utilized a cross-sectional design to evaluate relations among the percent of older women in a PCP's practice who have been screened recently, the characteristics of PCPs, and the characteristics of older women in physicians' practices. The study was approved by the University of Michigan Medical School Institutional Review Board for Human Subject Research (no. 1995-0343).

The study included 2527 PCPs who met all of the following criteria: 1) general practitioners, family physicians, internists, or obstetrician/gynecologists, 2) a practice office address in Michigan, 3) primarily in clinical practice, 4) graduated from medical school by 1993, and 5) submitted claims for treating at least 10 eligible women during 1997–1998. Eligible women met all of the following criteria: 1) Medicare beneficiaries, 2) age 65 or older by 1996, 3) alive and residing continuously in Michigan from 1996 to 1998, 4) no evidence of specified medical conditions that may alter mammography screening, and 5) not enrolled in managed care from 1996 to 1998.

Medicare beneficiary files for 1996, 1997, and 1998 were used to identify women age 65 years or older in 1996 with their primary residence in Michigan for all 3 years. Women were excluded if they had Medicare Part B claims in 1996–1998 with diagnosis or procedure codes for medical conditions likely to alter mammography use (e.g., significant malignancies, severe mental problems, and specified physical conditions, such as breast removal). ICD—9-CM diagnostic codes and Healthcare Common Procedure Coding System (HCPCS) codes used to classify women as having these conditions are available from the authors. A total of 485,292 women were eligible.

Medicare's Unique Physician Identification Number (UPIN) Master File for 1998 identified 4500 physicians who were eligible to provide care under Medicare coverage, with one of the four primary care specialties as their specialty and with a business office zip code in Michigan. The UPIN for these physicians was linked to the UPIN listing in the American Medical Association's Physician's Professional Data (AMA–PPD) file for all physicians in Michigan for 1998. Variables in the AMA–PPD file were used to exclude physicians likely not to be in stable clinical practice during recent years (e.g., Type of Practice = administration, research, retired). The UPINs for the remaining physicians were linked to patient claims: 2803 PCPs treated at least one of the eligible women and 2632 treated 10 or more of these women. We limited the current study to physicians who had treated 10 or more women to increase the reliability of characterizing physicians' practices when calculating means across treated patients. We further restricted the sample to physicians who graduated from medical school by 1993 and were licensed to practice by 1996 to ensure they were out of training during both 1997 and 1998. This resulted in 2527 PCPs identified for the study.

Measures

Mammography rate for a physician's practice

The score for a physician's practice on mammography was calculated as the percent of older women seen by the PCP in 1997 and 1998 who had a claim for mammography from 1996 to 1998. Medicare Part B claim files for 1997 and 1998 were used to identify women who had received a service from one or more of the PCPs in the study. We used Medicare Part B claim files for 1996, 1997, and 1998 to determine whether a woman had a mammogram during the 3 years. Both screening mammograms (two-view, HCPCS code 76092) and diagnostic mammograms (typically six-view, HCPCS codes 76090 and 76091) were counted as having a recent mammogram. Women with diagnosed cancer were already excluded from the study. Women referred for a screening mammogram may have a diagnostic mammogram performed if the intake interview at the mammography facility identifies risk factors (e.g., family history) for breast carcinoma.

Women may see more than one PCP, resulting in the women's status on mammography being included in the mammography scores of more than one PCP. We determined this was not a practical problem conceptually or empirically. Conceptually, PCPs are expected to include preventive care as part of their services. Even if one PCP is primarily responsible for a mammography occurring, each PCP seeing a woman should check her status. If the woman has been screened recently, other PCPs checking her status could have ordered a mammogram if it were needed. Empirically, we checked alternative ways of assigning women to be included in a physician's practice. For example, we defined subgroups of women based on the exclusivity of their relationship with a PCP: women who saw only this PCP, women who saw this PCP for the majority of visits to PCPs, and women who saw other PCPs more often than this PCP. We calculated separate practice scores on mammography based on all women the physician saw and based on women in each relationship subgroup. We compared the scores and found that they all had similar patterns in ranking physicians' practices. Therefore, we chose to use only the score for all women seen in subsequent analyses.

Physicians' characteristics

Physicians' characteristics were obtained from the AMA–PPD file for 1998. The eight measures are listed in Table 1.

Table 1. Personal Characteristics of Physicians (N = 2527)
inline image
Population characteristics of older women in a physician's practice

The population of older women in a practice was measured on the 12 characteristics listed in Table 2. For women treated by a PCP during 1997 or 1998, the values on a characteristic were combined (e.g., mean age) to provide an overall score on that characteristic for the physician's practice population.

Table 2. Measures of Physicians' Practice Population of Older Women Patientsa
inline image

Medicare Part B claims for 1997 and 1998 were used to identify the number of eligible older women in a practice. Medicare beneficiary files provided some demographic measures, such as age, race (coded to African-American or not), and urban or rural residence (zip code matched to Social Security Administration's designation of county as metropolitan statistical area or not). Additional demographic characteristics of women were approximated using census data at the zip code level. Census data from 1990 were obtained.21 A woman's zip code was used to link the mean values for a zip code to her characteristics. From the census measures available, we selected mean education level for adults of the individual's race, mean household income for the same race and age 65 years or older, and mean percent of individuals below poverty level for women of the same race and age 65 years or older. This use of population-based mean values has been shown to provide similar results to person-specific results for aggregate analyses.22

Medicare Part B claim files for 1997–1998 were used to create measures of a woman's interaction with the health care system: number of inpatient admissions, number of visits to physicians (inpatient claims 3 or more days apart counted as different visits and outpatient claims 2 or more days apart counted as different visits), and number of physicians billing for care (number of different physician billing numbers across all of the individual's claims). Interactions with PCPs were similarly calculated when the specialty code on the claim was for a PCP.

Analyses

All data were measured at the physician level, either as measures of a physician's personal characteristics or as summary measures (e.g., means) for characteristics of the population of older women patients in the physician's practice. All analyses were performed at the physician level. Standard descriptive statistics (e.g., means, distributions) were used to characterize the physicians and their practices. Pearson product-moment correlations evaluated the strength of associations between variables. For categoric variables, we recoded each category into a “yes” or “no” variable to determine the magnitude of each category's correlation with other variables. Multiple linear regression models evaluated the independent and combined effects of predictors on mammography rates in physician's practices. “Dummy” variable regression was used for categoric predictors. For a regression model, we report both the unstandardized coefficient (“b”) for the original scale of a predictor and the standardized transformation (mean = 0, standard deviation [SD] = 1) of that coefficient (“beta”).23 Beta weights estimate the effect on the dependent variable for a change of 1 SD on a predictor while holding the other predictors constant. Beta weights allow direct comparison of the magnitude of relations that different predictors have with the dependent variable. The correlation equals the beta weight for linear regressions having only one predictor.

The sample size of 2527 has the power to detect extremely weak relations, e.g., 90% power to detect a correlation of 0.07 as significant at the 0.05 level. A correlation of 0.07 reflects 0.49% shared variance, which is too small to be of practical consequence for most purposes. Although we report significance levels, we consider results of practically meaningful magnitude to be correlations of 0.20 or higher (r2 ≥ 4% shared variance) and for multiple regressions, beta values greater than or equal to 0.15 (approximately ≥ 2% variance in the dependent variable is associated with a predictor, independent of effects of all other predictors).

RESULTS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES

Characteristics of Physicians and Their Practice Populations

Distributions on eight characteristics of physicians are presented in Table 1. For the two noncategorical measures, the overall mean age of physicians is 50 (SD = 11) and the mean year of medical school graduation is 1978 (SD = 11). Distributions and means on 12 characteristics of older women in physicians' practice populations are presented in Table 2. Also shown in Table 2 is the distribution of physicians' practice populations on the percent of older women with a recent mammogram. The mean across practices is 59% of older women in a practice to have at least one mammogram within 3 years. Scores for individual practices ranged from 3% to 100%.

Relations of Predictors to Mammography Rates

Table 3 presents the correlations and multiple linear regression coefficients (all 20 predictors included in the model) for the physician and practice population characteristics in predicting the mammography rates in practices. Correlations show the magnitude of association of an individual predictor to mammography rate without accounting for associations between predictors. The regression coefficient reflects a predictor's independent association with mammography when other predictors are held constant.

Table 3. Correlation and Multiple Regression of Predictors with Percent Mammographya
PredictorCorrelation coefficient rbRegression coefficientc
Beta (standardized)B (unstandardized)P
  • PCP: primary care physician.

  • a

    N = 2527 primary care physicians. In bold font are r ≥ 0.20 and both beta and B when beta ≥ 0.15.

  • b

    For r ≥ 0.04, P < 0.05; for r ≥ 0.06, P < 0.01; for r ≥ 0.07, P < 0.001.

  • c

    A multiple linear regression model including the 20 predictors in this table.

  • d

    P < 0.05.

  • e

    P < 0.01.

  • f

    P < 0.001.

Physician's personal characteristics    
 Age−0.20−0.13−0.002e
 Medical school graduation year0.18−0.05−0.00008 
 Gender (male)−0.26−0.11−0.045f
 Urban practice location0.060.050.020e
 Board certification0.110.050.019f
 Specialty (reference: general practice)−0.16 
  Family practice−0.19−0.02−0.006 
  Internal medicine−0.040.060.022e
  Obstetrics/gynecology0.540.270.141f
 Medical school (reference: foreign)−0.03 
  University of Michigan0.080.020.0009 
  Wayne State University0.170.050.021 
  Michigan State University—allopathic0.000.010.008 
  Michigan State University—osteopathic−0.020.050.024 
  Other U.S. medical schools−0.170.010.003 
 Residency (reference: “missing”)−0.26 
  University of Michigan0.070.040.037e
  Wayne State University0.020.030.023d
  Michigan community hospital0.170.070.024f
  Outside Michigan0.060.090.041f
Practice population of older women    
 No. of older women seen 1997–1998−0.100.070.00008f
 Mean age−0.59−0.43−0.040f
 Percent African-American−0.13−0.04−0.033 
 Percent urban residence0.02−0.14−0.067f
 Mean education, zip0.330.250.126f
 Mean household income, zip0.35−0.12−0.000004f
 Mean percent below poverty, zip−0.28−0.01−0.040 
 Mean no. of inpatient admissions 1997–1998−0.31−0.31−0.26f
 Mean no. of visits to physicians 1997–1998−0.11−0.12−0.004f
 Mean no. of physicians billing for care 1997–1998−0.040.400.014f
 Mean no. of visits to PCPs 1997–1998−0.120.080.001f
 Mean no. PCPs involved in care 1997–1998−0.04−0.06−0.016f
R2 (all predictors)c  0.62f

Four physicians' characteristics and five practice characteristics are meaningfully correlated with mammography rates (r ≥ 0.20, P < 0.0001). Mammography rates are higher for physicians who are younger, female, obstetrician/gynecologists, and for whom residency location was available. Mammography rates are higher in practices where women are younger; live in zip codes with a higher education level, a higher household income, and a lower percent of population below poverty; and have a lower number of inpatient admissions compared to other practices.

Evaluating the regression coefficients to determine independent predictors of mammography rates, one physician characteristic and four practice characteristics have beta weights (standardized coefficients) greater than or equal to 0.15 (P < 0.0001). Mammography rates are independently higher for obstetrician/gynecologists compared to other practices. Rates are also independently higher for physicians' practices where women are younger, live in zip codes with a higher education level, have a lower number of inpatient admissions, and have a higher number of physicians billing for their care compared to other practices. Using the original scales for the predictors (Tables 1 and 2), the unstandardized coefficients show mammography rates averaging 14.1 percentage points higher for obstetrician/gynecologists than for general practitioners, 4 percentage points lower per additional year in mean patient age, 12.6 percentage points higher per unit increase in mean level of education in zip code (e.g., high school graduate vs. some college), 2.6 percentage points lower per additional admission in mean number of admissions, and 1.4 percentage points higher per additional physician in mean number of physicians billing for care.

Unadjusted and Adjusted Mammography Rates by Predictor

For each of the predictors with correlations greater than or equal to 0.20 or beta greater than or equal to 0.15, Table 4 presents unadjusted mammography rates for specific values of each predictor and these mammography rates adjusted to the mean values for all of the other predictors. For example, as the mean age of older women in a practice increases from 73 years to 78–83 years, the mammography rate decreases 37 percentage points, from 81% to 44% (reflecting the correlation of −0.59 in Table 3). When the mean mammography rate is adjusted for the effects of other predictors, it decreases only 19 percentage points, from 69% to 50% (reflecting the beta weight of −0.43 in Table 3).

Table 4. Unadjusted and Adjusted Percent of Older Women Patients with Mammograms by Physician and Practice Population Characteristicsa
PredictorPercentAdjusted percentb
  • a

    N = 2527 primary care physicians. Only values for predictors with either r ≥ 0.20 or beta ≥ 0.15 (in bold font in Table 3) are presented.

  • b

    A multiple linear regression model included all 20 predictors (values categorized as indicated in Table 4). The model adjusts the percentage of women with mammography that are associated with a particular category by adjusting all other predictors to the mean value for each predictor.

  • c

    Education of adults of the individual's race: 3 = high school graduate; 4 = some college but no degree, 5 = associate's degree.

  • d, f

    P < 0.05. P < 0.01, P < 0.001.

  • eP < 0.01.

  • f

    P < 0.001.

Physicians' personal characteristics
 Age (yrs)ff
  30–396360
  40–496161
  50–596060
  60–695457
  70–905153
 Genderff
  Female6863
  Male5758
 Specialtyff
  General practice5257
  Family practice5556
  Internal medicine5860
  Obstetrics/gynecology8473
 Residencyff
  (Missing)5357
  University of Michigan6763
  Wayne State University6160
  Michigan community hospital6360
  Outside of Michigan6262
Practice population of older women
 Mean age (yrs)ff
  738169
  74–756463
  76–775758
  78–834450
 Mean education, zipcff
  2.7–2.94654
  3.0–3.45457
  3.5–3.96160
  4.0–4.47064
  4.5–5.27265
 Mean household income, zipff
  $13,000–$19,9995459
  $20,000–$24,9995760
  $25,000–$29,9996459
  $30,000–$34,9997259
  $35,000–$50,0007960
 Mean percent below poverty, zipff
  3–4%7458
  5–9%6360
  10–14%5760
  15–19%5157
  20–34%4958
 Mean no. of inpatient admissions 1997–1998ff
  0.0–0.96664
  1.0–1.96060
  2.0–2.95757
  3.0–3.95252
  4.0–21.04650
 Mean no. of physicians billing 1997–1998df
  4–94952
  10–146159
  15–196560
  20–245062
  25–394559

For most of the predictors in Table 4, adjusting the mammography rate reduces the range of values. The exception is the mean number of physicians billing for care. In Table 3, this measure was not correlated with mammography rates, but had a beta value of 0.40. Table 4 shows an underlying curvilinear relation. The unadjusted mean rate is 49% when four to nine physicians bill for care, increases to 65% when 15–19 physicians are involved, then decreases to 45% when 25–39 physicians are involved. The adjusted mammography rate increases from 52% when four to nine physicians bill for care to approximately 60% when more physicians are involved.

Variation in Mammography Rates

Together, the 20 predictors in Table 3 account for 62% (R2) of the variation across physicians' practices in the percent of older women with recent mammography. We performed additional regression models and found that the 10 predictors with beta weights greater than or eqaul to 0.10 account for 60% of the variance and the five predictors with beta weights greater than or equal to 0.15 account for 55% of the variance.

DISCUSSION

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES

Previous studies have evaluated various characteristics of physicians associated with mammography rates for older women. The current study is unique in using a large representative sample of PCPs and non–self-report data to evaluate mammography rates and the joint effects of a number of characteristics of both physicians and of the population of older women in their practices.

Main Findings

Mammography rate in primary care physician practices

The overall average mammography rate and the distribution of mammography rates both indicate the need for interventions to improve mammography use for older women in PCP practices. Primary care practices have an overall mean rate of 59% of older women having mammography within 3 years (excluding women with cancer, major mental problems, and selected physical diagnoses). Although screening would not be appropriate for women with a short life expectancy, screening is likely to be appropriate for a substantial proportion of the 41% of women not screened. In addition, using a 3-year period for screening overestimates screening performance compared with a recommended screening interval of every 1–2 years. Although questions have been raised concerning the methodologic rigor of some studies of breast carcinoma screening,24 the U.S. Preventive Services Task Force recommends screening mammography every 1–2 years for “women age 70 and older (who face a higher absolute risk of breast cancer) if their life expectancy is not compromised by comorbid disease.”25

Mammography rates vary widely across physicians' practices. Eleven percent of practices have rates lower than 40% and 13% of practices have rates of 80% or higher (Table 2). This range and the known impact of physicians on patients' willingness to get mammography screening suggest that special efforts should be directed toward physicians with lower mammography rates.

Predictors of mammography rates in PCP practices

The predictors included in the current study account for the majority (R2 = 62%) of the variation in mammography rates across practices. The five strongest independent predictors suggest the most important processes and their relative importance.

The strongest independent predictor is the mean age of the women in the practice, even when factors associated with poor health (number of inpatient admissions, number of visits to physicians) are held constant. When asked, physicians reported that they recommended mammography less often to older women.8, 26 Presumably, both physicians and the older women in their practices believe that increasing age, independent of health status, is a reason not to be screened. The effect of age independent of health status has been found in other studies.9, 27

An almost equally strong independent predictor is mean number of physicians billing for care for older women in the practice. The major effect is that lower mammography rates occur when few physicians are involved and higher rates occur when additional physicians are involved. Presumably, as women in reasonable health are involved with more physicians, the likelihood increases that some physicians will recommend screening mammography.

The negative relation between the mean number of inpatient admissions and mammography rates is almost certainly due to more inpatient admissions reflecting poorer health. This association likely reflects appropriate clinical judgments not to perform tests for longer-term prevention.

Obstetrician/gynecologists have higher screening rates in their practices than other primary care specialties. The substantial decrease from a correlation of 0.54 to a beta weight of 0.27 indicates that most of the difference between obstetric/gynecologic practices and other primary care specialties is due to differences in characteristics of older women in those practices. The adjusted mammography rate of 73% in obstetric/gynecologic practices remains meaningfully higher than the adjusted rate of 58% across other primary care specialties (Table 4). This difference may reflect both obstetric/gynecologic practice and underlying attitudes and beliefs of older women who choose obstetric/gynecologic for primary care.

Practices with women living in zip codes with higher education tend to have higher mammography rates independent of other factors. The positive relation between education and mammography has been found in other studies, but with varying strength.2, 17 Among older women with regular physicians, education was the patient demographic variable most highly related to mammography.28 Individuals with more education are likely to be more aware of the importance of mammography screening and to initiate obtaining mammograms.

Physician characteristics versus practice population characteristics

When comparing mammography rates across physicians, rates should be adjusted for differences in patient populations. Differences in mammography rates across PCP practices are much more strongly associated with the characteristics of women in the practice than with commonly measured characteristics of physicians. Four of the five strongest independent predictors of mammography rates are patient characteristics. The only meaningful physician characteristic is being an obstetrician/gynecologist, which is as strongly related to mammography rates as the fourth strongest characteristic of patients.

Limitations

The current study necessarily has the strengths and limitations of the databases utilized. A major strength is the ability to measure actual mammography behavior and interactions with the health care system. Medicare enrolls 96% of U.S. women age 65 years or older.16 Most eligible individuals use their Medicare benefits, do so in fee-for-service arrangements, and use Medicare as the primary payer when other sources of payment are involved. Data on specific claims will not be available for women in managed care programs under Medicare or for older women receiving care through the Veterans Administration health care system. Data are not available until women enroll in Medicare, usually at age 65 years. Therefore, a claims history is not available until 2–3 years after that time. Although procedures are likely to be coded accurately, the coding of diagnoses is likely to be less uniform. The AMA–PPD file has parallel strengths and limitations. It is the most comprehensive set of national data on characteristics of physicians. However, at any given time, some physicians are changing practice locations, getting additional specialty certifications, or undergoing other changes that may not be reflected in the data file. Data are missing or inaccurate for some entries in all of these files. Overall, the data are sufficiently accurate for aggregate analyses and no better data sets are likely to be available for statewide and national studies of these characteristics.

Medicare claims do not provide important data on individual women relevant to mammography (e.g., family history of breast carcinoma, severity of comorbid disease). We used claims data to exclude women with some specific diagnoses likely to affect preventive screening. For the remaining women, health is reflected by number of hospital admissions and number of physicians involved in care. These measures do not precisely classify the health status of individual women. However, our study focuses on physicians and averages across all women in their practices. The measures may function reasonably in characterizing average health status for the group of women in a physician's practice.

The study was limited to the population of physicians in Michigan. An advantage of this geographic area is that Michigan is bounded by the Great Lakes on three sides, resulting in most physicians' practices being confined to patients within Michigan. Michigan broadly resembles many states in its composition by age, racial distribution, and distribution across urban, suburban, and rural communities. However, Michigan varies somewhat from overall U.S. averages. For example, among minorities, Michigan has a higher proportion of African-Americans and a lower proportion of Hispanics.29 Also, Michigan has a higher than average rate of mammography among women age 50 years or older.30 The overall pattern of relations found in Michigan is likely to apply generally across the U.S. However, specific rates of mammography may differ for other populations.

Implications

Interventions

The results suggest that cost-effective efforts to increase mammography among older women seeing PCPs will 1) measure performance and target physicians with low mammography rates and 2) encourage physicians to focus on the health status—not the age—of their older women patients when recommending mammograms.

With the exception of obstetrics/gynecology as a specialty, commonly measured characteristics of physicians are not associated with mammography rates and cannot be used to target interventions. Mammography rates must be measured for individual physicians. Mammography rates adjusted for patient population characteristics can be used in data-based feedback interventions to improve performance.31 The rates can also be used to target low-scoring physicians for more intensive interventions.

Messages to physicians should be worded to correct the apparent misperception that mammography screening is less important as women age. The results indicate that physicians (and the older women they treat) use age, independent of health status, as a reason not to be screened for breast carcinoma. A likely factor contributing to this age bias is the relative lack of emphasis on older women in information provided to physicians concerning mammography screening. The major randomized, controlled studies of mammography screening have been performed on women younger than 70 years, with discussion focusing primarily on younger women. Messages to physicians should highlight recent evidence-based recommendations regarding screening of older women, with the balance of benefits to harms growing more favorable as women age.25 The incidence of breast carcinoma increases with age, as does the sensitivity and positive predictive value of mammography.32 Older women have mammography less frequently and are more likely to be diagnosed with advanced breast carcinoma. Regular mammography use can eliminate age-related disparities in size and stage of breast carcinoma at diagnosis.33 Although overall life expectancy decreases with age, screening is moderately cost-effective for women 70–79 years old, who are at a higher risk for breast carcinoma.34 The cost-effectiveness will be even higher for older women who do not have comorbid conditions likely to shorten their life expectancy.

Messages regarding the importance of screening older women can be accompanied by supportive operational information. For example, a clinical framework for older women is available that ignores age and helps to determine the appropriateness of screening based on a patient's health status and preferences.35 Information is also available concerning office systems to monitor preventive care and approaches that enhance older women's acceptance of recommendations for mammography.36–38

The institutional responsibility and infrastructure for performing these interventions already exist within Medicare's current quality of care initiatives.39 The Centers for Medicare and Medicaid Services have charged the quality improvement organizations (QIOs, formerly peer review organizations) in each state with the responsibility for working with physicians on projects to improve care for Medicare beneficiaries. Biennial screening mammography is one of the specified quality measures. QIOs already have access to Medicare claims data and the capability to produce for their states the physician performance measure described in the current study.

Research

The methods and findings of the current study can guide the development of demonstration projects to increase mammography rates for older women in physicians' practices. The methods can also be used to measure physicians' mammography rates for older women in other states and nationally, as well as to monitor changes across time. Future research can explore the use of Medicare data to measure physicians' rates of other preventive care (e.g., colorectal carcinoma screening, immunizations) for older persons and factors associated with the rates for those services in physicians' practices.

REFERENCES

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES
  • 1
    Rimer BK. Improving the use of cancer screening for older women. Cancer. 1993; 72: 10841087.
  • 2
    Pearlman DN, Rakowski W, Ehrich B, Clark MA. Breast cancer screening practices among black, Hispanic, and white women: reassessing differences. Am J Prev Med. 1996; 12: 327337.
  • 3
    Ruchlin HS. Prevalence and correlates of breast and cervical cancer screening among older women. Obstet Gynecol. 1997; 90: 1621.
  • 4
    Bastani R, Marcus AC, Maxwell AE, Das IP, Yan KX. Evaluation of an intervention to increase mammography screening in Los Angeles. Prev Med. 1994; 23: 8390.
  • 5
    Caplan LS, Wells BL, Haynes S. Breast cancer screening among older racial/ethnic minorities and whites: barriers to early detection. J Gerontol. 1992; 47: 101110.
  • 6
    Fox SA, Murata PJ, Stein JA. The impact of physician compliance on screening mammography for older women. Arch Intern Med. 1991; 151: 5056.
  • 7
    Grady KE, Lemkau JP, McVay JM, Reisine ST. The importance of physician encouragement in breast cancer screening of older women. Prev Med. 1992; 21: 766780.
  • 8
    Herman CJ, Lengerich EJ, Stoodt G. Variation in recommendations for breast and cervical cancer screening among primary care physicians in North Carolina, 1991. South Med J. 1996; 89: 583590.
  • 9
    Harrison RV, Janz NK, Wolfe RA, Tedeschi PJ, Huang X, McMahon LF. 5-year mammography rates and associated factors for older women. Cancer. 2003; 97: 11471155.
  • 10
    Roetzheim RG, Vandurme DJ, Brownlee HJ, et al. Reverse targeting in a media-promoted breast cancer screening project. Cancer. 1992; 70: 11521158.
  • 11
    Schoen RE, Marcus M, Braham RL. Factors associated with the use of screening mammography in a primary care setting. J Community Health. 1994; 19: 239252.
  • 12
    Ackermann SP, Cheal N. Factors affecting physician adherence to breast cancer screening guidelines. J Cancer Educ. 1994; 9: 96100.
  • 13
    Saver B, Taylor T, Treadwell J, Cole W. Do physicians do as they say?: The case of mammography. Arch Fam Med. 1977; 6: 543548.
  • 14
    Anderson MR, Urban N. Physician gender and screening: Do patient differences account for differences in mammography use? Women Health. 1997; 26: 2939.
  • 15
    Lurie N, Margolis KL, McGovern PG, Mink PJ, Slater JS. Why do patients of female physicians have higher rates of breast and cervical cancer screening? J Gen Intern Med. 1997; 12: 3443.
  • 16
    Trontell AE. Use of mammography services by women aged ≥65 years enrolled in Medicare—United States, 1991-1993. MMWR Morb Mortal Wkly Rep. 1995; 44: 777781.
  • 17
    Ives DG, Lave JR, Traven ND, Schulz R, Kuller LH. Mammography and Pap smear use by older rural women. Public Health Rep. 1996; 111: 244250.
  • 18
    Breen N, Kessler L. Changes in the use of screening mammography: evidence from the 1987 and 1990 National Health Interview Surveys. Am J Public Health. 1994; 84: 6267.
  • 19
    National Center for Health Statistics. Health, United States, 2000 with adolescent health chartbook. Hyattsville, MD: National Center for Health Statistics, 2000.
  • 20
    Montano DE, Phillips WR. Cancer screening by primary care physicians: a comparison of rates obtained from physician self-report, patient survey, and chart audit. Am J Public Health. 1995; 85: 795800.
  • 21
    Bureau of the Census. Census of population and housing, 1990 [United States]: summary Table File 3B [Computer file]. Washington, DC: U.S. Department of Commerce, Bureau of the Census (producer), 1993.
  • 22
    Hofer TP, Wolfe RA, Tedeschi J, McMahon LF, Griffith JR. Use of community versus individual socioeconomic data in predicting variation in hospital use. Health Serv Res. 1998; 33: 219235.
  • 23
    Nunnally JC. Psychometric theory. New York: McGraw-Hill, 1967.
  • 24
    Olsen O, Gotzsche PC. Screening for breast cancer with mammography. Cochrane Database Systematic Rev. 2002 (2).
  • 25
    U.S. Preventive Services Task Force. Screening for breast cancer. Washington DC: Agency for Healthcare Research and Quality, 2002.
  • 26
    Hamblin JE. Physician recommendations for screening mammography: results of a survey using clinical vignettes. J Fam Pract. 1991; 32: 472477.
  • 27
    Burack RC, Gurney JG, McDaniel AM. Health status and mammography use among older women. J Gen Intern Med. 1998; 13: 366372.
  • 28
    Mayer-Oakes SA, Atchison KA, Matthias RE, De Jong FJ, Lubben J, Schweitzer SO. Mammography use in older women with regular physicians: what are the predictors? Am J Prev Med. 1996; 12: 4450.
  • 29
    U.S. Census Bureau. Rankings and comparisons population and housing tables: PHC-T-6 population by race and Hispanic or Latin origin for the United States, regions, divisions, states: 2000. Washington, DC: U.S. Census Bureau, 2001. Available from URL: www.census.gov/population/www/cen2000/tablist.html [accessed July 23, 2002].
  • 30
    Centers for Disease Control and Prevention. 2001 state health profiles. Atlanta: Centers for Disease Control and Prevention, 2001. Available from URL: www.cdc.gov/nchs/datawh/stprofiles.htm [accessed July 23, 2002].
  • 31
    Gustafson DH, Cats-Barli WL, Alemni F. Systems to support health policy analysis: theory, models and uses. Chicago: Health Administration Press, 1992.
  • 32
    Kerlikowske K, Grady D, Barclay J, Sickles EA, Ernster V. Effect of age, breast density, and family history on the sensitivity of first screening mammography. JAMA. 1996; 276: 3338.
  • 33
    Randolph WM, Goodwin JS, Mahnken JD, Freeman JL. Regular mammography use is associated with elimination of age-related disparities in size and stage of breast cancer at diagnosis. Ann Intern Med. 2002; 137: 783792.
  • 34
    Kerlikowske K, Salzmann P, Phillips KA, Cauley JA, Cummings SR. Continuing screening mammography in women aged 70 to 79 years: impact on life expectancy and cost-effectiveness. JAMA. 1999; 282: 21562163.
  • 35
    Walter LC, Covinsky KE. Cancer screening in elderly patients: a framework for individualized decision making. JAMA. 2001; 285: 27502756.
  • 36
    Agency for Healthcare Research and Quality. Put prevention into practice, a step-by step guide to delivering clinical preventive services: a systems approach. AHRQ pub. No. APPIP01-0001. Washington, DC: Agency for Healthcare Research and Quality, 2001.
  • 37
    Weinberger M, Saunders AF, Bearon LB, et al. Physician-related barriers to breast cancer screening in older women. J Gerontol. 1992; 47: 111117.
  • 38
    Preston JA, Scinto JD, Grady JN, Schulz AF, Petrillo MK. The effect of a multifaceted physician office-based intervention on older women's mammography use. J Am Geriatr Soc. 2000; 48: 17.
  • 39
    Centers for Medicare and Medicaid Services. Quality improvement organizations (QIOs) statement of work. Baltimore: Centers for Medicare and Medicaid Services, 2002. Available from URL: cms.hhs.gov/qio/2.asp [accessed July 23, 2002].