- Top of page
BACKGROUND: African-American and low-income women have lower rates of cancer screening and higher rates of late-stage disease than do their counterparts.
OBJECTIVE: To examine the effects of primary care, health insurance, and HMO participation on adherence to regular breast, cervical, and colorectal cancer screening.
DESIGN: Random-digit-dial and targeted household telephone survey of a population-based sample.
SETTING: Washington, D.C. census tracts with ≥30% of households below 200% of federal poverty threshold.
PARTICIPANTS: Included in the survey were 1,205 women over age 40, 82% of whom were African American.
MAIN OUTCOME MEASURES: Adherence was defined as reported receipt of the last 2 screening tests within recommended intervals for age.
RESULTS: The survey completion rate was 85%. Overall, 75% of respondents were adherent to regular Pap smears, 66% to clinical breast exams, 65% to mammography, and 29% to fecal occult blood test recommendations. Continuity with a single primary care practitioner, comprehensive service delivery, and higher patient satisfaction with the relationships with primary care practitioners were associated with higher adherence across the 4 screening tests, after considering other factors. Coordination of care also was associated with screening adherence for women age 65 and over, but not for the younger women. Compared with counterparts in non-HMO plans, women enrolled in health maintenance organizations were also more likely to be adherent to regular screening (e.g., Pap, odds ratio [OR] 1.89, 95% confidence interval [CI] 1.11 to 3.17; clinical breast exam, OR 2.04, 95% CI 1.21 to 3.44; mammogram, OR 1.95, 95% CI 1.15 to 3.31; fecal occult blood test, OR 1.70, 95% CI 1.01 to 2.83.)
CONCLUSIONS: Organizing healthcare services to promote continuity with a specific primary care clinician, a comprehensive array of services available at the primary care delivery site, coordination among providers, and better patient-practitioner relationships are likely to improve inner-city, low-income women's adherence to cancer screening recommendations.
Low-income women have disproportionate breast, cervical, and colorectal cancer morbidity and mortality. 1–7 African Americans are disproportionately represented among women with low incomes. A large portion of the income- and race-associated cancer morbidity and mortality is related to lower use of regular cancer screening. 6–11 This differential in screening adherence persists despite evidence that regular screening reduces mortality from 30% to 70%. 12–19 While “ever” and “recent” screening rates are increasing for all groups, 20–22 fewer data are available on factors associated with adherence to recommended 18 use of regular screening, especially for low-income and minority women. 23–32
Having a regular source of care 24–29 or a physician recommendation for screening 30–32 are two of the most consistent predictors of cancer screening among women of all income and demographic groups. If this “usual source of care” is a primary care site, then a “recent” cancer screening test is more likely to have occurred. 28,29 However, among those studies that assessed whether the woman had “a usual source of care,” most did not measure specific characteristics of primary care provider settings, 33 where screening is most often initiated. The studies that focused on the process and structure of primary care 28,29 did not concurrently measure patient attitudes and beliefs with respect to screening utilization, and target lower income persons. 34–37
We examined the effects of primary care and health insurance, including enrollment in health maintenance organizations, on adherence to breast (clinical breast exams and mammograms), cervical (Pap test), and colorectal (fecal occult blood test) cancer screening over time among a population-based sample of women living in low-income urban census tracts. We hypothesized that women with better primary care delivery sites, defined as care that is continuous, comprehensive, accessible, coordinated, and involving a strong patient-clinician relationship, would have higher rates of adherence to cancer screening, even in the face of strong socioeconomic, insurance and cultural belief barriers to screening.
- Top of page
Table 1 describes the sample. For comparison, we used the 1999 Current Population Survey 52 (CPS) data for Washington, D.C. to describe the universe of women over age 40 living in the census tracts sampled. Compared with the CPS estimates, the study population was older, had less formal education, was poorer, and was more likely to be African American. This reflects success in targeting a large subgroup of low-income women.
Table 1. Characteristics of the Study Sample ( N = 1,205) Compared with Women Living in the Same Census tract *
|Characteristic||Study Population ( N = 1,205)||Women in Same Census Tracts as the Study Population|
|Mean age, y||64.8||59.7|
| 41–49, %||16.3||25.5|
| 50–65, %||28.4||38.7|
| ≥65, %||55.3||35.8|
|Education (highest completed), %|
| <12 years||26.3||21.8|
| High school grad/GED||33.5||27.7|
| ≥Some college||40.2||50.5|
| Don't know/refused (most similar to the <$10K group)||26.9|| |
| <$ 10K||11.5||11.5|
|Self-identified ethnicity/race, %|
| Black/African American||82.7||67.0|
| Refused||7.0|| |
|Owns home (vs rents), %||66.2||63.5|
|Work status, %|
| Working full-time||24.3||41.1|
| Working part-time||6.4||9.1|
|Married/living as married, %||26.5||37.4|
|Mean family size||2.1||2.3|
| ≥4 Persons/household, %||12.4||15.6|
|Health status (self-assessed), %|| ||NA|
| Poor–fair||26.2|| |
| Good||36.4|| |
| Very good–excellent||37.4|| |
|Has a regular personal doctor/nurse, %||84.8||NA|
|Health insurance for any period during the past 12 mo, %|| ||NA|
| Public only||22.8|| |
| Private (may also have had Medicare/Medicaid)||67.9|| |
| Uninsured for the entire past 12 mo|
| <65 years old||13.2|| |
| ≥65 years old||6.3|| |
Study population rates of uninsurance for women age 41 to 64 (13.2%) were slightly higher than national rates (7%) for the same age group. 10 Seven percent of those with a regular clinician identified an obstetrician-gynecologist as that clinician. The majority of respondents (62%) used private doctors' offices or an HMO. Twenty-seven percent attended community health centers or other nonprofit community health clinics.
Only 1 of the cancer attitude and belief items was consistently associated with adherence across all 4 cancer screening tests: “Going to the doctor for check-ups when well” ( P≤ .01). None of these attitude or belief items were consistently associated with screening adherence in the logistic regression models (see Tables 3 and 4).
Table 3. Factors Significantly Associated with Adherence to Cancer Screening Recommendations for Women <65 Years Old ( N = 539) Living-income Census Tracts of Washington, D.C. 2000
|Significant Factors||Adherence to Screening-adjusted OR * (95% CI)|
|Pap Test||Clinical Breast Exam||Mammogram||FOBT †|
| Knowledge/attitudes/beliefs ‡|
| Avoid doctor even if sick|| ||0.56 § (0.34 to 0.91)||0.61 § (0.37 to 1.00)|| |
| Demographic and socioeconomic|
| Age (50–64 vs the 40–49 reference group)|| || ||2.14 ‖ (1.43 to 3.20)|| |
| Income higher (vs lower, reference group)||3.10 ‖ (1.59 to 5.98)|| ||1.78 § (1.10 to 3.16)|| |
| Owns home (vs rents, reference group)||1.95 § (1.12 to 3.41)|| || || |
| Education higher (vs lower reference group)|| ||2.0 ‖ (1.16 to 3.52)||2.68 ‖ (1.58 to 4.58)|| |
| African American (vs caucasian reference)|| || || ||3.15 ‖ (1.37 to 7.21)|
|Primary care attainment|
| Visit continuity|
| No site of care||0.42 (0.12 to 1.45)||0.19 ‖ (0.05 to 0.69)||0.24 § (0.06 to 0.96)||¶|
| Has a site, but no regular clinican||REF||REF||REF|| |
| Has site and regular clinican for some visits||2.66 ‖ (1.29 to 5.52)||1.30 (0.67 to 2.23)||1.49 ‖ (0.82 to 2.71)|| |
| Has site and sees regular clinican most||3.54 ‖ (1.69 to 7.40)||1.82 § (1.10 to 3.28)||2.10 ‖ (1.17 to 3.74)|| |
| Comprehensiveness, all needs met|
| Lower|| ||REF|| || |
| Higher|| ||0.43 § (0.19 to 0.98)|| || |
| Comprehensiveness, counseling|
| Lower||REF||REF|| || |
| Higher||2.70 ‖ (1.50 to 4.83)||2.17 ‖ (1.21 to 3.88)|| || |
| Patient-physician relationship, compassion|
| Lower|| || || ||REF|
| Higher|| || || ||2.91 § (1.10 to 7.78)|
| Patient-physician relationship, trust|
| Lower||REF|| || || |
| Higher||3.11 ‖ (1.30 to 7.43)|| || || |
|Health insurance and plan type|
| Uninsured|| || ||REF||REF|
| Public only|| || ||0.63 (0.31 to 1.28||2.24 (0.64 to 7.89)|
| Private HMO|| || ||2.29 ‖ (1.14 to 4.57)||6.39 ‖ (2.05 to 19.90)|
| Private non-HMO|| || ||1.00 (0.54 to 1.86)||2.22 (0.73 to 6.78)|
Table 4. Factors Significantly Associated with Adherence to Cancer Screening Recommendations for Women ≥65 Years Old ( N = 666) Living in Low-income Census Tracts of Washington, D.C. 2000
|Significant Factors||Adherence to Screening OR * (95% CI)|
|Pap Test||Clinical Breast Exam||Mammogram||FOBT †|
| Knowledge/Attitudes/Beliefs ‡|
| Surgery causes faster growth||0.59 § (0.40 to 0.86)|| || || |
| Prayer alone heals cancer|| ||0.67 ‖ (0.47 to 0.95)|| || |
| Demographic and socioeconomic|
| Owns home (vs rents reference group)|| ||1.61 § (1.10 to 2.34)||1.58 ‖ (1.09 to 2.31)|| |
| Education (higher vs lower reference group)|| || ||1.48 ‖ (1.02 to 2.17)||1.64 ‖ (1.02 to 2.64)|
| Marital status (vs single reference group)||1.60 ‖ (1.02 to 2.49)|| || || |
|Primary care attainment|
| Visit continuity|
| No site of care||0.32 ‖ (0.14 to 0.93)||0.19 § (0.05 to 0.68)||0.34 (0.13 to 0.90)|| |
| Has site but no regular clinician||REF||REF||REF||REF|
| Has site and regular clinician, sees some visits||0.94 (0.50 to 1.74)||0.97 (0.54 to 1.76)||1.38 (0.76 to 2.49)|| |
| Has site and regular clinician, sees most visits||0.76 (0.42 to 1.38)||1.22 (0.69 to 2.16)||1.38 (0.78 to 2.42)|| |
| Comprehensive counseling|
| Higher||1.68 ‖ (1.01 to 2.82)||1.46 ‖ (1.01 to 2.12)||1.53 § (1.09 to 2.15)||1.90 § (1.15 to 3.13)|
| Patient physician relationship communication|
| Lower||REF|| || || |
| Higher||2.37 § (1.31 to 4.28)|| || || |
| Coordination of specialist care ( n = 748)|
| Low|| ||REF||REF||REF|
| Mid|| ||1.47 ‖ (1.00 to 2.17)||1.63 ‖ (1.09 to 2.44)||1.27 (0.84 to 1.93)|
| High|| ||2.36 § (1.44 to 3.87)||1.78 § (1.11 to 2.85)||1.78 § (1.13 to 2.82)|
|Health insurance and plan type|
| Uninsured||1.37 (0.66 to 2.87)||1.75 (0.83 to 3.71)||1.15 (0.56 to 2.35)||0.68 (0.28 to 1.69)|
| Original Medicare (may have Medicaid too)||REF||REF||REF||REF|
| Medicare managed care||1.89 § (1.11 to 3.17)||2.04 § (1.21 to 3.44)||1.95 § (1.15 to 3.31)||1.70 ‖ (1.01 to 2.83)|
| Medicare plus private medigap||1.53 (1.00 to 2.31)||0.89 (0.60 to 1.32)||1.25 (0.83 to 1.87)||1.15 (0.75 to 1.77)|
Overall, 75% of respondents were adherent to regular Pap smears, 66% to clinical breast exams, 65% to mammography, and 29% to fecal occult blood test recommendations. Table 2 presents the unadjusted percentages of women adherent to screening recommendations according to respondents' primary care and insurance characteristics. Continuity of care was significantly associated with adherence to all 4 screening tests. For each test, the largest increase in adherence was found for those who had a usual source of primary care compared with those without one. However, there were also increases in screening adherence among the younger women ( Table 3) who had a specific practitioner whom they saw for most visits at their primary care delivery site, compared with those without a specific practitioner. Likewise, longer relationships with primary care practitioners were significantly associated with higher proportions of women adherent to screening recommendations ( Table 2). Respondents with more “comprehensive” primary care sites in terms of noncancer counseling, screening, and general health services were more likely to be adherent to all cancer screening tests.
Table 2. Screening Adherence by Features of the Primary Care Delivery Site, Insurance, and HMO Participation
| ||Women, n||Pap Test, %||CBE, %||Mammogram, %||FOBT, *%|
|Features of the primary care delivery site|
| Visit continuity with a primary care delivery site and clinician|
| No site of care||34||41.2 †||23.5 †||29.4 †||11.5 †|
| Has a site of care but no regular clinician||149||68.5 †||60.4 †||53.0 †||24.3 ‡|
| Has primary care site and regular clinician but sees for only some visits||391||76.9 †||64.7 †||66.5 †||33.3 ‡|
| Has primary care site and sees same regular clinician for most visits||631||76.7 †||71.5 †||69.7 †||28.9 ‡|
| Duration of the relationship|
| Has had primary care delivery site for <6 mo||71||61.9 †||57.7 ‡||50.7 †||21.0 ‡|
| Has had same primary care site for 6–24 mo||273||73.6 †||64.8 ‡||63.4 †||26.3 ‡|
| Has had same primary care site for >24 mo||826||77.7 †||69.7 ‡||69.0 †||31.9 ‡|
| Access, organizational|
| Lower||531||71.2 †||63.5 ‡||64.6||29.1|
| Higher||674||77.6 †||69.0 ‡||66.2||29.6|
| Access, geographic|
| Lower||588||73.5||63.9 ‡||62.1 †||27.9|
| Higher||598||76.9||69.2 ‡||69.4 †||31.0|
| Access, financial|
| Lower||308||70.8 †||63.3 ‡||62.7||29.1|
| Higher||692||78.7 †||70.6 ‡||66.9||30.0|
| Don't know/missing||205|| || || || |
| Comprehensiveness, all health needs met by regular provider|
| Lower||591||72.6 ‡||63.6 †||62.6 ‡||25.9 ‡|
| Higher||575||78.3 ‡||71.0 †||69.4 ‡||32.8 ‡|
| Comprehensiveness of non-cancer screening tests|
| Lower||190||55.8 †||40.5 †||47.4 †||13.8 †|
| Higher||1,007||78.9 †||72.0 †||69.4 †||32.2 †|
| Comprehesiveness, counseling, around diet, alcohol and tobacco|
| Lower||369||65.6 †||58.5 †||60.4 †||25.3 ‡|
| Higher||831||79.0 †||70.4 †||68.1 †||31.5 ‡|
| Patient-physician relationship, compassion|
| Lower||648||70.0 †||62.0 †||61.6 †||26.7 ‡|
| Higher||557||80.0 †||71.8 †||70.0 †||32.5 ‡|
| Patient-physician relationship, trust|
| Lower||321||72.6||62.0 ‡||63.8||25.2|
| Higher||871||75.9||68.3 ‡||66.5||30.9|
| Patient-physician relationship, communication|
| Lower||555||70.6 †||62.9 †||63.1||28.2|
| Higher||635||79.2 †||70.2 †||68.4||31.1|
|Coordination of specialist care ( n = 748)|
| Low||133||75.9||63.2 †||63.2||20.0 †|
| Mid||354||79.7||69.2 †||71.7||32.7 †|
| High||267||80.9||77.9 †||70.4||37.4 †|
|Health system/insurance status|
| <65 years old|
| Uninsured||70||71.4 †||61.4 †||50.0 †||10.0 †|
| Public only (Medicaid § and/or Medicare only)||88||78.4 †||63.6 †||43.2 †||26.5 †|
| Private HMO||145||90.3 †||75.8 †||80.0 †||43.4 †|
| Other private ‖ (may have public too)||229||84.7 †||78.6 †||65.9 †||23.6 †|
| ≥65 years old|
| Uninsured||41||60.9 †||58.5 ‡||53.7 †||17.1 †|
| Original Medicare (may have Medicaid too)||210||60.0 †||55.2 ‡||59.0 †||25.7 †|
| Medicare managed care||117||74.3 †||70.9 ‡||75.2 †||39.3 †|
| Medicare + private Medigap||295||72.5 †||61.7 ‡||70.5 †||31.9 †|
Logistic Regression Analyses
Tables 3 and 4 present the adjusted odds ratios from the logistic regression models done separately for women age <65 and ≥65. For women under age 65, continuity with a place and with a specific clinician was associated ( P≤ .01) with screening adherence for Pap tests, CBEs, and mammograms. For women ≥65, continuity of place was associated with adherence to Pap tests, CBEs, and mammograms. Respondents whose primary care delivery sites were more “comprehensive” were more likely to be adherent with recommended Pap tests and CBEs if aged <65 years, and with all 4 tests if ≥65 years. Better coordination of specialist care outside of the office was also associated with CBEs, mammograms and FOBT adherence among older women. The lack of association between coordination of specialty care and screening adherence among younger women was probably due to their better self-assessed health status and lower use specialists or of hospitalized care compared with older women's, and thus to their less-frequent responses to the items on coordination of care.
Although having any private insurance was significantly associated with adherent screening in the unadjusted analyses, once the primary care variables were entered into the models, only 1 of the insurance categories, i.e., private HMO, remained significant. Since we were interested in further exploring whether the higher adherence to cancer screening among private HMO enrollees was associated with the primary care performance of those arrangements, we compared women's reports of primary care performance across insurance and plan types. Women in private HMOs (<65 years old) and Medicare HMOs (≥65 years old) had lower ratings of their primary care characteristics than did women in private indemnity plans. For example, 37.6% of women in HMOs rated their continuity as highest whereas 64.3% of women in private indemnity plans rated their continuity as highest ( P = .001). Only 51.9% of women in HMOs rated their sites' organizational accessibility as highest versus 66.9% of women in private indemnity plans ( P = .001). Similar differences existed for the other characteristics of primary care: trust in the regular physician (60.5% HMO vs 77.5% private fee-for-service [FFS]), coordination of specialty care (27.4% HMO vs 41.4% private FFS), and comprehensiveness of services (42% HMO vs 55% in private FFS) ( P = .001 for each comparison).
Finally, we re-ran the final logistic regression models to examine whether there were differences for women whose regular clinician was an obstetrician-gynecologist versus other type of clinician. Inclusion of this covariate did not change the relationships between the primary care variables and screening adherence.
- Top of page
This study is the first to examine the role of specific characteristics of primary care delivery in adherence to cancer screening while incorporating a rich array of attitudinal, socioeconomic, and insurance barriers to screening for a population-based sample of inner-city, mostly low-income women. Despite numerous barriers, women with primary care delivery sites characterized by more continuity of care, comprehensive services, and coordination were more adherent to regular cancer screening. Finally, being in a private HMO was the only insurance category significantly associated with screening adherence after controlling for the primary care characteristics of women's delivery sites.
Additionally, higher levels of patient-clinician trust, compassion, and communication were associated with adherence to cancer screening for certain screening tests. It is possible that the 2 tests not done in the office, mammograms and FOBTs, might be associated with communication, but because they require an additional step for utilization (outside referral to a mammogram facility; and return of test cards for a home FOBT), they are more complicated than tests completely performed in the primary care office. Hence this added step may mediate the impact of communication on utilization for mammograms and FOBT. The answer to this question requires additional research.
While insurance was associated with screening adherence in unadjusted analyses, after controlling for primary care characteristics, being in a private HMO was the only insurance category that remained significant. Unlike the other insurance categories, the HMO category encompasses both insurance coverage and structural aspects of that subgroups' health care. Women in private HMOs and Medicare HMOs rated those sites significantly poorer on primary care performance (accessibility, continuity, comprehensiveness, coordination, patient-physician relationship) than did women in private indemnity plans. So, it is unlikely that the higher cancer screening rates among HMO participants were due to stronger primary care performance by those settings. One might speculate that women in HMOs were more likely to be adherent to screening than were women in nonHMOs because of participation by managed care plans in the Health Plan and Employer Data Information Set (HEDIS 3.0 and HEDIS 1999–2000 Reporting Sets; Washington, D.C.; National Committee for Quality Assurance) provides them incentives to emphasize cancer screening. This finding that HMOs perform well on the delivery of cancer screening services is supported by other studies. 53,54
In conjunction with the primary care findings of this study, the additional finding on insurance suggests several things about its relationship to cancer screening. Having insurance, in general, is an important facilitator of entry into healthcare. However, it seems that the characteristics—i.e., structure and process—of a woman's primary care delivery system are also important for assuring adherence to screening for low-income women. Among the primary care characteristics, continuity of care appears to have an impact on cancer screening adherence. Overall, this study adds to the growing literature suggesting that delivery system characteristics, in addition to the presence of insurance, are associated with preventive services use. 27,55
For women over 65, we did not find that financial accessibility of primary care, measured in this study as the perception of out-of-pocket costs for physician visits, prescriptions, and prescribed treatments, influenced screening utilization. Prior research found that lower cost-sharing is associated with greater use of preventive services. 56 However, state medical assistance, via the Qualified Medicare Beneficiary Program and the Specified Low-Income Medicare Beneficiary Program, helps low-income women with Medicare Part B premiums, deductibles, and coinsurance. 57 For women under 65, the National Breast and Cervical Cancer Early Detection Program's coverage of cervical and breast cancer screening for qualified women was probably an important contributor 58 to the absence of an association between perceived financial access and screening.
Because this study identified women randomly in the community, rather than using a convenience sample of clinic attendees, it has public health policy relevance, because it focuses on all women, not just those attending medical clinics or offices. This study also has limitations. These data may not generalize to persons without telephones or in rural areas. It is estimated that 94% of African-American households in the District of Columbia have phones. 59 The cross-sectional design limits interpretation of the directionality between independent variables and screening utilization. For example, women with a more recent CBE may be more likely to report having a more continuous relationship with a provider. However, prior work has found a potentially causal relationship between having a usual source of care and the receipt of screening services. 27 In addition, the exposure-response relationship between some of the primary care variables and adherence provides further support for a potentially causal association.
Although widely used and informative, self-report data generally overestimate screening rates. 60,61 We did not do a medical record validation of women's self-reported screening utilization because women received care at various different sites throughout the District of Columbia and it was not possible to do such a widespread medical record validation. Because urban and especially lower-income populations are mobile, women who received tests at various facilities may be undercounted or underrepresented in a primary care medical chart validation. 62 Characteristics that might influence self-report validity, such as education and socioeconomic status, were controlled for in analyses. Furthermore, the purpose of our study was not to validate screening rates, but to assess associations between primary care and self-reported screening. For a self-report bias to be operative, it needs to be unequally distributed among primary care groups. Although we think that this is unlikely, 63 we cannot rule out the possibility that women with a regular provider might be more likely to report adherence out of loyalty to that doctor.
The rates of screening adherence (last 2 tests at recommended intervals for age) for our entire sample were appropriately lower than rates of recent screening from a Behavioral Risk Factor Surveillance System (BRFSS) sample of women over age 40 in Washington, D.C. For example, rates of a “recent” mammogram (1 mammogram in the past 2 years) in the BRFSS for women over age 40 were about 85%. The overall rate of mammography adherence in our sample was 65.5%. Because we asked about not just the most recent mammogram but also the one prior to the most recent (adherence) we expected our percentages for mammography adherence to be somewhat lower, and they were. Similarly reassuring comparisons can be made for the other tests between our data and BRFSS data for women in Washington, D.C. 64
The validity of respondents' classification of their insurance status also cannot be ascertained. The percentage of our sample aged ≥65 with Medicare managed care (17.6%) was higher than among all women age ≥65 in Washington, D.C. (10.4%). 65 This probably reflects the fact that Medicare managed care plans primarily serve urban seniors with incomes under $20,000, 66 a group targeted by our sampling strategy.
We did not determine the proportion of our sample receiving care from both a primary care clinician and a gynecologist, the so-called ‘dual PCP’ arrangement. The likelihood of obtaining screening tests might increase among those with both types of clinicians, while continuity of care would decrease because of the need to see multiple providers for primary care. Nationally, about one third of women over age 18 receive care from both a gynecologist and a primary care provider. 67 This percentage is likely to be much lower in older, lower income women in medically underserved areas. 67,68 Finally, while the sample size of this study was adequate to detect subgroup differences in receipt of Pap smears, CBEs, and mammograms, we lacked power to fully assess all relationships between the independent variables of interest and FOBT.
An important association in our study was between continuity with a specific primary care clinician and cancer screening adherence. We calculated attributable risks to provide measures of the maximum amount that screening adherence could change if nonelderly women without a specific primary care clinician were linked to a specific clinician at their primary care delivery sites. Our findings revealed that if there were a causal association between continuity and screening, Pap smear adherence would increase by 30%, CBE by 15% and mammography by 12% among women without a primary care clinician. These are the maximum effects potentially attainable by reorganizing primary care services to promote continuity for inner-city, low-income women. Clearly continuity is not the only factor related to screening adherence. But our results indicate that there are potentially clinically meaningful effects of continuity. Future research needs to determine, through longitudinal observation studies or controlled clinical trials, the extent to which continuity impacts on screening adherence.
Although progress has been made in narrowing the gap in screening rates between minority and nonminority populations, barriers to screening persist even among the insured. Assessment of the relationship between mutable features of primary care that promote ongoing screening will help to target intervention efforts. For example, not all insured persons have their cancer screening coordinated through a primary care provider. Some obtain screening in mass screening programs but fail to obtain coordinated follow-up and repeat screening. The relationships found in this study between continuity of care and adherence to screening over time suggest that absence of such primary care features in screening programs may result in poorer adherence. Health care education aimed at patients should stress the importance of identifying a primary care provider and of obtaining screening through that provider.
Another implication of our study is that efforts to eliminate the disparities in late-stage cancer among lower income and minority women might focus on developing performance assessment tools to include process indicators of the quality of primary care delivery, (e.g., continuity of care, comprehensive service delivery). Disease-specific performance measures do not create adequate incentive among health plans to deliver “optimal” primary care to their members.
Our results also suggest that efforts to decrease the disparity in potentially avoidable cancer morbidity and mortality among low-income women should take a broad perspective of their health care systems. Cancer screening needs to be considered in the context of the totality of a woman's health care needs and in the context of the practitioner with whom she has an ongoing relationship. Regardless of the barriers women face to screening, specific features of primary care delivery are associated with adherence to screening for this at-risk group. Thus, providing health insurance is a necessary but not sufficient step toward improving adherence to recommended cancer screening services. Efforts to improve screening adherence via increased insurance coverage must be accompanied by additional research on the effectiveness of increasing primary care attainment among lower-income women.