SEARCH

SEARCH BY CITATION

Keywords:

  • breast neoplasms;
  • SEER;
  • Medicare;
  • primary health care;
  • mammography

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SUPPORT
  8. CONFLICT OF INTEREST DISCLOSURES
  9. REFERENCES

BACKGROUND

Primary care physician (PCP) services may have an impact on breast cancer mortality and incidence, possibly through greater use of screening mammography.

METHODS

The authors conducted a retrospective, 1:1 matching case-control study using the Surveillance, Epidemiology, and End Results (SEER)-Medicare–linked database to examine use of PCP services and their association with breast cancer mortality and incidence. SEER cases representing the 3 outcomes of interest (breast cancer mortality, all-cause mortality among women diagnosed with breast cancer, and breast cancer incidence) were matched to unaffected controls from the 5% Medicare random sample. Conditional logistic regression was used to examine associations between physician visits and breast cancer outcomes while controlling for other covariates.

RESULTS

Women who had 2 or more PCP visits during the 24-month assessment interval had lower odds of breast cancer mortality, all-cause mortality, and late-stage breast cancer diagnosis compared with women who had no PCP visits or 1 PCP visit while adjusting for other covariates, including mammography and non-PCP visits. Women who had 5 to 10 PCP visits had 0.69 times the odds of breast cancer mortality (95% confidence interval, 0.63-0.75), 0.83 times the odds of death from any cause having been diagnosed with breast cancer (95% confidence interval, 0.79-0.87), and 0.67 times the odds of a late-stage breast cancer diagnosis (95% confidence interval, 0.61-0.73) compared with those who had no PCP visits or 1 PCP visit.

CONCLUSIONS

The current findings suggest that PCPs play an important role in reducing breast cancer mortality among the Medicare population. Further research is needed to better understand the impact of primary care on breast cancer and other cancers that are amendable to prevention or early detection. Cancer 2013;119:2964-72. © 2013 American Cancer Society.


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SUPPORT
  8. CONFLICT OF INTEREST DISCLOSURES
  9. REFERENCES

Primary care physicians (PCPs) play an important role in reducing breast cancer (BC) mortality by detecting BC at earlier, more favorable stages.[1] There is significant empiric evidence that preventive care in general, and cancer screening in particular, is overwhelmingly delivered in the ambulatory setting.[2-4] The recommendations by PCPs for screening are often one of the strongest predictors of BC screening.[5, 6] PCPs also may reduce BC mortality by preventing diagnostic delays after screening.[7]

Understanding the effects of primary medical care is important because of the looming shortage of PCPs in the United States. Medical student interest in primary care[8] and the proportion of US physicians engaged in primary care[9] are both in decline. This problem may be worsened by heightened demand for PCP services expected under health care reform.[10]

Previous ecologic studies that linked higher PCP supply with improved cancer outcomes[11-14] are subject to an ecologic fallacy: it is not possible to determine whether individuals who had better outcomes in those studies are the same individuals who received care from PCPs. Cohort studies linking PCP visits with improved cancer survival also are subject to lead-time bias, in which longer survival is not associated with true reductions in mortality.[1, 15] Because of these limitations, it is unclear how PCP services impact cancer outcomes at the population level.

The current study overcomes the limitations of previous studies by examining the impact of the number of PCP visits on population-based measures of BC mortality and incidence among Medicare beneficiaries using the linkage of 2 large, population-based data sources. We extend our prior work by examining the association of PCP use with BC mortality and incidence among Medicare beneficiaries.[1] We hypothesize that PCP visits will have a protective effect against mortality, because PCP visits may increase detection at an earlier stage as a result of greater mammography use. Similarly, we hypothesize that PCP visits will decrease the incidence of late-stage cancers but will increase the incidence of early stage cancers (because of greater mammography screening).[1, 16] In addition, we hypothesize that PCP visits will increase the incidence of estrogen receptor (ER)-positive cancers (because long-term hormone replacement therapy [HRT] is associated with ER-positive tumors).[17, 18] Understanding the relation between primary medical care and cancer outcomes is important in preparing for policy changes necessary to improve our nation's health care system.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SUPPORT
  8. CONFLICT OF INTEREST DISCLOSURES
  9. REFERENCES

Study Design

For this study, we used a retrospective, 1:1 matching case-control study design to examine the relation between PCP visits and 3 outcomes: 1) BC-specific mortality, 2) all-cause mortality once diagnosed with BC, and 3) BC incidence using the 2008 Surveillance, Epidemiology, and End Results (SEER)-Medicare linked data set (available at: http://healthservices.cancer.gov/seermedicare/) (accessed April 25, 2012). This study was approved by the institutional review boards of the University of South Florida and Beth Israel Deaconess Medical Center. In our study, we examined PCP visits for cases and controls during a 2-year interval before the case's month of diagnosis. Because physician visit patterns are likely to change during the time that a potential BC is being diagnosed, we excluded the 3-month period immediately before the month of SEER diagnosis in our analysis and assessed physician claims during the preceding 24-month period (ie, the period 4-27 months before diagnosis).[19] Previous studies suggest that the overwhelming majority of patients complete diagnostic evaluations within this 3-month prediagnosis period.[7, 20] Our measures of PCP visits, thus, are intended to reflect routine care rather than care specifically associated with diagnosing BC.

Study Population:

Inclusion criteria: Cases

Women newly diagnosed with BC within the SEER Program from April 1994 to December 2005 represented the cohort of interest for identifying cases (n = 309,616). We excluded women without 27 months of continuous Medicare fee-for-service (FFS) insurance before diagnosis and whose diagnosis was not within their first continuous FFS interval from 1992 to 2005 (n = 190,641). Because screening procedures generally are not recommended for women aged <40 years or >85 years, we excluded those who were diagnosed before age 40 years or after age 85 years (n = 13,933), yielding a final number of 105,042 cases for the BC incidence analysis.

For our assessment of PCP visits on BC-specific mortality, we also excluded cases who did not die from BC by December 31, 2005, the last available date for cause-of-death information, yielding 10,369 cases for the BC-specific mortality analysis. For our assessment of PCP visits on all-cause mortality, we excluded incident BC cases who were alive at December 31, 2007, which was the latest available date of death, leaving 38,402 all-cause mortality cases.

Inclusion criteria: Controls

The candidate controls came from the 5% random sample of Medicare beneficiaries residing in the SEER registries and represented those who were at risk but did not develop the outcome of interest. Women whose first continuous FFS interval between 1992 and 2005 was longer than 27 months were considered. Those beneficiaries who did not develop BC through 2007 were eligible controls for the incidence analysis (n = 269,962). Those who did not die of BC by December 31, 2005 were eligible controls for the BC mortality analysis (n = 274,832). Finally, those who were alive at December 31, 2007 were eligible controls for the all-cause mortality analysis (n = 171,053).

Matching Criteria 1:1 for Cases and Controls

Cases were matched to candidate controls using a greedy algorithm,[21] matching each case to 1 control who had the same original reason for Medicare entitlement (age, disability, end-stage renal disease), who lived in the same SEER registry as the case during the year of the case's diagnosis, whose birth year was within 5 years of the case's, and who had at least 27 months of continuous FFS Medicare enrollment before the case's BC diagnosis month.

For the incidence and all-cause mortality analyses, 425 (0.4%) and 664 (1.7%) cases were not matched, respectively, because no available controls met the matching criteria. Unmatched cases were excluded from the analyses.

Physician Visits

We examined Medicare claims (National Claims History) for the following ambulatory-based evaluation and management services representing routine office visits: codes 99201 through 99205 and codes 99211 through 99215. Similar to prior research,[22, 23] we identified the physician specialty associated with each claim using the Medicare provider specialty field in National Claims History claims. We defined PCPs as providers who had the following specialties: general practice, family medicine, primary care internal medicine, geriatric medicine, and obstetrics-gynecology. We classified obstetrics-gynecology as primary care, because 64.3% of visits to such physicians by older women are for routine follow-up or preventive care. All other specialties were considered nonprimary care providers (non-PCPs). For each case and control, we determined the number of ambulatory claims for PCP and non-PCP office visits during the 4-month to 27-month period before the case's BC diagnosis. The numbers of PCP and non-PCP visits were categorized into 4 groups: 0 to 1 visits, 2 to 4 visits, 5 to 10 visits, and ≥11 visits (quartiles of PCP visits).

In addition, we determined if either a mammography procedure was done during this 4-month to 27-month period. Because codes for screening and diagnostic mammography often are used interchangeably, codes alone cannot be relied on to determine whether the test was done for screening or diagnosis.[19] We defined these procedures as Healthcare Common Procedure Coding System/Current Procedural Terminology codes G0202 through G0207, 76092, 77055, 77056, and 77052.

Statistical Analysis

Bivariate associations between 2 categorical variables were tested using Pearson chi-square tests. The relations between PCP visits and mammography receipt, BC-specific mortality, all-cause mortality, and BC incidence were examined using conditional logistic regression analysis conditioned on matched pairs.

We considered the following variables as potential confounders in multivariable models: race-ethnicity (reported by Medicare), age at case's diagnosis, census-derived measures of zip-code level median household income and educational attainment (percentage of individuals with less than a high school education) categorized by quintiles within each registry, metropolitan statistical area, Charlson comorbidity index[24, 25] (determined from both inpatient and outpatient claims), prior influenza vaccination in the 4-month to 27-month interval (as a marker of preventive behaviors),[15, 26] and the number of non-PCP visits. Missing values were not imputed. Because of the large sample size and modeling constraints, patients who were missing any confounders were excluded from the multivariable models.

Stratified analyses were conducted to investigate potential effect modifiers: age at diagnosis (ages 40-64 years, 65-75 years, and 76-85 years), ER status of the tumor (positive, negative), and stage at diagnosis (ductal carcinoma in situ [DCIS], early, late). Early stage disease was defined as stage I or II (models also were refit to include DCIS), and late stage was defined as stage III or IV.

All tests and 95% confidence intervals (CIs) were 2-sided with a significance level of .05. We adjusted for multiple comparisons of the P values of the 3 primary outcomes in the overall and stratified multivariable models using the step-down Bonferroni method. All analyses were performed using SAS software (version 9.3; SAS Institute Inc., Cary, NC).

RESULTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SUPPORT
  8. CONFLICT OF INTEREST DISCLOSURES
  9. REFERENCES

Sample Characteristics

The median age at cancer diagnosis was 74 years (interquartile range [IQR], 70-79 years). Most women who died of BC had late-stage incident cancers (stage III, 15.7%; stage IV, 27.1%), whereas fewer had early stage cancers (stage I, 10.3%; stage II, 26.9%) or DCIS (1%). Only 9.2% of incident cancers overall were ER-negative, whereas 20.4% of those who died of BC and 12.6% of those who died of any cause had incident ER-negative cancers.

For the BC mortality analysis, cases and controls visited a PCP within the 2 years before the case's month of diagnosis a median of 5 times (IQR, 1-11 visits). Non-PCP visits were fewer, with a median of 2 visits (IQR, 0-7 visits). The percentage of BC mortality cases with no PCP visits or 1 PCP visit was 36.2% and decreased to 26.4% and 21.7% for the all-cause mortality and incidence analyses, respectively. Conversely, the percentage of controls who had no PCP visits or 1 PCP visit remained relatively stable for all 3 analyses at 23.9%, 22.9%, and 23.5% for BC mortality, all-cause mortality, and incidence, respectively.

Breast Cancer-Specific Mortality

Having fewer than 2 PCP or non-PCP visits, being of African American race, living in a zip code in the most uneducated quintile or in the highest quintile of median household income, not receiving an influenza vaccination, and not receiving a mammogram were associated independently with increased BC-specific mortality (Table 1). When adjusting for potential confounders, all categories of multiple PCP visits decreased BC-specific mortality compared with no visits or 1 visit (2-4 visits: odds ratio [OR], 0.77; 95% CI, 0.70-0.84; 5-10 visits: OR, 0.69; 95% CI, 0.63-0.75; ≥11 visits: OR, 0.68; 95% CI, 0.62-0.74; all adjusted P < .0001).

Table 1. Associations With Mortality and Incidence
 OR (95% CI)
 BC Mortality, N = 20,738All-Cause Mortality, N = 75,476Incidence, N = 209,234
VariableUnadjustedAdjustedUnadjustedAdjustedUnadjustedAdjusted
  1. Abbreviations: BC, breast cancer; CI, confidence interval; OR, odds ratio; PCP, primary care physician; Ref, referent category.

PCP visits
0-11 (Ref)1 (Ref)1 (Ref)1 (Ref)1 (Ref)1 (Ref)
2-40.62 (0.57-0.67)0.77 (0.70-0.84)0.73 (0.69-0.76)0.82 (0.78-0.86)1.12 (1.09-1.15)1.08 (1.05-1.11)
5-100.53 (0.50-0.57)0.69 (0.63-0.75)0.77 (0.74-0.80)0.83 (0.79-0.87)1.13 (1.11-1.16)1.09 (1.07-1.12)
≥110.51 (0.48-0.56)0.68 (0.62-0.74)0.99 (0.95-1.03)0.89 (0.85-0.94)1.08 (1.05-1.11)1.06 (1.03-1.09)
Non-PCP visits
0-11 (Ref)1 (Ref)1 (Ref)1 (Ref)1 (Ref)1 (Ref)
2-40.69 (0.64-0.74)0.85 (0.78-0.92)0.93 (0.90-0.97)0.96 (0.91-1.00)1.16 (1.14-1.19)1.15 (1.12-1.18)
5-100.65 (0.60-0.70)0.82 (0.75-0.90)0.99 (0.95-1.03)0.96 (0.92-1.00)1.22 (1.20-1.25)1.21 (1.18-1.24)
≥110.66 (0.61-0.72)0.84 (0.77-0.93)1.30 (1.25-1.36)1.12 (1.06-1.17)1.29 (1.25-1.32)1.27 (1.23-1.30)
Age      
1-Y increase0.99 (0.93-1.06)0.99 (0.92-1.06)1.60 (1.57-1.64)1.59 (1.55-1.63)1.06 (1.04-1.08)1.06 (1.05-1.08)
Race/ethnicity      
White1 (Ref)1 (Ref)1 (Ref)1 (Ref)1 (Ref)1 (Ref)
Asian0.50 (0.40-0.62)0.43 (0.34-0.54)0.33 (0.29-0.36)0.26 (0.23-0.29)0.42 (0.40-0.45)0.40 (0.38-0.43)
Black1.32 (1.20-1.46)1.22 (1.08-1.36)1.12 (1.06-1.18)0.97 (0.90-1.03)0.90 (0.87-0.93)1.00 (0.97-1.04)
Hispanic0.61 (0.49-0.76)0.57 (0.45-0.72)0.49 (0.43-0.55)0.38 (0.34-0.44)0.51 (0.48-0.55)0.53 (0.49-0.57)
North American Native0.45 (0.27-0.77)0.36 (0.20-0.63)0.66 (0.49-0.88)0.48 (0.35-0.66)0.52 (0.44-0.62)0.59 (0.50-0.71)
Other0.75 (0.58-0.98)0.68 (0.51-0.91)0.51 (0.44-0.59)0.40 (0.34-0.47)0.56 (0.52-0.61)0.56 (0.51-0.60)
Unknown1.93 (1.00-3.69)1.63 (0.82-3.26)0.90 (0.66-1.22)0.69 (0.49-0.98)0.86 (0.69-1.07)0.83 (0.66-1.04)
Charlson Comorbidity Index      
01 (Ref)1 (Ref)1 (Ref)1 (Ref)1 (Ref)1 (Ref)
10.91 (0.85-0.97)1.04 (0.96-1.12)1.69 (1.63-1.76)1.78 (1.70-1.85)0.98 (0.96-1.00)0.94 (0.92-0.96)
≥21.06 (0.98-1.14)1.19 (1.09-1.31)3.31 (3.17-3.47)3.29 (3.12-3.47)0.90 (0.87-0.92)0.85 (0.83-0.87)
Metropolitan statistical area      
Large metropolitan1 (Ref)1 (Ref)1 (Ref)1 (Ref)1 (Ref)1 (Ref)
Less urban0.84 (0.73-0.96)0.92 (0.79-1.08)0.74 (0.69-0.80)0.80 (0.73-0.87)0.75 (0.72-0.78)0.89 (0.86-0.94)
Metropolitan0.83 (0.76-0.91)0.87 (0.79-0.96)0.86 (0.82-0.90)0.88 (0.83-0.93)0.86 (0.84-0.88)0.91 (0.88-0.93)
Rural0.81 (0.65-1.02)0.82 (0.63-1.05)0.69 (0.61-0.78)0.73 (0.64-0.84)0.73 (0.68-0.79)0.91 (0.85-0.98)
Urban0.84 (0.74-0.96)0.89 (0.77-1.03)0.84 (0.78-0.90)0.88 (0.81-0.95)0.84 (0.81-0.88)0.96 (0.91-1.00)
Education at zip-code level      
Quintile 11 (Ref)1 (Ref)1 (Ref)1 (Ref)1 (Ref)1 (Ref)
Quintile 20.77 (0.70-0.84)0.73 (0.66-0.80)0.83 (0.80-0.87)0.77 (0.73-0.82)0.87 (0.85-0.90)0.79 (0.77-0.82)
Quintile 30.72 (0.66-0.78)0.64 (0.57-0.71)0.83 (0.80-0.87)0.72 (0.68-0.76)0.91 (0.88-0.93)0.72 (0.70-0.75)
Quintile 40.73 (0.67-0.79)0.59 (0.53-0.66)0.79 (0.75-0.82)0.61 (0.58-0.65)0.94 (0.92-0.97)0.66 (0.64-0.68)
Quintile 5: Highest0.68 (0.63-0.75)0.46 (0.41-0.52)0.75 (0.72-0.79)0.52 (0.48-0.55)1.00 (0.97-1.03)0.57 (0.55-0.59)
Income at zip-code level      
Quintile 11 (Ref)1 (Ref)1 (Ref)1 (Ref)1 (Ref)1 (Ref)
Quintile 20.97 (0.88-1.06)1.22 (1.10-1.35)1.05 (1.00-1.10)1.23 (1.16-1.30)1.14 (1.11-1.17)1.26 (1.22-1.30)
Quintile 30.97 (0.88-1.05)1.35 (1.21-1.51)1.02 (0.98-1.07)1.28 (1.21-1.36)1.20 (1.17-1.23)1.39 (1.35-1.44)
Quintile 40.99 (0.91-1.09)1.54 (1.37-1.74)1.05 (1.00-1.10)1.49 (1.39-1.59)1.25 (1.21-1.28)1.56 (1.51-1.62)
Quintile 5: Richest1.21 (1.11-1.32)2.15 (1.89-2.45)1.21 (1.16-1.27)1.89 (1.76-2.03)1.59 (1.55-1.64)2.07 (1.99-2.15)
Influenza vaccination      
No1 (Ref)1 (Ref)1 (Ref)1 (Ref)1 (Ref)1 (Ref)
Yes0.66 (0.63-0.70)0.89 (0.84-0.95)0.92 (0.89-0.95)0.96 (0.92-0.99)1.14 (1.12-1.16)1.10 (1.08-1.12)
Mammogram      
No1 (Ref)1 (Ref)1 (Ref)1 (Ref)1 (Ref)1 (Ref)
Yes0.33 (0.30-0.35)0.38 (0.35-0.41)0.49 (0.48-0.51)0.51 (0.49-0.53)1.00 (0.98-1.01)0.91 (0.89-0.93)

The effect of PCP visits on BC-specific mortality stratified by potential effect modifiers are illustrated in Figure 1. Most categories of 2 or more PCP visits decreased BC mortality compared with no visits or 1 visit for most stratified analyses, and the most protective effect was observed with late-stage cancer. However, PCP visits did not affect BC mortality in those who had an early stage diagnosis or a diagnosis before age 65 years.

image

Figure 1. This chart illustrates the effect of primary care physician (PCP) visits on breast cancer (BC)-specific mortality for stratified analyses using multivariable conditional logistic regression. ER indicates estrogen receptor; DCIS, ductal carcinoma in situ; OR, odds ratio; CI, confidence interval.

Download figure to PowerPoint

All-Cause Mortality

All-cause mortality among women with a previous BC diagnosis was independently associated with increased age, being of African American race, having more comorbidities, living in a zip code in the most uneducated quintile, not receiving an influenza vaccine, and not receiving a mammogram (Table 1). In the overall adjusted analysis, the odds of getting diagnosed with BC and dying from any cause, compared with being alive through 2007 with or without BC, were lower for those who had 2 or more PCP visits in a 2-year span compared with those who had no visits or 1 visit (2-4 visits: OR, 0.82; 95% CI, 0.78-0.86; adjusted P < .0001; 5-10 visits: OR, 0.83; 95% CI, 0.79-0.87; adjusted P < .0001; ≥11 visits: OR, 0.89; 95% CI, 0.85-0.94; adjusted P = .0003).

Figure 2 demonstrates the effect of PCP visits on all-cause mortality for the stratified analyses. PCP visits were most protective against all-cause mortality with late-stage cancer and did not affect all-cause mortality in those with a BC diagnosis before age 65 years, whereas ≥11 PCP visits were associated with increased mortality among those with early stage BC.

image

Figure 2. This chart illustrates the effect of primary care physician (PCP) visits on all-cause mortality for stratified analyses using multivariable conditional logistic regression. ER indicates estrogen receptor; DCIS, ductal carcinoma in situ; OR, odds ratio; CI, confidence interval; BC; breast cancer.

Download figure to PowerPoint

Incidence and Mammography

The unadjusted results provided in Table 1 indicate that incidence was independently associated with having more than 1 PCP visit, having more than 1 non-PCP visit, having no comorbidities, being older, and living in a zip code with higher income. When adjusting for potential confounders, all categories of PCP visits increased the incidence of BC compared with no PCP visits or 1 PCP visit in a 2-year span. The effects of PCP visits on incidence stratified by potential effect modifiers are illustrated in Figure 3. Having 2 or more PCP visits increased the incidence of ER-positive tumors but was not associated with the incidence of ER-negative tumors. Having 2 or more PCP visits increased the incidence of DCIS and early stage BC but decreased the incidence of late stage BC.

image

Figure 3. This chart illustrates the effect of primary care physician (PCP) visits on breast cancer (BC) incidence for stratified analyses using multivariable conditional logistic regression. ER indicates estrogen receptor; DCIS, ductal carcinoma in situ; OR, odds ratio; CI, confidence interval.

Download figure to PowerPoint

Only 39.5% of women in the incidence analysis received a mammogram in a 2 year interval, and there was no statistical difference in the receipt of mammography between cases and controls (P = .66). Mammography use dwindled to 15.6% and 24% for cases who died of BC or of any cause, respectively. For mammography to be considered a mediator for PCP visits on incidence, mammography receipt must be associated with PCP visits. Increased PCP visits, compared with no visits or 1 visit, was associated with a greater likelihood of receiving a mammogram (2-4 visits: OR, 2.96; 95% CI 2.83-3.10; 5-10 visits: OR, 3.79; 95% CI, 3.63-3.95; ≥11 visits: OR, 3.67; 95% CI, 3.52-3.83).

Multivariable models were fit without and then with mammography receipt. The extent to which the ORs for PCP effects were attenuated after including mammography indicates how much this effect can be explained by mammography receipt. The addition of mammography attenuated the effects of PCP visits on incidence of DCIS and late-stage cancer (Table 2), although the effects remained statistically significant. Mammograms were protective against late-stage cancer (OR, 0.37; 95% CI, 0.35-0.40) but increased the incidence of DCIS (OR, 1.37; 95% CI, 1.30-1.45). Mammography did not attenuate the effect of PCPs on the incidence of early stage disease (stages I and II) or on the other potential effect modifiers (age at diagnosis, ER status).

Table 2. Effect of Primary Care Physician Visits and Mammography on the Incidence of Ductal Carcinoma In Situ and Late-Stage Breast Cancer
 OR (95% CI)
 DCIS, N = 30,206Late Stage, N = 20,614
VariableUnadjustedAdjusted Without MammographyaAdjusted With MammographyaUnadjustedAdjusted Without MammographyaAdjusted With Mammographya
  1. Abbreviations: CI, confidence interval; DCIS, ductal carcinoma in situ; MSA, metropolitan statistical area; OR, odds ratio; PCP, primary care physician; Ref, referent category.

  2. a

    Models were adjusted for non-PCP visits, age, race/ethnicity, comorbidities, MSA of residence, receipt of flu vaccination, and zip code-level education and income.

PCP visits      
0-11 (Ref)1 (Ref)1 (Ref)1 (Ref)1 (Ref)1 (Ref)
2-41.55 (1.44-1.67)1.43 (1.33-1.55)1.36 (1.25-1.47)0.56 (0.51-0.60)0.62 (0.57-0.68)0.71 (0.65-0.78)
5-101.67 (1.56-1.78)1.50 (1.39-1.61)1.39 (1.30-1.50)0.48 (0.44-0.51)0.56 (0.51-0.61)0.67 (0.61-0.73)
≥111.63 (1.53-1.75)1.49 (1.38-1.60)1.39 (1.29-1.50)0.47 (0.43-0.50)0.58 (0.53-0.63)0.70 (0.64-0.77)
Mammography receipt      
Yes vs no1.64 (1.56-1.72)1.37 (1.30-1.45)0.31 (0.29-0.34)0.37 (0.35-0.40)

DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SUPPORT
  8. CONFLICT OF INTEREST DISCLOSURES
  9. REFERENCES

Strengths/Importance

Our study had numerous strengths, including a population-based design and large sample sizes, giving us high power to detect small effects, even when adjusting for multiple comparisons. Conducting stratified analyses allowed us to explore relevant and biologically plausible mechanisms and associations. Overall, PCP visits significantly affected our 3 outcomes (BC-specific mortality, all-cause mortality, and BC incidence), even when adjusting for non-PCP visits and other factors. Our results suggest that ensuring access to PCPs is important in the early detection of BC and reduction of BC mortality among the US Medicare population. However, ensuring a PCP visit may prove challenging, even within this universally insured population, because we observed that 16.6% of cases and controls from the incidence analysis never visited a PCP. In addition, the PCP shortage currently experienced in the United States may worsen with increased demand stemming from the Affordable Care Act.[27]

Our data also support the important role of mammography in reducing late-stage diagnosis and mortality. We observed that increased PCP visits increased the incidence of DCIS and early stage BC and decreased the incidence of late-stage BC, and mammography explained some of this effect. When DCIS is diagnosed and treated, it may thwart the development of invasive cancer, although some have concerns about overdiagnosis.[28, 29] Despite mammography's importance, our findings revealed a general lack of screening over the entire period. Some physician services that may impact mortality and incidence, such as promoting healthy lifestyles, discussing risk-reducing medications, and addressing HRT, are considerably more complex than ordering a mammogram. For example, our results supported our hypothesis that PCP visits have a greater positive association with ER-positive BC. This finding is consistent with the hypothesis that factors affecting ER-positive tumors (eg, being prescribed HRT) are influenced by PCP visits. We also observed that more than 10 visits to a PCP were associated with increased mortality in the early stage analysis, which may be because patients who were sicker visited a PCP more often.

Limitations

Our results reflect the Medicare population living in SEER registries, and these trends may not apply to other populations or to those with other insurance plans. In addition, the SEER-Medicare records lack certain information on the nature of health care visits (eg, whether they were preventive visits, what recommendations or referrals were made, patient preferences and health beliefs), nor do they provide individual-level information on income or education. Although HRT use was postulated as a potential mediator for the positive effect of PCPs on BC incidence, we lacked information regarding HRT prescriptions and prevention strategies, so these theories could not be directly evaluated. Also, we were unable to differentiate between screening and diagnostic mammograms. However, we excluded the 3 months before diagnosis to avoid capturing diagnostic mammograms. Although our large sample size enabled us to have power to detect very small differences between groups, statistical significance may not necessarily translate to clinical significance. Nevertheless, small differences still may be important at the population level. Also, we did not estimate the underlying odds of mortality or incidence, but only the ORs, which may impede interpretation. Despite these limitations, our study is one of the first attempts to directly measure the impact of PCP visits on BC outcomes for Medicare beneficiaries. Our findings suggest that PCPs play an important role in reducing BC mortality among this population. Promoting and increasing access to primary care should be an important priority in health policy discussions if we are to decrease the burden of BC in the United States.

FUNDING SUPPORT

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SUPPORT
  8. CONFLICT OF INTEREST DISCLOSURES
  9. REFERENCES

This study was supported in part by the American Cancer Society (RSGHP-08-141-01-CPHPS). The sponsors had no role in the design or conduct of the study; in the collection, analysis, or interpretation of data; or in the preparation, review, or approval of the article.

REFERENCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. FUNDING SUPPORT
  8. CONFLICT OF INTEREST DISCLOSURES
  9. REFERENCES
  • 1
    Roetzheim RG, Ferrante JM, Lee JH, et al. Influence of Primary care on breast cancer outcomes among Medicare beneficiaries. Ann Fam Med. 2012;10:401-411.
  • 2
    Van Harrison R, Janz NK, Wolfe RA, et al. Characteristics of primary care physicians and their practices associated with mammography rates for older women. Cancer. 2003;98:1811-1821.
  • 3
    Harrison RV, Janz NK, Wolfe RA, Tedeschi PJ, Huang X, McMahon LF Jr. 5-Year mammography rates and associated factors for older women. Cancer. 2003;97:1147-1155.
  • 4
    Huth JF, O'Connor AP. Breast cancer screening practices for hospitalized women. J Cancer Educ. 1995;10:22-24.
  • 5
    Fox S, Murata P, Stein J. The impact of physician compliance on screening mammography for older women. Arch Intern Med. 1991;151:50-56.
  • 6
    Fox S, Siu A, Stein J. The importance of physician communication on breast cancer screening of older women. Arch Intern Med. 1994;154:2058-2068.
  • 7
    Ferrante JM, Rovi S, Das K, Kim S. Family physicians expedite diagnosis of breast disease in urban minority women. J Am Board Fam Med. 2007;20:52-59.
  • 8
    Bodenheimer T, Pham HH. Primary care: current problems and proposed solutions. Health Aff (Millwood). 2010;29:799-805.
  • 9
    Whitcomb ME, Cohen JJ. The future of primary care medicine. N Engl J Med. 2004;351:710-712.
  • 10
    Iglehart JK. Primary care update—light at the end of the tunnel? N Engl J Med. 2012;366:2144-2146.
  • 11
    Gorey KM, Luginaah IN, Fung KY, et al. Physician supply and breast cancer survival. J Am Board Fam Med. 2010;23:104-108.
  • 12
    Ferrante JM, Gonzalez EC, Pal N, Roetzheim RG. Effects of physician supply on early detection of breast cancer. J Am Board Fam Pract. 2000;13:408-414.
  • 13
    Roetzheim RG, Gonzalez EC, Ramirez A, Campbell R, van Durme DJ. Primary care physician supply and colorectal cancer. J Fam Pract. 2001;50:1027-1031.
  • 14
    Roetzheim RG, Pal N, Gonzalez EC, et al. The effects of physician supply on the early detection of colorectal cancer. J Fam Pract. 1999;48:850-858.
  • 15
    Ferrante JM, McCarthy EP, Gonzalez EC, et al. Primary care utilization and colorectal cancer outcomes among Medicare beneficiaries. Arch Intern Med. 2011;171:1747-1757.
  • 16
    Ernster VL, Ballard-Barbash R, Barlow WE, et al. Detection of ductal carcinoma in situ in women undergoing screening mammography. J Natl Cancer Inst. 2002;94:1546-1554.
  • 17
    Rossouw JE, Anderson GL, Prentice RL, et al. Risks and benefits of estrogen plus progestin in healthy postmenopausal women: principal results from the Women's Health Initiative randomized controlled trial. JAMA. 2002;288:321-333.
  • 18
    Chen WY, Hankinson SE, Schnitt SJ, Rosner BA, Holmes MD, Colditz GA. Association of hormone replacement therapy to estrogen and progesterone receptor status in invasive breast carcinoma. Cancer. 2004;101:1490-1500.
  • 19
    Gornick ME, Eggers PW, Riley GF. Associations of race, education, and patterns of preventive service use with stage of cancer at time of diagnosis. Health Serv Res. 2004;39:1403-1428.
  • 20
    Ferrante JM, Chen PH, Kim S. The effect of patient navigation on time to diagnosis, anxiety, and satisfaction in urban minority women with abnormal mammograms: a randomized controlled trial. J Urban Health. 2008;85:114-124.
  • 21
    Bergstralh EJ, Kosanke JL, Jacobsen SJ. Software for optimal matching in observational studies [letter]. Epidemiology. 1996;7:331.
  • 22
    Yu X, McBean AM, Virnig BA. Physician visits, patient comorbidities, and mammography use among elderly colorectal cancer survivors. J Cancer Surviv. Dec 2007;1:275-282.
  • 23
    Baldwin LM, Adamache W, Klabunde CN, Kenward K, Dahlman C, J LW. Linking physician characteristics and medicare claims data: issues in data availability, quality, and measurement. Med Care. 2002;40:82-95.
  • 24
    Charlson M, Pompei P, Ales K, Mackenzie C. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chron Dis. 1987;40:373-383.
  • 25
    Klabunde CN, Potosky AL, Legler JM, Warren JL. Development of a comorbidity index using physician claims data. J Clin Epidemiol. 2000;53:1258-1267.
  • 26
    Brookhart MA, Patrick AR, Dormuth C, et al. Adherence to lipid-lowering therapy and the use of preventive health services: an investigation of the healthy user effect. Am J Epidemiol. 2007;166:348-354.
  • 27
    Okie S. The evolving primary care physician. N Engl J Med. 2012;366:1849-1853.
  • 28
    Cady B, Chung MA. Preventing invasive breast cancer. Cancer. 2011;117:3064-3068.
  • 29
    Kalager M, Adami HO, Bretthauer M, Tamimi RM. Overdiagnosis of invasive breast cancer due to mammography screening: results from the Norwegian screening program. Ann Intern Med. 2012;156:491-499.