The Immediate Impact of the 2009 USPSTF Screening Guideline Change on Physician Recommendation of a Screening Mammogram: Findings from a National Ambulatory and Medical Care Survey-Based Study

Abstract

Regular screening is considered the most effective method to reduce the mortality and morbidity associated with breast cancer. Nevertheless, contradictory evidence about screening mammograms has led to periodic changes and considerable variations among different screening guidelines. This study is the first to examine the immediate impact of the 2009 US Preventive Services Task Force (USPSTF) guideline modification on physician recommendation of mammograms. The study included visits by women aged 40 years and older without prior breast cancer from the National Ambulatory and Medical Care Survey 2008–2010. Bivariate and multiple logistic regressions were used to determine the factors associated with mammography recommendation. Approximately 29,395 visits were included and mammography was recommended during 1350 visits; 50–64-year-old women had 72% higher odds, and 65–74-year-old women had twice the odds of getting a mammogram recommendation compared with 40–49-year-old women in 2009. However, there was no difference in recommendation by age groups in 2008 and 2010. Obstetricians and gynecologists did not modify their recommendation behavior in 2009, unlike all other specialists who reduced their recommendation for 40–49-year-old women in 2009. Other characteristics associated with mammogram recommendations were certain patient comorbidities, physician specialty and primary care physician status, health maintenance organization status of the clinic, and certain visit characteristics. This study demonstrated a temporary effect of the USPSTF screening guideline change on mammogram recommendation. However, in light of conflicting recommendations by different guidelines, the physicians erred toward the more rigorous guidelines and did not permanently reduce their mammogram recommendation for women aged 40–49 years.

Introduction

Breast cancer has the highest incidence and a relatively higher morbidity and mortality of all cancers among women in the United States. However, regular screening, early diagnosis, and timely treatment initiation can considerably reduce the morbidity and mortality associated with the disease.^1
–3 To promote diagnosis of breast cancer at an early stage, regular screening is strongly advocated by all national clinical guidelines.^4

–8 Despite the guideline recommendations, a large proportion of eligible women do not undergo regular annual or biennial breast cancer screening.

Mammography with 77%–95% sensitivity and 94%–97% specificity is considered the gold standard for breast cancer screening in the United States.⁹ Clinical breast examination (CBE) and breast self-examination (BSE) have a lower sensitivity and specificity compared with mammography.^3,10 However, over the years, contradictory evidence about the benefits of mammography has emerged, casting doubt on the usefulness of regular screening.^11
–13 Evidence from developing countries with limited access to mammography, such as India, suggests that CBE can be an equally effective screening technique.¹³ Clinical trials evaluating the long-term reduction in breast cancer mortality report no significant reduction in mortality due to mammography compared with CBE¹¹ and BSE.¹²

Evidence supporting and contradicting the effectiveness of mammography has led to periodic changes and considerable variation in recommendations for mammography screening frequency by the US Preventive Services Task Force (USPSTF), American Cancer Society (ACS), National Cancer Institute, National Comprehensive Cancer Network (NCCN), American College of Radiology (ACR), and Society of Breast Imaging (SBI).^{4

–8,14

–21} Specifically, evidence generated from a simulation study using 6 Cancer Intervention and Surveillance Monitoring Network models²² and a systematic review²⁰ informed the 2009 USPSTF guideline changes. The new guideline recommended against regular screening in women 40–49 years of age and suggested a lower frequency (biennial) of screening in women 50–74 years of age to reduce the burden associated with false-positive screening results. This decision by USPSTF stands in contrast to the annual mammography recommendations by NCCN, ACS, SBI, and ACR for women aged 40 years and older. The USPSTF decision has been widely criticized by other organizations.^23
–25

Evidence shows that physicians’ recommendation for mammography is extremely critical for improving patients’ screening behavior²⁶ and screening guidelines might influence physicians’ behavior for recommending a mammogram.²⁷ Previous research on the impact of guidelines on physician behavior provides mixed evidence, with most evidence showing limited to no impact of guideline changes on physician practice.^27,28 However, a previous National Ambulatory Medical Care Survey (NAMC) study reported an increase in physician recommendation of mammograms from 1998 to 2004 compared with 1996,²⁹ which probably corresponds with the change in ACS guideline from biennial to annual mammography in 1997.¹⁴ Given the significant role of physician recommendation on patient compliance with regular screening, it is important to understand the impact of guideline recommendations on physicians’ decisions to recommend a mammogram to screening-eligible patients.

This study aimed to examine the immediate impact of the 2009 USPSTF guideline change on physician recommendation for screening mammograms for women 40 years and older during 2008 to 2010. This study also examines other patient and physician characteristics associated with the likelihood of mammogram recommendation by physicians.

Five studies examining the immediate impact of the 2009 guideline on patient utilization using private claims data and patient surveys presented mixed results.^30

–34 The studies found inconsequential reduction to no changes in patient utilization because of the 2009 guideline change. However, some of these studies focused on only one insurer and consequently were not nationally representative, and other studies used self-reported patient data, which are subject to reporting biases. In addition, none of the studies examined the impact of the guideline change on physician recommendations. To the authors’ knowledge, this is the first national study examining the immediate impact of the 2009 USPSTF guideline change on physician recommendation for screening mammograms.

Methods

Data

This study uses the NAMCS data. NAMCS is a national probability sample survey that collects information on office-based physician visits. It is a cross-sectional survey by the National Center for Health Statistics for the Centers for Disease Control and Prevention. The sampling for NAMCS is done using a multistage probability design that includes a sample of primary sampling units from which physician practices are sampled. Finally, patient visits are sampled from the selected physician practices. Data collected from the survey are weighted to generate national estimates that help understand the utilization of physician office visits in the United States. This study used NAMCS data from 2008, 2009, and 2010.

Study design and subjects

This cross-sectional study included women aged 40 years and older who visited physician offices between 2008 and 2010. This age group was chosen because screening guidelines are aimed at this population. The years 2008–2010 were chosen because the authors wanted to examine the immediate impact of the 2009 USPSTF screening guideline. Women with a current diagnosis of breast cancer were excluded from the study. The study sample was a subset of all visits to office-based physicians who were not employed by the federal government and were involved in direct patient care.

Dependent variable

For this study, the outcome variable was a dichotomous variable measuring whether or not a woman received a physician recommendation for a screening mammogram during her outpatient visit.

Independent variables

Because the study focused on examining the effect of the 2009 USPSTF policy change on physician recommendation of mammograms, year was included as one of the main independent variables. Interaction between year and age was also examined to understand the differential impact of the USPSTF policy change on different age categories (because the policy change affected different age groups differently).

John M. Eisenberg's model of clinical decision making was used as the conceptual framework to identify independent variables that could influence a physician's likelihood to recommend a mammogram. This conceptual framework classifies factors that influence physicians’ recommendation behavior into 4 categories, namely (1) patient characteristics, (2) provider/physician characteristics (including physician and clinic characteristics), (3) physician's interaction with his/her profession and the health care system, and (4) physician's relationship and type of interaction with the patient.³⁵

Numerous independent variables capturing these 4 categories were tested. However, the final logistic regression model included those variables with a P value ≤0.20 from the unadjusted logistic regression of the dependent variable on each of the independent variables. In addition, forward and backward stepwise regressions were also performed using all independent variables under consideration, and any independent variable that met any of the selection criteria (based on unadjusted regression P values and stepwise regression selection) was included.

Independent variables capturing patient characteristics included demographic and clinical characteristics. Patient demographic characteristics tested comprised age, race/ethnicity, insurance status, urbanicity of the patient's residential zip code, and patient's zip code-level education, household income, and poverty measures (education, income, and poverty measures were not available at the individual level). Among these, only age, race/ethnicity, insurance status, and urbanicity were included in the final model based on the selection processes used. Age was categorized into 4 categories—40–49, 50–64, 65–74, and 75 years and older—to match the age groups specified by the USPSTF guidelines as well as to capture the change in socioeconomic status (income, employment, and insurance) at age 65. Race and ethnicity were categorized into 4 mutually exclusive categories: non-Hispanic white, non-Hispanic black, Hispanic, and non-Hispanic other. Insurance status was categorized into privately insured, Medicare, Medicaid, uninsured, and unknown insurance. Urbanicity was a dichotomous variable measuring whether or not the patient lived in an urban area based on the patient's residential zip code.

Patient clinical characteristics tested comprised variables capturing the current diagnosis of chronic comorbidities—namely arthritis, asthma, cerebrovascular disease, chronic renal failure, congestive heart failure, chronic obstructive pulmonary disease, depression, diabetes, hyperlipidemia, hypertension, ischemic heart disease, obesity, and osteoporosis—and another variable capturing the absence of all these chronic ailments. In addition to these clinical variables, body mass index (BMI) classification based on current height and weight and current tobacco usage were also tested for inclusion into the final model. The variables included in the final model were dichotomous variables measuring current diagnosis of diabetes, hyperlipidemia, and obesity.

Physician characteristics and physician's interaction with his/her profession and the health care system were captured using physician specialty and characteristics of clinics the physicians belonged to, which included region of clinic's location, clinic's location in a metropolitan statistical area, and whether the clinic had an electronic reminder system for guideline-based screening and interventions, had a computerized system for public health reporting, was a community health center, was a family planning clinic, was part of a health maintenance organization (HMO), and provided some kind of cervical cancer screening.

Dichotomous variables measuring whether or not the physicians were compensated based on clinical productivity, patient satisfaction, quality of care (which included provision of preventive care), and patterns of using screening services were also tested. The variables included in the final model were physician specialty (categorized into 4 groups: general/internal/family medicine, obstetrician–gynecologist, oncologist, and other), and dichotomous variables measuring whether or not the clinic was part of an HMO, and provided some kind of cervical cancer screening.

Physician's relationship and type of interaction with the patient were captured by examining whether or not the physician was the patient's primary care physician (PCP), the patient was an existing patient (vs. a new patient), the physician had seen the patient more than twice in the past 12 months, the physician spent 15 minutes or more with the patient during the visit, the type of visit, and other preventive services performed during the visit (including breast examination, pelvic examination, Pap test, pregnancy test, and HPV-DNA test). The only variables included in the final model were the dichotomous variables measuring whether or not the physician was the patient's PCP, whether or not the physician saw the patient more than twice in the past 12 months, whether or not the visit involved a CBE, and whether or not the visit involved a Pap test, and a 3-category variable about type of visit (categorized as preventive visit, routine visit for chronic ailment, and visit related to a problem such as injury or flare-up of a current condition, among others).

Statistical analysis

Descriptive statistics were performed to examine the general distribution of the abovementioned independent variables for the entire sample, among visits that involved a mammogram recommendation, and among visits that did not involve a mammogram recommendation. Sampling weights were used to obtain the frequencies in the descriptive statistics to make the comparison nationally representative. Unadjusted bivariate logistic regressions of the dependent variable on each of the independent variables were performed, followed by a multiple logistic regression to determine the effect of the patient and physician characteristics on the likelihood of mammogram recommendation.

Special emphasis was placed on year of visit and the interaction between patient age and year, to capture the immediate effect of the USPSTF policy change. Hence, a logistic regression with interaction between patient age and year, as well as subgroup logistic regression analyses by year, were performed to study the effect of the USPSTF policy change on different age groups by year. This helped examine if the likelihood to recommend a mammogram changed for the different age groups over time after the policy was introduced in 2009. The method proposed by Norton et al was used to estimate the marginal interaction effect in the nonlinear logistic regression model, because odds ratio (OR) interpretations for interactions are mathematically incorrect in these models.³⁶

Logistic regressions to detect differences in mammogram recommendation by age by physician specialty (interaction between age and physician specialty) and mammogram recommendation by age by physician specialty within each year (interaction between age, year, and physician specialty) were also examined, because physician specialty was the strongest predictor of mammogram recommendation (in terms of statistical significance and magnitude of effect). Subgroup logistic regression analyses by type of visit were also performed because mammogram recommendations are more likely during preventive visits and the difference by age, and by age and year, might be more discernable in the preventive visits.

All analyses were conducted using Stata version 14.0 (StataCorp LP, College Station, TX). This study was approved by the Institutional Review Board at the University of Texas Health Science Center for the Protection of Human Subjects.

Results

The 2008–2010 NAMCS consisted of 92,251 unweighted visits, of which 29,395 unweighted patient visits belonged to women aged 40 years and older without a history of breast cancer, representing ∼1 billion outpatient visits in the nation. A total of 1350 unweighted visits representing more than 50 million visits nationally involved a mammogram recommendation. As shown in Table 1, all 3 years (2008–2010) had a similar number of visits.

Table 1.

Patient and Physician Characteristics Associated with Visits Involving and Not Involving Mammogram Recommendations: 2008–2009 NAMCS Data

	Percent of weighted total visits	Visits with mammography recommendation			Visits with no mammography recommendation
Variables	Weighted N = 998,859,583 unweighted N = 29,395	Unweighted visits N = 1350	Weighted visits N = 50,221,518 (1000)	Percent of Weighted visits	Unweighted visits N = 28,045	Weighted visits N = 948,638,065 (1000)	Percent of Weighted visits
Year
2008	32.54	443	18,421	36.68	9016	306,583	32.32
2009	35.69	476	17,943	35.73	10,202	338,515	35.68
2010	31.78	431	13,857	27.59	8827	303,540	32.00
Patient demographic characteristics
Age
40–49 years	22.22	422	14,470	28.81	6516	207,479	21.87
50–64 years	37.20	622	23,901	47.59	10,545	347,674	36.65
65–74 years	19.72	199	7084	14.11	5314	189,841	20.01
≥75 years	20.86	107	4766	9.49	5670	203,644	21.47
Race/ethnicity
Non-Hispanic white	76.77	968	37,964	75.59	21,067	728,872	76.83
Non-Hispanic black	10.31	149	4964	9.88	2964	98,064	10.34
Hispanic	8.81	146	4529	9.02	2723	83,486	8.80
Other	4.10	87	2764	5.50	1291	38,215	4.03
Insurance status
Privately insured	49.07	835	34,651	69.00	12,508	455,477	48.01
Medicare	37.27	279	10,935	21.77	10,277	361,295	38.09
Medicaid	4.72	98	1480	2.95	1861	45,671	4.81
Uninsured	5.14	71	1876	3.74	1894	49,476	5.22
Unknown	3.80	67	1279	2.55	1505	36,719	3.87
Urbanicity
No	17.64	202	6789	13.52	4993	169,422	17.86
Yes	82.36	1148	43,433	86.48	23,052	779,216	82.14
Patient clinical characteristics
Diabetes
No	84.15	1202	45,513	90.62	23,713	795,032	83.81
Yes	15.85	148	4709	9.38	4332	153,606	16.19
Hyperlipidemia
No	76.73	1027	37,381	74.43	22,159	729,046	76.85
Yes	23.27	323	12,840	25.57	5886	219,592	23.15
Obesity
No	90.32	1161	43,678	86.97	25,359	858,525	90.50
Yes	9.68	189	6544	13.03	2686	90,114	9.50
Physician characteristics and interaction with his/her environment/profession/health care system
Physician specialty
General Medicine	42.82	638	25,155	50.09	9517	402,556	42.44
Obstetrics and Gynecology	7.15	549	20,685	41.19	1370	50,740	5.35
Oncology	0.39	13	212	0.42	217	3,685	0.39
Other	49.64	150	4169	8.30	16,941	491,656	51.83
Clinic part of HMO
No	97.96	1334	49,678	98.92	27,376	928,756	97.90
Yes	2.04	16	543	1.08	669	19,882	2.10
Clinic provides cervical cancer screening
No	73.85	596	20,408	40.64	22,124	717,229	75.61
Yes	26.15	754	29,814	59.36	5921	231,409	24.39
Physician's relationship with patient and type of interaction during visit
Primary care physician
No	58.69	660	22,565	44.93	18,819	563,647	59.42
Yes	41.31	690	27,656	55.07	9226	384,991	40.58
Past visit
≤2 in 12 months	50.81	943	35,218	70.12	14,713	472,253	49.78
>2 in 12 months	49.19	407	15,004	29.88	13,332	476,385	50.22
Visit involved breast examination
No	93.07	577	22,527	44.86	26,973	907,140	95.63
Yes	6.93	773	27,694	55.14	1072	41,498	4.37
Visit involved Pap test
No	95.05	716	26,016	51.80	27,379	923,371	97.34
Yes	4.95	634	24,205	48.20	666	25,267	2.66
Type of visit
Visit for a problem	53.15	322	10,928	21.76	15,143	519,977	54.81
Preventive visit	13.97	807	31,113	61.95	2960	108,439	11.43
Routine visit	32.88	221	8180	16.29	9942	320,222	33.76

HMO, health maintenance organization; NAMCS, National Ambulatory Medical Care Survey.

Descriptive statistics (Table 1) showed that women receiving a mammogram recommendation were younger, such that more than 76% of women receiving a mammogram recommendation were younger than 65 years versus less than 59% of women not receiving a mammogram recommendation were younger than 65 years. A greater percentage of women receiving a mammogram recommendation were privately insured compared with women not receiving a mammogram recommendation. A higher proportion of women receiving a mammogram recommendation lived in urban areas compared with women not receiving a mammogram recommendation. There were no racial/ethnic differences between women receiving and not receiving a mammogram recommendation.

Clinically, women receiving a mammogram recommendation were less likely to be diabetic and more likely to have hyperlipidemia and obesity compared with women not receiving a mammogram recommendation. In terms of physician and visit characteristics, 91% of visits involving a mammogram recommendation were to physicians practicing general/internal/family medicine or obstetrics and gynecology versus only 48% of visits not involving a mammogram recommendation were to physicians belonging to these specialties. The visits involving a mammogram recommendation were less likely to be in clinics that are part of an HMO, but more likely to be in clinics that provide cervical cancer screening. Visits involving a mammogram recommendation were more likely to be PCP visits, preventive/routine visits, visits involving breast examinations and Pap smears, and visits that were among the first 2 visits in the last 12 months.

The unadjusted odds of receiving a mammogram recommendation during a visit showed most of the abovementioned associations to be statistically significant (Table 2). Visits involving younger (<65 years), privately insured, nondiabetic, obese women living in urban areas had a higher likelihood of mammogram recommendation. Visits involving physicians practicing general/internal/family medicine, obstetrics and gynecology, and oncology were more likely to involve a mammogram recommendation compared with any other specialty. Also, visits involving physician practices that are not part of an HMO and that provided cervical cancer screening were more likely to involve a mammogram recommendation. PCP visits, preventive/routine visits, visits involving breast examinations and Pap smears, and visits that were among the first 2 visits in the last 12 months were more likely to involve a mammogram recommendation.

Table 2.

Unadjusted and Adjusted Odds of Receiving a Mammogram Recommendation

	Unadjusted ORs	Adjusted ORs
Variable	OR (95% CI)	OR (95% CI)
Year
2008	Reference	Reference
2009	0.88 (0.64–1.23)	0.96 (0.65–1.42)
2010	0.76 (0.56–1.02)	1.05 (0.67–1.66)
Patient demographic characteristics
Age
40–49 years	Reference	Reference
50–64 years	0.99 (0.84–1.16)	1.20 (0.95–1.50)
65–74 years	0.54^* (0.41–0.69)	1.07 (0.72–1.57)
≥75 years	0.34^* (0.26–0.43)	0.71 (0.44–1.15)
Race/ethnicity
Non-Hispanic white	Reference	Reference
Non-Hispanic black	0.97 (0.72–1.32)	0.99 (0.69–1.40)
Hispanic	1.04 (0.79–1.37)	1.23 (0.90–1.69)
Other	1.39 (0.75–2.57)	1.03 (0.55–1.91)
Insurance status
Privately insured	Reference	Reference
Medicare	0.40^* (0.33–0.48)	1.01 (0.70–1.46)
Medicaid	0.43^* (0.28–0.65)	0.73 (0.46–1.14)
Uninsured	0.50^* (0.32–0.77)	1.09 (0.73–1.64)
Unknown	0.46^* (0.30–0.70)	0.73 (0.43–1.23)
Urbanicity
No	Reference	Reference
Yes	1.39^* (1.04–1.85)	1.38 (0.95–2.00)
Patient clinical characteristics
Diabetes
No	Reference	Reference
Yes	0.54^* (0.43–0.67)	0.72^* (0.55–0.94)
Hyperlipidemia
No	Reference	Reference
Yes	1.14 (0.95–1.37)	1.43^* (1.14–1.79)
Obesity
No	Reference	Reference
Yes	1.43^* (1.12–1.82)	1.52^* (1.09–2.11)
Physician characteristics and interaction with his/her environment/profession/health care system
Physician specialty
Other	Reference	Reference
General Medicine	7.37^* (4.86–11.18)	2.51^* (1.33–4.74)
Obstetrics and Gynecology	48.07^* (31.05–74.43)	4.31^* (2.56–7.25)
Oncology	6.80^* (1.79–25.81)	5.39^* (1.26–23.09)
Clinic part of HMO
No	Reference	Reference
Yes	0.51 (0.26–1.02)	0.36^* (0.19–0.67)
Clinic provides cervical cancer screening
No	Reference	Reference
Yes	4.53^* (3.61–5.68)	1.42 (0.94–2.15)
Physician's relationship with patient and type of interaction during visit
Primary care physician
No	Reference	Reference
Yes	1.79^* (1.42–2.26)	2.08^* (1.31–3.31)
Past visit
≤2 in 12 months	Reference	Reference
>2 in 12 months	0.42^* (0.34–0.52)	0.65^* (0.50–0.83)
Visit involved breast examination
No	Reference	Reference
Yes	26.87^* (21.18–34.09)	5.16^* (3.71–7.19)
Visit involved Pap test
No	Reference	Reference
Yes	34.00^* (26.48–43.66)	3.64^* (2.51–5.30)
Type of visit
Visit for a problem	Reference	Reference
Preventive visit	13.65^* (10.56–17.65)	2.99^* (2.18–4.09)
Routine visit	1.22 (0.93–1.59)	1.36^* (1.02–1.80)

P value <0.05.

CI, confidence interval; HMO, health maintenance organization; OR, odds ratio.

Adjusted ORs from the multiple logistic regression analysis showed similar associations except that all patient demographic characteristics (age, insurance status, and urbanicity) were rendered statistically insignificant, having hyperlipidemia increased the odds of receiving a mammogram recommendation, and the association between mammogram recommendation and a clinic providing cervical cancer screening services reduced in statistical significance with a P value of 0.09 (Table 2).

This study primarily aimed to understand if change in the USPSTF guideline changed physician recommendations for screening mammograms for women aged 40 years and older during 2008 to 2010. Age and year had no significant effect on the likelihood of mammogram recommendation during 2008 to 2010 in the pooled logistic regression analysis (Table 2). The logistic regression analysis with age and year interaction and the subgroup logistic regression analyses by year established that, in 2009, visits involving women aged 40–49 years had a significantly lower likelihood of a mammogram recommendation compared with visits involving women aged 50–64 and 65–74 years (Table 3). This was the only statistically significant difference detected, and age had no significant effect on the likelihood of mammogram recommendation in 2008 and 2010.

Table 3.

Adjusted Odds and Marginal Effects by Year for Receiving a Mammogram Recommendation

	2008		2009		2010
Age groups	OR (95% CI)	AME ^a (95% CI)	OR (95% CI)	AME ^a (95% CI)	OR (95% CI)	AME ^a (95% CI)
40–49	Reference		Reference		Reference
50–64	1.01 (0.65–1.56)	−0.04 (−1.23–1.14)	1.72^* (1.11–2.67)	1.52^* (0.30–2.74)	0.94 (0.66–1.33)	−0.02 (−1.09–1.05)
65–74	0.98 (0.54–1.76)	−0.21 (−1.68–1.26)	2.03^* (1.06–3.89)	1.36 (−0.47–3.19)	0.49^* (0.25–0.95)	−0.92 (−2.67–0.82)
≥75	0.62 (0.31–1.24)	−1.33 (−2.73–0.07)	1.26 (0.58–2.72)	−0.01 (−1.41–1.38)	0.40 (0.15–1.07)	−1.29 (−3.27–0.70)

The ORs are from the subgroup regression analyses by year, and the average marginal interaction effects are from the logistic regression, including an interaction between age and year.

P value <0.05.

Units in percentage points.

AME, average marginal effect; CI, confidence interval; OR, odds ratio.

To detect any other baseline time trends before the 2009 guideline change, the authors also performed similar logistic regression analyses by including data from the years 2006 and 2007 to this study's data, and found no difference by year or by age during years 2006, 2007, 2008, and 2010. The only statistically significant differences were for the year 2009, as shown in Table 3.

This study also examined the difference in recommendation by age as well as by age and year for different subgroups. First, differences in mammogram recommendation by age, and by age and year, among different physician specialties (ie, general/internal/family medicine, obstetrics and gynecology, oncology, and other) were examined. This analysis was performed because a previous study conducted just before the USPSTF guideline change, examining differences in mammogram recommendation by age, found that obstetricians and gynecologists recommended mammograms more aggressively to younger women aged 40–49 years, compared with other specialties. However, the previous study also found no difference in mammogram recommendation by specialty among women aged 50–59 years.

This study found no difference in recommendation by age by physician specialty. However, when difference in recommendation for each age group was examined by year for each physician specialty (triple interaction between age, year, and physician specialty), significant differences in recommendation by age were observed in year 2009 for general/internal/family medicine and other specialties, but no differences by age were detected among obstetricians and gynecologists for any of the years. Oncologists could not be tested for a triple interaction because this group was introduced only in 2010 in the NAMCS data.

Second, differences in mammogram recommendation by age among different types of office visits (ie, preventive visit, routine visit for chronic ailment, and visit related to a problem) were examined. This was done because mammogram recommendation is more common among preventive office visits; hence, differences by age (and by age and year) due to guideline changes might be more discernable during preventive visits compared with other visits. However, the subgroup analysis by visit showed the same findings as the main analyses presented in Tables 2 and 3 and did not find any difference by age, by type of visit, and found similar differences by age, by year, for each type of visit.

All multiple logistic regression models performed by subgroups or with interactions had similar findings about associations for all other covariates, as presented in Table 2.

Discussion

The results of this study indicate that there was a reduction in mammogram recommendation rates for 40–49-year-old women compared with 50–64-year-old women in 2009, the year USPSTF guidelines were released.

It is surprising that mammogram recommendations during physician office visits started to decrease in 2009, even before the USPSTF guideline was released. This decrease is possibly because the results of lower benefits of mammograms among younger women were presented and discussed in professional circles before the formal guideline release. However, the pattern did not continue in 2010 possibly because many physician groups and advocacy organizations such as the ACS, Susan G. Komen Foundation, American College of Surgeons, American Society of Breast Disease, NCCN, ACR, American College of Obstetricians and Gynecologists, and others expressed their concern over the USPSTF guideline change and also continued their recommendation for annual mammogram screening for women aged 40 years and older.³⁷

Moreover, there was considerable criticism from the media and other stakeholders against the USPSTF guideline.²⁵ In addition, the publicity against the 2009 USPSTF guideline change led to concerns in the general population about reducing screening for 40–49-year-old women and, consequently, led to younger women requesting a mammogram in the months after the media backlash about the guideline.³⁸ Similar to the findings of a previously conducted physician survey, the current study showed that obstetricians and gynecologists (as opposed to general/internal medicine specialists, family practitioners, and others) did not agree with the USPSTF guideline and continued to consistently recommend mammograms to women between the ages of 40–79 years.³⁹

To the authors’ knowledge this is the first study that examined the impact of the 2009 USPSTF guideline change on physician recommendation of screening mammograms. Previous studies have examined actual utilization of mammograms by patients and have predominantly found no effects of the guideline on mammogram utilization.^31

–34 One study found a temporary drop in mammogram utilization among 40–49-year-old women soon after the guideline release in 2009, followed by an immediate reversal of this drop and no subsequent differences by age group in 2010, similar to the current study.³⁰

As already mentioned, the patient demographic characteristics—namely age, race/ethnicity, insurance status, and urbanicity—were not associated with the likelihood of mammogram recommendation. Other socioeconomic factors such as income, education, and poverty levels of the geographic area were tested and excluded from the model due to their statistical insignificance in all models.

On the other hand, clinical factors—namely not being diabetic and being hyperlipidemic and obese—were correlated with a higher likelihood of mammogram recommendation. These findings were consistent with previous studies that reported that women with chronic conditions, especially diabetes, received significantly lower referrals for screening mammograms and other preventive care.^40,41 A possible explanation is the complex nature of diabetes requiring time-intensive clinical management, with numerous aspects such as eye care, feet examination, insulin administration, medication management, and behavioral and lifestyle modifications requiring the clinician's attention; the complexity of this management poses competing demands and makes provision of other unrelated care challenging given the time constraints associated with a physician office visit.^40,42 In addition, the health care system, which predominantly focuses on disease-centered care, also hinders delivery of preventive care services even in primary care settings.⁴³ This finding about women with diabetes is of great concern as evidence suggests that type 2 diabetes increases the risk of breast cancer and breast cancer mortality.⁴⁴

This study found that women with hyperlipidemia were more likely to receive a mammogram recommendation. This association has not been reported in the past literature to the best of the authors’ knowledge and is contrary to the negative association between presence of chronic ailments and cancer screening due to competing risks. It is not clear if this association is intentional, and hence, it should be further investigated.

This study also found that women who were diagnosed as obese by their physicians were more likely to get a mammogram recommendation. Previous studies have focused on the association of obesity and patients’ use of and compliance with mammograms and other cancer screening. The studies found that obese women are less compliant with screening mammograms due to poor body image often causing embarrassment, small-sized mammography equipment causing discomfort, and concerns about disrespectful or negative attitudes of providers.^45,46

The current study focuses on physicians’ recommendations and not patient compliance of mammography and finds an inverse association. It is important to note that no association between patient BMI and physician recommendation was detected (and hence BMI was excluded from the analyses), suggesting that the physician recommendation is associated with being diagnosed with obesity but not the actual BMI status. This might imply that only when patients are identified as being obese, breast cancer screening recommendations increase, possibly because of the concern about increased breast cancer risk among obese women. In addition, the screening recommendation does not guarantee patient compliance.

One of the characteristics that had the strongest impact on the likelihood of mammogram recommendation is the specialty of the physicians. Compared with all other specialties, obstetricians and gynecologists, general/internal/family medicine specialists, and oncologists had a considerably higher likelihood of recommending mammograms. Previous studies have also documented that PCPs, and obstetricians, and gynecologists recommend more mammograms than most other specialties.^29,47 In addition, as noted before, this study found that obstetricians and gynecologists had no temporal change in the likelihood of recommending mammograms to different age groups between 2008 and 2010, unlike all other specialists who reduced their mammogram recommendation for women aged 40–49 years only during 2009. This finding was also consistent with a previous study.³⁹

Training associated with physician specialties, beliefs of peers within specialties, type of care provided by each specialty, and reason for visit might determine this strong association with mammogram recommendation and temporal changes by specialties. Competing health care demands and associated time constraints faced by specialties other than general/internal/family medicine, obstetrics–gynecology, and oncology, might also reduce the likelihood of mammogram recommendation by those specialties.^42,43

It is important to note that the current study did not have information on physician sex. Female physicians are known to perform and recommend more breast-related preventive services compared with male physicians,^47,48 possibly because women report embarrassment in discussing breast health issues with male physicians and female physicians are more trusting of the effectiveness of mammograms. In addition, female physicians’ orientation toward preventive services, tendency to spend more time with patients, and communication technique may influence their referral rates for preventive services.^47,48 Lack of data on physicians’ sex may have confounded the association between physician specialty and mammogram recommendation because specialties such as obstetrics and gynecology have a higher proportion of female physicians.

Other physician characteristics associated with mammogram recommendation were belonging to an HMO-based clinic and a clinic with cervical cancer screening services. In contrast to previous literature,⁴⁹ this study found that physicians from HMO-based clinics were less likely to recommend screening mammograms. This is an unusual finding and should be evaluated further. Not surprisingly, physicians belonging to clinics providing cervical cancer screening services had a higher likelihood of recommending mammograms, emphasizing the effect of environmental reinforcement of screening practices.

The relationship of the physician with the patient and the type of visit were also associated with the likelihood of mammogram recommendation. If the physician was the patient's PCP, the likelihood of mammogram referral was higher. Regular PCPs better understand a patient's medical history, have a more complete knowledge about a patient's need for different preventive services, and have a higher comfort level with women to discuss breast health issues.⁴⁷ In addition, patients might be referred to non-PCPs when they have a specific problem, thereby reducing the likelihood of preventive service recommendations because of other competing health care demands and associated time constraints.^42,43 Competing health care demands and time constraints might also explain why mammogram recommendation is higher during routine or preventive care visits compared with visits for flare-ups or new problems in this study as well as previous studies.^42,43,47

Similar to other studies, having 2 or fewer past visits in the preceding 12 months is associated with a higher likelihood of mammogram recommendation.⁴⁷ Greater numbers of previous visits could indicate the presence of chronic comorbid conditions requiring regular follow-up, which become the primary focus of the visit.⁴² In addition, the patient could have already received a mammogram recommendation in one of the visits during the preceding 12 months, thereby reducing the likelihood of the recommendation in the current visit. Receipt of a CBE or Pap test also increases the likelihood of a mammogram recommendation. Physicians often provide breast and cervical screening services as a package during a visit,⁵⁰ and provision of other cancer screening tests is indicative of a physician's extent of compliance with cancer prevention guidelines.

There are certain limitations to this study. First, NAMCS, being a cross-sectional dataset, provides information about patient visits rather than the individual patient. Therefore, the study cannot provide information on whether the women in the study are due or overdue for screening or are in fact up-to-date as per the guidelines. The study may also be affected by errors in data collection, data reporting, and also missing data.

In conclusion, the authors observed a significant drop in mammogram recommendation rate for women aged 40–49 years compared with 50–64 years in 2009, with no difference by age groups in 2008 and 2010. Obstetricians and gynecologists did not change their practice in accordance with the 2009 USPSTF guideline change during 2009 and 2010. The study established that patient demographic and socioeconomic factors, including race/ethnicity, were not associated with mammogram recommendation but a physician's training, peer opinions, practice environment, relationship with the patient, and nature of the visit strongly affected the likelihood of recommendation of a mammogram. To influence physicians’ practice, educating them about the importance of screening guidelines and striving to influence the professional groups of different physician specialties might assist in better influencing physician opinions and recommendations.

The findings of this study are particularly relevant in light of the Patient Protection and Affordable Care Act (PPACA). Previously, mammogram utilization by a patient was considerably influenced by the patient's ability to pay and the patient's insurance status.⁵¹ However, with increased coverage for mammograms under the PPACA, financial barriers will be significantly reduced and actual patient utilization will be driven more by physician recommendation. Hence, it is important to identify interventions, policies, and clinical guideline changes that would have the most impact on physicians’ practice, and would assist in improving the current biennial mammography rate of 66% for women aged 40 years and older.⁵¹

Footnotes

Author Disclosure Statement

Drs. Rajan and Lairson, Ms. Suryavanshi, and Mr. Karanth declared no conflicts of interest with respect to the research, authorship, and/or publication of this article. The authors received no financial support for this article.

References

Berry

, Cronin

, Plevritis

, et al. Effect of screening and adjuvant therapy on mortality from breast cancer. N Engl J Med, 2005; 353:1784–1792.

Richards

, Westcombe

, Love

, Littlejohns

, Ramirez

. Influence of delay on survival in patients with breast cancer: a systematic review. Lancet, 1999; 353:1119–1126.

Humphrey

, Helfand

, Chan

, Woolf

. Breast cancer screening: a summary of the evidence for the US Preventive Services Task Force. Ann Intern Med, 2002; 137:347–360.

US Preventive Services Task Force. Screening for breast cancer: US Preventive Services Task Force recommendation statement. Ann Intern Med, 2009; 151:716–726, W-236.

American Cancer Society. American Cancer Society recommendations for early breast cancer detection in women without breast symptoms. Breast Cancer Prevention and Early Detection website. www.cancer.org/cancer/breastcancer/moreinformation/breastcancerearlydetection/breast-cancer-early-detection-acs-recs Accessed November 24, 2015 .

Bevers

, Anderson

, Bonaccio

, et al. NCCN clinical practice guidelines in oncology: breast cancer screening and diagnosis. J Natl Compr Canc Netw, 2009; 7:1060–1096.

National Cancer Institute. Breast cancer screening—health professional version (PDQ®). National Cancer Institute website. www.cancer.gov/types/breast/hp/breast-screening-pdq Accessed November 24, 2015 .

Society of Breast Imaging. Screening mammography resources. www.sbi-online.org/RESOURCES/ScreeningMammography.aspx Accessed November 24, 2015 .

Mushlin

, Kouides

, Shapiro

. Estimating the accuracy of screening mammography: a meta-analysis. Am J Prev Med, 1998; 14:143–153.

10.

Barton

, Harris

, Fletcher

. The rational clinical examination. Does this patient have breast cancer? The screening clinical breast examination: should it be done? How?. JAMA, 1999; 282:1270–1280.

11.

Miller

, Wall

, Baines

, Sun

, To

, Narod

. Twenty-five year follow-up for breast cancer incidence and mortality of the Canadian national breast screening study: randomised screening trial. BMJ, 2014; 348:g366.

12.

First results on mortality reduction in the UK trial of early detection of breast cancer. UK Trial of Early Detection of Breast Cancer Group. Lancet, 1988; 2:411–416.

13.

Okonkwo

, Draisma

, der Kinderen

, Brown

, de Koning

. Breast cancer screening policies in developing countries: a cost-effectiveness analysis for India. J Natl Cancer Inst, 2008; 100:1290–1300.

14.

Leitch

, Dodd

, Costanza

, et al. American Cancer Society guidelines for the early detection of breast cancer: update 1997. CA Cancer J Clin, 1997; 47:150–153.

15.

Eastman

. NCI adopts new mammography screening guidelines for women. J Natl Cancer Inst, 1997; 89:538–539.

16.

US Preventive Services Task Force. Chapter 7. Screening for breast cancer. In: Guide to Clinical Preventive Services: Report of the U.S. Preventive Services Task Force. 2nd ed. www.ncbi.nlm.nih.gov/books/NBK15435 Accessed November 25, 2015 .

17.

Mettlin

, Smart

. Breast cancer detection guidelines for women aged 40 to 49 years: rationale for the American Cancer Society reaffirmation of recommendations. American Cancer Society. CA Cancer J Clin, 1994; 44:248–255.

18.

Smith

, Cokkinides

, Eyre

, American Cancer Society. American Cancer Society guidelines for the early detection of cancer, 2003. CA Cancer J Clin, 2003; 53:27–43.

19.

Woolf

. United States Preventive Services Task Force recommendations on breast cancer screening. Cancer, 1992; 69(7 suppl):1913–1918.

20.

Nelson

, Tyne

, Naik

, et al. Screening for breast cancer: an update for the US Preventive Services Task Force. Ann Intern Med, 2009; 151:727–737, W237–W242.

21.

Lee

, Dershaw

, Kopans

, et al. Breast cancer screening with imaging: recommendations from the Society of Breast Imaging and the ACR on the use of mammography, breast MRI, breast ultrasound, and other technologies for the detection of clinically occult breast cancer. J Am Coll Radiol, 2010; 7:18–27.

22.

Mandelblatt

, Cronin

, Bailey

, et al. Effects of mammography screening under different screening schedules: model estimates of potential benefits and harms. Ann Intern Med, 2009; 151:738–747.

23.

DeAngelis

, Fontanarosa

. US Preventive Services Task Force and breast cancer screening. JAMA, 2010; 303:172–173.

24.

American College of Radiology. USPSTF mammography recommendations will result in countless unnecessary breast cancer deaths each year. www.acr.org/About-Us/Media-Center/Position-Statements/Position-Statements-Folder/USPSTF-Mammography-Recommendations-Will-Result-in-Countless-Unnecessary-Breast-Cancer-Deaths Updated 2009. Accessed November 25, 2015 .

25.

Stein

. Federal panel recommends reducing number of mammograms. Washington Post website. www.washingtonpost.com/wp-dyn/content/article/2009/11/16/AR2009111602822.html Updated 2009. Accessed November 25, 2015 .

26.

Schueler

, Chu

, Smith-Bindman

. Factors associated with mammography utilization: a systematic quantitative review of the literature. J Womens Health (Larchmt), 2008; 17:1477–1498.

27.

Slanetz

, Moore

, Hulka

, et al. Screening mammography: effect of national guidelines on current physician practice. Radiology, 1997; 203:335–338.

28.

Klabunde

, O'Malley

, Kaluzny

. Physicians’ reactions to change in recommendations for mammography screening. Am J Prev Med, 1997; 13:432–438.

29.

Bhosle

, Samuel

, Vosuri

, Paskett

, Balkrishnan

. Physician and patient characteristics associated with outpatient breast cancer screening recommendations in the United States: analysis of the National Ambulatory Medical Care Survey data 1996–2004. Breast Cancer Res Treat, 2007; 103:53–59.

30.

Wang

, Fan

, Van Houten

, et al. Impact of the 2009 US Preventive Services Task Force guidelines on screening mammography rates on women in their 40s. PLoS One, 2014; 9:e91399.

31.

Wharam

, Landon

, Zhang

, Xu

, Soumerai

, Ross-Degnan

. Mammography rates 3 years after the 2009 US Preventive Services Task Force guidelines changes. J Clin Oncol, 2015; 33:1067–1074.

32.

Block

, Jarlenski

, Wu

, Bennett

. Mammography use among women ages 40–49 after the 2009 U.S. Preventive Services Task Force recommendation. J Gen Intern Med, 2013; 28:1447–1453.

33.

Pace

, He

, Keating

. Trends in mammography screening rates after publication of the 2009 US Preventive Services Task Force Recommendations. Cancer, 2013; 119:2518–2523.

34.

Howard

, Adams

. Mammography rates after the 2009 US Preventive Services Task Force breast cancer screening recommendation. Prev Med, 2012; 55:485–487.

35.

Eisenberg

. Sociologic influences on decision-making by clinicians. Ann Intern Med, 1979; 90:957–964.

36.

Norton

, Wang

, Ai

. Computing interaction effects and standard errors in logit and probit models. Stata J, 2004; 4:154–167.

37.

Barclay

. New breast cancer screening guidelines opposed by societies. Medscape website. www.medscape.com/viewarticle/712720 Updated November 19, 2009. Accessed December 1, 2015 .

38.

Davidson

, Liao

, Magee

. Attitudes of women in their forties toward the 2009 USPSTF mammogram guidelines: a randomized trial on the effects of media exposure. Am J Obstet Gynecol, 2011; 205:30.e1–e30.e7.

39.

Yasmeen

, Romano

, Tancredi

, Saito

, Rainwater

, Kravitz

. Screening mammography beliefs and recommendations: a web-based survey of primary care physicians. BMC Health Serv Res, 2012; 12:32.

40.

Lipscombe

, Hux

, Booth

. Reduced screening mammography among women with diabetes. Arch Intern Med, 2005; 165:2090–2095.

41.

Kiefe

, Funkhouser

, Fouad

, May

. Chronic disease as a barrier to breast and cervical cancer screening. J Gen Intern Med, 1998; 13:357–365.

42.

Nutting

, Baier

, Werner

, Cutter

, Conry

, Stewart

. Competing demands in the office visit: what influences mammography recommendations?. J Am Board Fam Pract, 2001; 14:352–361.

43.

Kottke

, Brekke

, Solberg

. Making “time” for preventive services. Mayo Clin Proc, 1993; 68:785–791

44.

Coughlin

, Calle

, Teras

, Petrelli

, Thun

. Diabetes mellitus as a predictor of cancer mortality in a large cohort of US adults. Am J Epidemiol, 2004; 159:1160–1167.

45.

Fontaine

, Heo

, Allison

. Body weight and cancer screening among women. J Womens Health Gend Based Med, 2001; 10:463–470.

46.

Amy

, Aalborg

, Lyons

, Keranen

. Barriers to routine gynecological cancer screening for white and African-American obese women. Int J Obes (Lond), 2006; 30:147–155.

47.

Sabatino

, Thompson

, Coughlin

, Schappert

. Predisposing, enabling, and reinforcing factors associated with mammography referrals in U.S. primary care practices. Open Health Serv Policy J, 2009; 2:57–70.

48.

Lurie

, Margolis

, McGovern

, Mink

, Slater

. Why do patients of female physicians have higher rates of breast and cervical cancer screening?. J Gen Intern Med, 1997; 12:34–43.

49.

Baker

, Phillips

, Haas

, Liang

, Sonneborn

. The effect of area HMO market share on cancer screening. Health Serv Res, 2004; 39:1751–1772.

50.

Murata

, Li

. Relationship between pap smear performance and physician ordering a mammogram. J Fam Pract, 1992; 35:644–648.

51.

National Center for Health Statistics. Health, United States, 2014: With Special Feature on Adults Aged 55–64. 2015. www.cdc.gov/nchs/data/hus/hus14.pdf#076 Accessed March 21, 2016 .