Abstract

Objective

Several studies raise questions about whether clinical practice guidelines actually guide practice. We evaluated patterns of use of breast-conserving surgery (BCS) over time to examine the effect of guideline publication.

Design

Retrospective analysis of time-series data on breast cancer treatment. Multiple logistic regression analysis was performed, adjusting for covariates including the patient's age, comorbidity status and admission year, to assess whether the use of BCS was higher after publication of treatment guidelines.

Setting

Five teaching hospitals participating in the Quality Improvement/Indicator Project (QIP) in Japan.

Participants

Female breast cancer patients who received surgical treatment at five teaching hospitals from January 1996 through December 2007 (n = 2199).

Main Outcome Measure

Rates of use of BCS.

Results

The proportion of BCS use increased from 26.4% before guideline publication to 59.9% after guideline publication in Japan. After controlling for other characteristics, the use of BCS has increased significantly over time, especially since 2001. Women aged 70 years and older (P=0.004) and those with any comorbidity (P < 0.001) were significantly less likely to receive BCS.

Conclusions

This study demonstrated that the adjusted proportion of BCS has increased dramatically since 2001, 2 years after guideline publication in Japan and this is consistent with a relationship between guideline publication and a change in this clinical practice.

Introduction

Clinical practice guidelines are defined as systematically developed statements to assist practitioners' and patients' decisions about appropriate health care for specific clinical conditions [1]. Successful implementation of clinical practice guidelines should improve quality of care by decreasing inappropriate practice variation and promoting reasonable decisions for effective practice [2]. A consistent finding in health services research, however, is the gap between scientific evidence and actual clinical practice [3]. Past studies suggest that 30–40% of patients do not receive current scientific evidence-based care [4]. These findings raise questions about whether clinical practice guidelines actually guide practice.

Randomized clinical trials performed in the 1980s demonstrated that women with early-stage breast cancer treated with either modified radical mastectomy or breast-conserving surgery (BCS) followed by radiation therapy had equivalent rates of survival and recurrence [5, 6]. In 1990, a National Institute of Health (NIH) Consensus Development Conference recommended BCS followed by radiation therapy for the majority of women with Stage I or II breast cancer [7]. After publication of the guideline, the use of BCS was statistically significantly higher, but not dramatically so [8–11].

These prior studies [8–11] cannot definitively separate effects of the 1990 NIH Consensus Development Conference from earlier events that might also increase the use of BCS; results of earlier clinical trials [5, 6] and popular publications may also have stimulated the increased use of BCS. Therefore, simple comparison of BCS use in the Western countries before and after the Consensus may not be appropriate. To separate the effects of Consensus recommendations on the use of BCS from the effect of scientific publications, it is necessary to account for scientific publications.

In July 1999, a steering committee convened by The Japanese Breast Cancer Society published evidence-based clinical practice guidelines for treatment of early-stage breast cancer in Japanese women [12]. To our knowledge, these Japanese guidelines were based on clinical evidence and existing guidelines from Western countries and not on the results of randomized clinical trials or re-analyses of previous studies that have evaluated the efficacy of BCS in Japan. In addition, because of language barriers, several large clinical trials published in Western countries seemed to have less impact on knowledge of the effectiveness of BCS in Japan compared with the impact in English-speaking countries [5, 6]. Before the publication of the Japanese guideline, therefore, it was possible that Japanese women might be unaware of this treatment choice. Because of this unique situation in Japan, any change in the use of BCS before and after guideline publication in Japan might reflect the effect of consensus recommendations rather than other factors. The aim of this study was to evaluate whether publication of clinical guidelines was associated with a change of treatment practices for breast cancer patients through the use of secondary administrative data from Japanese hospitals.

Methods

Database

We used a database from the Quality Indicator/Improvement Project (QIP), which includes more than 10 privately owned teaching hospitals in Japan. These hospitals are located in urban cities in Hokkaido (in north Japan), throughout Honshu (the main island of Japan), and in Kyushu (in south Japan). Records of all patients discharged from these hospitals have been kept since 1995. In this particular study, we selected 5 out of 10 tertiary care community hospitals that had more than 50 breast cancer patients between 1996 and 1999 [13], and have submitted medical claims data since 1996–2007. In 2000, the average number of general beds in the five hospitals was 671 (range 450–925). All had full-time surgeons and four out of five had radiotherapy machines. The average number of surgeons who worked in the hospitals was 10.6 (range 7–13) in 2000. All five hospitals were private teaching hospitals for board certification of general surgeons accredited by the Japan Surgical Society. This study was approved by the Institutional Review Board of Kyoto University Graduate School of Medicine Faculty of Medicine in Japan.

Study population

Because the QIP database involved all inpatient admissions, patients with either primary or recurrent breast cancer were included. A total of 2199 women who received either BCS or mastectomy between January 1996 and December 2007 were selected. Breast cancer surgeries are almost never performed in an outpatient setting in Japan. The database contains all clinical procedures with medical claims as well as demographic data on inpatients. Patients who underwent both surgeries and patients with distant metastasis were not included. The International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) and International Classification of Diseases, Tenth Revision (ICD-10) were used to identify women with breast cancer (ICD-9-CM codes 174.0–174.9 and ICD-10 codes C50) treated by either BCS (ICD-9-CM codes 85.20–85.23) or mastectomy (ICD-9-CM codes 85.41–85.48).

Statistical analysis

Data were analysed using either the Student's t-test for continuous variables or the χ2 test for categorical variables. We examined trends in the use of BCS between admission years by using a non-parametric test for trend. A multiple logistic regression analysis with robust standard errors (robust cluster) was performed to identify trends in the use of BCS over time, after adjusting for potential covariates. The dependent variable was the type of surgical procedure. Independent variables were the patient's age, comorbidity status and time period of admission. The hospital in which each patient was admitted was defined as a cluster. We used an adaptation of the Charlson comorbidity index to assess comorbidities [14, 15]. A patient was identified as having comorbidity if any of comorbidities were coded in the diagnosis (present vs. absent). All analytical procedures were performed using the STATA 9.2 statistical package (StataCorp. College Station, TX, USA). All reported P-values were two-tailed, and level of significance was P < 0.05.

Results

Table 1 shows the distribution of characteristics in patients treated by either BCS or mastectomy before or after guideline publication. In both time periods, women who underwent BCS were significantly younger than those who underwent mastectomy (P = 0.001 and <0.001, respectively). For women with BCS, comorbidity status was not significantly different between types of surgery before guideline publication (P = 0.49), whereas the difference in comorbidity status was statistically significant after guideline publication (P < 0.001).

Table 1

Characteristics of breast cancer patients who underwent either BCS or mastectomy at five teaching hospitals (n = 2199)

 Guideline publication (1999)
 
 Before publication (1996–99)
 
After publication (2000–07)
 
 BCS Mastectomy P-value BCS Mastectomy P-value 
Age [mean (SD)] 53.4 (12.4) 57.3 (13.2) 0.001a 55.9 (13.3) 59.1 (13.8) <0.001a 
Age group (%)   0.006b   <0.001b 
 <50 46.9 34.3  38.2 28.8  
 50–59 25.3 23.0  26.0 25.7  
 60–69 13.0 21.7  19.2 21.3  
 70+ 14.8 21.0  16.6 24.3  
Comorbidity (%)   0.49b   <0.001b 
 Absent 95.7 94.3  91.6 83.3  
 Present 4.3 5.8  8.4 16.7  
Total [n (%)] 162 (100%) 452 (100%)  950 (100%) 635 (100%)  
 Guideline publication (1999)
 
 Before publication (1996–99)
 
After publication (2000–07)
 
 BCS Mastectomy P-value BCS Mastectomy P-value 
Age [mean (SD)] 53.4 (12.4) 57.3 (13.2) 0.001a 55.9 (13.3) 59.1 (13.8) <0.001a 
Age group (%)   0.006b   <0.001b 
 <50 46.9 34.3  38.2 28.8  
 50–59 25.3 23.0  26.0 25.7  
 60–69 13.0 21.7  19.2 21.3  
 70+ 14.8 21.0  16.6 24.3  
Comorbidity (%)   0.49b   <0.001b 
 Absent 95.7 94.3  91.6 83.3  
 Present 4.3 5.8  8.4 16.7  
Total [n (%)] 162 (100%) 452 (100%)  950 (100%) 635 (100%)  

aStudent's t-test. bχ2 test.

Table 2 represents trends in BCS use among breast cancer patients with surgical treatment diagnosed before and after guideline publication. The percentage of patients receiving BCS almost doubled between the two time periods (P < 0.001). In all subgroups of women classified by age, the percentage receiving BCS increased substantially in the later time period. In addition, women with and without comorbidity had more BCS in the later time period (P < 0.05 and <0.001, respectively).

Table 2

Percentage of the use of BCS before and after guideline publication by age group, comorbidity and hospital among breast cancer patients (n = 1112)

 Guideline publication (1999)
 
P-valuea 
 Before publication (1996–99)
 
After publication (2000–07)
 
 n % BCS n % BCS 
Total 162 26.4 950 59.9 <0.001 
Age group      
 <50 76 32.9 363 66.5 <0.001 
 50–59 41 28.3 247 60.2 <0.001 
 60–69 21 17.7 182 57.4 <0.001 
 70+ 24 20.2 158 50.6 <0.001 
Comorbidity      
 Absent 155 26.7 870 62.2 <0.001 
 Present 21.2 80 43.0 <0.05 
Hospital      
 Hospital A 48 50.5 44 41.1 0.18 
 Hospital B 20 28.2 109 60.6 <0.001 
 Hospital C 50 41.3 115 73.7 <0.001 
 Hospital D 22 9.1 618 64.9 <0.001 
 Hospital E 22 25.9 64 33.7 0.20 
 Guideline publication (1999)
 
P-valuea 
 Before publication (1996–99)
 
After publication (2000–07)
 
 n % BCS n % BCS 
Total 162 26.4 950 59.9 <0.001 
Age group      
 <50 76 32.9 363 66.5 <0.001 
 50–59 41 28.3 247 60.2 <0.001 
 60–69 21 17.7 182 57.4 <0.001 
 70+ 24 20.2 158 50.6 <0.001 
Comorbidity      
 Absent 155 26.7 870 62.2 <0.001 
 Present 21.2 80 43.0 <0.05 
Hospital      
 Hospital A 48 50.5 44 41.1 0.18 
 Hospital B 20 28.2 109 60.6 <0.001 
 Hospital C 50 41.3 115 73.7 <0.001 
 Hospital D 22 9.1 618 64.9 <0.001 
 Hospital E 22 25.9 64 33.7 0.20 

aχ2 test.

Hospitals demonstrated wide variation in BCS use. Before guideline publication, Hospital A had the highest BCS use (50.5%), whereas Hospital D had the lowest (9.1%). The pattern of increased BCS after guideline publication differed between hospitals. Change in BCS at Hospitals A and E was not statistically significant, whereas the other three hospitals showed significant increases in BCS (P < 0.001). After guideline publication, an increase in BCS at Hospital D, which had exhibited the lowest use of BCS before guideline publication, was the second highest among all hospitals analysed.

Table 3 shows the results of a multiple logistic regression analysis to identify factors associated with utilization of BCS among all breast cancer patients who received surgical treatment in five hospitals analysed. For sensitive detection of changes after the 1999 guideline publication, we divided the 12-year interval annually, rather than into pre- and post-publication periods. A multiple logistic regression analysis demonstrated that the use of BCS has increased dramatically since 2001.

Table 3

Factors associated with the use of BCS among breast cancer patients (n = 2199)

Parameter Unadjusted
 
Adjusteda
 
Odds ratio 95% CI P-value Odds ratio 95% CI P-value 
Age group       
 <50 1.00   1.00   
 50–59 0.83 0.67–1.03 0.096 0.78 0.54–1.13 0.19 
 60–69 0.67 0.53–0.85 0.001 0.65 0.36–1.16 0.14 
 70+ 0.56 0.44–0.71 <0.001 0.56 0.38–0.83 0.004 
Comorbidity       
 Absent 1.00   1.00   
 Present 0.61 0.46–0.82 0.001 0.51 0.43–0.60 <0.001 
Admission yearb       
 1996 1.00   1.00   
 1997 1.42 0.79–2.57 0.25 1.46 0.99–2.16 0.058 
 1998 2.68 1.54–4.68 <0.001 2.92 1.67–5.08 <0.001 
 1999c 2.53 1.43–4.49 0.001 2.60 1.44–4.72 0.002 
 2000 3.31 1.82–5.99 <0.001 3.67 2.31–5.81 <0.001 
 2001 6.27 3.22–12.20 <0.001 6.68 4.57–9.77 <0.001 
 2002 7.98 4.53–14.04 <0.001 8.55 3.88–18.84 <0.001 
 2003 7.11 3.57–14.16 <0.001 7.63 2.23–26.17 0.001 
 2004 7.15 4.05–12.65 <0.001 8.06 2.77–23.45 <0.001 
 2005 9.66 5.77–16.15 <0.001 10.87 4.66–25.33 <0.001 
 2006 8.89 5.41–14.63 <0.001 9.89 3.22–30.42 <0.001 
 2007 8.12 4.91–13.44 <0.001 9.61 3.84–24.05 <0.001 
Parameter Unadjusted
 
Adjusteda
 
Odds ratio 95% CI P-value Odds ratio 95% CI P-value 
Age group       
 <50 1.00   1.00   
 50–59 0.83 0.67–1.03 0.096 0.78 0.54–1.13 0.19 
 60–69 0.67 0.53–0.85 0.001 0.65 0.36–1.16 0.14 
 70+ 0.56 0.44–0.71 <0.001 0.56 0.38–0.83 0.004 
Comorbidity       
 Absent 1.00   1.00   
 Present 0.61 0.46–0.82 0.001 0.51 0.43–0.60 <0.001 
Admission yearb       
 1996 1.00   1.00   
 1997 1.42 0.79–2.57 0.25 1.46 0.99–2.16 0.058 
 1998 2.68 1.54–4.68 <0.001 2.92 1.67–5.08 <0.001 
 1999c 2.53 1.43–4.49 0.001 2.60 1.44–4.72 0.002 
 2000 3.31 1.82–5.99 <0.001 3.67 2.31–5.81 <0.001 
 2001 6.27 3.22–12.20 <0.001 6.68 4.57–9.77 <0.001 
 2002 7.98 4.53–14.04 <0.001 8.55 3.88–18.84 <0.001 
 2003 7.11 3.57–14.16 <0.001 7.63 2.23–26.17 0.001 
 2004 7.15 4.05–12.65 <0.001 8.06 2.77–23.45 <0.001 
 2005 9.66 5.77–16.15 <0.001 10.87 4.66–25.33 <0.001 
 2006 8.89 5.41–14.63 <0.001 9.89 3.22–30.42 <0.001 
 2007 8.12 4.91–13.44 <0.001 9.61 3.84–24.05 <0.001 

CI, confidence interval. aA multiple logistic regression analysis with robust standard errors (robust cluster). Hospitals in which each patient admitted were defined as clusters. bA non-parametric test for trend across admission years: P < 0.001. cYear of guideline publication.

The use of BCS demonstrated linear correlation with patient age; younger women received more BCS than did older women. The odds ratio (OR) for patients with comorbidities was significantly less than for those lacking comorbidities (OR = 0.51 vs. no comorbidities; 95% confidence interval =0.43–0.60).

Discussion

We observed a significant increase in BCS during 2001–07 (after guideline publication) compared with 1996–99 (before guideline publication) consistent with a positive relationship between guideline publication and change in clinical practice. A change in practice patterns with a time lag (i.e. 2 years in our study) following guideline publication is consistent with the theory of diffusion of innovation [16], a process by which new technology or practices spread through social networks of physicians and patients over time. The trends in the use of BCS over time in this study were consistent with a study based on questionnaire data collected from 352 hospitals in Japan which showed an increase in BCS during 1995–2000, a plateau during 2000–01, followed by a second increase in the period after 2002, by using the crude proportion of BCS [17].

Although the impact of publication of guidelines or recommendations from expert consensus on practice changes is difficult to study in the absence of a control group, our study had two main advantages. First, because of the significant language barrier, it is unlikely that publication of results of several large clinical trials conducted in Western countries in the 1980s would have influenced practice in Japan, making the inference of causation from an epidemiological study more credible [5, 6]. Although the proportion of BCS use in Japan at the time of the Fisher's report [5] was 0.4% [17], BCS use in the USA had increased 16–17% [8, 9] during the same period. The NIH Consensus Development Conference recommended BCS for the majority of women with early-stage breast cancer [7]. Several studies have assessed effects of this recommendation on the use of BCS [8–11]. However, past studies could not definitively separate the effect of the Consensus recommendation from that of prior published clinical trials on BCS.

Second, considerable room for change existed in 1999. Several studies have addressed effects of the Consensus on treatment of breast cancer [8–11, 18], prostate cancer [19] and other conditions [20–22]. Only the breast cancer guidelines appeared to affect physicians' behaviour, perhaps because the practices recommended in guidelines for conditions other than breast cancer had already been adopted prior to publication of the guidelines. Our study demonstrated that only 26.4% of female breast cancer patients underwent BCS before guideline publication in Japan (Table 2); therefore, there was significant room for practice improvement following guideline publication.

Patient-, hospital- and surgeon-related factors have been associated with the use of BCS. Many previous researchers have considered the patient's age. Our observation that younger patients tended to undergo BCS rather than mastectomy is in accordance with previous reports [8, 23–25]. Several reasons may account for this association. First, younger patients seek more information than do older patients [26–29]. Second, older patients may be more sensitive to physicians' opinions and recommendations than are younger patients [28]. Third, older patients may wish to avoid the inconvenience of long courses of daily radiation therapy and, therefore, may prefer mastectomy [23, 27, 29].

As for other patient-related factors, we also examined whether comorbidity status was associated with type of surgery. In Table 1, the percentage of women who had any comorbidity after 2000 was higher than that before 1999 in both surgery groups, suggesting possible biases in the data. One possible source of bias could be an economic incentive. However, we feel that this may not be relevant. Although all participant hospitals in this study introduced the new per-diem payment system after 2004, which departed from the previously used fee-for-services system, our results show that the proportion of patients who had any comorbidity was lower before 1999 than after 2000. A more probable explanation for this inconsistency may be a difference in the format of our database. Before 2001, there was no column in the database specifically for patients' comorbidities; instead, there was the option of recording a maximum of five separate existing diseases for each patient at the time of admission. We identified the reported pre-existing diseases, excluding breast cancer, as each patient's comorbidities. However, since 2002, providers could input up to four comorbidities in our database.

Despite the variability in our measurement of comorbidity status, our multivariable logistic regression found that women who had any comorbidities at the time of diagnosis were less likely to receive BCS than those with no comorbidities, even after taking patient's age into account. Few previous studies have examined the relationship between comorbidity status or age and BCS use. Ballard-Barbash et al. [30] reported that older women and women who had any comorbidities at the time of diagnosis were more likely to undergo BCS without radiotherapy, and also noted that women were more likely to receive radiotherapy followed by BCS, after adjustment for other factors, if they were younger or had fewer comorbidities. Our results were concordant with previous findings [31] that younger women and women who had no comorbidity at the time of diagnosis were more likely to undergo BCS, without consideration of adjuvant radiotherapy status after BCS. There are many possible reasons why BCS use would be used less frequently in women with comorbidities. Women or their physicians may have wanted to avoid the challenge of radiotherapy after BCS due to severe symptoms of comorbidities, or to avoid the challenge of a second surgery if there were a recurrence of disease. Additional research using more detailed clinical data might clarify the relationship.

Also, we observed a variation in the use of BCS among participant hospitals. Percentage of BCS and pattern of increase in BCS differed significantly among five hospitals, perhaps due to surgeon- and hospital-related factors. According to medical literature, high-volume surgeons [24, 26, 29, 32] and female surgeons [24] favoured BCS. Causes of practice variation that were hospital-related included ownership of teaching hospitals [25, 32] and hospital location [24, 26, 32], as well as the type of teaching hospital (teaching hospital vs. non-teaching hospital). Regarding the association of hospital characteristics and BCS proportion in this study, we could not compare BCS use among different types of hospital ownership or reimbursement system because all five hospitals were private teaching hospitals, and had adopted the same reimbursement system at the same year. In terms of volume of surgery, however, our results suggest an association between high-volume hospitals and high proportion of BCS use (Table 2). Further research is necessary to identify reasons for practice variation among hospitals and to estimate the differences in guideline diffusion, such as rapidity and depth among different type of hospitals using a larger sample database.

Some limitations must be considered when interpreting the results of this study. First, since we estimated the impact of the surgery year on the use of BCS and separately assessed the year of guideline publication, this study cannot definitively demonstrate causality between the publication of guideline and practice changes. Second, we do not have information on tumour size, staging or other pathological features of breast cancer for each patient that might explain the choice of BCS, although it seems unlikely that these clinical characteristics would have changed as much the change in BCS rate suggests. According to The report of clinical statistical studies on registered mammary cancer patients in Japan edited by the Joint Committee for Mammary Cancer in Japan, the proportion of patients with early breast cancer (Stages 0–II) to which the BCS guidelines apply have not changed over the study period. Third, this study could not identify patients with recurrent breast cancer. Although the guideline covers local treatment of patients with Stages I and II breast cancer, our study population also involved patients with Stage III to whom the guidelines do not apply (patients with Stage IV have been excluded in our study by use of the information regarding distant metastasis).

Although these limitations make it difficult to know whether our estimates of BCS use by hospital reflect differences in clinical presentation or unwarranted variation in treatment, we believe that guideline publication could be associated with the change in clinical practice we observed. The proportion of patients with Stage III cancer, who are not eligible for BCS, has been <10% of all breast cancer patients during the study period (findings from the cancer registry reports). Therefore, even without clear information on stage, we believe that our results would not be essentially different if all data regarding patients' severity of cancer were available. Fourth, because the female patients received surgical treatment at five teaching hospitals in Japan, rather than being randomly selected from all hospitals in Japan, and the number of hospitals in our study was relatively small, this may reduce the external validity of our findings. However, our results regarding the use of BCS before and after guideline publication are consistent with a crude proportion of BCS use estimated by questionnaire survey at over 300 hospitals in Japan [17]. Therefore, the findings of our study may have face validity for hospitals throughout Japan. Last, since our database used administrative medical claims data, our analysis was not adjusted by surgeon factors and patients' preferences. Tarbox et al. [33] showed that not all surgeons believed that BCS and mastectomy had an equivalent survival rate. Especially in Japan, because there have been no clinical trials to examine whether early breast cancer patients in Japan are equally well treated with BCS as with mastectomy, we believe that the decision to use BCS for early breast cancer patients might depend on the preferences of surgeons and/or patients. Also, since both surgical procedures are equivalent in terms of survival, women may opt out of BCS by other factors such as a woman's individual preferences and personal reasons. For example, concerns about the burden of hospital visit for radiotherapy and the perceived risk of recurrence are supposable. On the contrary, of course, some patients with information regarding treatment strategy from media coverage would ask for BCS treatment and therefore the use of BCS might increase. The information, however, might be mostly derived from guideline publication in Japan.

Clarifying the impact of guideline dissemination alone cannot actually be achieved by the use of data from observational studies. Data from the clinical trials underlying these guidelines would have been published several years ahead of the guideline dissemination. Consequently, information from other sources such as medical literature, academic conference or seminars based on the results of clinical trials may spread to clinicians without passing through a specific guideline. Similarly, even after guideline dissemination, major events such as public education campaigns or patient advocacy campaigns could influence the rate of surgery. The results of several previous studies [8–11, 18–22], which aimed to explore the impact of guideline publication did not take into account biases involving information derived or available from other information sources (such as the results of clinical trials and dissemination of clinical guidelines). Our study is unique in potentially minimizing the direct and indirect impacts of published clinical trials on clinical practice (through a language barrier).

Conclusion

Several studies have assessed the effects of consensus recommendations on clinical practice, but have failed to show the impact of the consensus alone perhaps because information from scientific findings spread prior to the recommendations. Owing to the language barrier, however, we believe that by placing focus on the Japanese population, our study may reduce the impact of these publications. Our study suggested that the use of BCS has been increased dramatically in Japan since 2001, 2 years after guideline publication in Japan. This increase is consistent with a relationship between guideline publication and a change in clinical practice.

Funding

The work described in this article was funded in part by the Health Sciences Research Grants for the Research on Policy Planning and Evaluation from the Ministry of Health, Labor and Welfare of Japan and the Grant-in-aid for Scientific Research A from the Ministry of Education, Culture, Sports, Science and Technology of Japan.

Acknowledgements

The authors are grateful to the anonymous QIP participant hospitals.

References

Field
MJ
Lohr
MJ
Clinical Practice Guidelines: Directions for a New Program
 , 
1990
Washington, DC
National Academy Press
Audet
AM
Greenfield
S
Field
M
Medical practice guidelines: current activities and future directions
Ann Intern Med
 , 
1990
, vol. 
113
 (pg. 
709
-
14
)
Grol
R
Grimshaw
J
From best evidence to best practice: effective implementation of change in patients' care
Lancet
 , 
2003
, vol. 
362
 (pg. 
1225
-
30
)
Schuster
MA
McGlynn
EA
Brook
RH
How good is the quality of health care in the United States?
Milbank Q
 , 
1998
, vol. 
76
 (pg. 
517
-
63
)
Fisher
B
Bauer
M
Margolese
R
, et al.  . 
Five-year results of a randomized clinical trial comparing total mastectomy and segmental mastectomy with or without radiation in the treatment of breast cancer
N Engl J Med
 , 
1985
, vol. 
312
 (pg. 
665
-
73
)
Veronesi
U
Saccozzi
R
Del Vecchio
M
, et al.  . 
Comparing radical mastectomy with quadrantectomy, axillary dissection, and radiotherapy in patients with small cancers of the breast
N Engl J Med
 , 
1981
, vol. 
305
 (pg. 
6
-
11
)
NIH Consensus Conference
Treatment of early-stage breast cancer
JAMA
 , 
1991
, vol. 
265
 (pg. 
391
-
5
)
Lazovich
D
Solomon
CC
Thomas
DB
, et al.  . 
Breast conservation therapy in the United States following the 1990 National Institutes of Health Consensus Development Conference on the treatment of patients with early stage invasive breast carcinoma
Cancer
 , 
1999
, vol. 
86
 (pg. 
628
-
37
)
Du
X
Freeman
DH
Jr
Syblik
DA
What drove changes in the use of breast conserving surgery since the early 1980s?
Breast Cancer Res Treat
 , 
2000
, vol. 
62
 (pg. 
71
-
9
)
Gilligan
MA
Kneusel
RT
Hoffmann
RG
, et al.  . 
Persistent differences in sociodemographic determinants of breast conserving treatment despite overall increased adoption
Med Care
 , 
2002
, vol. 
40
 (pg. 
181
-
9
)
Gaudette
LA
Gao
RN
Spence
A
, et al.  . 
Declining use of mastectomy for invasive breast cancer in Canada, 1981–2000
Can J Public Health
 , 
2004
, vol. 
95
 (pg. 
336
-
40
)
Japanese Breast Cancer Society
Practice guideline: breast-conserving surgery
Jpn J Breast Cancer
 , 
2000
, vol. 
15
 (pg. 
147
-
56
[in Japanese]
Ishizaki
T
Imanaka
Y
Hirose
M
, et al.  . 
A first look at variations in use of breast conserving surgery at five teaching hospitals in Japan
Int J Qual Health Care
 , 
2002
, vol. 
14
 (pg. 
411
-
8
)
Charlson
ME
Pompei
P
Ales
KL
, et al.  . 
A new method of classifying prognostic comorbidity in longitudinal studies: development and validation
J Chronic Dis
 , 
1987
, vol. 
40
 (pg. 
373
-
83
)
Sundararajan
V
Henderson
T
Perry
C
, et al.  . 
New ICD-10 version of the Charlson comorbidity index predicted in-hospital mortality
J Clin Epidemiol
 , 
2004
, vol. 
57
 (pg. 
1288
-
94
)
Rogers
EM
Diffusion of Innovations
 , 
1995
4th edn
New York, NY
Free Press
Sonoo
H
Noguchi
S
Results of questionnaire survey on breast cancer surgery in Japan 2004–2006
Breast Cancer
 , 
2008
, vol. 
15
 (pg. 
3
-
4
)
Kosecoff
J
Kanouse
DE
Rogers
WH
, et al.  . 
Effects of the National Institutes of Health Consensus Development Program on physician practice
JAMA
 , 
1987
, vol. 
258
 (pg. 
2708
-
13
)
Sherman
CR
Potosky
AL
Weis
KA
, et al.  . 
The Consensus Development Program: detecting changes in medical practice following a consensus conference on the treatment of prostate cancer
Int J Technol Assess Health Care
 , 
1992
, vol. 
8
 (pg. 
683
-
93
)
Gleicher
N
Cesarean section rates in the United States: the short-term failure of the National Consensus Development Conference in 1980
JAMA
 , 
1984
, vol. 
252
 (pg. 
3273
-
6
)
Lomas
J
Anderson
GM
Domnick-Pierre
K
, et al.  . 
Do practice guidelines guide practice?
N Engl J Med
 , 
1989
, vol. 
321
 (pg. 
1306
-
11
)
Thamer
M
Ray
NF
Henderson
SC
, et al.  . 
Influence of the NIH Consensus Conference on Helicobacter pylori on physician prescribing among a Medicaid population
Med Care
 , 
1998
, vol. 
36
 (pg. 
646
-
60
)
Chagpar
AB
Studts
JL
Scoggins
CR
, et al.  . 
Factors associated with surgical options for breast carcinoma
Cancer
 , 
2006
, vol. 
106
 (pg. 
1462
-
6
)
Mandelblatt
JS
Berg
CD
Meropol
NJ
, et al.  . 
Measuring and predicting surgeons' practice styles for breast cancer treatment in older women
Med Care
 , 
2001
, vol. 
39
 (pg. 
228
-
42
)
Morrow
M
White
J
Moughan
J
, et al.  . 
Factors predicting the use of breast-conserving therapy in stage I and II breast carcinoma
J Clin Oncol
 , 
2001
, vol. 
19
 (pg. 
2254
-
62
)
Katz
SJ
Lantz
PM
Janz
NK
, et al.  . 
Surgeon perspectives about local therapy for breast carcinoma
Cancer
 , 
2005
, vol. 
104
 (pg. 
1854
-
61
)
Liang
W
Burnett
CB
Rowland
JH
, et al.  . 
Communication between physicians and older women with localized breast cancer: implications for treatment and patient satisfaction
J Clin Oncol
 , 
2002
, vol. 
20
 (pg. 
1008
-
16
)
Cyran
EM
Crane
LA
Palmer
L
Physician sex and other factors associated with type of breast cancer surgery in older women
Arch Surg
 , 
2001
, vol. 
136
 (pg. 
185
-
91
)
Mandelblatt
JS
Hadley
J
Kerner
JF
, et al.  . 
Patterns of breast carcinoma treatment in older women: patient preference and clinical and physical influences
Cancer
 , 
2000
, vol. 
89
 (pg. 
561
-
73
)
Ballard-Barbash
R
Potosky
AL
Harlan
LC
, et al.  . 
Factors associated with surgical and radiation therapy for early stage breast cancer in older women
J Natl Cancer Inst
 , 
1996
, vol. 
88
 (pg. 
716
-
26
)
Hall
SE
Holman
CD
Hendrie
DV
, et al.  . 
Unequal access to breast-conserving surgery in Western Australia 1982–2000
ANZ J Surg
 , 
2004
, vol. 
74
 (pg. 
413
-
9
)
Nattinger
AB
Gottlieb
MS
Veum
J
, et al.  . 
Geographic variation in the use of breast-conserving treatment for breast cancer
N Engl J Med
 , 
1992
, vol. 
326
 (pg. 
1102
-
7
)
Tarbox
BB
Rockwood
JK
Abernathy
CM
Are modified radical mastectomies done for T1 breast cancers because of surgeon's advice or patient's choice?
Am J Surg
 , 
1992
, vol. 
164
 (pg. 
417
-
20
)