Abstract Session C2: Women's Health Moderators: Wendy L. Bennett, - - PDF document

Abstract Session C2: Women's Health Moderators: Wendy L. Bennett, MD, MPH Discussant: Melissa McNeil, MD, MPH, SGIM 2014 Distinguished Professor in Women’s Health

PROGRESS OR STALEMATE IN ACADEMIC GENDER DISPARITIES? 15 YEAR FOLLOW UP OF THE NATIONAL FACULTY SURVEY Karen Freund1,2; Anita Raj3; Norma Terrin1,2; Samantha Kaplan4,5; Tracy Urech2; Subash Pathak2; Phyllis Carr6. 1Tufts University School of Medicine, Boston, MA; 2Institute for Clinical Research and Health Policy Studies, Tufts Medical Center, Boston, MA; 3University of California, San Diego, San Diego, CA; 4Boston University School of Medicine, Boston, MA; 5Boston Medical Center, Boston, MA; 6Massachusetts General Hospital, Boston, MA. (Tracking ID #1931176) BACKGROUND: Our prior research found that women compared to men received lower compensation, published fewer articles, and did not advance at the same rate despite similar grant funding productivity. Efforts over the past 20 years have sought to reduce such inequities, but long-term gender differences on these outcomes over the course of the academic career remain unknown. The objective of our current research is to assess gender differences in rank, productivity, compensation and career satisfaction among a national cohort of medical faculty followed from 1995 to 2012. METHODS: We conducted a follow-up of the 1995 National Faculty Survey cohort, which was drawn from a representative sample of 24 US medical schools, and oversampled generalists as well as underrepresented minority faculty and senior women. We supplemented on line survey responses with publically available data on publications, rank, and leadership on grants. We compared unadjusted median and mean responses on variables between men and women. We adjusted analyses by duration of time in academic career and academic department type (categorized as basic science, medical specialty, surgical specialty or generalist). RESULTS: We report on 288 men and 301 women. At the time of the follow up survey 62% of women and 77% of men reached the rank of full professor (p<.0002). Women had fewer median peer-reviewed publications than men (unadjusted median 43 versus 68, p=.0001), with a 15 publication deficit after adjustment for duration of the academic career and department type (Wald 95% confidence interval (CI) 6 - 25, p<.005). The proportion of faculty with funded grants in the prior two years was similar between men and women (50% versus 47%, p=.39). Women had an unadjusted $26,060 lower mean income compared with men ($169,190 versus $195,250, p<.003) which was reduced to $18,830 after adjustment. The difference in a 4 item career satisfaction measure (scale ranged from 4 - 20) was statistically significant, but was a relatively small effect (women, mean=14.7, standard deviation (SD)=3.9 versus men, mean=15.6, SD=3.5, p<.0001). Adjustment for years in academics and department type explained some but not all gender differences. CONCLUSIONS: Women report similar satisfaction with academic careers, similar grant funding productivity but lower peer-reviewed publication productivity. Women across the course of their careers continue to report lower compensation, and to be less likely to achieve the rank of professor. These findings parallel the status of this group of women compared with their male counterparts in 1995, indicating that rank and compensation gaps remain over the course of the academic career, despite decades to address these disparities.

Variable Analysis Adjusted for Duration in Academics and Department Type Women versus Men 95% CI and p-value Number of Peer-Reviewed Publications Adjusted mean number of publications

• 15

95% CI -25, -6; p=.002 Leadership on Grants Adjusted OR .98 95% CI .69, 1.39; p=.9 Achieving Rank of Professor Adjusted OR .56 95% CI .36, .88; p=.01 Career Satisfaction Adjusted parameter estimate

• .81

95% CI -1.40, -0.22; p=.008 Compensation Adjusted annual compensation

• $18,830

95% CI -$35,647, -$2,013; p=.03

PHYSICIAN VISITS AND SURVEILLANCE TESTING AMONG BREAST CANCER SURVIVORS IN TAIWAN Yong A. Wang1,2; An-Chen Feng2; Patricia Ganz1. 1UCLA Jonsson Comprehensive Cancer Center, Los Angeles, CA;

2Koo Foundation Sun Yat-Sen Cancer Center, Taipei, Taiwan. (Tracking ID #1933474)

BACKGROUND: The resource demands from the expanding cancer survivor population pose a significant challenge to health care systems worldwide. Guidelines recommend regular history, physical examination and mammography, but against routine surveillance imaging for detecting distant metastasis in survivors of breast cancer treated with curative

• intent. Studies of health services utilization of cancer survivors in Asia are limited. We examined patterns of physician

visits and surveillance testing in a major cancer center in Taiwan, and assessed their relationship to various patient and physician factors and trends over time. METHODS: All stage 0, I, II breast cancer cases were identified from the Taiwan Cancer Registry that were treated in the cancer center during 2002 to 2009. After excluding cases with recurrence, cases and institutional claims data were merged to measure the number of outpatient visits, breast imaging, and surveillance tests for detecting metastatic recurrence including serum tumor markers, abdominal imaging, chest X-ray, bone scan and PET, during the disease-free follow-up period from 13 to 60 months after initiation of breast cancer treatment. Poisson regression and logistic regression analyses were used to examine factors contributing to physician visits and testing, and separate analyses were conducted to look at individual patient cohorts from years 2002 to 2008 where sufficient follow up was available. RESULTS: A total of 3,488 women ages 19 to 88 were identified from the registry. The number of breast cancer-specific physician visits in the 48-month follow up period (mean 18.7 per patient, SD 8.5) and the rates of surveillance testing to detect distant metastasis were greater than those recommended by ASCO/NCCN guidelines. Variables predicting the number of physician visits included age, cancer stage, year of treatment, hormonal therapy, chemotherapy, radiotherapy, whether the patient had seen psychiatry or medical subspecialties other than oncology, and the primary physician/specialty following the patient for breast cancer. After adjusting for covariates, the number of visits was highest among patients primarily followed by medical oncologists and lowest among those followed by breast surgeons (Figure 1). The adjusted 48-month visit rates decreased progressively from the 2003 cohort (mean 22.7) to the 2008 cohort (mean 17.7), related to an institutional effort in care coordination for breast cancer survivors. There was a significant decrease in the adjusted rate of tumor marker testing from the 2002 cohort (99.7%) to the 2008 cohort (38.0%) associated with the change in the institutional guideline (Figure 2). Greater number of visits was correlated with more frequent testing. Patients primarily followed by medical oncologists had the highest adjusted testing rates for tumor markers, bone scan, and abdominal imaging. CONCLUSIONS: Early stage breast cancer survivors in Taiwan had high utilization of physician visits and surveillance testing for distant metastasis. This might be improved by greater coordination of care among breast cancer specialists. Implementation of an evidence-based guideline regarding the number of visits, imaging and tumor marker testing may be necessary to reduce the overutilization of testing.

MAMMOGRAPHY FACILITY CHARACTERISTICS ASSOCIATED WITH TIMELY FOLLOW-UP OF ABNORMAL MAMMOGRAMS Leah S. Karliner; Celia P. Kaplan; Lin Ma; Karla Kerlikowske. UCSF, San Francisco, CA. (Tracking ID #1934230) BACKGROUND: Delays in follow-up of abnormal mammograms are common, particularly for minority

• women. Delays can contribute to anxiety and women being diagnosed and treated for breast cancer at more

advanced stages. We investigated whether mammography facilities vary in their timeliness of follow-up and if facility characteristics are associated with length of follow-up times. METHODS: We examined time to follow-up of abnormal mammograms collected from 2005-2011 for 12 facilities participating in the San Francisco Mammography Registry (SFMR). We categorized facilities based on the proportion of exams without follow-up. For abnormal mammograms with a Breast Imaging Reporting and Data System (BI-RADS) 0 assessment, we categorized facilities with <=15% without imaging follow-up at 60- days as ‘short' follow-up and >15% as ‘long' follow-up. For abnormal BI-RADS 4/5 assessments, we categorized facilities with <=25% without biopsy follow-up at 30-days as ‘short' follow-up and >25% as ‘long' follow-up. For the BI-RADS 4/5 analysis, we excluded four facilities for which the SFMR does not have complete capture of biopsy data. We then examined facility characteristics pertaining to access and staffing, communication, and population served. RESULTS: There were 70,393 BI-RADS 0 assessments and 17,750 BI-RADS 4/5 assessments. For BI-RADS 0 assessments, the facilities in the ‘short' follow-up group completed diagnostic imaging for 95.5% of examinations at 60-days compared to 74.7% for the ‘long' follow-up group (p<.0001). For the BI-RADS 4/5 assessments, the ‘short' follow-up group completed biopsies for 81.6% of examinations at 30-days compared to 60.7% for the ‘long' follow-up group (p<.0001). For both BI-RADS 0 and 4/5 abnormal mammogram follow- up, the ‘short' follow-up facilities had higher screening and diagnostic mammography volume, more radiologists' FTE reading mammography, and shorter times to scheduling diagnostic tests compared to the ‘long' follow-up facilities. More ‘short' follow-up facilities contacted women directly about their abnormal

• result. Additionally, ‘long' follow-up facilities served more vulnerable populations, including a higher

proportion of minority, limited English proficient, and less educated women compared to ‘short' follow-up facilities. CONCLUSIONS: Clinically significant differences in follow-up of abnormal mammograms exist. Facilities serving more vulnerable populations have longer and less adequate follow-up completion that is likely related to availability of resources. Providing mammography facilities serving vulnerable populations with additional resources may decrease disparities in abnormal mammogram follow-up, and in turn timely diagnosis and treatment of breast cancer.

PAP VS. HPV VS CO-TEST FOR CIN2+ DETECTION: A RETROSPECTIVE STUDY OF PERFORMANCE AND COST Andrea L. Sikon1; Xian W. Jin1; Laura D. Lipold2; Sarah Schramm3; Kelly Nottingham3; Bo Hu4; Michael B. Rothberg1,3. 1Cleveland Clinic, Cleveland, OH; 2Cleveland Clinic, Cleveland, OH; 3Cleveland Clinic, Cleveland, OH; 4Cleveland Clinic, Cleveland, OH. (Tracking ID #1936679) BACKGROUND: Existing cervical cancer screening guidelines allow for two options for women aged 30 and

• lder: pap every three years (with reflex testing for HPV when cytology shows ASCUS) or pap and HPV co-

testing every 5 years. Some is known about the performance of these two strategies but comparisons of cost are even more limited. To understand the performance of existing as well as a novel strategy (i.e. initial HPV screening with reflex pap if positive,) we compared the expected costs per case of cervical intraepithelial neoplasia 2 or worse (CIN2+) detected by these strategies amongst a large cohort of women. METHODS: A retrospective chart review was performed of all women 30 years or older who had primary screening with both HPV and cervical cytology (i.e. co-testing) during the same encounter at the Cleveland Clinic from 2004 to 2010. A pathology database was utilized to identify biopsy-proven CIN2+ cases. Information on age, race, marital status, insurance type, and smoking history was also collected. For women who had abnormal screening tests but did not have documented colposcopy results, we performed data imputation based on co-testing results. Sensitivity, specificity, positive and negative predictive values were estimated for 3 strategies: Pap with reflex HPV, HPV with reflex pap and co-testing. For each strategy we calculated costs based on the number of HPV tests, pap tests and colposcopies that would be performed. RESULTS: The dataset included 99,549 women who had simultaneous cytology and HPV collected during the study period. Subjects were mostly white (78.4%), married/domestic partner (70.7%), nonsmokers (61.3%) and with private insurance (86.1%). Overall, 5121 subjects (5.14%, 95% CI=[5.01%, 5.28%]) were infected with HPV and 6115 (6.14%, 95% CI=[5.99%, 6.29%]) had cytology that demonstrated ASCUS or worse. Of these, 1,347 underwent colposcopy, and 605 subjects (0.61%; 95% CI=[0.56%. 0.66%]) had CIN2+. An additional 709 women had a combination of HPV+ and ASCUS or higher on cytology but did not have documented colposcopy results. Of these women, 183 had CIN2+ assigned by imputation, for an overall expected incidence rate of CIN2+ of 0.79% (95% CI=[0.74%. 0.85%]). Sensitivity for CIN2+ was 89.9% for cytology, 94.0% for HPV and 97.7% for co-testing; specificity was 97.7% for cytology, 98.3% for HPV and 97.7% for co-testing. The negative predictive value was 99.92% for cytology alone, 99.95% for HPV alone and 99.98% if both tests were negative. Pap with reflex HPV was least expensive, costing $31.6M and detecting 708 cases of CIN2+ , with 2965 colposcopies per 100,000 women screened. HPV with reflex pap would detect an additional 33 cases

• f CIN2+, and require 508 fewer colposcopies, at a marginal cost of $11.3M, or $342K per additional case
• detected. Compared to HPV with reflex pap, co-testing would detect an additional 29 cases, and require an

additional 570 colposcopies, at a marginal cost of $21.2M or $732K/case. The marginal cost per case would be higher in lower risk populations and vice versa. CONCLUSIONS: Screening women using HPV initially rather than pap could detect slightly more cases of CIN 2+ at a marginal cost of $342K/case in this large cohort. Co-testing all patients could detect even more cases, but at a marginal cost of $732K/case. Acceptable thresholds of additional cost per case to capture a relatively small number of additional cases are currently lacking and further cost effectiveness studies should be

• considered. In this era of value based care and personalized health care, considerations of varied screening

strategies according to individual risk as well as to cost per case is necessary. More sophisticated assessment of risk may help to better modify recommended screening methods and intervals and allow for concentration of more costly resources on those at highest risk while still providing sound screening for those at lower risk.

OVARIAN CANCER TREATMENT AND SURVIVAL TRENDS OVER PAST DECADE Jenny J. Lin1; Natalia Egorova2; Rebeca Franco2; Nina A. Bickell1,2. 1Icahn School of Medicine at Mount Sinai, New York, NY; 2Icahn School of Medicine at Mount Sinai, New York, NY. (Tracking ID #1939426) BACKGROUND: While death rates from ovarian cancer appear to be decreasing and five-year survival rates are increasing, it is unclear whether overall survival is improving among women with advanced ovarian cancer. We undertook this study to assess trends of overall survival, cancer recurrence and treatment patterns among women with advanced stage ovarian cancer. METHODS: Using SEER-Medicare claims data from 2000-2009, we identified a cohort of women age 65+ years with Medicare diagnosed with stage III and IV epithelial ovarian cancer. Recurrent cancer was identified using claims for chemotherapy, hospice or secondary cytoreductive surgery following a 6 month treatment-free window after primary treatment for ovarian cancer. Persistent cancer was defined as receipt of treatment (surgery or chemotherapy) without a treatment-free window of at least 6 months. Optimal treatment was defined as surgery within 120 days of diagnosis followed by chemotherapy within 120 days of initial surgery. Survival time was calculated from time of diagnosis. Parametric and semi-parametric multivariate survival analyses were used to assess comparative treatment survival rates and factors affecting survival and recurrence. RESULTS: Of the 46,004 women with ovarian cancer, 6,814 met the study criteria. Of these, only 3,413 women (50%) underwent primary debulking surgery and chemotherapy in an optimal time frame, 970 (14%) received surgery alone and 1,799 (26%) received chemotherapy alone, 214 received no treatment and 418 did not receive surgery and chemotherapy in an optimal time frame. During the past decade, overall median survival time remained relatively constant, although the time to first recurrence increased by 260 days and the probability of having a recurrence grew from 2000 to 2004 and stabilized during 2005-2009. Ovarian cancer was the cause of death for 85% of women diagnosed in 2000 and 72% for those who diagnosed in 2008. Optimal treatment was associated with longer survival time (p<0.001, median survival 1,411 days), a lower proportion of persistent cancers and higher proportion of patients with recurrences (see figure). Survival of patients receiving optimal treatment improved over the decade, while for those who received chemotherapy or surgery alone, survival probability did not change. However, the proportions of women who received optimal treatment versus other treatments did not vary over time. Multivariate survival models controlling for patient demographics, cancer type and comorbidities showed that patients who received optimal primary treatment or who had multiple recurrences had longer survival but there were not substantial changes in overall survival during the last decade. Propensity analyses are underway. CONCLUSIONS: Among elderly women diagnosed with advanced stage ovarian cancer, patterns of overall survival and treatment types have not changed in the past decade. Optimal cancer treatment is associated with improved survival time. Further work should evaluate barriers to receipt of optimal treatment among this group

• f patients.