Clinician Knowledge, Attitudes, and Barriers to Management of Vulvovaginal Atrophy

Variations in Primary Care and Gynecology

Kimberly K. Vesco, MD, MPH; Kate Beadle, NP, NCMP; Ashley Stoneburner, MPH; Joanna Bulkley, PhD; Michael C. Leo, PhD; Amanda L. Clark, MD, MCR, NCMP


Menopause. 2019;26(3):265-272. 

In This Article


Kaiser Permanente (KPNW) is a large nonprofit, integrated healthcare system serving over 570,000 members in the states of Oregon and southwest Washington. Members include individuals and families covered by commercial group and individual self-pay health plans, Medicare Advantage, and Medicaid. In preparation for a cluster randomized trial to test an intervention to improve the diagnosis and treatment of symptomatic VVA within the KPNW healthcare system, we collaborated with the Northwest Permanente Continuing Medical Education and Professional Development (NWP CME) Department to conduct a survey of PC and GYN clinicians practicing ambulatory medicine at KPNW.

The target population for the survey and intervention was clinicians who performed ''well care'' (annual or routine) visits for adult women. In our health system, these visits are conducted by physicians and advanced practitioners (nurse practitioners, certified nurse midwives, and physician assistants) in the departments of internal medicine, family medicine, and obstetrics and gynecology (OBGYN). For the survey and planned intervention, we categorized all clinicians who had an ambulatory clinic practice within internal and family medicine as primary care clinicians and all clinicians practicing within the department of OBGYN as gynecology clinicians. A query of the KPNW electronic health record in 2013 showed that 83% of visits coded as ''well care'' for all women aged 55 and older are conducted in PC and 17% are conducted in GYN. When stratified by age, 73% of women aged 55 to 59 and 94% of women aged ≥75 receive their ''well care'' visit in primary care (Table 1).

We developed the clinician survey in the context of a quality improvement initiative to improve the diagnosis and management of symptomatic VVA.[13] The goal of the initiative was to test whether providing targeted clinician education and electronic health record (EHR) patient care tools increases the detection and treatment of VVA among postmenopausal women. The study design for the intervention was a cluster randomized trial within the PC and GYN departments at KPNW. This survey was designed to inform the development of the clinician education materials and the EHR patient care tools for the intervention trial. The survey had three content areas related to diagnosis and treatment of VVA: (1) clinician knowledge, (2) practice behavior and confidence, and (3) perceived barriers. Because GSM terminology was not in active use at the time we developed the survey, we used the terminology ''vulvovaginal atrophy'' and ''VVA'' rather than GSM. We did, however, include survey questions about urinary symptoms. See Supplementary Digital Content 1, for the full contents of the survey questionnaire.

We assessed knowledge about VVA prevalence, diagnosis, and treatment by asking eight multiple-choice questions. We adapted these questions from a 2014 CME course, ''Postmenopausal Vulvovaginal Atrophy: Breaking Down the Barriers to Identification and Treatment,'' which was produced by Haymarket Medical Education and sponsored by NAMS (content used with permission from NAMS). We reported the knowledge score as the proportion of correct responses.

The practice behavior section contained three questions regarding clinicians' likelihood of assessing for VVA symptoms at a routine (well) visit, their confidence in advising patients about VVA symptoms, and their confidence in counseling about the risks and benefits of local estrogen therapy. The responses contained a five-point scale ranging from highly unlikely to highly likely and from very low to very high.

We also asked clinicians, ''Which of the following are barriers to identification and treatment of VVA among patients in your practice?'' From a 12-item list, clinicians were advised: ''Check all that apply.'' We developed the list of barriers based on an existing survey that asked physicians to identify challenges to communicating with patients about menopause[12] and on surveys of postmenopausal women regarding VVA symptoms and treatment.[10,14] Prior clinician-identified barriers included lack of time, complicated discussions regarding treatment options, lack of available information about treatment options, and patient embarrassment with asking questions.[12] Prior patient-identified barriers included concerns about the long-term safety of prescription therapy use for the treatment of VVA symptoms, including breast cancer risk,[10,14] concerns about side effects of therapy and ineffective symptom relief,[10] and lack of information about symptoms and treatment of vaginal discomfort.[14] Because part of our planned intervention included the development of EHR tools, we added a statement to our list of barriers about lack of support tools for diagnosis and treatment of VVA. We also asked whether incorporation of systemic estrogen products into the American Geriatric Society (AGS) and NCQA HEDIS list of drugs that are high risk for use in the older population[15,16] was a barrier to treatment of VVA. We were concerned that this warning against systemic estrogen use was being applied by clinicians to vaginal estrogen products as well.

We submitted the survey to the NWP CME Department, which programmed it using Survey Monkey (, San Mateo, CA). In August 2014, the CME Department sent an e-mail request to all PC and GYN clinicians (n=360) with a valid e-mail address asking them to complete the survey and included the survey link. We introduced the survey to clinicians as a request for information to guide development of clinician education materials and support tools for the diagnosis and treatment of symptomatic vulvovaginal atrophy. All data were collected electronically. There was a 4-week time frame for completion, which included two e-mail reminders to complete the survey. We received IRB approval to obtain and analyze the data collected by the CME Department and to disseminate the survey results (approval date 6/5/2014 and Institutional Review Board #PRO 4685). Because the survey was an internal quality improvement initiative, informed consent was not applicable.

The NWP CME Department prepared a dataset of survey responses and shared them with our team for analysis. To ascertain potential nonresponse bias, we compared responders to nonresponders by age, sex, department of practice, and clinician type (physician vs advanced practitioner) using Fisher's exact test for categorical variables and t test for age. We examined descriptive statistics for all survey items to check for unrealistic or outlying values and assess missing data.

To determine whether PC and GYN clinicians differed in their answers to the three Likert-scale-based practice questions, we performed multiple regression. For binary outcomes, which included the individual knowledge and barrier questions, we used multiple logistic regression. In all regression models, the independent variable was an indicator for clinician department (GYN=1, PC=0) and covariates were age, sex, and clinician type (physician vs advanced practitioner). Because a high proportion of participants answered certain knowledge items correctly (ie, situation with small sample sizes in certain cells), conventional logistic regression can lead to estimation problems and result in biased estimates,[17] thus we applied the Firth method, which uses penalized likelihood to generate the regression estimates.[18,19] We reported the unstandardized regression coefficients for the multiple regression analyses and odds ratios (OR) for the logistic regression analyses, along with 95% CIs for these analyses.