A Systematic Review of the Bimanual Examination as a Test for Ovarian Cancer Mark H. Ebell, MD, MS, MaryBeth Culp, BS, Krista Lastinger, BS, Tara Dasigi, MPH This activity is available for CME credit. See page A4 for information.
Context: An annual bimanual pelvic examination remains widely recommended for healthy women, but its inclusion may discourage attendance. Our goal was to determine the accuracy of the pelvic examination as a screening test for ovarian cancer and to distinguish benign from malignant lesions. Evidence acquisition: PubMed was searched to identify studies evaluating the accuracy of the bimanual pelvic examination for ovarian cancer diagnosis. Data regarding study design, study quality, and test accuracy were abstracted. Heterogeneity was evaluated and meta-analysis performed where appropriate, including bivariate receiver operating characteristic curves. Evidence synthesis: Eight studies in screening populations (n¼36,599) and seven studies in symptomatic patients (n¼782) were identified. Search was completed in November 2013; included studies were published between 1988 and 2009. Screening studies were homogeneous; the summary estimates of sensitivity and specificity of the pelvic examination as a screening test for ovarian cancer were 0.44 and 0.98 (positive likelihood ratio, 24.7; negative likelihood ratio, 0.57). For distinguishing benign versus malignant lesions, there was considerable heterogeneity, with a range of sensitivity from 0.43 to 0.93 and specificity from 0.53 to 0.91.
Conclusions: The bimanual pelvic examination lacks accuracy as a screening test for ovarian cancer and as a way to distinguish benign from malignant lesions. In a typical screening population, the positive predictive value of an abnormal pelvic examination is only 1% (95% CI¼0.67%, 3.0%). Its inclusion in a health maintenance examination cannot be justified on the basis of using it to screen for ovarian cancer. (Am J Prev Med 2015;48(3):350–356) & 2015 American Journal of Preventive Medicine
Introduction
A
pelvic examination is routinely performed on women as part of the “annual physical” or “periodic health exam.” It typically includes inspection of the external genitalia, a speculum examination of the vagina and cervix (almost always accompanied by a Pap test), and a bimanual examination to palpate the uterus and adnexa. The American College of Obstetrics and Gynecology1 recently reiterated its support for performing a bimanual examination annually in otherwise healthy women aged Z21 years, although they note that it may be omitted if desired for otherwise
From the Department of Epidemiology and Biostatistics, College of Public Health, University of Georgia, Athens, Georgia Address correspondence to: Mark H. Ebell, MD, MS, 233 Miller Hall, University of Georgia Health Sciences Campus, Athens GA 30602. E-mail: [email protected]. 0749-3797/$36.00 http://dx.doi.org/10.1016/j.amepre.2014.10.007
350 Am J Prev Med 2015;48(3):350–356
healthy women who have had a total hysterectomy and bilateral salpingo-oophorectomy for benign indications. Reasons given by physicians for performing the bimanual pelvic examination include that it is a standard part of a well woman examination, to screen for ovarian cancer, to screen for other gynecologic cancers, to screen for sexually transmitted infections (STIs),2 and as a requirement for hormonal contraception.3 A study4 surveyed physicians and found that 47% of obstetrician-gynecologists believed that the pelvic examination is an effective screening test for ovarian cancer. However, the bimanual examination is not needed to screen for STIs1 and is not recommended as a screening test for gynecologic cancer by the American Cancer Society5 or the U.S. Preventive Services Task Force.6,7 It is also no longer recommended as a requirement before receiving oral contraceptives.1,8,9 Although many women have had an annual pelvic examination in association with screening for cervical cancer, the interval for this screening has increased to 3–5 years for average-
& 2015 American Journal of Preventive Medicine
Published by Elsevier Inc.
Ebell et al / Am J Prev Med 2015;48(3):350–356
risk women, calling into question the need to perform an annual bimanual pelvic examination that is not accompanied by a Pap test.7 In fact, the knowledge that an annual health maintenance visit is likely to include a bimanual pelvic examination, which many women view as invasive and uncomfortable, can cause anxiety10 and may actually discourage attendance.11 As a consequence, some women may not receive other important clinical preventive services. Perhaps even more important is the opportunity cost incurred by performing a pelvic during a periodic health examination, time that might be better spent delivering clinical preventive services that have been proven effective.12 The objective of the current study is to systematically review the literature regarding the accuracy of the bimanual examination as a screening test for ovarian cancer. A secondary goal is to examine the ability of the pelvic examination to distinguish benign from malignant lesions in women with a known abnormality, the other reason for using the bimanual examination related to ovarian cancer.
Figure 1. Prisma flow diagram. March 2015
351
Methods Inclusion Criteria Two types of studies were identified, screening studies and studies of symptomatic patients. In screening studies, the goal of the bimanual examination was to detect ovarian cancer in an asymptomatic woman. Screening studies that performed a bimanual pelvic examination in asymptomatic women and reported enough information to calculate the sensitivity, specificity, or both for the diagnosis of ovarian malignancy were included. Acceptable reference standards included ultrasound, cancer antigen (CA)-125 testing followed by biopsy if suspicious, or both; laparotomy or laparoscopy; or repeated screening or long-term follow-up in r12 months. A second group of studies of symptomatic women was identified, where the intention of the bimanual pelvic examination was to distinguish malignant from benign abnormalities. Prospective cohort studies of women with a known pelvic abnormality who underwent bimanual examination prior to surgery and that reported enough information to calculate the sensitivity and specificity for malignant versus benign lesion were included. Studies where bimanual examinations were performed under anesthesia were excluded, as this is not representative of usual
March 2001– June 2003
Greece Palpable adnexal mass
Palpable mass
Mean 58.1 years, range 45–80 years
Mean 58.4 years, range 50–85 years
Asymptomatic women aged Z45 years without any evidence of adnexal pathology
Asymptomatic peri- and postmenopausal women aged Z50 years
2,000
3,201
Adonakis (1996)20
Kurjak (2005)21
([bimanual OR pelvic exam OR pelvic examination] AND [ovarian cancer OR ovarian malignancy]) OR ([physical examination OR physical exam OR clinical exam OR clinical examination OR pelvic examination OR bimanual examination OR pelvic exam OR bimanual exam] AND [sensitivity OR specificity OR accuracy OR predictive value OR likelihood ratio] AND [ovarian neoplasms [MeSH Terms]]) OR ([pelvic examination OR bimanual examination] AND [sensitivity OR specificity] AND [ovary OR ovarian OR adnexal]) OR ([pelvic examination OR bimanual examination] AND [sensitivity OR specificity OR likelihood ratio])
The search was performed in August 2013 and updated in November 2013. Two researchers reviewed each abstract separately to identify studies that appeared to report the accuracy of bimanual pelvic examination for the detection of ovarian malignancy. The full text of each study identified by at least one of the researchers was reviewed by both for final inclusion or exclusion in the study. In addition, the reference lists of review articles, including a 2006 evidence report from the Agency for Healthcare Research and Quality,13 were reviewed to identify additional studies that may have been missed by the PubMed search.
Data Abstraction Data regarding study design, study quality, and diagnostic accuracy were abstracted in tandem by at least two researchers. Inconsistencies were reconciled by consensus, and the final data abstraction was reviewed in detail by the lead investigator (MHE). Studies were classified as being in a screening population if the intent was to detect cancer using the bimanual pelvic examination, or as being in a symptomatic population of women with a known pelvic mass if the intent was to distinguish benign from malignant lesions. Screening studies were further classified as average risk or high risk. High risk was defined as having a BRCA1 or 2 mutation or having a strong family history for breast or ovarian cancer. Studies enrolling predominantly average-risk postmenopausal women but also enrolling some younger women with a family history were classified as average risk.
Quality Assessment
GI, gastrointestinal; NR, not reported.
2,623 Grover (1995)19
van Nagell (1995)18
Search Strategy
Croatia
Australia Bulky/fibroid uterus or adnexal abnormality Mean 51 years, range 25–92 years Asymptomatic volunteers
NR
U.S. Palpable mass Mean 56 years, range 25–92 years Asymptomatic women aged Z50 years and women Z25 years with a family history of ovarian cancer
1987–2005
United Kingdom Palpable adnexal mass of any size separate from uterus/GI tract Median 54 years, range 45–83 years Apparently healthy women aged 445 years and amenorrheic 412 months 1,010 Jacobs (1988)17
Screening average risk
25,327
Country Definition of abnormal exam
NR
Date of data collection
outpatient clinical practice. For both clinical questions, case–control studies, surveillance studies in patients with known cancer, and studies in children were excluded.
The following search strategy for PubMed, limited to “Humans” and “has abstract” was used:
Age Patient population n Author (year)
Table 1. Characteristics of Included Studies
March 1991– June 1993
Ebell et al / Am J Prev Med 2015;48(3):350–356
352
Only cohort studies were included, whereas case–control studies were excluded, as the latter previously have been shown to be at high risk of bias.14 The following study quality characteristics were evaluated: the reference standard, whether data collection was prospective or retrospective, presence of verification bias, and blinding of those performing the index test and of those performing the reference standard test. Each potential threat to validity was rated as yes if present, no if absent, and unclear if it could not be determined after careful review of the source material.
Analysis The I2 statistic was used to assess the heterogeneity of included studies for each question. The I2 statistic has a value between 0 and 100, www.ajpmonline.org
I ¼0.00 I
2,a
Summary effect
Van der Velde (2009)
Olivier (2006)
The I2 statistic has a value between 0 and 100; 0 represents perfect homogeneity and 100 is perfect heterogeneity. Although I2 4 0.5, this is an artifact of the small number of studies but large size and precision of each study. Therefore, the summary measure of specificity is reported with its 95% CI. b
a
I2¼0.00 I ¼0.00
0.57 (0.45, 0.72)
I ¼73.7
2
24.7 (15.8, 37)
b 2 2
0.44 (0.32, 0.57)
0.98 (0.97, 0.99)
0.67 (0.30, 1.5) 27.6 (4.2, 180) 0.99 (0.97, 1.0) 0.33 (0.01, 0.91) 1/3
4/331
19.3 (7.8, 48) 0.40 (0.12, 0.74)
24
4/10
15/722
0.98 (0.97, 0.99)
0.61 (0.37, 1.0)
0.76 (0.34, 1.0) 29.8 (2.5, 352) 0.99 (0.99, 1.0) 0.00 (0.00, 0.98) 0/1
23
Oei (2006)22
Kurjak (2005)
Screening (high-risk population)
3/5
2/3
11/1,371
0.60 (0.15, 0.95)
0.67 (0.09, 0.99)
23.4 (10.1, 54.1) 21
Adonakis (1996)
Van Nagell (1995)
20
19/44
57/1,997
0.43 (0.28, 0.59)
0.97 (0.96, 0.98)
0.34 (0.07, 1.7)
0.76 (0.34, 1.7) 16.2 (1.4, 182) 0.00 (0.00, 0.98)
18
Grover (1995)
19
Jacobs (1988)17
Screening (average-risk population)
0/1
40/2,622
0.98 (0.98, 0.99)
0.26 (0.02, 1.0) 27.6 (11.4, 66) 0.97 (0.96, 0.98) 1.00 (0.03, 1.0)
March 2015
1/1
27/999
Likelihood ratio positive (95% CI) Specificity (95% CI) Sensitivity (95% CI) Abnormal pelvic exam/ total without cancer Abnormal pelvic exam/ total with cancer Author (year)
Table 2. Accuracy of Pelvic Exam for Diagnosis of Ovarian Cancer in Screening Studies
Likelihood ratio negative (95% CI)
Ebell et al / Am J Prev Med 2015;48(3):350–356
353
where 0 represents perfect homogeneity and 100 is perfect heterogeneity. As this statistic is unreliable when there is a small number of studies,15 the authors also used visual inspection of forest plots to assess heterogeneity. If there was adequate homogeneity (I2o0.5), bivariate metaanalysis was performed using the meta-analytical integration of diagnostic accuracy (MIDAS) procedure16 in Stata, version 13.1, to calculate summary measures of sensitivity, specificity, and likelihood ratios. MIDAS uses a bivariate mixed-effects regression model that has been modified for meta-analysis of diagnostic accuracy studies. It jointly analyzes the pairs of sensitivity and specificity, taking into account correlation between these parameters. Bivariate summary receiver operating characteristic curves were drawn, again using the MIDAS procedure in Stata, version 13.1, and includes 95% CIs and predictive intervals. The MIDAS procedure graphs the mean logit sensitivity and specificity with their SEs and 95% CIs. Where there were zero events, a continuity correction of 0.5 was used by the MIDAS procedure.
Results The PubMed search returned a total of 303 articles, and previous systematic reviews and reference lists of included studies identified another 24 studies, for a total of 315 unique studies after elimination of duplicates. A total of 56 studies met the initial review and were examined in full. Forty-one studies were excluded because they did not meet the inclusion criteria, most often because of failure to report sensitivity or specificity, failure to address one of our study questions, or because they were studying surveillance in patients with known malignancy. Ultimately, 15 studies met the inclusion criteria and were included in the analysis (Figure 1, Table 1, and Appendix Table 1). Included studies were published between 1988 and 2009.
Characteristics of Included Studies Eight screening studies with 36,599 asymptomatic women17–24 and seven studies of 782 symptomatic women25–31 met the inclusion and exclusion criteria. Study characteristics are summarized in Table 1 and Appendix Table 1. Three of the screening studies enrolled women at high risk for ovarian cancer based on their family history or the presence of a BRCA mutation,22–24 whereas the other five studies predominantly enrolled postmenopausal asymptomatic women at average risk for ovarian cancer.17–21 All of the studies with symptomatic women enrolled those scheduled for exploratory laparotomy for adnexal
354
Ebell et al / Am J Prev Med 2015;48(3):350–356
The summary estimates for sensitivity and specificity were 0.44 (95% CI¼0.32, 0.57) and 0.98 (95% CI¼0.97, 0.99), respectively. The respective summary estimates for the positive and negative likelihood ratios were 24.7 (95% CI¼15.8, 37) and 0.57 (95% CI¼0.45, 0.72). Thus, an abnormal pelvic examination greatly increased the likelihood of ovarian cancer, whereas a normal examination was of little value in ruling out cancer. Less than half of ovarian cancers were detected using the pelvic examination. The bivariate summary receiver operating characteristic curve is shown in Figure 2. The c-statistic of 0.81 (95% CI¼0.77, 0.84) showed fair to good discrimination, but the predictive interval was broad, encompassing sensitivities from approximately 25% to 62%.
Accuracy of Bimanual Examination for Distinguishing Malignant from Benign Disease Figure 2. Summary bivariate receiver operating characteristic (ROC) curve for pelvic exam as a screening test in asymptomatic women. Note: The 95% confidence contour defines the 95% CI for the summary estimates of sensitivity and specificity, whereas the 95% prediction contour indicates the range where a future study is likely to lie.
mass of unknown origin. Three were set in the U.S., and they largely enrolled women aged 445 years.28,30,31
Quality Assessment The quality of included studies is summarized in Appendix Table 2. All had a cohort design and all but one used prospective data collection. All of the screening studies suffered from verification bias, as women with normal pelvic examination, ultrasound, and with or without CA-125 (the protocols varied) had repeat screening 1 year later, whereas those with an abnormal screening test had more intensive follow-up that often included laparotomy or laparoscopy. Blinding was inconsistently reported, although is unlikely to be of consequence for those performing surgery or evaluating the histopathology.
Accuracy of Bimanual Pelvic Examination as a Screening Test The accuracy of the bimanual examination as a screening test for ovarian cancer is summarized in Table 2. There was excellent homogeneity between studies, with the exception of specificity. However, this was likely an artifact due to the small number of studies and narrow CI for each study.15 The specificities ranged from 0.97 to 0.99, and each CI ranged from 0.96 to 1.0. Therefore, summary estimates are reported for this specificity.
Data from studies of symptomatic women undergoing laparotomy for a known pelvic mass are summarized in Table 3. There was significant heterogeneity for the sensitivity, specificity, and negative likelihood ratio, and moderate heterogeneity for the positive likelihood ratio. Therefore, only a summary measure of accuracy for the positive likelihood ratio was calculated (3.4, 95% CI¼2.8, 4.3). The sensitivity (0.43–0.93), specificity (0.53–0.91), and negative likelihood ratio (0.12–0.62) had broad ranges. The bivariate summary receiver operating characteristic curve is shown in Figure 3, with a c-statistic of 0.83. However, there was significant heterogeneity and the CIs and predictive intervals were broad.
Discussion The bimanual pelvic examination has very low sensitivity for the diagnosis of ovarian cancer. Although the specificity of 0.98 appears high, it is important to remember that the examination is being used to detect a rare disease. The Prostate Lung Colon Ovary (PLCO) trial detected approximately 5 cases of ovarian cancer per 10,000 person-years of follow-up.32 Thus, if 10,000 women are screened, the 44% sensitivity means that approximately two of the five cancers would be detected. However, the false-positive rate of 2% would result in 200 false-positive abnormal pelvic examinations, a positive predictive value of only 1.0%. Using the 95% CIs for specificity of 97%–99% and sensitivity of 32%–57%, the range for positive predictive value is 0.67%–3.0%. This study did not assess the impact of screening using the pelvic examination on mortality or other clinical outcomes; the focus was on accuracy. However, the PLCO trial found that even annual ultrasound did not reduce ovarian cancer specific or all-cause mortality.32 Thus, it seems unlikely that www.ajpmonline.org
Where I2, a measure of heterogeneity, exceeded 0.5, we report only the range. Where it is less than 0.5, we report the summary measure of effect and the 95% CI. The I2 statistic has a value between 0 and 100; 0 represents perfect homogeneity and 100 is perfect heterogeneity. b
a
I2,b
Summary effect
a
Ghaemmaghami (2008)
Balbi (2001)
25
Buckshee (1998)
Roman (1997)
30
Buist (1994)
27
355
I2¼85.0 I2¼40.8 I2¼87.9
I2¼83.8
0.12, 0.62 3.4 (2.8, 4.3) 0.43, 0.93
0.53, 0.91
0.35 (0.24, 0.49) 4.0 (2.3, 7.0) 63/88
11/62
0.72 (0.61, 0.81)
0.82 (0.70, 0.91)
0.12 (0.03, 0.47) 3.5 (2.2, 5.7)
March 2015
26
29
20/22
13/50
0.91 (0.71, 0.99)
0.74 (0.60, 0.85)
0.25 (0.7, 0.85) 7.0 (2.3, 21.5) 7/9
3/27
0.78 (0.40, 0.97)
0.89 (0.71, 0.98)
0.58 (0.43, 0.80) 3.1 (2.0, 4.8) 22/43
30/183
0.51 (0.35, 0.67)
0.84 (0.77, 0.89)
0.59 (0.32, 2.1) 1.5 (0.87, 2.4) 31/45
9/19
0.69 (0.53, 0.82)
0.53 (0.29, 0.76
0.12 (0.06, 0.24) 2.5 (2.0, 3.2) 88/95 Schutter (1994)31
47/127
0.93 (0.85, 0.97)
0.63 (0.54, 0.71)
0.62 (0.46, 0.83) 5.0 (2.1, 11.6) 0.91 (0.82, 0.97) 16/37 Finkler (1988)28
6/70
0.43 (0.27, 0.61)
Likelihood ratio positive (95% CI) Specificity (95% CI) Sensitivity (95% CI) Abnormal pelvic exam/ benign disease Abnormal pelvic exam/ malignant disease Author (year)
Table 3. Accuracy of Pelvic Exam for Distinguishing Benign from Malignant Lesions in Patients with Pelvic Mass
Likelihood ratio negative (95% CI)
Ebell et al / Am J Prev Med 2015;48(3):350–356
Figure 3. Summary bivariate receiver operating characteristic (ROC) curve for pelvic exam as a test to distinguish benign from malignant lesions in women with pelvic mass. Note: The 95% confidence contour defines the 95% CI for the summary estimates of sensitivity and specificity, whereas the 95% prediction contour indicates the range where a future study is likely to lie.
the pelvic examination, a much less sensitive test, would be able to have any impact on mortality. The pelvic examination is also an unreliable test for distinguishing between benign and malignant lesions. This is not surprising, and it is why researchers have developed predictive indices such as the Risk of Malignancy Index that incorporate ultrasound, CA-125, and menopausal status.33,34 The most important limitation of this study is related to the quality of the included studies. Verification bias was an important concern in all of the screening studies, as patients with a negative screening test did not undergo further investigation until the next screening round. However, it is unlikely that future studies would overcome this bias by performing a surgical biopsy in all women with normal screening tests, or even a random sample of such women. Masking was also rarely reported, although we assumed it when the examination took place prior to surgery. Studies of symptomatic patients were generally small with uncertain inclusion criteria. Our finding that the bimanual examination is not an effective screening test for ovarian cancer supports the increasingly common practice of foregoing the traditional annual bimanual pelvic examination in favor of other clinical preventive services for which there is better evidence.12,35 It is also consistent with guidelines5–7 that do not recommend the bimanual examination as a screening test for ovarian cancer. It will be important
356
Ebell et al / Am J Prev Med 2015;48(3):350–356
to clearly communicate that not doing a bimanual pelvic examination every year is not a way to save money or limit access to healthcare services. Rather, it is an attempt to refocus the encounter with the physician on the full spectrum of clinical preventive services. Doing this has the potential to improve women’s health, remove a barrier to access, and reduce the cost of care. This study was not funded or sponsored. It was conceived, carried out, and written by the authors under the lead of Dr. Ebell. The authors do not have any relevant conflicts of interest, either financial or professional. Dr. Ebell is a member of the U.S. Preventive Services Task Force (USPSTF). This article does not necessarily represent the views and policies of the USPSTF. No financial disclosures were reported by the authors of this paper.
References 1. Committee on Gynecologic Practice. Committee opinion no. 534: wellwoman visit. Obstet Gynecol 2012;120(2 Pt 1):421–4. 2. Schachter J, Schafer M, Young M, Ott M. Routine pelvic examinations in asymptomatic young women. N Engl J Med 1996;334:1847–8. 3. Stormo AR, Hawkins NA, Cooper CP, Saraiya M. The pelvic examination as a screening tool: practices of US physicians. Arch Intern Med 2011;171(22):2053–4. 4. Henderson JT, Harper CC, Gutin S, et al. Routine bimanual pelvic examinations: practices and beliefs of US obstetrician-gynecologists. Am J Obstet Gynecol 2013;208(2):109.e1–7. 5. American Cancer Society guidelines for the early detection of cancer, 2013. www.cancer.org/healthy/findcancerearly/cancerscreeningguidelines/ american-cancer-society-guidelines-for-the-early-detection-of-cancer. 6. U.S. Preventive Services Task Force. Screening for ovarian cancer: recommendation statement. Ann Fam Med 2004;2(3):260–2. dx.doi.org/ 10.1370/afm.200. 7. Moyer VA. U.S. Preventive Services Task Force. Screening for ovarian cancer: USPSTF reaffirmation recommendation statement. Ann Intern Med 2012;157(12):900–4. 8. Stewart FH, Harper CC, Ellertson CE, et al. Clinical breast and pelvic examination requirements for hormonal contraception: current practice vs evidence. JAMA 2001;285(17):2232–9. 9. Henderson JT, Sawaya GF, Blum M, et al. Pelvic examinations and access to oral hormonal contraception. Obstet Gynecol 2010;116(6): 1257–64. 10. Reddy DM, Wasserman SA. Patient anxiety during gynecologic examinations. Behavioral indicators. J Reprod Med 1997;42(10):631–6. 11. Lacy NL, Paulman A, Lovejoy B. Why we don’t come: patient perceptions on no-shows. Ann Fam Med 2004;2(6):541–5. 12. Nguyen GT, Cronholm PF. The annual pelvic examination: preventive time not well spent. Am Fam Physician 2013;87(1):8–9. 13. Myers ER, Bastian LA, Havrilesky LJ, et al. Management of adnexal mass. Evidence report/technology assessment No. 130 (prepared by the Duke Evidence-based Practice Center under Contract No. 290-020025.) AHRQ Publication No. 06-E004. Rockville MD: Agency for Healthcare Research and Quality, February 2006. 14. Lijmer JG, Mol BW, Heisterkamp S, et al. Empirical evidence of designrelated bias in studies of diagnostic tests. JAMA 1999;282(11):1061–6. 15. Thorlund K, Imberger G, Johnston BC, et al. Evolution of heterogeneity (I2) estimates and their 95% confidence intervals in large metaanalyses. PLoS One 2012;7(7):e39471.
16. Dwamena BA. MIDAS: a program for meta-analytical integration of diagnostic accuracy studies in Stata. Ann Arbor, MI: Division of Nuclear Medicine, Department of Radiology, University of Michigan Medical School; 2007. 17. Jacobs I, Bridges J, Reynolds C, et al. Multimodal approach to screening for ovarian cancer. Lancet 1988;331(8580):268–71. 18. van Nagell JR Jr, DePriest PD, Ueland FR, et al. Ovarian cancer screening with annual transvaginal sonography. Cancer 2007;109(9):1887–96. 19. Grover SR, Quinn MA. Is there any value in bimanual pelvic examination as a screening test? Med J Aust 1995;162(8):408–10. 20. Adonakis GL, Paraskevaidis E, Tsiga S, et al. A combined approach for the early detection of ovarian cancer in asymptomatic women. Eur J Obstet Gynecol Reprod Biol 1996;65(2):221–5. 21. Kurjak A, Prka M, Arenas JM, et al. Three-dimensional ultrasonography and power Doppler in ovarian cancer screening of asymptomatic peri- and postmenopausal women. Croat Med J 2005;46(5):757–64. 22. Oei AL, Massuger LF, Bulten J, et al. Surveillance of women at high risk for hereditary ovarian cancer is inefficient. Br J Cancer 2006;94(6):814–9. 23. Olivier RI, Lubsen-Brandsma MA, Verhoef S, van Buerden M. CA125 and transvaginal ultrasound monitoring in high-risk women cannot prevent the diagnosis of advanced ovarian cancer. Gynecol Oncol 2006;100(1):20–6. 24. van der Velde NM, Mourits MJ, Arts HJ, et al. Time to stop ovarian cancer screening in BRCA1/2 mutation carriers? Int J Cancer 2009;124 (4):919–23. 25. Balbi GC, Musone R, Menditto A, et al. Women with a pelvic mass: indicators of malignancy. Eur J Gynaecol Oncol 2001;22(6):459–62. 26. Buckshee K, Temsu I, Bhatla N, Deka D. Pelvic examination, transvaginal ultrasound and transvaginal color Doppler sonography as predictors of ovarian cancer. Int J Gynaecol Obstet 1998;61(1):51–7. 27. Buist MR, Golding RP, Burger CW, et al. Comparative evaluation of diagnostic methods in ovarian carcinoma with emphasis on CT and MRI. Gynecol Oncol 1994;52(2):191–8. 28. Finkler NJ, Benacerraf B, Lavin PT, et al. Comparison of serum CA 125, clinical impression, and ultrasound in the preoperative evaluation of ovarian masses. Obstet Gynecol 1988;72(4):659–64. 29. Ghaemmaghami F, Fakour F, Zarchi MK, et al. Clinical assessment, gross examination, frozen section of ovarian masses: do patients benefit? Arch Gynaecol Obstet 2008;278(3):209–13. 30. Roman LD, Muderspach LI, Stein SM, et al. Pelvic examination, tumor marker level, and gray-scale and Doppler sonography in the prediction of pelvic cancer. Obstet Gynecol 1997;89(4):493–500. 31. Schutter EM, Kenemans P, Sohn C, et al. Diagnostic value of pelvic examination, ultrasound, and serum CA125 in postmenopausal women with a pelvic mass. Cancer 1994;74(4):1398–406. 32. Buys SS, Partridge E, Black A, et al. for the PLCO Project Team. Effect of screening on ovarian cancer mortality. The Prostate, Lung, Colorectal and Ovarian (PLCO) cancer screening randomized controlled trial. JAMA 2011;305(22):2295–303. 33. Jacobs I, Oram D, Fairbanks J, et al. A risk of malignancy index incorporating CA 125, ultrasound and menopausal status for the accurate preoperative diagnosis of ovarian cancer. Br J Obstet Gynaecol 1990;97(10):922–9. 34. Yamamoto Y, Yamada R, Oguri H, et al. Comparison of four malignancy risk indices in the preoperative evaluation of patients with pelvic masses. Eur J Obstet Gynecol Reprod Biol 2009;144(2):163–7. 35. Stewart RA. Routine pelvic examination for asymptomatic women: exploring the evidence. Aust Fam Physician 2006;35(11):873–6.
Appendix Supplementary data Supplementary data associated with this article can be found at http://dx.doi.org/10.1016/j.amepre.2014.10.007. www.ajpmonline.org
Context: An annual bimanual pelvic examination remains widely recommended for healthy women, but its inclusion may discourage attendance. Our goal was to determine the accuracy of the pelvic examination as a screening test for ovarian cancer and to distinguish benign from malignant lesions. Evidence acquisition: PubMed was searched to identify studies evaluating the accuracy of the bimanual pelvic examination for ovarian cancer diagnosis. Data regarding study design, study quality, and test accuracy were abstracted. Heterogeneity was evaluated and meta-analysis performed where appropriate, including bivariate receiver operating characteristic curves. Evidence synthesis: Eight studies in screening populations (n¼36,599) and seven studies in symptomatic patients (n¼782) were identified. Search was completed in November 2013; included studies were published between 1988 and 2009. Screening studies were homogeneous; the summary estimates of sensitivity and specificity of the pelvic examination as a screening test for ovarian cancer were 0.44 and 0.98 (positive likelihood ratio, 24.7; negative likelihood ratio, 0.57). For distinguishing benign versus malignant lesions, there was considerable heterogeneity, with a range of sensitivity from 0.43 to 0.93 and specificity from 0.53 to 0.91.
Conclusions: The bimanual pelvic examination lacks accuracy as a screening test for ovarian cancer and as a way to distinguish benign from malignant lesions. In a typical screening population, the positive predictive value of an abnormal pelvic examination is only 1% (95% CI¼0.67%, 3.0%). Its inclusion in a health maintenance examination cannot be justified on the basis of using it to screen for ovarian cancer. (Am J Prev Med 2015;48(3):350–356) & 2015 American Journal of Preventive Medicine
Introduction
A
pelvic examination is routinely performed on women as part of the “annual physical” or “periodic health exam.” It typically includes inspection of the external genitalia, a speculum examination of the vagina and cervix (almost always accompanied by a Pap test), and a bimanual examination to palpate the uterus and adnexa. The American College of Obstetrics and Gynecology1 recently reiterated its support for performing a bimanual examination annually in otherwise healthy women aged Z21 years, although they note that it may be omitted if desired for otherwise
From the Department of Epidemiology and Biostatistics, College of Public Health, University of Georgia, Athens, Georgia Address correspondence to: Mark H. Ebell, MD, MS, 233 Miller Hall, University of Georgia Health Sciences Campus, Athens GA 30602. E-mail: [email protected]. 0749-3797/$36.00 http://dx.doi.org/10.1016/j.amepre.2014.10.007
350 Am J Prev Med 2015;48(3):350–356
healthy women who have had a total hysterectomy and bilateral salpingo-oophorectomy for benign indications. Reasons given by physicians for performing the bimanual pelvic examination include that it is a standard part of a well woman examination, to screen for ovarian cancer, to screen for other gynecologic cancers, to screen for sexually transmitted infections (STIs),2 and as a requirement for hormonal contraception.3 A study4 surveyed physicians and found that 47% of obstetrician-gynecologists believed that the pelvic examination is an effective screening test for ovarian cancer. However, the bimanual examination is not needed to screen for STIs1 and is not recommended as a screening test for gynecologic cancer by the American Cancer Society5 or the U.S. Preventive Services Task Force.6,7 It is also no longer recommended as a requirement before receiving oral contraceptives.1,8,9 Although many women have had an annual pelvic examination in association with screening for cervical cancer, the interval for this screening has increased to 3–5 years for average-
& 2015 American Journal of Preventive Medicine
Published by Elsevier Inc.
Ebell et al / Am J Prev Med 2015;48(3):350–356
risk women, calling into question the need to perform an annual bimanual pelvic examination that is not accompanied by a Pap test.7 In fact, the knowledge that an annual health maintenance visit is likely to include a bimanual pelvic examination, which many women view as invasive and uncomfortable, can cause anxiety10 and may actually discourage attendance.11 As a consequence, some women may not receive other important clinical preventive services. Perhaps even more important is the opportunity cost incurred by performing a pelvic during a periodic health examination, time that might be better spent delivering clinical preventive services that have been proven effective.12 The objective of the current study is to systematically review the literature regarding the accuracy of the bimanual examination as a screening test for ovarian cancer. A secondary goal is to examine the ability of the pelvic examination to distinguish benign from malignant lesions in women with a known abnormality, the other reason for using the bimanual examination related to ovarian cancer.
Figure 1. Prisma flow diagram. March 2015
351
Methods Inclusion Criteria Two types of studies were identified, screening studies and studies of symptomatic patients. In screening studies, the goal of the bimanual examination was to detect ovarian cancer in an asymptomatic woman. Screening studies that performed a bimanual pelvic examination in asymptomatic women and reported enough information to calculate the sensitivity, specificity, or both for the diagnosis of ovarian malignancy were included. Acceptable reference standards included ultrasound, cancer antigen (CA)-125 testing followed by biopsy if suspicious, or both; laparotomy or laparoscopy; or repeated screening or long-term follow-up in r12 months. A second group of studies of symptomatic women was identified, where the intention of the bimanual pelvic examination was to distinguish malignant from benign abnormalities. Prospective cohort studies of women with a known pelvic abnormality who underwent bimanual examination prior to surgery and that reported enough information to calculate the sensitivity and specificity for malignant versus benign lesion were included. Studies where bimanual examinations were performed under anesthesia were excluded, as this is not representative of usual
March 2001– June 2003
Greece Palpable adnexal mass
Palpable mass
Mean 58.1 years, range 45–80 years
Mean 58.4 years, range 50–85 years
Asymptomatic women aged Z45 years without any evidence of adnexal pathology
Asymptomatic peri- and postmenopausal women aged Z50 years
2,000
3,201
Adonakis (1996)20
Kurjak (2005)21
([bimanual OR pelvic exam OR pelvic examination] AND [ovarian cancer OR ovarian malignancy]) OR ([physical examination OR physical exam OR clinical exam OR clinical examination OR pelvic examination OR bimanual examination OR pelvic exam OR bimanual exam] AND [sensitivity OR specificity OR accuracy OR predictive value OR likelihood ratio] AND [ovarian neoplasms [MeSH Terms]]) OR ([pelvic examination OR bimanual examination] AND [sensitivity OR specificity] AND [ovary OR ovarian OR adnexal]) OR ([pelvic examination OR bimanual examination] AND [sensitivity OR specificity OR likelihood ratio])
The search was performed in August 2013 and updated in November 2013. Two researchers reviewed each abstract separately to identify studies that appeared to report the accuracy of bimanual pelvic examination for the detection of ovarian malignancy. The full text of each study identified by at least one of the researchers was reviewed by both for final inclusion or exclusion in the study. In addition, the reference lists of review articles, including a 2006 evidence report from the Agency for Healthcare Research and Quality,13 were reviewed to identify additional studies that may have been missed by the PubMed search.
Data Abstraction Data regarding study design, study quality, and diagnostic accuracy were abstracted in tandem by at least two researchers. Inconsistencies were reconciled by consensus, and the final data abstraction was reviewed in detail by the lead investigator (MHE). Studies were classified as being in a screening population if the intent was to detect cancer using the bimanual pelvic examination, or as being in a symptomatic population of women with a known pelvic mass if the intent was to distinguish benign from malignant lesions. Screening studies were further classified as average risk or high risk. High risk was defined as having a BRCA1 or 2 mutation or having a strong family history for breast or ovarian cancer. Studies enrolling predominantly average-risk postmenopausal women but also enrolling some younger women with a family history were classified as average risk.
Quality Assessment
GI, gastrointestinal; NR, not reported.
2,623 Grover (1995)19
van Nagell (1995)18
Search Strategy
Croatia
Australia Bulky/fibroid uterus or adnexal abnormality Mean 51 years, range 25–92 years Asymptomatic volunteers
NR
U.S. Palpable mass Mean 56 years, range 25–92 years Asymptomatic women aged Z50 years and women Z25 years with a family history of ovarian cancer
1987–2005
United Kingdom Palpable adnexal mass of any size separate from uterus/GI tract Median 54 years, range 45–83 years Apparently healthy women aged 445 years and amenorrheic 412 months 1,010 Jacobs (1988)17
Screening average risk
25,327
Country Definition of abnormal exam
NR
Date of data collection
outpatient clinical practice. For both clinical questions, case–control studies, surveillance studies in patients with known cancer, and studies in children were excluded.
The following search strategy for PubMed, limited to “Humans” and “has abstract” was used:
Age Patient population n Author (year)
Table 1. Characteristics of Included Studies
March 1991– June 1993
Ebell et al / Am J Prev Med 2015;48(3):350–356
352
Only cohort studies were included, whereas case–control studies were excluded, as the latter previously have been shown to be at high risk of bias.14 The following study quality characteristics were evaluated: the reference standard, whether data collection was prospective or retrospective, presence of verification bias, and blinding of those performing the index test and of those performing the reference standard test. Each potential threat to validity was rated as yes if present, no if absent, and unclear if it could not be determined after careful review of the source material.
Analysis The I2 statistic was used to assess the heterogeneity of included studies for each question. The I2 statistic has a value between 0 and 100, www.ajpmonline.org
I ¼0.00 I
2,a
Summary effect
Van der Velde (2009)
Olivier (2006)
The I2 statistic has a value between 0 and 100; 0 represents perfect homogeneity and 100 is perfect heterogeneity. Although I2 4 0.5, this is an artifact of the small number of studies but large size and precision of each study. Therefore, the summary measure of specificity is reported with its 95% CI. b
a
I2¼0.00 I ¼0.00
0.57 (0.45, 0.72)
I ¼73.7
2
24.7 (15.8, 37)
b 2 2
0.44 (0.32, 0.57)
0.98 (0.97, 0.99)
0.67 (0.30, 1.5) 27.6 (4.2, 180) 0.99 (0.97, 1.0) 0.33 (0.01, 0.91) 1/3
4/331
19.3 (7.8, 48) 0.40 (0.12, 0.74)
24
4/10
15/722
0.98 (0.97, 0.99)
0.61 (0.37, 1.0)
0.76 (0.34, 1.0) 29.8 (2.5, 352) 0.99 (0.99, 1.0) 0.00 (0.00, 0.98) 0/1
23
Oei (2006)22
Kurjak (2005)
Screening (high-risk population)
3/5
2/3
11/1,371
0.60 (0.15, 0.95)
0.67 (0.09, 0.99)
23.4 (10.1, 54.1) 21
Adonakis (1996)
Van Nagell (1995)
20
19/44
57/1,997
0.43 (0.28, 0.59)
0.97 (0.96, 0.98)
0.34 (0.07, 1.7)
0.76 (0.34, 1.7) 16.2 (1.4, 182) 0.00 (0.00, 0.98)
18
Grover (1995)
19
Jacobs (1988)17
Screening (average-risk population)
0/1
40/2,622
0.98 (0.98, 0.99)
0.26 (0.02, 1.0) 27.6 (11.4, 66) 0.97 (0.96, 0.98) 1.00 (0.03, 1.0)
March 2015
1/1
27/999
Likelihood ratio positive (95% CI) Specificity (95% CI) Sensitivity (95% CI) Abnormal pelvic exam/ total without cancer Abnormal pelvic exam/ total with cancer Author (year)
Table 2. Accuracy of Pelvic Exam for Diagnosis of Ovarian Cancer in Screening Studies
Likelihood ratio negative (95% CI)
Ebell et al / Am J Prev Med 2015;48(3):350–356
353
where 0 represents perfect homogeneity and 100 is perfect heterogeneity. As this statistic is unreliable when there is a small number of studies,15 the authors also used visual inspection of forest plots to assess heterogeneity. If there was adequate homogeneity (I2o0.5), bivariate metaanalysis was performed using the meta-analytical integration of diagnostic accuracy (MIDAS) procedure16 in Stata, version 13.1, to calculate summary measures of sensitivity, specificity, and likelihood ratios. MIDAS uses a bivariate mixed-effects regression model that has been modified for meta-analysis of diagnostic accuracy studies. It jointly analyzes the pairs of sensitivity and specificity, taking into account correlation between these parameters. Bivariate summary receiver operating characteristic curves were drawn, again using the MIDAS procedure in Stata, version 13.1, and includes 95% CIs and predictive intervals. The MIDAS procedure graphs the mean logit sensitivity and specificity with their SEs and 95% CIs. Where there were zero events, a continuity correction of 0.5 was used by the MIDAS procedure.
Results The PubMed search returned a total of 303 articles, and previous systematic reviews and reference lists of included studies identified another 24 studies, for a total of 315 unique studies after elimination of duplicates. A total of 56 studies met the initial review and were examined in full. Forty-one studies were excluded because they did not meet the inclusion criteria, most often because of failure to report sensitivity or specificity, failure to address one of our study questions, or because they were studying surveillance in patients with known malignancy. Ultimately, 15 studies met the inclusion criteria and were included in the analysis (Figure 1, Table 1, and Appendix Table 1). Included studies were published between 1988 and 2009.
Characteristics of Included Studies Eight screening studies with 36,599 asymptomatic women17–24 and seven studies of 782 symptomatic women25–31 met the inclusion and exclusion criteria. Study characteristics are summarized in Table 1 and Appendix Table 1. Three of the screening studies enrolled women at high risk for ovarian cancer based on their family history or the presence of a BRCA mutation,22–24 whereas the other five studies predominantly enrolled postmenopausal asymptomatic women at average risk for ovarian cancer.17–21 All of the studies with symptomatic women enrolled those scheduled for exploratory laparotomy for adnexal
354
Ebell et al / Am J Prev Med 2015;48(3):350–356
The summary estimates for sensitivity and specificity were 0.44 (95% CI¼0.32, 0.57) and 0.98 (95% CI¼0.97, 0.99), respectively. The respective summary estimates for the positive and negative likelihood ratios were 24.7 (95% CI¼15.8, 37) and 0.57 (95% CI¼0.45, 0.72). Thus, an abnormal pelvic examination greatly increased the likelihood of ovarian cancer, whereas a normal examination was of little value in ruling out cancer. Less than half of ovarian cancers were detected using the pelvic examination. The bivariate summary receiver operating characteristic curve is shown in Figure 2. The c-statistic of 0.81 (95% CI¼0.77, 0.84) showed fair to good discrimination, but the predictive interval was broad, encompassing sensitivities from approximately 25% to 62%.
Accuracy of Bimanual Examination for Distinguishing Malignant from Benign Disease Figure 2. Summary bivariate receiver operating characteristic (ROC) curve for pelvic exam as a screening test in asymptomatic women. Note: The 95% confidence contour defines the 95% CI for the summary estimates of sensitivity and specificity, whereas the 95% prediction contour indicates the range where a future study is likely to lie.
mass of unknown origin. Three were set in the U.S., and they largely enrolled women aged 445 years.28,30,31
Quality Assessment The quality of included studies is summarized in Appendix Table 2. All had a cohort design and all but one used prospective data collection. All of the screening studies suffered from verification bias, as women with normal pelvic examination, ultrasound, and with or without CA-125 (the protocols varied) had repeat screening 1 year later, whereas those with an abnormal screening test had more intensive follow-up that often included laparotomy or laparoscopy. Blinding was inconsistently reported, although is unlikely to be of consequence for those performing surgery or evaluating the histopathology.
Accuracy of Bimanual Pelvic Examination as a Screening Test The accuracy of the bimanual examination as a screening test for ovarian cancer is summarized in Table 2. There was excellent homogeneity between studies, with the exception of specificity. However, this was likely an artifact due to the small number of studies and narrow CI for each study.15 The specificities ranged from 0.97 to 0.99, and each CI ranged from 0.96 to 1.0. Therefore, summary estimates are reported for this specificity.
Data from studies of symptomatic women undergoing laparotomy for a known pelvic mass are summarized in Table 3. There was significant heterogeneity for the sensitivity, specificity, and negative likelihood ratio, and moderate heterogeneity for the positive likelihood ratio. Therefore, only a summary measure of accuracy for the positive likelihood ratio was calculated (3.4, 95% CI¼2.8, 4.3). The sensitivity (0.43–0.93), specificity (0.53–0.91), and negative likelihood ratio (0.12–0.62) had broad ranges. The bivariate summary receiver operating characteristic curve is shown in Figure 3, with a c-statistic of 0.83. However, there was significant heterogeneity and the CIs and predictive intervals were broad.
Discussion The bimanual pelvic examination has very low sensitivity for the diagnosis of ovarian cancer. Although the specificity of 0.98 appears high, it is important to remember that the examination is being used to detect a rare disease. The Prostate Lung Colon Ovary (PLCO) trial detected approximately 5 cases of ovarian cancer per 10,000 person-years of follow-up.32 Thus, if 10,000 women are screened, the 44% sensitivity means that approximately two of the five cancers would be detected. However, the false-positive rate of 2% would result in 200 false-positive abnormal pelvic examinations, a positive predictive value of only 1.0%. Using the 95% CIs for specificity of 97%–99% and sensitivity of 32%–57%, the range for positive predictive value is 0.67%–3.0%. This study did not assess the impact of screening using the pelvic examination on mortality or other clinical outcomes; the focus was on accuracy. However, the PLCO trial found that even annual ultrasound did not reduce ovarian cancer specific or all-cause mortality.32 Thus, it seems unlikely that www.ajpmonline.org
Where I2, a measure of heterogeneity, exceeded 0.5, we report only the range. Where it is less than 0.5, we report the summary measure of effect and the 95% CI. The I2 statistic has a value between 0 and 100; 0 represents perfect homogeneity and 100 is perfect heterogeneity. b
a
I2,b
Summary effect
a
Ghaemmaghami (2008)
Balbi (2001)
25
Buckshee (1998)
Roman (1997)
30
Buist (1994)
27
355
I2¼85.0 I2¼40.8 I2¼87.9
I2¼83.8
0.12, 0.62 3.4 (2.8, 4.3) 0.43, 0.93
0.53, 0.91
0.35 (0.24, 0.49) 4.0 (2.3, 7.0) 63/88
11/62
0.72 (0.61, 0.81)
0.82 (0.70, 0.91)
0.12 (0.03, 0.47) 3.5 (2.2, 5.7)
March 2015
26
29
20/22
13/50
0.91 (0.71, 0.99)
0.74 (0.60, 0.85)
0.25 (0.7, 0.85) 7.0 (2.3, 21.5) 7/9
3/27
0.78 (0.40, 0.97)
0.89 (0.71, 0.98)
0.58 (0.43, 0.80) 3.1 (2.0, 4.8) 22/43
30/183
0.51 (0.35, 0.67)
0.84 (0.77, 0.89)
0.59 (0.32, 2.1) 1.5 (0.87, 2.4) 31/45
9/19
0.69 (0.53, 0.82)
0.53 (0.29, 0.76
0.12 (0.06, 0.24) 2.5 (2.0, 3.2) 88/95 Schutter (1994)31
47/127
0.93 (0.85, 0.97)
0.63 (0.54, 0.71)
0.62 (0.46, 0.83) 5.0 (2.1, 11.6) 0.91 (0.82, 0.97) 16/37 Finkler (1988)28
6/70
0.43 (0.27, 0.61)
Likelihood ratio positive (95% CI) Specificity (95% CI) Sensitivity (95% CI) Abnormal pelvic exam/ benign disease Abnormal pelvic exam/ malignant disease Author (year)
Table 3. Accuracy of Pelvic Exam for Distinguishing Benign from Malignant Lesions in Patients with Pelvic Mass
Likelihood ratio negative (95% CI)
Ebell et al / Am J Prev Med 2015;48(3):350–356
Figure 3. Summary bivariate receiver operating characteristic (ROC) curve for pelvic exam as a test to distinguish benign from malignant lesions in women with pelvic mass. Note: The 95% confidence contour defines the 95% CI for the summary estimates of sensitivity and specificity, whereas the 95% prediction contour indicates the range where a future study is likely to lie.
the pelvic examination, a much less sensitive test, would be able to have any impact on mortality. The pelvic examination is also an unreliable test for distinguishing between benign and malignant lesions. This is not surprising, and it is why researchers have developed predictive indices such as the Risk of Malignancy Index that incorporate ultrasound, CA-125, and menopausal status.33,34 The most important limitation of this study is related to the quality of the included studies. Verification bias was an important concern in all of the screening studies, as patients with a negative screening test did not undergo further investigation until the next screening round. However, it is unlikely that future studies would overcome this bias by performing a surgical biopsy in all women with normal screening tests, or even a random sample of such women. Masking was also rarely reported, although we assumed it when the examination took place prior to surgery. Studies of symptomatic patients were generally small with uncertain inclusion criteria. Our finding that the bimanual examination is not an effective screening test for ovarian cancer supports the increasingly common practice of foregoing the traditional annual bimanual pelvic examination in favor of other clinical preventive services for which there is better evidence.12,35 It is also consistent with guidelines5–7 that do not recommend the bimanual examination as a screening test for ovarian cancer. It will be important
356
Ebell et al / Am J Prev Med 2015;48(3):350–356
to clearly communicate that not doing a bimanual pelvic examination every year is not a way to save money or limit access to healthcare services. Rather, it is an attempt to refocus the encounter with the physician on the full spectrum of clinical preventive services. Doing this has the potential to improve women’s health, remove a barrier to access, and reduce the cost of care. This study was not funded or sponsored. It was conceived, carried out, and written by the authors under the lead of Dr. Ebell. The authors do not have any relevant conflicts of interest, either financial or professional. Dr. Ebell is a member of the U.S. Preventive Services Task Force (USPSTF). This article does not necessarily represent the views and policies of the USPSTF. No financial disclosures were reported by the authors of this paper.
References 1. Committee on Gynecologic Practice. Committee opinion no. 534: wellwoman visit. Obstet Gynecol 2012;120(2 Pt 1):421–4. 2. Schachter J, Schafer M, Young M, Ott M. Routine pelvic examinations in asymptomatic young women. N Engl J Med 1996;334:1847–8. 3. Stormo AR, Hawkins NA, Cooper CP, Saraiya M. The pelvic examination as a screening tool: practices of US physicians. Arch Intern Med 2011;171(22):2053–4. 4. Henderson JT, Harper CC, Gutin S, et al. Routine bimanual pelvic examinations: practices and beliefs of US obstetrician-gynecologists. Am J Obstet Gynecol 2013;208(2):109.e1–7. 5. American Cancer Society guidelines for the early detection of cancer, 2013. www.cancer.org/healthy/findcancerearly/cancerscreeningguidelines/ american-cancer-society-guidelines-for-the-early-detection-of-cancer. 6. U.S. Preventive Services Task Force. Screening for ovarian cancer: recommendation statement. Ann Fam Med 2004;2(3):260–2. dx.doi.org/ 10.1370/afm.200. 7. Moyer VA. U.S. Preventive Services Task Force. Screening for ovarian cancer: USPSTF reaffirmation recommendation statement. Ann Intern Med 2012;157(12):900–4. 8. Stewart FH, Harper CC, Ellertson CE, et al. Clinical breast and pelvic examination requirements for hormonal contraception: current practice vs evidence. JAMA 2001;285(17):2232–9. 9. Henderson JT, Sawaya GF, Blum M, et al. Pelvic examinations and access to oral hormonal contraception. Obstet Gynecol 2010;116(6): 1257–64. 10. Reddy DM, Wasserman SA. Patient anxiety during gynecologic examinations. Behavioral indicators. J Reprod Med 1997;42(10):631–6. 11. Lacy NL, Paulman A, Lovejoy B. Why we don’t come: patient perceptions on no-shows. Ann Fam Med 2004;2(6):541–5. 12. Nguyen GT, Cronholm PF. The annual pelvic examination: preventive time not well spent. Am Fam Physician 2013;87(1):8–9. 13. Myers ER, Bastian LA, Havrilesky LJ, et al. Management of adnexal mass. Evidence report/technology assessment No. 130 (prepared by the Duke Evidence-based Practice Center under Contract No. 290-020025.) AHRQ Publication No. 06-E004. Rockville MD: Agency for Healthcare Research and Quality, February 2006. 14. Lijmer JG, Mol BW, Heisterkamp S, et al. Empirical evidence of designrelated bias in studies of diagnostic tests. JAMA 1999;282(11):1061–6. 15. Thorlund K, Imberger G, Johnston BC, et al. Evolution of heterogeneity (I2) estimates and their 95% confidence intervals in large metaanalyses. PLoS One 2012;7(7):e39471.
16. Dwamena BA. MIDAS: a program for meta-analytical integration of diagnostic accuracy studies in Stata. Ann Arbor, MI: Division of Nuclear Medicine, Department of Radiology, University of Michigan Medical School; 2007. 17. Jacobs I, Bridges J, Reynolds C, et al. Multimodal approach to screening for ovarian cancer. Lancet 1988;331(8580):268–71. 18. van Nagell JR Jr, DePriest PD, Ueland FR, et al. Ovarian cancer screening with annual transvaginal sonography. Cancer 2007;109(9):1887–96. 19. Grover SR, Quinn MA. Is there any value in bimanual pelvic examination as a screening test? Med J Aust 1995;162(8):408–10. 20. Adonakis GL, Paraskevaidis E, Tsiga S, et al. A combined approach for the early detection of ovarian cancer in asymptomatic women. Eur J Obstet Gynecol Reprod Biol 1996;65(2):221–5. 21. Kurjak A, Prka M, Arenas JM, et al. Three-dimensional ultrasonography and power Doppler in ovarian cancer screening of asymptomatic peri- and postmenopausal women. Croat Med J 2005;46(5):757–64. 22. Oei AL, Massuger LF, Bulten J, et al. Surveillance of women at high risk for hereditary ovarian cancer is inefficient. Br J Cancer 2006;94(6):814–9. 23. Olivier RI, Lubsen-Brandsma MA, Verhoef S, van Buerden M. CA125 and transvaginal ultrasound monitoring in high-risk women cannot prevent the diagnosis of advanced ovarian cancer. Gynecol Oncol 2006;100(1):20–6. 24. van der Velde NM, Mourits MJ, Arts HJ, et al. Time to stop ovarian cancer screening in BRCA1/2 mutation carriers? Int J Cancer 2009;124 (4):919–23. 25. Balbi GC, Musone R, Menditto A, et al. Women with a pelvic mass: indicators of malignancy. Eur J Gynaecol Oncol 2001;22(6):459–62. 26. Buckshee K, Temsu I, Bhatla N, Deka D. Pelvic examination, transvaginal ultrasound and transvaginal color Doppler sonography as predictors of ovarian cancer. Int J Gynaecol Obstet 1998;61(1):51–7. 27. Buist MR, Golding RP, Burger CW, et al. Comparative evaluation of diagnostic methods in ovarian carcinoma with emphasis on CT and MRI. Gynecol Oncol 1994;52(2):191–8. 28. Finkler NJ, Benacerraf B, Lavin PT, et al. Comparison of serum CA 125, clinical impression, and ultrasound in the preoperative evaluation of ovarian masses. Obstet Gynecol 1988;72(4):659–64. 29. Ghaemmaghami F, Fakour F, Zarchi MK, et al. Clinical assessment, gross examination, frozen section of ovarian masses: do patients benefit? Arch Gynaecol Obstet 2008;278(3):209–13. 30. Roman LD, Muderspach LI, Stein SM, et al. Pelvic examination, tumor marker level, and gray-scale and Doppler sonography in the prediction of pelvic cancer. Obstet Gynecol 1997;89(4):493–500. 31. Schutter EM, Kenemans P, Sohn C, et al. Diagnostic value of pelvic examination, ultrasound, and serum CA125 in postmenopausal women with a pelvic mass. Cancer 1994;74(4):1398–406. 32. Buys SS, Partridge E, Black A, et al. for the PLCO Project Team. Effect of screening on ovarian cancer mortality. The Prostate, Lung, Colorectal and Ovarian (PLCO) cancer screening randomized controlled trial. JAMA 2011;305(22):2295–303. 33. Jacobs I, Oram D, Fairbanks J, et al. A risk of malignancy index incorporating CA 125, ultrasound and menopausal status for the accurate preoperative diagnosis of ovarian cancer. Br J Obstet Gynaecol 1990;97(10):922–9. 34. Yamamoto Y, Yamada R, Oguri H, et al. Comparison of four malignancy risk indices in the preoperative evaluation of patients with pelvic masses. Eur J Obstet Gynecol Reprod Biol 2009;144(2):163–7. 35. Stewart RA. Routine pelvic examination for asymptomatic women: exploring the evidence. Aust Fam Physician 2006;35(11):873–6.
Appendix Supplementary data Supplementary data associated with this article can be found at http://dx.doi.org/10.1016/j.amepre.2014.10.007. www.ajpmonline.org
