M. DUEHOLM*, A. FORMAN*, M. L. JENSEN, H. LAURSEN and P. KRACHT*

Ultrasound Obstet Gynecol 2001; 18: 54 61 Transvaginal sonography combined with saline contrast Blackwell Original Paper Science, Ltd sonohysterography in evaluating the uterine cavity in premenopausal patients with abnormal uterine bleeding M. DUEHOLM*, A. FORMAN*, M. L. JENSEN, H. LAURSEN and P. KRACHT* *Department of Obstetrics and Gynaecology and Department of Pathology, Aarhus University Hospital, Aarhus and Department of Obstetrics and Gynaecology, Randers Central Hospital, Randers, Denmark KEYWORDS: Abnormal uterine bleeding, Saline contrast sonohysterography, Transvaginal ultrasonography ABSTRACT Objectives To evaluate whether saline contrast sonohysterography (SCSH) adds additional information to that obtained by transvaginal sonography (TVS) for predicting endometrial abnormality in premenopausal patients with abnormal uterine bleeding. Patients and methods This was a two-center prospective study at a university clinic and a central hospital in Denmark. The uterine cavity was evaluated with TVS and SCSH in 470 premenopausal patients with abnormal uterine bleeding. One hundred and eighty-nine of the patients had operative hysteroscopy or hysterectomy within 4 months which provided a detailed description of the uterine cavity and was used as the true value for exclusion of polyps and submucous myomas. Results Based on normal endometrial morphology alone, the results for detection of an abnormal uterine cavity were as follows: sensitivities of TVS 0.92, SCSH 0.99; specificities of TVS 0.62, SCSH 0.72; positive predictive values of TVS 0.80, SCSH 0.85; negative predictive values of TVS 0.82, SCSH 0.98. Transvaginal sonography combined with SCSH was superior to TVS for detection of intracavitary abnormalities (McNemar test, P = 0.008). The post-test probability of there being an abnormal cavity after normal findings on TVS alone was 0.18 (0.10 0.32) and after TVS and SCSH it was 0.02 (0.01 0.11). When normal endometrial morphology was combined with an endometrial thickness of < 12 mm for evaluation of all abnormalities including hyperplasia, the diagnostic potential of TVS or SCSH was almost unchanged except for specificities, which were markedly lower (TVS 0.54; SCSH 0.57). In all the patients referred, TVS had a negative predictive value of 0.94 for identification of polyps and myomas when findings at subsequent SCSH were accepted as the true value. Transvaginal sonography reduced the pretest probability of polyps or submucous myomas from 0.35 to a post-test probability of 0.06, but missed 21% of the polyps. Conclusions Sonohysterography was a sensitive tool and was superior to TVS used alone for evaluation of the uterine cavity in patients who underwent operative surgery for abnormal uterine bleeding. All abnormalities except one were found at SCSH, while TVS alone missed polyps and had almost one in four equivocal findings. The use of TVS, without saline contrast, left one in five of the polyps undiagnosed in referred patients with abnormal bleeding. INTRODUCTION Hysteroscopy with biopsy has become the gold standard for evaluation of the uterine cavity, as a reliable and safe method in routine outpatient settings 1. Transvaginal sonography (TVS) is an alternative non-invasive method that is commonly used for examination of the endometrium and uterine cavity in postmenopausal patients 2,3. In pre- and perimenopausal patients with abnormal bleeding, the diagnostic potential of TVS in experienced hands has been found to be in line with that of hysteroscopy 4, but not when less experienced operators performed the examination 5. When TVS was combined with saline contrast in the uterine cavity (saline contrast sonohysterography, SCSH), the diagnostic accuracy was markedly improved and was found to equal that of hysteroscopy when performed by skilled investigators 6 11. Nevertheless, most of the studies 8 13 on SCSH have involved a limited number of both premenopausal and postmenopausal patients and in only one study were the premenopausal patients evaluated separately from the postmenopausal patients 10. In the present study the diagnostic potential of TVS, and that of TVS combined with saline contrast, were compared with the findings at operative hysteroscopy or hysterectomy in a group of premenopausal patients. The study was performed by investigators who had routine experience in performing transvaginal ultrasound. The aim was to evaluate whether Correspondence: Dr M. Dueholm, Department of Obstetrics and Gynecology, Aarhus University Hospital, DK-8000 Aarhus, Denmark (e-mail: hjorth@dadlnet.dk) Received 17-1-00, Revised 1-12-00, Accepted 22-1-01 54 ORIGINAL PAPER

SCSH improved the diagnostic accuracy of TVS for identification of the uterine cavity. Firstly, we evaluated whether the presence of a straight regular endometrial lining at TVS or SCSH excluded polyps and myomas. Secondly, a straight regular endometrial lining was combined with an endometrial thickness of < 12 mm for exclusion of all abnormalities including hyperplasia. Moreover, findings of polyps or myomas at TVS were compared to findings at subsequent SCSH in an unselected group of premenopausal patients referred for abnormal uterine bleeding. METHODS The study was carried out in two different centers, between January 1st 1994 and October 1st 1995 in one center and between March 1st and October 1st 1995 in the other. A total of 470 women were referred to the bleeding disorder clinics during this period, and met the inclusion criteria. Four senior registrars and one consultant with routine experience in transvaginal ultrasound and who had performed 50 150 SCSHs were involved. Data from 95 patients from one of the centers have been presented in a previous paper 14, which describes the organization of a simple one stop menstrual problem clinic with the use of SCSH. Patients were recruited if they had abnormal uterine bleeding (menorrhagia, metrorrhagia, and menometrorrhagia), were premenopausal (defined as being within 1 year of arrest of bleeding) and were below the age of 55 years. Patients on hormone replacement therapy (HRT) and who had an indefinite menopausal status were included when the duration of HRT was less than 3 years. By these selections we tried to avoid cases with endometrial cancers, as SCSH has a theoretical but not proven risk of spreading an endometrial cancer by transport of endometrial cells through Fallopian tubes 15 17. Patients below 35 years of age had a negative chlamydia test. Those with an intrauterine contraceptive device agreed to have it removed prior to recruitment into the study. Moreover patients with serious cardiopulmonary disease which required cardiac surveillance, and those in whom pregnancy or infection-related bleeding disorders were suspected, were excluded. Abnormal uterine bleeding caused by cervical factors was diagnosed by colposcopy prior to the consultation. In each patient, the gynecological history was taken, the procedures explained and consent obtained. Bimanual palpation of the pelvis was performed with the patient in the dorsal lithotomy position. Transvaginal sonography (TVS) Transvaginal sonography was performed using a 5 7.5-MHz transvaginal transducer (Siemens Sonoline SI-400, Ballerup, Denmark). Measurement of the endometrium included both endometrial layers (double layer) as described by Granberg et al. 18. The contours of the endometrial cavity were studied from the internal os to the fundus in the longitudinal and transverse planes. The midline echo was considered to be normal when a straight regular endometrial lining, with well-defined margins and without echodense foci, was found. When the midline echo was disturbed, polyps were defined as echogenic masses with a fairly homogeneous texture without disruption of the myometrial endometrial interface, while submucous myomas had an inhomogeneous texture with possible continuity with the myometrium. Myomas disturbing the midline echo or exceeding a diameter of 15 mm in the myometrium were counted. Submucous myomas were classified according to the European Society of Gynaecologic Endoscopy classification: type 0 (pedunculated submucous myomas without intramural extension), type I (sessile and with an intramural part of less than 50%) and type II (with an intramural part of 50% or more). The investigators classified the quality of the examinations as sufficient or insufficient for evaluation of the uterine cavity. Before SCSH, the presence or absence of abnormalities was recorded, and all abnormalities were described at the standard record. Following TVS, SCSH was performed by the same investigator. Saline contrast sonohysterography (SCSH) A small flexible sterile catheter (baby-feeding tube Vigor 310.04, Belgium) mounted with a 50-mL syringe was introduced into the uterine cavity. During instillation of isotonic saline, the pressure was manually adjusted until sufficient for expansion of the uterine cavity. Concomitantly, the distension was observed by transvaginal sonography (equipment as above) and continued until the entire uterine cavity was clearly visible. The uterine cavity was evaluated in sagittal and coronal views and pictures were taken for documentation. Findings at SCSH were noted according to a standard form. A normal finding implied the presence of a straight regular endometrial lining without echodense foci and projections from the myometrium. Endometrial polyps and submucous myomas were defined following the criteria published by Widrich et al. 8 and Parsons and Lense 19. Again the investigators noted the quality of the examination. Finally the patients were asked whether they had experienced discomfort during SCSH (yes or no). Less than 10 min was required to complete each examination. Endometrial sampling Endometrial sample specimens were obtained with either the Vabra device (3 mm flexible suction curette and tissue trap, Berkeley Medevices, CA, USA) or dilatation and curettage (D&C). Samples were obtained according to local guidelines 20,21 in patients older than 45 years with menorrhagia, in patients older than 40 years with metrorrhagia and in patients older than 35 years with obesity, anovulation or previous oligomenorrhagia. Samples were always obtained prior to endometrial resection. With normal findings at SCSH, no endometrial sample was taken when a sufficient normal sample had been obtained within the last year. When operative hysteroscopic removal Ultrasound in Obstetrics and Gynecology 55

of solitary myomas or polyps was planned, the endometrial sample was taken at this stage. In 97 patients there was no indication for endometrial sampling or surgery, and they were treated medically. In 166 patients SCSH was followed by endometrial sampling only. Operative hysteroscopy/hysterectomy During operative hysteroscopy or hysterectomy the uterine cavity was described according to a standard form. The number of polyps and myomas was recorded and the mean diameter of the largest measured. Again myomas were classified according to the European Society of Gynaecologic Endoscopy classification. Operative hysteroscopy using a resectoscope was performed according to general guidelines 22. Three experienced hysteroscopists performed these procedures. The resected material was sent for pathological examination, and curettage was performed. At hysterectomy the presence and size of abnormalities and the percentage of myomas in the uterine cavity were described. The operative procedures were performed within 3 months of the sonographic examination. Hysterectomy was performed in 74 patients, while 79 underwent hysteroscopic resection of polyps, myomas or endometrium. Hysteroscopy using a resectoscope and curettage were completed in 36 cases, the others having equivocal findings or abnormalities not suitable for resection. Eight patients with an insufficient sonographic examination and 18 patients with failed SCSH were given a separate appointment for hysteroscopy. Five patients subsequently presented to the clinic with slight (not exceeding 38 C) temperature elevation. Oral antibiotic treatment was given, and recovery was uneventful. Analysis We analyzed findings at TVS and SCSH by two protocols. 1 Normal endometrial morphology alone, i.e. a straight regular endometrial lining, with well-defined margins and with no echodense foci, was seen irrespective of the endometrial thickness. (TVS or SCSH were combined with endometrial samples as the standard for diagnosis of hyperplasia or malignancies). 2 Normal endometrial morphology and thickness, i.e. a straight regular endometrial lining, with well-defined margins and with no echodense foci was seen and the endometrial thickness was < 12 mm (TVS or SCSH were used for exclusion of all abnormalities including hyperplasia and endometrial cancer). Sensitivity, specificity and positive and negative predictive values of TVS and SCSH were calculated with findings at hysteroscopy and hysterectomy being used as the standard for both algorithms. Equivocal sonographic findings were calculated as abnormal in the evaluation of diagnostic potential. The McNemar test was used for paired comparisons. Post-test probabilities were determined to evaluate the diagnostic potential of TVS and SCSH, when operative hysteroscopy or hysterectomy was indicated. Pretest probabilities of abnormalities were the proportion of patients with abnormality in the population of interest. Given a positive test result, the post-test probability of abnormalities was the positive predictive value. Given a negative test result, the post-test probability of abnormalities was 1 negative predictive value. Finally we calculated likelihood ratios. Likelihood ratios for a normal and abnormal test outcome show the diagnostic potential of the tests in a prevalence-independent way. The likelihood ratio of a positive test result is calculated as sensitivity/(1-specificity). The negative likelihood ratio is calculated as (1-sensitivity)/specificity. A positive likelihood ratio of 2 5 indicates a fair clinical test, 5 10 is good, and above 10 is excellent. A negative likelihood ratio of 0.5 0.2 is fair, 0.2 0.1 is good, and less than 0.1 is excellent 23,24. For all parameters, 95% confidence intervals were estimated. RESULTS The age distributions of patients in the different groups of follow up are shown in Figure 1. The mean age of all patients was 44.2 ± 5.7 (range, 22 55) years. Four hundred and ten (91%) patients did not complain of pain during the examination, while acceptable discomfort was reported in 42 (9%) patients. Four hundred and seventy patients met the inclusion criteria during the study period. Saline contrast sonohysterography could not be performed in 18 cases, leaving findings from 452 patients for analysis. In 28 other patients visualization of the uterine cavity was not optimal, but considered sufficient for evaluation. In eight cases SCSH was inconclusive. In 355 patients (189 who also underwent operative hysteroscopy or hysterectomy and 166 who underwent endometrial sampling only) microscopic specimens were available for pathologic diagnosis. The numbers of abnormalities diagnosed in the different groups of follow up are listed in Table 1. Number of patients 80 60 40 20 0 34 35 40 41 45 46 50 51 55 Age (years) Figure 1 Age distribution of patients with operative follow up ( ), saline contrast sonohysterography with endometrial samples ( ), and saline contrast sonohysterography without endometrial samples ( ). 56 Ultrasound in Obstetrics and Gynecology

Table 1 Numbers of patients with various abnormalities in the groups with operative follow up (operative hysteroscopy/hysterectomy), saline contrast sonohysterography with endometrial samples, and saline contrast sonohysterography without endometrial samples Submucous myomas Polyps Hyperplasia Endometrial cancer Other No abnormality Total Operative hysteroscopy/hysterectomy 79 41 8* 0 67 189 SCSH + endometrial sample 16 1 19 0 1 129 166 SCSH + no endometrial sample 8 0 1 88 97 Total number 103 42 27 0 284 452 *1 with atypia; 6 patients had two abnormalities at the same time; bicornuate uterus; SCSH, saline contrast sonohysterography. No patients had endometrial cancers, and only eight patients with operative surgery had hyperplasia. Samples in 196 patients with no abnormalities showed the endometrium to be secretory or proliferative in 145 cases, dysfunctional or menstruating in 30 cases, atrophic in 15 cases, inflammatory in three cases, and insufficient but normal in three cases. In 258 of 452 patients (57%) SCSH revealed myomas. In 11% of the patients ovarian cysts with a diameter of > 4 cm were found. No ovarian malignancies were found. Transvaginal sonography and saline contrast sonohysterography vs. operative follow up In 189 of 452 patients TVS was followed by operative hysteroscopy or hysterectomy (Table 2). The assessments of diagnostic potential were based on the findings at operative follow up as the gold standard. Equivocal sonographic findings were classified as abnormal in the evaluation of diagnostic potential. The discrepancies between findings at TVS and at operative follow up are listed in Table 3. In seven patients polyps were not identified at TVS, whereas eight false-positive findings occurred. The diagnostic sensitivity for detection of polyps at TVS was poor. The positive likelihood ratio of TVS (2.4; 95% CI, 1.7 3.5) was only fair, but the negative likelihood ratio of TVS (0.14; 95% CI, 0.06 0.31) was good. The pretest probability of an abnormal cavity was 0.62 (95% CI, 0.55 0.70) and that of a normal cavity was 0.38 (95% CI 0.31 0.45). Table 2 Diagnostic potential of transvaginal sonography for diagnosis of polyps and submucous myomas compared with hysteroscopy/ hysterectomy Findings at transvaginal sonography Findings at hysteroscopy/hysterectomy Uterine cavity Abnormal Normal Total Abnormalities 74 8 82 No abnormalities 10 44 54 Possible abnormality 34 19 53 Total 118 71 189 Sensitivity 0.92 (0.85 0.96) Specificity 0.62 (0.50 0.73) Positive predictive value 0.80 (0.72 0.86) Negative predictive value 0.82 (0.68 0.90) Abnormalities at hysteroscopy and hysterectomy did not include 8 cases of hyperplasia. CI, confidence interval. Findings at SCSH and at operative follow up were compared (Table 4). Findings at SCSH were classified as normal or abnormal including equivocal results. Saline contrast sonohysterography detected all abnormal cavities except one. In this patient a type 0 myoma of 10 mm in diameter was found on pathologic examination by hysterectomy. As was the case for TVS, the positive likelihood ratio of SCSH (3.5; 95% CI, 2.4 5.6) was only fair. Nevertheless, the negative likelihood ratio of SCSH (0.014; 95% CI, 0 0.08) was excellent. Table 3 Disagreement between transvaginal sonography and hysteroscopy/hysterectomy (n = 136) False results at TVS (n) Findings at sonography Findings at SCSH Hysteroscopy/ hysterectomy results Endometrial pathology results False negative 5 Normal Polyp Polyp Polyp 1 Normal Polyp Polyp Menstruation 1 Normal Polyp Submucous myoma Myoma 1 Normal Polyp Polyp Hyperplasia 1 Intramural myoma Intramural myoma Submucous myoma Proliferative 1 Normal Submucous myoma Submucous myoma Myoma False positive 5 Submucous myoma Submucous myoma (Type II) Intramural myoma Secretory, proliferative 2 Submucous myoma Intramural myoma Intramural myoma Menstruation, proliferative 1 Polyp Polyp Normal Secretory Seven polyps and 3 submucous myomas were not found at vaginal sonography. Equivocal sonograms (53) are not included. TVS, transvaginal sonography; SCSH, saline contrast sonohysterography. Ultrasound in Obstetrics and Gynecology 57

Patients with equivocal results of SCSH, or disagreement between SCSH and findings at operative hysteroscopy/ hysterectomy, and the corresponding pathologic findings are listed in Table 5. In 14 patients with abnormal SCSH, normal cavities were seen at operative follow up. These cases included six patients with slight impression of intramural myomas at SCSH. In a further six of these 14 patients, polyps were suspected at SCSH, while no abnormality was found at operative hysteroscopy or hysterectomy. In two of these six cases hyperplasia was present at pathologic examination. One normal cavity was interpreted as having a type 0 myoma in the presence of multiple myomas near the uterine cavity. Another normal cavity was perceived to have a type I myoma at SCSH. In 14 cases there were difficulties in discrimination between myomas and polyps, and uncertainty in the detection of two abnormalities in the same patient (Table 6). Table 4 Diagnostic potential of saline contrast sonohysterography for diagnosis of polyps and submucous myomas compared with hysteroscopy/hysterectomy Findings at saline contrast sonohysterography Findings at hysteroscopy/hysterectomy Uterine cavity Abnormal Normal Total Abnormalities 115 14 129 No abnormalities 1 51 52 Possible abnormality 2 6 8 Total 118 71 189 Sensitivity 0.99 (0.95 1.00) Specificity 0.72 (0.60 0.82) Positive predictive value 0.85 (0.78 0.91) Negative predictive value 0.98 (0.88 1.00) Abnormalities at hysteroscopy and hysterectomy did not include 8 cases of hyperplasia. CI, confidence interval. The sensitivity and negative predictive value for SCSH in the detection of the normal/abnormal uterine cavity were 0.99 and 0.98, respectively (Table 4). This implies that SCSH detected almost all abnormalities. The lower specificity (0.72) in this population of mostly abnormal cavities was mainly a consequence of small irregularities in the uterine cavity being assessed as polyps or myomas with slight impressions in the uterine cavity. The pretest probability of polyps or submucous myomas was 0.62 (95% CI, 0.55 0.70). The posttest probability of polyps or submucous myomas after normal findings at SCSH was 0.02 (95% CI, 0.01 0.11). In Table 7 findings at TVS and SCSH were again evaluated against findings at subsequent operative hysteroscopy or hysterectomy, but with hyperplasia (n = 8) included as an abnormality and an endometrial thickness of 12 mm counted as an abnormal finding at TVS and SCSH. The sensitivities of SCSH and TVS and the negative predictive values had almost the same values as seen in Tables 2 and 4, in which sonograms were considered normal when a straight regular endometrial lining was seen independent of the endometrial thickness. The specificities, however, were lower as only a small number of patients had an endometrial thickness of < 12 mm and no abnormalities after TVS and SCSH. Transvaginal sonography vs. saline contrast sonohysterography For detection of any abnormality in the uterine cavity SCSH showed a higher sensitivity than TVS. In 118 cases of abnormal cavity at operative follow up, nine cases were falsely abnormal at TVS and truly abnormal at SCSH, while no cases were false normal at SCSH and normal at TVS (McNemar test, P = 0.008). When hyperplasia was included as an abnormality in 122 cases of abnormal cavity the difference (7 at TVS vs. 0 at SCSH) was still significant (McNemar test, P = 0.02). In contrast no significant differences were found between TVS and SCSH in identification of the normal uterine cavity. Table 5 Disagreement between saline contrast sonohysterography and operative hysteroscopy/hysterectomy (n = 189) False/equivocal results at SCSH (n) Findings at SCSH Hysteroscopy/hysterectomy results Endometrial pathology results False negative 1 Normal cavity, multiple intramural myoma 1 type 0 myoma, multiple intramural myomas Proliferative False positive 4 Polyp Normal Secretory or proliferative 1 Polyp Normal Hyperplasia, simple 1 Polyp Normal Hyperplasia, complex, no atypia 6 Myoma type II Normal Secretory or proliferative 1 Myoma type I Normal Atrophy, normal 1 1 myoma type 0 14 mm, multiple myomas at the cavity line Normal Dysfunction, normal Equivocal results 3 Possible polyp Normal Secretory or proliferative 1 Possible polyp Normal Dysfunction, normal 1 Possible polyp Normal Hyperplasia, simple 2 Possible polyps Polyps 1 polyp, 1 secretory 1 Possible intramural myomas Normal Proliferative 58 Ultrasound in Obstetrics and Gynecology

Table 8 presents all referred patients. In 107 (24%) of 452 patients the midline echo could not be described as definitely abnormal or normal by TVS. Transvaginal sonography was compared to the findings of polyps or myomas at SCSH as the true value in 444 patients where SCSH was conclusive. Abnormalities of the uterine cavity caused by myomas or polyps were found at SCSH in 156 (35%) cases. Transvaginal sonography missed at least nine polyps and one submucous myoma; these findings were verified at operative hysteroscopy or hysterectomy. Thus, most findings missed at TVS were polyps and 21% of the polyps were missed. The negative predictive value of TVS Table 6 Saline contrast sonohysterography vs. operative hysteroscopy/ hysterectomy; patients with disagreement between findings of polyps vs. myomas Polyp vs. myoma at SCSH (n) Findings at SCSH Hysteroscopy/ hysterectomy results Pathology results 5 Polyp MyomaMyoma 1 Polyp MyomaSecretory, normal 2 Myomas Polyp Polyp 1 MyomaPolyp Myoma 1 Myoma2 polyps Dysfunction, normal 1 MyomaMyoma Adenomyoma 3 Myomas Myoma + polyp Myoma, polyp There was disagreement between findings of polyps vs. myomas at saline contrast sonohysterography and at hysteroscopy/hysterectomy in 10 patients. One finding of myoma both at SCSH and at operative hysteroscopy was found to be an adenomyoma at microscopic examination. In three patients, myomas were found at SCSH, but at hysteroscopy/hysterectomy both myomas and polyps were discovered. SCSH, saline contrast sonohysterography. Table 7 Diagnostic potential of transvaginal ultrasound and saline contrast sonohysterography compared with hysteroscopy/hysterectomy when hyperplasia was included as an abnormality and an endometrial thickness of 12 mm was considered an abnormal finding at TVS and SCSH Operative hysteroscopy or hysterectomy Abnormal cavity Normal cavity Transvaginal sonography (n = 189) Abnormalities at TVS 114 31 No abnormalities at TVS 8 36 Sensitivity 0.93 (0.87 0.97) Specificity 0.54 (0.41 0.66) Positive predictive value 0.79 (0.71 0.85) Negative predictive value 0.82 (0.67 0.91) Saline contrast sonohysterography (n = 189) Abnormalities at SCSH 121 29 No abnormalities at SCSH 1 38 Sensitivity 0.99 (0.95 1.00) Specificity 0.57 (0.44 0.69) Positive predictive value 0.81 (0.73 0.87) Negative predictive value 0.97 (0.85 0.99) TVS, transvaginal ultrasound; SCSH, saline contrast sonohysterography; CI, confidence interval. was at least 0.94 (95% CI, 0.90 0.96) and at most 0.96, and reduced the pretest probability of polyps or submucous myomas from 0.35 (95% CI, 0.30 0.39) to 0.06 (95% CI, 0.04 0.10) or at least 0.04 (95% CI, 0.02 0.08) after TVS. DISCUSSION In this prospective study TVS and SCSH were used to examine the uterine cavity in a population of patients with premenopausal bleeding disorders. Both methods were compared to operative hysteroscopy or hysterectomy as the gold standard. While TVS was the least invasive of the two procedures SCSH displayed a higher diagnostic potential. Thus, SCSH showed a higher accuracy than TVS for detection of the abnormal cavity. Moreover, the midline echo could not be described as definitely abnormal or normal in 24% of the patients during TVS. In these cases the uterine cavity had to be examined in another way. Finally, TVS displayed a negative predictive value of 0.82, and in patients scheduled for surgery, routine use of TVS without further investigations might leave a significant number of abnormalities undiagnosed. Saline contrast sonohysterography displayed a high sensitivity for diagnosis of abnormalities in the uterine cavity, in accordance with previous studies 7 10. In addition, a high negative predictive value was found. Thus routine use of this method even in non-expert hands implies a low number of undiagnosed lesions. A normal sonogram combined with an endometrial thickness of < 12 mm excluded even abnormalities such as hyperplasia, but the small number of patients with hyperplasia made the benefits and evidence limited for exclusion of this abnormality. The main disadvantage of SCSH was that small irregularities caused by blood clots or endometrial protrusions were frequently interpreted as polyps. In addition slight projections of myomas in the intramural portion of the uterus were seen at SCSH, but in some cases these projections might disappear after marked distension during operative hysteroscopy. These factors may have contributed to the relatively low specificity for SCSH. Moreover, detailed mapping of more than one lesion was difficult, and the capacity of SCSH for discrimination between polyps and myomas was poor. Table 8 Abnormal findings at transvaginal ultrasound in all referred patients (452), when findings of polyps or submucous myomas at saline contrast sonohysterography was regarded as the true value Findings at TVS Findings at SCSH Abnormal cavity Normal cavity Equivocal cavity Total Abnormal cavity 96 7 103 Normal cavity 15* 226 1 242 Equivocal cavity 45 55 7 107 Total 156 288 8 452 *12 patients with polyps and 3 patients with submucous myomas (3 polyps < 10 mm found at SCSH, not at operative follow up, 8 polyps found at SCSH and at operative follow up, 1 polyp not verified at operative follow up but at microscopy, 1 myoma verified at operative follow up and 2 submucous myomas without operative follow up). TVS, transvaginal ultrasound; SCSH, saline contrast sonohysterography. Ultrasound in Obstetrics and Gynecology 59

The same examiner performed TVS and SCSH in the same patients. This study design eliminated the difference within patients caused by differences in skill between observers. The use of examiners who were not blinded at SCSH and at operative hysteroscopy may have biased the results slightly, but operative hysteroscopy could not be performed blinded to the ultrasound diagnosis for ethical reasons, and the results appear more reliable as the true values compared to the possible alternative of diagnostic hysteroscopy 9. Transvaginal sonography was compared to SCSH in all patients referred for abnormal bleeding. In accordance with other studies 4 polyps or myomas were diagnosed in 35% of cases. At least 10 false-negative findings occurred at subsequent SCSH, when the midline echo was found to be normal at TVS. Nevertheless SCSH is an invasive method which had a 15% false-positive rate in the referred population; it may thus have resulted in unnecessary procedures in 15% of the performed hysteroscopies. Moreover, only a few submucous myomas, but several polyps were missed. Thus, the efficacy of TVS combined with endometrial samples as a diagnostic procedure in premenopausal patients with abnormal bleeding might be related to the clinical importance of finding polyps 25 27, which are rarely malignant in premenopausal patients 28. Efficient removal of polyps requires hysteroscopy and is not achieved by curettage alone 25 27. No controlled trial has been performed, but retrospective data 29,30 suggest hysteroscopic removal of polyps to be effective in the treatment of bleeding abnormalities. In infertile patients removal of small polyps did not seem to improve fertility 31, and detection of small polyps might be clinically unimportant. Nevertheless, it seems rational to remove polyps by hysteroscopy until control trials have been performed. Thus in relatively inexpert hands the higher diagnostic accuracy of SCSH at the expense of a few unnecessary hysteroscopies and slight patient discomfort may motivate the combined use of TVS with SCSH in a population of patients referred with abnormal bleeding. In conclusion, especially when operative treatment is planned, SCSH is superior to TVS. Normal findings at TVS reduce intervention rates but fail to eliminate particularly polyps, while submucous myomas seem to be identified. The use of TVS without SCSH would leave 21% of polyps undiagnosed in patients referred for abnormal uterine bleeding. Thus, SCSH was found to be a very sensitive tool for prediction of the abnormal uterine cavity, and routine use of this method as an alternative to diagnostic hysteroscopy in the primary investigation of patients with bleeding disorders would potentially lead to two in three hysteroscopies being avoided. ACKNOWLEDGMENT We would like to thank Krista Skejaa, Department of Obstetrics and Gynaecology, Aarhus University Hospital, for his help with the data analysis. REFERENCES 1 de Jong P, Doel F, Falconer A. Outpatient diagnostic hysteroscopy. Br J Obstet Gynaecol 1990; 97: 299 303 2 Smith BR, Kerlikowske K, Feldstein VA, Subak L, Scheidler J, Segal M, Brand R, Grady D. Endovaginal ultrasound to exclude endometrial cancer and other endometrial abnormalities. JAMA 1998; 280: 1510 7 3 Karlsson B, Granberg S, Wikland M, Ylostalo P, Torvid K, Marsal K, Valentin L. Transvaginal ultrasonography of the endometrium in women with postmenopausal bleeding a Nordic multicenter study. Am J Obstet Gynecol 1995; 172: 1488 94 4 Emanuel MH, Verdel MJ, Wamsteker K, Lammes FB. A prospective comparison of transvaginal ultrasonography and diagnostic hysteroscopy in the evaluation of patients with abnormal uterine bleeding: clinical implications [see comments]. Am J Obstet Gynecol 1995; 172: 547 52 5 Towbin NA, Gviazda IM, March CM. Office hysteroscopy versus transvaginal ultrasonography in the evaluation of patients with excessive uterine bleeding. Am J Obstet Gynecol 1996; 174: 1678 82 6 Fukuda M, Shimizu T, Fukuda K, Yomura W, Shimizu S. Transvaginal hysterosonography for differential diagnosis between submucous and intramural myoma. Gynecol Obstet Invest 1993; 35: 236 9 7 Cicinelli E, Romano F, Anastasio PS, Blasi N, Parisi C, Galantino P. Transabdominal sonohysterography, transvaginal sonography, and hysteroscopy in the evaluation of submucous myomas. Obstet Gynecol 1995; 85: 42 7 8 Widrich T, Bradley LD, Mitchinson AR, Collins RL. Comparison of saline infusion sonography with office hysteroscopy for the evaluation of the endometrium. Am J Obstet Gynecol 1996; 174: 1327 34 9 Schwarzler P, Concin H, Bosch H, Berlinger A, Wohlgenannt K, Collins WP, Bourne TH. An evaluation of sonohysterography and diagnostic hysteroscopy for the assessment of intrauterine pathology. Ultrasound Obstet Gynecol 1998; 11: 337 42 10 Bronz L, Suter T, Rusca T. The value of transvaginal sonography with and without saline instillation in the diagnosis of uterine pathology in pre- and postmenopausal women with abnormal bleeding or suspect sonographic findings. Ultrasound Obstet Gynecol 1997; 9: 53 8 11 Bernard JP, Lecuru F, Darles C, Robin F, de Bievre P, Taurelle R. Saline contrast sonohysterography as first-line investigation for women with uterine bleeding. Ultrasound Obstet Gynecol 1997; 10: 121 5 12 Dubinsky TJ, Parvey HR, Maklad N. The role of transvaginal sonography and endometrial biopsy in the evaluation of peri- and postmenopausal bleeding. AJR Am J Roentgenol 1997; 169: 145 9 13 Gaucherand P, Piacenza JM, Salle B, Rudigoz RC. Sonohysterography of the uterine cavity: preliminary investigations. J Clin Ultrasound 1995; 23: 339 48 14 Dueholm M, Laursen H, Knudsen UB. A simple one-stop menstrual problem clinic with use of hysterosonography for the diagnosis of abnormal uterine bleeding. Acta Obstet Gynecol Scand 1999; 78: 150 4 15 Benifla JL, Darai E, Filippini F, Walker-Combrouze F, Crequat J, Madelenat P. Operative hysteroscopy may transport endometrial cells into the peritoneal cavity: report of a prospective longitudinal study. Gynecol Endoscopy 1997; 6: 151 3 16 Romano S, Shimoni Y, Muralee D, Shalev E. Retrograde seeding of endometrial carcinoma during hysteroscopy. Gynecol Oncol 1992; 44: 116 8 (Abstract) 17 Alcazar JL, Errasti T, Zornoza A. Saline infusion sonohysterography in endometrial cancer: assessment of malignant cells dissemination risk. Acta Obstet Gynecol Scand 2000; 79: 321 2 18 Granberg S, Wikland M, Karlsson B, Norstrom A, Friberg LG. Endometrial thickness as measured by endovaginal ultrasonography for identifying endometrial abnormality. Am J Obstet Gynecol 1991; 164: 47 52 19 Parsons AK, Lense JJ. Sonohysterography for endometrial abnormalities: preliminary results. J Clin Ultrasound 1993; 21: 87 95 20 Ash SJ, Farrell SA, Flowerdew G. Endometrial biopsy in DUB. J Reprod Med 1996; 41: 892 6 21 Danish Health Research Council. Blødningsforstyrrelser hos kvinder. Koncensus rapport. Copenhagen: Danish Hospital Institute, 1993 22 Dueholm M, Forman A, Ingerslev J. Regression of residual tissue after incomplete resection of submucous myomas. Gynecol Endoscopy 1998; 7: 309 14 23 Møller-Petersen J. Evaluation of diagnostic tests. Design Phases Scan J Lab Invest 1992; 208: 35 50 60 Ultrasound in Obstetrics and Gynecology

24 Simel DL, Samsa GP, Matchar DB. Likelihood ratios for continuous test results making the clinicians job easier or harder? J Clin Epidemiol 1993; 46: 85 93 25 Dijkhuizen FP, Brolmann HA, Potters AE, Bongers MY, Heinz AP. The accuracy of transvaginal ultrasonography in the diagnosis of endometrial abnormalities. Obstet Gynecol 1996; 87: 345 9 26 Giusa-Chiferi MG, Goncalves WJ, Baracat EC, de Albuquerque Neto LC, Bortoletto CC, de Lima GR. Transvaginal ultrasound, uterine biopsy and hysteroscopy for postmenopausal bleeding. Int J Gynaecol Obstet 1996; 55: 39 44 27 Loffer FD. Hysteroscopy with selective endometrial sampling compared with D & C for abnormal uterine bleeding: the value of a negative hysteroscopic view. Obstet Gynecol 1916; 73: 16 20 28 Van BL. Clinicopathologic findings in endometrial polyps. Obstet Gynecol 1988; 71: 771 3 29 Nagele F, Mane S, Chandrasekaran P, Rubinger T, Magos A. How successful is hysteroscopic polypectomy? Gynecol Endoscopy 1996; 5: 137 40 30 Cravello L, Dércole C, Azoulay P, Boubli L, Blanc B. Hysteroscopic treatment of endometrial polyps. Gynecol Endoscopy 1995; 4: 201 5 31 Mastrominas M, Pistofidis GA, Dimitropoulos K. Fertility outcome after outpatient hysteroscopic removal of endometrial polyps and submucous fibroids. J Am Assoc Gynecol Laparosc 1996; 3: 29 Ultrasound in Obstetrics and Gynecology 61