Key observations
The present study showed that the “POP” scoring system (Past history of OBGY, no Other symptoms, and Peritoneal irradiation sign) had good screening ability for OBGY disease, with good discrimination and calibration with internal validation in the ED setting.
Previous literature and the present study’s strengths
Compared to previous studies, our study has some strengths for generalizability. A previous prospective multicenter study in five OBGY departments in Paris (N = 516) developed and validated a clinical prediction rule for identifying life-threatening diseases (e.g. ectopic pregnancy, adnexal torsion or tubo-ovarian abscess which can lead to hemodynamic instability, organ failures, severe morbidity, and death) in gynecological emergency rooms for patients with acute pelvic pain. Vomiting, sudden onset of fever, and pain from palpation are significantly associated with life-threatening disease.(1) However, the setting of this previous study was specialized for gynecological emergency, which was substantially different from the primary care or general emergency departments as the setting in the previous study did not include various types of diseases (i.e., digestive or urological diseases). Thus, the generalizability may be limited (spectrum bias). Conversely, our study setting was general ED in an urban area. Therefore, our study had the strength in terms of generalizability as compared to other ED settings.
Other previous prospective studies in the United States developed and validated a prediction model for ectopic pregnancy in the ED.(8) In this previous study, patients were limited to early pregnant women who visited the ED, and predictors included cervical motion tenderness and fetal heart rate. For non-gynecologist physicians, the opportunity to perform vaginal examinations or transvaginal ultrasonography is extremely limited in Japan. Thus, this model to predict ectopic pregnancy cannot be applied in general emergency departments. Accordingly, we believe that our prediction rule may be more suitable for diagnosing or excluding OBGY diseases in general ED.
Interpretation
We suggested the possible explanation of this prediction model. The present study evaluated clinically relevant variables that can be summarized as “POP” (past history of OBGY disease, other symptoms, and peritoneal irradiation sign). In terms of the past history of OBGY diseases, it is reported that ovarian tumor rupture and adnexal torsion are likely to occur and recur in patients with a history of ovarian tumor.(2, 10, 11) Accordingly, past history of OBGY diseases is important clinical information for prediction. Moreover, previous studies reported that vomiting is associated tubal rupture and adnexal torsion.(12, 13) However, the study population was only composed of patients who were diagnosed with OBGY diseases. Conversely, most patients in our study (489/740: 66%) were diagnosed with digestive diseases; half of them (250/489: 51%) complained of vomiting, while only 12% (8/65) with OBGY diseases had vomiting. Hence, it may be reasonable that no vomiting was more associated with OBGY diseases than other cases especially related to digestive disease in general ED. Moreover, in terms of fever, there was no association between fever and OBGY diseases.(12) On peritoneal irritation signs, most patients with ectopic pregnancy had abdominal peritoneal signs.(8) Thus, we assumed that ovarian bleeding and ectopic pregnancy cause bleeding in the pelvic cavity, and PID causes localized inflammation in the pelvis. Similarly, we found that most patients hospitalized for OBGY disease or those who underwent emergency surgery for OBGY disease also had peritoneal irritation sign. Atypical genital bleeding can be expected to be associated with OBGY diseases. However, in this study, there were only 2 cases out of 740 cases with an atypical genital bleeding. Therefore, the association between atypical genital bleeding and OBGY diseases was unknown in our study.
Hence, it is reasonable that we selected these contents to be the clinical predictors.
Clinical implications
The clinical implications of this study are that OBGY diseases can diagnosed or excluded by this simple scoring system. When score cut-off was set at 0/1 point, the negative likelihood ratio was 0.08 in our findings, which is useful to rule out OBGY diseases. If the prior probability is 8.8% was the same as in our setting, the posterior probability decreased to 1.3% when the score is 0. As an expected advantage of easy screening to exclude OBGY disease diagnosis, we presumed that there would be decrease in unnecessary consultation and number of transfers from the hospital without obstetricians and gynecologists, thereby improving the workload of these specialist physicians. When score cut-off was set at 2/3 points, the positive likelihood ratio was 17.25. The posterior probability increased to 55% in the abovementioned setting, when the score was 3. It may be useful for rule-in, leading to appropriate consultation. We suggest consultation with gynecologists if the POP score is 3 points. Meanwhile, if the POP score was 1 or 2 points, we considered evaluating the results from other additional tests (e.g., blood test, transabdominal ultrasonography, and computed tomography). Accordingly, the POP score may be useful for rule-out or rule-in of OBGY disease in an ED setting, similar to the results of our study.
Limitations
Our study has several limitations. Firstly, this is a retrospective study based on chart review, wherein the validity of the diagnosis, measurement factors, and the missed diagnosis might have led to the information bias. Secondly, direct visitation to OBGY department may have led to the selection bias. Thirdly, we could not assess the external validation as our study was conducted in a single center, with a relatively small sample size. Despite using the bootstrap procedure, our results indicated a low risk of bias by overfitting. Thus, further research is necessary to evaluate the external validation and applicability in other areas and multi-centers.