Several decades ago, women with early pregnancy complications were empirically treated with surgery, either to stop potential haemorrhage from suspected ectopic pregnancy, or to stop vaginal bleeding from ‘inevitable abortion’. This was neither reassuring for patients nor safe for mothers as we have all seen ‘inevitable abortions’ delivered at term. Thankfully, these dark ages are behind us. Today, we have sensitive urine pregnancy tests, high resolution, transvaginal ultrasound and logistic regression hCG models, whilst suspected ectopic pregnancy has been replaced by pregnancy of unknown location (PUL). The concern is that, despite all these advances, mothers and babies may still not be quite safe. In our desire to diagnose ectopic pregnancy as early as possible, we may cause harm by over-diagnosing ectopic pregnancy. False-positive diagnosis of ectopic pregnancy inevitably leads to unnecessary interventions and some normal intrauterine pregnancies being treated with methotrexate or curettage. Mothers may be subjected to ‘negative’ laparascopies, despite the fact that they suffered a miscarriage which would resolve spontaneously without any intervention. Constructing evidence-based practice from data synthesis requires careful scrutiny and critical analysis of existing literature. Precise tools applied to imprecise data do not lead to good progress. In order to make our practice safer, we need to accept that there should be zero tolerance for mistakenly diagnosing miscarriage, or ectopic pregnancy in cases where very early and viable intrauterine pregnancy exists. For any diagnostic algorithm that aims to detect normally developing intrauterine pregnancy, any false-negative results are simply unacceptable. To find out how the current hCG algorithms perform in this respect, we should turn to the systematic review by van Mello et al. Such reviews with meta-analysis are considered mandatory in the evaluation of effectiveness and safety of medical and surgical interventions, yet they are quite rare when it comes to diagnostic tests. The common excuse has been that the methodology for literature searching and data synthesis has not been robust enough for such reviews to be clinically useful, but this is no longer the case. Van Mello et al. provide a good example of methodologically robust meta-analysis which, hopefully, others in our speciality will follow. Having looked at 980 citations and 23 included study, the authors concluded that the definitions of intrauterine pregnancy were so inconsistent that meta-analysis was not possible. This may be disappointing, but is not unexpected. The same problem has been encountered in meta-analyses of interventions, time and again. The solution is simple—we need prospective studies with internationally agreed definitions and registered protocols. Cognisant of the present predicament, we need to recognize that a zero tolerance for false-negative diagnosis of normally progressing intrauterine pregnancy means that in clinically stable women with non-diagnostic ultrasound findings we should act simply to wait and watch—as long as it takes—to reach a conclusive diagnosis of intraor extrauterine pregnancy. And herein lays the dilemma. We are not really prepared to wait too long in the presence of PUL. When balancing fears of mistakenly treating intrauterine pregnancy against missing a ‘life-threatening ectopic’, the second one usually prevails. This inevitably leads to overdiagnosis of ectopic pregnancy and employment of ‘preventative’ management strategies. van Mello et al. offer an insight in the current definitions of ‘clinically important’ ectopic pregnancy that most clinicians believe is worth detecting and treating. Purists would argue that only histological confirmation of chorionic villi in a symptomatic patient should be labelled as a true ectopic pregnancy. However, in line with current thinking, van Mello et al. were compelled to accept in their meta-analysis as a ‘gold standard’ not only ectopic pregnancies confirmed histologically after laparoscopy or laparotomy, but also visual confirmation during laparoscopy without histology, hCG increase after D&C or systemic metorexate
Read full abstract