Richardson, Alison
(2017)
Minimising diagnostic uncertainties in early pregnancy.
PhD thesis, University of Nottingham.
Abstract
Introduction
Approximately one in five women experience abdominal pain and/or vaginal bleeding in early pregnancy. This usually prompts referral to an Early Pregnancy Assessment Unit where an ultrasound scan will be performed. Following the ultrasound, either a certain or uncertain diagnosis will be made. Certain diagnoses may be positive, i.e. a viable intrauterine pregnancy, or negative, i.e. a non-viable or ectopic pregnancy. Uncertain diagnoses occur when there is ambiguity regarding either the location or the viability of the pregnancy. Up to 25% of women attending an Early Pregnancy Assessment Unit are given such a diagnosis at their initial visit.
All women with a diagnosis of either a pregnancy of unknown location or uncertain viability need to be followed-up until a definitive diagnosis can be made. At present this is haphazard and protracted, commonly taking up to two weeks to resolve and requiring multiple visits for various different investigations. This takes a considerable amount of time and costs a not insignificant amount of money. Furthermore, in the time taken to make a definitive diagnosis, a stable woman with an unknown miscarriage or ectopic pregnancy may become unstable and require immediate resuscitation, life-saving blood transfusion and/or emergency surgery.
Aims
The aim of this PhD was to develop methods to minimise the number of women given uncertain diagnoses in early pregnancy, or to at least minimise the duration of uncertainty if the diagnosis is unavoidable. Several different studies were undertaken in an attempt to accomplish this.
Studies
We initially undertook a prospective cohort study to determine the levels of anxiety generated by uncertain diagnoses in early pregnancy was undertaken. Women with uncertain diagnoses were found to be significantly more anxious (as measured using the standardized short form of Spielberger’s state-trait anxiety inventory) than their counterparts given certain diagnoses (23±0.79 versus 14±6.6), even if these certain diagnoses were not associated with an ongoing pregnancy. This study served to further justify the main objective.
We then performed a systematic review and meta-analysis to identify and determine the diagnostic accuracy of various different ultrasonographic features to predict (a) an intrauterine pregnancy prior to visualization of embryonic contents and (b) a tubal ectopic pregnancy in the absence of an obvious extra-uterine embryo. This study identified the double decidual sac sign as a potential marker to be able to accurately differentiate a true gestation sac from a pseudosac with a sensitivity of 82% (95% CI, 68-90%), specificity of 97% (95% Ci, 76-100%), positive likelihood ratio of 30 (95% CI, 2.8-331) and negative likelihood ratio of 0.19 (95% CI, 0.10-0.35). The quality of the studies included in the meta-analysis however precluded the use of the double decidual sac sign in clinical practice without further validation
As a consequence, we carried out a prospective study following STARD guidelines to determine the diagnostic accuracy of the double decidual sac sign to predict an intrauterine pregnancy prior to visualization of embryonic contents using modern, high-resolution transvaginal ultrasound. This study found that the double decidual sac sign predicted an intrauterine pregnancy with a sensitivity of 94% (95% Ci, 85-98%), specificity of 100% (95% CI, 16-100%) and overall diagnostic accuracy of 94% (95% CI, 88-100%). The positive and negative predictive values are 100% (95% CI, 94-100%) and 33% (95% CI, 4.3-78%) respectively whilst the positive likelihood ratio was infinite and the negative likelihood ratio was 0.06 (95% CI, 0.02-0.16). These results suggest that the meta-analysis under-estimated the ability of the double decidual sac sign to differentiate between a true gestation sac and a pseudosac.
Subsequently, we conducted a study incorporating off-line analysis of ultrasonographic images to determine the inter- and intra-observer reliability of the double decidual sac sign to predict an intrauterine pregnancy prior to ultrasonographic visualization of embryonic contents. This involved fifteen observers from around the United Kingdom remotely assessing a selection of two-dimensional images from 25 of the diagnostic accuracy study participants. There was significant (p<0.01) agreement amongst the observers but the level of agreement was only ‘fair’, reflected by kappa statistics of 0.25, 0.33 and 0.21. Following a period of focused training, the inter-observer reliability significantly increased demonstrated by kappa statistics of 0.70, 0.63 and 0.53. The intra-observer reliability ranged from ‘substantial’ (K=0.65) to ‘almost perfect’ (K=0.92). These results demonstrate that the double decidual sac sign has the potential, after training, to be both reliable and precise, making it a very useful ultrasonographic sign in clinical practice.
Finally, we undertook a prognostic research study, following REMARK recommendations, investigating the ability of five serum biomarkers to predict pregnancy outcome in women with pregnancies of uncertain viability. Candidate biomarkers included Angiopoietin-1 (Ang-1), Angiopoietin-2 (Ang-2), soluble FMS-like Tyrosine Kinase-1 (Flt-1), serum TNF-Related Apoptosis Inducing Ligand and Interleukin-15. Serum concentrations of Ang-2 and Flt-1 were significantly lower in pregnancies of uncertain viability that were subsequently proven to be viable than those that were subsequently proven to be non-viable (Ang-2 1510pg/ml versus 2365pg/ml and Flt-1 103pg/ml versus 202pg/ml). Furthermore, there were statistically significant (p<0.01), linear (p-value for trend <0.01) associations between Ang-2 and Flt-1 concentrations and subsequent pregnancy viability such that women with a pregnancy of uncertain viability and Ang-2 concentrations greater than or equal to 2666pg/ml were 64% less likely to have a viable pregnancy than those with Ang-2 concentrations less than or equal to 1382pg/ml and women with a pregnancy of uncertain viability and Flt-1 concentrations greater than or equal to 142pg/ml were 50% less likely to have a viable pregnancy than those with Flt-1 concentrations less than or equal to 87pg/ml. These findings suggest that Ang-2 and Flt-1 may be useful in the prediction of pregnancy viability in cases of uncertainty.
Discussion
One of the biggest challenges in early pregnancy ultrasonography is accurate differentiation between a true gestation sac and a pseudosac. Pseudosacs, although rare, are strongly suggestive of an ectopic pregnancy, hence it is an important distinction to make, ideally as soon as possible. Both appear initially as intrauterine fluid collections or ‘empty sacs’. Whilst experts may claim that it is not difficult to differentiate between the two structures, in clinical practice, many of the individuals undertaking the scans in early pregnancy do not claim to be experts. Traditional teaching has always been to wait until either a yolk sac or fetal pole are visualized within the sac before confirming a definite intrauterine pregnancy. Although safe, inherent with this approach is that an intrauterine fluid collection is visible using transvaginal ultrasound from around day 28 but a yolk sac is not visible until at least day 35. If an ultrasound is undertaken during this time, an ‘empty sac’ will be seen and uncertainty will ensue.
Application of the results from the studies described above could potentially revolutionize the care of women with diagnostic uncertainties in early pregnancy. Firstly, the confirmation that uncertain diagnoses in early pregnancy are highly anxiogenic, means that Early Pregnancy Assessment Units can now justify the allocation of resources to help alleviate this distress. This is crucial if we are to improve the holistic nature of the care provided to women with complications of early pregnancy.
Furthermore, the discovery that the double decidual sac sign can accurately predict an intrauterine pregnancy prior to visualization of embryonic contents (and therefore effectively exclude an ectopic pregnancy) means that we can rationalise follow-up, improve consistency and minimise error in the management of women with ultrasonographic evidence of an empty sac in early pregnancy.
Although it could be argued that utilization of the double decidual sac sign does not minimise the number of women given uncertain diagnoses in early pregnancy, merely swap concerns regarding location to ones regarding viability, in clinical practice it is the potential consequences of pregnancies of unknown location that are most hazardous, both because of the immediate threat to health and the future threats to fertility. Furthermore, if the findings from our prognosis study are confirmed, and appropriate threshold levels for our biomarkers determined, it may be possible to minimise the duration of uncertainty for women with pregnancies of uncertain viability to hours rather than weeks.
Using a combination of approaches therefore, we have achieved the overall aim of this thesis in minimising diagnostic uncertainties in early pregnancy, the clinical benefits of which are multifold. Not only does it reduce anxiety for women, but also prevents unnecessary investigations from being performed in those with an intrauterine pregnancy and minimise morbidity and mortality, permit earlier, potentially less invasive intervention and possibly preserve future fertility for women with an ectopic pregnancy.
Actions (Archive Staff Only)
 |
Edit View |
Tools
Tools
