Rationalizing the management of pregnancies of unknown location: temporal and external validation of a risk prediction model on 1962 pregnancies

• Published in: Human reproduction (Oxford) Vol. 28; no. 3; pp. 609 - 616
• Main Authors: Van Calster, B., Abdallah, Y., Guha, S., Kirk, E., Van Hoorde, K., Condous, G., Preisler, J., Hoo, W., Stalder, C., Bottomley, C., Timmerman, D., Bourne, T.
• Format: Journal Article
• Language: English
• Published: England Oxford University Press 01-03-2013
• Subjects: Adult; Chorionic Gonadotropin - blood; Embryo Loss - etiology; Embryo Loss - prevention & control; False Positive Reactions; Female; Hospitals, Teaching; Hospitals, Urban; Humans; London; Models, Biological; Pregnancy; Pregnancy Complications - etiology; Pregnancy Complications - prevention & control; Pregnancy, Ectopic - blood; Pregnancy, Ectopic - diagnosis; Pregnancy, Ectopic - epidemiology; Pregnancy, Ectopic - physiopathology; Prognosis; Prospective Studies; Retrospective Studies; Risk; Sensitivity and Specificity; Young Adult; London; pregnancy of unknown location; ectopic pregnancy; miscarriage; hCG; ultrasound
• ISSN: 0268-1161; 1460-2350
• DOI: 10.1093/humrep/des440

STUDY QUESTION
Can we accurately define a group of pregnancies of unknown location (PULs) as low risk in order to safely reduce follow-up for these pregnancies and allocate resources to pregnancies at an increased risk of being ectopic?
SUMMARY ANSWER
Prediction model M4 classified around 70% of PULs as low risk, of which around 97% were later characterized as failed PULs or intrauterine pregnancies (IUPs), while still classifying 88% of ectopic pregnancies as high risk.
WHAT IS KNOWN ALREADY
Depending on the level of suspicion of ectopic pregnancy (EP), women with a PUL receive a lengthy follow-up in order to confirm the location and viability of the pregnancy.
STUDY DESIGN, SIZE, DURATION
A multi-centre diagnostic accuracy study of 1962 patients was carried out between 2003 and 2007 for retrospective temporal validation and between 2009 and 2011 for prospective external validation. The reference standard is the final characterization of PUL as failed pregnancies or IUPs (low risk), or as ectopic pregnancies (high risk). M4 is a multinomial logistic regression model based on the serum human chorionic gonadotrophin (hCG) levels at presentation and 48 h later.
PARTICIPANTS/MATERIALS, SETTING, METHODS
Temporal validation data from 1341 PULs collected at St George's Hospital in London were available, of which 53% were failed, 39% were intrauterine and 8% were ectopic pregnancies. External validation data from 621 PULs collected at four other London-based teaching hospitals were available, of which 63% were failed, 22% were intrauterine and 15% were ectopic pregnancies.
MAIN RESULTS AND THE ROLE OF CHANCE
The EP rate varied between 8 and 16% across the five hospitals. At St George's, 980 [73.1%, 95% confidence interval (CI): 70.5–75.4] PULs were considered low risk. Of these, 963 were failed PULs or IUPs (98.3%, 95% CI: 97.2–98.9) and 17 were ectopic pregnancies. At the other four hospitals, 62–75% were considered low risk, with 96–98% of these turning out to be failed PUL or IUP. Eighty-five percent (95% CI: 76.8–90.2) of the ectopic pregnancies were considered high risk at St George's, compared with 80–92% in the other hospitals.
LIMITATIONS, REASONS FOR CAUTION
Of total, 120 patients had been excluded due to loss to follow-up, and a further 102 patients because of missing hCG levels due to differences in local clinical practice. There are variations in the definition of a PUL used in different countries.
WIDER IMPLICATIONS OF THE FINDINGS
The suggested protocol could safely reduce the follow-up in the majority of PUL such that units could increase the focus on women at a risk of complications. This would lead to a change in the management of the majority of women with a PUL and a more efficient use of resources. At the end of the manuscript, we provide a link to enable clinicians to use the protocol.
STUDY FUNDING/COMPETING INTEREST(S)
B.V.C. is supported by a postdoctoral fellowship from the Research Foundation Flanders (FWO). K.V.H. is supported by a fellowship from the Flanders' Agency for Innovation by Science and Technology (IWT-Vlaanderen), by the Research Council KU Leuven (GOA MaNet), by the Flemish Government (iMinds) and by the Belgian Federal Science Policy Office (IUAP P7/DYSCO). T.B. is supported by the Imperial Healthcare NHS Trust NIHR Biomedical Research Centre. No competing interests are declared.