In a randomized multicenter trial the investigators want to compare the ability of two hCG-based models in correctly classifying EPs as high risk among PULs and correctly classifying IUPs and failed PULs as low risk after two hCG measurements.

After the classification of PULs into high or low risk of EP, the clinical management will be the same within each risk group (high and low) regardless of randomization group (which model classified the patients). The clinical management will be according to the management protocol published in a NICE guideline.

Pregnancy of unknown location (PUL) is a term used to describe a situation where there is no evidence of a gestation either intra-or extrauterine on transvaginal sonography (TVS) in women seeking medical attention for various reasons in early pregnancy. The principal tool in PUL management besides TVS is the interpretation of the change in serial serum human chorionic gonadotropin (hCG) levels. A recent meta-analysis concluded that there is not enough evidence to recommend a specific strategy using hCG when managing PULs. In a NICE (National Institute of health and care excellence) guideline however women with a PUL acceptable for outpatient surveillance, where the rate of change in rising hCG levels surpassing 63 % in 48 hours (h) highly likely represents an IUP and the patient is triaged to a TVS after one week. Those with a decline in hCG levels of more than 50 % are predicted to be failed PULs and a urinary pregnancy test (UPT) after two weeks is appropriate. If the rate of change in hCG levels fail to meet these cut-offs a re-visit within 24 h is warranted in order to possibly diagnose an EP and suitable treatment selected. In a previous study; an hCG based logistic regression model, M4 classified 70 % of all PULs as low risk of being an EP, hence qualifying for a reduced follow-up while correctly identifying 88 % of EPs. A study from Sahlgrenska University hospital (SU) evaluated the M4 model with similar results, where also the cut-offs by NICE similarly identified 86 % of EPs but to a lesser extent correctly classified low risk PULs. Another finding in the study was that EPs identified by M4 mainly depicted rising hCG levels opposed to the NICE model identifying mainly EPs with declining hCG levels which could have clinical consequences since EPs with rising hCG levels more often seem to necessitate laparoscopic surgery and maybe a postponed diagnosis could be deleterious in these cases. Hence, the M4 model potentially would be better than the NICE cut-offs for PUL management. None of these findings however have been evaluated in a randomized controlled trial (RCT). An updated version of M4 named M6 was first published 2016 and has not been externally validated. Neither has prior cut-offs of declining hCG levels currently used to define a failed PUL.

The handling of these women is often lengthy with a high level of uncertainty before a final diagnosis can be made. The psychological morbidity in this group is largely unknown, while well acknowledged in women following miscarriage, were up to 41% self-reporting clinically significant levels of anxiety and 36% reporting depression within 1 month. In a recent study 24 % (CI 21-28) and 11 % (8-14) respectively of women with either a miscarriage or ectopic pregnancy experienced moderate or severe anxiety and depression according to the Hospital Anxiety Depression Scale (HADS), compared to 13 % (7-21) and 2 % (1-8) in a control group without complications in early pregnancy. We aim to determine the presence and severity of psychological distress and quality of life among women with a pregnancy of unknown location.

A secondary objective is to test following hypothesis: Women with an early normal pregnancy starting as a pregnancy of unknown location have twice as high probability of having HADS-scores above or equal to eight on both subscales, than women with an early normal pregnancy.

We calculated that we would need 121 women in each group to reach a power of 80 % to detect an absolute difference of 15 percentage points after one week, with a 15 % incidence of HADS anxiety score ≥ 8 among control subjects and 30 % among IUP starting as a PUL at a two-sided alpha level of 0.05 (20). For protection against that 10 % of patients will be lost to follow-up a total of 130 women are included in each group. In a second calculation based on using HADS as a continuous scale, 94 women would need to be included in each group to detect an absolute two points difference on the anxiety subscale of the HADS assuming a mean score of 6.0 and a standard deviation of 3.4 in control subjects, at a significance level 0.05 and power 0.80. This minimally important difference is based on mean scores on the HADS anxiety subscale reported for women in early pregnancy in prior studies (20-22). We plan to include a total of 210 women with a 1:1 ratio to account for an expected 10 % loss to follow-up.

Inclusion Criteria:

1. Patients with a PUL
2. Patients with mild or no symptoms
3. Patients willing to be randomized
4. First hCG value <10,000 IU/L
5. Patients with an interval between two hCG measurements of 44-56 h

Exclusion Criteria:

1. Hemodynamically unstable patients
2. Hemoperitoneum
3. Patients not managed as outpatients during the course of the initial two hCG measurements
4. Non-understanding of the oral or written study information