Materials and Methods
Participants
The study population consisted in a consecutive series of subjects who attended the gynecologic emergency department (GED) of Hospital de Clínicas de Porto Alegre between April 14, 2011, and Oct. 31, 2013.
Participant recruitment
Consecutive pregnant women between 14 and 49 years old who attended the GED for any reason were invited to participate in the study.
Inclusion and exclusion criteria
Subjects were included if they had a pregnancy <12 weeks according to the last menstrual period or previous gestational ultrasonography. Subjects were excluded if their pregnancy was not confirmed by urinary β-hCG, if they had a pregnancy of ≥12 weeks, or if they did not give written consent to participate in the study. Pregnancy and gestational age were confirmed by urinary and/or serum β-hCG and ultrasound, respectively.
Data collection
After signing the written consent, a standard questionnaire was obtained. Next, a fresh urine sample was collected from the patient for the index text (Inexscreen). Subjects were followed with serial plasma β-hCG and transvaginal ultrasound. In case of pregnancy of unknown location, subjects were followed every 48 hours or every week until the outcome of pregnancy was identified.
Reference standard and its rationale
A transvaginal ultrasound was used as the reference standard to confirm the presence of a viable, or a nonviable pregnancy. VUS was performed within 4 hours after the Inexscreen test. Pregnancy viability was defined as the presence of intrauterine embryo/fetus with cardiac activity. Miscarriage was defined by the absence of visible heartbeat in an embryo with crown-rump length ≥7 mm, or if the mean gestational sac diameter was ≥25 mm and no structure was visualized inside. The diagnostic algorithm published by Mol et al was used for diagnosis of ectopic pregnancy. Ectopic pregnancy was confirmed by surgery and pathology report.
Technical specifications
Inexscreen was the index test for screening viable or nonviable first-trimester pregnancy. Urine specimens were collected in a clean, dry, plastic container and Inexscreen was run within 10 minutes of collection. With the disposable pipette provided with the Inexscreen kit, 5 drops of urine were loaded into the sample well of the Inexscreen test device. Interpretation of the test result was performed after 5 minutes, according to the manufacturer’s instructions. Briefly, Inexscreen test has 2 windows (A and B), and 2 lines in each window (C and T). Line C is the internal control; if it is absent in one of the windows, the test is discarded and a new test used. The intensity of the T line was defined by visual comparison with standards given by the manufacturer. The intensity of the lines was graduated in a grading system of whole numbers between 0 and 10. The cut-off between normal and abnormal followed the manufacturer instructions. A ratio of the intensity of lines A > B was considered abnormal; a ratio A ≤ B was considered normal.
Persons executing and reading Inexscreen
Two experts in Inexscreen reading (J.L.G., R.F.S.) trained and supervised the clinical physicians (n = 8), the residents and medical students that were in rotation every month (total n = 37/30.5 months). The physicians, residents, and medical students evaluated the intensity of line T in windows A and B.
Persons executing ultrasound and pathologic analysis
Board certified radiologists and pathologists were responsible for executing transvaginal ultrasound and pathologic analysis, respectively. Readers of the Inexscreen test, radiologists, and pathologists were blind to the results of each other’s test.
Statistical methods and sample size
The performance of the Inexscreen test was calculated using a 95% confidence interval (CI) for sensitivity, specificity, positive and negative predictive values, and positive and negative likelihood ratios. Kappa index was used to assess inter- and intraobserver agreement. Reproducibility of the test, ie, normal or abnormal, was verified between 2 independent observers using 60 cases comprising the whole spectrum of results (ie, from 0 to 10). Digital pictures of lines A and B of these 60 cases were taken and stored in a file. Results from a senior researcher (R.F.S.) were compared with the results of a naïve researcher for interobserver agreement. The senior researcher scored the same 60 cases in 2 different occasions for intraobserver agreement. Analyses were performed using Prism 6.0 (GraphPad, San Diego, CA), online kappa calculator ( http://vassarstats.net/kappa.html ) and online diagnostic test calculator ( http://araw.mede.uic.edu/cgi-bin/testcalc.pl ).
Sample size was calculated according to the nomogram described in the literature. The following parameters were used: an estimated incidence of ectopic pregnancy of 10% (±5%) and an estimated specificity of 95% (with a precision of ±5%). Sample size calculation yielded a total number of 730 subjects. For cases of miscarriage, where the estimated incidence of miscarriage was 40%, the minimal number of subjects was 180. With these parameters we were able to verify with a 95% CI, that Inexscreen has specificity between 90 and 100% to diagnose ectopic pregnancy in a population where the prevalence is between 5 to 15%.
Ethical issues
This study was submitted and approved by Comite de Ética em Pesquisa of Hospital de Clínicas de Porto Alegre, the local institutional review board (no. 11-0113). Inexscreen was evaluated in the GED of the Hospital de Clínicas de Porto Alegre, in Porto Alegre, RS, Brazil.
Results
Beginning and end dates of recruitment
Data collection was performed between April 14, 2011, and Oct. 28, 2013.
Clinical and demographic characteristics of the study population
Most of the participants were white and the mean gestational age was 8.1, 6.5, and 7.9 weeks for intrauterine pregnancy, ectopic pregnancy and miscarriage, respectively ( Table 1 ).
| Characteristics | Viable pregnancy n = 450 | Ectopic pregnancy n = 73 | Miscarriage n = 280 |
|---|---|---|---|
| Age, mean (SD) | 26.7 (6.5) | 28 (6.7) | 28.3 (7.8) |
| Ethnicity, n (%) | |||
| White | 315 (70) | 44 (60) | 182 (65) |
| Black | 131 (29) | 29 (40) | 88 (31) |
| Native-Brazilian | 3 (0.7) | 4 (1.7) | |
| Asian | 1 (0.3) | 4 (1.7) | |
| Not available | 2 (0.6) | ||
| Gestational age, mean (SD) | 8.1 (2.1) | 6.5 (2) | 7.9 (2.1) |
| Risk factors (no/yes) | |||
| Hx ectopic pregnancy | 425/25 | 59/14 | 269/11 |
| Hx tubal surgery | 429/21 | 64/9 | 257/13 |
| Smoking | 334/116 | 47/26 | 214/66 |
| Hx PID | 414/36 | 58/15 | 258/22 |
| Hx ≥3 miscarriages | 425/25 | 70/3 | 271/9 |
| Hx infertility a | 405/45 | 52/21 | 228/52 |
| >5 sexual partners | 338/112 | 48/25 | 197/83 |
| Hx intrauterine dispositive | 408/42 | 66/7 | 261/19 |
| Signs and symptoms, n (%) | |||
| Asymptomatic | 71 (16) | 5 (7) | 49 (18) |
| Pelvic pain | 163 (36) | 8 (11) | 31 (11) |
| Vaginal bleeding | 75 (17) | 14 (19) | 54 (19) |
| Pelvic pain + bleeding | 141 (31) | 46 (63) | 146 (52) |
a Defined as failure to achieve a clinical pregnancy after ≥12 mo of regular unprotected sexual intercourse.
Number of participants
A total of 815 Inexscreen tests were used in the study and none of them were discarded for being faulty (ie, negative control line). Six subjects did not satisfy the inclusion criteria and were excluded; six were lost to follow-up. A total of 803 subjects satisfied inclusion criteria and had a VUS or were submitted to surgery ( Figure ). The median follow-up was 3 days (range, 1–60 days).
Time interval between tests results
The time interval between the Inexscreen test and the vaginal ultrasound was 4 hours or less. No treatment was administered in between.
Distribution of the severity of the disease
A total of 125 subjects were asymptomatic with no or ≥1 risk factor. Asymptomatic subjects without risk factor comprised 6.6% (53 of 803) of the study population. From these 53 subjects, 39.6% had an abnormal pregnancy (3 cases were ectopic pregnancy; 18 cases of miscarriage). The majority of subjects (93.4%) had pelvic pain and/or vaginal bleeding.
Cross tabulation of the Inexscreen and transvaginal ultrasound
All 803 subjects included in the study underwent transvaginal ultrasound and had a urine sample analyzed by Inexscreen. Positive Inexscreen (abnormal) in abnormal pregnancy was seen in 47 cases; abnormal Inexscreen in normal pregnancy was seen in 81 cases. Normal Inexscreen in abnormal pregnancy was seen in 306 cases, and normal Inexscreen in normal pregnancy was seen in 369 cases. There were no missing or indeterminate results.
Adverse events from Inexscreen and vaginal ultrasound
No adverse event occurred by performing the index test (Inexscreen) or reference standard (transvaginal ultrasound).
Estimates of the test
The test performance to identify an abnormal first trimester pregnancy was as follows: sensitivity: 13% (95% CI, 10–17%); specificity: 82% (95% CI, 78–85%); positive predictive value: 37% (95% CI, 28–46%); negative predictive value: 55% (95% CI, 50–58%); accuracy: 52%; positive likelihood ratio: 0.74 (95% CI, 0.53–1.03); negative likelihood ratio: 1.06 (95% CI, 1–1.12). The pretest probability of an abnormal pregnancy was 44%; the posttest probability after a positive and negative test was 37 and 45%, respectively ( Table 2 ). Similar results were observed in a subgroup of symptomatic subjects with gestational age between 5 and 8 weeks of pregnancy ( Table 3 ).
