Background and Significance:
Point-of-Care testing to rapidly detect pregnancy in women of reproductive age is common practice in Emergency Medicine departments. The patient history may not always reliably identify the possibility of pregnancy1, and a large proportion of these patients are exposed to potential teratogens because of their emergency department visit. In one study, 33% of such patients received a radiographic evaluation and 75% were prescribed a medication2. Additionally, a pregnancy test is required as the first step in the evaluation of a suspected, potentially life threatening, ectopic pregnancy3.
Commercially available rapid immunochemical assays detect human chorionic gonadotropin (hCG) concentrations \>25 mlU/ml, and yield results at 3 minutes using urine and at 5 minutes using serum4-6. However, in clinical practice these tests may take much longer to perform due to significant delays in obtaining urine specimens from patients, or time delays required for separating serum from whole blood. By prolonging patient waiting time for further diagnostic testing and treatment, these delays may adversely impact on patient outcomes, emergency department crowding, and patient satisfaction.
Using whole blood for hCG immunoassay, instead of urine or serum, would facilitate much more rapid emergency department identification of pregnant patients. Whole blood can be obtained immediately during intravenous insertion of phlebotomy, or by finger stick puncture using a lancet in a manner similar to obtaining blood for bedside glucose determination.
Previous study of hCG immunoassays (which were less rapid than the immunoassays which are now commercially available and in use today) utilized whole blood as well as urine and serum, with equal sensitivities7. However, the accuracy of whole blood identifying pregnancy using presently available rapid immunochemical hCG test kits has not been validated.
Objective:
To evaluate the sensitivity and specificity of a commercially available rapid immunochemical hCG assay kit to detect pregnancy using whole blood specimens.
Material \& Methods:
Design: A prospective, blinded study for validation of a diagnostic test
Setting: Two Clinical Sites:
1. Primary Care Women's Center of the Maimonides Medical Center, located at 5008 Fort Hamilton Parkway, Brooklyn, New York.
2. Adult Emergency Department of Maimonides Medical Center, located at Fort Hamilton Parkway and 49th Street, Brooklyn, New York.
Sample:
A convenience sample of healthy women from the above two settings, who meet the following inclusion and exclusion criteria:
Inclusion Criteria:
1. Healthy adult women of child bearing age.
2. Who either request a pregnancy test, or require a pregnancy test for diagnostic studies or treatment.
Exclusion Criteria:
1. Minors
2. Adults who are unable to independently give consent for study
3. Known pregnancy (i.e. - patient appears obviously pregnant)
4. Hemodynamically unstable.
5. Moderately to severely ill-appearing or in moderate to severe pain.
6. Technically not possible to obtain blood from the study subject.
Sample Size:
If we consider the quantitative hCG as the parameter for the final outcome: A sample of 279 from the positive group and 279 from the negative group achieve 80% power to detect a difference of 0.0500 between the area under the ROC (receiver operating characteristic) curve (AUC) under the null hypothesis of 0.8500 and an AUC under the alternative hypothesis of 0.9000 using a two-sided z-test at a significance level of 0.0500. The data are discrete (rating scale) responses. The AUC is computed between false positive rates of 0.000 and 1.000. The ration of the standard deviation of the responses in the negative group to the standard deviation of the responses in the positive group is 1.000.
If we consider a dichotomous outcome for the qualitative result of hCG, instead of testing a hypothesis: A sample size of 384 from each (Positive and Negative test) group will be required to be 95% confident that the results would be within a true population parameter using the following formula:
(Py)(Pn)
= N Std Error2 Py and Pn represent the proportion of people responding to each of the categories in a dichotomous variable and N represent the sample size. We assumed a 50/50 split approach to maximize the variation (i.e. Py = 0.5 and Pn = 0.5) and a Type l error = .05 to obtain a confidence interval of 95 percent.
Methods:
Prior to enrollment in this study, written informed consent will be obtained from all subjects. After valid consent is granted, subjects will complete a brief standardized data collection sheet to determine whether they meet eligibility criteria for participation in this study. Study investigators will review this data. Subjects who meet eligibility criteria and agree to participate in this study will be enrolled as participants. Each participant will submit 5ml of whole blood to be used exclusively for this research study. This blood specimen will be dram from an antecubital or other large upper extremity vein by a physician, nurse, or professional phlebotomist. Each blood sample will be labeled with a unique identification number. One or two drops from each blood sample will be used immediately by the study investigator to perform an ICON 25 rapid hCG immunoassay test (Beckman Coulter, Miami, Florida). The result will be read at 3 and 5 minutes and the specimen identification number and corresponding result recorded in a log. The remainder of each whole blood specimen will be sent to the Maimonides Medical Center chemistry laboratory and processed to perform the reference gold standard test, a quantitative serum total hCG (ADVIA Centaur® System total hCG chemiluminometric immunoassay; Bayer Corporation, Tarrytown, New York) The Beckman Coulter ICON 25 kits is 100% sensitive and 100% specific at hCG values \> 25mlU/ml. Therefore, the reference gold standard for a positive pregnancy test will be a quantitative serum hCG \> 25 mlU/ml. Investigators who perform the whole blood pregnancy test and laboratory technicians performing reference gold standard testing of serum will be blinded to one another's results.
Confidentiality:
This study is not intended to evaluate the performance of any individual resident. Date from this study will be used only for the statistical analysis for this study. The participant's privacy will be maintained by giving each a unique ID number in the database for analysis. Access to the database file will be protected by encryption and a password. All data forms will be secured in a locked file cabinet and access will be limited to the investigators only.
Data Analysis:
All data will be initially entered and analyzed via SPSS. Sensitivity and specificity will also be estimated using 2X2 table.
