Accuracy of home pregnancy tests at the time of missed menses

Transcription

Accuracy of home pregnancy tests at the time of missed menses
American Journal of Obstetrics and Gynecology (2004) 190, 100e5
www.elsevier.com/locate/ajog
Accuracy of home pregnancy tests at the time of
missed menses
Laurence A. Cole, PhD,* Sarah A. Khanlian, Jaime M. Sutton, Suzy Davies, PhD,
William F. Rayburn, MD
USA hCG Reference Service, Department of Obstetrics and Gynecology, School of Medicine, University of New
Mexico, Albuquerque, NM
Received for publication May 12, 2003; revised August 7, 2003; accepted August 27, 2003
––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
KEYWORDS
Human chorionic
gonadotropin
Home pregnancy test
Missed menses
Objective: The purpose of this study was to investigate the validity of home pregnancy tests
(HPTs) around the time of the missed menses.
Study design: Levels of human chorionic gonadotropin (hCG) were determined in urine around
the time of the missed menses, and 5th and 95th centiles were calculated. In a blind study evaluating 18 brands of HPTs, each was tested six times with five concentrations of hCG, 0 to 100
mIU/mL. Sensitivity was defined as the concentration at which six of six brands gave positive
results, at the suggested reading time or at an extended time.
Results: It was estimated that a sensitivity of 12.5 mIU/mL was needed to detect 95% of pregnancies at the time of missing menses. Only 1 of 18 HPTs had this sensitivity. If faintly discernible
results considered, 2 of 18 brands gave false-positive or invalid results. Clear positive results were
given by only 44% of the brands at 100 mIU/mL hCG. If one accepted faintly discernible and
extended reading times, all brands responded at 100 mIU/mL. A test with 100 mIU/mL sensitivity would detect approximately 16% of pregnancies at the time of missed menses.
Conclusion: The utility of HPTs is questioned. Clinicians need to be aware of the limitations of
current HPT brands.
Ó 2004 Elsevier Inc. All rights reserved.
–––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––
Home pregnancy tests (HPTs) have become a widely
accepted first step in the detection of early pregnancy.
More than 6 million women conceive annually in the
United States, and many more wish to know whether
a desired or inadvertent conception has occurred.1 Preg-
Supported in part by National Institutes of Health grant No.
HD35654.
* Reprint requests: Laurence A. Cole, PhD, Department of
Obstetrics and Gynecology, 2211 Lomas Blvd NE, Albuquerque,
NM 87131.
E-mail: [email protected]
0002-9378/$ - see front matter Ó 2004 Elsevier Inc. All rights reserved.
doi:10.1016/j.ajog.2003.08.043
nancy testing at home offers the advantages of privacy
and early detection without the need to go to a clinic.
It is no wonder that HPTs are among the most commonly purchased over-the-counter disposable health
kits.
Since their introduction in 1975, the number of commercially available HPTs has increased from 4 to more
than 20 brands. All manufacturers continue to report
their HPT as being more than 99% accurate.2 Two major changes have occurred during this period: (1) HPTs
now use modern immunometric assay (antibody-dye/
hCG/immobilized antibody) and (2) the tests claim
101
Cole et al
use on the first day of the missed period or earlier rather
than use after 5 weeks’ gestation.2
The objective of this investigation was to determine
whether there is validity in the claim advertised by all
HPT brands of O99% accuracy and use of the test at
the first day of the missed period.
Material and methods
Urine hCG concentrations at missed menses
After obtaining approval from our Human Research
Review Committee, we first determined the urine concentrations of hCG in women during early pregnancy.
Urine samples were gathered between 4 weeks 0 days
and 4 weeks 3 days from the last menstrual period of
25 women achieving clinical pregnancy while attending
an infertility clinic. Samples were random and not selected in any way. All were singleton pregnancies. Four
of 25 ended in spontaneous abortion or ectopic pregnancy (16%). None received hCG therapy to induce
ovulation. In all cases, early pregnancy was confirmed
by measurement of serum hCG.
Figure 1 displays a scattergram of hCG concentrations during these 4 days after the missed menses. Individual hCG values are logarithmically distributed.3,4
Values were examined by logarithmic regression using
multiple of the median methods. These are established
methods for comparing distributions of normal pregnancy hCG concentrations in the first3 and second trimesters4 of pregnancy. Results were ranked and centiles
calculated as previously described.3,4 By use of these
methods, permitted centiles can be calculated for 0, 1, 2,
and 3 days after missing the period data from a regression equation established from all 25 patient results.
From these data the calculated median result at the time
of missed menses was 49 mIU/mL, and the 5th and
95th centiles were calculated as 12.4 and 241 mIU/mL,
respectively. An HPT brand, therefore, would need to
measure an hCG concentration of 12.4 mIU/mL to detect 95% of intrauterine pregnancies at the time
of missed menses. Furthermore, sensitivities of 21, 35,
and 58 mIU/mL would be necessary to detect 95% of
intrauterine pregnancies at days 1, 2, and 3 after missed
menses.
Preparation of urine samples
Urine samples were pooled from five healthy nonpregnant women. Currently, the purest and most homogeneous form of hCG available is recombinant hCG
(Sigma Chemical Co, St Louis, Mo). This hCG was
added to the urine by one of the investigators (L. C.)
in amounts calculated to obtain concentrations of approximately 0, 12.5, 25, 50, and 100 mIU/mL. Samples
Figure 1 Scattergram displaying 25 urine hCG concentrations
during the first 4 days after the missed menses. Logarithmic
regression analysis and multiples of the median analysis were used
to determine centiles. Lines represent the 95th, 50th, and 5th
centiles for the 25 results.
of 7 mL (required for each evaluation, at each hCG concentration for each HPT) were blindly coded according
to hCG concentrations, then frozen at 80(C. Additional 7-mL urine samples, 13 for each hCG concentration, were thawed and warmed to 37(C multiple times
on each day of this study to affirm the constancy of
the hCG concentration. No significant variations were
observed between the calibrated and actual hCG concentrations during the study period, at 0 (all !1.0),
12.5 (12.6 G 0.5), 25 (27 G 2), 50 (52 G 2), and 100
(99 G 3) mIU/mL.
HPTs and testing protocol
This study evaluated three different lots of 18 brands of
HPT, purchased at multiple retail stores. This permitted
consideration of lot-to-lot variations and normal shipping and storage variations into our results. All HPT
brands work using similar multi-antibody immunometric assay methods with a result window in the HPT to
show a positive test and a control window to show that
the HPT is functioning properly.
Figure 2 displays a flow sheet illustrating the experimental design of the brands to be tested. Each was tested
with five different concentrations of hCG (0, 12.5, 25,
102
Cole et al
Figure 2 Flow sheet illustrating how the 18 HPT brands were tested with 5 different urine hCG concentrations by 3 technicians (total, 465
HPT brands evaluated). A random testing protocol was developed for the 3 technicians or readers, for evaluating all HPT brands. Under these
protocols, the 3 technicians evaluated 6 samples for each data point.
50, and 100 mIU/mL). Each urine concentration was
blindly tested 6 times by the three professionally trained
clinical technicians. The readings included three different lots each HPT to consider interlot variation. Five
brandsdEckerd One-Step, Rite Aid One-Step, Sav-On
Osco One-Step, Target Brand, and Walgreens OneStepdare very similar HPTs produced by one manufacturer and labeled for specific retailers. Considering the
similarity between these brands, they were only tested
in triplicate (three HPTs at each hCG concentration,
three lots). For validation, the 50 mIU/mL hCG urines
were repeated for a total of six valuations at each hCG
concentration for each product.
Experiments throughout the course of the study were
undertaken by use of identical conditions in the same
testing laboratory. The study was blinded as to the urine
sample hCG concentration and the order in which the
kit was tested. It was not possible to blind the individual
brands because each had a distinctive shape or label and
needed to be maintained in a sealed package until the
time of testing.
Approximately every 30 minutes the investigator who
prepared and coded the urine concentrations (L. C.)
thawed a batch of urine samples in a water bath for the
three technicians and allowed evaluation only when the
samples reached 37(C. The HPT brands were tested
strictly according to the package insert instructions. Each
HPT was read twice, first at the suggested reading time
(1 to 5 minutes depending on the kit instructions) and
again at an extended time, 10 minutes, to examine maximal results. This was the longest reading time suggested
by 14 of 18 HPT instructions and was not considered detrimental by the remaining 4 HPT instructions. Test results
that were clearly positive were recorded as ‘‘2,’’ faintly discernible as ‘‘1,’’ and negative as ‘‘0.’’
Sensitivity was defined as the urine concentration at
which six of six HPTs of a specific brand gave a positive
result (or 3/3 HPTs in the case of the 5 brands where only
3 HPT brands tested at 12.5, 25, and 100 mIU/mL concen-
trations), whether faintly discernible (score = 1) or
clearly positive (score = 2). Sensitivity was only calculated for legitimate test results, those giving a valid control
line (score = 1 or 2) to show the HPT was functioning
properly, as well as a positive test line (score = 1 or 2).
Results
The Table displays the 18 HPT brands in alphabetic
order and the corresponding recommended or claimed
earliest time to read the test results. Overall, the brands
were unreliable in providing a clearly positive result at
any of the four hCG concentrations at the suggested
reading time. Brands with a longer suggested reading
time (5 minutes) did not have higher sensitivity than
those with a shorter suggested reading time (1 minute).
Clearly positive results were present in only 44% of
the brands at the highest hCG concentration (100
mIU/mL). When a positive result included a faintly discernible line, the sensitivity was improved (83% of
brands gave a positive result with the 100 mIU/mL
hCG urine) but still inexact.
Test sensitivity was improved when HPT brands were
reread at an extended reading time of 10 minutes. Despite this, the 18 HPT brands were 100% accurate only
when the urine hCG concentration was at the highest
level tested (100 mIU/mL), the test was evaluated at
the extended time, and when faintly discernible results
were included (Table).
Certain brands were more sensitive at lower hCG
concentrations than others. None provided clearly positive results at the suggested reading time using the 12.5
or 25 mIU/mL concentration. One brand (First Response, Early Results) gave consistent faintly discernible
results at 12.5 mIU/mL. Three brands gave some form
of a positive result, whether clear or faintly discernible
at the suggested reading time using the 25 mIU/mL concentration of hCG (Clear Blue Easy, One Minute; Clear
Plan Easy; First Response, Early Results). Three brands
103
Cole et al
Table
Sensitivity of available HPT brands
Results at suggested read time
Home pregnancy test
Answer
Answer, Quick and Simple OneStep
Clear Blue Easy, One Minute
Clear Choice
Clear Plan Easy
Confirm
CVS
E.P.T
Eckerd, One-Step
Equate
First Response, Early Result
Inverness Medical, Early
Inverness Medical, Early, Cassette
K-Mart, American Fare, Easy to Read
Rite Aid, One-Step
Sav-On, Osco, One-Step
Target Brand
Walgreens, One-Step
Sensitivity, clearly positive onlyy
Results at extended read time
Suggested read 12.5
25
50
100
12.5
25
50
100
time (min)
mIU/mL mIU/mL mIU/mL mIU/mL mIU/mL mIU/mL mIU/mL mIU/mL
2
2
1
5
3
5
3
3
3
1
3
3
5
3
3
3
3
3
Sensitivity, clearly positive or faintly discernibley
4D1
0D0
2D2
0D3
0D1
0D0
0D0
0D2
0D0
0D0
0D6
0D0
0D2
0D0
0D0
0D0
0D0
0D0
0/18
(0%)
1/18
(6%)
3D2
3D2
3D3
1D3
3D3
0D0
0D0
0D5
0 D 0*
0D0
3D3
0D1
2D1
1D3
0 D 0*
0 D 0*
0 D 0*
0 D 0*
0/18
(0%)
3/18
(17%)
6D0
6D0
3D3
2D3
6D0
0D1
0D5
2D4
0D0
0D3
6D0
0D2
5D1
3D2
0D3
1D4
0D6
0D0
4/18
(22%)
8/18
(44%)
6D0
6D0
6D0
3D3
6D0
0D5
5D1
3D3
0 D 3*
0D3
6D0
2D4
6D0
2D4
2 D 1*
1 D 1*
3 D 0*
3 D 0*
8/18
(44%)
15/18
(83%)
3D1
0D0
2D4
0D4
0D6
0D1
0D3
2D0
0 D 0*
4D1
0D6
0D4
0D6
0D4
0 D 1*
0 D 1*
0 D 0*
0 D 3*
0/18
(0%)
5/18
(28%)
6D0
3D2
6D0
0D5
4D2
0D0
2D2
5D1
0 D 2*
6D0
5D1
3D3
3D0
4D2
0 D 3*
2 D 1*
1 D 2*
3 D 0*
4/18
(22%)
12/18
(67%)
6D0
6D0
6D0
4D1
6D0
0D2
6D0
6D0
6D0
6D0
6D0
6D0
6D0
5D1
6D0
5D0
6D0
3D0
13/18
(72%)
14/18
(78%)
6D0
6D0
6D0
4D2
6D0
6D0
6D0
6D0
3 D 0*
6D0
6D0
6D0
6D0
5D1
3 D 0*
3 D 0*
3 D 0*
3 D 0*
16/18
(89%)
18/18
(100%)
In each column, the first number represents the clearly positive results, while the second number represents a faintly discernible positive result. The
2 reading times are those suggested by the manufacturer and the extended time of 10 minutes (recommended as maximum read time by 14 of 18
manufacturers).
* Eighteen HPT products widely available at US retail stores were blindly evaluated: Answer, Answer Quick and Simple OneStep, and First Response
Early Result, produced by Church and Dwight Inc, Princeton, NJ; Clear Blue Easy One Minute and Clear Plan Easy, produced by Unipath Diagnostics,
Princeton, NJ; E.P.T., manufactured by Warner Lambert Consumer Healthcare, Morris Plains, NJ; Confirm, distributed by Durex Consumer Products Inc,
Norcross, Ga; Clear Choice, manufactured by Phamatech Inc, San Diego, Calif; Equate, produced by LifeCare Medical International, Philadelphia, Pa; Eckerd
One-Step, Rite Aid One-Step, Sav-On Osco One-Step, Target Brand, Walgreens One-Step, Inverness Medical Early, and Inverness Medical Early, Cassette,
manufactured by Inverness Medical, Waltham, Mass; K-Mart American Fare Easy to Read, manufactured by American Fare, Troy, Mich.
y
Only 3 HPT brands evaluated at 12.5, 25, and 100 mlU/mL concentrations.
z
Sensitivity was defined as the urine hCG concentration at which 6 of 6 HPT brands (or 3 of 3 HPT brands, in the case of 5 brands where only 3 HPT
brands tested at 12.5, 25, and 100 mlU/mL concentrations) gave a positive result, whether faintly discernible (questionable) or clearly positive.
failed to show consistently positive results (6/6 positive
results), using both the highest hCG concentration
(100 mIU/mL) and the suggested reading time (Confirm,
Equate, Sav-On Osco One-Step). It was not until the
extended time of 10 minutes that all 18 HPT brands
became clearly or faintly positive at the highest hCG
concentration (100 mIU/mL).
Several brands had indications of technical or design
problems. One in six of the Confirm tests gave a clear
falsely positive result with the 0 mIU/mL urines at the
suggested reading time. This is a false-positive result.
Similarly, three of six Clear Choice HPTs tested at
0 mIU/mL yielded false-positive results at the suggested
reading time. It was also observed that several tests
failed to display a control band, which indicates to users
that the HPT is functioning correctly. This occurred at
the suggested reading time and at the extended reading
time in 1 of 30 evaluations (those at 0, 12.5, 25, 50, and
100 mIU/mL) with the Answer Quick and Simple, 1 of
18 with the Sav-on Osco, 9 of 30 with the Clear Choice,
and 10 of 30 evaluations with the Confirm product.
Comment
Home pregnancy testing was introduced almost 3 decades ago. To appeal to consumers in this competitive
market, manufacturers have shortened the time of early
diagnosis to the same day of the missed period. One
brand (First Response, Early Result) has been reported
to be accurate 4 days before the missed menses. Several
new brands or new adaptations of existing brands, not
evaluated here, now make this earlier use claim. Our research found that in many circumstances HPTs were not
sensitive enough to detect the low hCG levels reported
104
at the onset of a missed period, or they provided faintly
discernible results at multiple test concentrations. A
faintly discernible result is open to misinterpretation.
We found in well-dated pregnancies that a wide
range of hCG (from 23 to 2438 mIU/mL) is produced
by pregnant women by 28 to 30 days after the onset of
the last known menstrual period. It was calculated that
for a test to detect 95% of pregnancies at the time of
missed menses, it would need to consistently detect at
least 12.4 mIU/mL hCG. If a HPT brand was used
at days 1, 2, or 3 after the onset of the missed menses,
then to detect 95% of pregnancies it would need to
consistently detect 21, 35, and 58 mIU/mL hCG, respectively. It should be noted that these values are
estimates based upon regression lines from only 25
women achieving pregnancy.
All 18 HPT brands evaluated use proven multiantibody technology with very similar testing instructions.
Tests were credited as being positive regardless of
whether the line was clearly positive or only faintly discernible in the result window. A total of 6%, 17%, 44%,
and 83% of brands tested at the suggested reading time
gave clear or faintly discernible positive results at hCG
concentrations of 12.5, 25, 50, and 100 mIU/mL, respectively. In contrast, only 0%, 0%, 22%, and 44% of these
brands tested gave clear or indisputably positive results
at the suggested reading time at these same concentrations. These results, together with the data indicating
hCG concentration around the time of missing menses,
clearly indicate that a high percentage of pregnant
women will have a negative test result during the first
few days after the missed period. We found that most
tests were consistently accurate only after the urine
hCG concentration was at least 100 mIU/mL, or the
highest hCG concentration evaluated, allowing for both
clearly positive and faintly discernible readings and allowing for readings extended to 10 minutes, rather that
at the 1- to 5-minute suggested reading time. The poor
findings reported here may in part be due to consideration of multiple lots of brands purchased from multiple stores and the shipping and storage variations. We
estimate that 100 mIU/mL is equivalent to the 84th
centile at the time of the missed menses, so that a test
with this sensitivity would detect only about 16% of
pregnancies.
Realizing our findings, how can manufacturers claim
such a high accuracy in very early gestation? The
reason for this product claim relates to an arcane
Food and Drug Administration (FDA) regulation.
The manufacturer needs only to demonstrate that its
test results, using a nonpregnant woman’s urine supplemented with commercial hCG, agree with those of an
existing test more than 99% of the time. Today’s
HPT brands are more sensitive than previous products,
so they would be expected to give as many positive
results as older products. This FDA evaluation, how-
Cole et al
ever, has no bearing on the ability to detect early pregnancy.
In addition to the inability of these HPT brands to
detect the low hCG concentrations associated with very
early pregnancies, other reasons may explain a negative
result using these brands. The person may have erroneously calculated the time of onset of her missed period,
she may have a delayed menses due to an early pregnancy loss5 or delayed ovulation, or implantation may
have prolonged the onset of significant hCG production.6-8 Although the test instruction sheets were easy
to understand, an erroneously performed test or faintly
discernible result may also explain a falsely negative
result.
The purpose of this investigation was not to recommend specific brands of HPT. Nonetheless, we found
that First Response, Early Result was the most sensitive
test, consistently detecting 12.5 mIU/mL of hCG (in
a supplementary study by Cole et al9 this HPT detected
6.3 mIU/mL hCG), considering the manufacturer’s suggested read time, and both clear and faintly discernible
positive results. This is the sensitivity needed to detect
95% of pregnancies at the times of the missed menses
(12.4 mIU/mL). It is inferred that this test is the most
likely of the 18 brands evaluated to detect a pregnancy
at the time of the missed menses.
Two of the 18 brands had clear technical or design
problems (Clear Choice and Confirm). These tests both
gave falsely positive hCG test results with urine containing no hCG and also gave numerous invalid results as
indicated by the absence of a confirmation or validity
line. Such brands may be unacceptable and may need
to be withdrawn and fixed. They should probably be
avoided because they may generate false hope and great
confusion among users.
Costs of the different HPT brands varied from $8 to
$16. The more expensive brands contained a second test
to be performed later, ‘‘in case the first test was negative.’’ This option to purchase a kit with two tests would
be reasonable if the patient understands in advance that
testing at the day of the missed period or earlier can
yield negative or only faintly positive results (even
though the user maybe pregnant). Use of two tests is
also helpful in determining an early pregnancy loss (ie,
test on day of missed menses is positive, whereas test
a few days later is negative).
Home pregnancy testing continues to be a dominant
method for early pregnancy detection despite questions
about clarity of positive results, false pregnancy detection as a result of early pregnancy loss, and problems
with interpretation of a negative result. Considering the
analytic detection limits of HPTs and the wide variability
in hCG concentrations, we question the greater than
99% accuracy claims made by the manufacturers. In fairness to the consumer, we suggest that these guidelines be
re-examined by the manufacturers. FDA 510(k) regula-
Cole et al
tions should insist on measurements of clinical sensitivity
for detecting pregnancy on the day of the missed menstrual period and at other specific time points, as support
for an accuracy claim. By doing so, a clearer understanding of diagnostic accuracy can be reported on each
brand’s package insert rather than the printing boldly
‘‘over 99% accuracy’’ on the outside of the package.
All these issues need to be considered to avoid delays
in detection of pregnancy, not only for early commencement of prenatal care, but to allow for appropriate
changes in medication and behavior consistent with
healthy pregnancy (avoidance of alcohol, tobacco, and
inappropriate medicines), or to seek earliest pregnancy
termination if desired.
References
1. Ventura SJ, Mosher WD, Curtin SC, Abma JC, Henshaw S.
Highlights of trends in pregnancies and pregnancy rates by
outcome: estimates for the United States 1976-1996. Natl Vital
Stat Rep 1999;47:1-9.
105
2. Butler SA, Khanlian SA, Cole LA. Detection of early pregnancy
forms of human chorionic gonadotropin by home pregnancy test
devices. Clin Chem 2001;47:2131-6.
3. Cole LA, Isozaki T, Jones EE. Urine b-core fragment, a potential
screening test for ectopic pregnancy and spontaneous abortion.
Fetal Diagn Ther 1997;12:336-9.
4. Cole LA, Shahabi S, Oz UA, Bahado-Singh RO, Mahoney MJ.
Hyperglycosylated human chorionic gonadotropin (invasive trophoblast antigen) immunoassay: a new basis for gestational Down
syndrome screening. Clin Chem 1999;45:2109-19.
5. Wilcox AJ, Weinberg CR, O’Connor JF, Baird DD, Schlatterer JP,
Canfield RE, et al. Incidence of early loss of pregnancy. N Engl J
Med 1988;319:189-94.
6. O’Connor JF, Ellish N, Kakuma T, Schlatterer J, Kovalevskaya G.
Differential urinary gonadotrophin profiles in early pregnancy and
early pregnancy loss. Prenat Diagn 1998;18:1232-40.
7. Bjercke S, Tanbo T, Dale PO, Morkrid L, Abyholm T. Human
chorionic gonadotrophin concentrations in early pregnancy after invitro fertilization. Hum Reprod 1999;14:1642-6.
8. Wilcox AJ, Baird DD, Dunson D, McChesney R, Weinberg CR.
Natural limits of pregnancy testing in relation to the expected
menstrual period. JAMA 2001;286:1759-61.
9. Cole LA, Khanlian SA, Sutton JM, Davies S, Stephens N.
Hyperglycosylated hCG (invasive trophoblast antigen, ITA) a key
antigen for early pregnancy detection. Clin Biochem 2003;36:647-55.