J Clin Epidamiol Vol. 47, No. 11, pp.

121S-1228, 1994
Copyright0 1994 Elsevier Science Ltd
Printedin Great Britain. All rights reserved

Pergamon

0895-4356/94

$7.00 + 0.00

PREFERENCES
OF PREGNANT WOMEN FOR
AMNIOCENTESIS
OR CHORIONIC VILLUS SAMPLING
FOR PRENATAL TESTING: COMPARISON OF PATIENTS
CHOICES AND THOSE OF A DECISION-ANALYTIC
MODEL
PAUL S. HECKERLING,* MARION S. VERP~ and TERESA A. HADRO~
Department of Medicine, University of Illinois, Chicago, IL, and 2,3Departmentof Obstetrics and
Gynecology, University of Chicago, Chicago, IL, U.S.A.
(Received in revised form

17 May 1994)

Abstract-Decision
analytic models have suggested that the choice of amniocentesis or
chorionic villus sampling for prenatal genetic testing is a utility-driven decision. We
compared preferences for prenatal testing among 1.56pregnant women who had chosen
either amniocentesis (n = 82) or chorionic villus sampling (n = 74) for the indication of
maternal age. We also compared their choices with those of a decision-analytic model
based on their preferences, and age-specific rates of spontaneous abortion and chromosomal abnormalities. Preferences were assessed using written scenarios describing
potential outcomes of prenatal testing, and were recorded on linear rating scales. The
differences in preference ratings for first- vs second-trimester prenatal diagnosis of a
normal child (4.2 vs - 1.6, p = 0.0004), and for first- vs second-trimester abortion of
an abnormal fetus (4.4 vs - 1.6, p = O.Ol), were significantly greater among women
choosing chorionic villus sampling than among women choosing amniocentesis. There
were no significant differences between chorionic villus sampling and amniocentesis
patients in their preference ratings for test-related miscarriage, disconfirmed results
at pregnancy termination, or maternal morbidity from therapeutic abortion. After
adjusting for demographic and obstetric factors, the difference in preferences for early
vs late prenatal diagnosis was an independent predictor of the choice of chorionic
villus sampling in a multivariate model. Among women whose decision analyses
selected amniocentesis,
56.8% had chosen amniocentesis,
and among women
whose analyses selected chorionic villus sampling, 63.2% had chosen chorionic villus
sampling (p = 0.05). We conclude that the preferences of pregnant women for the
outcomes of prenatal testing were associated with their choice of amniocentesis or
chorionic villus sampling. In addition, the choice of prenatal test made by the majority
of women was concordant with that of a decision-analytic model that incorporated their
preferences. Nevertheless, because many women made choices that were discordant with
their decision-analytic results, further research into the bases for their choices is
warranted.
Prenatal testing
analysis

Amniocentesis

*Reprint requests should be addressed

to: Paul

Chorionic villus sampling

Utilities

Decision

S. Heckerling, M.D., Department of Medicine, University of Illinois, Box

6998 M/C 787, Chicago, IL 60680, U.S.A.

1215

PAUL S. HECKERLINGet al.

1216
INTRODUCTION

Amniocentesis and chorionic villus sampling are

tests used for prenatal diagnosis of chromosoma1 abnormalities. Each test has advantages
and disadvantages when compared to the other.
Amniocentesis has a lower risk of procedurerelated spontaneous
abortion [l-6], lower
false-positive [3,4,7-201 and indeterminacy
[4, 13-l 5, 16-l 8,2 1,221 rates, and possibly a
lower risk of induced malformations [23,24],
compared to chorionic villus sampling. However, because results are usually not available
before the 20th gestational week, pregnancy
termination after an abnormal amniocentesis
requires a second-trimester abortion, with attendant risk of maternal morbidity [25]. In
addition, delay in genetic diagnosis may result
in increased maternal anxiety [26,27] and diminished maternal-fetal bonding [28]. Chorionic villus sampling provides genetic diagnosis
in the first trimester, thereby decreasing maternal anxiety [26,27] while providing the option of a safer [29], less traumatic, first-trimester
therapeutic abortion if the results are abnormal.
In our previous work [30], using a decisionanalytic model, we showed that the choice of
amniocentesis or chronic villus sampling for
maternal age was driven by two factors: the
difference in the decision makers utilities (values) for first- and second-trimester therapeutic
abortion; and the anxiety cost of awaiting second-trimester genetic diagnosis. These results
remained true over a wide range of plausible
values for the rates of spontaneous abortion
villus
and chorionic
after amniocentesis
sampling, abnormal and indeterminate chorionic villus sampling results, false-positive and
false-negative test results, and maternal morbidity after first- and second-trimester therapeutic
abortion. This suggested that for pregnant
women with differing values for first- vs secondtrimester therapeutic abortion, and for early vs
late genetic diagnosis, the optimal choice of
prenatal test might differ. However, studies
comparing these values, and values for other
outcomes of prenatal testing, among women
choosing amniocentesis and chorionic villus
sampling have not been performed. In addition,
how the choices of prenatal test made by pregnant women compare with those prescribed by
decision analytic models incorporating their
values is not known.
To answer these questions, we studied preferences for the outcomes of prenatal genetic testing among women choosing amniocentesis or

chorionic villus sampling for the indication of

maternal age. We compared preferences of
women choosing each test, and compared their
choices with those of a decision-analytic model
based on their preferences, and age-specific rates
of spontaneous abortion and chromosomal
abnormalities. We also examined demographic
factors, obstetric factors, and factors related to
locus of decision making (patient vs physician),
to determine their effect on test choice.
METHODS

Study population

We studied 156 patients who were referred to

the obstetrical genetic clinics of the University
of Chicago Hospitals for prenatal testing
because of maternal age. Patients presenting
because of an abnormal fetus or child in a
previous pregnancy, a family history of genetic
disease, or an abnormal maternal serum alpha
fetoprotein, were excluded. These 156 patients
were referred by 73 different physicians, of
whom 21 (28.8%) were university-based practitioners, and 52 (71.2%) were private practitioners from the local community
and
surrounding
suburbs. Study patients were
enrolled over a 10 month period, from midFebruary to mid-December 1991. The enrollment period preceded local media publicity
concerning the possible association of chorionic
villus sampling and limb reduction defects by
several months. Thus the preferences of study
participants, and their choice of prenatal test,
were not influenced by this factor. All eligible
patients agreed to participate in the study.
The obstetrical genetic clinics consist of an
amniocentesis clinic and a chorionic villus
sampling clinic. In the chorionic villus sampling
clinic, patients present at 10-l 1 weeks gestation
for counseling and chorionic villus sampling. In
the amniocentesis clinic, patients present at
15-16 weeks gestation for counseling and
amniocentesis. Patients presenting to the clinics
typically have already decided to have the
respective procedure performed. The decision to
have chorionic villus sampling or amniocentesis
will have been made in some cases by the patient
(and her partner), and in others by the referring
physician.
Demographic and obstetric data

Prior to counseling and testing, patients completed a questionnaire containing demographic

and obstetric data. Demographic data included
age, ethnic background, level of education,

Amniocentesis or Chorionic Villus Sampling for Prenatal Testing

occupation, marital status, annual income, and

type of medical insurance. Obstetric data included the number of prior amniocenteses,
chorionic villus samplings, spontaneous abortions, stillbirths, elective abortions, and living
children.
Preference and decision -making data

Patients were also asked to indicate on the

questionnaire their preference ratings for each
of the potential outcomes of prenatal testing.
Potential outcomes considered were first- and
second-trimester diagnosis (by chorionic villus
sampling and amniocentesis, respectively) of a
normal child; spontaneous
abortion
after
amniocentesis, chorionic villus sampling, and
without prenatal testing (with genetic status of
the abortus unknown); first- and second
trimester therapeutic abortion after an abnormal test result, with abortus confirmed as
abnormal, and with disconfirmed results (abortus normal); and maternal morbidity after
therapeutic abortion (i.e. infection requiring
hospitalization
or bleeding requiring transfusion). In a subset of 142 patients, the preference ratings for spontaneous abortion after an
abnormal amniocentesis and chorionic villus
sampling were also determined. All preferences
were recorded prior to counseling and testing.
Neither assessed preference ratings, nor expected values derived from the decision-analytic
model (see below), were used to counsel patients
concerning the test they had chosen. Counseling
sessions focused on the likelihood of an abnormal pregnancy, the accuracy of prenatal testing,
and the risk of spontaneous abortion or fetal
injury from prenatal testing. The likelihood of
maternal morbidity after therapeutic abortion
of an abnormal pregnancy was not discussed
unless the patient raised the issue.
Each preference rating was assessed using a
written vignette describing a hypothetical
situation in which the patient experienced the
outcome in question. For example, the preference for first-trimester therapeutic abortion was
assessed by a vignette that asked the patient to
assume that she had a chorionic villus sampling;
that the result was abnormal and she elected to
have a therapeutic abortion; that the abortion
was performed before she was visibly pregnant
or had experienced fetal movement, and did not
involve the induction of labor; and that the
abortus was found to be chromosomally abnormal, as predicted. The preference for secondtrimester therapeutic abortion was assessed by a

1217

similar vignette, except that the patient was

asked to assume that she had an amniocentesis;
and that therapeutic abortion was performed in
the second trimester, after she was visibly pregnant and had experienced fetal movement, and
involved the induction of labor. (Although
abortion by dilatation and evacuation is available for some second-trimester terminations at
many centers, including ours, abortion by induction is more commonly used to terminate
abnormal pregnancies following genetic amniocentesis.) Other vignettes from the preference
assessment instrument are shown in Appendix
A. To control for the effects of vignette order on
patients preferences, patients were randomized
to receive questionnaires presenting amniocentesis-related or chorionic villus sampling-related
vignettes first.
Patients preferences were recorded on linear
rating scales [31] anchored at 100 units (the best
possible outcome) and 0 units (the worse possible outcome). Patients indicated their preferences for each outcome by placing an x on the
appropriate place on the scale. The distance
from the 0 mark to the x, as measured by
a ruler, was used to calculate each preference
[i.e. preference = (distance
from
0 to
x/length of rating scale line] x 100). On this
scale, preferences that differed by x3.75 units
(3/16 in. on the rating scale line) were difficult to
distinguish reproducibly by linear measurement.
Therefore, when preference differences between
paired amniocentesis- and chorionic villus
sampling-related outcomes differed by c 3.75
units, they were assigned identical preferences
(the mean of the two values), under the assumption that in these cases, the patient had intended
the outcomes to be valued equivalently. This
conservative assumption was designed to avoid
assigning preference differences for amniocentesis and chorionic villus sampling when
none existed. This approximation procedure has
not, to our knowledge, been tested in studies
using rating scales, and its validity remains
uncertain. Of the 156 study participants, 150
(96.2%) completed the rating scale part of the
preference questionnaire.
Patients were also asked whether the decision
to have an amniocentesis or chorionic villus
sampling was made principally by themselves,
by their physicians, or was shared equally
between them. Responses were recorded on a
five-point Likert-type scale, with a spectrum
ranging from decision entirely mine to decision entirely physicians.

1218

PAUL S. HECKERLINC er al.

Data analysis

From each patients recorded preferences, we

derived four preference differences (costs): (1)
the difference in preferences between first- and
second-trimester diagnosis of a normal child.
This represents a cost of waiting for prenatal
testing results (cost of waiting); (2) the difference
in preferences between spontaneous abortion
without prenatal testing and spontaneous abortion after amniocentesis or chorionic villus
sampling. This represents a cost imputed to the
possibility that the testing procedure caused the
abortion (cost of test-related miscarriage). To
calculate this cost, we used preferences for spontaneous abortion after testing of a fetus of
unknown genetic status, because in the absence
of testing, the status of an aborted fetus would
generally not be known; (3) the difference in
preferences between first- and second-trimester
therapeutic abortion of an abnormal fetus. This
represents a cost of therapeutic abortion performed after, rather than before, a woman is
visibly pregnant and has experienced fetal
movement, and that involves the induction of
labor (cost of late therapeutic abortion); and
(4) the difference in preferences of therapeutic
abortion without vs with attendant maternal
morbidity (cost of maternal morbidity).
Comparisons
of demographic
variables,
obstetric variables, preferences, and derived
costs between patients choosing amniocentesis
and chorionic villus sampling were made using
x2 with Yates correction for continuity for
categorical data, and Students t for continuous
data. Based on the variances in preference ratings, our study had 99.9 and 98.9% power to
detect 5-unit differences between the amniocentesis and chorionic villus sampling groups in the
preference costs of waiting for genetic diagnosis
and late therapeutic abortion, respectively. Our
study also had 99.4 and 95.3% power to detect
IO-unit differences in the preference costs of
test-related spontaneous abortion and maternal
morbidity, respectively. These differences can
determine the choice of amniocentesis or chorionic villus sampling in decision models [30], and
therefore are clinically important to detect.
To compare prenatal testing choices made by
patients and those prescribed by formal decision
analytic models, we used our previously
published decision tree [30]. The tree models
the choice of amniocentesis, chorionic villus
sampling, and no prenatal testing, and includes
probability and utility data important in making

that choice. Because all patients in this study had

prenatal testing performed, we restricted our
analysis to the amniocentesis and chorionic villus
sampling branches of the tree (Figs 1 and 2). For
each patient, we used age-adjusted prevalences of
chromosomal abnormalities [32-341 and spontaneous abortion rates [35,36] (Appendix B),
and the patients rating scale preferences, to
evaluate the decision tree. Expected values for
amniocentesis and chorionic villus sampling
were calculated by folding back the decision tree
[37], and the test with the higher expected value
was considered the decision-analytic choice for
that patient. We also calculated the difference in
expected values between the amniocentesis and
chorionic villus sampling strategies to determine
how much better one strategy was than the other.
Comparisons of decision-analytic based choices,
and the test choices actually made by patients,
were made using x2. Stepwise logistic regression
[38] was used to determine the independent
contributions of demographic factors, obstetric
factors, preference ratings, expected values, and
decision-maker status, to the choice of prenatal
test. Odds ratios and 95% confidence limits were
calculated from the logistic coefficients and their
SEs. A two-tailed a level of 0.05 was considered
significant.
RESULTS

Demographic and obstetric data for the 156

women in the study populations are shown in
Table 1. The mean age was 37.5 years; 149
women (95.5%) were 35 years of age or older.
Seven women <35 years of age (five at age 34;

IAbortion

Abnormal
Abortus

Morbidity
(0
Z
2
B__(,
::
2

Positive: Second
Trimester
Therapeutic
Abortion

Normal
Abortus
Abnormal

2
r

No Morbidity

Abortus
Normal

Continue

Abortus

Abnormal

Negative
[

Fig. I. The amniocentesis

branch

Normal
Child

of the decision

tree.

Amniocentesis or Chorionic Villus Sampling for Prenatal Testing

1219

Spontaneous
Abortion
Abnormal
Abortus

Morbidity
0

Normal

Chorionic Villus

Positive: First
Trimester

sampling

Continue

Indeterminate

Abortus

Amniocentesis

Abnormal
Vegative

Fig. 2. The chorionic villus sampling branch of the decision tree.

one at age 33; one at age 32) received prenatal

testing solely because of maternal age, and were
therefore included. Eight-two women (52.6%)
chose amniocentesis for prenatal testing, and 74
women (47.4%) chose chorionic villus sampling.
In 79 cases (52.7%) the choice of prenatal test
was made entirely or mostly by the patient, in
45 cases (30.0%) the choice was shared equally
between patient and physician, and in 26 cases
(17.4%) the choice was made entirely or mostly
by the physician.
Table 2 compares demographic and obstetric
factors between women choosing amniocentesis
and women choosing chorionic villus sampling.
Women choosing chorionic villus sampling were
more frequently college educated, professionally
employed, and covered by commercial insurance, than women choosing amniocentesis,
but the differences were not significant. In
addition, women choosing chorionic villus
sampling showed a trend toward more frequently having had a prior amniocentesis or
chorionic villus sampling. There were no significant differences in the frequencies of spontaneous abortion, elective abortion, or living
children, among women choosing amniocentesis
or chorionic villus sampling. Among women
choosing amniocentesis, the choice of prenatal
test was made by the patient in 51.9% of cases,
by the physician in 20.3%, and shared between
patient and physician in 27.8%; among women
choosing chorionic villus sampling, the choice

was made by the patient in 54.3% of cases, by

the physician in 14.3%, and shared between
patient and physician in 31.4% (p = 0.62).
Table 3 compares rating scale preferences and
derived costs (preference differences) between
women choosing amniocentesis and chorionic
villus sampling. The cost of waiting for genetic
diagnosis (the difference in preferences for firstand second-trimester diagnosis of a normal
child), and the cost of late therapeutic abortion
(the difference in preferences for first- and second-trimester therapeutic abortion of an abnormal fetus), were significantly greater among
women choosing chorionic villus sampling than
among women choosing amniocentesis. There
were no significant differences in the costs of
test-related spontaneous abortion or maternal
morbidity between the two groups. Stepwise
logistic regression identified the cost of waiting
for genetic diagnosis as a significant independent predictor of choosing chorionic villus
sampling (odds ratio for a l-unit change in
utility cost, 1.16; 95% confidence limits,
1.01-1.33; odds ratio for a 5-unit change, 2.06;
95% confidence limits, 1.03-4.11).
Table 3 also compares expected values of
amniocentesis and chorionic villus sampling,
determined by decision analysis, among women
choosing each test. The expected value of chorionic villus sampling, and the difference in expected values between chorionic villus sampling
and amniocentesis, were significantly greater

PAULS. HECKERLING
et al.

1220

villus
among women choosing chorionic
sampling. There was no difference in the expected value of amniocentesis between the two
groups. When expected values were included as
candidate variables for the logistic model, the
difference in expected values between chorionic
villus sampling and amniocentesis (odds ratio
for a l-unit change, 1.21; 95% confidence limits,
1.02-1.43), and having had a prior amniocentesis (odds ratio 8.98; 95% confidence limits,
1.03-78.6), were significant independent predictors of choosing chorionic villus sampling for
prenatal testing.
Table 4 stratifies the comparisons of preferences, costs, and expected values between the
amniocentesis and chorionic villus sampling
groups by locus of decision making. Among
women who either made or shared in the choice
of test, the cost of waiting for genetic diagnosis,
the cost of late therapeutic abortion, and the
difference in expected values between chorionic
villus sampling and amniocentesis, were significantly greater among those choosing chorionic
villus sampling. Among women whose physicians made the choice of test, there were no

Table 1. Demographic

and obstetric data for the study

population

Demographic
31.5 * 2.5

Age W

Ethnicity

White non-Hispanic
Hispanic
Black
Other

Occupation

Housewife
Non-professional
Professional

31 (20.0)?
78 (50.3)
46 (29.7)

Education

< 16 yr
> 16 yr

85 (54.5)t
71 (45.5)

Marital

Not married
Married

29 (18.7)t
126 (81.3)

Income (thousands of dollars)

Insurance

103 (66.0)?
9 (5.8)
38 (24.4)
6 (3.8)

61.5 _+39.2

Commercial
Public aid

141 (91.6)t
13 (8.4)

Obstetric

Amniocenteses

0
21

147 (93.6)t
10 (6.4)

Chorionic villus samplings

0
31

I53 (97.5)?
4 (2.5)

Spontaneous abortions

0
>l

98 (62.4)t
59 (37.6)

Elective abortions

0
21

123 (78.3)t
34 (21.7)

Children

0
31

51 (32.5)-l
106 (67.5)

*Mean f SD.
tNumber (%).

significant differences in preferences, costs, or

expected values between the two groups.
Based on patients preference
ratings,
decision analysis showed amniocentesis to be
the choice in 74.7% of cases, and chorionic
villus sampling to be the choice in 25.3% of
cases. The test choices made by women and
prescribed by their decision analyses agreed in
58.4% of cases, and disagreed in 41.6% of cases.
Among women whose decision analyses
selected an amniocentesis strategy, 56.8% had
chosen amniocentesis, and 43.2% had chosen
chorionic villus sampling; among women
whose decision analyses selected a chorionic
villus sampling strategy, 36.8% had chosen amniocentesis, and 63.2% had chosen chorionic
villus sampling (p = 0.05). Among women who
either made or shared in the choice of test,
54.4% of those whose decision analyses selected
amniocentesis chose amniocentesis, and 63.6%
of those whose decision analyses selected chorionic villus sampling chose chorionic villus
sampling (p = 0.11). Among women whose
physicians made the choice of test, 60% of those
whose decision analyses selected amniocentesis
chose amniocentesis, and 50% of those whose
decision analyses selected chorionic villus
sampling chose chorionic
villus sampling
(p = 0.68).
Preference ratings for spontaneous abortion
of a fetus of unknown genetic status were
generally low, and were less than preference
ratings for therapeutic abortion of an abnormal
fetus (Tables 3 and 4). This probably reflects a
belief by our patients that most such spontaneously aborted fetuses would be normal.
Preference ratings for spontaneous abortion of
an abnormal fetus were higher (Tables 3 and 4).
When we reanalyzed the decision trees using
these preference ratings for spontaneous abortion, 57.1% of women whose decision analyses
selected amniocentesis chose amniocentesis, and
56.8% of those whose decision analyses selected
chorionic villus sampling chose chorionic villus
sampling (p = 0.18). Among women who made
or shared in the choice of test, 54.8% of those
whose decision analyses selected amniocentesis
chose amniocentesis, and 60.5% of those whose
decision analyses selected chorionic villus
sampling chose chorionic
villus sampling
(p = 0.18). Among women whose physicians
made the choice of test, 61.9% of those whose
decision analyses selected amniocentesis chose
amniocentesis, and 28.6% of those whose decision analyses
selected chorionic
villus

Amniocentesis or Chorionic Villus Sampling for Prenatal Testing

1221

Table 2. Comparison of demographic and obstetric variables among women

choosing amniocentesis (AMN) and chorionic villus sampling (CVS)
Variable

AMN

Demographic
Age (yr)

cvs

p value

37.5

37.6

0.71

Ethnicitv

White non-Hisnanic
Hispanic
*
Black
Other

63.4%
6.1%
26.8%
3.7%

69.9%
5.5%
20.5%
4.1%

0.82

Occupation

Non-professional
Professional

76.5%
23.5%

63.0%
37.0%

0.067

Education

c 16 yr
3 16 yr

60.5%
39.5%

47.3%
52.1%

0.14

Marital

Not married
Married

22.5%
11.5%

14.9%
85.1%

0.32

Income (thousands of dollars)

60.9

61.9

0.89

Insurance

87.5%
12.5%

95.9%
4.1%

0.12

91.6%
2.4%

89.2%
10.8%

0.07

100%
0%

94.6%
5.4%

0.10

64.6%
35.4%

60.8%
39.2%

0.74

82.9%
17.1%

74.3%
25.1%

0.26

34.1%
65.9%

31.1%
68.9%

0.81

Commercial
Public aid

Obstetric
Amniocenteses
,l
Chorionic villus samplings
,l
Spontaneous abortions
,l
Elective abortions
,l
Children
,l

sampling chose chorionic

villus sampling
(p = 0.65).
There were no significant differences in demographic or obstetric factors between women
who did and did not choose the test prescribed
by their decision-analytic model. Women whose
decision analyses prescribed amniocentesis but
who chose chorionic villus sampling had had a
prior amniocentesis somewhat more frequently
(10.4% vs 1.6%, p = 0.11) than women who
chose amniocentesis. Women whose decision
analyses prescribed chorionic villus sampling
but who chose amniocentesis assigned significantly lower costs to waiting (2.3 vs 12.5,
p < 0.03), and to late therapeutic abortion (2.7
vs 12.7, p < 0.05), compared with women who
chose chorionic villus sampling. Among women
whose decision analyses prescribed amniocentesis, there were no significant differences in
preference costs between those choosing amniocentesis or chorionic villus sampling. Among
women who chose the test prescribed by their
decision analysis, the choice had been theirs in
57.0%, shared with their physicians in 23.3%,
and made by their physicians in 19.8% of cases.
Among women who did not choose the test

prescribed by their decision analysis, the choice

had been theirs in 44.8%, shared with their
physicians in 39.7%, and made by their
physicians in 15.5% of cases (p = 0.11).

DISCUSSION

We found that the preferences of pregnant

women for the outcomes of prenatal genetic
testing were associated with their choice of
amniocentesis or chorionic villus sampling. The
cost of waiting for genetic diagnosis (the difference in preferences for first- and secondtrimester diagnosis of a normal child), and the
cost of late therapeutic abortion (the difference
in preferences for first- and second-trimester
abortion of an abnormal fetus), were significantly higher among women choosing chorionic
villus sampling than among women choosing
amniocentesis. In addition, the cost of waiting
for diagnosis was an independent predictor of
the choice of chorionic villus sampling in a
multivariate model. In the majority of cases,
womens choices of prenatal test agreed with
the choices of decision analytic models that
incorporated their preferences. Nevertheless, in

1222

PAULS. HECKERLING
et al.

41.6% of cases, the choices of women and their

decision models differed.
Greater utility increments for early vs late
genetic diagnosis, and first- vs second-trimester
therapeutic abortion, led to a choice of chorionic villus sampling in a decision analytic model
[30], and in our patients. Other studies that did
not use formal preference assessment methods
have also shown that testing during the first
trimester [39-411, shorter waiting period for
results [39,40], and type of abortion available
for abnormals [39,40], were the most important factors in the choice of chorionic villus
sampling over amniocentesis. Although risk
of spontaneous abortion was the most important factor in the choice of amniocentesis in
some studies [39,40], we found no difference
in the cost of test-related abortion between
women choosing amniocentesis or chorionic
villus sampling. In addition, although women
choosing amniocentesis are at greater risk
for maternal
morbidity
from therapeutic
abortion [25,29], we found no difference in

the cost of maternal morbidity between the

groups.
Demographic and obstetric factors may also
affect the choice of prenatal genetic test. Some
studies have shown a higher proportion of
minority patients among women choosing
amniocentesis [42], and a higher level of education among women choosing chorionic villus
sampling [42,43], although other studies have
not confirmed these associations [39,44,45]. We
found that women choosing chorionic villus
sampling were somewhat more frequently college educated, professionally employed, and
covered by commercial insurance, compared
with women choosing amniocentesis, but the
differences were not significant. In one study of
women seeking prenatal diagnosis for advanced
maternal age, 36% of those preferring amniocentesis compared with 22% of those preferring
chorionic villus sampling had experienced a
spontaneous abortion in a prior pregnancy [40].
In another study, 25% of women choosing
chorionic villus sampling compared with none

Table 3. Comparison of rating scale preferences, costs, and expected

values among women choosing amniocentesis (AMN) and chorionic
villus sampling (CVS)
Variable

AMN

Preference ratings
Normal child after AMN
Normal child after CVS
Spontaneous abortion
Spontaneous abortion after AMN
Spontaneous abortion after CVS
Spontaneous abortion after AMN
(abortus abnormal)
Spontaneous abortion after CVS
(abortus abnormal)
Therapeutic abortion after AMN
(abortus abnormal)
Therapeutic abortion after CVS
(abortus abnormal)
Therapeutic abortion after AMN
with maternal morbidity
Therapeutic abortion after CVS
with maternal morbidity
Therapeutic abortion after AMN
(abortus normal)
Theraneutic abortion after AMN
(abortus normal)
Therapeutic abortion after CVS
(abortus normal)
Costs
Cost of
Cost of
Cost of
Cost of

waiting
test-related miscarriage
late therapeutic abortion
maternal morbidity

Expected values
Expected value of AMN
Expected value of CVS
A Expected value (AMN - CVS)

cvs

p value

94.1
93.1
28.4
21.9
21.3

93.7
97.1
32.6
32.2
33.7

0.64
0.02
0.31
0.34
0.13

64.9

61.5

0.51

65.9

61.5

0.39

63.7

58.3

0.32

62.1

62.1

0.92

38.8

36.1

0.66

39.4

37.0

0.62

15.9

15.3

0.89

15.9

15.3

0.89

15.7

17.5

0.68

- 1.6
0.83
- 1.6
23.8

4.1
0.41
4.4
23.7

0.0004
0.93
0.01
0.98

91.34
89.80
1.54

90.31
93.96
- 3.64

0.62
0.031
0.0006

Amniocentesis or Chorionic Villus Sampling for Prenatal Testing

1223

Table 4. Comparison of rating scale preferences, costs, and expected values among women choosing
amniocentesis (AMN) and chorionic villus sampling (CVS), stratified by decision maker
Patient/shared
AMN

Variable
Preference ratings
Normal child after AMN
Normal child after CVS
Spontaneous abortion
Spontaneous abortion after AMN
Spontaneous abortion after CVS
Spontaneous abortion after AMN
(abortus abnormal)
Spontaneous abortion after CVS
(abortus abnormal)
Therapeutic abortion after AMN
(abortus abnormal)
Therapeutic abortion after CVS
(abortus abnormal)
Therapeutic abortion after AMN
with maternal morbidity
Therapeutic abortion after CVS
with maternal morbidity
Therapeutic abortion after AMN
(abortus normal)
Therapeutic abortion after CVS
(abortus normal)

cvs

Physician
p value

AMN

cvs

p value

94.5
93.3
28.7
29.7
29.3

93.8
97.7
33.1
31.5
33.4

0.77
0.04
0.32
0.71
0.38

95.8
93.3
29.4
23.2
22.0

93.3
95.8
36.8
39.6
36.7

0.51
0.56
0.47
0.14
0.15

64.7

60.9

0.49

65.5

64.4

0.94

65.9

61.3

0.42

65.5

57.3

0.53

65.3

56.1

0.12

57.9

66.3

0.54

63.4

60.9

0.67

57.6

64.5

0.62

39.7

37.2

0.63

36.8

30.6

0.62

40.8

37.5

0.52

35.1

30.6

0.70

15.2

13.3

0.66

16.6

24.6

0.49

14.9

15.7

0.86

16.8

24.6

0.50

- 1.2
- 0.75
- 1.9
24.1

4.2
1.58
4.8
21.2

0.0008
0.65
0.02
0.62

- 2.6
6.7
- 0.3
21.8

:::
- 1.8
34.8

0.25
0.29
0.68
0.32

91.16
90.20
0.96

90.27
94.02
- 3.75

0.71
0.07
0.002

89.97
91.76
- 1.79

0.54
0.60
0.26

costs
Cost
Cost
Cost
Cost

of
of
of
of

waiting
test-related miscarriage
late therapeutic abortion
maternal morbidity

Expected values
Expected value of AMN
Expected value of CVS
A Expected value (AMN - CVS)

of those choosing amniocentesis, had previously

had an elective abortion (p = 0.014) [42]. These
results suggested that the lower abortion rate
following amniocentesis may have been particularly important to women who had experienced
spontaneous abortion, and that the earlier diagnosis of chorionic villus sampling may have
been important to women who had undergone
elective abortion. However, we, as others [43],
found no differences in spontaneous or elective
abortions among women choosing amniocentesis or chorionic villus sampling.
In our study, women who chose chorionic
villus showed a trend toward having had prior
amniocentesis more frequently than women
choosing amniocentesis (p = 0.07). In addition,
prior amniocentesis emerged as a predictor of
the choice of chorionic villus sampling in a
multivariate model that controlled for patients
preferences. These results suggest that independent of their values for the outcomes of prenatal
testing, some women choose chorionic villus
sampling because of negative experiences with
amniocentesis from a prior pregnancy. How-

92.31
89.26
3.05

ever, others have found that the frequency of

prior amniocentesis was similar among women
choosing amniocentesis and chorionic villus
sampling [42].
Decision analysis is a method for explicitly
defining choices and outcomes under conditions
of uncertainty
[37,46]. Decision analysis
incorporates probabilities and utilities for each
uncertain outcome into a structured tree, and
selects the best choice according to the theory of
maximum expected utility. It therefore allows
the decision maker to trade-off the risks and
benefits that are normally involved in a decision.
Our results show that decision-analytic results
were predictive of womens choices of prenatal
test. The difference in expected values between
chorionic villus sampling and amniocentesis
from each womans decision model was the
most important predictor of choice of test in a
multivariate model that considered each of their
preferences and preference costs separately.
Thus some women in our study appeared to be
able to consider all relevant preferences in
making their choice of prenatal test. Among

1224

PAUL S. HECKERLINGet al.

women whose decision analyses selected amniocentesis, 56.8% had chosen amniocentesis, and
among women whose decision analyses selected
chorionic villus sampling, 63.2% had chosen
chorionic villus sampling, a difference that was
significant (p = 0.05). Thus the majority of
test were
womens choices of prenatal
concordant with those of decision models that
incorporated their preferences. Others have successfully used decision analysis to guide the
choices of pregnant women for amniocentesis or
no prenatal testing [47,48].
Nevertheless, almost 42% of the women in
our study did not choose the prenatal test
prescribed by their decision-analytic model.
There are several reasons why this may have
occurred. One possibility is that these women
considered all relevant outcome values but
weighted them improperly due to inaccurate
estimates of their likelihood of occurrence. The
choice of amniocentesis or chorionic villus
sampling in our decision analytic model was
relatively insensitive to a wide range of outcome
probabilities [30]. Therefore, inaccurate estimates of probabilities are unlikely to entirely
explain the difference between women and decision models. Nevertheless, we cannot exclude
the possibility that womens probability estimates were poorly
calibrated,
or were
confounded with their preferences.
Another possibility is that despite lack of
agreement with an expected value model, these
women made choices consistent with their
preferences. In this regard, it is notable that
women whose decision analysis prescribed
chorionic villus sampling but who chose amniocentesis had significantly lower preference costs
for waiting for genetic diagnosis, and for late
therapeutic abortion, than women who chose
chorionic villus sampling. Indeed, their preference costs for these outcomes were not different
from those of women whose choice of amniocentesis matched that of their decision models
(data not shown). These results suggest that
despite their decision-analytic outcomes, these
women made a choice that was in some sense
consonant with their preferences. However,
there were no differences in preference costs for
patients who decision analysis prescribed amniocentesis and who did and did not choose this
test. Thus the decision making of the latter
group of women cannot be explained on this
basis.
A third possibility is that in cases where test
choice and decision analysis did not agree, the

preferences used to select the test were the

physicians rather than the patients. Because we
did not assess physicians preferences for the
outcomes of prenatal testing, we cannot address
this possibility directly. However, it is notable
that in these cases, physicians either made, or
shared in, the choice of test somewhat more
frequently (p = 0.11) than in the cases where
test choice and decision analysis agreed. Thus it
is possible that it was the physicians, rather
than the patients, preferences that weighed
more heavily in the cases where patients and
decision analyses differed. Studies have shown
that patients and physicians utilities for various patient outcomes may differ considerably
[49, 501. It is also possible that our patients
partners preferences determined the choice of
test, but we have no data to support or refute
this possibility.
Finally, it is possible that factors not captured
by decision-analytic models determined our
patients choices of test. Studies have shown
that patients [51] and physicians [52,53] often
make choices that differ from those of decisionanalytic models based on their utilities. Reasons
for discrepancies between decision making and
expected utility models are probably multifactorial, and include the overweighting of small
probabilities
[54], considerations
of regret
[52,55, 561, perceptions of short- vs long-term
expectation [57], as well as other factors.
There are several limitations to our study.
First, and most important, women provided
preferences for prenatal testing after they had
decided which test to have. Thus it is possible
that their preferences were confounded by stage
of decision making, and served as a post hoc
justification of their choice. The fact that some
women choosing amniocentesis assigned higher
preference ratings to second- as compared to
first-trimester diagnosis of a normal child, and
to second- as compared to first-trimester therapeutic abortion of an abnormal fetus, supports
this possibility. In addition, women presenting
for testing when their pregnancy is too advanced
for chorionic villus sampling may favor amniocentesis because it is their only viable option.
Others have found that up to 71% of women
choosing amniocentesis do so because their
pregnancy has progressed too far for chorionic
villus sampling to be performed [45].
Second, we assessed preferences using linear
rating scales rather than standard reference
gambles, to improve comprehension of the task
[58]. Rating scales have shown intra-rater

Amniocentesis or Chorionic Villus Sampling for Prenatal Testing

[59-611 and test-retest [62] reliabilities comparable to those of standard gambles [62,63], and
have demonstrated acceptable inter-rater reliability [59]. Moderately high correlations
between rating scale preferences and standard
gamble utilities have been found in some studies
[64], although not in others [63]. Although
standard gambles, derived from the axioms of
utility theory [37,65], are the reference method
for measuring health outcomes, and are most
appropriate for use in decision trees, they may
demonstrate internal inconsistencies depending
on the outcomes of the gamble [66].
Finally, we studied women who presented for
prenatal testing for the indication of maternal
age. Our results therefore cannot be generalized
to women who seek testing because of a prior
genetic abnormality. Because of their greater
likelihood of requiring therapeutic abortion,
these women often choose chorionic villus
sampling [28,67], as decision-analytic models
suggest that they should [30].
In conclusion, the preferences of pregnant
women for the outcomes of prenatal testing
were associated with their choice of amniocentesis or chorionic villus sampling. The
choices of prenatal test made by the majority of
women were concordant with those of a decision analytic model that incorporated their
preferences. Therefore, most pregnant women
were able to convert their preferences for
delayed diagnosis, therapeutic abortion, test-related spontaneous abortion, and maternal morbidity, into an appropriate choice of prenatal
test. These results suggest that womens preferences for prenatal testing may be useful for
genetic counseling [47,48], and for qualityadjusting outcomes of clinical trials [68]. However, because many women made choices that
were discordant with their decision-analytic
results, further research into the bases for their
choices is warranted.
Acknowledgement-This
project was supported by grant
No. H.506945Ol Al from the Agency for Health Care Policy
and Research.

REFERENCES
NICHD National Registry for Amniocentesis Study
Group. Midtrimester amniocentesis for prenatal diagnosis: safety and accuracy. JAMA 1976; 236:
1471-1476.
Crandall BF, Howard J, Lebherz TB, Rubinstein L,
Sample WF, Sarti D. Follow-up of 2000 secondtrimester amniocenteses. Obstet Gynecol 1980; 56:
625-628.
Tabor A, Philip J, Madsen M, Bang J, Obel EB,

4.

5.

6.

7.
8.

9.

10.

11.

12.

13.
14.

IS.
16.

1225

Norgaard-Pedersen B. Randomized controlled trial of

genetic amniocentesis in 4606 low-risk women. Laneet
1986; 1: 1287-1293.
Crane JP, Beaver HA, Cheung SW. First trimester
chorionic villus sampling versus mid-trimester genetic
amniocentesis: preliminary results of a controlled
prospective trial. Prenat Diag 1988; 8: 355-366.
Rhoads GG, Jackson LG, Schlesselman SE et al. The
safety and efficacy of chorionic villus sampling for
early prenatal diagnosis. N Engl J Med 1989; 320:
609-617.
Canadian Collaborative CVS-Amniocentesis Clinical
Trial Group. Multicentre randomized clinical trial of
chorionic villus sampling and amniocentesis. Laneet
1989; 1: 1-6.
Golbus MS, Loughman WD, Epstein CJ, Halbasch G,
Stephens JD, Hall BD. Prenatal genetic diagnosis in
3000 amniocenteses. N Engl J Mad 1979; 300: 157-163.
Polani PE, Alberman E, Alexander BJ er al. Sixteen
years experience in counseling, diagnosis, and prenatal
detection in one genetic centre: progress, results, and
problems. J MedCenet 1979; 16: 166-175.
Sauire JA. Nauth L. Ridler MAC. Sutton S. Timberlake C. Prenatal diagnosis and o&ome of pregnancy
in 2036 women investigated by amniocentesis. Hum
Genet 1982; 61: 215-222.
Young SR, Wade RV, Watt GW, Hixson ET, Dennis
EJ. The results of one thousand consecutive prenatal
diagnoses. Am J Obstet Gynecol 1983; 147: 181-188.
Dacus JV, Wilroy RS, Summitt RL et al. Genetic
amniocentesis: a twelve years experience. Am J Med
Genet 1985; 20: 443-452.
Benn PA, Hsu LYF, Carlson A, Tannenbaum HL.
The centralized prenatal genetics screening program of
New York City. III. The first 7000 cases. Am J Med
Genet 1985; 20: 369-384.
Hogge WA, Schonberg SA, Golbus MS. Chorionic
vill;s sampling: experience in the first 1000 cases. Am
J Obstet Gvnecol 1986: 154: 1249-1252.
Sachs ES, jahoda MGJ, Kleijer WJ, Pijpers L, Galjaard H. Impact of first-trimester chromosome, DNA,
and metabolic studies on pregnancies at high genetic
risk: experience with 1000 cases. Am J Med Genet
1988; 29: 293-303.
Jackson L. Prenatal genetic diagnosis by chorionic
villus sampling (CVS). Bern Perinat 1985; 9: 209-218.
Martin AO, Simpson JL, Rosinsky BJ, Elias S. Chorionic villus sampling in continuing pregnancies. II.
Cvtoaenetic reliabilitv. Am J Obstet Cvnecol 1986: 154:
ljS3~1362.

17. Therkelsen AJ, Jensen PKA, Hertz JM, Smidt-Jensen

S, Hahnemann N. Prenatal cytogenetic diagnosis after
transabdominal chorionic villus sampling in the first
trimester. Prenat Diag 1988; 8: 19-3i. _
18. NICHD CVS Studv Grouo. Diagnostic accuracv in
chorionic villus saipling (CVS): iiitial findings from
the U.S. Collaborative Study. Am J Hum Gmet 1988;
43: A242.
P,
19. Metaxotou C, Antsaklis A, Panagiotopoulou
Mavrou A, Aravantinos D, Matsaniotis N. Prenatal
diagnosis from chorionic biopsy samples. Clin Genet
1985; 28: 450.
20. Simoni G, Fraccaro M, Gimelli G, Maggi F, Dagna
Bricarelli F. False-positive and false-negative findings
on chorionic villus sampling.
_ _ Prenat Diag 1987; 7:
671-672.
21. Leschot NJ, Wolf H, Verjaal M et al. Chorionic villi
sampling: cytogenetic and clinical findings in 500 pregna&es.-Br- Med J 1987; 295: 407-410.22. Green JE. Dorfmann A, Jones SL. Bender S, Patton
L, Schulman JD. Chorionic villus sampling: experience
with an initial 940 cases. Obstet Gynecol 1988; 71:
208-2 12.

1226
23.

24.

25.

26.

21.

28.

29.

30.

31.

32.

33.

34.

35.

36.
31.

38.

39.

40.

41.
42.

PAUL S. HECKERLING ef al.

Firth HV, Body PA, Chamberlain
P, MacKenzie IZ,
Lindenbaum,
RH, Huson SM. Severe limb abnormalities after chorion
villus sampling
at 56-66 days
gestation. Lancet 1991; 337: 762-763.
Schloo R, Miny P, Holzgreve W, Horst J, Lenz W.
Distal
limb deficiency
following
chorionic
villus
sampling? Am J Med Genet 1992; 42: 402-413.
Peterson WF, Berry FN, Grace MR, Gulbranson
CL.
Second-trimester
abortion by dilatation
and evacuation: an analysis of 11,747 cases. Ohstet Gynecol 1983;
62: 1855190.
Robinson GE, Garner DM, Olmsted MA, Shime J,
Hutton EM, Crawford
BM. Anxiety reduction after
chorionic villus sampling and genetic amniocentesis.
Am J Obstet Gynecol 1988; 159: 953-956.
Spencer JW, Cox DN. Emotional
responses of pregnant women to chorionic villi sampling or amniocentesis. Am J Obatet Gyneeol 1987; 157: 1155-l 160.
Spencer JW, Cox DN. A comparison
of chorionic villi
sampling
and amniocentesis:
acceptability
of procedure and maternal attachment
to pregnancy. Obstet
Gynecol 1988; 72: 714-718.
Burnhill MS. Reducing the morbidity of vacuum aspiration abortion.
In: Zatuchni GI, Sciarra JJ, Speidel
JJ, Eds Pregnancy Termination: Procedures, Safety,
and New Developments. Hagerstown:
Harper & Row;
1979: 136-148.
Heckerling PS, Verp MS. Amniocentesis
or chorionic
villus sampling for prenatal genetic testing: a decision
analvsis. J Clin Enidemiol 1991: 44: 6577670.
Torrance
GW. -Utility
approach
to measuring
health-related
quality of life. J Chron Dis 1987; 40:
593600.
Hook EB, Cross PK, Schreinemachers
DM. Chromosomal abnormality
rates at amniocentesis
and in live
born infants. JAMA 1983; 249: 2034-2038.
Hook EB, Cross PK, Jackson L, Pergament E, Brambati B. Maternal
age-specific rates of 47, f21 and
other cytogenetic abnormalities
diagnosed in the first
trimester of pregnancy in chorionic villus biopsy specimens: comparison
with rates expected from observation at amniocentesis.
Am J Hum Genet 1988; 42:
797-807.
Hook EB. Evaluation
and projection
of rates of
chromosome
abnormalities
in chorionic villus studies
(CVS). Am J Hum Cenet 1988; 43: A108.
Wilson RD, Kedrick V, Wittman BK, McGillivray
B.
Spontaneous
abortion
and pregnancy
outcome after
normal first-trimester
ultrasound
examination.
Obstet
Gynecol 1986; 67: 352-355.
Gilmore DH, McNay MB. Spontaneous
fetal loss in
early pregnancy.
Lancet 1985; 1: 107.
Weinstein MC, Fineberg HV, Elstein AS, Frazier HS,
Neuhauser
D, Neutra RR, McNeil BJ. Clinical Decision Analysis. Philadelphia, PA: (W. B.) Saunders;
1980.
Engelman L. Stepwise logistic regression.
In: Dixon
WJ, Engelman L, Hill MA, Jennrich MI, Eds BMDP
Statistical Software Manual. Berkeley, CA: University
of California Press: 1988: 941-969.
McGovern
MM, Goldberg JD, Desnick RJ. Acceptability of chorionic villi sampling for prenatal diagnosis. Am J Obstet Gynecol 1986; 155: 25-29.
Lippman A, Perry TB, Mandel S, Cartier L. Chorionic
villi sampling: womens attitudes. Am J Med Genet
1985; 22: 395-401.
Abramsky
L, Rodeck CH. Womens choices for fetal
chromosome
analysis. Prenat Diag 1991; I I: 23-28.
Tunis SL, Golbus MS, Copeland KL, Fine BA,
Rosinsky
BJ, Seely L. Patterns
of mood states
in pregnant
women
undergoing
chorionic
villus
sampling or amniocentesis.
Am J Med Genet 1990; 37:
191-199.

43.

44.

45.

46.
47.

48.

49.

50.

51.

52.

53.

54.

55.
56.

57.
58.

59.

60.

61.

Sjogren B, Uddenberg
N. Prenatal
diagnosis
and
psychological
distress: amniocentesis
or chorionic villus biopsy? Prenat Diag 1989; 9: 477-487.
Caccia N, Johnson
JM, Robinson
GE, Barna T.
Impact of prenatal testing on maternal-fetal
bonding:
chorionic villus sampling versus amniocentesis.
Am J
Obstet Gynecol 1991; 165: 1122-I 125.
Brandenburg
H, Van Der Zwan L, Jahoda
MGJ,
Stijnen T, Wladimiroff
JW. Prenatal
diagnosis
in
advanced
maternal
age. Amniocentesis
or CVS, a
Prenat Diag
patients choice or lack of information?
1991; 11: 6855690.
Pauker SG, Kassirer JP. Decision analysis. N Engl J
Med 1987; 316: 250-258.
Pauker SP, Pauker SG. Prenatal diagnosis: a directive
approach to genetic counseling using decision analysis.
Yale J Biol Med 1977; 50: 2755289.
Pauker SP, Pauker SG. The amniocentesis
decision:
ten years of decision analytic experience. In: EveersKiebooms
G, Cassiman
J-J, Van den Berghe H,
dYdewalle G. Eds Genetic Risk, Risk Perception, and
Decision Making. Birth Defects: Original Article Series
23. New York: Alan R. Liss; 1987: 151-169.
Holmes MM, Rovner DR, Rothert ML, Elstein AS,
Holzman GB, Hoppe RB, Metheny WP, Ravitch MM.
Womens and physicians utilities for health outcomes
in estrogen replacement
therapy. J Gen Intern Med
1987; 2: 1788182.
Boyd NF, Sutherland HJ, Heasman KZ, Tritchler DL,
Cummings
BJ. Whose utilities for decision analysis?
Med Decis Making 1990; IO: 58867.
Pauker SP, Pauker-SG. The amniocentesis decision: an
explicit guide for parents. In: Eustein CJ. Currv CJR.
Packman S, Sherman S, Hall SD, Eds Risk, Commnnication, and Decision Making in Genetic Counseling.
Birth Defects: Original Article Series XV(SC). New
York: Alan R. Liss; 1979; 289-324.
Elstein AS, Holzman GB, Ravitch MM, Metheny WA,
Holmes MM, Hoppe RB, Rothert ML, Rovner DR.
Comparison
of physicians decisions regarding estrogen replacement
therapy for menopausal
women and
decisions derived from a decision analytic model. Am
J Med 1986; 80: 246-258.
Clancy CM, Cebul RD, Williams SV. Guiding individual decisions: a randomized controlled trial of decision
analysis. Am J Med 1988; 84: 2833288.
Tversky A, Kahneman
D. The framing of decisions
and the psychology
of choice. Science 1981; 211:
453-458.
Bell DE. Regret in decision making under uncertainty.
Operations Res 1982; 30: 961-981.
Feinstein AR. The chagrin
factor and qualitative
decision
analysis.
Arch Intern Med 1985; 145:
125771259.
Lopes LL. Decision making in the short run. J Exp
Psycho1 (Human Learn Mem) 1981; 7: 3777385.
dYdewalle G, Evers-Kiebooms
G. Experiments
on
genetic risk perception and decision making: explorative studies. In: Evers-Kiebooms
G, Cassiman J-J, Van
den Berghe H, dYdewalle G, Eds Genetic Risk, Risk
Perception, and Decision Making. Birth Defects: Original Article Series 23. New York: Alan R. Liss; 1987;
209-225.
Patrick
DL, Bush JW, Chen MM. Methods
for
measuring levels of well-being for a health status index.
Health Serv Res 1973; 8: 228-245.
Cadman
D, Goldsmith
C. Construction
of social
value or utility-based
health indices: the usefulness of
factorial experimental design plans. J Chron Dis 1986;
39: 643-651.
Torrance GW. Preferences for health states: a review
of measurement methods. Mead Johnson Symp Perinat
Develop Med 1982; 20: 37-45.

Amniocentesis or Chorionic Villus Sampling for Prenatal Testing

OConnor AM, Boyd NF, Till JE. Influence of elicitation technique, position order and test-retest error
on preferences for alternative cancer drug therapy.
Nnrshsa Reseueh: Science for Oualitv of Care: Proc
Tenth lhhmi Nursing Reaearch~Confknce. Toronto:
University of Toronto; 1985.
Torrance GW. Social preference for health states: an
empirical evaluation of three measurement techniques.
Socioecon Planning Sci 1976; 10: 129-136.
Read JL, Quinn RJ, Berwick DM, Fineberg HV,
Weinstein MC. Preferences for health outcomes: comparison of assessment methods. Med Decis Making
1984; 4: 315-329.
Keeny RL, Raiffa H. Decisiolls with Multiple Objectives: Preferences and Value Tradeoffs. New York:
Wiley, 1976.
Llewellyn-Thomas H, Sutherland HJ, Tibshirani R,
Ciampi A, Till JE, Boyd NF. The measurement of
uatients values in medicine. Med Decis Maklne_ 1982:
2: 449-462.
McCormack MJ, Rylance ME, MacKenzie WE, Newton J. Patients attitudes following chorionic villus
sampling. PrenatWag 1990; 10: 253-255.
Feeny DH, Torrance GW. Incorporating utility-based
quality-of-life
measures in clinical trials: Two
examples. Med Care 1989; 27: Sl90-S204.

1227

chromosomally normal fetus. At birth, the child is

normal.
I________________________l________________________
1
0
100
(4 You decide not to have any prenatal genetic testing.
Your pregnancy ends in a miscarriage.
*______________________--I_________________________
1
63.
0
100
(4 You decide to have an amniocentesis at 4 months. An
64.
amniocentesis can cause a miscarriage, although most
miscarriages that occur after amniocentesis are due to
other causes. Sometime after the test is performed, your
pregnancy ends in a miscarriage.
I________________________l________________~_______
1
65.
0
100
(e) You decide to have a chorionic villus sampling at 2f
66.
months. A chorionic villus sampling can cause a miscarriage, although most miscarriages that occur after
chorionic villus sampling are due to other causes.
Sometime after the test is performed, your pregnancy
67.
ends in a miscarriage.
I________________________l_____-___________________
l
0
100
68.
0-J You decide to have an amniocentesis at 4 months. The
result shows a chromosomally abnormal fetus. While
you are deciding whether or not to have an abortion,
the pregnancy ends in a miscarriage. The fetus is found
to be abnormal, as predicted.
*________________________I_________________________
l
APPENDIX A
0
100
Preference Rating Questionnaire
(iid You decide to have a chorionic villus sampling at 2;
months. The result shows a chromosomally abnormal
This questionnaire is designed to determine your attitudes
fetus. While you are deciding whether or not to have an
towards various possible outcomes of pregnancy related to
abortion, the pregnancy ends in a miscarriage. The fetus
prenatal genetic testing for chromosomal abnormalities.
is found to be abnormal, as predicted.
Amniocentesis and chorionic villus sampling are tests
I________________________I_________________________
*
to evaluate the possibility of a chromosomally-abnormal
0
100
fetus.
01) You decide to have an amniocentesis at 4 months. The
Amniocentesis is performed during the 4th month of
result shows a chromosomally abnormal fetus, and you
pregnancy, at a time when a woman is visibly pregnant and
decide to have an abortion. The aborted fetus is found
may have felt the fetus move. The results are available about
to be abnormal, as predicted.
2 weeks after the test. An abortion performed because of
I_____________________.__l________________________
1
an abnormal amniocentesis result would have to be per0
100
formed after this time, and would possibly require inducing
labor.
(9 You decide to have a chorionic villus sampling at 2f
months. The result shows a chromosomally abnormal
Chorionic villus sampling is performed at 25 months of
fetus, and you decide to have an abortion. The aborted
pregnancy, at a time when a woman is not yet visibly
fetus is found to be abnormal, as predicted.
pregnant and has not felt the fetus move. The results are
l________________________l___-_____________________
1
available about 1 week after the test. An abortion performed
0
100
because of an abnormal chorionic villus sampling result
could be performed before a woman was visibly pregnant or
(3 The same as question (h), but after the abortion you
experience pain and vaginal bleeding, and require hoshad felt the fetus move, and would not require inducing
pitalization and antibiotic therapy for infection.
labor.
I________________________l____-____________________
*
We assume that for you the best outcome would be to give
0
100
birth to a normal child, and that the worst outcome would
be to give birth to a genetically abnormal child. If, on the
04 The same as question (i), but after the abortion you
experience pain and vaginal bleeding, and require hosvisual scales shown below, the best outcome (a normal child)
pitalization and antibiotic therapy for infection.
is worth 100 points, and the worst outcome (an abnormal
*________________________I______________~__________
l
child) is worth 0 points, please indicate the worth of each of
0
100
the following outcomes by marking an x on the appropriate point on the scale:
(1) You decide to have an amniocentesis at 4 months. The
test result shows a chromosomally abnormal fetus, and
You decide to have an amniocentesis. As a result, for
you decide to have an abortion. It turns out that the
4-5 months of your pregnancy, you dont know
aborted fetus was normal, and that the test result was
whether your fetus is normal or abnormal. Two weeks
incorrect.
l________________________I_________________________
1
after the test, the result is available, and shows a
chromosomally normal fetus. At birth, the child is
0
100
normal.
(m) You decide to have a chorionic villus sampling at 2;
l________________________l________________________
l
months. The test result shows a chromosomally abnor0
100
mal fetus, and you decide to have an abortion. It turns
out that the aborted fetus was normal, and that the test
(W You decide to have a chorionic villus sampling. As a
result, for 253 months of your pregnancy, you dont
result was incorrect.
l________________________I_____-____________________
*
know whether your fetus is normal or abnormal. One
week after the test, the result is available, and shows a
0
100

62.

PAULS. HECKERLING
et al.

228

APPENDIX B
Probabilities

Used in the Decision Analyses

Probability

AMN

Spontaneous abortion*
Ages 30-34
Ages 35-39
Ages B 40
Chromosomal abnormalityt

cvs

&
0.028
0.039
0.074

0.033
0.044
0.079

(excess over abnormal liveborn rate)$t

0.007 1 (0.002 1)
0.0089 (0.0027)
0.0113 (0.0035)
0.0142 (0.0044)
0.0178 (0.0056)
0.0224 (0.0069)
0.0283 (0.0087)
0.0356 (0.0108)
0.0448 (0.0134)
0.0564 (0.0165)
0.0710 (0.0204)

0.0085 (0.0035)
0.0108 (0.0046)
0.0139 (0.0061)

Indeterminate test result

0.011

Chromosomal abnormality after an indeterminate test

0.08

False-negative test result

0.004

0.004

False-positive test result

0.0007

0.001

Morbidity after therapeutic abortion

0.014

0.006

Ages <35
36
37
38
39
40
41
42
43
44
245

0.0178
0.0227
0.0291
0.0373
0.0477
0.0611
0.0782
0.1001

(0.0080)
(0.0105)
(0.0136)
(0.0177)
(0.0229)
(0.0297)
(0.0383)
(0.0495)

*Assuming ultrasonic viability at 8-10 weeks gestation. With amniocentesis, 45% of spontaneous
abortions were assumed to occur in the first trimester, and 55% were assumed to occur in the
second and third trimesters. With chorionic villus sampling, 55% of spontaneous abortions were
assumed to occur in the first trimester, and 45% were assumed to occur in the second and third
trimesters.
tRepresents the rates of chromosomal abnormalities at the time of amniocentesis and chorionic villus
sampling 132-341.
$The excess over abnormal liveborn rates represents abnormal fetuses that would have spontaneously
aborted, but were instead terminated by therapeutic abortion because they were detected by
prenatal testing. The probabilities of an abnormal child after a negative test result, and of a normal
abortus after a positive test result, were calculated as Bayesian probability revisions using maternal
age-adjusted abnormal liveborn rates, and the false-negative and false-positive rates of amniocentesis and chorionic villus sampling.