- Email: [email protected]
An economic evaluation of patch testing in the diagnosis and management of allergic contact dermatitis
An Economic Evaluation of Patch and Management of Allergic Rukmini
Rajagopalan,
Testing Contact
in the Diagnosis Dermatitis
Roger T. Anderson, Syam Sarma, John Kallal, Carol Retchin, Joseph F. Fowler, Jr, and Elizabeth F. Sherertz
John
Jones,
Background: A previous retrospective study that patch testing is cost-effective and well by patients. Objective: The objective of this observational prospendy was to show the cost-effectiveness of patch testing in patients suspected of allergic contact dermatitis (ACD) and to determine the order in which different severity groups rank in terms of cost-effectiveness. Methods: This observational study was conducted-patients from IO investigator sites over a period of 1 year. All patients with a suspicion of contact allergy who exhibited at least moderate disease activity were included in the study and were stratified according to disease severity and whether or not they were patch tested. In each severity category, the cost-effectiveness of patch testing was evaluated. Patients who were ruled out for contact allergy by the first 6 months after admission were excluded. A validated dermatology-specific quality of life instrument was administered to all the patients at entrance
into the study and at 6 and 12 months after that. The cost-effectiveness analysis is shown using a decision analysis model. Results: Patch testing was performed on 22% of patients mild disease, 41% of patients with moderate disease, and 50% of patients with severe disease. As a result of changes made in their lifestyle, 66% in the patch-tested group and 51% in the non-patch-tested group reported 75% or more improvement in disease symptoms after 6 months. Early confirmation of diagnosis helped reduce the prediagnosis costs of treatment, which was mostly based on preliminary diagnosis. The greatest quality of life benefits from patch testing, relative to no patch testing, occurred in subjects with recurrent or chronic ACD. Conclusion: Patch testing is most cost-effective and reduces the cost of therapy in patients with severe ACD. Copyright 0 1998 by W.B. Saunders Company
A
testing is a valuable tool to provide an accurate, relatively simple means of diagnosis that allows the physician to initiate appropriate management to alleviate subjects’ suffering without losing precious time.7 In a retrospective study, data were collected on 260 subjects with suspected ACD.8 Results from that study indicated that patch-tested subjects experienced a shorter postdiagnosis duration of the disease than subjects who were not patch tested. This finding was particularly noted in subjects who had more than 2 months’ history ofACD before the diagnosis. Following these results from the retrospective study and on consultation with opinion leaders in contact dermatitis, a prospective study was designed to evaluate the benefits of patch testing in obtaining an early confirmed diagnosis in subjects with a suspicion of contact allergy.
indicated accepted
LLERGIC CONTACT DERMATITIS (ACD) is a common allergic condition affecting the skin, and it presents as an acute or subacute eruption characterized by erythema, vesicles, and scaling after exposure to the contact allergen. The diagnosis of ACD is traditionally based on the subject’s history of symptoms and substance contact, physical examination, and use of patch testing to confirm suspected allergens. Patch testing is a practical tool available to the clinician for confirming the diagnosis of ACD’ caused by common substances. It is indicated as an aid in the diagnosis of ACD, but it is not a substitute for taking the subject’s history or for direct physical examination.2 The overall usefulness of patch testing depends on the experience of the physician interpreting the results3J4 General clinical experience in the use of patch testing assists the physician in defining the clinical relevance of a subject’s reaction to test allergen(s).s In 1995, the National Center ofHealth Statistics survey had recorded more than 697 million physician visits, and out of them about 3.6% (25.0 million) were attributed to dermatology-related causes. Furthermore, 7.6 million visits were for contact dermatitis, atopic dermatitis, and other eczema.6 In 1989, Rietschel showed that patch
American
Journalof
ContactDermatitis,
Vol9,
No 3 (September),
Rexarch
From Glaxo Dermatology, Triangle Park; Bowman
a division of Glaxo Wellcome Inc, Gray School of Medicine, Winston-
Salem; and PharmaResearch Corporation, Wilmington, sig ofLouisville School of Medicine, Louisville, KY
NC; and Univer-
Presented at the Annual Meetings of the Drug Information Association, June 24, 1997, Montreal, Quebec, Canada. Address reprint requests to Rukmini Rajagopalarz, DrPH, MBA, m, Glaxo Dennatoloa, a diuision of Glaxo TVellcome kc, 5 Moore Drive, Research Triangle Park, NC 27709. Copyright 0 I998 by WB. Saunders Comnppay 1046.199Xl98/0903-0003$03.00/O
1998: pp 149-154
149
150
RAJAGOPALAN
OBJECTIVES
No hypothesis
The objectives of this 1%month studywere as follows: 1. To identify patient populations for whom patch testing is cost-effective as a diagnostic tool 2. To determine the cost-effectiveness of patch testing in these populations by collecting data at 6 and 12 months on a. Cost of disease and cost of treatment b. The frequency of patch testing c. The duration of disease, frequency of episodes, and number of offke visits d. Diagnosis and treatment outcomes e. Subjects’ perception of diagnosis and treatment f. Quality of life STUDY DESIGN AND METHODS
observational
study
ment, comes impact assessed
by the proportion
(based on physician positively identified
of subjects
with
confirmed
assessment), the proportion allergen, the rate of subjects
naire
specific quality were collected
of life (DSQL) on investigator’s
management using case report subjects’ medical charts, which
on subjects’
instrument. Additionally, data global assessment and disease form to extract information were the source documents.
DSQL
quality
instrument
of life had items
used for collecting
on the following
ACD. No treatment subjects. A variety
were taken verbatim out of SF-36 instrument.g patch testing on DSQL in patients with ACD
clinician
based on the history
test is left to the discretion
of the
and physical
of the
characteristics
condition; to reduce bias and reflect reality, patients suspected of having a contact dermatitis unrelated to poison ivy or oak were included. Cost-effectiveness studies intervention as possible best reflect produce
performed the real-life
with as little scenario and
results closest to those seen in actual practice. The study was performed in 10 outpatient dermatology
clinics and was conducted for a period of 12 months with an evaluation at the end of 6 and 12 months. Investigators were chosen with various investigators patch
levels ofpractice experience with ACD. Four tested 15% to 30% of the subjects, another
four patch tested 30% to 60% of the subjects, and the remaining two patch tested 90% to 100% of the subjects. To obtain a representative sample ofdermatologists in the United States, we chose
to recruit
two
investigators
from
the North
American
Contact Dermatitis Group, four dermatologists who had considerable experience in patch testing, and four who recently started patch testing. These 10 investigators recruited a total of 567 subjects Sample provide
to acquire data for at least 500 evaluable subjects. size was chosen in the hope that each investigator would evaluable data on at least 50 subjects. The number of
subjects actually enrolled, however, were unevenly distributed across sites. The number of subjects per site ranged from 3 1 to 76. Consequently, some results may vary by site according to this unequal distribution.
from
Method
self-perception, (6) mental, under each of the headings.
in the study protocol. The decision to patch
global
Data were collected using a pharmacoeconomic questionof which one component was a validated dermatology-
data prospectively on the use of the commercially available patch-testing products as diagnostic aids in the management of
specified
diagnosis
of subjects with advised to avoid
was assessed by results of patch testing, investigator’s evaluation, and disease management summary.
(1) physical symptoms, (2) activities activities and functioning, (4) work
use and the outcomes of care provided to ACD subjects during the 12-month follow-up period were collected and analyzed as
outthe was
allergenic substances, and the rate of subjects who were benefited by avoidance of possible allergens. Disease management
cost-effectiveness of patch testing in the diagnosis and treatment of moderate-to-severe ACD. The study was designed to collect
plan or diagnostic method was assigned to of data that represent health care resource
calcula-
and pharmacoeconomic assessments. Secondary were the quality-of-life measures used to estimate of ACD on the subjects’ quality of life. Diagnosis
Quality ofl$ to evaluate
and no power
evaluated for outcomes related to ACD and patch testing. The primary outcome measures included diagnosis, disease manage-
The validated
was a multicenter,
was planned
tion for sample size was used for comparison of cost-effectiveness. Prospectively collected data for all eligible subjects were
Study Design This
testing
ET AL
data
dimensions:
of daily living, (3) social or school functioning, (5)
(7) vitality, The mental
and (8) global items and vitality domains The impact is discussed
of in
greater detail elsewhere.1° In evaluating the cost-effectiveness of patch testing, improvement in quality of life from baseline was the primary variable of effectiveness used in the decision analysis model
to calculate
the cost-effectiveness
ratio.
Statistical Method To assess the associated clinical,
pharmacoeconomic,
impact
of patch
and quality-of-life
testing
assessments,
on the sub-
jects were divided into two groups: (1) patch tested-at screening or before or during the 6-month (or discontinuation) visit, and (2) non-patch tested-those who do not qualify as patch tested. sented
For statistical by patch-testing
summaries and analyses, data are prestatus. All statistical tests are two-sided
and performed at 5% significance level. Sample size was chosen expecting that each investigator would provide evaluable data on at least 50 subjects. An evaluable
patch-tested
subject
should
patch tested at any time between screening evaluation, and an evaluable non-patch-tested not have been ruled out as having ACD between &month evaluation. All pleted a pharmacoeconomic
have
been
and 6-month subject should screening and
evaluable subjects should have com(PE) questionnaire at the 6-month
visit and a DSQL questionnaire at the screening and at the 6-monthvisit. Two-sample t-tests were performed on the qualityof-life changes from baseline to determine whether (1) patch testing appeared to result in better outcomes and (2) whether prognostic outcome with patch testing is different among diagnostic subgroups.
PATCH
TESTING
IN ALLERGIC
CONTACT
DERMATITIS
Cost-Efktiueness Ana&si.r Total costs per month (prescription costs and cost of physician visits) from the day of confirmed diagnosis were compared over the period of the study between the subjects who had confirmed diagnosis within 8 days of screening and those who were confirmed after 8 days of screening. These costs were compared with the prediagnosis costs. When subjects were not patch tested in the beginning of the study, delay in diagnosis and identification of the allergen contributed to the worsening of the disease and chronicity. When such subjects did not respond to treatment and when they were patch tested at a later date, they crossed into the patch-tested group, and such subjects were liable to increase the postdiagnosis costs for the patch-tested group. Based on this rationale, the patch-tested group was further defined as only those subjects who were patch tested within the first 8 days after screening.
Decision Ana&s
Model
Additionally, a decision analysis model was created to identify the rank order of cost-effectiveness of patch testing for the groups of subjects who were studied. For purposes of the decision model, the total costs were defined as the sum of drug therapy costs without generic substitutions and physician office visits (patch-test assessment visits were not included). Two models, one for analyzing the costs and the other for analyzing the changes in quality-of-life scores, were constructed. The results of the two models were used to create a cost-effectiveness ratio. Total costs were defined as costs of drug therapywithout generic substitution and office visits unrelated to patch-test assessments. The patch-test costs and related visits were not included in the total costs because of the variety of patch-testing packages (both commercial and noncommercial) being used and the wide range of their prices (approximately $96 to $520 per application). Inasmuch as the decision analytic model is sufficiently robust over a nominal range of costs, it is sensitive to the wide disparity in prices of these products. Therefore the cost of patch testing is not included in the total costs.
RESULTS Population
Characteristics
For this study, 567 subjects were recruited. A total of 136 subjects, including the 124 subjects who discontinued, were excluded from the 6-month analysis. A total of 121 subjects were nonevaluable for reasons such as lack of quality-of-life or PE data, ACD being ruled out, or 6-month visit being outside 6 months _t 1 month window. Fifteen subjects were excluded because of extraneous protocol violations, and improper administration of quality-of-life instrument. The remaining 431 subjects (76%) were considered evaluable-186 patch tested and 245 non-patch tested. The overall population was 71% female and 87% white. The mean age was 45.5 years. For patch-tested and non-patch-tested subject groups,
151
Table 1. Diagnostic With
Effectiveness
Suspected
Allergic
of Patch
Testing
Contact
Dermatitis Patch Tested
Total no. of subjects Subjects with confirmed Median time to confirmed Subjects advised to avoid
186 diagnosis diagnosis sub-
stances based on the identification of allergen Subjects
who
made
changes
as a result
8 d
Non-Patch Tested 245 168(69%) 175 d
141 (76%)
106(43%)
125(67%)
114(47%)
in life-
style 75% or more
164(88%)
in Subjects
improvement reported of changes in lifestyle
66%
51%
the percentage of time with bothersome signs or symptoms before enrollment averaged 64% and 65%, respectively. The prestudy dermatitis duration was 10 and 12 months (median), respectively.‘O The most common diagnoses reported were (1) acute dermatitis with pattern of contact dermatitis-sharp cutoffs and patterns combined with ACD, (2) chronic dermatitis or recurring dermatitis with questionable (unknown) origin, and (3) chronic or recurrent hand eczema. Evaluable subjects were stratified by disease severity based on the subjects’ perception of average severity of their skin problem over the month before screening. In the assessment of quality of life, subjects were asked to rate the severity on a scale from 1 (very mild) to 10 (very severe). A score from 1 to 3 was considered mild, and 49 (11.4%) subjects rated their disease mild; 180 (42%) subjects scored their disease moderate with a rating of 4 to 7; and a score of 8 to 10 was considered severe, and that is how the rest of the 200 (46.6%) subjects rated themselves.1° Cost-Effectiveness Diagnosis. There was a significant difference between the patch-tested and non-patch-tested groups in terms of the time to obtain a confirmed diagnosis* (medians = 8 days and 175 days, P < .OOl) and a significant difference in the ratio of subjects that had a confirmed diagnosis (88% and 69%, P < .OOl) (Table 1). These findings show the effectiveness of patch testing as a tool to diagnose the cause of contact dermatitis and thereby reduce *For non-patch-tested subjects, physicians used their routine customary methods to confirm diagnosis, which may have included clinical morphology, history, response to treatment, biopsy, or other factors.
152
RAJAGOPALAN
Table 2. Comparison
of Mean
Change
Non-Patch-Tested
in Score
for Dermatology-Specific
Groups
Within
Recurrent
Non-Patch-Tested DSQL Scale Physical
symptoms
Activities
of daily
N
Quality Allergic
of Life (DSQL)
Contact
Dermatitis
Group
Mean Change
Scales
Patch-Tested
and
(N = 169)
Patch-Tested SE
Between
ET AL
Group
Mean Change
SE
PValue
97 96
-0.58 -0.09
0.10 0.09
69 68
-1.23 -0.44
0.12 0.10
,001 ,001
Social functioning Mental health
98 94
-0.30 -0.18
0.10 0.09
69
-0.63 -0.45
0.12
.039
Vitality Work/school Self-perception
functioning
93 64
-0.13 -0.15
0.11 0.08
67 66 33
-0.37 -0.17
0.10 0.12 0.12
,052 ,137 ,916
Self-perceived
severity
97 98
-0.47 -1.85
0.11 0.33
67 69
-0.77 -3.59
0.13 0.36
.081 .OOl
living
(global
item)
the duration of disease. In the patch-tested group, the investigators advised the subjects to avoid substances based on the identification of the allergen in 76% of the subjects, whereas only 43% in the non-patch-tested subjects received that advice. As a result of changes made in their lifestyle, 66% in the patch-tested group and 51% in the non-patchtested group reported 75% or more improvement in their disease symptoms after 6 months. Treatment. Subjects who were diagnosed within 8 days of screening incurred lower medical expenses (prescriptions and office visits), an average of $99 per month in the prediagnosis period, compared with those who were diagnosed more than 8 days after screening. Once the diagnosis was confirmed, both groups had incurred fairly similar expenses (ranging from $15 to $60 per month). When total costs were comparedwithin the severity groups, however, the differences between patchtested and non-patch-tested groups varied considerably. Particularly in subjects with severe dermatitis, costs were much less for the patch-tested group compared with the non-patch-tested group. Early confirmation of diagnosis helped reduce the prediagnosis costs of treatment, which was mostly based on preliminary diagnosis. Quality of Lz@. Use of patch testing among subjects with moderate to severe ACD is associated Table 3. Comparison
of Mean
Change
in Score Chronic
for DSQL Scales Allergic Contact Non-Patch-Tested
DSQL Scale
N
Physical symptoms Activities of daily living Social functioning
74 72
health
75 74 73
Work/school functioning Self-perception Self-perceived severity
47 74 75
Mental Vitality
(global
item)
Mean Change
with greater quality-of-life benefit at 6-month evaluation than when patch testing was not used.*OThis result is obtained despite the fact that those not patch tested, as a group, have less severe ACD than those patch tested. The greatest quality-of-life benefit from patch testing, relative to non-patch testing, occurred in subjects with recurrent or chronic ACD (Tables 2-4). Less benefit was found for subjects with mild or moderate disease. DECISION
ANALYSIS
The calculation of expected values of costs and effects in terms of change in quality of life using the decision analysis model are in Tables 5 and 6. Actual decision analysis model is presented elsewhere.” The subjects were first classified under low-utility and high-utility clinics in terms of patch testing and further subclassified by the selfreported severity of the disease. For each of the subclasses, the total costs were evaluated for the decision alternatives-patch test within 8 days or not. Regardless of disease severity, patch-tested subjects showed a greater than 75% rate of confirmed diagnosis within 8 days. In both groups of clinics, the “severe” group of subjects showed the highest degree of costeffectiveness for the patch testing. In the highutility clinics, the “moderate” group of subjects had Between Patch-Tested Dermatitis (N = 139) Group
and Non-Patch-Tested
Patch-Tested N
Mean Change
0.11 0.10
62 56
-0.06 -0.01
0.10 0.09 0.11
60 60
-0.98 -0.56 -0.54
-0.01 -0.21 -2.09
0.05 0.12 0.37
63 30 56 61
Groups
Within
Group
SE
-0.71 -0.18 -0.19
MODEL
SE
P VallX
0.16 0.11
,157 ,009
-0.40 -0.31
0.14 0.12 0.14
,034 ,026 .I01
-0.45 -0.86 -3.18
0.12 0.17 0.44
,001 .002 .050
PATCH
TESTING
IN ALLERGIC
Table 4. Comparison
CONTACT
of Mean
DERMATITIS
Change
in Score With
153
at 12 Months
Chronic
for DSQL
Allergic
Contact
Scales
Between
Dermatitis
Early
Patch Tested Within 7 Days of Screening DSQL Scale
Physical Activities
N
symptoms of daily living
Mean Change
Versus
Late Patch-Tested
All Others Patch Tested
SE
N
Mean Change
SE
PValue
,001 ,016
58 58
-0.74 -0.43
0.14 0.13
36 37
-0.01 -0.03
0.14 0.12
Social functioning Mental health
58 58
-0.75 -0.50
0.14 0.12
37 38
-0.01 -0.02
0.13 0.17
Vitality Work/school
58 38
-0.21 -0.55
0.16 0.15
38 21
-0.32 0.15
0.16 0.13
56 60
PO.88 -3.83
0.16 0.36
37 37
-0.19 -1.91
0.15 0.52
functioning
Self-perception Self-perceived
severity
(global
item)
no difference in the cost-effectiveness between the decisions of performing and not performing patch testing within 8 days. The “mild” group of subjects, however, showed higher cost-effectiveness than the “moderate” group. Therefore, patch testing is shown to be definitely cost-effective in a descending order for subjects with severe ACD and then those with mild ACD; in moderate ACD, there is an indifference, meaning that the physician could decide either way. This can be explained by the experience and expertise of the dermatologists in the highutility clinics who can be intuitive in counseling and treating their subjects even with mild ACD. In low-utility clinics, however, there is no cost-effectiveness shown for patch testing in moderate and mild ACD. DISCUSSION Utility of patch testing to the patients was shown in the retrospective study cited earlier.8 Utility of patch testing to the physician is evident from the higher probability of a confirmed diagnosis and the shorter time to a confirmed diagnosis. Table 5. Expected Cost and Effect* and Cost-Effectiveness Ratio in 1 Year for Clinics That Patch Tested <50% of Their Subjects Decision to Patch Early
Severity Groupt
Mild
PT 58 days No PT c8 days
Moderate Severe
Expected cost* ($)
PT 58 days No PT 58 days PT 58 days No PT 58 days
*Expected
Table 6. Expected Cost and Effect* Ratio in 1 Year for Clinics That of Their
cost per Unit Change in QOL ($1
Severity Groupt
669.76 1,070.76
-0.56 -0.10
1.196.00 10,708.OO
999.42 905.51 999.45
-2.00 -2.05 -3.31
500.00 442.00
1,689.18
-1.77
960.00
Mild
1,431.oo 832.00
PTs8d No PT 58 d
Moderate
692.00 800.77
-1.93 -6.61
359.00 121.00
PT 58 d NoPTs8d PTc8d
983.64
-2.69
366.00
severe before
their skin problem screening. Severity
are based
*Expected abilities
perception
of
how
has been on average over the month is derived from QOL assessments.
QOL, quality
No PT 58 d
on prob-
of life; PT, patch
test.
severe before
Subjects Cost per Unit Change in QOL ($1
2,280.OO
Severe
and Cost-Effectiveness Patch Tested ~50%
Expected Effect* (A QOL)
-0.68 -2.21
effect
,002
Expected cost* I$)
-0.48
cost and expected
,001
Decision to Patch Early
972.87 1,839.27
from the decision analysis model. ISeverity is based on subjects’
,001 ,021 ,648 ,001
Further, the longer the delay in confirming the diagnosis, the longer the subjects had to undergo symptomatic disease activity; the earlier the diagnosis was confirmed, the earlier they could have disease activity alleviated by avoiding the causative substances. Further, the costs of treating the disease without a firm diagnosis can be avoided. Clinical practice patterns also affected qualityof-life outcomes. Patch testing fewer than one half of such subjects classified as having chronic or recurrent ACD would leave many of those not patch tested with little or no quality-of-life improvement over 6 months. Increased use of patch testing may result in better quality-of-life outcomes across subjects. Data also show, however, that patch testing all, or nearly all, such subjects probably reflects overutilization from a quality-of-life effectiveness standpoint because a proportion of subjects (~50%) with chronic or recurrent ACD in some dermatology practices apparently can be selected to have good prognoses without patch testing. Quality-oflife benefits from patch testing appear to be moderated by subject selection processes, which guide
1,094.23
abilities
Abbreviations:
Expected Effect” (A QOL)
Subjects
(N = 139)
cost and expected
effect
are based
302.00
on prob-
from the decision analysis model. tseverity is based on subjects’ perception of how their skin problem has been on average over the month screening. Severity is derived from QOL assessments. Abbreviations: QOL, quality of life; PT, patch test.
RAJAGOPALAN
154
treatment allocation, and that at some point between 50% and 100% utilization, optimal effectiveness is reached. When quality-of-life improvements were compared between chronic ACD subjects patch tested in the first week of the study and chronic ACD subjects who were patch tested later in the study period, the former group showed a remarkable improvement superior to the latter, and the differences were statistically significant. This shows the importance of early confirmed diagnosis. This difference in quality-of-life improvement also influences differences in frequency of visits, number of prescriptions, and other direct and indirect costs. From the perspective of the audience, who are traditionally used to randomized clinical trials, this design of an observational study might seem to introduce bias because the option to patch test is totally based on the discretion of the dermatologist. Because the purpose of the study was to describe the use and outcomes of patch testing in actual practice, an unobtrusive observational study was most appropriate. The absence of randomization should be taken into consideration, however, when the comparison between patch-tested and nonpatch-tested groups is made. Another point is that 431 (76% of the enrolled) subjects were evaluable with PE data. It is worth mentioning that in an observational study, which is almost like a survey, a 50% response is considered quite impressive.
tional cost involved. Patch testing also helps confirm diagnosis of ACD at a higher rate and at an earlier time point. More patch-tested subjects than non-patch-tested subjects were appropriately advised to avoid a substance to aid in management of ACD. At the end of 6 months, a larger proportion of patch-tested subjects reported better than or equal to 75% improvement in their symptoms compared with the non-patch-tested subjects. By confirming the diagnosis earlier, the prediagnosis costs on prescriptions and office visits were reduced; the postdiagnosis costs on prescriptions and office visits were comparatively less for the subjects diagnosed within the first 8 days of the screening than those diagnosed later. Overall, the improvement in quality of life is greater in subjects who had their diagnosis confirmed earlier with patch testing than those with no patch testing or late patch testing. On the whole, the data showed clearly the importance of early confirmation of diagnosis of ACD, identifying and avoiding the causative allergens, and treatment and the importance of patch testing for early confirmation of diagnosis. ACKNOWLEDGMENTS The
authors
acknowledge
W. Loss, MD,
search
Patch testing has been considered worthwhile by subjects despite the discomfort, time, and addi-
the valuable
contributions
of
all the investigators-William Abramovits, MD, Ponciano D. Cruz, MD, Zoe Draelos, MD, Joseph F. Fowler, Jr., MD, Robert David
M. Pariser,
MD, Joel Shavin, MD, Schlessinger, MD-and
CONCLUSION
ET AL
Corporation The
Wellcome
toward
MD,
FACP,
Marvin
Rapaport,
Elizabeth F. Sherertz, MD, and Joel the clinical monitors from PharmaRethe success
of this study.
study was conducted with a grant Inc, Research Triangle Park, NC.
from
Glaxo
REFERENCES 1. Fisher 36:451, history
1985 2. Kieffer and patch
AA: The use and abuse of patch
testing.
Bethesda, M: Nickel sensitivity: test reaction. Contact
son between 50,1994 bility nous
A, Seidenari
two application
S: Patch
methods.
tact Dermatitis Group. 1994 5. Ruhnek-Forsbeck
J Am
Acad
M,
Fischer
for Health
between 5:398-401,
contact
test reading:
a compari-
Dermatol21:812-815, 1995 8. RajagopalanR, K&al@,
Contact
4. Brasch J, Henseler T, Aberer of patch tests: a multicenter left-versus-right-sided patch tests
Center
Dermatitis
30:49-
Statistics:
Advance
data.
MD, 1995 7. Rietschel I& Human and economic impact of allergic dermatitis and the role of patch testing. J Am Acad
relationship Dermatitis
1979. 3. Motolese
6. National
Cutis
tive evaluation allergic contact
of patch dermatitis.
FowlerJF,
et al: Aretrospec-
testing in subjects diagnosed Cutis 57:360-364, 1996
with
W, et al: Reproducistudy of synchroby the German Con-
9. Ware JE Jr: SF-36 Health Survey. The Health Institute. Boston, MA, New England Medical Center, 1993 10. Rajagopalan R, Anderson RT: Impact of patch test-
Dermatol
ing on dermatology-specific lergic contact dermatitis.
31:584-591,
T, Meding
Comparative multi-center study with TRUE Chamber patch test methods in eight Swedish DefmVenereol (Stockh) 68:123-128,1988
B, et al:
Test and hospitals.
Finn Acta
quality of life in patients with alAm J Contact Dermat 8:215-221, 1997
11, Rajagopalan R, Anderson RT, Sarma S, et al: Use of decision analysis modeling in economic evaluation of patch testing in allergiccontact dermatitis. Submitted toPharmacoEconomics,
1997
Rajagopalan,
Testing Contact
in the Diagnosis Dermatitis
Roger T. Anderson, Syam Sarma, John Kallal, Carol Retchin, Joseph F. Fowler, Jr, and Elizabeth F. Sherertz
John
Jones,
Background: A previous retrospective study that patch testing is cost-effective and well by patients. Objective: The objective of this observational prospendy was to show the cost-effectiveness of patch testing in patients suspected of allergic contact dermatitis (ACD) and to determine the order in which different severity groups rank in terms of cost-effectiveness. Methods: This observational study was conducted-patients from IO investigator sites over a period of 1 year. All patients with a suspicion of contact allergy who exhibited at least moderate disease activity were included in the study and were stratified according to disease severity and whether or not they were patch tested. In each severity category, the cost-effectiveness of patch testing was evaluated. Patients who were ruled out for contact allergy by the first 6 months after admission were excluded. A validated dermatology-specific quality of life instrument was administered to all the patients at entrance
into the study and at 6 and 12 months after that. The cost-effectiveness analysis is shown using a decision analysis model. Results: Patch testing was performed on 22% of patients mild disease, 41% of patients with moderate disease, and 50% of patients with severe disease. As a result of changes made in their lifestyle, 66% in the patch-tested group and 51% in the non-patch-tested group reported 75% or more improvement in disease symptoms after 6 months. Early confirmation of diagnosis helped reduce the prediagnosis costs of treatment, which was mostly based on preliminary diagnosis. The greatest quality of life benefits from patch testing, relative to no patch testing, occurred in subjects with recurrent or chronic ACD. Conclusion: Patch testing is most cost-effective and reduces the cost of therapy in patients with severe ACD. Copyright 0 1998 by W.B. Saunders Company
A
testing is a valuable tool to provide an accurate, relatively simple means of diagnosis that allows the physician to initiate appropriate management to alleviate subjects’ suffering without losing precious time.7 In a retrospective study, data were collected on 260 subjects with suspected ACD.8 Results from that study indicated that patch-tested subjects experienced a shorter postdiagnosis duration of the disease than subjects who were not patch tested. This finding was particularly noted in subjects who had more than 2 months’ history ofACD before the diagnosis. Following these results from the retrospective study and on consultation with opinion leaders in contact dermatitis, a prospective study was designed to evaluate the benefits of patch testing in obtaining an early confirmed diagnosis in subjects with a suspicion of contact allergy.
indicated accepted
LLERGIC CONTACT DERMATITIS (ACD) is a common allergic condition affecting the skin, and it presents as an acute or subacute eruption characterized by erythema, vesicles, and scaling after exposure to the contact allergen. The diagnosis of ACD is traditionally based on the subject’s history of symptoms and substance contact, physical examination, and use of patch testing to confirm suspected allergens. Patch testing is a practical tool available to the clinician for confirming the diagnosis of ACD’ caused by common substances. It is indicated as an aid in the diagnosis of ACD, but it is not a substitute for taking the subject’s history or for direct physical examination.2 The overall usefulness of patch testing depends on the experience of the physician interpreting the results3J4 General clinical experience in the use of patch testing assists the physician in defining the clinical relevance of a subject’s reaction to test allergen(s).s In 1995, the National Center ofHealth Statistics survey had recorded more than 697 million physician visits, and out of them about 3.6% (25.0 million) were attributed to dermatology-related causes. Furthermore, 7.6 million visits were for contact dermatitis, atopic dermatitis, and other eczema.6 In 1989, Rietschel showed that patch
American
Journalof
ContactDermatitis,
Vol9,
No 3 (September),
Rexarch
From Glaxo Dermatology, Triangle Park; Bowman
a division of Glaxo Wellcome Inc, Gray School of Medicine, Winston-
Salem; and PharmaResearch Corporation, Wilmington, sig ofLouisville School of Medicine, Louisville, KY
NC; and Univer-
Presented at the Annual Meetings of the Drug Information Association, June 24, 1997, Montreal, Quebec, Canada. Address reprint requests to Rukmini Rajagopalarz, DrPH, MBA, m, Glaxo Dennatoloa, a diuision of Glaxo TVellcome kc, 5 Moore Drive, Research Triangle Park, NC 27709. Copyright 0 I998 by WB. Saunders Comnppay 1046.199Xl98/0903-0003$03.00/O
1998: pp 149-154
149
150
RAJAGOPALAN
OBJECTIVES
No hypothesis
The objectives of this 1%month studywere as follows: 1. To identify patient populations for whom patch testing is cost-effective as a diagnostic tool 2. To determine the cost-effectiveness of patch testing in these populations by collecting data at 6 and 12 months on a. Cost of disease and cost of treatment b. The frequency of patch testing c. The duration of disease, frequency of episodes, and number of offke visits d. Diagnosis and treatment outcomes e. Subjects’ perception of diagnosis and treatment f. Quality of life STUDY DESIGN AND METHODS
observational
study
ment, comes impact assessed
by the proportion
(based on physician positively identified
of subjects
with
confirmed
assessment), the proportion allergen, the rate of subjects
naire
specific quality were collected
of life (DSQL) on investigator’s
management using case report subjects’ medical charts, which
on subjects’
instrument. Additionally, data global assessment and disease form to extract information were the source documents.
DSQL
quality
instrument
of life had items
used for collecting
on the following
ACD. No treatment subjects. A variety
were taken verbatim out of SF-36 instrument.g patch testing on DSQL in patients with ACD
clinician
based on the history
test is left to the discretion
of the
and physical
of the
characteristics
condition; to reduce bias and reflect reality, patients suspected of having a contact dermatitis unrelated to poison ivy or oak were included. Cost-effectiveness studies intervention as possible best reflect produce
performed the real-life
with as little scenario and
results closest to those seen in actual practice. The study was performed in 10 outpatient dermatology
clinics and was conducted for a period of 12 months with an evaluation at the end of 6 and 12 months. Investigators were chosen with various investigators patch
levels ofpractice experience with ACD. Four tested 15% to 30% of the subjects, another
four patch tested 30% to 60% of the subjects, and the remaining two patch tested 90% to 100% of the subjects. To obtain a representative sample ofdermatologists in the United States, we chose
to recruit
two
investigators
from
the North
American
Contact Dermatitis Group, four dermatologists who had considerable experience in patch testing, and four who recently started patch testing. These 10 investigators recruited a total of 567 subjects Sample provide
to acquire data for at least 500 evaluable subjects. size was chosen in the hope that each investigator would evaluable data on at least 50 subjects. The number of
subjects actually enrolled, however, were unevenly distributed across sites. The number of subjects per site ranged from 3 1 to 76. Consequently, some results may vary by site according to this unequal distribution.
from
Method
self-perception, (6) mental, under each of the headings.
in the study protocol. The decision to patch
global
Data were collected using a pharmacoeconomic questionof which one component was a validated dermatology-
data prospectively on the use of the commercially available patch-testing products as diagnostic aids in the management of
specified
diagnosis
of subjects with advised to avoid
was assessed by results of patch testing, investigator’s evaluation, and disease management summary.
(1) physical symptoms, (2) activities activities and functioning, (4) work
use and the outcomes of care provided to ACD subjects during the 12-month follow-up period were collected and analyzed as
outthe was
allergenic substances, and the rate of subjects who were benefited by avoidance of possible allergens. Disease management
cost-effectiveness of patch testing in the diagnosis and treatment of moderate-to-severe ACD. The study was designed to collect
plan or diagnostic method was assigned to of data that represent health care resource
calcula-
and pharmacoeconomic assessments. Secondary were the quality-of-life measures used to estimate of ACD on the subjects’ quality of life. Diagnosis
Quality ofl$ to evaluate
and no power
evaluated for outcomes related to ACD and patch testing. The primary outcome measures included diagnosis, disease manage-
The validated
was a multicenter,
was planned
tion for sample size was used for comparison of cost-effectiveness. Prospectively collected data for all eligible subjects were
Study Design This
testing
ET AL
data
dimensions:
of daily living, (3) social or school functioning, (5)
(7) vitality, The mental
and (8) global items and vitality domains The impact is discussed
of in
greater detail elsewhere.1° In evaluating the cost-effectiveness of patch testing, improvement in quality of life from baseline was the primary variable of effectiveness used in the decision analysis model
to calculate
the cost-effectiveness
ratio.
Statistical Method To assess the associated clinical,
pharmacoeconomic,
impact
of patch
and quality-of-life
testing
assessments,
on the sub-
jects were divided into two groups: (1) patch tested-at screening or before or during the 6-month (or discontinuation) visit, and (2) non-patch tested-those who do not qualify as patch tested. sented
For statistical by patch-testing
summaries and analyses, data are prestatus. All statistical tests are two-sided
and performed at 5% significance level. Sample size was chosen expecting that each investigator would provide evaluable data on at least 50 subjects. An evaluable
patch-tested
subject
should
patch tested at any time between screening evaluation, and an evaluable non-patch-tested not have been ruled out as having ACD between &month evaluation. All pleted a pharmacoeconomic
have
been
and 6-month subject should screening and
evaluable subjects should have com(PE) questionnaire at the 6-month
visit and a DSQL questionnaire at the screening and at the 6-monthvisit. Two-sample t-tests were performed on the qualityof-life changes from baseline to determine whether (1) patch testing appeared to result in better outcomes and (2) whether prognostic outcome with patch testing is different among diagnostic subgroups.
PATCH
TESTING
IN ALLERGIC
CONTACT
DERMATITIS
Cost-Efktiueness Ana&si.r Total costs per month (prescription costs and cost of physician visits) from the day of confirmed diagnosis were compared over the period of the study between the subjects who had confirmed diagnosis within 8 days of screening and those who were confirmed after 8 days of screening. These costs were compared with the prediagnosis costs. When subjects were not patch tested in the beginning of the study, delay in diagnosis and identification of the allergen contributed to the worsening of the disease and chronicity. When such subjects did not respond to treatment and when they were patch tested at a later date, they crossed into the patch-tested group, and such subjects were liable to increase the postdiagnosis costs for the patch-tested group. Based on this rationale, the patch-tested group was further defined as only those subjects who were patch tested within the first 8 days after screening.
Decision Ana&s
Model
Additionally, a decision analysis model was created to identify the rank order of cost-effectiveness of patch testing for the groups of subjects who were studied. For purposes of the decision model, the total costs were defined as the sum of drug therapy costs without generic substitutions and physician office visits (patch-test assessment visits were not included). Two models, one for analyzing the costs and the other for analyzing the changes in quality-of-life scores, were constructed. The results of the two models were used to create a cost-effectiveness ratio. Total costs were defined as costs of drug therapywithout generic substitution and office visits unrelated to patch-test assessments. The patch-test costs and related visits were not included in the total costs because of the variety of patch-testing packages (both commercial and noncommercial) being used and the wide range of their prices (approximately $96 to $520 per application). Inasmuch as the decision analytic model is sufficiently robust over a nominal range of costs, it is sensitive to the wide disparity in prices of these products. Therefore the cost of patch testing is not included in the total costs.
RESULTS Population
Characteristics
For this study, 567 subjects were recruited. A total of 136 subjects, including the 124 subjects who discontinued, were excluded from the 6-month analysis. A total of 121 subjects were nonevaluable for reasons such as lack of quality-of-life or PE data, ACD being ruled out, or 6-month visit being outside 6 months _t 1 month window. Fifteen subjects were excluded because of extraneous protocol violations, and improper administration of quality-of-life instrument. The remaining 431 subjects (76%) were considered evaluable-186 patch tested and 245 non-patch tested. The overall population was 71% female and 87% white. The mean age was 45.5 years. For patch-tested and non-patch-tested subject groups,
151
Table 1. Diagnostic With
Effectiveness
Suspected
Allergic
of Patch
Testing
Contact
Dermatitis Patch Tested
Total no. of subjects Subjects with confirmed Median time to confirmed Subjects advised to avoid
186 diagnosis diagnosis sub-
stances based on the identification of allergen Subjects
who
made
changes
as a result
8 d
Non-Patch Tested 245 168(69%) 175 d
141 (76%)
106(43%)
125(67%)
114(47%)
in life-
style 75% or more
164(88%)
in Subjects
improvement reported of changes in lifestyle
66%
51%
the percentage of time with bothersome signs or symptoms before enrollment averaged 64% and 65%, respectively. The prestudy dermatitis duration was 10 and 12 months (median), respectively.‘O The most common diagnoses reported were (1) acute dermatitis with pattern of contact dermatitis-sharp cutoffs and patterns combined with ACD, (2) chronic dermatitis or recurring dermatitis with questionable (unknown) origin, and (3) chronic or recurrent hand eczema. Evaluable subjects were stratified by disease severity based on the subjects’ perception of average severity of their skin problem over the month before screening. In the assessment of quality of life, subjects were asked to rate the severity on a scale from 1 (very mild) to 10 (very severe). A score from 1 to 3 was considered mild, and 49 (11.4%) subjects rated their disease mild; 180 (42%) subjects scored their disease moderate with a rating of 4 to 7; and a score of 8 to 10 was considered severe, and that is how the rest of the 200 (46.6%) subjects rated themselves.1° Cost-Effectiveness Diagnosis. There was a significant difference between the patch-tested and non-patch-tested groups in terms of the time to obtain a confirmed diagnosis* (medians = 8 days and 175 days, P < .OOl) and a significant difference in the ratio of subjects that had a confirmed diagnosis (88% and 69%, P < .OOl) (Table 1). These findings show the effectiveness of patch testing as a tool to diagnose the cause of contact dermatitis and thereby reduce *For non-patch-tested subjects, physicians used their routine customary methods to confirm diagnosis, which may have included clinical morphology, history, response to treatment, biopsy, or other factors.
152
RAJAGOPALAN
Table 2. Comparison
of Mean
Change
Non-Patch-Tested
in Score
for Dermatology-Specific
Groups
Within
Recurrent
Non-Patch-Tested DSQL Scale Physical
symptoms
Activities
of daily
N
Quality Allergic
of Life (DSQL)
Contact
Dermatitis
Group
Mean Change
Scales
Patch-Tested
and
(N = 169)
Patch-Tested SE
Between
ET AL
Group
Mean Change
SE
PValue
97 96
-0.58 -0.09
0.10 0.09
69 68
-1.23 -0.44
0.12 0.10
,001 ,001
Social functioning Mental health
98 94
-0.30 -0.18
0.10 0.09
69
-0.63 -0.45
0.12
.039
Vitality Work/school Self-perception
functioning
93 64
-0.13 -0.15
0.11 0.08
67 66 33
-0.37 -0.17
0.10 0.12 0.12
,052 ,137 ,916
Self-perceived
severity
97 98
-0.47 -1.85
0.11 0.33
67 69
-0.77 -3.59
0.13 0.36
.081 .OOl
living
(global
item)
the duration of disease. In the patch-tested group, the investigators advised the subjects to avoid substances based on the identification of the allergen in 76% of the subjects, whereas only 43% in the non-patch-tested subjects received that advice. As a result of changes made in their lifestyle, 66% in the patch-tested group and 51% in the non-patchtested group reported 75% or more improvement in their disease symptoms after 6 months. Treatment. Subjects who were diagnosed within 8 days of screening incurred lower medical expenses (prescriptions and office visits), an average of $99 per month in the prediagnosis period, compared with those who were diagnosed more than 8 days after screening. Once the diagnosis was confirmed, both groups had incurred fairly similar expenses (ranging from $15 to $60 per month). When total costs were comparedwithin the severity groups, however, the differences between patchtested and non-patch-tested groups varied considerably. Particularly in subjects with severe dermatitis, costs were much less for the patch-tested group compared with the non-patch-tested group. Early confirmation of diagnosis helped reduce the prediagnosis costs of treatment, which was mostly based on preliminary diagnosis. Quality of Lz@. Use of patch testing among subjects with moderate to severe ACD is associated Table 3. Comparison
of Mean
Change
in Score Chronic
for DSQL Scales Allergic Contact Non-Patch-Tested
DSQL Scale
N
Physical symptoms Activities of daily living Social functioning
74 72
health
75 74 73
Work/school functioning Self-perception Self-perceived severity
47 74 75
Mental Vitality
(global
item)
Mean Change
with greater quality-of-life benefit at 6-month evaluation than when patch testing was not used.*OThis result is obtained despite the fact that those not patch tested, as a group, have less severe ACD than those patch tested. The greatest quality-of-life benefit from patch testing, relative to non-patch testing, occurred in subjects with recurrent or chronic ACD (Tables 2-4). Less benefit was found for subjects with mild or moderate disease. DECISION
ANALYSIS
The calculation of expected values of costs and effects in terms of change in quality of life using the decision analysis model are in Tables 5 and 6. Actual decision analysis model is presented elsewhere.” The subjects were first classified under low-utility and high-utility clinics in terms of patch testing and further subclassified by the selfreported severity of the disease. For each of the subclasses, the total costs were evaluated for the decision alternatives-patch test within 8 days or not. Regardless of disease severity, patch-tested subjects showed a greater than 75% rate of confirmed diagnosis within 8 days. In both groups of clinics, the “severe” group of subjects showed the highest degree of costeffectiveness for the patch testing. In the highutility clinics, the “moderate” group of subjects had Between Patch-Tested Dermatitis (N = 139) Group
and Non-Patch-Tested
Patch-Tested N
Mean Change
0.11 0.10
62 56
-0.06 -0.01
0.10 0.09 0.11
60 60
-0.98 -0.56 -0.54
-0.01 -0.21 -2.09
0.05 0.12 0.37
63 30 56 61
Groups
Within
Group
SE
-0.71 -0.18 -0.19
MODEL
SE
P VallX
0.16 0.11
,157 ,009
-0.40 -0.31
0.14 0.12 0.14
,034 ,026 .I01
-0.45 -0.86 -3.18
0.12 0.17 0.44
,001 .002 .050
PATCH
TESTING
IN ALLERGIC
Table 4. Comparison
CONTACT
of Mean
DERMATITIS
Change
in Score With
153
at 12 Months
Chronic
for DSQL
Allergic
Contact
Scales
Between
Dermatitis
Early
Patch Tested Within 7 Days of Screening DSQL Scale
Physical Activities
N
symptoms of daily living
Mean Change
Versus
Late Patch-Tested
All Others Patch Tested
SE
N
Mean Change
SE
PValue
,001 ,016
58 58
-0.74 -0.43
0.14 0.13
36 37
-0.01 -0.03
0.14 0.12
Social functioning Mental health
58 58
-0.75 -0.50
0.14 0.12
37 38
-0.01 -0.02
0.13 0.17
Vitality Work/school
58 38
-0.21 -0.55
0.16 0.15
38 21
-0.32 0.15
0.16 0.13
56 60
PO.88 -3.83
0.16 0.36
37 37
-0.19 -1.91
0.15 0.52
functioning
Self-perception Self-perceived
severity
(global
item)
no difference in the cost-effectiveness between the decisions of performing and not performing patch testing within 8 days. The “mild” group of subjects, however, showed higher cost-effectiveness than the “moderate” group. Therefore, patch testing is shown to be definitely cost-effective in a descending order for subjects with severe ACD and then those with mild ACD; in moderate ACD, there is an indifference, meaning that the physician could decide either way. This can be explained by the experience and expertise of the dermatologists in the highutility clinics who can be intuitive in counseling and treating their subjects even with mild ACD. In low-utility clinics, however, there is no cost-effectiveness shown for patch testing in moderate and mild ACD. DISCUSSION Utility of patch testing to the patients was shown in the retrospective study cited earlier.8 Utility of patch testing to the physician is evident from the higher probability of a confirmed diagnosis and the shorter time to a confirmed diagnosis. Table 5. Expected Cost and Effect* and Cost-Effectiveness Ratio in 1 Year for Clinics That Patch Tested <50% of Their Subjects Decision to Patch Early
Severity Groupt
Mild
PT 58 days No PT c8 days
Moderate Severe
Expected cost* ($)
PT 58 days No PT 58 days PT 58 days No PT 58 days
*Expected
Table 6. Expected Cost and Effect* Ratio in 1 Year for Clinics That of Their
cost per Unit Change in QOL ($1
Severity Groupt
669.76 1,070.76
-0.56 -0.10
1.196.00 10,708.OO
999.42 905.51 999.45
-2.00 -2.05 -3.31
500.00 442.00
1,689.18
-1.77
960.00
Mild
1,431.oo 832.00
PTs8d No PT 58 d
Moderate
692.00 800.77
-1.93 -6.61
359.00 121.00
PT 58 d NoPTs8d PTc8d
983.64
-2.69
366.00
severe before
their skin problem screening. Severity
are based
*Expected abilities
perception
of
how
has been on average over the month is derived from QOL assessments.
QOL, quality
No PT 58 d
on prob-
of life; PT, patch
test.
severe before
Subjects Cost per Unit Change in QOL ($1
2,280.OO
Severe
and Cost-Effectiveness Patch Tested ~50%
Expected Effect* (A QOL)
-0.68 -2.21
effect
,002
Expected cost* I$)
-0.48
cost and expected
,001
Decision to Patch Early
972.87 1,839.27
from the decision analysis model. ISeverity is based on subjects’
,001 ,021 ,648 ,001
Further, the longer the delay in confirming the diagnosis, the longer the subjects had to undergo symptomatic disease activity; the earlier the diagnosis was confirmed, the earlier they could have disease activity alleviated by avoiding the causative substances. Further, the costs of treating the disease without a firm diagnosis can be avoided. Clinical practice patterns also affected qualityof-life outcomes. Patch testing fewer than one half of such subjects classified as having chronic or recurrent ACD would leave many of those not patch tested with little or no quality-of-life improvement over 6 months. Increased use of patch testing may result in better quality-of-life outcomes across subjects. Data also show, however, that patch testing all, or nearly all, such subjects probably reflects overutilization from a quality-of-life effectiveness standpoint because a proportion of subjects (~50%) with chronic or recurrent ACD in some dermatology practices apparently can be selected to have good prognoses without patch testing. Quality-oflife benefits from patch testing appear to be moderated by subject selection processes, which guide
1,094.23
abilities
Abbreviations:
Expected Effect” (A QOL)
Subjects
(N = 139)
cost and expected
effect
are based
302.00
on prob-
from the decision analysis model. tseverity is based on subjects’ perception of how their skin problem has been on average over the month screening. Severity is derived from QOL assessments. Abbreviations: QOL, quality of life; PT, patch test.
RAJAGOPALAN
154
treatment allocation, and that at some point between 50% and 100% utilization, optimal effectiveness is reached. When quality-of-life improvements were compared between chronic ACD subjects patch tested in the first week of the study and chronic ACD subjects who were patch tested later in the study period, the former group showed a remarkable improvement superior to the latter, and the differences were statistically significant. This shows the importance of early confirmed diagnosis. This difference in quality-of-life improvement also influences differences in frequency of visits, number of prescriptions, and other direct and indirect costs. From the perspective of the audience, who are traditionally used to randomized clinical trials, this design of an observational study might seem to introduce bias because the option to patch test is totally based on the discretion of the dermatologist. Because the purpose of the study was to describe the use and outcomes of patch testing in actual practice, an unobtrusive observational study was most appropriate. The absence of randomization should be taken into consideration, however, when the comparison between patch-tested and nonpatch-tested groups is made. Another point is that 431 (76% of the enrolled) subjects were evaluable with PE data. It is worth mentioning that in an observational study, which is almost like a survey, a 50% response is considered quite impressive.
tional cost involved. Patch testing also helps confirm diagnosis of ACD at a higher rate and at an earlier time point. More patch-tested subjects than non-patch-tested subjects were appropriately advised to avoid a substance to aid in management of ACD. At the end of 6 months, a larger proportion of patch-tested subjects reported better than or equal to 75% improvement in their symptoms compared with the non-patch-tested subjects. By confirming the diagnosis earlier, the prediagnosis costs on prescriptions and office visits were reduced; the postdiagnosis costs on prescriptions and office visits were comparatively less for the subjects diagnosed within the first 8 days of the screening than those diagnosed later. Overall, the improvement in quality of life is greater in subjects who had their diagnosis confirmed earlier with patch testing than those with no patch testing or late patch testing. On the whole, the data showed clearly the importance of early confirmation of diagnosis of ACD, identifying and avoiding the causative allergens, and treatment and the importance of patch testing for early confirmation of diagnosis. ACKNOWLEDGMENTS The
authors
acknowledge
W. Loss, MD,
search
Patch testing has been considered worthwhile by subjects despite the discomfort, time, and addi-
the valuable
contributions
of
all the investigators-William Abramovits, MD, Ponciano D. Cruz, MD, Zoe Draelos, MD, Joseph F. Fowler, Jr., MD, Robert David
M. Pariser,
MD, Joel Shavin, MD, Schlessinger, MD-and
CONCLUSION
ET AL
Corporation The
Wellcome
toward
MD,
FACP,
Marvin
Rapaport,
Elizabeth F. Sherertz, MD, and Joel the clinical monitors from PharmaRethe success
of this study.
study was conducted with a grant Inc, Research Triangle Park, NC.
from
Glaxo
REFERENCES 1. Fisher 36:451, history
1985 2. Kieffer and patch
AA: The use and abuse of patch
testing.
Bethesda, M: Nickel sensitivity: test reaction. Contact
son between 50,1994 bility nous
A, Seidenari
two application
S: Patch
methods.
tact Dermatitis Group. 1994 5. Ruhnek-Forsbeck
J Am
Acad
M,
Fischer
for Health
between 5:398-401,
contact
test reading:
a compari-
Dermatol21:812-815, 1995 8. RajagopalanR, K&al@,
Contact
4. Brasch J, Henseler T, Aberer of patch tests: a multicenter left-versus-right-sided patch tests
Center
Dermatitis
30:49-
Statistics:
Advance
data.
MD, 1995 7. Rietschel I& Human and economic impact of allergic dermatitis and the role of patch testing. J Am Acad
relationship Dermatitis
1979. 3. Motolese
6. National
Cutis
tive evaluation allergic contact
of patch dermatitis.
FowlerJF,
et al: Aretrospec-
testing in subjects diagnosed Cutis 57:360-364, 1996
with
W, et al: Reproducistudy of synchroby the German Con-
9. Ware JE Jr: SF-36 Health Survey. The Health Institute. Boston, MA, New England Medical Center, 1993 10. Rajagopalan R, Anderson RT: Impact of patch test-
Dermatol
ing on dermatology-specific lergic contact dermatitis.
31:584-591,
T, Meding
Comparative multi-center study with TRUE Chamber patch test methods in eight Swedish DefmVenereol (Stockh) 68:123-128,1988
B, et al:
Test and hospitals.
Finn Acta
quality of life in patients with alAm J Contact Dermat 8:215-221, 1997
11, Rajagopalan R, Anderson RT, Sarma S, et al: Use of decision analysis modeling in economic evaluation of patch testing in allergiccontact dermatitis. Submitted toPharmacoEconomics,
1997