Abstract:
Understanding study design is the foundation for accurately interpreting research. Health care professionals should be able to distinguish well-conducted research from poorly conducted research and know how to apply that knowledge to the care of their community. After participating in this exercise students will be able to recognize the advantages and disadvantages of experimental versus observational studies. This case discusses the proposed association between the MMR vaccine and autism. Students will be prompted to create a study that investigates this claim and to utilize the fundamentals of epidemiology to measure the strength of association between these two variables.
Recommended Reading:
. Madsen KM et al. A population-based study of measles-mumps-rubella vaccination and autism. N Engl J Med 2002;347(19)1477-82.
. Offit, PA. & Coffin, SE. Communicating science to the public: MMR vaccine and autism: Vaccine 2003: 22(1) 1-6.
. Smeeth L et al. MMR vaccination and pervasive developmental disorders: a case-control study. Lancet 2004;364:963-9.
Objectives: At the end of this session, students will be able to:
. Recognize criteria for initiating experimental and observational studies
. Identify key design components of studies
. Describe the advantages and disadvantages of various study designs
. Calculate and interpret outcome measures in cohort and case-control studies
. Interpret data from observational studies looking at MMR vaccination and autism
Introduction
1. What have you heard about an association between vaccines and autism? What are some of the hypotheses you have heard about or read about?
2. What types of studies would you recommend to look at an association between vaccines and autism?
Section A: Study Design
Design a study to look at the association between MMR vaccine and autism.
1. Cohort study
a. How will you assemble your cohort ?
b. What are your inclusion/exclusion criteria for participation in the study?
c. Exposure status: How will you define and measure vaccination status?
d. Outcome status: How will you define and measure autism?
e. What measure of association will you calculate in a cohort study?
2. Case-control study
a. What is your definition of a case?
b. How will you identify cases?
c. What is your definition of a control?
d. How will you identify controls?
e. Exposure status: How will you define and measure vaccination status?
f. What measure of association will you calculate in a case-control study?
Section B: Calculating Measures of Association
Over 25 studies have been conducted which have failed to show an association between MMR vaccine and autism. Below are some data from two of these studies (one cohort and one case-control). Calculate the measure of association for each study after completing the 2x2 table and interpret the measure of association.
1. Cohort study by Madsen et al (NEJM 2002) Of the 537,303 children in the cohort, 440,655 (82.0 percent) had received the MMR vaccine. We identified 316 children with a diagnosis of autistic disorder; 263 of the cases had been vaccinated with MMR (83%); 53 children with autism had not been vaccinated.
2. Case control study by Smeeth et al (Lancet 2004) 1294 cases of pervasive developmental disorder (PDD) and 4469 controls were included. 1010 cases (78.1%) had MMR vaccination recorded before diagnosis, compared with 3671 controls (82.1%) before the age at which their matched case was diagnosed.
Section C: Interpreting results
Review the following data from the aforementioned studies and answer the following questions.
FIGURE 1: Table 1 from Madsen et al., 2002
TABLE 1. CHARACTERISTMS OF THE 537,303 CHILDREN IN THE DANISH COHORT.
|
|
VACC INATED
CHILDREN
|
UNVACC INATED
CHILDREN
|
|
|
|
CHARACTERISTIC
|
W=440,6581
|
(N-88,668)
|
P Vane
|
|
|
|
number (Percent)
|
|
|
|
Sex
|
|
|
0 55
|
|
|
|
|
|
Male
|
226,042 (51.3)
|
49480 (51.4)
|
|
|
Female
|
214,613 (48.7)
|
46968 (4&6)
|
|
|
|
Birth weight
|
|
|
<0.001
|
|
|
c2499 g
|
21,633 (4.9)
|
5,164 (5.3)
|
|
|
|
2500-2999 g
|
53,874 (12.2)
|
12,062 (12.5)
|
|
|
|
3000 -3499 g
|
135,630 (30.8)
|
29,262 (30.3)
|
|
|
|
3500-3999 g
|
135,255 (30.7)
|
29,143 (30.2)
|
|
|
|
.4000 g
|
66,358 (15.1)
|
14,563 (151)
|
|
|
|
Data missing
|
27$05 (6.3)
|
6,454 (6.7)
|
|
|
|
Gestational age
|
|
|
<0.001
|
|
|
Co wk
|
19,029 (4.3)
|
3,129 (3.2)
|
|
|
|
37-41 wk
|
272,345 (61.8)
|
40,609 (42.0)
|
|
|
|
s42 wk
|
27,349 (6.2)
|
3 986 (4.1)
|
|
|
|
Data missing!'
|
121,932 (27.7)
|
48524 (50.6)
|
|
|
|
Socioeconomic statust
|
|
|
<0.001
|
|
|
Manager (very high)
|
41,367 (9.4)
|
9940 (10.3)
|
|
|
|
Wage tamer (high)
|
85,772 (19.5)
|
16,187 (16.7)
|
|
|
|
Wage canter (medium)
|
70$06 (16.1)
|
13,753 (14.2)
|
|
|
|
Wage earner (low)
|
116,503 (26.4)
|
26,699 (27.6)
|
|
|
|
Wage earner (minimal)
|
57,408 (13.0)
|
10,996 (11.4)
|
|
|
|
Unemployed
|
67,841 (15.4)
|
18,519 (19.2)
|
|
|
|
Data missing
|
858 (0.2)
|
554 (0.6)
|
|
|
|
Mother's education
|
|
|
<0.001
|
|
|
Postgraduate education
|
26,118 (5.9)
|
5,856 (6.1)
|
|
|
|
College
|
67,776 (15.4)
|
14,599 (15.1)
|
|
|
|
Vocational training
|
178,553 (40.5)
|
34,006 (35.2)
|
|
|
|
Secondary school
|
42,667 (9.7)
|
10,164 (10.5)
|
|
|
|
Primary school
|
114768 (26.0)
|
28,680 (29.7)
|
|
|
|
Data missing
|
10,773 (2.4)
|
3,343 (3.5)
|
|
|
|
Age at diagnosis of autistic disorder
|
|
|
0.87
|
|
|
<2 yr
|
48 (0.01)
|
9(0.01)
|
|
|
|
3-5 yr
|
187 (0.04)
|
31 (0.03)
|
|
|
|
.6 yr
|
34 (0.01)
|
7(0.01)
|
|
|
|
Age at diagnosis of another autistic-spectrum disorder
|
|
|
0.19
|
|
|
<2 yr
|
32 (0.01)
|
3 (0.003)
|
|
|
|
3-5 yr
|
202 (0.05)
|
37 (0.04)
|
|
|
|
.6 yr
|
118 (0.03)
|
30 (0.03)
|
|
|
1. Describe the characteristics in Table 1 in terms of what type of data they are (discrete / qualitative versus continuous / quantitative).
Using the chi-square test (see footnote of FIGURE 1), authors found significant differences in certain characteristics of vaccinated and unvaccinated children.
2. What characteristics showed no differences between the two groups? What characteristics showed differences between the vaccinated and unvaccinated groups?
3. What do these differences mean?
The New England Journal of Medicine
TABLE 2. ADJUSTED RELATIVE RISK OF AUTISTIC DISORDER AND OF OTHER AUTISTIC- SPECTRUM
DISORDERS IN VACCINATED AND UNVACCINATED CHIDREN.
|
|
|
|
|
OTher Auntnearearral
|
|
VAcenannart
|
Person-year
|
Autistic Disorder
|
Owns
|
|
|
|
|
AMSTED
|
|
ADIOS-OD
|
|
|
|
|
MIAMI IUSX
|
|
MIAMI RIO(
|
|
|
|
NO. Of OWLS
|
(95% CI)
|
NO. Of cuss
|
(95% CI)
|
|
Total
|
2,129,864
|
316
|
|
422
|
|
|
Vaccination
|
|
|
|
|
|
|
No
|
482,360
|
53
|
1.00
|
77
|
1.00
|
|
Yes
|
1,647,504
|
263
|
0.9210.68-1.24)
|
345
|
0.83(0.65-1.07)
|
|
Age a vaccination
|
|
|
|
|
|
|
Not vaccinated
|
482,360
|
53
|
1.00
|
77
|
1.00
|
|
C14 mo
|
200,003
|
38
|
1.18 (078-1.80)
|
43
|
0.88(0.60-1.28)
|
|
15-19 mo
|
1,320753
|
195
|
0.86 (0.63-1.17)
|
270
|
0.83(0.64-1.08)
|
|
20-24 mo
|
69,242
|
17
|
1.19 (0.69-2.07)
|
12
|
0.62(0.33-1.13)
|
|
25-35 mo
|
40,935
|
|
1.20 (0.63-2.31)
|
15
|
1.09 (0.63-1.91)
|
|
g 36 mo
|
16,572
|
2
|
0.56 (0.14-2.30)
|
5
|
0.64(0.26-1.59)
|
|
Interval since vaccination
|
|
|
|
|
|
|
Not vaccinated
|
482,360
|
53
|
1.00
|
77
|
1.00
|
|
<6 mo
|
212,805
|
3
|
0.39 (0.11-1.32)
|
8
|
1.18 (0.51-2.75)
|
|
6-11 mo
|
197,931
|
21
|
1.38 (076-2.51)
|
4
|
0.31(0.10-0.91)
|
|
12-17 mo
|
183,460
|
22
|
1.07 (0.59-1.95)
|
16
|
0.92(0.47-1.80)
|
|
18-23 mo
|
168,045
|
31
|
0.86 (0.52-1.41)
|
16
|
0.47(0.26-0.86)
|
|
24-29 mo
|
154,290
|
42
|
0.99 (0.61-1.58)
|
32
|
0.77(0.46-1.27)
|
|
30-35 mo
|
139,258
|
33
|
0.86 (0.54-1.38)
|
27
|
0.69(0.43-1.11)
|
|
36-59 mo
|
406,320
|
90
|
0.99 (0.66-1.50)
|
158
|
1.05 (0.77-1.45)
|
|
a60 mo
|
185,396
|
21
|
0.67(0.34-1.33)
|
84
|
0.75 (0.51-1.09)
|
|
Date of vxcination
|
|
|
|
|
|
|
Not vaccinated
|
482,360
|
53
|
1.00
|
77
|
1.00
|
|
1991-1992
|
248,646
|
31
|
1.00 (0.59-170)
|
61
|
0.75 (0.51-1.09)
|
|
1993-1994
|
659,152
|
81
|
073 (0.50-1.06)
|
146
|
0.74(0.56-0.99)
|
|
1995-1996
|
475,990
|
96
|
0.91 (0.63-1.30)
|
116
|
1.13(0.81-1.56)
|
|
1997-1999
|
263,716
|
55
|
1.35(0.84-2.17)
|
22
|
0.71 (0.40-1.24)
|
4. Do the data in Figure 2 above show an association between MMR vaccine and risk of autism?
5. Do the data in Figure 3 from Smeeth et al show an association between MMR vaccine and risk of autism?
FIGURE 3: Tables 1 and 2 from Smeeth et al 2004
Table 1: Association between Pervasive Developmental Disorder (PDD) and MMR vaccination before index date, before and after third birthday, and before and after age 18 months.
|
|
Unadjusted OR (95% CI) Adjusted OR (95% CI)* p (for adjusted OR)
|
|
MMR vaccination before index date At any age
|
|
|
|
|
No MR vaccination
|
(1.0)
|
|
|
|
Vaccinated with WAR
|
0.73(059-091)
|
P86(0.68-1.09)
|
021
|
|
Before and after third birthday
|
|
|
|
|
No MMR vaccination
|
(1,0)
|
|
|
|
MMR vaccination before third birthday 0.75(060495)
|
OM (0-70-1.15)
|
039
|
|
MMR vaccination after third birthday
|
(168(0-50-09)
|
077 (0 55.148)
|
013
|
|
Before and after age 18 months
|
|
|
|
|
No ?AMR vaccination
|
(1.0)
|
|
|
|
MMR vaccination before 18 months
|
0.76 (0-60496)
|
0-90 (0.70-1-15)
|
0.39
|
|
Mrs* vaccination after 18 months
|
0.69(054-089)
|
080(041-1.05)
|
till
|
|
|
Unadjusted OR (95% CI) Adjusted OR (95% CI)'
|
p (for adjusted OR)
|
|
MMR vaccination before index date Autism only
No WAR vaccination Vaccinated with WAR Other PDDs only
No WAR vaccination Vaccinated with 0.4MR
|
(1-0)
0-77 (060-0-98)
(1-0)
040 (l) 39-0,92)
|
01113 (067-119
0.75 (046- 1.23)
|
0.25
|
Section D: Conclusion
Since 1998, numerous well-designed studies have found no link between vaccines and autism (see table below). Why do you think some parents are still fearful of vaccines? What is your role as future healthcare providers in counseling patients about vaccines?
Table 1. Studies that fail to support an association between measles-mumps-rubella vaccine and autism.
| Source |
Study design |
Study location |
| Taylor et al., 1999 [5] |
Ecological |
United Kingdom |
| Farrington et al., 2001 [6] |
Ecological |
United Kingdom |
| Kaye et al., 2001 [7] |
Ecological |
United Kingdom |
| Dales et al., 2001 [8] |
Ecological |
United States |
| Fombonne et al., 2006 [9] |
Ecological |
Canada |
| Fombonne and Chakrabarti, 2001 [10] |
Ecological |
United Kingdom |
| Taylor et al., 2002 [11] |
Ecological |
United Kingdom |
| DeWilde et al., 2001 [12] |
Case-control |
United Kingdom |
| Makela et al., 2002 [13] |
Retrospective cohort |
Finland |
| Madsen et al., 2002 [14] |
Retrospective cohort |
Denmark |
| DeStefano et al., 2004 [15] |
Case-control |
United States |
| Peltola et al., 1998 [16] |
Prospective cohort |
Finland |
| Patja et al., 2000 [17) |
Prospective cohort |
Finland |