SLIDE 1 Descriptive Study Aya Goto
Dec 29, 2015
SLIDE 1
Descriptive StudyAya Goto
Key concepts
Nicely presented quantitative data tells a lot WITHOUT p-values.
Graphs and tables
SLIDE 2
Grades of evidence
SLIDE 3
Concato J, Shah N, Horwitz RI. Randomized, controlled trials, observational studies, and the hierarchy of research designs.N Engl J Med. 2000 Jun 22;342(25):1887-92.
This is where research starts!
Descriptive and Analytical study
Descriptive study: Describe distribution of each factor individually.
Analytical study: Analyze associations among factors.
In any study design, you must describe distribution of all factors you investigate.
This is the very basic and most important of all study designs.
SLIDE 4
SLIDE 5
Abortion trends in Japan, 1975-95. Goto A, Fujiyama-Koriyama C, et al. Stud Fam Plann. 2000; 31(4): 301-8.
0
10
20
30
40
<2020-2425-2930-3435-3940-44
1980 1985 1990 1995
Abortion rate (per 1000 women)
SLIDE 6
100
1000
10000
<2020-2425-2930-3435-3940-44
1980 199519901985
Abortion ratio (per 1000 live births)
SLIDE 7
N(%)
Infections Nghi Thuy
Nghi Khanh
Nghi Thuan
Nghi Xa
Hung Thong
(N = 46) (N = 35) (N = 47) (N = 36) (N = 44)Any infection 19 (41.3) 14 (40.0) 14 (29.8) 14 (38.9) 20 (45.5)Endogenous infectionsCandidiasis 5 (10.9) 6 (17.1) 8 (17.0) 8 (22.2) 12 (27.3)Bacterial vaginosis 2 ( 4.4) 4 (11.4) 4 ( 8.5) 2 ( 5.6) 3 ( 6.8)Group B streptococcus 3 ( 6.5) 1 ( 2.9) 3 ( 6.4) 3 ( 8.3) 1 ( 2.7)STDTrichomoniasis 2 ( 4.4) 0 ( 0.0) 0 ( 0.0) 0 ( 0.0) 0 ( 0.0)Gonorrhea 0 ( 0.0) 0 ( 0.0) 0 ( 0.0) 0 ( 0.0) 1 ( 2.7)Syphilis 0 ( 0.0) 0 ( 0.0) 0 ( 0.0) 0 ( 0.0) 0 ( 0.0)Hepatitis B 12 (26.1) 4 (11.4) 2 ( 4.3) 2 ( 5.6) 4 ( 9.1)
Prevalence of and factors associated with reproductive tract infections among pregnant women in ten communes in Nghe An Province, Vietnam.
Goto A, Nguyen QV, et al. J Epidemiol. 2005; 15(5): 163-72.
SLIDE 8
Ecologic fallacy
Is protein causing colon cancer?
Table and figure format
SLIDE 9
TableTable 1. Age-specific abortion rates and ratios in Japan and Fukushima, 2000
Age groupTotal <20 20-24 25-29 30-34 35-39 40-44 45-49
Abortion rate (per 1000 women)Japan 11.6 12.2 20.1 15.1 14.2 13.2 6.2 0.5Fukushima 20.7 18.4 30.8 23.8 22.9 19.9 10.2 0.7
Abortion ratio (per 1000 live births)Japan 285 2249 512 154 156 420 1624 5775Fukushima 390 2627 462 203 237 607 2324 13500
SLIDE 10
Bar chart
0.02.04.06.08.0
10.012.014.016.0
1995 1996 1997 1998 1999 2000 2001
Abortion rate (1000 women)
Figure 1. Trends in abortion rate, Japan, 1995-2001
SLIDE 11
Line graph
0
5
10
15
20
25
1980-84 1985-89 1990-94 1995-99
FukushimaJapan
Abortion rate ( 1000 women )
Figure 2. Trends in abortion rate, Japan and Fukushima, 1980-1999
SLIDE 12
HistogramFigure 3. Age-specific proportion of abortion cases, Japan, 1999
05
1015
2025
Per
cent
1 2 3 4 5 6var1 <20 20-24 25-29 30-34 35-39 40-44
SLIDE 13
USAGETable Precise dataBar chart TrendLine graph Trend
(Comparing several groups)
Histogram Distribution
Summary measures
SLIDE 14
Table 1. Basic characteristics of pregnant womenN (%) or Median (min, max)
Age (years) 27 (18, 44)Occupation
Housewife 27 ( 5.4) Agriculture 317 (62.8) Business 62 (12.3) Government employee 71 (14.1) Others 28 ( 5.5)Education
None to primary school 64 (12.7) Secondary school 271 (53.7) High school 101 (20.0) Professional school 53 (10.5) University 15 ( 3.0)Household assets (0-6)* 3 (1, 6)
* Electricity, bicycle, motorbike, running water, television, and telephone were the assets measured.
SLIDE 15
Arithmetic rules can NOT be applied. N (%)• Categorical data
Blood type: 1= O, 2=A, 3=B, and 4=ABInjury: 1=fatal, 2=severe, 3=moderate, 4=minor
Arithmetic rules can be applied. Summary measures• Continuous data
Hb level Number of births that woman has given.
SLIDE 16
Summary measures of continuous dataMean=average
Standard Deviation(SD)
Mean±2SDs = a range in which “most” of your subjects fit. (About 95% lie within 2 SDs)
Median=50th percentile
Range=Min/MaxMode=most frequent value(s)
SLIDE 17
Age: 6, 6, 7, 8, and 25Mean =Median =Mode =
SLIDE 18
Selection of Measures1. Sample size: Large (>30) --> Mean (SD)
Small --> Median (Min,Max)
2. Shape
Bell-shape Mean(= Median = Mode)
More than one peaks Mode
Skewed to the right (left) Median
Median/ModeMean
Clinical descriptive epidemiologyModel case from a private clinic in JP
Shironishi Clinic Dr. Ogawa, Yutaka General practitioner in a remote area
Practice Diary
(Year / Month / Day) No. Out-patients ( ) No. Home visits ( )
ID NAME TEL NO. TREATMENT
Other Notes
Check points
Daily practice notes
Date ID Name Birth date
Referral hospital
Reason Follow-up response #1
#2 #3
Referral notes
SLIDE 23
Total Follow-up response received No responseTotal Brief visit
record onlyTreatment record
Total Satisfactory Not satisfactory
268 (100%)
221 (82%)
36 (13%)
185 (69%)
157 (59%)
28 (10%)
47 (18%)
“A systematic referral system should be established.” A referral card was introduced in the region. The paper was awarded.
Kosei No Sihyo2000: 47(8): 23-28.
Simple clinical-oriented descriptive study
SLIDE 24
Better
practice