Upload
arnold-leyes
View
215
Download
0
Tags:
Embed Size (px)
Citation preview
Designing effective tables
Kostas Danis
Competency to be gained from this lecture
Lay out data effectively in tables
Key areas
• Essential rules when arranging a table• Common tables in field epidemiology
Communicating patterns and messages contained in your data
• Show the patterns inherent in the data• Focus attention on these patterns• Serve as a basis for narrative or
discussion• Lead observer to insight, discussion,
conclusions
Avoid visual puzzles in tables
• Poorly organized data• Series of complicated numbers• Important data obscured• Unnecessary frames, lines, coloring• Decoration
Basic table rules
Column headings
Data
Footnotes
Title
Row headings
Typical table layout with components
Making sure that a table is understandable without referral to other
material • Title
Person Time Place Content of cells (any measurement found in all
columns)
• Row and column headings Content of the row or column Any modifier applied to all cells of a row or column Unit of measurement Abbreviations, if necessary
• Eliminate acronyms, unless standard (eg.OR)• Avoid excessive use of capitals
Basic table rules
Using footnotes in a table
• Clarify points of potential ambiguity• Explain all:
Abbreviations Symbols Codes
• Note exclusions • Mention data source if applicable
Basic table rules
Table 2. Cases And Controls Among Customers at UMFS
Cases Controls TotalOR
(95%CI)
Swordfish 34 20 54 13.2 (5.3-33.0)
Paella 8 62 70 0.1 (0.01-0.95)
Chicken 12 23 35 1.0 (0.4-1.9)
Flan caramel 20 40 60 0.9 (0.2-2.9)
Crema catalan 10 22 32 0.3 (0.1-1.4)
Lemon tarte 0 80 120 -
Incomplete title
Absence of necessary footnotes
Excess use of capitals Acronyms
REVISED Table 2. Frequency of exposures among 42 cases of gastrointestinal illness and
82 controls by fish consumption, ”Uncle Mike’s Fish & Chips”, Berlin, 2005
ExposureCases*
n=42Controls
n=82Odds Ratio(95CI% †)
Swordfish 34 20 13 (5.3-33)
Paella 8 62 0.1 (0.0-0.9)
Chicken 12 23 1.0 (0.4-1.9)
Flan caramel 20 40 0.9 (0.2-2.9)
Crema catalan 10 22 0.3 (0.1-1.4)
Lemon tarte 0 80 Reference
* 2 cases were excluded† 95% Confidence Interval
*ASC Ehrenberg, J R Statis Soc A, 140(3):277-297, 1977
Suggestions for data arrangement in tables*
1. Round data to 2 meaningful figures2. Summarize rows and columns3. Compare numbers in columns 4. Arrange key data by magnitude5. Help the reader with easy table layout6. Align numbers by decimalures
Basic table rules
Table with excessive number of meaningful figures
Factor Cases RateRate Ratio pa
None 27451 2.345 1.000 Refb
A 34211 3.433 1.464 0.1011
B 11002 5.661 2.414 0.0133
C 5643 6.001 2.559 0.0005
a. p-value
b. Reference exposure category
Up to five meaningful figures
Rate ratios difficult to compare
1. Round data to 2 meaningful figures
Basic table rules
Rounding data in a table to 2 meaningful figures
Factor
Cases(1000s
) RateRate ratio p
None 27 2.3 1.0 Ref*
C 34 3.4 1.5 >0.100
A 11 5.7 2.4 <0.050
B 06 6.0 2.6 <0.001
a. p-value
b. Reference exposure category
2 meaningful figures
Rate ratios easier to compare
1. Round data to 2 meaningful figures
Basic table rules
Rounding tips
• Cut decimals for percentages, eg 56.78 %• Use of thousand dividers, eg 18,526• Round up measures of associations to 2 meaningful
figures: 2 decimals between 0-0.99 1 decimal between 1-9.9 0 decimals between 10-99 round to nearest 10 between 100-999ORs symmetrical around 1 on log scale134 same precision as 13.4 or 1.34 or 0.134
X
Rounding tips: p-values
Basic table rules
P-valueNumber of decimals Example
>0.10 2 0.21
<0.10-0.001 3 0.041
<0.001 3 p<0.001
Year M FBoth Sexes
1973 500 99 600
1970 580 87 670
1968 460 89 550
1966 260 71 330
Mean 430 86 520
Summary of the columns
Summary of the rows
2. Summarize rows and columns
Summarizing rows and columns with totals, averages or other statistics
Basic table rules
Compare numbers in columns
23 42 34 109 87 42 27 98 114 75
Difficult to compare numbers in rows
23 42 34
109 87 42 27 98
114 75
1st improvement: Right-justify numbers
vertically 2327
34424275
87 98109
114
2nd improvement: Sort numbers
3. Compare numbers in columns
Basic table rules
Organize data by magnitude
Exposure
Cases(1000
s) RateRate ratio Pa
A 11 2.9 1.3 > 0.100
B 06 9.9 4.3 < 0.001
C 34 5.4 2.3 > 0.100
None 27 2.3 1.0 Refb
4. Arrange key data by magnitude
a. p-value
b. Reference exposure category Basic table rules
Organize data by magnitude
Exposure
Cases(1000s
) RateRate ratio pa
B 6 9.9 4.3 < 0.010
C 34 5.4 2.3 < 0.050
A 11 2.9 1.3 > 0.001
None 27 2.3 1.0 Refb
a. p-value
b. Reference exposure category
4. Arrange key data by magnitude
Basic table rules
YearBothsexes Male Female
1973 600 500 99
1970 670 580 87
1968 550 460 89
1966 330 260 71
Spaced out table layout: Comparisons difficult for the reader
5. Help the reader with easy table layout
Basic table rules
YearBothsexes Male Female
1973 600 500 99
1970 670 580 87
1968 550 460 89
1966 330 260 71
5. Help the reader with easy table layout
Drawing columns and rows close together facilitates comparisons
Basic table rules
Intervening statistics: Separated numbers are harder to compare
Rate per 1000 (SE)
Year Male Female All
1993 83 78 80
2.3 2.2 1.9
1994 62 66 63
2.5 2.7 1.8
1995 58 54 56
2.1 2.0 1.7
1996 55 45 51
2.0 2.0 1.7
5. Help the reader with easy table layout
Basic table rules
Rate per 1000 (SE)
Year Male Female All
1993 83 (2.3) 78 (2.2) 80 (1.9)
1994 62 (2.5) 66 (2.7) 63 (1.8)
1995 58 (2.1) 54 (2.0) 56 (1.7)
1996 55 (2.0) 45 (2.0) 51 (1.7)
Moving and minimizing intervening numbers facilitates readability
5. Help the reader with easy table layout
Basic table rules
Rate per 1000a
Year M F All
1993 83 78 80
1994 62 66 63
1995 58 54 56
1996 55 45 51
a. Standard errors for all rates less than 5% of rate.
Remove intervening numbers entirely if consequence minimal
5. Help the reader with easy table layout
Basic table rules
Align columns by decimal
23 42 34
10.9 8.7 42 27 9.8 114 75
23.0 42.0 34.010.9 8.7
42.0 27.0
9.8 114.0
75.0
Keeping the zeros or not is a question of personal style
6. Align numbers by decimal
Basic table rules
More suggestions
1. Use one column for each of figures2. Use only horizontal lines between sections of
table3. Avoid redundant (duplicated) data4. Use landscape format to display more
information, if needed5. Merge tables that share the same
denominator, but do not mix data from different populations, denominators, indicators (medians/proportions)
Basic table rules
Table 1: Distribution of the Households (n=506) by per capita monthly income,
Place X, 20012Monthly income per
capita (Euros)
Number
(%)
Up to 500 268
(53.0)
501 – 1000 131
(25.94)
1001 – 2000 75
(14.82)
>2000 32
(6.31)
Place number and % in separate columns
Excessive use of formatting lines, vertical divider not needed
Text not aligned to the left
Proportions not rounded
REVISED
Table 1: Distribution of the households (n=506) by per capita monthly income,
Place X, 2012Monthly income
Per capita (Euros)
Numbe
r
Percentag
e
Up to 500 268 53
500-1,000 131 26
1,001-2,000 75 15
>2,000 32 6
Table 2- Baseline characteristics of parents/guardians and their children, vaccination
coverage survey, Greece, 2006
Common tables
Sex Number Percentage
Female 1,919 49.6
Male 1,949 50.4
Total 3,868 100
Redundant:
Proportion of females will indicate proportion of males
X
Table 4- Complete vaccination coverage of children by place of residence, vaccination coverage survey, Greece, 2006
Place of residence
n (weighted %) 95% CI
UrbanRural
1676 (65%)448 (58%)
63.2-67.552.5-61.3
Table 3- Complete vaccination coverage of children by maternal belief, vaccination coverage survey, Greece, 2006
n (weighted %) [95%CI]
Positive attitude of mother towards her child’s vaccinationNo
1993 (64.5) [62.2-70.5]
24 (52.3) [49.1-62.1]
Row heading takes more than one line-too wordy
Use one column for each figure
Consider landscape format
Merge tables with identical structrure
Use thousand dividers
Cut decimals from percentages
Explain 95%CI in a footnote
n % * 95% CI†
Place of residence
UrbanRural
1,676 448
6558
63-6752-61
Maternal attitude
PositiveNegative
1,99324
6552
62-7149-62
REVISEDTable 3- Complete vaccination coverage of children by selected characteristics, vaccination coverage survey,
Greece, 2006
* Weighted % allowing for clustering † 95% Confidence Interval
Table 2. Clinical characteristics of 102 cases of campylobacteriosis, Ireland, 2002
Characteristics Value
Total cases 102
Median age (years)Range (years)
FeverDiarrhoeaJoint pain
355-83
65 (65.6 %)102 (100 %)
4 (4.3)
HeadacheMuscle pain
Isolation of organism
12 (12.4%)4(4.4%)
Stool samples (5/93%)
Text must be alighned to the left
The table presents frequency of symmptoms
Quantitave variables/other info should not be here
Sort rows. Decreasing order
REVISED
Table 2. Frequency of clinical characteristics of 102 cases of campylobacteriosis, Ireland, 2002
Symptoms n %
Diarrhoea 102 100
Fever 65 66
Headache 12 13
Joint pain 4 4
Muscle pain
4 4
Arranging common types of tables in epidemiology
• Line listing• Two variable table• Complex table• Cohort study• Case-control study
Common tables
State Age1 Sex Days2 Dose
New York 02 M 03 1
California 03 M 03 1
Pennsylvania 06 M 03 1
Pennsylvania 02 M 04 1
Colorado 04 F 04 1
California 07 M 04 2
Kansas 02 F 05 1
Colorado 03 M 05 1
New York 03 F 05 1
North Carolina 04 F 05 1
Missouri 11 M 05 1
Pennsylvania 03 F 07 1
California 04 F 14 2
Pennsylvania 02 M 29 1
California 05 M 59 1
1. Age in months
* MMWR, 48 (27):577
2. Days from dose to symptom onset
Reported cases of intussusception among recipients of rotavirus vaccine, by state, United
States, 1998-1999*
a. Line listing
Common tables
New cases of primary and secondary syphilis by age group and sex, United States, 1989
Age group Cases (100’s)
(years) Male Female Total
14 0.4 1.9 2.3
15-19 17.4 27.9 44.3
20-24 51.4 53.9 100.3
25-29 53.4 42.9 96.3
30-34 55.4 31.9 86.3
35-44 50.4 19.9 69.3
45-54 21.4 49.9 26.3
55 11.4 13.9 13.3
Total 260.4 180.9 440.3
b. Two variable table
Common tables
Complex table
Children
Character Exp %(n=205)
Not exp %(n=8729) p
Gestational age (weeks) at birth
<25 5.8 14 0.04
25-29 18.0 19 NS
Birthweight (kg)
1.5 15 .0 15 NS
2.5 39 .0 43 NS
c. Complex table
Common tables
ate ham
did not ham
ill not ill
49 49 98
4 6 10
2x2 table for calculation of measure of effect
d. Cohort study
Tab. IV Fish consumption and gastro-intestinal illness among customers at ”Uncle Mike’s Fish & Chips”, Berlin, 2005
Ill Total Attack rate
Relativerisk
Ate fish 42 58 72% 9.3 (3.9-22)
Did not eat fish
5 64 8% Ref
Total 47 122 39%
d. Cohort study
2x2 table fo
r caclu
lations
Not for p
resentatio
n
Exposed
Exposure%
Res. a
Yes No RRc (95% CId)
n ARb n ARb
Type 1
Sub Type 1-A
( - )
Sub Type 1-B
( - )
Sub Type 1-C
( - )
Type 2 ( - )
Type 3 ( - )
Type 4:
a. Res. = Respondedc. RR = Risk Ratio
b. AR = Attack Rate – cases per ___d. 95% CI = 95% confidence interval of the RR
d. Cohort study
Risk of ______ by exposure, among #### residents of Place, time
Common tables
Exposed
Exposure n AR a RR b 95% CI c
Type or Level 3
Type or Level 2
Type or Level 1
None or Level 0 1.0 Referent
b. RR = Risk Ratio
c. 95% CI = 95% confidence interval of the RR
a. AR = Attack Rate – cases per ___
Risk of ______ by exposure, among #### residents of Place, time
d. Cohort study (reference group)
Common tables
Exposed
Not exposed
Cases Controls Odds ratio
Case control study
50 20 4 a b
50 80 c d
Total 100 100
Exposed % (n) a
Exposure Cases Controls OR b 95% CI c
Type 1 (n) (n) ( – )
Sub Type 1-A (n) (n) ( – )
Sub Type 1-B (n) (n) ( – )
Sub Type 1-C (n) (n) ( – )
Type 2 (n) (n) ( – )
Type 3 (n) (n) ( – )
c. 95% CI = 95% confidence interval of the OR
a. n = subjects responding b. OR = Odds Ratio
Exposures (%) among ### cases and ### controls, Place, Time
e. Case control study
Common tables
Table from a case control study
Table 5. Association between exposures and campylobacteriosis in case-control
study, Oslo, Norway, 1998. Univariate, matched analysis.
Exposure Cases Controls Odds 95% conf.
ratio interval
Eaten at pizza restaurant 9/37 12/70 1.8 0.62 - 5.0
Eaten at party 10/36 9/74 3.2 0.97 - 11
Eaten foods from deli 23/37 42/74 1.2 0.56 - 2.7
Eaten unpeeled fruits 19/37 54/74 0.27 0.10 - 0.78
Close contact with a case 7/35 2/72 13 1.5 - 110
Drank >4 glas of water per day 21/37 33/74 1.7 0.73 - 3.9
Customer of water company B 27/37 33/74 4.0 1.3 - 7.3
Food Specific Attack Rates, Outbreak of Salmonellosis, Prison X, Dover, Delaware,
September 1992
6.21.51.3 1.1
REVISED for oral presentationFood specific attack rates, outbreak of
Salmonellosis, prison X, Dover, Delaware, September 1992
Take home message
Design your table around the message that is contained in your data
Practical 1
Spot the errors of the following tables
2.3. Reported laboratory diagnosis methods for chronic and acute infections
Lab. method
anti-HCV anti-HCV+ RNA-HCV
RNA-HCV Data missin
g
Chronic cases
(n=10403)
4084 (40.3%)
2659(25.4%)
2057(20%)
1603 (15%)
Acute cases
(n=956)
383(40.4%)
260(27.3%)
199(21.2%)
114(12.4
%)
SmiNet database 2005-2011
Seroconversion could not be verified for the VHC acute cases.- Place acute and chronic vertically to facilitate
comparison- Round up proportions- Add thousand dividers
Reported laboratory diagnosis methods for chronic and acute HCV infections, SmiNet database 2005-2011
Seroconversion could not be verified for acute hepatitis C cases.
Information available among cases
Acute cases Chronic cases
n % n %
Anti HCV 4,084 40 383 40
Anti HCV + RNA 2,659 25 260 27
RNA HCV 2,057 20 199 21
Data missing 1,603 15 144 12
Total 10,403 100 956 100
- Vertical comparisons- Rounded proportions
- Thousand dividers
CMOs reporting procedures19/21 CMOs replied the questionnaire
Easy to apply case definitions?
Yes (Both chronic and acute)
Yes (Only chronic)
Yes (Only acute)
No
Replies (n=19) 9 (47.5%) 1 (5%) 0 9 (47.5%)
Reporting instructions
for labs
Report after confirmation by
imunoblot positive test
Report after any
antibody positive
test
Wait for RNA
confirmation test
Other
Replies (n=19)
12 (63%) 2 (10%) 1 (5%) 4 (21%)
- Two tables with identical structure- Incomplete title
Hepatitis C reporting procedures described by 19 of the 21 Chief Medical Officers (CMOs) surveyed, Sweden, 2012
Item Answers N %
Case definition easily applicable
For chronic and acute cases 9 47
For chronic cases only 1 5
For acute cases only 0 0
No 9 47
Reporting instruction for laboratory
After confirmation (Iblot) 12 63
After any antibody test 2 10
Wait for RNA 1 5
Other 4 21
Total 19 100
- Merged table- Time, place and person title
Sexually transmitted infections (STIs)• Main public health concern
• Prevention of STI transmission is a major PH challenge
Number of new STI diagnoses in 2009-11, and changes in trend in 2002-11, England
New STI diagnoses
Year % Change
2009 2010 2011 2009-10 2010-11 2002-11
Chlamydia 189,356 189,314 186,196 0% -2% 135%
Gonorrhoea 16,144 16,835 20,965 4% 25% -13%
Syphilis* 2,851 2,650 2,915 -7% 10% 87%
Herpes** 27,536 29,794 31,154 8% 5% 81%
Warts** 77,845 75,415 76,071 -3% 1% 21%
Total*** 426,735 419,773 426,867 -2% 2% 49%
*Syphilis: primary, secondary & early latent **Anogenital herpes / warts
***Total includes diagnoses stated in the table, plus ‘Non-specific genital infection’, ‘Pelvic inflammatory disease & epididymitis’ and ‘Other new STI diagnoses’
Source: http://www.hpa.org.uk/webc/HPAwebFile/HPAweb_C/1215589015024 53
- Two parts in table: Values and changes- Footnote too small / detailed
- Heterogeneous content indicator-wise
Practical 2
Prepare dummy tables for a:•case-control study•cross-sectional study
Practical 2a
Prepare dummy tables for a:•case-control study to identify risk factors for Campylobacter infection•Exposures:
travel food consumption (chicken, lettuce) domestic animals
•Demographics
Practical 2b
Prepare dummy tables for a:•Sero-prevalence study to identify risk factors for West Nile virus infection•Exposures:
rural place of residence mosquito protection employment status
Exposed % (n) a
Exposure Cases Controls OR b 95% CI c
Age>medianFood (n) (n) ( – )
Chicken (n) (n) ( – )
Lettuce (n) (n) ( – )
Travel abroad (n) (n) ( – )
Domestic animal (n) (n) ( – )
c. 95% CI = 95% confidence interval of the OR
a. n = subjects responding b. OR = Odds Ratio
Exposures (%) among ### cases of campylobacter and ### controls, Place,
Time
Exposed
Exposure % P a PR b 95% CI c
Population size Urban Rural 1.0 Referent
Mosquito protection
Often Rarely Never 1.0 Referent
b. RR = Prevalence Ratio
c. 95% CI = 95% confidence interval of the RR
a. P = Prevalence– cases per ___
Prevalence of West Nile virus infection by exposure, among #### residents of
Place, time
Common tables
Group ERS
Group KLO
Group MGI
Group NHO
Group NEA
Group KLN
Day 1-2 Day 1-2 Day 1-2 Day 3-4 Day 3-4 Day 3-4
Factory Atada NDPH
13** 1245 53 3467 3462 2425
Factory Seuda 457 2351 6589 9i0 569 43§
Factory Desda 111 (56) 43 (96) 35 (97) 46 (53) 56 (75) 567 (42)
Factory Rioja 1 1 0 3
Mean age 23 34 23 45 23 32
Travel hours 64 45 56 678 89 890
H C AB level 67 70 890 4356 56 76
HIV + 54 56 678 567 890 9080
Primary school 345 34 45 e65 56 78
Secondary school 234 54 65 568 76 878
BACK-UP SLIDES
61
Results
• Number of cases submitted to USISSAge group
Type<1 1 to 4 5 to 14 15 to 44 45 to 64 65+ Total
A(H1N1) 2 4 1 6 5 7 25A(H3N2) 5 8 3 16 18 17 67A(unknown) 10 20 8 34 21 40 133B 2 0 4 2 8 3 19Total 19 32 16 58 52 67 244
Information hard to follow as table
62
Number of cases submitted to USISS, by age and virus, {Place},
{Time}
Data presented at as graph
Exposed Unexposed
Exposure Total Cases AR% Total Cases AR% RRConfidence
interval P
blood glucose monitoring 30 8 26.67 56 0 0.00 20.26 [3.18-∞] <0.001
diabetes mellitus 40 8 20 46 0 0.00 11.05 [1.71-∞] 0.003
insulin injection 25 6 24.00 61 2 3.28 7.32 [1.58-33.84] 0.003
chiropody 53 8 15.09 33 0 0.00 6.45 [1.01-∞.] 0.048
upper floor 57 8 14,04 29 0 0.00 5.62 [0.87-∞] 0.056
ground floor 43 1 2.33 43 7 16.28 0.14 [0.02-1.11] 0.058
urethral catheter 4 1 25.00 76 7 9.21 2.71 [0.43-17.06] 0.350
eye drops 10 1 10.00 73 7 9.59 1.04 [0.14-7.62] 1.000
sex 26 2 7.69 60 6 10.00 0.77 [0.17-3.56] 1.000
dialysis 1 0 0.00 78 8 10.26 0.00 [.-.] 1.000
63
Multivariable analysis: only blood glucose monitoring significant
Results
- Redundant stats
- Alignment- Decimals
- Neutral title
Exposed Unexposed
Exposure Total Cases AR% Total Cases AR%Relative
riskConfidence
interval
Glucose monitoring 30 8 27 56 0 0 20 3.2-∞
Diabetes mellitus 40 8 20 46 0 0 11 1.7-∞
Insulin injection 25 6 24 61 2 3 7.3 1.6-34
Chiropody 53 8 15 33 0 0 6.4 1.0-∞.
Upper floor 57 8 14 29 0 0 5.6 0.87-∞
Ground floor 43 1 2 43 7 16 0.14 0.0-1.1
Urethral catheter 4 1 25 76 7 9 2.7 0.43-17
Eye drops 10 1 10 73 7 10 1.0 0.14-7.6
Sex 26 2 7 60 6 10 0.77 0.17-3. 6
Dialysis 1 0 0 78 8 10 0.0 .-.
64Multivariable analysis: Only blood glucose
monitoring significant
Risk of hepatitis B according to selected exposures, nursing home, Saxony, Germany, 2011
- Full title- Rounding off
- Alignment- P values deleted