2. Methodology
• How the study was conducted
• Should be described in enough detail to
permit an experienced investigator to
replicate the study
• Three labeled subsections:
– Participants
– Materials
– Desgin and procedures
3. Participants
• Define inclusion and exlusion criteria
• These should be supported in Intoduction
– e.g. population, gender, age, comorbidity etc.
• Verification methods should be described
– E.g. if hypertension is inclusion/exclusion criteria,
how do you define hypertension ?
4. • When planning a RTC, determination of
inclusion and exclusion criteria is a part of
screening
• Search for simple and cost-effective methods
to determine inclusion/exclusion criteria
– E.g. “patients with distant metastases were
excluded…”
– How do you determine distant metastases ? Chest
X-ray ? Ultrasound ? PET-CT ?
5. • It may be hard or impossible to detect certain
conditions during screening
• Consider later detection and separate
grouping during dana analysis
– E.g. peritoneal carcinomatosis may be vera hard to
detect before surgery
– Solutions:
• carcinomatosis ans exclusion criteria (requires
expensive and invasive procedures to detect)
• carcinomatosis as a separate group in data analysis
6. Measurement scales
• Nominal
• The lowest measurement level you can use, from a
statistical point of view, is a nominal scale.
• Placing of data into categories, without any order or
structure.
• A physical example of a nominal scale is the terms we use
for colours. The underlying spectrum is ordered but the
names are nominal.
• In research activities a YES/NO scale is nominal. It has no
order and there is no distance between YES and NO.
• The statistics which can be used with nominal scales are in
the non-parametric group. The most likely one is
crosstabulation - with chi-square
8. • Ordinal
• An ordinal scale is next up the list in terms of power of
measurement. The simplest ordinal scale is a ranking
• There is no objective distance between any two points on
your subjective scale.
• An ordinal scale only lets you interpret gross order and
not the relative positional distances.
• Ordinal data would use non-parametric statistics. These
would include:
– Median and mode
– rank order correlation
– non-parametric analysis of variance
9. • Example:
• When you ask participants to rank 5 types of
exercise from hardest to easiest
• There is no objective distance between any
two points on your subjective scale. For you
the hardest exercise may be far harder than
the second one, to another respondant with
the same top and second exercise, the
distance may be subjectively small.
11. • Interval
• The standard survey rating scale is an interval scale.
• It is an interval scale because it is assumed to have
equidistant points between each of the scale elements.
We contrast this to an ordinal scale where we can only
talk about differences in order, not differences in the
degree of order.
12. • Interval scale data would use parametric statistical
techniques:
– Mean and standard deviation
– Correlation – r
– Regression
– Analysis of variance
– Factor analysis
• You can use non-parametric techniques with interval and ratio
data. But non-paramteric techniques are less powerful than
the parametric ones.
13. • When you are asked to rate your satisfaction on a
7 point scale, from Dissatisfied to Satisfied, you
are using an interval scale.
• It is an interval scale because it is assumed to
have equidistant points between each of the
scale elements. This means that we can interpret
differences in the distance along the scale. We
contrast this to an ordinal scale where we can
only talk about differences in order, not
differences in the degree of order.
15. • Ratio
• Top level of measurement and is not often available in
social research.
• The factor which clearly defines a ratio scale is that it has
a true zero point.
• Statistical techniques:
– The same as for Interval data
17. • Always use the most accurate scale !
• Example:
– Age (years)
– Age (age group)
Ratio Interval Nominal
Age (years) Age group Young / Old
23.7 >20 and ≤30 0
35 >30 and ≤40 0
42.3 >40 and ≤50 0
67 >65 1
18. Prepare Case Report File
• Information for participants (description of
the study, sideffects, benefits, contact details)
• Copy of the Ethics Committee approval ?
• Signed informed consent sheet
• Dana entry sheet
19. • Data entry sheet
• General information (demographics)
• Screening results
• Determination of inclusion/exclusion
• Unique identifier
• Group code
20. • Data entry sheet
ID Age Height Duration of
symtoms
Tumor size Operation
234 60 years old 5’2” 1 year 3x4x7 Miles
783 74 y 6 months 182 cm 83 days 5 cm in
diameter
Rectal
amputation
987 84.5 176 >5 months 62 mm Rectal
resection
21. • Data entry sheet
ID Age
(years)
Height
(cm)
Duration of
symtoms
(months)
Tumor -
largest
diameter
(cm)
Operation
(code)
234 60.0 157.4 12.00 7.0 APR
783 74.5 182.0 2.77 5.0 APR
987 84.5 176.0 5.00 6.2 AR
22. • Always collect most accurate data
– Would you collect:
• Age at operation
• Date of birth and date of surgery
• Patient was 83 years old at operation. He died
on Sep 12th 2012.
– How long did he survive after surgery ?
– How old was he when he died ?
23. • If you use groups, clearly define them and
assign codes (numerical or alphabetical)
Type of surgery
MAJOR Colon resection, Liver resection, stomach resection …
MINOR Appendectomy, cholecystectomy …
N stage
0 No positive lymph nodes
1 1-5 positive lymph nodes
2 >5 positive lymph nodes
24. • If you use intervals, define them
• What about the patient who is 30 years old ?
Age group
1 20 – 30 years
2 30 – 40 years
3 40 – 50 years
25. • If you use intervals, define them CLEARLY !
• What about the patient who is 30 years old ?
• Group 1
Age group
1 > 20 and ≤ 30 years
2 > 30 and ≤ 40 years
3 > 40 and ≤ 50 years
27. • Define how to handle missing data
– E.g. no data on tumor size. What do you enter ?
28. Assuring baseline comparability
• In order to avoid bias, even after
randomization, you should present relevant
baseline data comparison
29. Participant flow diagram
• Number of patients screened
• Number of included / excluded
• Separation into groups
• Dropouts (reasons ?)
• Number of patients per group available for
analysis
33. Presentation of data
• How to present data ?
• Present relevat dana to enable independent
statistical analysis and meta analysis
• Parametric data:
– Mean, standard deviation (SD) or standard error
(SE)
• Non-parametric data:
– Median, range
34. • Presentation in text
• Avoid presenting large amount of data in text
• Combine text and tables / graphs
• Do not repeat results in text and tables /
graphs
• Provide sufficient ammount of data !
35. • Two groups did not differ in height (p=NS).
• Two groups did not differ in height (p=0.320).
• Two groups did not differ in height (175 cm
[SD 12] vs. 178 cm [SD 11], p=0.320).
36. • Text :
– Two groups did not differ in height (Table 1).
• Table 1
Group A Group B P (two-
sided)
N Mean SD N Mean SD
Height
(cm)
25 175 12 35 178 11 0.320
37. • Presenting data in tables:
– Provide variable name and a unit of measurement
– Provide group size and mean (SD) or median (min,
max)
– Provide exact p value (avoid NS, p<0.05)
– Unless clearly defined in Methods, provide the
name of statistical test under the table (in
explanation)
– Use indexes to clarify abbreviations
38. • Presenting results in graphs
• Better visual appearence
• Easier to spot differences
• Does not provide exact data !
39. • Give general data in text and refere to tables
or graphs for details.
• Results should follow the logic of your study,
explain why you did certain comparisons and
what do the results mean
– e.g. “Pain scores were significantly lower in group
A compared to group B at every measurement,
except during activity at 24h and day 7 (Table 2).”
40. • Pictures:
– Use only if necessary !
– Consider schematics !
– Use color when you need it !
– Label the picture !
– Protect patient’s privacy !