For Pediatric Residents and Medical Students
General Pediatrics, the
Basic Steps in Clinical Research
1. Find a research topic
·
Based on motivation, knowledge, observation and analytic thinking
·
Answer one question and solve one problem each time
·
Knowledge: experiences based upon observations, internal thought
processes
2. Support the topic: Review the following information:
Systematic review of a clinical topic
·
Text books and clinical guideline- standard care
·
Reviews, summaries, or meta-analyses
Meta-analysis combines data from different studies to increase the statistical power of the findings and to arrive at a summary estimate.
Literature - new trends
·
Top news in Internet- most recent trends
·
Journals- usually at least half to one year delay
·
Review – usually 2-3 years delay
·
Read the top news and the medical literature intelligently.
·
Take notes whenever read and save time later.
“A single study may present only one piece of a complex clinical puzzle, and different studies of the same therapy or intervention may reach conflicting conclusion.”7
3. Generate a hypothesis to formulate new ideas that, if supported, will
solve the problem.
4. Design a procedure to test the hypothesis.
·
Time consume: rare
case vs. common medical problems.
·
Do not include
extraneous procedures that are not directed toward the research problem at
hand.
·
Find fact from fancy
in everyday life.
·
Quantifiable, if
possible: more objective in nature; allow for greater simplicity; make the
research results more readily understandable.
·
Experimental design vs. observing design.
·
Affirms vs. rejects the hypothesis.
5. Execute the research procedure, collect data and interpret data in the objective,
uniform methodical manner.
6. Conclusion based upon the information.
·
Clarity of result.
·
Simplicity of
conclusions.
·
Truthfully reported
whether or not the outcome coincides with the researcher's expectations.
Research & Statistics
1.
Observational studies: Investigate the causes of diseases
·
Cohort- prospective studies: Sample are chosen on the basis of the presence or
absence of a suspected cause and then followed over time to compare the
frequency of disease development in each sample.
·
Historical cohort:
Conducted totally on the basis of past records.
·
Case control-retrospective studies: Sample are chosen on the basis of presence or absence
of disease and compared for possible cause in their past. The choice of
controls in case -control studies is critical.
Chronology
(time)
Geographic
distribution- e.g. hospital vs. population based (place)
Matching
for demographic characteristics (persons) such as
Genetic, Age, Sex, Ethnic group
Physiologic
state- Intercurrent or preexisting disease
Prior
immunology experience
Human behavior- Hygiene, food handling, diet, tobacco
& drug,
Interpersonal contact, occupation, recreation,
Utilization of health resources
Other
factors
2.
Experimental studies: Intervention (stimulus or treatment) on study unit, then observe
response.
·
Randomization to
allocate study units to the different treatment groups.
Ideally, a randomized, controlled trial will constitute two or more groups that are entirely comparable in size and other characteristics. To maintain balance in-group size, researchers often employ block randomization. Within each block, an equal number of assignments are made to each of the two or more groups (treatment and control) in entirely random sequencing.
·
Clinical trial- study
units are people.
·
Choice of control:
placebo.
·
Health-care workers,
and study personnel blind to the intervention: single blinded vs.
double-blinded.
Blinding is an important step that ensures that the finding of a study are accurate. Unblinded
studies are more likely to be subjective and more likely to be positive. “The blinding process
should be described in the same level of detail as the randomization process so that readers can
determine how effective they find it to be.”7
·
Ethical consideration
playing an important role.
·
Consider compliance
and adherence to a regular regimen.
In general, losing more than 10% to 15% of patients to follow-up bring into question the validity of finding.
3.
Population and sample
·
Definition: Target population-
whole group of study units
Study population- possible study units
Study unit- individual study member
·
Select sample from a
population
Simple random sampling
Stratified
random sampling- population is heterogeneous.
Random
cluster sampling
Split-plot design
·
Sample size
The reasonable larger the sample the better.
Small size- may be more difficult to arrive at
significant conclusions.
Too larger size- time consume and economical difficult.
Research & Statistics
4.
Number needed to treat (NNT)
·
The number of patients who need to the treated
before a statistically significant benefit can be expected to accurate to one
additional patient. Expressed as 1 over the risk difference (1/RD).
·
The risk difference (RD) involves subtraction
(absolute difference); the relative difference is a ratio.
·
“Consider the risk of hospitalization for RSV
among premature infants treated with palivizumab. A
study found that 10.6% of patients treated with placebo were hospitalized,
compared with 4.8% treated with palivizumab. This
difference can be reported as a 5.8% risk difference (subtracting 4.8% from
10.6%- absolute difference) or as a 55% relative reduction (relative
difference) in the risk of hospitalization.”7
Treatment
hospitalized rate |
Placebo hospitalized rate |
Relative difference |
Absolute difference |
Number needed to treat NNT (1/RD) |
4.8%
|
10.6% |
55% |
5.8% |
1/5.8% = 17.24 |
The efficacy of treatment vs. placebo7
Treatment
cure rate |
Placebo “cure” rate |
Relative difference |
Absolute difference |
Number needed to treat |
|
20% |
10% |
50% |
10% |
10 |
|
2% |
1% |
50% |
1% |
100 |
|
0.2% |
0.1% |
50% |
0.1% |
1,000 |
5.
Data
·
Interval, ordinal, and
nominal scales
·
Quantitative and
Qualitative data
·
Continuos data: arise
only from quantitative data and measured on interval scale.
·
Categorical or
discrete data: arise from quantitative and qualitative traits and measured on ordinal or nominal scale
6.
Descriptive statistics: summarize the essential characteristics of a set of data.
Table
·
Relatively simple
& easy to read.
·
Title- clear, concise,
and indicate what is being tabulated.
·
Give units of measurement.
·
Label each row and
column as appropriate.
·
Show the totals if
appropriate.
·
Explain codes,
abbreviations and symbols in a footnote.
·
If the data are not
original, their source should be given in a footnote.
Graph
·
Simple, clear,
accurate, consistent with its purpose.
·
Every graph should be
completely self explanatory, labeled with title scales, and explanatory keys.
·
The title is commonly
placed below the graph.
·
Generally proceeds
form left to right and from bottom to top.
·
Histogram; Bar graph;
pie graph; frequency polygon; cumulative plot; scatter plot or scatter diagram;
tree diagram.
7.
Proportions and rates
Proportions or percentages
Rates
·
Incidence = Only new disease cases
over a period of time divided by a population at risk, so, incidence means new
cases
·
Prevalence = Total number of disease
cases at a given time divided by total population, so, prevalence means all
cases
·
Relative risk (RR)= incidence of disease in an exposed group divided by incidence of disease in an unexposed group
·
Attributable risk (AR) = the incidence of disease in an exposed group minus the incidence of disease in an unexposed group
·
Secondary attack rate = number of exposed persons developing the disease within the range of
the incubation period divided by total number of persons exposed to the primary
case
·
Case fatality rate = number of individuals dying during a specified period of time after
disease onset divided by number of individuals with the specified disease
during that period of time
·
Odds = the
ratio of an occurrence of an event to the non-occurrence of that event. For
example: The odds of a person drawing the ace of spades from
a deck of playing cards is 1/51.
8.
Measures of central tendency
·
Mean- arithmetic average
·
Median- 50th percentile
·
Mode- one or more peak values
9.
Measures of spread or
variability
·
Range- largest value minus
smallest value
·
Standard deviation- square root of the
variance
10.
The standard normal
(Gaussian)distribution and confidence limits
·
68% observations are within 1 SD
·
95% observations are within 2 SD
·
Mean ( x ) + 2 standard deviation(SD) is often
calculated as a "normal
range" that contains about 95% of the population values.
11. Sensitivity and specificity: (test of VDRL for syphilis):
Disease present Disease
absent
Positive
Screening A=
True positive (90) B=
False positive (20)
Negative
Screening C=
False negative (10) D= True negative (80)
Sensitivity = A/A+C = 90/100= 90%
Specificity = D/B+D = 80/100= 80%
Positive Predictive value = true positives/ all those
with a positive test result
= 90/110=82%
Negative Predictive value = true negatives/ all those
with a negative test result
=
80/90=89%
Sensitivity
= Persons with the disease and positive screening test divided by Number of
persons tested with disease ( x100)
i.e.,
sensitivity of a test means it gives a positive finding when the person has the
disease.
Specificity
= Persons without the disease and negative screening test is divided by Number
of persons tested without diseases ( x100)
i.e.,
specificity of a test means it gives a negative finding when the person does
not have the disease.
12.
Statistical tests
·
F test: Test for differences
between groups
·
T test: Test for two groups, pair,
matched pair with normal distribution
·
Chi-square test: This test most commonly
used for differences between proportions, i.e., comparing effects of two
different medications used in two different groups
·
Chi- Square McNemar's test: 2x2 table
·
Chi- Square goodness-of-fit test: test whether a sample of data is
consistent with any specified probability distribution
·
Rank sum test- two populations with
similarly shaped distributions
13.
Hypothesis testing- Null
hypothesis
When
we compare two groups in a study and notice some differences between them, the
null hypothesis may suggest that the observed differences are due just to random variations in the data.
·
Type I (alpha) error means null hypothesis is
true, but rejected
Which is akin to a false-positive result in diagnostic testing
The conventional standard is that
a type I error should occur no more often than 5% of the time (expressed as
P< 0.05).
·
Type II (beta) error means null hypothesis is
false, but accepted
Which is akin to a false negative result in diagnostic testing
A type II error occurring in determining the sample size
Typical power of study is 80%, which means in a negative study of 80% power, there is 20%
chance that, in fact, the differences between groups are real even though they are statistically
insignificant.
14. The P value:
·
P value <0.01 means the result is
statistically highly significant because the probability of random variation
alone is very small. There is <1% chance that the observed results are due
to chance alone.
·
P value <0.05 means " There is
<5% chance that the observed results are due to chance alone.
·
The meaning of "significant" in terms of
probability is far different from biological significance.
·
The power of a statistical
test can be
increased by
One
sided test
Large
sample size- even a biologically trivial difference can be made to be
statistically
significant
Detect
a larger difference
Higher
statistical significant level and increase alpha error
Decreased by
greater variability due to measurement error
heterogeneity of study units
15.
Correlation and regression
·
Covariance- a measure of how two
random variables vary together
·
Correlation- Simple liner
Multiple
correlation
Partial
correlation
Rank
correlation
·
Regression- Simple liner
Multiple
Logistic
·
Survival times and survival
curves
Other Glossary
95% Confidence interval: The range of values within which one can be 95% confident that the true result lies.
Block randomization: A procedure by which subjects are randomized in blocks such that, at the completion of each block, equal numbers will be present in each arm.
Effectiveness: Answers the question of whether a given therapy works under real world circumstances.
Efficacy: Answers the question of whether a given therapy can work under optimal circumstances.
Evidence-based medicine (EBM): The conscientious, explicit, and judicious use of current best evidence to make a decision about the care of individual patients.
Meta-analysis: The statistical combination of many, different studies to arrive at a single summary estimate of an effect.
References
1. Cacha CA. Research design and statistics for the safety and health professional.
1st ed.
2.
Glantz SA. Primer of
Biostatistics. 4th ed.
3.
Armenian HK, Shapiro S. Epidemiology and health
services. 1st ed.
4.
5.
Elston RC, Johnson WD.
Essentials of Biostatistics. 2nd ed.
Davis Company, 1994.
6. Sackett DL, Haynes RB, Guyatt GH, Tugwell P. Clinical Epidemiology : A basic
Science for clinical Medicine.
7. Christakis DA. Your turn to learn to read: Evaluating articles about treatment in the
medical literature. Contemporary
Pediatrics. 2003;20:79-95.