1. CLINICAL TRIALS
THE DRUG CHARACTERISTICS ARE GENERALLY
ASSESSED THROUGH ANIMAL EXPERIMENTS OR
LABORATORY TESTS BEFORE THEY ARE
RECOMMENDED FOR USE
BUT, HOWEVER SUCCESSFUL THESE EXPERIMENTS
MAY TURN TO BE ,THE DRUG ULTIMATELY HAS TO BE
TRIED ON HUMAN BEINGS TO ASSESS THEIR
EFFICACY AS COMPARED TO THE EXISTING LINES OF
TREATMENT ,THEIR SIDE EFFECTS AND DOSAGES
SUCH TRIALS ARE CALLED CLINICAL TRIALS
2. IN A CLINICAL TRIAL THE EFFECT OF
EXPOSURE / INTERVENTION ON THE
OUTCOME ON A GROUP OF SUBJECTS IS
STUDIED
EXPOSURE / INTERVENTION: ----------
DRUG,DIET,SURGERY,EXERCISE OR
HEALTH EDUCATION
OUTCOME: ------------------------------------
RECOVERY , IMPROVEMENT,SURVIVAL,
INCREASE / DECREASE IN THE VALUE OF
THE PARAMETERS ETC.
3. EFFICACY:
IS A MEASURE OF THE BENEFIT RESULTING FROM AN
INTERVENTION FOR A GIVEN HEALTH PROBLEM
UNDER THE IDEAL CONDITIONS OF AN
INVESTIGATION.
IT ANSWERS THE QUESTION,
"DOES THE INTERVENTION DO MORE GOOD THAN
HARM TO PEOPLE WHO FULLY COMPLY WITH THE
RECOMMENDATIONS?"
4. EFFECTIVENESS:
IS A MEASURE OF THE BENEFIT RESULTING FROM
AN INTERVENTION FOR A GIVEN HEALTH PROBLEM
UNDER USUAL CONDITIONS OF CLINICAL CARE FOR
A PARTICULAR GROUP.
UNDER THIS EVALUATION, IN ADDITION TO
MEASURING THE EFFICACY OF AN INTERVENTION, IT
ALSO MEASURES ITS ACCEPTANCE BY THOSE TO
WHOM IT IS OFFERED.
THUS, EFFECTIVENESS ANSWERS THE QUESTION,
"DOES THE PRACTICE DO MORE GOOD THAN HARM
TO PEOPLE TO WHOM IT IS OFFERED?"
5. FOUR PHASES OF TRIALS :--
PHASE - I : -TOXICOLOGY , PHARMACO-
KINETICS,SAFETY ETC-ON HUMAN VOLUNTEERS
PHASE- II : TO STUDY TREATMENT EFFECT-
SMALL NUMBER OF PATIENTS
PHASE- III : RANDOMIZED (MULTICENTRIC )
CONTROLLED TRIALS ( RCT )
PHASE- IV : MARKETING THE DRUGS-
STRATEGIES AND MODALITIES-to study long term
effects
6. RANDOMIZED TRIALS --- STEPS:---
1) ESTIMATION OF MINIMUM SAMPLE SIZE
2) SELECTION OF PATIENTS
(a)EXCLUSION & INCLUSION CRITERIA
(b) COMPARABILITY OF SUBJECTS IN THE DIFFERENT
GROUPS
(c) CONTROL GROUP (PLACEBO) AND THEIR SELECTION
(d) METHOD OF SELECTION (APPROPRIATE SAMPLING
METHOD RANDOM ALLOCATION )
7. 3) TREATMENT SPECIFICATIONS
4) ETHICAL CONSIDERATIONS
5) FOLLOW-UP OF PATIENTS AND RECORDING OF
VARIOUS INFORMATION SYSTEMATICALLY IN
SUITABLY DESIGNED PROFORMAE
6) SORTING OUT THE PROBLEMS ARISING FROM DROP
OUTS AND WITHDRAWALS
7) DATA ANALYSIS APPLYING APPROPRIATE STATISTICAL
METHODS
8) INTERPRETATION OF THE RESULTS VALIDLY AND
MEANINGFULLY , MENTIONING THE DRAWBACKS OF
THE TRIAL AND CAUTIONING THE INTERPRETATIONS
8. SAMPLE SIZE - n
INFORMATION REQUIRED:---
(1) p1-SURVIVAL RATE WITH CHEMOTHERAPY
(2) p2-SURVIVAL RATE WITH RADIATION
(3) CONFIDENCE REQUIRED-C
(4) POWER REQUIRED-P
IF p1=60% , p2= 80%, C=95% AND P=80% n=85
IF 10% DROP OUT RATE IS EXPECTED , n=100
IF P=90% , n=110 ;
IF C=99% & P=90% n=165
9. SAMPLE SIZE
ADJUVANT STUDY OF THE DRUG-LEVAMISOLE FOR
COLON CANCER PATIENTS
RELAPSE FREE SURVIVAL:--
SUCCESS RATE WITH LEVAMISOLE ----60%
SUCCESS RATE WITH PLACEBO --- 50%
C= 95% ,P= 80%
n = 135 RELAPSES IN EACH GROUP
n’ ( NO. OF PATIENTS ) = 335 IN EACH GROUP
10. SAMPLE SIZE
MEDIAN SURVIVAL TIME :---
DRUG : PLACEBO-----1.35 : 1
C = 95% P = 80%
n = 140 RELAPSES IN EACH GROUP
n’ = 200 PATIENTS IN EACH GROUP
11. RANDOM ALLOCATION
(1) BIASED RANDOMIZATION
(ALTERNATE ,ODD / EVEN )
(2) BALANCED RANDOMIZATION(RANDOM NUMBER
TABLE FROM BOOKS OR COMPUTER GENERATED
NUMBERS )
STRATIFICATION :--
WITH RESPECT TO, SAY,AGE,SEVERITY ,TYPE ETC.
WHICH COULD AFFECT THE OUTCOME
12. DESIGN:--
(1)PARALLEL--:
a) TREATMENT - A
b) TREATMENT - B
(2) CROSS-OVER :-
A-----------------> B
( PERIOD-I) ( PERIOD-II)
B----------------- > A
(PERIOD-I) (PERIOD-II)
14. SHORT AND LONG TERM
CLINICAL TRIALS
IN A TRIAL TO FIND OUT THE EFFICACY OF A NEW
DRUG IN COMPARISON TO THE STANDARD DRUG IN
THE TREATMENT OF COMMON COLD OR INFLUENZA ,
THE OUTCOME IS EXPECTED WITHIN A SHORT TIME
BUT, IN TRIALS ON CANCER PATIENTS , THE
OUTCOME WILL TAKE LONG TIME TO SHOW
SPECIFIC PRACTICAL PROBLEMS MAY HAVE TO BE
FACED IN SUCH TRIALS
15. PRACTICAL PROBLEMS IN LONG TERM
CLINICAL TRIALS
1) NECESSITY OF DEDICATED INVESTIGATORS
BECAUSE OF THE LONG PERIOD OF STUDY
2) SYSTEMATICALLY MAINTAINED REGISTERS
3) DROP OUTS / WITHDRAWALS DUE TO SIDE EFFECTS /
PARTIAL IMPROVEMENT
4) PATIENT CONSENT & COMPLIANCE
16. 5) NECESSITY OF CHANGE IN TREATMENT DUE TO
ETHICAL REASONS
6) IF MULTICENTRIC TRIAL ,PROBLEM OF KEEPING
UNIFORMITY IN THE METHODOLOGY & EXECUTION
OF THE TRIAL AND DATA ANALYSIS
7) NECESSITY OF INTERIM EVALUATION
8) IN MULTICENTRIC TRIALS , COPING WITH
CONFLICTING RESULTS
9) SPECIFIC STATISTICAL METHODS TO ANALYSE THE
END POINT RESULTS---SURVIVAL ANALYSIS
18. PROGNOSTIC ( CO ) FACTORS
Effect of the Treatment could be related to many
variables :---
Sex,Age,Severity of disease,Duration of the disease,
Personality Variables( Diet,Smoking habit,Use of
Alcohol),
Clinical & Laboratary variables ( BP,Heart rate,
Blood Sugar level) etc.
While comparing the Response variable between the
Treatment Groups,the effect of these Co-Factors on the
Response Variable has to be studied and adjusted.
19. FACTOR SUB-GROUPANALYSIS
Response Group
SD ND
Improved 17(45.9% ) 20(60.3 % )
Not-improved 38 25
χ2 = 1.41 ( p =0.24 ) –Not Significant
Age Group
SD ND
15-30 20(36.4%) 30(66.7%)
31-60 35 15
χ2 = 7.92 ( p =0.005 ) --Significant
21. Statistical difference was not observed when the
analysis was done for the combined age group.
When the analysis was done separately for each age
group, the result was different. While there was no
significant difference in the response between the two
treatments in the younger age group , the difference
was statistically different in the older age group.
SD gave better response rate in the younger age group
( not statistically significant ) but, ND gave better
response rate in the older age group ( statistically
significant ) .
22. MULTIPLE REGRESSION ANALYSIS
( Quantitative Response Variable )
In Multiple regression analysis , the regression Model
is given as :---
Y = b0 + b1X1 + b2 X2 + b3 X3 + …….+ ei
Where the bis are the regression coefficients
of the Factors ( Co-variates ) - Xis
b0 is the regression equation constant
and the eis are the residuals
( difference between the observed & predicted value of
Y for each subject).
23. DATA FROM CLINICAL TRIAL ON
HYPERTENSION PATIENTS
NO. Gr. SEX AGE DIET SM BS BD AS AD RESP.
1 0 0 0 0 0 140 90 130 90 0
2 0 0 1 0 1 160 95 150 90 0
3 0 1 1 1 1 165 100 160 95 0
4 1 1 1 0 0 150 95 140 80 0
Group: 0-SD 1-ND Sex: 0-Female 1-Male
Age: 0-<40 1->40 Diet: 0-Veg. 1-Non-veg.
Smoking: 0-No 1-Yes Response: 0-No 1-Yes
24. RESULTS OF THE MULTIPLE REGRESSION
ANALYSIS
ASBP ADBP
Variable b p b p
Group -7.809 0.008 -4.655 0.007
Sex 4.353 0.142 -0.595 0.723
Age 2.738 0.335 1.448 0.379
Diet -4.428 0.161 1.559 0.389
Smoking 1.694 0.526 0.321 0.835
BSBP 0.650 0.0001 -0.053 0.330
BDBP 0.300 0.105 0.553 0.0001
Constant 17.766 0.425 46.697 0.001
R2 87.8% 84.5%
R2 ( Adjusted ) 69.7% 62.3%
26. The regression coefficient of -7.809 for the Treatment
group shows that the mean ASBP with ND ,after
adjusting for the prognostic factors is lesser by 7.8 units
compared to SD.
In case of ADBP ,the mean with ND was lesser by about
4.7 units , compared to SD .
As for the influence of the prognostic factors ,except the
BSBP in case of ASBP( p<0.0001) and BDBP in case of
ADBP ( p<0.0001) ,all other factors were statistically not
significant .
27. MULTIPLE LOGISTIC REGRESSION ANALYSIS
(Dependent variable---Qualitative )
If p is the probability of no improvement (unfavourable
event ) , Multiple Logistic Regression model can be written
as :---
Log [ p/(1-p)] =b0+b1X1+b2X2+b3X3+……….+ei
Where b1,b2,b3 etc are the logistic regression coefficients .
Log[p/(1-p)]is called the log –odds of No improvement.
28. RESULTS OF THE MULTIPLE LOGISTIC
REGRESSION ANALYSIS
Variable b p Exp(b)
Group -1.83 0.0986 0.1604
Sex 1.95 0.0987 7.0286
Age -0.41 0.7312 0.6636
Diet -1.00 0.4301 0.3679
Smoking 1.05 0.3580 2.8577
BSBP 0.06 0.1429 1.0618
BDBP -0.04 0.6063 0.9608
Constant -6.25 0.5134
Exp(b)---Quantification of the impact of the Variable
on the Response variable( improved / not- improved )
29. INTERPRETATIONS OF THE RESULTS
The negative value of b ( -1.83) for the Treatment group indicates
that the log –odds ( probability ) of no improvement is smaller
with ND compared to SD.
ie; probability of improvement is higher with ND than with SD
The value of Exp(b): --- 0.1604 for the Treatment group indicates
that the odds of improvement with ND was 6.23 times (1/0.1604)
higher than that with SD after accounting for the effect of the
prognostic factors on the response.
However the Treatment group and none of the prognostic factors
contributed significantly on the response variable.
30. INTENTION TO TREAT ANALYSIS
Drop-out,Withdrawal,change of Treatment (within or
outside the CT Protocol) —due to serious side
effects,general negligence etc.
Ideally to be avoided,but,in practice,may not be
possible
Affects the balance of Randomization and introduces
bias in the Treatment comparisons
To avoid this,analysis may be done as per the original
Grouping itself
This analysis is called –INTENTION TO TREAT
ANALYSIS
31. INTENTION TO TREAT ANALYSIS
May not sound Logical
Not recommended for all Clinical Trials
When Treatment A is not effective and for ethical
reasons another Treatment has to be given ,for the
benefit of the patient ,this type of analysis may be
recommended
32. INTENTION TO TREAT ANALYSIS
Coronary by-pass surgery & Medication for Unstable
Angina pectoris----Medication may not be effective
in some patients and for ethical reasons and keeping
the treatment for the benefit of the patient, coronary
by-pass surgery may have to be done in such patients .
Advisable to do the analysis in both ways---
ie; Original grouping & According to the grouping
after the change over
33. KAPLAN-MEIER
( ACTURIAL SURVIVALANALYSIS )
SURVIVAL PERIOD ,IN YEARS , OF 18 CANCER PATIENTS
OF WHOM 8 HAD RECEIVED CHEMOTHERAPY ( C ) AND
10 HAD UNDERGONE SURGERY ( S ) ARE :---
C:---2 , 2* , 4 , 4* ,6* ,8 , 8 , 8
S:---1 , 2 , 6* , 8* , 10* , 10* , 12* , 12* , 12* , 12
* CENSORED ( LIVING AT THE TIME OF LAST FOLLOW-
UP )
36. HAZARD FUNCTION CURVES( EXPONENTIAL )
THE INSTANTANEOUS RISK OF DEATH OR FAILURE
FOR INDIVIDUAL WHO HAS SURVIVED TO TIME ‘t’---
NEGATIVE SIDE OF SURVIVAL FUNCTION
COX PROPORTIONAL HAZARD MODELS
FOR COMPARISON OF SURVIVAL CURVES AND TO
TEST THE STATISTICAL SIGNIFICANCE AND
RELATIVE IMPORTANCE OF THE CO-FACTORS WHICH
INFLUENCE THE SURVIVAL EXPERIENCES IN THE
DIFFERENT GROUPS
38. ADJ. KAPLAN- MEIER
SURVIVAL CURVE
Kaplan-Meier survival estimates, by group
analysis time
0 5 10 15
0.00
0.25
0.50
0.75
1.00
group 1
group 2
39. COX-PROPORTIONAL HAZARD MODEL
VARIABLE HAZARD RATIO p
GROUP 4.2996 0.0914
ALCOHOL 0.7310 0.6842
SMOKING 1.1393 0.8825
DIET 1.0891 0.9290
NO SIGNIFICANT DIFFERENCE IN HAZARD RATIO
W.R.T. TREATMENT , ALHOHOL USE , SMOKING HABIT
AND TYPE OF DIET
WITHOUT ADJUSTMENT OF CO-FACTORS:-
HAZARD RATIO = 4.0934
p-VALUE:---0.0958
