Biostatguide

Sorafenib in advanced hepatocellular carcinoma-Critique of the NEJM article

statguru — Tue, 03 Feb 2009 02:47:00 GMT

Hypothesis
Does sorafenib increase the survival in cancer patient with hepatocellular carcinoma?

Sorafenib- oral multikinase inhibitor of vascular endothelial growth factor receptor, inhibits tumor cell proliferation

Methods
Study population- patients with advanced hepatocellular carcinoma
None of these people received previous systemic chemotherapy
Used child pugh classification??? Why not MELD score?? This is a definite drawback.

Study design
Multi center, randomized double blind, placebo controlled
Phase 3 trial

400 mg daily sorfenib
Placebo

Primary outcome
Symptomatic progression
Decrease in 4 or more points from the baseline score on patients response
Death

Secondary outcome
Time to radiologic progression

Statistical analysis
Intention to treat analysis- analysis is based on the intention to treat in the beginning
This is to avoid the cross over and drop out effects.
Every one who begins he treatment considered as a part of the trial

Patients
902 patients’ screened
602 met the eligibility criteria
Overall survival
143 vs 178 in the sorafenidb vs placebo
Total of 321 died
Survival 44% vs 33% sorafenib vs placebo
31% relative risk reduction
Overall survival by cox proportional hazard significant
Radiologic progression was lower in sorafenib group

Discussion
Bottom line. You don’t have any choice. Take it.
FDA is dragging the foot in approving the cancer drugs, people are fed up
Not a great trial.
Some benefit is definitely there.

PROVE-IT trial: A critique

statguru — Thu, 15 Jan 2009 02:47:00 GMT

Introduction did not do justice this to article. Authors did a poor job in introduction, Not much information about the previous studies.

Well planned study. Patient population was selected from 8 different countries and impressive number of 4162 patients. Lot of ineligibility criteria about he previous statin use. A creatinine level of 2.0 is a cut off, why? Every one knows renal insufficiency people are at high risk

Pravastatin 40mg daily vs atorvastatin 80mg daily
Non-inferiority trial

10 day course of Gatifloxacin was also mentioned given to the patients. If you are mentioning in the trial, please tell us why?
When will the authors report this?

Average follow up time 2 years. Not good. 5 years may be better.

Statistical analysis - Looked at the sample size for 87% power, Please mention which software packet was used for analysis. This is not acceptable.

Primary endpoint-
Pravastatin group was 26.3%
Atorvostatin was 22.4%
Again primary end point was death
So clearly statistically significant benefit is here.

One other finding is clearly in favor is the reduction of C reactive protien
base line was 12.1
Pravastatin group it was 2.1
Atorvostatin group it was 1.3
but the dose difference is very significant. Pravastatin 40 vs Atorvostatin 80 mg

Now let us look at the HDL level
Pravastatin increased HDL by 8.1%
Atorastatin only 6.5%
P value is .001

Main points

Lower LDL goal is better.
Reduction of CRP is achieved by any statin.
Is there any benefit in increasing the HDL by drug therpay?- Probably not in this tudy group

Whats the best way to learn Biostats?

statguru — Thu, 08 Jan 2009 03:52:00 GMT

everybody asks me this question.
Answer is simple.

read the journals.
look at the methods section.

read the critique I post here.

NY times love affair with R

statguru — Thu, 08 Jan 2009 03:05:00 GMT

All praises for this powerful software.
Future is R according to NYT.

SAS are they worried?
They should be worried.

Some guy at SAS compared to R with an aircraft engine wihtout any guarantee.

he was stupid enough to say that o NY times correspondent.

A 4 year trial of Tiotropium in Chronic Obstructive Pulmonary disease: A critique of this article. UPLIFT trial

statguru — Wed, 07 Jan 2009 02:59:00 GMT

The authors did a good job in the back ground section describing the previous studies. It would been nicer it they can mention any studies comparing Ipatropium and Titropium

Methods
Good study design. sample size calculated.Multi center. did not mention the race percentage in the demographic section. Does it matter to the authors? Coe n guys, wake up
Demographic table 45% on anticholinergics. But in page 2 people on other anitchlinergic were excluded?

smokers changed their pattern during the study. So the result are screwed??

Over all head to head Tiotropium and Placebo - P value not significant

Where are the benefits.
significant decrease in exacerbations, hospital admissions.

Interesting findings
Decreased Mi in Tiotropium group

final advice to COPD patients

Take the drug

The standard deviation

statguru — Sat, 26 Apr 2008 15:16:00 GMT

What is standard deviation?

It is the measure of spread in a data.

The standard deviation is usually denoted with the letter σ (lower case sigma). It is defined as the square root of the variance.

A simple example

Suppose we wished to find the standard deviation of the set of the numbers 4 and 8.

Step 1: find the arithmetic mean (or average) of 4 and 8,

(4 + 8) / 2 = 6.

Step 2: find the deviation of each number from the mean,

4 - 6 = - 2

8 - 6 = 2.

Step 3: square each of the deviations (amplifying larger deviations and making negative values positive),

( - 2) 2 = 4

22 = 4.

Step 4: sum the obtained squares (as a first step to obtaining an average),

4 + 4 = 8.

Step 5: divide the sum by the number of values, which here is 2 (giving an average),

8 / 2 = 4.

Step 6: take the non-negative square root of the quotient (converting squared units back to regular units),

So, the standard deviation is 2.

The Variance

statguru — Sat, 26 Apr 2008 14:44:00 GMT

What is the variance?

answer is in the comments section

What is a Quantile?

statguru — Sat, 26 Apr 2008 14:33:00 GMT

This is same as percentiles.
It is a series of cut of points.
It divides the group of data into groups containing equal number of observations

Range

statguru — Fri, 25 Apr 2008 02:22:00 GMT

What is Range?

Measures of spread

statguru — Fri, 25 Apr 2008 02:18:00 GMT

Range
Quantiles
Variance
Standard deviation

Classifying distribution by the number of the mode

statguru — Fri, 25 Apr 2008 01:48:00 GMT

Unimodal - Distribution with one mode.

Bimodal - Distribution with two modes.

Trimodal- Distribution with three modes.

Computation of the Mode

statguru — Fri, 25 Apr 2008 01:38:00 GMT

Compute the mode of the following data set.

1 3 5 7 8 9 2 4 4 3 3 2 4 3

Mode

statguru — Fri, 25 Apr 2008 01:36:00 GMT

What is mode?

Negatively skewed

statguru — Fri, 25 Apr 2008 01:21:00 GMT

This means skewed to the left.

Positively skewed

statguru — Fri, 25 Apr 2008 01:20:00 GMT

This means skewed to the right

Distribution is symmetric

statguru — Fri, 25 Apr 2008 01:12:00 GMT

This means that position of the points are distributed on each side of the sample median are the same.

Weakness of the Median

statguru — Wed, 23 Apr 2008 02:21:00 GMT

What is the weakness of the median?

Main strength of the median

statguru — Wed, 23 Apr 2008 02:08:00 GMT

What is the main strength of the median?

Median

statguru — Wed, 23 Apr 2008 01:54:00 GMT

Calcualte the median of the following numbers

2 4 6 8 10 12

Median

statguru — Wed, 23 Apr 2008 01:49:00 GMT

calculate the median of the following data set

2 4 6 8 10