• How likely is it that my patient will experience an adverse drug reaction?
• How likely is it that my patient will experience an adverse drug reaction that is so serious that it may be life threatening?
• How will the risk of an adverse drug reaction vary with different doses of the drug?
• Are there specific parameters that should be monitored more closely in my patient?

Best information available upon which clinician can form an answer to these questions is the information gathered during pre-approval clinical trials

In usual case, any apparent finding emerges from assessment of dozens of adverse events, making description of statistical uncertainty very difficult. Approach taken is best described as one of exploration & estimation of event rates. It should be appreciated that exploratory analyses are a critical & essential part of a safety evaluation.

Safety evaluations focus primarily on estimating the risk of unwanted events associated and on risk of those events relative to what would be expected in the patient population as a whole if the drug were to be approved.

• Any untoward medical occurrence which does not have a causal relationship with the treatment
• Any unfavorable & unintended sign, symptom, or disease temporally associated with the use of a medicinal (investigational) product

• By study investigators on the basis of their own observations
• By the study participant as a self-reported event

Calculating the proportion of individuals reporting any AE for all treatment groups in a study enables us to see whether AEs are more or less likely in the test treatment group than in other groups

• To do this in a standardized manner it is necessary to “code” the AE descriptions
  
  • e.g., 25 participants received treatment A & 5 reported a headache, the proportion is 5/25 = 0.20, which can also be expressed as 20%

• Only 1% of participants in placebo group reported dizziness compared with 3–4% of participants treated with the active drug.
• How might a regulatory reviewer interpret these data?
• The first conclusion is, dizziness was not reported very often, so, if drug is approved, most patients treated would probably not have a problem. However, the difference might generate some concern.
• Initially, absolute difference may not seem extreme. However, in relative terms, those treated with the investigational drug are 3 to 4 times as likely to experience dizziness. This measure of risk is called a relative risk

Interval estimation is the standard error of estimator, which quantifies how much sample estimate would vary from sample to sample if we were to conduct same clinical study over & over again. Larger the sample size, smaller the standard error.

Reliability factor is determined by shape of the sampling distribution of interest & is the value that defines an area under the curve of (1-α). In case of a two-sided interval, reliability factor defines lower & upper tail areas of size α/2

• Is probability of experiencing a headache after treatment with active drug, higher than risk after treatment with placebo?

• Step 1: Point estimate = 0.06-0.08=-0.02
• Step 2: Standard error =
• Step 3: Reliability factor - 1.96 and for this interval, we use continuity factor 0.5(1/98 + 1/302) = 0.007
• Step 4: Confidence interval -
  Lower limit is -0.02-1.96(0.03)-0.007 = -0.09
  Upper limit is -0.02+1.96(0.03)-0.007 = 0.04