Other issues

### ITT (interntion-to-treat) analysis
- includes all subjects.
- account for treatment effects, difficulties in administering the drug and compliance issues

### PP (per protocol) analysis
- excludes subjects who stop being compliant
- evaluate the maximum benefit possible from a treatment, given perfect compliance

### Subgroup Analysis
- examine whether the treatment effect or side-effects of treatments are the same or greater in patients with a specific feature or risk factor so that more specific treatment decisions can be made
- multiple comparison

### Regression Analysis
- adjust for differences in baseline characteristics
- identify the predictors of an outcome variable
- identify prognostic factors while controlling for potential confounders
- determine prognosis (prognositic models)
- determine diagnosis (diagnostic models)

### Confounding
- confounder is associated with the outcome and is unequally distributed between the treatment goups
- use stratification and regression mdoeling to control confounders

### Missing Data
1. Missing completely at random (MCAR)
2. missing at random (MAT)
3. missing not at random (MNAT)

### Interim Monitoring and Stopping Rules

# main reasons
- ensure that adverse event frequency and toxicity levels are acceptable
- ensure that patients are not recruited into a trial that is going to be unable to reach a definitive result
- ensure that randomization of patients is stopped as soon as there is sufficiently clear evidence either for or against the treatment being evaluated
- address unexpected problems with the study protocol such as exclusion criteria delaying recruitment

# most popular statistical methods - group sequencial methods
- Pocock
- O'Brien and Fleming
- DeMets

Cluster Randomized Trials

Randomization unit is not individual subject but group of subjects

Design Effect: 1+(m-1)*ICC
ICC: intra-cluster correlation coefficent

#Advantage
- optimal design for evaluating qualit improvement strategies in health care intervention and education program studies
- account for contamination between patients within a cluster
- easier to administrate the randomization and centers

# Disadvantages
- larger sample size is required than for a simple randomized trial
- the patients and clinicians may recognize whether they are in the active or placebo arm
- clinicians might transfer information between clusters

Equivalence Trials

- to check that change in formulation does not change the efficacy of the compound
- to assess whether generic and original drugs have identical therapeutic effectiveness

### Clinical Equvalence
- based on clinical outcomes

### Bioequvalence
- based on pharmacokinetics (PK) parameters, which is clearly defined and have lower variability
- basic assumption: the same # of drug compound molecules occupying he same # of receptors will have similar clinical effects.

### Noninferiority Trials
- where superiority trials would not be appropriate due to ethical or practical constraints
- where efficacious treatments already exits and the demonstration of noinferiority, as well as secondary benefits, is sufficient for regulatory approval.
- to demonstrate that the new treatment is not worse than the comparator by more than a specified margin, delta.
- the choice of delta is critically important and needs to be prespecified and justified when designing the trial.

Factorial Design

# Consider more than 2 treatments at a trial

# Advantage
- cost
- sample size
- exploring interaction effects

# Limitations
- interactions
- individuals randomized to several different interventions wil find it difficult to comply with treatment.

Crossover Design

# Paralled Study Design: each subject is randomized to one and only one treatment

# Crossover Design: each subject receives more than one treatment in a specified sequene

# Crossover Trials assumes that patients usually have a chronically stable condition that will not vary between when they are taking the 1st and 2nd treatments.

# p*q crossover design: p sequences of treatments administered over q different dosing periods.

### Advantages of crossover trials over parallel studies

# Treatment differences can be based on within-subject comparisons instead of between-subject comparisons (less variability is expected).

# require less sample size
N(cross over) = (1-r)*N(paralle) / 2 , where r is correlation coefficient

# a crossover design provides the least-biased estimates for the difference between treatments assuming that the response of subjects to treatment is consistent

### Main limitations of crossver trials

# crossover trials pose greater inconvenience to the subjects
# censored observations due to subject withdrawal have a higher impact
# subjects should be in a comparable condition at the start of each treatment period; it is more appropriate for chronic disease that have a stable set of symptoms, while it is not suitable either for acute conditions, for primary outcomes that are permanent or for terminal events (pregnancy or death).
# it requires fewer patients, but not for Phase III study in which needs large sample size anyway.
# carryover effect
# period effect (treatment by period interaction)

basics of pharmacokinetics

# Pharmacokinetics equations describe the relationships between the dosage regimen and the profile of drug concentration in the blood over time.

# Pharmacodynamic equations describe the relationships between the drug concentration-time profile and therapeutic and adverse effects.

## Initial Treatment

# volume of distribution
volume (V) = D / C
D: dose, C: concentration

# target concentration
Loading dose = (Target C - Measured C)*V

# salt correction factor (s)
# molar correction factor (m)

## Continuing Treatment

# Maintenance dose and clearance
Css (steady state concentation) = (s*m*IR) / CI
CI: clearance rate, IR: infusion rate

# Factors affecting clearance
# Bioavailability and oral maintenance dosage regiments
bioavailability: the proportion of the administered dose that reaches the systemic circulation
Css average = Dosing rate / CI = F*D*s*m / (CI*tau)
Oral Dose = (target Css average * CI*tau) / (F*s*m)
tau: dosage interval, F: bioavailability

## Stopping Treatment

# Elimination rate constant (k) and concentration-time profile
Ct = (D/V)*exp(-kt)
Ct: concentration at any time after the dose
D/V: maximum concentration that would be achieved

#Elimination half-life
t(1/2) = ln(2)/k

From "Back to basics: pharmacokinetics by Alison Thomson"