Treatment Considerations And Bias

Dosages, potency, inclusion/exclusion, study setting

What is Bias?

Systematic (or introduced) error that leads to data distortion (skewing) and hence, to an incoherent conclusion Subject selection bias

Study design Random sampling Randomization – are groups equal Inclusion/exclusion criteria

Data Collection Bias

Method selected for data collection Measuring method

Validity and reproducibility of instrument used to measure

Blinding

Investigator Bias

Journal study is published in Drug company funding?– (don’t

automatically reject the study!) Value statements in results/discussion

that are unsupported by literature (citations, statistics from current study)

Statistical vs. clinical significance Use of the word “significance”

Bias with “Significance”

Obstructive symptom score in Men with BPH treated with Finasteride or Placebo

Group Baseline score 12 mo. Score Placebo 6.7 +3.5 5.9 + 3.8 1 mg Finas. 7.4 + 3.8 6.0 + 3.8 * 5 mg Finas. 7.0 + 3.6 5.1+ 3.6 * * Statistically significantly different from

placebo.

Confounding variables

Something the investigators did NOT do or failed to consider, which may have influenced the outcome/results. Pharmacokinetics Side effects Demographics of study population Pharmacogenetics – race, gender, age Compliance Dosages Study setting

Dosages

Must be administered in a dose likely to produce benefit

There is no single or standard relationship between the intensity of response and dose (dose response curve) which can be uniformly applied to all drugs.

Variation small increase in dose can cause large response large increase in dose can cause small response

Potency Because 1 drug is more potent than

another on a mg to mg basis, this does not imply clinical superiority.

Potency is not important when comparing efficacy if they are equivalent doses

Doses for active controls should also be comparable to those of the study drug

Use of fixed doses could be inappropriate

Fixed Doses

Problems with a fixed dose study might not be applicable to “real world” a therapeutic dose in one patient might not

be adequate in another. If patients have widely varying body

weights, the amount of drug each receives (on a mg to mg basis) could vary widely.

Therapeutic Concentration

For many drugs there is no proven exact correlation between the concentration of the drug and clinical response. Beta blockers- propranolol, metoprolol anti-inflammatories- ibuprofen, naprosyn Some antidepressants- Paxil, Celexa, Zoloft Cholesterol/Triglyceride lowering agents- Clonidine (BP), Ketoconazole (antifungal)

Therapeutic Concentration

For many, it is easier and more clinically relevant to measure the response, or another marker Warfarin- PT/ INR, Heparin- PTT Cholesterol lowering drugs- blood cholesterol Diuretics (Dyazide, Diuril)- urinary output Anti-diabetic drugs (actos, miglitol, Glipizide,

glucotrol, metformin)- blood glucose Antibiotics (PCN, TCN, Ery., keflex)- blood bacteria

levels

Therapeutic Concentration

For some drugs, assaying levels is expensive or inconvenient Free phenytoin levels vs. bound phenytoin

levels Drug concentrations can be used to help

indicate whether or not the patient has been compliant with medication regimen BUT, no guarantee of compliance with regimen.

Therapeutic Range If drug has a therapeutic range, then doses

should be employed which will assist patients achieving concentrations in the therapeutic range

Drugs with narrow therapeutic ranges: Digoxin, Lithium, Dilantin, Theophylline, Tegretol, Vancomycin, Gentamicin, Tobramycin, Procainamide, Lidocaine, Quinidine

Pharmacokinetic Properties

Absorption: Efficacy measurements taken at appropriate times, ie. after absorption

Bioavailability can be altered by: Concomitant ingestion of food- antibiotics Diurnal variation- pravastatin, prednisone Alterations in GI tract motility- Reglan (ac) pH differences- enteric coated: ASA,

bisacodyl

Absorption in relation to meals

Pen VK with food slows rate but not overall absorption

Pen G has decreased absorption w/ food Ketoconazole has increased absorption

w/ food Griseofulvin absorption is increased w/

high fat meal.

Onset of Action

Immediate release tablet/capsule has fast onset

Delayed release has slower onset Liquids have fast onset Injectables have fastest onset Transdermal patches have slow onset

Time Until Onset of Effect Differences

Nicotine patch- 2-4 hrs Clonidine patch - 6-12 hrs Duragesic patch 8-12 hrs

Timing of Dose in Relation to Blood Levels

Antibiotics should be dosed around the clock for maximum bioavailability

Gentamicin is renal toxic if levels are not allowed to drop before re-dosing

Nitrates need to have blood levels drop to a drug free period to be most effective

Time To Therapeutic Effect

Need blood levels drawn at appropriate time

Dependent on half life of drug Steady state achieved w/in 4-5 half lives Takes 4-5 half lives to remove drug from

body Peak and/or trough blood levels are

appropriate for some toxic drugs (Gentamicin, Vancomycin, Tobramycin)

Concurrent Medications

Subjects may take non study drugs: because they have another medical condition if they were taking them before this study began.

Consider drug interactions with study med increase bioavailability may make study drug

look more effective than it is affect underlying disease state being studied taken equiv. amounts in tx and control group? Do they contribute to adverse effects observed?

Study Setting- Inpatient Setting

The more rigidly controlled the environment is, the more difficult to extrapolate the findings to the outside population

Inpatient close supervision - improved compliance less exercise/ activity - more tests needed more evaluations needed - less protocol variation less protocol variation less environmental variability

Outpatient Setting

More environment variability Less compliance control More diet variability Less testing less supervision more protocol variation

Artificial Setting

Arranged environment (NH, VA hospital, clinical research labs, dorms, hotel rooms, prisons)

Used more often in studies which recruit volunteers.

This environment is primarily used to control a greater # of variables

Compliance of Drugs

“80% of patients take less than 80% of their prescribed doses”

Those with mild disease have worst compliance

Those with severe disease have moderate compliance

Those with moderate disease have best compliance

Compliance Bias Occurs when differences in tx lead to different

degrees of compliance by patient Bias exists when it can be concluded that one

of the drugs lacked efficacy when in actuality, it might not have been taken.

Example: Using Nitroglycerin oint vs. patch Ointment messy, more frequent applications Patch once per day at bedtime (easier) Compliance is usually a greater concern in

studies involving out-patients than in-patients

Methods to Measure and Assess Compliance

Observing patients take their doses pill counts use of a non-toxic inert marker in study med asking the patient directly and frequently measuring the concentration of drug in

blood stream use of an electronic device in the vial cap review of Rx records- refills, patient diaries Physiological evidence- HR, BP, urine color

How Can Non-Compliance Effect the Study Results

A drugs therapeutic response will be less than expected or absent

A drug will be considered less potent than it actually is

One drug could be assumed to be similar or less effective than another drug

Non-compliance increases the sample size needed to detect a difference between groups

How to Analyze Non-Compliance

Exclusion: Drop non-compliant patients from evaluation of the results.

Intent-to-treat :Include everyone even if they had their therapy altered (non-compliant) taking patients last score at time they

dropped out taking average score for entire group taking worst score for entire group

Number Needed to Treat (NNT)

NNT: number of individuals that need to be treated in order to prevent one adverse event or one outcome. NNT = 1

ARR Ex: study determine efficacy of drug

preventing cancer. Incidence of cancer in placebo 15%, in treatment group 5%

15%-5% = 10% 1/10% = 10=NNT (10 pts needed to treat to prevent 1 case of cancer

NNT= 1/ placebo - treatment group

Number Needed to Harm (NNH)

NNH= 1/ treatment- placebo group Ex: Headache occurred in 25% of

placebo patients and 75% of patients taking drug X.

The NNH = 75%-25% = 50% 1/0.5 = 2 Only 2 patients would need to be treated

with drug X in order to cause a headache occurrence.