Scientific Research Design In Drug TrialsPosted on 02/21/2011
In drug trials, the sponsoring pharmaceutical company seeks to determine if a new formulation is effective in alleviating the symptoms of the targeted disease, is as effective as another of the firm’s medications that has lost profitability due to retired patents leading to generic manufacture and low-price competition, or is more effective than existing medications. The tests must also demonstrate that the new formulation is safe, based on guidelines set by the Federal Food and Drug Administration (FDA) of acceptable percentages of patients who experience certain negative side effects from taking the medication. Since people are genetically different and react differently to chemical alternations in their bodies, proving the safety to a high level of confidence is very difficult.
Drug trial researchers design double-blind, placebo-controlled tests that compare the responses of typically two randomized groups of patients to a treatment regimen: the control group (those patients receiving a placebo) and the experimental group (those patients receiving the medication whose effectiveness and safety are being studied). If a large sample size can be procured, a third group may be included whose members might be given a competing medication whose efficacy is known, making the placebo control group unnecessary: “Where there is no effective treatment available for a medical condition, novel treatments ought to be compared with placebo. Once effective treatment exists, however, novel interventions should usually be tested against best available standard treatment.” (Weijer, p. 70)
Single-blind studies mean that only the subject is unaware of the nature of the manipulated factors, but in double-blind studies, the subject or patient, the researcher, and the outcome assessors, or data collectors, are all unaware of the nature of the administered treatment and of which group received medication or a sugar pill. The capsules or tablets administered to the members of both groups have an identical appearance, with no markings, except possibly an encoded number whose code is broken only after the completion of the testing.
The nature of double-blind studies reduces the researcher’s bias for expected outcomes, which might otherwise be inadvertently signaled to subjects. The more information that subjects or patients know, or think they know, the more likely they are to change their response to align with what they interpret is expected by the investigators or assumed to be normal.
It is important to note that the subjects in both groups are experiencing the same medical conditions or disease. All who agree to participate in the trial must cease taking all medications for their condition so that a true and accurate result can be achieved from the study. They all have a vested interest in seeing that a new and better treatment may become available to them soon. They agree to the deprivation of their medications as a criteria for participation in the study, which indicates that they have a higher than normal level of desperation that a cure be found and that they lack trust in the effectiveness of their current treatment to improve their condition, which, in some cases, may be a life-threatening. They all also have been informed that there is a 50/50 chance that they have been assigned to the placebo-control group, whose members are receiving no active pharmacological agents to combat their disease. Subjects want the new drug to cure their condition, and they may make themselves believe that the drug has worked, reporting that it has, and their bodies may even produce the hoped for release of corrective antigens or hormones.
Since the subjects are sick patients, some question the ethics of forced discontinuance of current medication and replacement with placebos.
The placebo effect operates in both the placebo control group and the experimental test group, because randomized participants in the experimental group will feel the same psychological improvement levels, ideally, as those in the placebo control group. “The use of a placebo control group balances the placebo effect in the treatment group, allowing for independent assessment of the treatment effect” (Schulz and Grimes, p. 698). The percentage of those reporting improvement in the placebo control group is assumed to be the same as the percentage of those in the experimental group, so this percentage is factored out, and the remaining percentage in the experimental group reporting improvement is the actual percentage of improvement presumed to be caused by the new medication under study.
When the difference is big enough to have 95% certainty that the drug effect isn’t simply due to chance, that is, when it is “statistically significant,” it is presumed that the medicine has a positive effect. But researchers must also allow for individual patient psychology, normal disease regression, latent variables introduced outside of the experimental structure, such as the participant’s biology, and other uncontrolled or unknown factors that impede equivalency among members in assigned groups. The rate of improvement in both groups should ideally be at least the same, and if the medicated group sees a vast decline in patient health, the medication is causing far worse physiological damage in side effects than the value it can provide as a treatment, or the cessation of other medication has caused negative physiologically responses that must also be identified and statistically factored out.
For more on scientific methodology, see Atheism.
Pitfalls in Scientific ResearchPosted on 02/21/2011
Limitations in scientific research are self-limitations, because researchers know what the confounding factors are, but within the scope of a study’s funding and time, they decide not to set up controls to limit those factors.
In research studies, there can be a number of confounding factors, which include investigator bias and small sample sizes that limit the general application and validity of the conclusions that can be drawn from a study. In general, a confounding factor is any extraneous variable in a statistical model that correlates (positively or negatively) with both the dependent variable and the independent variable, making the supposition of causal relationship unreliable, uncertain, and possibly erroneous, posing a major threat to the validity of inferences made about cause and effect.
In designing experiments to study the effectiveness of a new drug, for instance, the following factors must be considered as possibly being in play:
A patient’s original medical condition may have been misdiagnosed, so subsequent tests after dosing yield negative results for correction of the condition that the drug was intended to treat. If a placebo has the same effect as arthroscopic knee surgery, then either the surgery was unnecessary to the patient who received it or the condition of the knee of the patient who received the placebo didn’t merit corrective surgery.
Time in between trials may have caused the misdiagnosed condition to propagate and/or subside due to lower or higher healing rate that would be unexpected for the condition that the drug is hypothesized to treat.
When testing symptoms following introduction of medicine, researchers misinterpret measuring device readouts; symptoms masked underlying, undiagnosed medical condition; or the wrong measure was used to measure the suspected but incorrect condition. The experimenter’s level of expertise in conducting unbiased studies is a variable that leads to questionable results.
Patients can have multiple undiagnosed conditions, some amenable to placebo, some amenable only to valid and effective drug therapy or surgery.
The patient’s medical condition was psychosomatic.
Happiness at the prospect of cure or alleviation of pain caused the body’s release of cortisol, norepinephrine, endorphins, opioids, and other pain-receptor blocking hormones, which can be sampled and tested to rule out the cause as being the medicine, if the experiment were properly structured.
Subjects are lying to delude themselves and make the doctors happy (fulfilling demand characteristics that oblige people to respond in kind when a doctor or researcher seems to be doing them a solid favor by permitting participation in the trial).
All self-reports of pain reduction are suspect and based on expectations that pain should have receded.
Subjects were lying to begin with to gain attention for themselves (weak Munchhausen Syndrome effect).
The medical condition spontaneously remitted, or the body healed itself over time, or there was a natural regression of pain. Many diseases run a natural course, are self-limiting in duration, and are cyclical in their harshest effects, so seeming to improve can’t be attributed either to the placebo or the new drug’s effects.
Relief of symptoms are confused with cessation of a disease’s progress.
Some subjects secretly continue to take their prescribed medications during the trial, to hedge their bets or double their chances of getting better.
The subjects in the study were declaring pain to scam their insurance companies and doctors to both pay for and obtain prescriptions for pain medication because the subjects enjoy the effect of the drugs—they are drug-seeking and thought that participation in the study would prove to health insurance providers that they are actively seeking relief for their pain. They will pretend nonrelief, regardless of whether they are in the control group.
Most results favoring the conclusion that placebo effects are statistically significant are based on flawed clinical trial and research methodology and failure to account for lies in self-reports and testimonials. Flawed and limited thinking make determining an actual physiological effect from a placebo difficult, regardless of tests that measure chemical compositions in bodily fluids (the act of being tested causes a level of stress in some subjects that could incite the release of the exact chemicals to be measured as alleged to result from a placebo effect). “The regressive fallacy is the failure to take into account natural and inevitable fluctuations of things when ascribing causes to them.” (Gilovich, p. 26)
People forget that results must be statistically significant; that we are not genetically identical and so will respond differently to drugs, some of us healing slower, which would have us wrongly placed in the unhealed category at the conclusion of a study; that not all drugs work on all people who have the same medical condition; that the human response to disease is too complex to be assigned causality or cure to placebos; that the alleged medical practice/elixir works only when the patient is aware that it is being administered, making the results dependent on the strength of each patient’s belief system and immune system response that branches from belief; that the results from multiple administrations of a placebo over time often do not reproduce the same response within the same subject or across all subjects given specific circumstances; that environmental factors play a role in health; and that special-interest groups who pay for research pressure researchers to support illegitimate claims about a drug’s effectiveness.