Allopathic medicine presents itself as fundamentally different from traditional medicine. Traditional medicine developed largely through experience, observation, and trial and error: a person became ill, a remedy was administered, an outcome was observed, and practices evolved through accumulated experience.
Allopathic medicine, by contrast, presents itself as science-based. Diseases are said to be identified scientifically, medicines are characterized scientifically, analytical methods are validated, clinical trials establish safety and efficacy, and regulators evaluate the scientific evidence.
I am familiar with this framework. I spent approximately three decades as a research scientist at Health Canada, focusing on pharmaceutical assessment and analytical methodology. My concern is not that requirements called characterization, specifications, validation, quality control, bioavailability/bioequivalence, and clinical testing do not exist. They do. My concern is more fundamental: when we look beneath these labels and ask what has actually been measured and experimentally established, does the scientific substance match the terminology?
A Scientific Label Does Not Establish Science
Consider pharmaceutical analytical testing. Regulatory language includes identity, purity, potency, accuracy, precision, specificity, reference standards, and validation. These are legitimate scientific concepts, but their presence in a guideline does not itself establish the validity of the underlying claim.
The scientific inquiry should begin by asking: What exactly is being measured? How was the
Does the procedure actually measure the property represented by its name? Was validation performed against an independently established reference, or merely against another procedure or predefined criterion?
A method may produce repeatable numbers, acceptable precision, and a linear response. Those observations establish properties of the procedure.
They do not automatically establish that the biological or medical entity attributed to the signal has itself been correctly identified. Validation cannot rescue an inadequately established analyte. Regulatory compliance and scientific characterization are not necessarily the same thing.
Drug Dissolution: A Concrete Example
Drug dissolution testing illustrates the distinction. For tablets and capsules, a dosage form is placed in a specified liquid medium under controlled temperature and agitation, and drug release is measured. The equipment can be standardized, the analytical procedure validated, and the specification met.
But if Product A is tested under Condition A and Product B under Condition B, what caused a difference in their dissolution profiles—the products or the different experimental conditions? Once both the test object and test conditions change, their effects become confounded. Precision does not correct experimental design.
Product-specific dissolution methods can serve a quality-control purpose by showing that successive batches behave consistently against an established specification. But that is a different claim from demonstrating physiologically relevant drug-release behavior.
Product-dependent testing conditions can undermine that interpretation because the physiological environment is product-independent — the human gastrointestinal tract does not change its conditions simply because a different pharmaceutical product is administered.
Therefore, a product-specific regulatory quality-control procedure should not automatically be treated as a scientifically validated model of what occurs in the human body.
Clinical Trials and the Upstream Question
Clinical trials are routinely described as the “gold standard” for determining whether a medicine works. Subjects receive an intervention, outcomes are observed, groups are compared, and statistics estimate whether differences are associated with treatment. This is more structured than casual observation, but the basic pattern remains: intervention → observation → statistical comparison.
Randomization, blinding, placebo controls, larger samples, and statistical analysis can reduce bias and help manage variability. They do not eliminate the substantial inter- and intra-individual variability inherent in human subjects.
This baseline variability—the experimental “noise”—may be large relative to the treatment effect being measured, making reliable interpretation of the treatment signal difficult. More importantly, statistics cannot answer scientific questions that were never adequately established upstream.
What exactly is the illness? What physical or chemical abnormality defines it? How was that abnormality measured? What does the endpoint represent?
Does it directly measure correction of the claimed abnormality, or is it a symptom, score, surrogate, diagnosis, or other indirect outcome?
Statistical analysis can quantify group differences; it cannot by itself establish the physical identity of an inadequately characterized entity or validate an uncertain premise.