Standard research is rarely conclusive because:
1 Changes in the cycling pattern can be confused with improvement.
2 A measure which elicits a good acute response can, over time, increase deterioration.
Problems in accounting for biochemical individuality
1 Intricate interactions of treatment with individual biochemistry and health are difficult to control for.
2 Responsive subgroups may not be readily apparent.
Issues specific to pharmaceutical studies
1 Data on adverse effects is often lost due to the high drop-out rate common in pharmaceutical studies (Horrobin 2002).*
2 The effects of drugs, polypharmacy, and drug discontinuation are not easily separated from effects of illness.**
3 The high cost of large double blinds largely shapes what is convincingly studied to proposals for which large pharmaceutical companies and other wealthy concerns are willing to pay.*** (Abramson 2004)
* Horrobin points out that typically, 40-60% of subjects drop out of six-to-eight-week studies of psychiatric medications (mostly to avoid side effects); in one-year studies, 60-90%. Benefits are thus validated for only a small percentage of patients.
** For example, in many of the early studies, lithium was abruptly withdrawn from controls, a practice now known to promote mania. (Ketter 2001, Bauer 1996, Gold 1987)
*** Abramson (2004) claims that studies financed by drug companies are five times more likely to find in favor of the company's drug of choice.
To limit such consequences, current mainstream studies often compare a proposed medication to one already in use. This approach, unfortunately, makes conclusions about the new drug dependent on the lack of flaws in studies of the original drug.
This issue over depriving the control group also shapes nutritional studies. Doctors are loathe to deny bipolars what they believe are effective nutrients, for the sake of experimental results.
Evolving experimental design
Long-term naturalistic studies
Because of such factors, sufficiently controlled double-blinds studying bipolar treatment are unlikely. (Bauer 1996, Compton 2001) (So, for instance, despite prior "controlled" double-blind research, decades passed before recognizing that significant antidepressant use often fosters rapid-cycling, or otherwise worsens long-term outcome. (Post 2003, Ketter, Frye 2001))
Consequently, some mainstream researchers now recommend open, naturalistic, longitudinal studies, i.e., following patients over many years, probably decades. (Bauer 1996, Compton 2001 Ketter, Frye 2001)
This type of design particularly lends itself to exploring how bipolars fare over the long term, with multiple nutrients, tailored to health, diet, medication, evolving biochemistry, cycling pattern, etc.
3 Other clinical data, especially that with lengthy follow-up.
5 Research on nutritional, herbal, and medical therapies for confounding physical conditions (thyroid disorder, allergy, Candida, sugar imbalance, neurotoxicity, etc.)