Sacks et al. (2005) reported the case of a 72-year-old man, described as professionally successful, intelligent, and cultivated, with polymyalgia rheumatica, who after being treated with prednisone developed a psychosis and dementia , which several behavioral neurology and neuropsychiatry consultants initially diagnosed as early dementia or Alzheimer's disease .  Large dosage variations in the patient's medication (including a self-increased dosage from 10 mg/day to as much as 100 mg/day for at least 3 months) produced extreme behavioral changes, from missed appointments to physical altercations, and eventually admission to a psychiatric ward and later to a locked Alzheimer facility. During this time, neuropsychological testing showed a decline in the patient's previously superior IQ as well as deficits in memory, language, fluency, and visuospatial function, which given the patient's age was considered to be compatible with early dementia. When the steroid treatment ended after a year, the patent's confusion and disorganized appearance stopped immediately. Within several weeks, testing showed strong improvement in almost all cognitive functions. His doctors were surprised at the improvement, since the results were inconsistent with a diagnosis of dementia or Alzheimer's. Testing after 14 months showed a large jump in Full Scale IQ from 87 to 124, but mild dysfunction in executive function, memory, attentional control, and verbal/nonverbal memory remained. 
Some of the approved drugs are synthetic versions of the natural hormones, such as trenbolone acetate and zeranol. Just like the natural hormone implants, before FDA approved these drugs, FDA required information and/or toxicological testing in laboratory animals to determine safe levels in the animal products that we eat (edible tissues). Furthermore, FDA required that the manufacturers demonstrate that the amount of hormone left in each edible tissue after treatment is below the appropriate safe level. As described above, a safe level is a level which would be expected to have no harmful effect in humans.