Earlier quoted context omitted.
To be clear, this is referring to a very specific subset of police: "crisis intervention team" officers within the 3% of departments that use a "police-based specialized police response" to mental health issues.
That is not a correct reading of the quoted analysis. In tabular form: CIT 3% police-based specialized mental health response 12% mental- health-based specialized mental health response 30% no formal policy 55%
> CIT officers make up about 15 to 20 percent of patrol officers and receive 40 hours of training that focuses on scenarios derived from actual incidents. Officers receive extensive training in de-escalation techniques.
It's ambiguous whether "officers" in the second sentence refers to all officers or only CIT officers (I read it as the latter). Regardless, according to this analysis only 0.45–3% of all patrol officers receive this training that is specifically with regard to people with mental illness.
Edit in case my math is unclear: according to this research, 3% of all department use CIT or similar. Assuming 100% of officers in those departments receive the training, that works out to 3% of all patrol officers. Assuming only the ~15% of CIT officers receive the training, that works out to 0.45% of all patrol officers.