Some observers have hypothesized that the prisons were calmer, with fewer incidents of violence, because of the improved psychiatric care available through telemedicine. To test this hypothesis, the Bureau’s research division provided the research team with counts of incidents or acts of aggression (assaults) by inmates against either inmates or correctional officers -- for the year preceding the demonstration and for the demonstration period -- for each of the three Pennsylvania prisons.¹ Incident rates were compared using the monthly prison census as the denominator to account for differences in the size of the populations at each prison.

Major disturbances occurred in the prisons during a few months in the year preceding the demonstration and in the demonstration period. After removing the effect of these unusual months, a pattern emerges. At the two Allenwood facilities, but not at USP-Lewisburg, there were significantly fewer acts of aggression per inmate-month during the demonstration period than in the preceding year.² The decline began before the demonstration was implemented but continued during the demonstration period.

Similarly, researchers obtained counts of incidents in which correctional officers used force to subdue an inmate or defuse a situation. These two measures -- assaults and use of force -- may at times be counting the same incident in two ways (once from the perspective of the guard and again from the perspective of the inmate), while at other times they indicate different events. At USP-Allenwood and FCI-Allenwood, use of force began declining well before the demonstration began -- 6 months before at FCI-Allenwood -- and the new lower level continued throughout the demonstration. A new warden began service at FCI-Allenwood several months before the demonstration began, and her new practices regarding use of force may have contributed to the decline that was observed at that prison. In contrast, use of force increased slightly at USP-Lewisburg.

The research team concludes, with a 20-percent chance of measurement error, that fewer assaults occurred at the two Allenwood prisons after the demonstration began than in the prior year. It is possible that this decline was related to improved psychiatric care at the prisons, but it may also have been due to a combination of other factors that the team did not attempt to identify. Researchers also conclude that although use of force by guards declined at the Allenwood prisons, the decline preceded telemedicine by several months and was probably unrelated. Because neither pattern was observed at USP-Lewisburg, the team is unable to draw any consistent conclusion about the value of telemedicine or telepsychiatry in improving the social climate of the demonstration prisons.

¹ There were few homicides or suicides in any of the prisons -- too few upon which to base any conclusion.

² Rather than the usual precision level of 0.05 or 0.10 (that is, a 5- to 10-percent chance that a finding happened due to chance alone), the research team chose a 0.20 precision level for this analysis. With small numbers of very important events, the team believes it is appropriate to accept more risk of measurement error to be able to observe a change over time.