Of the 16,370 stays in segregation, 1,314, or 8 per cent lasted for 15 days or longer, the maximum duration stipulated in the Mandela Rules. This equates to 13.4 instances of segregation of 15 days or more per 100 prisoners. Women were much more likely (69 per cent) than men to be segregated (269.4 instances per 100 women, compared to 159.7 instances per 100 men), and for longer times: 20.7 for instances of segregation lasting 15 days or longer per 100 prisoners for women, compared to 12.9 instances for men, (or a total of 138 instances for women and 1176 for men). By contrast, in England and Wales, there was little difference between men and women in the use of segregation: 38.1 instances per 100 of women, compared to 36.8 per 100 for men. New Zealanders of non-European descent were much more likely to be segregated compared to their counterparts of European descent. Between May and October 2016, Mäori and Pacific Islanders made up approximately 80 per cent of Directed Segregations (Management units and Disciplinary segregation). By comparison, New Zealanders of European descent accounted for a mere 15 per cent of prisoners in Directed Segregation. The picture was rather different in At Risk units, where the same ethnic groups made up approximately 54 per cent and New Zealanders of European descent made up 37 per cent of the population.26 Taking all forms of segregation together, Mäori and Pacific Islanders made up for approximately 62 per cent of those segregated, and New Zealanders of European descent made for 30 per cent of the segregated population. While these figures are almost identical to the representation of these groups in prison more generally (62 per cent and 32 per cent respectively), they are grossly disproportionate to Mäori representation in New Zealand’s population, where Europeans make for 69 per cent of the population, Mäori make for 14.6 per cent, and Pacific Islanders make for 6.9 per cent of the population. Forms of restraint According to data provided to us by the Department of Corrections, in the six months May-October 2016, a total of 423 incidents involving the use of mechanical restraint were recorded, some involving the use of more than one form of restraint. The recorded incidents primarily involved the use of handcuffs, but they also included five uses of a restraint bed; 9 uses of spit hoods; 16 uses of waist restraints, and; three uses of head protectors. This data needs to be approached 26 Analysis of data provided to this review by Corrections, though the number of instances recorded compared to the total number of segregations instances set above suggests that these datasets are incomplete. 26