Skip to main content
. 2016 Oct 20;27(1):51–61. doi: 10.1017/S2045796016000731

Table 3.

Seclusion cohort: sample characteristics and association with seclusion status in multivariable analyses

Risk factor Seclusion cases (n = 990) Seclusion controls (n = 1032) OR (95% CI) p
Age: n (%)
<25 years 323 (32) 219 (21) 3.30 (1.85–5.90) <0.001
25–34 years 347 (35) 366 (35) 1.92 (1.12–3.28) 0.017
35–44 years 185 (19) 216 (21) 2.06 (1.15–3.69) 0.015
≥45 years 135 (14) 231 (22) (ref)
Female sex: n (%) 354 (36) 203 (20) 2.07 (1.35–3.17) 0.001
Ethnicity: n (%)
White 199 (20) 253 (24) (ref)
Black African/Caribbean 690 (70) 657 (64) 1.13 (0.711.79) 0.609
Other 101 (10) 122 (12) 0.71 (0.341.46) 0.347
Diagnosis: n (%)
Schizophrenia 323 (33) 435 (42) (ref)
Other psychotic 273 (28) 262 (25) 0.78 (0.501.23) 0.287
Bipolar disorder 283 (28) 217 (21) 1.14 (0.701.86) 0.590
Personality disorder 24 (2) 18 (2) 1.44 (0.424.94) 0.563
Other diagnosis 87 (9) 100 (10) 1.14 (0.592.23) 0.696
Admission status: n (%)
Informal 44 (4) 119 (12) (ref)
Civil section 417 (42) 170 (16) 2.29 (1.13–4.68) 0.022
Section 3/forensic 529 (53) 743 (72) 1.34 (0.692.62) 0.387
Time since admission: n (%)
≤7 days 395 (40) 98 (10) (ref)
8–21 days 235 (24) 215 (21) 0.24 (0.15–0.39) <0.001
22–60 days 204 (21) 380 (37) 0.18 (0.11–0.31) <0.001
>60 days 150 (16) 339 (33) 0.26 (0.15–0.45) <0.001
Behavioural precursors: n (%)
Abus* 570 (58) 231 (22) 1.37 (0.941.98) 0.100
Aggress* 654 (66) 192 (19) 1.85 (1.30–2.64) 0.001
Agitat* 714 (72) 286 (28) 1.99 (1.42–2.78) <0.001
Arous* 472 (48) 124 (12) 1.87 (1.29–2.71) 0.001
Assault* 229 (23) 25 (2) 3.31 (1.81–6.06) <0.001
Demand* 531 (54) 276 (27) 1.24 (0.891.72) 0.212
Hit* 266 (27) 41 (4) 1.98 (1.17–3.35) 0.011
Restrain* 554 (56) 67 (6) 6.54 (4.36–9.81) <0.001
Shout* 573 (58) 217 (21) 3.75 (2.58–5.43) <0.001
Threat* 689 (70) 190 (18) 1.53 (0.982.37) 0.060
Threw/throw* 328 (33) 77 (7) 1.17 (0.821.66) 0.385
Violen* 247 (25) 28 (3) 2.09 (1.17–3.73) 0.012

MHA, Mental Health Act; OR, odds ratio; CI, confidence interval.

*Indicates wildcard, used to identify words with 0 or more characters in that position.

Patient ID included as a random effect in order to account for clustering at the patient level. Results significant at the 0.05 level indicated in bold.