The effectiveness of the offender personality disorder pathway: a propensity score-matched analysis

Figures & data

Figure 1. Plot of the linear mixed model results for OGP scores. Analysis included a total of n = 19,440 individuals, with data from 3 years before to 3 years after referral to the Pathway.

Table 1. Propensity weighted linear mixed model results for OGP scores. Analysis included a total of n = 19,440 individuals, with data from 3 years before to 3 years after referral to the Pathway.

Group	Before/After referral to the Pathway	Estimated Slope	Standard Error	p-value	95% Confidence Interval
Comparison	Before	0.142	0.068	0.036	0.009, 0.274
	After	−0.786	0.077	<0.001	−0.937, −0.634
	Difference (after minus before)	−0.927	0.109	<0.001	−1.140, −0.714
Treatment	Before	0.135	0.083	0.105	−0.028, 0.298
	After	−0.402	0.073	<0.001	−0.545, −0.258
	Difference (after minus before)	−0.537	0.116	<0.001	−0.764, −0.310
	Difference of difference (Treatment minus Comparisonl)*	0.390	0.171	0.023	0.054, 0.726

*Effectiveness of the Pathway in terms of OGP scores.

Figure 2. Plot of the linear mixed model results for OVP scores. Analysis included a total of n = 19,440 individuals, with data from 3 years before to 3 years after referral to the Pathway.

Table 2. Propensity weighted linear mixed model results for OVP scores. Analysis included a total of n = 19,440 individuals, with data from 3 years before to 3 years after referral to the Pathway.

Group	Before/After referral to the Pathway	Estimated Slope	Standard Error	p-value	95% Confidence Interval
Comparison	Before	0.234	0.069	0.001	0.098, 0.370
	After	−1.074	0.078	<0.001	−1.220, −0.921
	Difference (after minus before)	−1.309	0.115	<0.001	−1.535, −1.082
Treatment	Before	0.243	0.071	0.001	0.104, 0.383
	After	−0.777	0.074	<0.001	−0.922, −0.631
	Difference (after minus before)	−1.020	0.112	<0.001	−1.239, −0.801
	Difference of difference (Treatment minus Comparison)*	0.288	0.173	0.095	−0.051, 0.627

*Effectiveness of the Pathway in terms of OVP scores.

Table 3. Propensity weighted negative binomial regression for adjudication incidence counts. Analysis included a total of n = 12,998 individuals, with data from 3 years before to 3 years after referral to the Pathway.

Group	Before/After referral to the Pathway	Incidence Rate	Standard Error	p-value	95% Confidence Interval
Comparison	Before	1.123	0.026	<0.001	1.072, 1.173
	After	0.918	0.020	<0.001	0.880, 0.957
	Incidence rate ratio (after/before)	0.818	0.023	<0.001	0.772, 0.864
Treatment	Before	1.346	0.044	<0.001	1.259, 1.433
	After	1.059	0.027	0.024	1.006, 1.111
	Incidence rate ratio (after/before)	0.787	0.025	<0.001	0.737, 0.836
	Ratio of ratios (Treatment/Comparison)*	0.962	0.033	0.252	0.896, 1.026

*Effectiveness of the Pathway in terms of adjudication events.

NB. Incidence rates for each time period (after/before referral to the Pathway) represent the number of events per person per unit exposure time (year) within that period. A rate ratio with a value greater than one represents a higher rate of adjudication events for the period after referral compared to that before referral to the Pathway. A rate ratio with a value less than one represents a higher rate of adjudication events for the period before referral compared to that after referral to the Pathway. All incidence rates can be converted to incidence per 1000 people within the exposure period, by simply multiplying their value by 1000.

Table 4. Propensity weighted Poisson regression for number of recalls. Analysis included a total of n = 6440 individuals, with data from 3 years before to 3 years after referral to the Pathway.

Group	Before/After referral to the Pathway	Incidence Rate	Standard Error	p-value	95% Confidence Interval
Comparison	Before	0.196	0.003	<0.001	0.189, 0.202
	After	0.191	0.003	<0.001	0.185, 0.196
	Incidence rate ratio (after/before)	0.972	0.022	0.218	0.929, 1.015
Treatment	Before	0.198	0.004	<0.001	0.189, 0.206
	After	0.196	0.004	<0.001	0.188, 0.203
	Incidence rate ratio (after/before)	0.991	0.029	0.746	0.934, 1.047
	Ratio of ratios (Treatment/Comparison)*	1.019	0.041	0.640	0.938, 1.099

*Effectiveness of the Pathway in terms of number of recalls.

NB. Incidence rates for each time period (after/before referral to the Pathway) represent the number of events per person per unit exposure time (year) within that period. All incidence rates can be converted to incidence per 1000 people within the exposure period, by simply multiplying their value by 1000.

Figure 3. Plot of incidence rates from propensity weighted negative binomial model for number of self-harm reports. Analysis included a total of n = 5513 individuals, with data from 3 years before to 3 years after referral to the Pathway

Table 5. Propensity weighted negative binomial regression for number of self-harm reports. Analysis included a total of n = 5513 individuals, with data from 3 years before referral to 3 years after referral to the Pathway.

Group	Before/After referral to the Pathway	Incidence Rate	Standard Error	p-value	95% Confidence Interval
Comparison	Before	0.443	0.011	<0.001	0.421, 0.465
	After	0.396	0.012	<0.001	0.372, 0.420
	Incidence rate ratio (after/before)	0.895	0.027	<0.001	0.841, 0.949
Treatment	Before	0.527	0.019	<0.001	0.490, 0.564
	After	0.450	0.014	<0.001	0.424, 0.477
	Incidence rate ratio (after/before)	0.855	0.032	<0.001	0.793, 0.917
	Ratio of ratios (Treatment/Comparison)*	0.955	0.045	0.326	0.867, 1.042

*Effectiveness of the Pathway in terms of number of self-harm reports.

NB. Incidence rates for each time period (after/before referral to the Pathway) represent the number of events per person per unit exposure time (year) within that period. All incidence rates can be converted to incidence per 1000 people within the exposure period, by simply multiplying their value by 1000.

Figure 4. Plot of incidence rates from propensity weighted negative binomial model for number of non-violent offences. Analysis included a total of n = 402 individuals, with data from 2 years before referral to 2 years after referral to the Pathway

Table 6. Propensity weighted negative binomial regression for number of non-violent offences. Analysis included a total of n = 402 individuals, with data from 2 years before to 2 years after referral to the Pathway.

Group	Before/After referral to the Pathway	Incidence Rate	Standard Error	p-value	95% Confidence Interval
Comparison	Before	0.678	0.110	0.016	0.462, 0.893
	After	0.890	0.181	0.568	0.535, 1.245
	Incidence rate ratio (after/before)	1.314	0.282	0.203	0.762, 1.867
Treatment	Before	0.623	0.102	0.004	0.422, 0.823
	After	0.746	0.132	0.097	0.488, 1.004
	Incidence rate ratio (after/before)	1.199	0.307	0.479	0.597, 1.800
	Ratio of ratios (Treatment/Comparison)*	0.912	0.300	0.779	0.324, 1.499

*Effectiveness of the Pathway in terms of number of non-violent offences.

NB. Incidence rates for each time period (after/before referral to the Pathway) represent the number of events per person per unit exposure time (year) within that period. All incidence rates can be converted to incidence per 1000 people within the exposure period, by simply multiplying their value by 1000.

Figure 5. Plot of incidence rates from propensity weighted negative binomial model for number of sexual offences. Analysis included a total of n = 1359 individuals, with data from 2 years before to 2 years after referral to the Pathway

Table 7. Propensity weighted negative binomial regression for number of sexual offences. Analysis included a total of n = 1359 individuals, with data from 2 years before to 2 years after referral to the Pathway.

Group	Before/After referral to the Pathway	Incidence Rate	Standard Error	p-value	95% Confidence Interval
Comparison	Before	0.222	0.021	<0.001	0.181, 0.263
	After	0.404	0.079	<0.001	0.249, 0.559
	Incidence rate ratio (after/before)	1.819	0.411	0.008	1.014, 2.624
Treatment	Before	0.234	0.029	0.001	0.178, 0.290
	After	0.540	0.097	0.001	0.349, 0.731
	Incidence rate ratio (after/before)	2.307	0.403	<0.001	1.517, 3.096
	Ratio of ratios*	1.268	0.354	0.395	0.574, 1.962

*Effectiveness of the Pathway in terms of number of sexual offences.

NB. Incidence rates for each time period (after/before referral to the Pathway) represent the number of events per person per unit exposure time (year) within that period. All incidence rates can be converted to incidence per 1000 people within the exposure period, by simply multiplying their value by 1000.

Figure 6. Plot of incidence rates from propensity weighted negative binomial model for number of violent offences. Analysis included a total of n = 10,977 individuals, with data from 2 years before to 2 years after referral to the Pathway

Table 8. Propensity weighted negative binomial regression for number of violent offences. Analysis included a total of n = 10,977 individuals, with data from 2 years before to 2 years after referral to the Pathway.

Group	Before/After referral to the Pathway	Incidence Rate	Standard Error	p-value	95% Confidence Interval
Comparison	Before	0.891	0.030	0.001	0.832, 0.950
	After	0.924	0.034	0.033	0.856, 0.991
	Incidence rate ratio (after/before)	1.036	0.041	0.362	0.957, 1.116
Treatment	Before	0.840	0.028	<0.001	0.786, 0.894
	After	0.957	0.034	0.223	0.890, 1.024
	Incidence rate ratio (after/before)	1.139	0.043	0.001	1.054, 1.224
	Ratio of ratios*	1.099	0.057	0.066	0.988, 1.210

*Effectiveness of the Pathway in terms of number of violent offences.

NB. Incidence rates for each time period (after/before referral to the Pathway) represent the number of events per person per unit exposure time (year) within that period. All incidence rates can be converted to incidence per 1000 people within the exposure period, by simply multiplying their value by 1000.

Figure 7. Plot of incidence rates from propensity weighted negative binomial model for number of sexual and violent offences. Analysis included a total of n = 3468 individuals, with data from 2 years before to 2 years after referral to the Pathway.

Table 9. Propensity weighted negative binomial regression for number of sexual and violent offences. Analysis included a total of n = 3478 individuals, with data from 2 years before to 2 years after referral to the Pathway.

Group	Before/After referral to the Pathway	Incidence Rate	Standard Error	p-value	95% Confidence Interval
Comparison	Before	0.665	0.032	<0.001	0.602, 0.728
	After	1.063	0.071	0.360	0.924, 1.202
	Incidence rate ratio (after/before)	1.599	0.118	<0.001	1.367, 1.831
Treatment	Before	0.747	0.046	<0.001	0.658, 0.837
	After	1.077	0.083	0.338	0.914, 1.239
	Incidence rate ratio (after/before)	1.441	0.115	<0.001	1.216, 1.665
	Ratio of ratios*	0.901	0.088	0.289	0.728, 1.074

*Effectiveness of the Pathway in terms of number of sexual and violent offences.

NB. Incidence rates for each time period (after/before referral to the Pathway) represent the number of events per person per unit exposure time (year) within that period. All incidence rates can be converted to incidence per 1000 people within the exposure period, by simply multiplying their value by 1000.

Data availability statement

The use of Ministry of Justice data was addressed by a time limited Data Sharing Agreement between the Ministry of Justice and Kings College London. Due to the sensitive nature of the research (high risk, high harm offending data), and due to legal restrictions, supporting data are not available.