A Study of the Efficiency and Effectiveness of the Integrated Correctional Program Model (ICPM)

Executive Summary

The following is an assessment of the efficiency and effectiveness of the recently piloted Integrated Correctional Program Model (ICPM).  The performance of ICPM is examined relative to the traditional suite of Nationally Recognized Correctional Programs (NRCP) in the treatment of offenders serving federal sentences in Canada. As an alternative to the approach of assigning offenders to a series of programs aimed at individual criminogenic factors, ICPM was conceived and developed as one comprehensive program designed to address multiple and overlapping areas of need. Ultimately, the goals of introducing ICPM were to maintain the positive outcomes yielded by the traditional suite of correctional programming (i.e., ensure equal to lower rates of recidivism and institutional charges), while affording a more efficient and timely delivery of services to ensure that offenders will have completed their programming prior to discretionary release eligibility date.

The Correctional Service of Canada (CSC) introduced ICPM in the Pacific and Atlantic regions beginning in January 2010 and January 2011, respectively. The initial subject pool provided to the contractors consisted of (1) male offenders delivered ICPM in Pacific or Atlantic regions and (2) male comparison subjects delivered Nationally Recognized Correctional Programming (NRCP) from the traditional model. The latter had at least one admission to federal custody in either Pacific or Atlantic regions, and all sentence commencement dates resided within a 10 year timeframe of data extraction date (August 19, 2002 – August 19, 2012). A series of eight criminogenic and demographic indicators were incorporated into a case-control matching procedure, resulting in a sample of 940 ICPM and 940 NRCP cases.

Across samples, offender mean age was approximately 34 years. In terms of ethnic distribution, the majority of offenders were Caucasian (≈70%), followed by Aboriginal (≈20%), and Other (≈10%). Based on the most serious crime identified on one's criminal record, over 50% of offenders in the sample were categorized as violent offenders.

Several measures of program efficiency were compared for ICPM and NRCP participants, including time to program commencement, time to program completion, and relative proportion of discretionary releases. Through exposure to a Primer, offenders in the ICPM group were afforded programming 81 days sooner following admission to CSC than offenders in the NRCP group. Partially due to involvement in a Primer, ICPM subjects commenced their main program an average of 29 days later than NRCP subjects.

Although NRCP subjects completed their first correctional program more expediently than ICPM subjects (286.26 days from CSC admission vs. 339.90 days from CSC admission), time to complete one main program in ICPM versus all NRCP program enrollments on a sentence favours ICPM (332.11 days vs. 489.05 days, respectively). Disaggregation of the ICPM sample suggested superior time efficiencies across all measures for the Multi-target Stream compared with either the Aboriginal or Sex Offender Streams; essentially, offenders in the Multi-target program are commencing and ending their respective programs more expediently than offenders enrolled in alternative streams.

In addition to reducing the time required to complete correctional programming, a related consideration in developing ICPM was specifically to ensure that offenders completed their programming in advance of release eligibility. Across the entire sample of offenders, proportions of discretionary releases granted to ICPM and NRCP groups were statistically equivalent at 35.5% and 37.7%, respectively. However, a cohort effect was identified whereby the frequency of discretionary releases granted to NRCP subjects steadily decreased from 2002 to present. Considering only the 2010-2012 unmatched cohorts, a significantly greater proportion of discretionary releases were granted to ICPM versus NRCP subjects (33.6% vs. 21.1%). However, because it was not possible to simultaneously control for the cohort effect and retain a sufficient number of matched subjects, it would not be prudent to draw any firm conclusions from these results.

The first intermediate outcome considered in our comparisons of ICPM and NRCP groups was a measure of pre- versus post-program change in ratings on the Generic Program Performance Measure (GPPM; Stewart, 2005), an index reflecting the extent to which an offender has successfully completed a program on the basis of performance and effort. ICPM and NRCP were on par in this respect, with GPPM ratings increasing to an equivalent degree as a result of respective programming.

The second intermediate outcome of interest involved a comparison of institutional misconducts resulting in guilty verdicts and occurring exclusively during the 1-year timeframe preceding the end of the observation period dictated either by release from custody, death/deportation, or data extraction date. When considering relative frequency of minor charges and the relative proportion of offenders incurring minor institutional charges over said timeframe, the ICPM sample demonstrated better outcomes. Specifically, 38.1% of ICPM offenders versus 60.3% of NRCP offenders committed minor institutional infractions. However, rates of serious institutional charges were nearly equivalent across groups (31.7% for ICPM vs. 31.0% for NRCP).

To reiterate, the expectation in designing ICPM was for the program model to be at least as effective as NRCP in attenuating recidivism outcomes. From the initial ICPM-NRCP matched groups (n = 940 in each group), a total of 346 ICPM participants had been released from custody and consequently, only 346 matched NRCP counterparts were included so as to preserve group equivalence. Despite some variation in magnitude associated with respective Offender Person Year (OPY) rates (i.e., an incidence rate calculated by dividing the total number of events over the observation period by the total number of years at risk within each offender subgroup), rate ratios indicate that ICPM and NRCP groups are statistically equivalent across all outcomes of interest. Consideration of recidivism rates over fixed timeframes confirmed findings from OPY analysis; although there appears to be a general trend illustrating lower recidivism rates for ICPM compared to NRCP (e.g., returns to custody being 32.0% vs. 38.0% at 1 year), no statistically significant differences emerged.

Notably, in the subsample of Aboriginal offenders, the reverse trend occurred whereby recidivism rates were higher for offenders exposed to ICPM compared to NRCP; for example, at 9 months post-release, 37.5% of Aboriginal offenders enrolled in ICPM had returned to custody, versus only 12.5% of Aboriginal offenders enrolled in NRCP (c2 = 4.00, p < .05). Although the difference in these percentages is likely inflated due to small sample sizes, a series of Cox regression survival analyses that controlled for time-at-risk (i.e., did not restrict sample size through a fixed follow-up period), age at release, and reintegration potential confirm all above findings.  Specifically, while survival rates for offenders in ICPM and NRCP were statistically identical across recidivism outcomes (i.e., return to custody for any reason and return to custody with an offence), the survival rate for Aboriginal offenders participating in NRCP was greater than for their ICPM counterparts in examining any return to custody, with a strong tendency towards statistical significance (Wald c2 = 3.70, p = .055).  In sum, it appears that while ICPM and NRCP models are statistically equivalent in reducing recidivism outcomes overall, ICPM may be less effective in the treatment of Aboriginal offenders.

An important finding to emerge was that offenders enrolled in the ICPM Multi-target Stream evidenced more favourable recidivism outcomes than matched offenders in the NRCP group. For illustrative purposes, at the 2-year follow-up mark, the survival rate for offenders in the ICPM Multi-target Stream was 87% compared to 77% for NRCP participants (re. returns to custody with an offence). As such, these preliminary results suggest that the Multi-Target Stream of ICPM appears to be an effective model in the reduction of recidivism. In addition, while high intensity ICPM and NRCP models were equally effective, at moderate intensity, the survival rate of ICPM participants significantly exceeded that of NRCP participants (c2 = 5.56, p < .05).

An attempt was made to examine recidivism according to treatment dosage, where dosage was conceptualized as the cumulative number of prescribed hours associated with an offender's completed programs. Interestingly, despite the fact that the traditional program model (NRCP) is expected to treat offenders via multiple programs tailored to each need area, the median number of program completions (excluding maintenance programs) in the NRCP group was 1.00, with a mean of 1.23 (SD = 0.80). Notwithstanding the possibility that NRCP participants may not have completed the programming required to target their multiple needs, recidivism outcomes were still comparable to those of ICPM participants.

Controlling for dosage, results of survival analyses examining returns to custody and returns to custody for the commission of an offence revealed that in each case, dosage conceptualized as prescribed program hours was not predictive of recidivism, nor did dosage significantly interact with program type. However, with maintenance and community programs included in the calculation of dosage, within-group comparisons did in fact demonstrate that increased dosage based on number of program completions predicts lower recidivism.

While the Multi-target Stream of ICPM and ICPM delivered at moderate intensity demonstrated promising results in terms of both time efficiency and effectiveness in recidivism reduction, a major concern identified by the current research surrounds the results of recidivism outcomes for Aboriginal offenders involved in ICPM.  Further research is required to elucidate the potential barriers of ICPM in effectively treating Aboriginal offenders. As a key sub-population within federal corrections in Canada, it is important that an effective programming model be available to address the multiple needs of this group. As currently implemented, the traditional program model (NRCP) appears to be a preferable option for the treatment of Aboriginal offenders.

A number of caveats and methodological limitations must be underscored in interpreting the results of this study. The use of retrospective control cases introduced a cohort effect in the examination of discretionary releases. Our ability to compare ICPM cases to a matched cohort of NRCP cases was precluded by low numbers and consequently, it was not possible to provide a reasonable response to the question of whether the new program model produced higher discretionary release rates.

Given the relatively recent implementation of ICPM, the available follow-up period for these subjects was limited to an average of only 217.54 days (vs. 357.67 days in the case of NRCP). Therefore, recidivism base rates pertinent to specific reoffence types – namely violent and sexual reoffending – were extremely low and could not be reliably examined. In a related vein, follow-up periods extended only to an offender's Warrant Expiry Date (WED) such that unless the offender received a subsequent federal sentence, activity following WED was unknown. Although both ICPM and NRCP were equally biased by this methodological constraint, an ideal approach would have involved the triangulation of provincial and federal records to provide a more complete and accurate record of recidivism outcomes.

While ICPM aims to target multiple criminogenic needs in the context of one comprehensive program, the model was not explicitly tailored to address the particularities of specialized offending patterns (e.g., domestic violence). Ultimately, based on small sample sizes and the unavailability of detailed offence information, it was not possible to gauge the effectiveness of ICPM with specific offender subtypes. From this perspective, discontinuation of existing NRCPs designed to treat very specific groups (including Aboriginals) is ill-advised at this juncture.

The operative question underlying this research is whether ICPM is a preferable alternative to the traditional NRCP model in preparing offenders for release as efficiently as possible while contributing to recidivism reduction. There is evidence that, on average, ICPM participation begins earlier than NRCP with enrollment in an ICPM Primer. There is also evidence that ICPM programming (particularly within the Multi-target Stream) is completed sooner than NRCP when the latter is gauged by all program completions on a sentence. Results on relative discretionary release rates were inconclusive, and while the overall number of institutional charges incurred was significantly lower in the ICPM group, the disparity between groups was attributable only to minor infractions. With the exception of the results pertaining to Aboriginal offenders where ICPM produced inferior outcomes, we provide evidence that in general, ICPM either "holds its own" or shows a marginal improvement over NRCP with respect to recidivism reduction.

The interpretation of this comparative analysis should take into account some temporal issues that differentiate the delivery of the two treatment models.  The broader correctional literature provides evidence to show that program impacts are most positive at the earliest stages of development and implementation; that is, program delivery quality has been shown to decline over time, resulting in an attenuation of treatment benefits (e.g., Welsh, Sullivan, & Olds, 2010). Although both ICPM and NRCP yielded comparable results on a number of measures, we compared a newly implemented program to one that has almost certainly suffered implementation drift. As such, these preliminary comparisons of ICPM to NRCP should be interpreted with the caution that ICPM effects may be inflated for this reason.

Importantly, the current research could not address all of the factors that should be considered in determining whether ICPM is a superior alternative to NRCP.  A thorough cost-benefit analysis of ICPM versus NRCP is advisable, as well as further research that incorporates longer follow-up periods. Mixed-methods approaches that supplement quantitative results with qualitative components are also highly suggested. For example, soliciting feedback through interviews and surveys from correctional staff and offenders involved in ICPM would be extremely useful in determining their level of satisfaction with the new model and in identifying potential areas of concern.

Taken together, the results of this study and their caveats do not strongly support one model over the other. Given the already well-validated status of the traditional cadre of programming, proceeding with an expansion of ICPM based on these preliminary and sometimes equivocal results is not recommended at this time.

For more information, please e-mail your request to  Gen-nhqPrograms/programmes@csc-scc.gc.ca