An Examination of Predictive Validity and Change in Risk Factors for Stalking over Time

Stephanie R. Penney; Roy Ulrich; Margaret Maheandiran

doi:10.29158/JAAPL.220110-22

Article Figures & Data

Figures

Tables

Download figure
Open in new tab
Download powerpoint
Figure 1.
Study design.
Download figure
Open in new tab
Download powerpoint
Figure 2.
Change in dynamic risk scores over time.

Tables

Figures

View popup

Table 1

Prevalence of Outcomes Following Forensic Admission

Outcome variable^a	One year (n = 86)		Two years (n = 80)		Three years (n = 70)		Four years (n = 56)		All time periods (n = 86)
Outcome variable^a	n	%	n	%	n	%	n	%	n	%
Stalked previous victim	14	16.3	8	10.0	10	14.3	6	10.7	23	26.7
Stalked new victim	7	8.1	7	8.8	5	7.1	3	5.4	13	15.1
Stalking recidivism (any)	19	22.1	14	17.5	13	18.6	9	16.1	30	34.9
Nonstalking violence	10	11.6	11	13.8	5	7.1	5	8.9	22	25.6
Medication noncompliance	26	30.2	13	16.3	15	21.4	15	27.3	42	48.8
Substance use	10	11.6	11	13.8	11	15.7	10	17.9	24	27.9
Rehospitalization^b	11	44.0	9	23.1	11	25.6	11	33.3	28	32.6

↵a All outcomes were coded from the health record and Ontario Review Board disposition documents at each time point.
↵b Of the 25, 39, 43, and 33 individuals with some duration of community tenure during these periods, respectively.

View popup

Table 2

Internal Consistency and Descriptive Statistics

			Risk coding (baseline)	Risk coding follow-up 1	Risk coding follow-up 2
	ICC_A,1/Kw (95% CI)	α	M (SD)			Diff Paired t (Cohen’s d)^a
HCR-20V3
Historical	.90 (.19, .98)	.75	13.26 (3.90)	—	—	—
Clinical	.94 (.42, .99)	.73–.82	6.76 (2.47)	5.05 (2.93)	4.93 (3.09)	4.30 (.53)
Risk management	.85 (.72, .98)	.57–.63	3.22 (1.50)	3.49 (1.69)	3.46 (1.85)	−1.47 (−.18)
Total	.86 (.74, .98)	.76	23.24 (5.72)	—	—	—
			% low / moderate / high
SRR _{Future violence}	.68 (.26, 1.0)	—	36.0 / 40.7 / 23.3	56.0 / 21.3 / 22.7	32.8 / 37.3 / 29.9	z = −.72–.43
			M (SD)
SAM
Previous – Nature of stalking	.92 (.65, .98)	.44	13.63 (3.46)	—	—	—
Previous – Perpetrator factors	.98 (.93, .99)	.13	14.38 (2.47)	—	—	—
Current – Nature of stalking	.88 (.49, .97)	.80–.86	1.99 (3.52)	1.79 (3.63)	1.52 (3.17)	0.99 (.12)
Current – Perpetrator factors	.96 (.88, .99)	.61–.74	10.13 (3.33)	8.85 (4.10)	8.54 (4.36)	3.78 (.47)
			% low / moderate / high
SRR _{Same victim}	.76 (.35–1.0)	—	39.5 / 41.9 / 18.6	44.0 / 36.0 / 20.0	43.3 / 37.3 / 19.4	z = −0.47–0.57
SRR _{Case prioritization}	.79 (.50–1.0)	—	36.3 / 43.8 / 19.9	42.1 / 35.7 / 22.2	39.2 / 37.4 / 23.4	z = −0.40–0.27
			M (SD)
SRP
Stalking violence	.75 (.48–.92)	.56–.59^b	—	—	—	—
Recurrence	.75 (.39–.97)	.39–.52^b	—	—	—	—
Persistence	.77 (.62–.99)	.66–.71	—	—	—	—
Psychosocial damage	.68 (.47–.98)	.47–.66	—	—	—	—
			% low / moderate / high
SRR _{Stalking violence}	.31 (.00–.80)	—	93.1 / 6.9 / 0.0	—	—	—
SRR _{Future stalking same} (Persistence + Recurrence _same)	.74 (.47–1.0)	—	42.5 / 39.1 / 18.4	—	—	—
SRR _{Psychosocial damage}	.60 (.10–.91)	—	28.7 / 44.9 / 26.4	—	—	—

Typology	Incompetent	Intimacy Seeking	Rejected	Resentful	Predatory
n (%)	1 (1.1)	40 (40.6)	12 (13.8)	30 (34.5)	4 (4.6)

HCR-20V3 = Historical, Clinical, Risk Management-20, Version 3. Maximum/minimum values = 0–20 (H), 0–10 (C & R). SAM = guidelines for Stalking Assessment and Management. Maximum/minimum values = 0–20 (Nature of stalking), 0–20 (Perpetrator risk factors). SRP = Stalking Risk Profile. SRR = Summary risk rating. ICC_A,1 = intraclass correlation coefficient, single measure (absolute agreement). Kw = weighted kappa.
Bolded items: p ≤ .001.
↵a Difference scores calculated as the difference between the last follow-up and baseline risk scores.
↵b Scale reliability calculated on those items common across motivational types (n = 12 for Stalking violence; n = 4 for Recurrence). The Persistence (12 items) and Psychosocial damage (11 items) scales contain the same items across each typology.

View popup

Table 3

Classification Accuracy Statistics for the HCR-20V3, SAM, and SRP Measures

	Stalking (n = 80)	Violence (n = 80)	Rehospitalization (n = 65)
	AUC (95% CI)
HCR-20V3
H scale	.54 (.39, .70)	.77 (.66, .87)^b	.67 (.54, .80)^a
C scale	.62 (.48, .77)	.67 (.52, .83)^a	.61 (.47, .75)
R scale	.65 (.52, .78)^a	.59 (.45, .74)	.65 (.52, .79)^a
SRR_{Future violence}	.72 (.59, .86)^b	.49 (.34, .65)	.58 (.43, .73)
SAM
Previous – nature of stalking	.45 (.32, .58)	.48 (.34, .62)	.41 (.27, .55)
Previous – perpetrator risk factors	.63 (.50, .76)	.73 (.62, .85)^b	.68 (.54, .81)^a
Current – nature of stalking	.71 (.58, .84)^b	.53 (.38, .67)	.65 (.51, .78)^a
Current – perpetrator risk factors	.69 (.55, .83)^b	.65 (.50, .80)^a	.63 (.49, .76)
SRR_{same victim}	.73 (.61, .86)^b	.52 (.37, .67)	.59 (.45, .74)
SRR_{case prioritization}	.72 (.60, 84)^b	.54 (.38, .67)	.58 (.42, .71)
SRP
SRR_{future stalking same} (persistence + recurrence_same)	.72 (.59, .85)^b	.53 (.39, .68)	.61 (.47, .76)

AUC = area under the curve.
↵a p ≤ .05.
↵b p ≤ .01.

View popup

Table 4

Cox Proportional Hazard Regression Testing the Predictive Validity of the HCR-20V3 Static and Dynamic Risk Indicators

	Stalking			Violence			Rehospitalization
	β	SE	HR (95% CI)	β	SE	HR (95% CI)	β	SE	HR (95% CI)
Step 1
H scale	.01	.05	1.01 (.92, 1.11)	.17	.06	1.19 (1.06, 1.34)^b	.09	.05	1.09 (.99, 1.20)
Step 1
H scale	−.01	.06	.99 (.89, 1.11)	.17	.07	1.19 (1.04, 1.34)^b	.13	.06	1.14 (1.02, 1.27)^a
C scale	.06	.11	1.06 (.86, 1.30)	.06	.11	1.07 (.86, 1.34)	−.14	.10	.87 (.72, 1.05)
R scale	.27	.16	1.31 (.95, 1.81)	.17	.19	1.19 (.83, 1.71)	.34	.16	1.40 (1.03, 1.92)^a

HR = Hazard ratio. All scores reflect baseline ratings.
Model results: Stalking: χ² (3, n = 80) = 5.37, p = .15. Violence: χ² (3, n = 80) = 11.50, p = .009. Rehospitalization: χ² (3, n = 65) = 7.55, p = .05.
↵a p ≤ .05.
↵b p ≤ .01.

View popup

Table 5

Cox Proportional Hazard Regression Testing the Predictive Validity of the SAM Static and Dynamic Risk Indicators

	Stalking			Violence			Rehospitalization
	β	SE	HR (95% CI)	β	SE	HR (95% CI)	β	SE	HR (95% CI)
Step 1
Previous – Nature of Stalking	−.06	.06	.95 (.84, 1.07)	−.07	.07	.93 (.82, 1.07)	−.13	.06	.88 (.78, .99)^a
Previous – Perpetrator Risk Factors	.11	.08	1.12 (.95, 1.31)	.32	.11	1.37 (1.10, 1.70)^b	.16	.08	1.18 (1.00, 1.39)^a
Step 1
Previous – Nature of Stalking	−.03	.06	.97 (.86, 1.10)	−.04	.07	.96 (.84, 1.11)	−.18	.07	.84 (.74, .96) b
Previous – Perpetrator Risk Factors	.09	.10	1.09 (.89, 1.33)	.28	.12	1.32 (1.05, 1.67)^a	.23	.09	1.25 (1.04, 1.51)^a
Current – Nature of Stalking	.14	.06	1.15 (1.03, 1.29)^b	−.04	.07	.96 (.84, 1.11)	.11	.06	1.12 (.99, 1.27)
Current – Perpetrator Risk Factors	.07	.08	1.07 (.92, 1.26)	.11	.08	1.12 (.95, 1.31)	−.11	.07	.90 (.78, 1.03)

↵HR = Hazard ratio.
Model results: Stalking: χ² (4, n = 80) = 19.37, p = .001. Violence: χ² (4, n = 80) = 10.13, p = .04. Rehospitalization: χ² (4, n = 65) = 10.46, p = .03.
↵a p ≤ .05.
↵b p ≤ .01.