Table 1 Findings from papers.

Author (year)	Subjects, setting, and study design	Quality score (max. 14)	Abuse type	Follow up	Reabuse rate	Factors associated with reabuse
DePanfilis and Zuravin (1999a)10	446 selected families from all cases (2902) with substantiated index maltreatment in 1988. Mother was primary or shared caregiver at index and living with children for some of follow up period and in contact with CPS prior to recurrence. Baltimore. Retrospective (described as “non‐concurrent prospective”).	12	Physical abuse or neglect. (CSA or mixed abuse at index excluded)	5 years from index maltreatment 1181/2902 fulfilled initial selection criteria; 654/1181 available after sampling and stratification; 446/654 available to recurrence study.	5 year family recurrence rate 34.8% 155/446	Any child in family placed. Risk Ratio (RR) = 1.9 (p = 0.002) Child vulnerability construct. RR = 1.4 (p = 0.02) Family stress construct RR = 1.2 (p = 0.02) Social support deficit construct RR = 1.4 (p = 0.0001) Partner abuse construct RR = 1.5 (p = 0.03) NB: other constructs/factors (number of prior abuse episodes; index abuse type; severity index; mother factors; survival stress) were not significantly associated with reabuse.
DePanfilis and Zuravin (1999b)11	1167 selected families from all cases (2902) with substantiated index maltreatment in 1988. Mother was primary or shared caregiver at index and living with children for some of follow up period. Includes cases closed, opened and continued for CPS after index episode. Baltimore. Retrospective (“non‐concurrent prospective”).	10	Physical abuse or neglect. (CSA or mixed abuse at index excluded)	5 years from index maltreatment 1181/2902 fulfilled initial selection criteria; 1167/1181 available after further exclusions.	5 year family recurrence rate: 497/1167 (42.6%);	Risk of recurrence highest during first 30 days after index abuse. 26.8% of recurrences occurred during CPS intake or ongoing CPS input. After CPS closure: 85% of families had no recurrence within 2 years. Neglect cases had consistently higher recurrence rates across all 3 service statuses (closed, open, or continued).
DePanfilis and Zuravin (2001)12	As per 1999b plus “Any child placed at any time” (probably most severe cases) removed from follow up. 747 cases (252 closed at intake; 495 provided services.) Baltimore. Retrospective (“non‐concurrent prospective”).	11	Physical abuse or neglect. (CSA or mixed abuse at index excluded)	5 years from index maltreatment As per 1999b plus 420 families excluded because of placement episodes.	5 year family recurrence rate: “closed at intake” 11/252 (4.37%) “open for services” 127/495 (25.7%)	Mean time to recurrence: Closed at intake group: 1748 days (SE 22.90; 95% CI: 1703–1793) Opened at intake group: 1423 days (SE 31.9; 96% CI: 1360–1485)
DePanfilis and Zuravin (2002)13	Sample of 434 families from same sample as (1999 above). 12 further families excluded because service variables not available. Baltimore. Retrospective (“non‐concurrent prospective”).	12	Physical abuse or neglect. (CSA or mixed abuse at index excluded)	5 years from index maltreatment As per 1999a plus 12 further families because service variables not available	5 year family recurrence rate: 34.8% (151/434)	Results as per 1999a paper plus addition of service, compliance and improvement variables. Attendance at CPS services: RR = 0.688 (p = 0.05): “attendance reduces risk of recurrence by 32%”. NB. other service variables examined revealed no relationship with recurrence
English et al (1999)14	4437 families with “founded initial or subsequent” referrals from sample of 42000 CPS referrals (recurrence dataset). Washington state. Database extraction; retrospective.	9	Any type except mixed	18 months following initial referral 4437 followed through. (No precise figures were given for numbers or reasons for dropouts.)	Family recurrence rates: 6.4% at 6 months, 9.1% at 12 months, and 10.6% at 18 months.	Bivariate associations (p⩽0.05)* between recurrence and: History of domestic violence Chronicity of maltreatment Child's age risk level Parent's history of maltreatment as child Caregiver impairments Parental substance abuse Social support Protection of child *No other evidence of strength of associations given
Fluke et al (1999)15	556224 (our calculation) “substantiated or indicated” child reports from 10 states in the USA (placed children excluded). Database extraction study; retrospective.	9	Any type including mixed abuse types.	Maximum 2 year fixed follow up period (range = <1–24 months). Only percentages used, thus no information re dropouts/missing information available.	Child recurrence rate (weighted mean + range): 6 months: 14.7% (3.5–22.2) 12 m:19.6%(4.8–29.3)	Neglect most likely maltreatment type to recur (log rank p<0.001) in 9 out of 10 states. Likelihood of recurrence increases after each subsequent maltreatment event. Children already receiving CPS at index at higher risk of first recurrence. No significant age or race‐ related findings.
Fryer and Miyoshi (1994)16	24507 records of substantiated abuse from Colorado CAN registry; first recurrence data collected only. Database extraction study; retrospective.	8	Any maltreatment.	Maximum follow up 4 years (1–1470 days). “All records complete and of high quality.”	Child recurrence rate during study 9.34%.	66.9% of recurrences within 330 days of initial incident (23.6% of recurrences within 30 days). Hazard rate for first 30‐day interval 11.6× greater than that for 25th interval (at about 2 years). Younger children more vulnerable (p<0.001). Revictimisation rates: physical neglect 13.17%; emotional neglect 12.02%; lack of supervision 10.99%; cuts/welts/bruises 8.8%; sexual abuse 8.26%.
Herrenkohl et al (1979)17	286 families with verified incidents from 328 families who were provided services for child abuse. All cases either closed or received more than one year of service. Demography of sample said to match that of locality. Two Pennsylvania counties. Case‐note study; retrospective	4	Any maltreatment.	Fixed 10 year period (1967–76); little data re follow up in individual cases. No data re selection or dropouts.	Family recurrence rates: 191/286 (66.8%) for verified incidents. Child recurrence rate: 253/511 (49.5%).	Recurrence rates for families in which physical abuse occurs 54.1%, gross neglect 44.4%, sexual abuse 29.6%, emotional abuse 21.4%. Recurrence estimate for single type abuse 51.9%; mixed abuse types: double 66.3%, triple 85.0%. Higher rates of recurrence in families with a child aged 0–5 years and lower rates with children over 11 (χ2 = 23.37, p<0.01, df = 2).
Johnson and L'Esperance (1984)18	120 randomly selected cases from a large urban California county. Database extraction study; retrospective	9	Physical only (CSA excluded unless physical force).	Maximum 2 years Little data given on selection or dropouts	Family recurrence rate 55/120 (45.8%)	Correlations with reabuse: “stress” (>1 child in home): r = 0.26, p<0.001; time abuser spends with child: r = 0.28, p<0.001); “parenting skills” (mothers skills): r = 0.36, p<0.0005; reasonableness of expectations of child: r = 0.33, p<0.0005); “client's capacity to use resources”: r = 0.33, p<0.0005. (NB: some strong correlations between individual factors)
Littel (2001)19	2681 families entering FPS during 15 months. Illinois Database extraction: retrospective	11	Any maltreatment necessitating FPS involvement.	From FPS referral to 1 year after ending use of FPS services. In FPS services for average of 107 days (range 1–617 days; sd 55.4) Usable data on 2194/2681 (81.8%) families.	Family recurrence rate: during FPS (mean 107 days): (372/2194) 17%; during FPS and within 1 year after FPS: (419/2194) 19.1%.	Participation in FPS broken down into two researcher‐derived indices: “collaboration” (participation in treatment planning and agreement with treatment plans), and “compliance” (keeping appointments, completing tasks and cooperation). Collaboration was not associated with substantiated reabuse. Compliance was associated with small reduction in substantiated reports during FPS (b = −0.08) but not after FPS.
Littel et al (2002)20	2681 families entered FPS during 15 months. Illinois. Database extraction; retrospective.	10	Any maltreatment necessitating FPS involvement.	From FPS referral to 1 year after ending use of FPS services. In FPS services for average of 102.8 days. Data available for 1911 families (71.3%).	Child recurrence rate during FPS and within 1 year after FPS 564/1911 (29.5%).	In families with cocaine problems: no variable was related to subsequent maltreatment. In families with housing problems previous chronic neglect relates to subsequent maltreatment (p<0.001). In families with parental mental illness prior reports of abuse and neglect , presence of children <2 years both relate to subsequent maltreatment p<0.05. In families with poor child care skills reports of child abuse (p<0.01) and child neglect (p<0.001) and Chicago area (p<0.001) all related to subsequent maltreatment. NB: service characteristics not linked to outcome in the above groups.
Murphy et al (1992)21	206 cases of serious maltreatment heard at Boston Court. Unclear whether these are selected or consecutive but represent 40% of cases filed in years 1985 and 1986. Boston Juvenile Court. Database linkage study; retrospective.	4	Neglect and/or physical abuse of under 13 year olds (excludes CSA).	Mean follow up 3.5 years from onset of court case (not from the index episode). Dropouts: 13 missing cases.	Child recurrence rates: 18/63 (28.6%) of cases dismissed by the court; 8/130 (6.2%) of permanently removed cases.	In previously dismissed group returning to court there were more prior appearances and parental mental disorder than group who did not return to court (χ2 = 6.4, df = 1, p<0.01 and χ2 = 5.4, df = 1, p<0.05 respectively). (Severity and type of maltreatment, child's age and parents' rejection of court services were not associated with return to court.) In permanently removed group return to court was more likely to be associated with age >3 years at first hearing and >6 previous reports (χ2 = 6.1, df = 1, p<0.01 and χ2 = 4.9, df = 1,p<0.05 respectively). (Severity and type of maltreatment, parents' rejection of court services were not associated with return to court). (Review)
Rittner (2002)22	447 children living with parents for minimum 6 months after initial maltreatment. All had CPS contact. “Randomly selected” 200 from active, 205 from closed cases, 42 from a transferred to foster care group. Metropolitan area in New York state. Database study: retrospective.	10	Any maltreatment.	Followed up every 6 months until 18 months. Dropouts: “initial sample” 500 children from child welfare database. 487 records located; 25 used to test instrument and a further 15 for inter‐rater reliability.	18 month child recurrence rate: 101/447 (22.6%)	Strongest predictors of recurrent maltreatment: CPS investigation in last 5 years (χ2 = 25.912, df = 5, p<0.0001), and caretaker's own abuse history ‐ especially neglect (χ2 = 11.08, df = 1, p<0.001). No association with cooperation/compliance with court orders. Predictors of reabuse in multivariate analysis were: mental health problems in parent no income at intake, and parental history as victim of abuse (however these only accounted for a very small part of the variance).
Rivara (1985)23	71 children <24 months old referred for physical abuse to specialist hospital service.	6	Physical abuse only.	Follow up mean 30.8 months (sd 21.5) Dropouts: data missing in 3 cases	Child recurrence rate: 21/71 (30%).	Reabuse after evaluation occurred a mean of 5 weeks later (range 1–12 weeks) “There was no relationship between[recurrent maltreatment and] compliance to treatment, child's age at time of evaluation or IQ.*” *no substantiating statistics given.
Swanston et al (2002)24	183 sexually abused children aged 5–15 years presenting to 2 hospital child protection units. Metropolitan Sydney, Australia. Prospective.	9	Sexual abuse only.	6 years from initial referral. No dropouts evident: some precise data missing regarding nature of initial CSA in 14 cases.	6 year child recurrence rate: 28% received notifications for re‐abuse and/or neglect; 11% received 2 or more notifications.	Significant relationship between notifications for abuse/neglect (esp. emotional abuse) before study and notifications after intake RR = 6.44 (3.02–13.76).
						Severity of index CSA related to subsequent notification for abuse/neglect (χ2 = 29.54, df = 3, p = 0.02). Demographics and maternal history of CSA not significantly related.
						Intake variables most related to subsequent notifications for maltreatment: caregiver change before intake, social worker's rating of family function and multiple changes in caregivers (χ2 = 17.77; df = 2; p = 0.001; χ2 = 11.27, df = 4,p = 0.02; χ2 (MH) = 17.44, df = 1, p<0.001); parental mental health problems (RR = 4.23, (95% CI 2.01–8.89)), parental conflict (RR = 2.25 (1.1–4.62)) and parental alcohol abuse variables (RR = 2.67 (1.24–5.74)). Multivariate model including prior notifications for emotional abuse, multiple changes in caregiver and severity of index CSA accounts for 13% variance. (Similar trends obtained for relationships of individual variables to recurrence of CSA).
Wood (1997)25	409 validated cases of child maltreatment selected in chronological order from January 1990. No apparent exclusion criteria. El Paso, Texas. Retrospective case note analysis.	10	Any maltreatment including mixed	Follow up: 2 years after index event. No dropouts recorded. However, 8 /27 crucial risk indicators were unscorable because of insufficient data (e.g. prior placement for child; changes of household and drug or alcohol abuse).	2 year child recurrence rates: abuse 9%; neglect 5%. Overall maltreatment rates unclear as overlap between categories not described.	a) Substantiated subsequent abuse associated with: prior history of episodes of abuse /neglect in family (χ2 = 19.4, df = 2, p<0.01); authors' estimated odds ratio = 5.96; number of victims involved in incident (χ2 = 8.8, 1, p = 0.01); primary caretaker abused as child (χ2 = 6.0, 1, p = 0.01). b) Substantiated subsequent neglect associated with: current report of neglect (χ2 = 11.8, 2, p<0.01); prior history of episodes of abuse /neglect in family (χ2 = 13.6, 2, p = 0.01); primary caretaker intellectual limits (χ2 = 8.8, 1, p = 0.01); inadequate supervision (χ2 = 4.6, 1, p = 0.03).

CPS, Child Protective Services; CSA, child sexual abuse; FPS, Family Preservation Services.