Systematic Review

Psychosocial Interventions for Children and Young People With Visible Differences Resulting From Appearance Altering Conditions, Injury, or Treatment Effects

Elizabeth Jenkinson, DHEALTHPSY, CPSYCHOL; Heidi Williamson, DHEALTHPSY, CPSYCHOL; James Byron-Daniel, PHD, CPSYCHOL; Timothy P. Moss, PHD, CPSYCHOL

Disclosures

J Pediatr Psychol. 2015;40(10):1017-1033. 

In This Article

Results

Study Selection

Database searches yielded 11,422 abstracts, and 6 were identified through other sources. After duplicates were removed, 9,904 titles were screened, with 1,149 abstracts retained for further review. One hundred and five full papers were reviewed and 12 studies met the eligibility criteria. Reasons for exclusion were documented using a PRISMA flow chart (Figure 1).

Study Characteristics

Studies included participants with a range of visible differences. Seven evaluated the effectiveness of interventions for those with burns (Arnoldo, Crump, Burns, Hunt, & Purdue, 2006; Bakker, Van der Heijden, Van Son, Van de Schoot, & Van Loey, 2011; Biggs, Henrich, Jekel, & Cuono, 1997; Blakeney et al., 2005; Gaskell, 2007; Rimmer et al., 2007; Rosenberg et al., 2013), two with cranio-facial conditions (Devine & Dawson, 2010; Kapp Simon, McGuire, Long, & Simon, 2005), one with psoriasis patients (Scheewe, Schmidt, Peterman, Stachow, & Warshburger, 2001), one with cancer patients (Varni, Latz, Colegrove, & Dolgin, 1993), and one with patients with a range of conditions (Maddern et al., 2006). Across studies, 606 participants entered intervention groups and sample sizes were generally small (range 13–169, mean = 50.5). Studies had a male (Arnoldo et al., 2006; Bakker et al., 2011; Biggs et al., 1997; Gaskell, 2007; Rimmer et al., 2007; Rosenberg et al., 2013; Varni et al., 1993) or female (Blakeney et al., 2005; Devine & Dawson, 2010; Kapp Simon et al., 2005; Maddern et al., 2006; Scheewe et al., 2001) sample bias with an overall greater number of male participants (n = 342); participant ages ranged from 5 to 18 years old (mean could not be calculated owing to missing data). Ethnicity was not documented in six studies (Arnoldo et al., 2006; Bakker et al., 2011; Biggs et al., 1997; Gaskell, 2007; Maddern et al., 2006; Scheewe et al., 2001). Where reported, White participants accounted for between 51.3 and 90% of the sample (Blakeney et al., 2005; Kapp Simon et al., 2005; Rimmer et al., 2007; Varni et al., 1993) with the exception of Rosenberg's study where 97% were Hispanic and 3% were White.

Intervention Characteristics

Across studies, the rationale for interventions was to improve psychological well-being by targeting low self-esteem, low self-confidence and/or social interaction difficulties and/or poor body image and/or behavioral outcomes. Six studies evaluated residential social camp programs (Arnoldo et al., 2006, Bakker et al., 2011; Biggs et al., 1997; Devine & Dawson, 2010; Gaskell, 2007; Rimmer et al., 2007), two group SIST (Blakeney et al., 2005; Kapp Simon et al., 2005), one individual CBT plus SIST (Maddern et al., 2006), one group education/behavioral therapy (BT) (Scheewe et al., 2001), one individual SIST (Varni et al., 1993), and one exercise and counseling (Rosenberg et al., 2013). Interventions were delivered face-to-face by teams that included health-care professionals, recreational specialists, or other appropriately trained professionals. In five studies, psychologists, psychiatrists, counselors, and mental health professionals led or supervised the interventions (Blakeney et al., 2005; Kapp Simon et al., 2005; Maddern et al., 2006; Rimmer et al., 2007; Varni et al., 1993).

Outcomes and Measures Employed

Fifteen questionnaire measures were used to evaluate psychosocial outcomes in the 12 included studies (see Table I and Table II for full details). The primary outcome in nine studies was self-assessment of global self-esteem, evaluated using the Rosenberg (1965) Self-esteem Scale (Arnoldo et al., 2006; Bakker et al., 2011; Biggs et al., 1997; Devine & Dawson, 2010; Rimmer et al., 2007) and the Self-Perception Profile for Children (Harter, 1985a; Bakker et al., 2011; Gaskell, 2007; Varni et al., 1993).

General Findings

Results of individual studies evaluating the effectiveness of residential social camp studies are presented in Table I with statistics reported to two decimal places (for full data see original papers cited). Five of the six residential camp interventions showed little or no effect on global self-esteem. However, four camp interventions reported short-term improvements in other measures of psychological well-being. Bakker et al. (2011) demonstrated significant differences in dissatisfaction with appearance at 1 week post camp for the attendees as compared with the nonattenders, with a very small effect size (p = .02, d = .08). Multiple regression showed camp participation accounted for small variation in decrease in dissatisfaction with appearance (F(1, 146) = 7.28, p < .01, R 2 = .05, adj. R 2 = .05). However, these gains were not maintained at follow-up at 16 weeks post camp. Similarly, Devine and Dawson (2010) demonstrated significant improvements in social acceptance immediately post camp for attendees (p = .15) but these were not maintained at 6-week follow-up. Rimmer et al. (2007) noted improvements in self-esteem for first-time campers with a small effect size calculated (p = .01, d = .27) and among those who attended both camps (p = .02, d = .29) compared with no significant change for nonattenders. Gaskell (2007) demonstrated no significant impact on primary outcomes such as self-esteem, social anxiety/competence, or strengths and difficulties, but identified a positive impact of camp attendance on subscales of the SDQ and SPPC 1 out of the 4 years of camp recruitment (conduct p < .01, school p < .05) and physical appearance (p < .05).

Results of individual studies evaluating the effectiveness of individual and group SIST, CBT, BT, and exercise and counselling are presented in Table II with statistics reported to two decimal places (for full data see original papers cited). Of these six studies, the one study that measured global self-esteem showed significant improvements for the intervention group at 9 months follow-up with a medium effect size (p = .03, d = .72; Varni et al., 1993). Four included studies showed significant positive effects for participants receiving the interventions on one or more measures or subscales of psychological well-being, although some studies only demonstrated an impact for children and young people in one aspect of their psychosocial functioning and effect sizes were generally small. Maddern et al. (2006) measured the impact of CBT/SIST using a teasing visual analog scale and the Child Behavior Checklist (CBCL; Achenbach, 1991) completed by parents. A significant reduction in self-reported teasing was reported after intervention (playground p = .013, classroom p = .03). Gains were maintained at 6 months follow-up (p < .1). Subscales of the CBCL were also improved postintervention (internalizing behavior = p < .00; somatic complaints = p < .01; anxious/depressed = p < .05). SIST was also shown to be beneficial for young cancer patients (Varni et al., 1993), with improvements in self-esteem at 9 months with a medium effect size (p = .03, d = .72), and a significant time main effect for state anxiety and parental social support at 6 months (which was not maintained at 9 month follow-up). A nonsignificant improvement was also found in total behavioral problems as measured by the CBCL at 6 months for those receiving the intervention (p = .08), which was significant at 9 months (p = .03, d = .22). Kapp-Simon et al. (2005) adopted a novel approach to evaluating a SIST intervention, measuring behavioral outcomes by observational methods, and showed a significant effect on overall communication with peers postintervention for those in the intervention group (p = .01, p = .31) and an interaction between pre- and postmeasures and group status (p = .02, p = .03) with medium-large effect sizes (d = .60; .90), although it is likely that these large effect sizes were impacted by small sample bias in calculation. Blakeney et al. (2005) also demonstrated some improvements for patients with burn injuries receiving an SIST intervention with children in the intervention showing reduced self-reported behavioral problem scores compared with a comparison group, with a very small effect size (p < .03, d = .12).

However, BT and education showed no significant effect on anxiety or estimation of attractiveness for young people with psoriasis (Scheewe et al., 2001), and although both the control and intervention group showed improvements in self-efficacy (p < .00), this was not influenced by group status. Exercise and counseling (Rosenberg et al., 2013) was found to only have an impact on subscales of the Child Health Questionnaire (Landgraf, Abetz, & Ware, 1996), with family cohesion showing significant improvements for the intervention group postintervention compared with the control group (p < .01). Within the intervention group there were also significant improvements postintervention in physical functioning, role/social physical, and mental health, as well as parent-reported physical and psychosocial summary score (p < .01).

Risk of Bias Within Studies

Table III and Table IV present the risk of bias for each included study, split by intervention type. Overall, studies evaluating both camp interventions and SIST, CBT, BT, and counseling and exercise were deemed to be at high risk of bias.

Selection Bias

Two studies described randomization but it was unclear how random allocation was achieved and as such, the risk of bias is unclear for these studies (Blakeney et al., 2005; Varni et al., 1993). Seven studies did not include a control group or comparison group and therefore, did not include randomization within the study design (Arnoldo et al., 2006; Biggs et al., 1997; Devine & Dawson, 2010; Gaskell, 2007; Maddern et al., 2006; Rimmer et al., 2007; Rosenberg et al., 2013) and the remaining three studies did not use random allocation (Bakker et al., 2011; Kapp Simon et al., 2005; Scheewe et al., 2001). Eleven studies did not report inclusion of methods of allocation concealment, or owing to a lack of control/comparison group, it was inappropriate. One study cited using allocation concealment, but it was not clear if this method was appropriate (Varni et al., 1993).

Performance and Detection Bias

Appropriate blinding of participants and personnel was not reported in any study resulting in a high risk of bias in all. One study employed blinding of outcome assessment for participants, as the study used naturalistic observational ratings as the primary outcome measure, limiting detection bias (Kapp Simon et al., 2005). However, it is acknowledged that given the nature of included studies and interventions under scrutiny, both performance and detection bias may have been difficult to achieve in these studies.

Attrition Bias

Three studies included appropriate information regarding attrition rates and showed either no attrition (Kapp Simon et al., 2005) or no difference between attrition from the control and intervention group (Bakker et al., 2011; Blakeney et al., 2005). In two studies, the risk of bias was unclear (Gaskell, 2007; Varni et al., 1993) and seven were deemed to be high risk (Arnoldo et al., 2006; Biggs et al., 1997; Devine & Dawson, 2010; Maddern et al., 2006; Rimmer et al., 2007; Rosenberg et al., 2013; Scheewe et al., 2001).

Selective Reporting

All but one study (Biggs et al., 1997) reported the results of all measures used and included clear descriptions of data analysis. Across 10 out of 11 studies, there was a low risk of selective reporting bias. Nevertheless, no studies included intention to treat analyses.

Other Sources of Bias

Six studies did not include a comparison group, which included five out of six of the studies evaluating residential social camp interventions (Arnoldo et al., 2006; Biggs et al., 1997; Devine & Dawson, 2010; Gaskell, 2007; Maddern et al., 2006; Rimmer et al., 2007). Ten studies did not formally record adherence and/or treatment fidelity (Arnoldo et al., 2006; Bakker et al., 2011; Biggs et al., 1997; Blakeney et al., 2005; Devine & Dawson, 2010; Gaskell, 2007; Kapp Simon et al., 2005; Rimmer et al., 2007; Rosenberg et al., 2013; Varni et al., 1993). In three studies, one or more scales were not standardized or validated (Devine & Dawson, 2010; Maddern et al., 2006; Scheewe et al., 2001).

Risk of Bias Across Studies

No risk of bias analysis across studies was completed (see synthesis of results). However, given the very small sample sizes used in included studies, calculations of effect size are likely to be subject to small study bias and as a result, an unreliable measure of impact.

Comments

3090D553-9492-4563-8681-AD288FA52ACE

processing....