Assessing Recovery and Disability After Physical Trauma: The Pediatric Injury Functional Outcome Scale

Abstract

Objective To establish reliability and validity of the Pediatric Injury Functional Outcome Scale (PIFOS), a brief injury-specific rating scale covering motor, self-care, communication, social–emotional, cognition, physical, and academic areas. Methods In a prospective longitudinal study, the PIFOS structured interview was administered to parents of children 3–15 years of age at 3 and 12 months after hospitalization for traumatic brain injury (TBI) or orthopedic injury (OI). Results The total score had good internal consistency (α = .90–.93) and inter-rater reliability (α = .90) and correlated significantly with injury severity and neurodevelopmental outcomes. Generalized linear modeling showed the PIFOS was sensitive to the type and severity of injury, showed specific initial and persisting difficulties following TBI and OI, and was responsive to change during the first year after injury. Both groups had residual difficulties with coordination, emotionality, social participation, and discomfort. Conclusion The PIFOS is useful in examining recovery in natural history and intervention studies.

disability, functional outcome, orthopedic injury, recovery, traumatic brain injury

Pediatric trauma is a key public health concern. Injuries are the leading cause of death and disability in children and adolescents (Langlois, Rutland-Brown, & Thomas, 2005; Winthrop et al., 2005). Survivors of major trauma, particularly children with moderate to severe traumatic brain injury (TBI), often experience significant short-term and long-term alterations in health-related quality of life (HRQOL) (Janssens, Gorter, Ketelaar, Kramer, & Holtslag, 2008). Following moderate to severe TBI, long-term consequences often persist in a broad array of areas, including neuropsychological, psychological health, and academic areas (Babikian & Asarnow, 2009; Di Battista, Anderson, Catroppa, & Soo, 2012; Fay et al., 2009). Improvement in cognitive domains is greatest during the first 6 months after TBI; recovery curves reflecting longer term outcomes are significantly below estimated preinjury levels (Chadwick, Rutter, Brown, Shaffer, & Traub, 1981; Fay et al., 1994). Sequelae may persist for years (Anderson, Brown, Newitt, & Hoile, 2011; Ewing-Cobbs et al., 1997; Taylor et al., 2002; Yeates et al., 2004) and extend into adulthood (Cattelani, Lombardi, Brianti, & Mazzucchi, 1998; Di Battista et al., 2012; Nybo, Sainio, & Muller, 2004). Adaptive behavior outcomes, including communication and self-care skills, are particularly vulnerable to disruption (Rivara et al., 2011), with deterioration or persistent deficits noted in 26–46% of cases (Fay et al., 2009). Psychological health outcomes are variable. Emotional and behavioral symptoms may increase (Taylor et al., 2001) or remain stable (Max et al., 2005, 2006) during the first 2 years after injury. Academic deficits are among the most significant long-term sequelae of TBI, with high rates of academic failure and need for academic support services (Ewing-Cobbs et al., 2006; Ewing-Cobbs, Fletcher, Levin, Iovino, & Miner, 1998; Jaffe, Polissar, Fay, & Liao, 1995).

Increasingly, the adverse long-term impact of major trauma and orthopedic injury (OI) on both physical and psychosocial outcomes is beginning to be appreciated (Ding et al., 2006; Holbrook et al., 2005, 2007; Winthrop, 2010). Impairments in HRQOL persisting throughout a 2-year follow-up were documented in children and adolescents sustaining major trauma (Holbrook et al., 2007). Children with either upper or lower extremity fractures experienced poorer HRQOL in physical and psychosocial domains 3 months to 1 year after the injury (Ding et al., 2006; Stancin et al., 2001). Long-term adaptive behavior deficits were reported in 29% of children examined 3–5 years after OI; neuropsychological and academic outcomes were not significantly affected (Fay et al., 2009).

Given the high incidence and prevalence of pediatric injury, it is increasingly important to assess the quality of long-term outcomes and areas of persisting disability to inform health care, academic, and vocational systems needed to support persons with long-term disabilities. However, recent reviews highlight the paucity of evidence-based measures evaluating outcomes after injury (Ardolino, Sleat, & Willett, 2012), particularly those validated to quantify pediatric outcomes (Winthrop, 2010). Outcome measures may be categorized as generic or disease specific. Several measures have been used to evaluate generic outcome and HRQOL following pediatric injury (Crouchman, Rossiter, Colaco, & Forsyth, 2001; Fiser et al., 2000; Landgraf, Abetz, & Ware, 1996; McCarthy et al., 2006; Varni, Seid, & Rode, 1999; Willis, Gabbe, Butt, & Cameron, 2006). Other commonly used instruments, such as the Pediatric Evaluation of Disability Inventory (Haley, Coster, Ludlow, Haltiwanger, & Andrellos, 1992) and the Wee Functional Injury Measure (Msall et al., 1994), were developed to assess adaptive, physical, and/or daily living skills in children with greater injury severity. However, there are few measures of injury-specific outcomes that cover a broad range of injury severity and have been validated for a broad age range. Consequently, there is a need for a brief quantitative measure assessing injury-specific sequelae following pediatric injury.

The purpose of the present study is to examine reliability and validity of the Pediatric Injury Functional Outcome Scale (PIFOS), a new rating scale based on a caregiver interview that quantitates motor, cognitive, communication, social–emotional, self-care, physical, and academic functional outcomes during postacute and long-term follow-up stages of recovery. We examined the following hypotheses:

1. The PIFOS was expected to demonstrate acceptable reliability and to correlate significantly with indices of injury severity.

2. PIFOS subscale score profiles would differ in children with TBI and OI at the postacute evaluation. Improvement was expected across time in motor, self-care, physical changes, and academic areas.

3. Long-term disability would be characterized by a gap between preinjury and postinjury level of functioning; disability was expected to be greater in the TBI than OI groups in communication, cognitive, and academic subscales.

4. The PIFOS total score was expected to correlate significantly with global outcome measures and neurodevelopmental scores.

Method

Participants

This study combines participants from two prospective longitudinal studies examining functional and neuropsychological outcomes after pediatric injury. Eighty children aged 3–15 years with TBI and 46 children with OI were recruited from an urban level 1 trauma center. Children aged 3–7 years were recruited from consecutive admissions from 1994 to 1998 (n = 26); children aged 8–15 years were recruited from 2004 to 2007 (n = 100).

Inclusionary criteria for children with TBI and OI were (a) hospitalization following acceleration–deceleration or blunt impact injuries caused by vehicular accidents, falls, or impact with a blunt object; (b) no history of previous TBI; (c) bilingual or primarily English speaking; (d) legal alien status and residing within 125 miles of the catchment area; (e) no known premorbid neurologic, metabolic, major developmental, or psychiatric disorders that would interfere with assessment of the impact of TBI on outcomes (e.g., intellectual deficiency, low-functioning autism spectrum disorders), and participation in the study at initial and 1-year follow-up evaluations. Children with OI also had no head or facial injuries to minimize the possibility of co-occurring mild TBI.

The lowest postresuscitation Glasgow Coma Scale (GCS) score (Teasdale & Jennett, 1974) and the new Injury Severity Score (ISS) (Osler, Baker, & Long, 1997) were used as indices of injury severity. The ISS is a standard summary measure of anatomic injury based on the sum of squares of the Abbreviated Injury Scale score (Association for the Advancement of Automotive Medicine, 1990) from the three most severely injured body regions. The severity of TBI was categorized based on the score and acute neuroimaging findings. Mild TBI was defined by GCS scores from 13 to 15; complicated–mild TBI was defined by GCS scores from 13 to 15 and neuroimaging evidence of extra-axial hemorrhage or parenchymal injury (Levin et al., 2008). Moderate and severe TBI had GCS scores of 9–12 and 3–8, respectively, with or without positive neuroimaging findings. The sample included children with mild (n = 2), complicated–mild (n = 12), moderate (n = 16), and severe (n = 50) TBI.

Procedure

Informed written consent was obtained from each child’s guardian. In accordance with university institutional review board guidelines, oral assent was obtained from children aged 6–7 years and written consent was obtained from children aged 8–15 years. Study participants were treated in accordance with ethical standards of the American Psychological Association.

Outcomes were evaluated an average of 3 and 12 months after injury. Each parent was interviewed and completed questionnaires evaluating adaptive behavior, executive functions, and behavioral outcomes. Neuropsychological tests were administered to each child individually by a trained examiner in an outpatient clinic setting.

Measure Development

The PIFOS was developed using recommended procedures for new assessment instruments (DeVellis, 1991). Items were generated based on a formal consensus process incorporating input from a multidisciplinary team of psychological, psychiatric, surgical, and rehabilitation professionals with expertise in acute care and long-term follow-up of pediatric injury. After piloting, items were refined and redundant items were eliminated. Remaining items were combined into a summated rating scale (Spector, 1992) completed by the interviewer based on a structured caregiver interview containing 26 items rated on a 4-point scale. Each item evaluated change from preinjury levels. A score of 1 indicated no change from preinjury functioning; response options from 2 to 4 were behaviorally anchored and reflected increasing need for support and limitations in daily activities. Total scores ranged from 26 to 104. Outcome subscales and the range of points per subscale are as follows: Motor, Communication, and Physical Changes (3–12 points), Self-care and Social–Emotional (5–20 points), Cognition (6–24 points), and Academic (1–4 points). The content of items in each subscale is listed in Table VI. To provide an example of question format and item content, the Leisure item from the Social–Emotional subscale is below:

“Has there been any change in your child’s ability to participate in his/her usual play, sports, or recreational activities (e.g., art, piano) because of changes in motor or thinking skills?”

1 = No change from preinjury, 2 = Resumes participation in most preinjury sports and activities with slight reduction in quality of performance, 3 = Unable to participate in competitive sports, significantly reduced participation in age-appropriate or preinjury activities, 4 = Engages only in passive recreational activities (e.g., watches TV).

The PIFOS interview was completed in <10 min for children with few injury-related changes and in ∼15 min for children with more injury sequelae. Fifteen consecutive parent interviews based on the PIFOS were recorded to allow estimation of inter-rater reliability.

Relation With Other Outcomes

To establish construct validity, PIFOS scores were examined in relation to the following functional, neuropsychological, and behavioral measures recommended as common data elements following pediatric TBI (McCauley et al., 2012): Global outcome—Glasgow Outcome Scale (GOS) (Jennett, Teasdale, Braakman, Minderhoud, & Knill-Jones, 1976) is a 5-point scale ranging from death to good recovery, regarded as the gold standard in TBI outcome research; Adaptive behavior—Vineland Adaptive Behavior Scales II (Sparrow, Cicchetti, and Balla, 2009) Interview Form (ages: 3–7 years) or Adaptive Behavior Assessment System II (Harrison & Oakland, 2003) (ages: 8–15 years) composite scores; Executive functions—Behavior Rating Inventory of Executive Functions (Gioia, Espy, & Isquith, 2000; Gioia, Isquith, Kenworthy, & Barton, 2002); general executive composite T-score; Psychosocial—Child Behavior Checklist (Achenbach, 1991) internalizing and externalizing behavior problem T-scores for children aged 3–18 years plus school problems ratings for children aged ≥6 years; General cognition—Stanford–Binet Intelligence Scale, 4th edition (Thorndike, Hagen, & Sattler, 1986) composite area standard score was used for children aged 3–6 years, and the Wechsler Abbreviated Scale of Intelligence (Wechsler, 1999) was used for children aged 7–15 years; Reading—Gray Oral Reading Tests-4 (Wiederholt & Bryant, 2001) (ages: >7 years); Processing speed—Coding subtest (Wechsler, 2003) (ages: >4 years); Fine motor coordination—Grooved Pegboard (Klove, 1963) (ages: >5 years).

Statistical Approach

Descriptive statistics for PIFOS scores and outcome variables were examined to determine whether scores were normally distributed. The PIFOS scores were positively skewed; consequently, group comparisons were completed using generalized linear models with a Poisson distribution and log link function.

Reliability

Internal consistency was assessed using Cronbach’s alpha for each of the six subscales with multiple items. There were a total of 26 items divided into seven subscales. However, the academics item was the only item for its scale, so it was not possible to examine internal consistency for that item except for its contribution to the total score. We also examined the corrected item–total score correlations and the alpha if the item was deleted to determine if any items needed elimination. We did this for each time point because it is important to keep the items on the same subscale to promote measuring change over time. Inter-rater reliability was evaluated using generalizability analysis (Brennan, 1983; Cronbach, Gleser, Nanda, & Rajaratnam, 1972).

Validity

Several approaches were used to evaluate construct validity of the PIFOS. First, sensitivity to injury severity was examined by correlating the PIFOS total score with ISS for the total sample and the GCS for children with TBI. For participants with TBI, generalized linear model analysis examined the effect of severity of TBI (mild/complicated mild, moderate, severe), time (3 and 12 months), and their interaction on PIFOS total and subscale scores. Second, known-group validity was examined using profile analysis to examine the pattern of subscale scores and differences in the profiles over time in the two injury groups. This repeated-measures generalized linear modeling analysis examined group (TBI, OI), time (3- and 12-month evaluations), and the Group × Time interaction using the seven PIFOS subscale scores as dependent variables. To assess specific areas contributing to persisting disability, we used chi-square analysis to examine group differences on individual items at the 12-month evaluation. Third, concurrent validity was examined using Spearman correlations to assess the strength of relations between PIFOS scores and the outcome measures. Sample sizes varied across outcome measures depending on age at which standardized scores are available.

Results

Injury Characteristics

As indicated in Table I, the TBI and OI groups did not differ significantly on age at injury, gender, ethnicity, or socioeconomic status (Hollingshead, 1975). Table II provides injury characteristics of each group and reports group comparisons where appropriate. The external cause of injury varied by group; motor vehicle collisions predominated in the TBI group, and falls and sport/recreational injuries occurred more often in the OI group. GOS scores did not differ at the 3-month evaluation [(Χ² (2, N = 126) = 5.45, p = .066]); approximately half of each group had a good recovery. By the one 1-year follow-up, a good recovery was noted in 63% of the TBI group and 98% of the OI group, [(Χ² (2, N = 126) = 19.66, p < .001]). The new ISS (Osler et al., 1997) was significantly higher in the TBI group than in the OI group. Injuries in the OI group were distributed as follows: Upper extremity (n = 26), thorax (n = 4), and/or lower extremity/pelvis (n = 35).

Table I.

Demographic Information for Traumatic Brain and Orthopedic Injury Groups

	Group
Demographic variable	Traumatic brain injury	Orthopedic injury	Statistic (p)
(n = 80)	(n = 46)
Months of age at injury M (SD)	108.1 (46.5)	119.9 (34.2)	F(1, 124) = 2.31 (.131)
Gender % male	69	57	χ2 (1, N = 126) = 1.90 (.168)
Ethnicity (n)			χ2 (4, N = 126) = 5.51 (.239)
African American	8	10
Asian	2	2
Hispanic	20	13
Other/multiethnic	6	1
Caucasian	44	20
Socioeconomic status (n)			χ2 (4, N = 126) = 7.47 (.113)
1 (high)	12	10
2	25	16
3	24	11
4	10	9
5 (low)	9	0

Table II.

Injury Characteristics for Brain Injury and Orthopedic Groups

	Injury group
	Traumatic brain injury	Orthopedic injury
Injury variable	(n = 80)	(n = 46)	Statistic (df) (p)
External cause of injury (n)			χ2 (4, N = 126) = 29.32 (<.001)
Fall	15	20
Motor vehicle collision	38	4
Vehicle–pedestrian	14	8
Sports/recreational	5	12
Other	8	2
Injury-test interval (days) M (SD)
Initial	103.0 (8.2)	73.8 (19.6)	F(1, 124) = 15.30 (<.001)
Follow-up	394.9 (44.3)	388.7 (44.0)	F(1, 124) = 00.57 (.451)
Injury severity score M (SD)	22.2 (10.3)	6.7 (2.5)	F(1, 124) = 100.90 (<.001)
Glasgow Coma Scale score (n)
3–8	50
9–12	16
13–15	14
Days Glasgow motor score <6
M (SD)	4.8 (8.2)	0

Comparison of the children in the younger and older cohorts revealed that the mean GCS score of 10.15 (SD = 4.2) was significantly higher in the younger children than in the older cohort, who had a mean score of 5.5 (SD = 3.7), F(1, 79) = 25.63, p < .001. Similarly, the ISS score differed across cohorts, with the younger cohort having less severe injuries, F(1, 79) = 4.86, p = .03. This difference in severity of the cohorts reflects the epidemiology of TBI because preschool-aged children are more likely to be injured in falls whereas older children and adolescents are more likely to be injured in transportation-related and assault incidents (Faul, Xu, Wald, & Coronado, 2010). The cohorts did not differ significantly on socioeconomic status, F(1, 79) = 0.13, p = .72.

PIFOS Reliability

Table III presents the results of the internal consistency reliability using Cronbach’s alpha. The total score reliabilities were excellent and ranged from .90 at 3 months to .93 at 12 months. As expected for a brief scale, the internal consistency of the subscales was lower than for the total score. Item total score correlations at 3 months indicated that the fine motor and receptive language items detracted from the reliability of the Motor and Communication subscales, respectively. All other items contributed to the reliability of both their subscale and the total score. By 12 months, the fine motor item did not add or detract from the reliability of the Motor scale. The toileting item detracted from the reliability of the Self-care scale. The sleep and discomfort items did not add or detract from the reliability of the total score. All other items contributed both to their subscale and the total scale reliability. The reliability of the subscales is adequate to good for all scales except Physical Changes. Cronbach’s alpha increased from 3 to 12 months for the total score and for each of the subscale scores. The changes were notable for the Motor, Communication, and Social–Emotional subscales; increases in alpha ranged from 8 to 11 units.

Table III.

Reliabilities and Corrected Item to Total Score Correlations for the Subscales and Total Score at 3 and 12 Months After Injury

Scale	Number of items	Median item to total r	Range of item to total r	Cronbach’s alpha
3 months
Motor	4	.70	.27–.83	.78
Daily Living Skills	4	.56	.43–.74	.76
Communication	4	.46	.23–.58	.65
Social–Emotional	5	.45	.43–.49	.66
Cognition	5	.64	.44–.68	.79
Physical Changes	3	.31	.30–.46	.54
Total	26*	.51	.30–.67	.90
12 months
Motor	4	.76	.59–.79	.86
Daily Living Skills	4	.72	.36–.79	.79
Communication	4	.57	.48–.63	.76
Social–Emotional	5	.55	.41–.62	.75
Cognition	5	.63	.51–.75	.83
Physical Changes	3	.36	.32–.49	.57
Total	26a	.61	.29–.78	.93

Note. ^aThe total score contains the academic item that was the only item for its subscale.

Inter-rater reliability was determined for the total score using generalizability analysis (Brennan, 1983; Cronbach et al., 1972). There is a single error facet to the model, that of the rater. One may determine the reliability per rater considering only relative error (the rater by examinee interaction) or by considering both relative and absolute error (rater by examinee interaction and rater effect). The latter accounts for consistent differences in raters as error, whereas the former does not. In the case where a single rater is likely to be used, the absolute error is the better estimate to consider. In this case, the inter-rater reliability for the total score using only relative error was .94, but considering the rater’s absolute error, it remained high at .90.

PIFOS Validity

Convergent Validity: Relations With Injury Severity and Demographic Variables

First, we examined whether the total PIFOS score correlated significantly with indices of injury severity. The GCS score correlated significantly with the PIFOS total score in participants with TBI at the 3- (r = −.47, p < .001) and 12-month (r = −.43, p < .001) follow-up evaluations. Similarly, the number of days the GCS motor scale was <6 also correlated significantly with PIFOS scores at the two intervals, r = .60 and r = .50, ps < .001. The PIFOS total score correlated with ISS scores for the total group at 3 and 12 months after injury (r = .32 and r = .49, respectively, ps < .001).

The total score was significantly related to age at testing for the total group (r = .25, p = .004) but not when age at testing was restricted to children who were at least 6 years of age (r = .14, p = .167). The Hollingshead Index was not significantly related to the PIFOS total score (r = .08, p = .373).

Generalized linear models examining the impact of TBI severity on PIFOS scores did not identify any interaction of group and time of testing. As noted in Table IV, TBI severity influenced both total and subscale scores. For all analyses, multiple comparisons revealed that the mild–complicated, mild, and moderate groups did not differ significantly from each other and had significantly more favorable scores than the severe group. Motor and Social–Emotional scores improved significantly over time.

Table IV.

Impact of Traumatic Brain Injury Severity on Longitudinal Pediatric Injury Functional Outcome Scale Scores

	Time of evaluation		Factors
	3 month	12 month	Group	Time	Group × time
			df (1, 77)	df (1, 77)	df (2, 77)
Total and scores by TBI severity			F (p)	F (p)	F (p)
Total score M (SD)			9.72 (.000)	6.51 (.013)	1.10 (.338)
Mild/complicated–mild	29.9 (3.97)	29.6 (4.36)
Moderate	34.7 (9.50)	30.4 (6.82)
Severe	41.6 (11.6)	38.7 (10.4)
Motor skills			5.99 (.004)	9.45 (.003)	1.12 (.333)
Mild/complicated–mild	4.71 (1.44)	4.43 (0.938)
Moderate	5.06 (2.11)	4.25 (0.577)
Severe	6.78 (2.85)	5.68 (2.25)
Self-care			4.21 (.018)	1.20 (.276)	1.74 (.183)
Mild/complicated–mild	4.00 (0.000)	4.00 (0.000)
Moderate	4.13 (0.500)	4.13 (0.342)
Severe	5.08 (1.93)	4.52 (1.15)
Communication			7.68 (.001)	1.26 (.266)	.820 (.444)
Mild/complicated–mild	4.36 (0.842)	4.21 (0.426)
Moderate	5.00 (1.03)	4.50 (0.894)
Severe	5.60 (1.55)	5.62 (1.77)
Social–Emotional			5.58 (.005)	6.68 (.012)	.374 (.689)
Mild/complicated–mild	6.50 (1.74)	6.07 (1.69)
Moderate	7.56 (2.87)	6.50 (2.28)
Severe	8.62 (2.43)	7.98 (2.62)
Cognition			6.42 (.003)	.445 (.507)	1.70 (.190)
Mild/complicated–mild	5.64 (1.08)	6.00 (1.52)
Moderate	7.25 (3.00)	6.19 (2.23)
Severe	8.20 (2.86)	8.28 (3.14)
Physical Changes			4.90 (.010)	1.66 (.201)	1.73 (.184)
Mild/complicated–mild	3.57 (.756)	3.86 (1.03)
Moderate	4.19 (1.68)	3.69 (1.20)
Severe	5.20 (2.03)	4.66 (1.80)
Academic			8.58 (.000)	3.84 (.054)	.408 (.666)
Mild/complicated–mild	1.14 (0.535)	1.07 (0.267)
Moderate	1.50 (1.10)	1.13 (0.342)
Severe	2.14 (1.20)	1.86 (0.969)

Note. Significant effects are bolded. All pairwise comparisons indicated mild/complicated–mild = moderate < severe based on least significant difference procedure.

Known-Groups Validity: Profiles of TBI and OI Groups

To examine the profile of subscale scores, we used generalized linear modeling. A significant three-way interaction of group, time of evaluation, and PIFOS subscale scores was obtained, Wilks’s lambda (6, 119)=5.43, p < .0001. Table V provides descriptive and inferential statistics for models examining total and subscale scores at each time interval. At baseline, subscale scores indicated significantly greater impairment in the TBI group than in the OI group in communication, cognition, physical changes, and academic areas; effect sizes were large for all of these group comparisons except for the academic item, which had a medium effect size. Significant time of testing effects indicated improvement in the Motor, Self-care, Social–Emotional, Physical Changes, and Academic scores. Group × Time interactions were obtained for Motor, Self-care, and Social–Emotional subscales. For these three subscales, scores were initially elevated in the OI group and showed greater resolution in the OI group than in the TBI group by the 12-month follow-up. One year after injury, ratings of motor skills did not differ in the OI and TBI groups. In contrast, the TBI group had significantly higher scores in all other subscales, indicating continuing impairment in a variety of social, cognitive, adaptive, and academic areas relative to the subjects’ preinjury level of functioning. With the exception of the Self-care subscale, which had a medium effect size, effect sizes for the remaining subscales were large, suggesting that group membership exerted a major impact on the level of PIFOS scores.

Table V.

Longitudinal Pediatric Injury Functional Outcome Scale Total and Subscale Scores for Traumatic Brain and Orthopedic Injury Groups

	Time point						Statistic df (1, 124)
	Initial			Follow-up			Group	Time	Group × Time
PIFOS Score M (SD)	TBI	OI	Cohen’s d	TBI	OI	Cohen’s d	F (p)	F (p)	F (p)
Total score	38.19 (11.2)	36.41 (6.6)	.20	35.39 (9.8)	29.07 (3.1)	.98	8.01 (.005)	71.53 (<.001)	17.52 (.001)
Motor skills	6.08 (2.7)	7.70 (2.7)	−.60	5.18 (1.9)	4.91 (1.2)	.16	2.24 (.137)	110.28 (<.001)	24.73 (<.001)
Self-care	4.70 (1.6)	4.93 (1.5)	−.14	4.35 (0.9)	4.02 (0.2)	.43	0.13 (.720)	28.63 (<.001)	5.83 (.017)
Communication	5.26 (1.4)	4.24 (0.5)	.86	5.15 (1.6)	4.07 (0.3)	.85	29.65 (<.001)	1.90 (.171)	0.19 (.661)
Social–Emotional	8.04 (2.5)	8.37 (2.0)	−.14	7.35 (2.5)	6.17 (1.4)	.54	1.78 (.185)	60.35 (<.001)	17.98 (<.001)
Cognition	7.56 (2.8)	5.46 (0.9)	.91	7.46 (2.9)	5.28 (0.7)	.92	32.60 (<.001)	0.67 (.413)	0.12 (.733)
Physical Changes	4.71 (1.9)	4.28 (1.3)	.25	4.33 (1.6)	3.59 (0.8)	.53	5.15 (.025)	17.65 (<.001)	1.43 (.234)
Academic	1.84 (1.2)	1.43 (1.0)	.38	1.58 (0.9)	1.02 (0.2)	.80	12.67 (<.001)	19.24 (<.001)	2.71 (.102)

Note. Significant effects are bolded.

Chi square was used to compare specific items within subscales at 12 months after injury. Table VI shows the percentage of each group who had ratings of continuing disability, as indicated by scores >1, on each test item in relation to preinjury levels. In both groups, >20% of participants had persistent problems in ambulation, reduced participation in leisure activities, and physical discomfort; >30% had continuing gross motor incoordination and increased emotionality. Compared with the OI group, caregiver ratings indicated a higher proportion of the TBI group showed significant difficulties in fine motor and grooming activities as well as in all communication and cognitive areas. Behavioral changes characterized by decreased impulse control were seen in nearly half of the TBI group but occurred infrequently in the OI group. Of the 35% of children requiring school support, 15% received academic accommodations, 17.5% were receiving formal special educational support, and 2.6% were in self-contained classrooms.

Table VI.

Percentage of Each Group With Residual Disability 1 Year After Injury on Individual Pediatric Injury Functional Outcome Scale Items

				Chi square
df = 1, N = 126
Subscale	Item content	TBI	OI	χ2 (p)
Motor	Balance	17.5	10.9	1.00 (.317)
	Ambulation	21.3	32.6	1.99 (.158)
	Gross motor	35.0	37.0	0.05 (.825)
	Fine motor	17.5	2.2	6.54 (.011)
Self-care	Feeding	1.3	0	0.58 (.446)
	Dressing	5	0	2.38 (.123)
	Toileting	8.8	2.2	2.12 (.145)
	Grooming	12.5	0	6.25 (.012)
Communication	Articulation	26.3	0	14.49 (<.001)
	Receptive language	21.3	0	11.30 (.001)
	Expressive language	18.8	2.2	7.24 (.007)
	Social language	27.5	4.4	10.51 (.001)
Social–Emotional	Leisure	45.0	28.3	3.44 (.064)
	Emotional	38.8	32.6	0.48 (.491)
	Behavioral	46.3	6.6	23.31 (<.001)
	Sleep	17.5	17.4	0.00 (.988)
	Social	27.5	10.9	4.80 (.028)
	Judgment and safety	22.5	0	12.08 (.001)
Cognition	Memory	33.8	2.2	16.93 (<.001)
	Attention	32.5	13.0	5.84 (.016)
	Speed of processing	32.5	0	18.84 (<.001)
	Executive functions	36.3	6.6	13.62 (<.001)
Physical Changes	Discomfort	36.3	39.1	0.10 (.748)
	Sensory changes	28.8	0	16.18 (<.001)
	Fatigue	26.3	15.2	2.06 (.152)
Academic	School supports	35.0	2.2	17.76 (<.001)

Note. Significant effects are bolded.

Concurrent Validity: Relations With Global and Neuropsychological Outcomes

Spearman correlation analysis examined relations of the PIFOS total and subscale scores with other established measures known to be sensitive to the effects of TBI. Table VII provides the correlation coefficients for the TBI group; coefficients in bold font were significant at p ≤ .01. At the 3- and 12-month evaluations, the PIFOS total score was significantly correlated with 8 of the 10 outcome measures. The GOS score, which is the standard measure of global outcome after TBI, was strongly related to both the PIFOS total and subscale scores at initial and follow-up evaluations. The subscales were moderately correlated with the majority of the neurodevelopmental and behavioral outcomes at both evaluations. Although TBI often is associated with widespread changes in cognition and psychological health, there was some specificity in the pattern of relations of subscores and outcomes. For example, 1 year after injury, the executive function score was significantly related only to the PIFOS Cognition subscale, which contains items assessing judgment, attention, and executive processes. Internalizing behavior was significantly related to Communication, Social–Emotional, and Cognitive scales, but not to Motor skills, Self-care, or Academic subscales.

Table VII.

Spearman Correlation Coefficients for Pediatric Injury Functional Outcome Scale Total and Domain Scores of the Brain Injury Group With Global and Neurodevelopmental Outcomes at Initial and Follow-up Evaluations

		Pediatric injury functional outcome scale scores
Outcome scores by time	N	Total score	Motor skills	Self-care	Communication	Social– emotional	Cognition	Physical changes	Academic
3 months
Glasgow Outcome scale	80	.88	.75	.68	.67	.66	.74	.73	.78
Adaptive behavior	73	−.41	−.32	−.27	−.46	−.30	−.39	−.23	−.43
Fine motor speed	56	−.27	−.23	−.27	−.21	.03	−.21	−.29	−.28
General cognition	54	−.38	−.30	−.34	−.41	−.17	−.39	−.27	−.26
Executive function	73	.55	.37	.33	.50	.57	.55	.27	.46
Processing speed	51	−.41	−.33	−.45	−.44	−.03	−.31	−.40	−.43
Reading comprehension	52	−.26	−.20	−.14	−.26	−.18	−.21	−.18	−.18
School problems	44	−.50	−.26	−.29	−.47	−.19	−.47	−.54	−.56
Internalizing behavior	74	.55	.46	.31	.40	.61	.49	.26	.36
Externalizing behavior	74	.39	.21	.23	.31	.44	.42	.19	.28
1 year
Glasgow Outcome scale	80	.75	.63	.56	.53	.58	.71	.48	.76
Adaptive behavior	78	−.32	−.26	−.35	−.23	−.22	−.42	−.08	−.41
Fine motor speed	79	−.38	−.47	−.28	−.32	−.41	−.15	−.36	−.31
General cognition	54	−.40	−.35	−.36	−.33	−.28	−.28	−.25	−.41
Executive function	77	.20	.02	.23	.17	.21	.33	.07	.09
Processing speed	54	−.56	−.48	−.30	−.29	−.44	−.48	−.43	−.54
Reading comprehension	52	−.46	−.29	−.35	−.35	−.33	−.40	−.26	−.52
School problems	62	−.52	−.34	−.41	−.35	−.37	−.53	−.35	−.52
Internalizing behavior	77	.38	.21	.16	.30	.48	.33	.32	.08
Externalizing behavior	77	.24	.07	.17	.19	.27	.30	.17	.06

Note. All bolded values are significant p ≤ .01.

Because the general cognition and adaptive behavior scores were composed of different tests for the two age cohorts, we completed subgroup analyses. The Wechsler Abbreviated Scale of Intelligence scores were significantly related to the PIFOS scores and are included in Table VII. In contrast, the Stanford–Binet composite score was not related to PIFOS scores at either time point, r_s = −.04 to −.20. It is unclear whether the differential sensitivity of the general cognition scales is related to the scales themselves or to the milder injuries in the younger participants. The relation between adaptive behavior and the PIFOS scores was the same across cohorts.

For the OI group, several of the coefficients examining strength of the relation of PIFOS scores with outcomes were statistically significant. At the initial evaluation, the GOS score was significantly related to the PIFOS total score, r_s = .77, p < .001, and to the Motor, Social–Emotional, and Physical Changes subscales, r_s = .44 to .66, ps < .002. The Communication subscale was positively correlated with the Internalizing score, r_s = .41, p = .007. At the 1-year follow-up, PIFOS scores were not correlated significantly with GOS or adaptive behavior scores. The Cognition subscale was moderately correlated with the executive function score, r_s = .40, p < .007, and the Motor scale was significantly correlated with the Coding score, r_s = −.36, p < .01. As expected, neurodevelopmental outcome scores were not significantly related to PIFOS subscale scores at follow-up because of the improvement of participants with OI in most areas.

Discussion

The PIFOS, a new injury-specific multidimensional rating scale, assesses performance in relation to preinjury levels of functioning in areas relevant to children’s everyday functioning during the subacute to chronic stages of recovery from physical trauma. The PIFOS is based on a brief caregiver interview that provides a time-efficient and cost-effective method for assessing a broad range of cognitive, physical, and psychological health areas commonly impacted by pediatric injury in children aged 3–15 years. The PIFOS has a number of attributes important in health outcomes assessment instruments, including item development based on a conceptual framework, psychometric properties including reliability and validity, responsiveness to change over time, ease of interpretability, and low respondent burden (Scientific Advisory Committee of the Medical Outcomes Trust, 2002). The PIFOS is appropriate for use across a large age range, was sensitive to a broad range of injury severity, showed specificity in patterns of disability in children with TBI and OI, and was responsive to recovery during the first year after injury. Given its sensitivity to change, the PIFOS may be a useful injury-specific measure for tracking outcomes in natural history studies and intervention trials.

Psychometric analyses provided support for all hypotheses. The PIFOS met standards for psychometric properties of internal consistency and inter-rater reliability. Cronbach’s alpha increased from 3 to 12 months for the total score and for each of the subscale scores. The changes were notable for the Motor, Communication, and Social–Emotional subscales; alpha increased from 8 to 11 units. Both physical and psychological health are in flux during the early stages of recovery from both TBI and OI. The increased reliability of the subscales at 1 year may reflect the greater stability of the children’s symptoms and/or more sensitive recognition of the consequences of injury by caregivers.

Content validity was enhanced by developing items based on a formal consensus process incorporating input from specialists with expertise in both acute and long-term treatment of injured children. Validity analyses revealed favorable convergent, concurrent, and known-groups validity. The PIFOS total score was significantly correlated with other indices of injury severity, including the GCS, days of impaired consciousness, and the ISS. Concurrent validity was established through significant correlation of PIFOS scores with both global and neuropsychological outcome measures sensitive to the effects of TBI and OI. The PIFOS total score was strongly correlated with the GOS score, which is the gold standard of outcome assessment following TBI. Known-groups validity was indicated by the different profiles and patterns of change over the first year after injury in children with TBI or OI.

Given its sensitivity to the specific sequelae of different types of injuries and patterns of recovery, the PIFOS shows clinical utility. At the initial assessment, the TBI group was rated as having significantly greater disability in Communication, Cognitive, and Academic subscales; conversely, the OI group had significantly greater disability only on the Motor subscale. Social–emotional disability occurred frequently in both groups. Group × Time interactions revealed significant recovery 1 year after injury in Motor and Social–Emotional subscales, particularly in participants with OI. Both groups showed significant recovery in Self-care, Physical Changes, and Academic support areas. Despite this improvement over time, both groups showed residual disability. More than 20% of children in both groups had continuing impairment in Motor, Physical, and Social–Emotional subscales. Specific areas of disability included ambulation, gross motor coordination, physical discomfort, increased emotionality, and decreased leisure participation.

One year after injury, the effect sizes comparing TBI and OI scores for most subscales were in the medium to large range, indicating significant continuing difficulty in the TBI group. Residual disability specific to TBI was present in Cognition, Communication, Academic, and Physical Changes subscales. More than 25% of these patients had challenges in several areas, including using language in social contexts, articulation difficulties, fatigue, and sensory changes. More than 30% had difficulties with impulse control, social interaction, memory, attention, speed of processing, executive functions, and academic performance that contribute to reduced HRQOL. These residual neuropsychological and psychological health symptoms mirror findings of studies using direct assessment and parent report to characterize late outcomes after pediatric TBI (Anderson et al., 2011; Babikian & Asarnow, 2009; Ewing-Cobbs et al., 1998; Fay et al., 2009; Rivara et al., 2011; Taylor et al., 2002).

The profile in the OI group was consistent with reports of reduced physical and/or psychological health during the subacute stage of recovery (Stancin et al., 1998) and persisting throughout the first year of recovery in many children hospitalized for extremity fractures (Ding et al., 2006; Stancin et al., 2001) or major trauma (Holbrook et al., 2007). Even though motor and social–emotional symptoms persisted in at least 30% of children with OI, the GOS was not sensitive to these residual difficulties, as all but one child were rated as having a good recovery.

The GOS (Jennett & Bond, 1975) and the recent pediatric modification (Beers et al., 2012) classify outcomes into five or eight categories ranging from death to good recovery. Used in conjunction with the GOS, the PIFOS has the potential to enhance outcome assessment following pediatric injury owing to the greater breadth of item coverage, content relevant to pediatric populations, and subscale scores that may highlight specific areas of recovery or persisting difficulty. Brief global outcome assessments that cover a large age range are needed to quantify injury-specific outcomes across developmental epochs. The PIFOS items assess change in current performance relative to preinjury performance. Therefore, they may be administered to a broad age range. Because the items evaluate change, they may be used to assess the impact of injury in children at different developmental levels and in children with developmental disabilities.

Limitations of the present study include the lack of data on participation rates. The generalizability of the findings to the spectrum of injury severity is also limited by the few participants at the extremes of the severity distribution. Although children hospitalized following mild and complicated–mild TBI were included, they were underrepresented. Children with devastating TBI, such as those who were minimally conscious or who were not testable by 3 months after injury, were not included. The cohorts of younger and older children differed in terms of injury severity; the younger children had significantly milder injuries, which is consistent with the epidemiology of injuries in the preschool age range. There may be differences in the care and management of the children injured within the two recruitment phases of the study that could impact outcome. This study is also limited by the small sample size, which precluded using factor analytic approaches to examine the factor structure. Although the total score showed excellent internal consistency, reliability of subscales was variable. The profile of subscale scores may be useful for highlighting areas of disability for intervention planning and for assessment of unmet rehabilitation and academic needs; however, the total score provides a more reliable assessment of posttraumatic deficits and change over time than the subscales. In particular, additional items need to be crafted for the Physical Changes and Academic scales to improve the internal consistency. Doubling the number of items, assuming similar intercorrelations, would raise the reliability to more acceptable levels. Future studies should expand the normative base for injury populations and assess applicability of the PIFOS to children with other acute and chronic health conditions.

Acknowledgments

We gratefully acknowledge the contribution of the Children’s Memorial Hermann Hospital Trauma Team in providing expert consultation regarding item development, and Jennifer Hendricks, MEd, and Linda Winzeler, MA, for completing the interviews.

Funding

National Institutes of Health NIH-R01 NS046308 and R01 NS029462 awarded to L.E.C. The content is solely the responsibility of the authors and does not necessarily represent the official views of the granting institutes.

Conflict of interest: None declared.