Recovery of Olfactory Function following Pediatric Traumatic Brain Injury: A Longitudinal Follow-Up

Abstract

There is increasing evidence that disruption of olfactory function after pediatric traumatic brain injury (TBI) is common. Olfactory dysfunction (OD) has been linked to significant functional implications in areas of health, safety, and quality of life, but longitudinal research investigating olfactory recovery is limited. This study aimed to investigate recovery trajectories for olfaction following pediatric TBI and explore predictors of early and late olfactory outcomes. The olfactory function of 37 children with TBI ages 8–16 years was assessed on average at 1.5, 8.0, and 18.0 months post-injury using the University of Pennsylvania Smell Identification Test. A significant improvement in olfactory performance was seen over time in those with acute OD, however, only 16% of those with the most severe OD showed recovery to normal olfactory function, with the remainder demonstrating ongoing olfactory impairment at the 18 month follow-up. Predictors of early (0–3 month) and late (18 month) olfactory outcomes varied with site of impact, a significant predictor of later olfactory performance. In summary, while there was evidence of recovery of OD over time in pediatric TBI, the majority of children with severe OD did not show any recovery. In light of limited recovery of function for more severely affected children, the importance of appropriate education and implementation of rehabilitation management strategies is highlighted.

Introduction

Research investigating olfactory outcomes in pediatric traumatic brain injury (TBI) is limited, with a recent systematic review¹ identifying only a handful of published pediatric studies. Results from these studies indicate widely varying estimates of the frequency of olfactory dysfunction (OD) post–pediatric TBI, with figures ranging from 3–58%.^2
–4 More recently, in a prospective study of olfactory dysfunction in a pediatric TBI cohort of mixed severity (Bakker and colleagues, in submission), almost 20% of children demonstrated impaired olfaction in the acute recovery phase, with 5% documented as having anosmia. These results are in line with contemporary estimates of OD in adult TBI in the first few months post-injury,^5,6 and add to the growing evidence that OD post–pediatric TBI is a relatively common clinical sequelae.

Despite this, knowledge of recovery patterns in OD following TBI remain unclear. Early reports and studies involving large adult cohorts provide evidence of limited recovery of olfaction, ranging from 8–39% of those with OD,^7

–10 with timing of recovery ranging from 5 days to 5 years post-injury. The majority of adults demonstrating OD showed recovery in the first 3–6 months post-injury^8,9 and, while late recovery was reported, this was slow and increasingly unlikely to occur after 12 months post-injury.⁹ A number of methodological challenges in the existing adult TBI literature make it difficult to ascertain olfactory recovery trajectories after TBI, with studies limited by the use of outcome measures with limited psychometric properties,^7
–9 reliance on self-report,¹⁰ and variability in follow-up times.

A more recent body of research using standardized olfactory assessment methods in adults with TBI recruited via otorhinolaryngology clinics or specialist smell and taste centers report rates of improvement in olfactory performance from 10–42% post-TBI.^11

–15 Despite documented improvement in olfactory performance, rates of recovery to normal olfactory function are much lower, around 5–15% when reported.^16,17 Given the recruitment setting of specialist smell and taste centers, these studies are biased towards cohorts with long-term or chronic OD, which limit their generalizability. Time from injury to initial assessment is often extended, thus, those with early recovery of OD are unlikely to be included. In addition, wide variation in follow-up times within and between studies make it difficult to gain a clear picture of recovery trajectories.

More recently, a prospective longitudinal study of an adult TBI cohort incorporating standardized assessment of olfactory function demonstrated recovery in 39% of the cohort with OD at the 12 month follow-up.⁵ Recovery appeared to be related to symptom severity, with those who were categorized as anosmic acutely, exhibiting no recovery over time.

The underlying mechanisms for recovery of olfactory function post-TBI are not well understood. Sumner⁸ proposed different underlying injury mechanisms to explain his finding of early and late recovery in his adult cohort. Recovery within the first 3 months post-injury was felt to be due to resolution of underlying transient pathology, such as contusion, hemorrhage, inflammation, and edema,^8,11,16 while later recovery was considered to be in line with slower underlying neural recovery.^8,16 While ongoing neurogenesis is known to occur in the olfactory epithelium,¹⁶ recovery of olfactory function also may be dependent on the location and degree of injury to olfactory system components and damage to certain components of the olfactory system, such as basal cells layers in the neuroepithelium, may not be reversible.¹⁶ The role of scar tissue and gliosis after lesions to olfactory nerves and olfactory bulb in impeding axonal growth from nerve cells to the olfactory bulb also has been suggested as a rationale for failure of recovery in olfactory function post-injury.^15,16

Contradictory findings have been reported for factors that may be predictive of acute and longer-term olfactory outcomes following TBI.¹⁷ Injury variables such as severity,^9,18

–21 site of impact^8,11 and presence of skull fractures^22,23 have been linked to olfactory outcomes, although findings have not been consistently demonstrated.^{5,11,22,24,25} These variable findings may be consistent with theories of multiple underlying mechanisms for olfactory dysfunction post-TBI. Sigurdardottir and colleagues'⁵ findings of effects of severity at 3 months post-injury, although no significant findings of severity group differences at the 12 month follow-up, may suggest that different injury variables may be predictive of early versus later outcomes in adults in line with differences in underlying causes of OD, and support the importance of longitudinal versus cross-sectional outcomes research.

While it is tempting to generalize adult findings directly to children with TBI, previously identified differences in child and adult TBI outcomes in a variety of neurocognitive areas make extrapolating findings in adult literature to pediatric TBI cohorts problematic.^26,27 Unfortunately, to date, olfactory recovery patterns following pediatric TBI remain unclear, with existing pediatric studies limited to a small number of case reports^28,29 and a large single group study.² Hagan²⁸ reported a single case study of a 16-year-old girl with hyposmia, evident at 1 month post-TBI, who showed resolution of OD by re-assessment 2 months later. In contrast Jimenez,²⁹ in a case series of five patients, reported on a 17-year-old male with anosmia followed up 2 years post-injury who showed no gains in olfactory function over time. Evidence of recovery of olfaction at a group level was reported by Jacobi and colleagues,² with the percentage of children demonstrating anosmia reducing from 3.3% at initial testing to 1.2% at the 6 month follow-up. The limited generalizability of single case reports,^28,29 use of unstandardized²⁹ or unknown² assessment measures, and retrospective study designs² in the of small body of existing pediatric research severely limits our ability to gain a clear picture of olfactory function recovery in a pediatric TBI population.

Without knowledge of recovery trajectories of olfactory function following pediatric TBI, it is difficult to provide evidence-based rehabilitative management and education for families and sufferers. Such management is particularly important in light of the significant functional implications known to be associated with OD.^30

–33 Understanding which injury variables predict early and late olfactory outcomes is important from a prognostic and management perspective in identifying those most at risk of olfactory deficits and thus in need of clinical follow-up services.

To our knowledge, this is the first prospective longitudinal study to investigate recovery of OD following pediatric TBI using a standardized olfactory assessment. The main aim of the study was to document recovery patterns of OD to 18 months post–pediatric TBI, and to ascertain whether severity of injury influences recovery trajectory. To achieve this, we explored recovery across the 18 months post-injury for the total group (Aim 1a), different injury severity groups (mild vs. moderate/severe; Aim 1b), and for participants with OD acutely (Aim 1c). A secondary aim was to understand which injury factors are predictive of early (0–3 post-injury; Aim 2a) and late (18 months post-injury; Aim2b) olfactory outcome in pediatric TBI.

Methods

Participants

In this prospective, longitudinal, single site observational study, children ages 8–16 who sustained a TBI were recruited after presentation to the emergency department or referral to the rehabilitation department, of the Royal Children's Hospital (RCH) between April 2010 and August 2012. Inclusion criteria were 1) age 8–16 at time of injury and 2) documented evidence of traumatic brain injury (period of altered or loss of consciousness, period of coma, period of post-traumatic amnesia or retrograde amnesia, and/or neuroimaging evidence of brain trauma).³⁴ Exclusion criteria included 1) history of documented prior brain injury, developmental, learning, neurological or neuropsychiatric disorder and 2) pre-morbid difficulties with smell or taste.

Two hundred and twenty-three children and their families were invited to participate. Sixty-two families were not able to be contacted and were excluded. A further 42 children met exclusion criteria and five failed to attend follow-up appointments. Seventy-seven families declined participation. The resultant sample included 37 children. The un-enrolled eligible group (n = 77) did not differ significantly from the enrolled group (n = 37) in terms of age at injury (t [112] = −0.38; p = 0.71), socio-economic factors, as measured by the Index of Relative Social Advantage and Disadvantage³⁵ (Z = − 0.6; p = 0.55), gender (χ2 [2] = 3.78; p = 0.14), or severity of injury (χ2 [2] = 5.07; p = 0.07).

Participants were followed up at three time-points post-injury: T1 = 0–3 months; T2 = 8 months; and T3 = 18 months. Criteria for inclusion in longitudinal data analysis required participation in at least two of the three assessment time-points. Of the original 37 participants, three were excluded due to missing data at T2 and T3. Of those excluded, two had declined further participation and one was lost to follow-up. All three excluded participants sustained moderate-severe TBI.

Complete olfactory function data were available for 34 participants at T1, 28 at T2, and 32 at T3. Missing data at each time-point were due to participants being temporarily uncontactable at the time of assessment, although overall study retention rate from T1 to T3 was high at 87%.

Children were assigned to two groups on the basis of severity of injury. Mild TBI was defined as: initial Glasgow Coma Scale³⁶ (GCS) score 13–15 and/or period of post-traumatic amnesia (PTA) <24 h, with or without neuroimaging evidence of intracranial injury, and not requiring surgical intervention. Moderate/severe TBI was defined as initial GCS of ≤12 and/or period of PTA ≥24 h.

Materials

Demographic and injury information

Information regarding gender, age at injury, premorbid medical and developmental history, and premorbid smell and taste function were collected from medical records and from parents via questionnaire. Medical records were reviewed to provide information on injury indices, such as initial GCS, period of PTA, neuroimaging information, site of impact, cause of injury, neurosurgical intervention, presence or absence of skull fracture, and dysphagia.

General intellect

Intellectual functioning was assessed using the Wechsler Abbreviated Scale of Intelligence (WASI)³⁷ or Wechsler Intelligence Scale for Children, 4th Edition.³⁸ Full scale IQ scores (mean [M] = 100, SD = 15, calculated using the two subtest method for the WASI)³⁷ and T scores (M = 50, SD = 50) for vocabulary and matrices subtests were derived to determine general intellectual functioning (FSIQ) and estimates of verbal (VIQ) and perceptual reasoning (PRIQ).

Socio-economic status

Relative social risk for participants was calculated using the Social Risk Index (SRI).³⁹ The index uses information regarding parental age, education, occupation, family constellation, and language to derive a measure of social status and risk. Higher scores indicate increased social risk, with scores ranging from 0 to 12.

Primary outcome measure (smell identification)

The University of Pennsylvania Smell Identification Test (UPSIT)⁴⁰ was used as the primary outcome measure of olfactory function. The UPSIT is a 40-item forced choice “scratch and sniff” test assessing smell identification. Possible scores range from 0–40. The UPSIT provides an absolute index of smell loss, as well as percentile data, for 5–90 year olds. Categorizations include normosmia (normal olfactory functioning), mild, moderate, and severe microsmia (diminished olfactory function), and anosmia (absence of olfactory function).⁴⁰ Previously identified performance differences between Australian cohorts and the North American standardization sample have resulted in a recommended correction of +2 raw score points.⁴¹ As well as assigning participants into categories on the basis of their olfactory function (normosmic, microsmic, anosmic), performance below the 5th percentile was taken to indicate abnormal olfactory function. Percentile scores obtained from the manual allow for assignment of percentile scores down to 5th percentile but do not provide specific percentiles below this point. Percentile scores below the 5th percentile were coded as at the 2nd percentile (as the midpoint of possible scores below 5th percentile) to minimize floor effects. Corrected percentile scores and categorization were collected for use in analysis.

Procedure

The study was approved by the Human Ethics Research Committee, RCH, Melbourne, Australia, and carried out in accordance with National Privacy Legislation and the Code of Ethical Principles for Medical Research Involving Human Subjects of the World Medical Association (Declaration of Helsinki).

Three recruitment strategies were employed, with recruitment via 1) emergency department; 2) inpatient rehabilitation program; and 3) outpatient brain injury clinic at RCH. For those recruited via the emergency department, a letter outlining details of the study was sent to all potentially eligible families inviting them to participate in the research project. Follow-up phone calls were made to those who did not respond to the original invitation within 2 weeks. Potential participants admitted to the inpatient rehabilitation program were approached directly on the ward, once medically stable and out of PTA, and families were provided with detailed information sheets and consent forms. Children seen for routine outpatient brain injury clinic follow-up 4–8 weeks post-injury were approached directly and families provided with detailed information and consent forms. For consenting families in all recruitment strategies above, children were screened for exclusion criteria and those deemed eligible completed testing immediately or an appointment was scheduled for initial testing at a later date.

Follow-up occurred at three time-points post-injury: Time 1 (T1), 0–3 months post-injury (M = 1.58, SD = .63); Time 2 (T2), 8 months post-injury (M = 8.42, SD = 2.08); and Time 3 (T3), 18 months (M = 18.12, SD = 3.83) post-injury. At each assessment, children were screened for conditions that could impact on smell abilities (e.g., upper respiratory tract infection) and testing rescheduled if necessary.

All assessments were conducted either at the hospital or in the child's home. Tests were administered by an experienced neuropsychologist in a fixed order. The UPSIT was administered following standard administration procedures with each odorant presented to both nostrils. As a concession to age, the four alternative answers were read aloud to children, although children also were able to read each option. At T1, parents completed the demographic questionnaire while children were assessed. The UPSIT was completed by children at each of the three time-points (T1-T3). Intellectual assessment was completed at T2 only along with measures of executive function and day-to-day olfactory function that are not presented in this article.

Statistical analysis

Data were analyzed using SPSS Statistical Package Version 22.0 (IBM Corp., Armonk, NY). Independent t-tests and chi-square analysis were used to compare participating and non-participating children and severity groups on demographic and injury variables. Effect sizes were calculated using Cohen's d (0.3 = small, 0.5 = moderate and 0.8 = large) and phi (φ) coefficient (0.1 = small, 0.3 = moderate and 0.5 = large) where appropriate.⁴²

Due to missing data, a linear mixed model with an unstructured covariance matrix was used to examine change over time in olfactory function. Analysis was completed to investigate changes in olfactory function for Total group (Aim 1a) and to compare recovery trajectories between the mild and moderate/severe TBI groups (Aim1b) over the 18 months post-injury. Preliminary analysis indicated a significant difference between the mild and moderate/severe TBI groups in VIQ and age at injury; as a result, these were used as covariates in the linear mixed model. A linear mixed model was then used to explore recovery patterns for the sub-group exhibiting olfactory dysfunction at T1 (n = 20; Aim 1c).

Linear regression was conducted to examine predictors of UPSIT performance at T1 and T3 for total group (n = 34; Aim 2a and 2b). Exploratory univariate regression analysis, with alpha set at 0.1, was conducted with theoretically-derived predictor variables. Variables meeting significance criteria were then put into regression model to examine their relative contributions with standard alpha of 0.05.

Results

Demographics and injury characteristics

Demographic and injury data for eligible participants are presented in Table 1. Comparison of mild and moderate/severe TBI groups indicated no significant difference in terms of gender or socioeconomic status. The two groups were significantly different in terms of age at injury with the moderate/severe group being older. The groups differed, as expected, on measures of injury severity including, period of PTA, initial GCS, need for neurosurgery, and presence of acute dysphagia.

Table 1.

Comparison of Mild and Moderate/Severe Traumatic Brain Injury Groups on Demographic and Injury Characteristics

	Total	mTBI	m/sTBI	Significance	Effect size
n (Assessment 1)	34	22	12
Gender, female, n	12	6	6	χ ² (1) = 1.76, p = 0.27	φ = 0.23
Age at injury (years), M (SD)	12.31 (1.93)	11.68 (1.81)	13.45 (1.65)	t (32) = −2.81, p = 0.008	d = 1.02
SRI, M (SD)	1.94 (2.18)	1.48 (1.91)	2.75 (2.45)	t (31) = −1.66, p = 0.11	d = 0.51
Injury characteristics
GCS, M (SD)	13.42 (2.6)	14.59 (0.59)	11.09 (3.45)	t (31) = 4.7, p < 0.0005	d = 1.73
PTA days, M (SD)	4.46 (8.87)	0.10 (0.16)	12.46 (11.31)	t (32) −5.19, p < 0.0005	d = 2.16
Positive neuroimaging, n	13	4	9	χ ² (1) = 3.49, p = 0.1	φ = 0.39
Neurosurgery, n	3	0	3	χ ² (1) = 6.03, p = 0.04	φ = 0.42
Dysphagia, n	3	0	3	χ ² (1) = 6.03, p = 0.04	φ = 0.42
Skull fracture, n	8	3	5	χ ² (1) = 3.39, p = 0.1	φ = 0.31

mTBI, mild traumatic brain injury; m/sTBI, moderate/severe traumatic brain injury; M, mean; SD, standard deviation; SRI, Social Risk Index; GCS, Glasgow Coma Scale; PTA, post-traumatic amnesia.

Mean FSIQ, VIQ, and PRIQ scores for total, mild, and moderate/severe group are reported in Table 2. Analysis indicated a significant difference between the mild and moderate/severe groups on VIQ, with children in the moderate/severe group demonstrating significantly lower VIQ (M = 43.5; SD = 6.57) than those in the mild TBI group (M = 50.95; SD = 7.49; t[23] = 2.18; p = 0.04; d = 1.06). There were no significant differences between the groups in terms of FSIQ (t [25] = 1.83; p = 0.08; d = 0.86), or PRIQ (t [23] = 1.48; p = 0.15; d = 0.68).

Table 2.

Comparison of Mild and Moderate/Severe Traumatic Brain Injury Groups on Measures of Intellectual Function

	Total M (SD)	mTBI M (SD)	m/sTBI M (SD)
FSIQ (St. score), n = 27	99.78 (11.15)	102 (11.21)	93.43 (8.83)
VIQ^* (T score), n = 25	49.16 (7.85)	50.95 (7.49)	43.5 (6.57)
PRIQ (T score), n = 25	50.72 (8.21)	52.05 (7.88)	46.5 (8.48)

p = 0.04.

mTBI, mild traumatic brain injury; m/sTBI, moderate/severe traumatic brain injury; M, mean; SD, standard deviation; FSIQ, full scale intelligence quotient; VIQ, verbal intelligence quotient; PRIQ, perceptual reasoning intelligence quotient.

Recovery of olfaction (Aim 1)

Total group

The Total group demonstrated a significant improvement in olfactory performance over time (F [2, 30.96] = 4.76; p = 0.02).

Injury severity groups

Exploration of differential olfactory function recovery between the mild and moderate/severe TBI groups indicated no significant findings. No significant main effect of time (F [2, 23.1] = 1.49; p = 0.25), group (F [1, 20.1] = .26; p = 0.61), or interaction (F [2, 23.1] = .41; p = 0.67) was found, although there was a significant main effect of the covariate VIQ (F [1,18.64] = 16.39; p = 0.001).

Olfactory dysfunction group

Exploration of recovery of olfaction in the group demonstrating OD at T1 (n = 20) indicated a significant main effect of time (F [2, 17.93] = 9.1; p = .002. The pattern of recovery observed for the OD group is illustrated in Figure 1. Post hoc pairwise comparisons with Bonferroni correction for multiple comparison indicated that significant recovery occurred from T1 to T2 (p = 0.003, d = 1.42) and T1 to T3 (p = 0.009, d = 0.97), with large effect sizes observed. Comparison of olfactory function from T2 to T3 indicated no significant difference (p = 0.7, d = 0.56), with a medium effect size observed.

FIG. 1.

Change in olfactory function over time.

Of the 20 participants demonstrating OD at T1, which ranged from mild microsmia to anosmia, six demonstrated more significantly impaired olfaction (with performance below the 5th percentile). Information about performance and injury characteristics for this more significantly impaired group is outlined in Table 3. Only one of these six participants showed improvement over time, with evidence of return to normal olfactory function at T3, while the majority continued to demonstrate significant impairment in olfactory function, with performance remaining below the 5th percentile at the 18 month follow-up.

Table 3.

Injury and Demographic Characteristics and Examination of Recovery of Olfaction for Those with Initial UPSIT Score below 5th Percentile

						UPSIT T1		UPSIT T3
Id	Age at injury (years)	Gender	Injury severity	Site of impact	Skull fracture	crs (%ile)	category	crs (%ile)	category
3	12.3	M	MIL	UN	N	23 (< 5)	MOD MIC	19 (< 5)	SEV MIC
7	15	F	SEV	UN	N	21 (< 5)	SEV MIC	23 (< 5)	SEV MIC
9	11.33	F	SEV	PAR	Y	26 (< 5)	MOD MIC	33 (16)	NORM
25	14.25	F	MOD	FRO	N	24 (< 5)	MOD MIC	25 (< 5)	SEV MIC
29	10.97	M	MOD	OCC	Y	19 (< 5)	SEV MIC	16 (< 5)	ANO
37	12.94	M	MOD	OCC	Y	13 (< 5)	ANO	14 (< 5)	ANO

UPSIT, University of Pennsylvania Smell Identification Test; T1, time 1 (0–3 months post injury); T3 time 3 (18 months post injury); crs, corrected raw score; %ile, percentile score; M, male; F, female; MIL, mild; MOD, moderate; SEV, severe; UN, unknown; PAR, parietal; FRO, frontal, OCC, occipital, N, no; Y, yes; MOD MIC, moderate microsmia; SEV MIC, severe microsmia; ANO, anosmia, NORM, normosmia.

Predictors of olfactory outcome—acute and late (Aim 2)

Acute (0–3 months post-injury)

Univariate linear regression was conducted to examine the independent contribution of theoretically-derived predictor variables, including PTA, age at injury, time post-injury, presence of skull fracture, evidence of intracranial injury, and occipital and frontal site of injury impact to olfactory performance at T1. Those variables meeting pre-set criteria of significance (p ≤ 0.1) were included in the final regression model.

Duration of PTA was the only variable making a substantial contribution to variance in UPSIT performance at T1, with the effect of PTA duration approaching significance (p = 0.05) and explaining 11.2% of the variance in UPSIT score acutely (adjusted r² = 0.09), with those with longer duration of PTA showing poorer olfactory performance at T1.

Late (18 months post-injury)

A similar procedure was followed to examine predictors of UPSIT performance at T3, with additional predictor variables of FSIQ, VIQ, PRIQ, and UPSIT T1 included in initial exploratory univariate regression. The final regression model included PTA, occipital impact, and VIQ as independent variables. While FSIQ, positive neuroimaging, and frontal impact all met pre-set criteria for inclusion, these were excluded from the model given significant inter-correlations with other independent variables to ensure no violation of the assumptions of multicollinearity.

Overall the model including PTA, VIQ, and occipital impact was significant (p = 0.03) and explained 39% of the variance in UPSIT performance at T3 (adjusted, r² = 29.1%). Occipital impact was the only variable that made a significant individual contribution to olfactory performance at T3 (p = 0.04). Table 4 outlines characteristics of group with and without occipital impacts. Neither PTA nor VIQ showed significant independent contributions to UPSIT performance at T3.

Table 4.

Comparison of Participants with Occipital and Non-Occipital Impacts on Injury Variables and Olfactory Outcome at T1 and T3

	OCC (n = 13)	NON-OCC (n = 17)
PTA, M (SD)	2.39 (3.23)	5.45 (11.11)
GCS, M (SD)	13.83 (1.7)	13.24 (3.03)
Gender, n (males)	8	12
Age at injury, years, M (SD)	12.04 (2.12)	12.16 (1.85)
Positive neuroimaging, n	7	4
UPSIT corrected percentile
T1, M (SD)	16.08 (14.4)	23.18 (23.02)
T3, M (SD)^*	11.85 (10.30)	25.67 (15.02)

p = 0.04.

OCC, occipital; NON-OCC, non-occipital (includes frontal, n = 11; temporal, n = 4; parietal, n = 2); PTA, post-trauamtic amnesia; M, mean; SD, standard deviation; GCS, Glasgow Coma Scale; UPSIT, University of Pennsylvania Smell Identification Test; T1, time 1 (0–3 months post-injury); T3, time 3 (18 months post-injury).

Discussion

To our knowledge, this is the first prospective longitudinal study using standardized olfactory assessment to investigate olfactory recovery in pediatric TBI. The results of our study indicate overall group recovery, but no differential effects across injury severity groups in line with Aims 1a and 1b. For those initially presenting with OD (Aim 1c), there was evidence of significant improvement in olfactory function at T2 (8 month follow-up). Recovery for our cohort occurred in the early stage post-injury with no significant further recovery in olfactory function observed at T3 (18 months post-injury).

While recovery occurred at a group level for those showing acute OD, examination of the subgroup demonstrating more significant impairments (performance below the 5th percentile) indicated that recovery was more limited in this group. In our cohort only one participant with performance below the 5th percentile at T1 showed improvement in olfactory function at T3, with performance returned to normal levels. The remaining five participants continued to be categorized with performance below the 5th percentile at 18 months post-injury.

The recovery trajectory observed in our pediatric cohort is consistent with previous literature in adult TBI indicating recovery within the first 6 months post-TBI, but limited recovery or change in performance after this time.^8,9 Similarly the low recovery rate observed for those with more severe OD (16%) are consistent with findings of Sigurdardottir and colleagues⁵ in their adult TBI group showing evidence of recovery at the 12 month follow-up for their total cohort, but no evidence of recovery in the anosmic group, and these findings are consistent with the common report of poor prognosis for those suffering OD post-TBI.^8,9 We found no evidence of a significant main effect of group or any interaction effects, indicating that there was no differential pattern of recovery relative to severity once language differences were controlled for in the mild and moderate/severe TBI groups.

The longitudinal design of our study allowed for the investigation of predictors of olfactory outcome in both acute and long-term stages post-injury (Aim 2), in contrast to earlier cross-sectional study designs.^{18,20,22,24,25} Severity of injury—as measured by duration of PTA—was the only injury factor to make a substantial contribution to olfactory function at 0–3 months post-injury (Aim 2a). Those with more severe injury tended to demonstrate poorer olfactory function acutely. While non-significant, this finding is consistent with previous literature demonstrating evidence of a dose–response relationship between injury severity and olfactory performance.^9,18

–21 Other acute injury factors that have been linked to olfactory performance, such as site of injury impact,^8,11 evidence of intracranial injury,²² and presence of skull fracture,^22,23 were not significantly predictive of acute olfactory outcomes in our pediatric cohort.

We found no overlap between early and late predictors of olfactory outcome. Occipital site of impact was a significant predictor of later (18 month) olfactory performance, with injury severity and VIQ failing to make a significant individual contribution to prediction of late olfactory performance (Aim 2b). Our results extend adult findings to pediatric TBI, that those with occipital impacts are more likely to exhibit olfactory impairments,^8,11 with OD possibly a result of contre-coup injury mechanisms, resultant shearing of olfactory nerve filaments, and damage to orbitofrontal cortex in these individuals.

The results suggest the significance of different predictors of olfactory outcome in acute versus long-term post-injury. These results are interesting given the inconsistent findings in previous largely cross-sectional studies regarding predictors of OD post-TBI. While not a significant finding, the observed relationship between injury severity and early olfactory performance would be consistent with Sumner's⁸ theory of olfactory disturbance at this acute stage as having a more generalized cause related to transient pathology such as contusion, hemorrhage, inflammation, and edema. As these generalized effects of injury and underlying transient pathology clear an improvement in olfactory function may be seen. Ongoing OD, in contrast, may relate to direct neural damage involving the olfactory system, and this is supported by the finding that site of impact positively predicts later olfactory performance, as opposed to injury severity. In these individuals with direct neural damage, the prospects of recovery appear more limited.

Our study is not without its limitations. The participation rate of potentially eligible subjects was low; however, those who declined participation did not significantly differ from the final cohort and so our sample is representative of a general pediatric TBI cohort. The small size of our original study cohort means that our group classified as having impaired olfaction is necessarily small. This limits the analysis that could be done in terms of change over time, specifically within the impaired group. Moreover as the impaired participants often performed at floor levels on the UPSIT, relative changes or relative differences within the impaired group or over time could not be examined.

A common limitation of longitudinal research is that of missing data, and this was a factor in our study, as well. The use of linear mixed models allowed for analysis with missing observations, thus not requiring further exclusion of participants. A high number of missing observations at T2 may impact on the interpretation of results at that time-point; however, the high overall retention rate of 87% does allow for examination of recovery over time. A control group was not included for comparison with the OD group in terms of recovery trajectory. The use of a well-validated standardized test of smell identification with age norms to track trajectory of olfactory function post-TBI is a strength of the study and allows for a picture of change in olfactory function over time. The inclusion of an appropriate control group in future studies would strengthen the evidence of olfactory recovery trajectories in pediatric TBI.

Variability in our sample in terms of neuropathology meant that this could not be meaningfully examined in our cohort in terms of its ability to predict olfactory outcomes. Location of neuropathology has been linked to olfactory outcomes in adult TBI with frontal neuropathology related to olfactory impairment.²² It would be reasonable to hypothesize that evidence of frontal neuropathology may be an accurate predictor of olfactory impairment in pediatric TBI and will be important to consider in future pediatric TBI research.

While the above limitations make us necessarily guarded about the generalizability of our results, in the absence of previous prospective longitudinal studies of olfactory recovery following pediatric TBI, the findings of our study are significant. In general, it appears that recovery of olfaction can occur following TBI in children and that this recovery tends to occur in the early stages post-injury within 9 months. Recovery for those presenting with more severe OD is limited. Severity of injury (as measured by duration of PTA) may impact on the expression of OD acutely, and early recovery may be related to resolution of early, generalized effects of injury. Children at greatest risk of poorer late outcomes in olfactory function are those with an occipital site of impact—which may be an indicator of direct underlying damage to olfactory connections or neural damage in orbitofrontal areas via contre-coup injury mechanisms.

Given the longstanding nature of olfactory deficits post–pediatric TBI, the limited prospects of recovery in those more severely impaired, and the known functional implications of OD, the importance of screening assessment and educational and rehabilitative management is paramount. While a proportion of those exhibiting acute olfactory dysfunction are likely to show improvements over time, those who exhibited olfactory impairments more than 9 months post-injury are likely to have ongoing deficits; therefore counseling about likely prognosis, functional implications, appropriate management, and safety precautions should be offered as a matter of course.

Footnotes

Acknowledgments

This work was supported by funding from the Victorian Neurotrauma Initiative, Murdoch Childrens Research Institute, the Victorian Government Operational Infrastructure Scheme, and an Australian Postgraduate Award from the University of Melbourne.

The authors thank Dr. Peter Barnett, Mr. Ken Robinson, Dr. Kevin Dunne, and the staff of the Emergency Department and Victorian Paediatric Rehabilitation Service, RCH, for assistance with recruitment, and Dr. Stephen Hearps for assistance with data analysis.

Author Disclosure Statement

No competing financial interests exist.

References

Bakker

, Catroppa

, and Anderson

(2014). Olfactory dysfunction in pediatric traumatic brain injury: a systematic review. J. Neurotrauma, 31, 308–314.

Jacobi

, Ritz

, and Emrich

(1986). Cranial nerve damage after paediatric head trauma: a long-term follow-up study of 741 cases. Acta Paediatr. Hung., 27, 173–187.

Roberts

M.A.

and Simcox

A.F.

(1996). Assessing olfaction following pediatric traumatic brain injury. App. Neuropsychol., 3, 86–88.

Sandford

A.A.

, Davidson

T.M.

, Herrera

, Gilbert

, Magit

A.E.

, Haug

, Gutglass

, and Murphy

(2006). Olfactory dysfunction: a sequela of pediatric blunt head trauma. Int. J. Pediatr. Otorhinolaryngol., 70, 1015–1025.

Sigurdardottir

, Jerstad

, Andelic

, Roe

, and Schanke

A.-K.

(2010). Olfactory dysfunction, gambling task performance and intracranial lesions after traumatic brain injury. Neuropsychology, 24, 504–513.

de Kruijk

J.R.

, Leffers

, Menheere

P.P.C.

, Meerhoff

, Rutten

, and Twijnstra

(2003). Olfactory function after mild traumatic brain injury. Brain Inj. 17, 73–78.

Leigh

A.D.

(1943). Defects of smell after head injury. Lancet, 241, 38–40.

Sumner

(1964). Post-traumatic anosmia. Brain, 87, 107–120.

Costanzo

R.M.

and Becker

D.P.

(1986). Smell and taste disorders in head injury and neurosurgery patients. In: Clinical Measurements of Taste and Smell. Meiselman

H.L.

and Rivlin

R. S.

(eds). Macmillan Publishing Company: New York, pps. 565–578.

10.

Zusho

(1982). Posttraumatic anosmia. Arch Otolaryngol. 108, 90–92.

11.

Doty

R.L.

, Yousem

D.M.

, Pham

L.T.

, Kreshak

A.A.

, Geckle

, and Lee

W.W.

(1997). Olfactory dysfunction in patients with head trauma. Arch. Neurol., 54, 1131–1140.

12.

London

, Nabet

, Fisher

A.R.

, White

, Sammel

M.D.

, and Doty

R.L.

(2008). Predictors of prognosis in patients with olfactory disturbance. Ann. Neurol., 63, 159–166.

13.

Welge-Lussen

, Hilgenfeld

, Meusel

, and Hummel

(2012). Long-term follow-up of posttraumatic olfactory disorders. Rhinology. 50, 67–72.

14.

Duncan

H.J.

and Seiden

A.M.

(1995). Long-term follow-up of olfactory loss secondary to head trauma and upper respiratory tract infection. Arch. Otolaryngol. Head Neck Surg., 121, 1183–1187.

15.

Reden

, Mueller

, Konstantinidis

, Frasnelli

, Landis

B.N.

, and Hummel

(2006). Recovery of olfactory function following closed head injury or infections of the upper respiratory tract. Arch. Otolaryngol. Head Neck Surg., 132, 265–269.

16.

Costanzo

R.M.

(2005). Regeneration and rewiring the olfactory bulb. Chem. Senses, 30, 133–134.

17.

Kobayashi

and Costanzo

R.M.

(2009). Olfactory nerve recovery following mild and severe injury and the efficacy of dexamethasone treatment. Chem. Senses, 34, 573–580.

18.

Callahan

C.D.

and Hinkebein

J.H.

(2002). Assessment of anosmia after traumatic brain injury: Performance characteristics of the University of Pennsylvania Smell Identification Test. J. Head Trauma Rehabil., 17, 251–256.

19.

Costanzo

R.M.

and Zasler

N.D.

(1992). Epidemiology and pathophysiology of olfactory and gustatory dysfunction in head trauma. J. Head Trauma Rehabil., 7, 15–24.

20.

Green

, Rohling

M.L.

, Iverson

G.L.

, and Gervais

R.O.

(2003). Relationships between olfactory discrimination and head injury severity. Brain Inj. 17, 479–496.

21.

Fujiwara

, Schwartz

M.L.

, Gao

, Black

S.E.

, and Levine

(2008). Ventral frontal cortex functions and quantified MRI in traumatic brain injury. Neuropsychologia. 46, 461–474.

22.

Haxel

B.R.

, Grant

, and Mackay-Sim

(2008). Olfactory dysfunction after head injury. J. Head Trauma Rehabil., 23, 407–413.

23.

Renzi

, Carboni

, Gasparini

, Perugini

and Becelli

(2002). Taste and olfactory disturbances after upper and middle third facial fractures: a preliminary study. Ann. Plast. Surg., 48, 355–358.

24.

Callahan

C.D.

and Hinkebein

(1999). Neuropsychological significance of anosmia following traumatic brain injury. J. Head Trauma Rehabil., 14, 581–587.

25.

Neumann

, Zupan

, Babbage

D.R.

, Radnovich

A.J.

, Tomita

, Hammond

, and Willer

(2012). Affect recognition, empathy, and dysosmia after traumatic brain injury. Arch. Phys. Med. Rehabil., 93, 1414–1420.

26.

Anderson

, Spencer-Smith

, Leventer

, Coleman

, Anderson

, Williams

, Greenham

, and Jacobs

(2009). Childhood brain insult: can age at insult help us predict outcome?. Brain, 132, 45–56.

27.

Crowe

L.M.

, Catroppa

, Babl

F.E.

, Rosenfeld

J.V.

, and Anderson

(2012). Timing of traumatic brain injury in childhood and intellectual outcome. J. Pediatr. Psychol., 37, 745–754.

28.

Hagan

P.J.

(1967). Posttraumatic anosmia. Arch Otolaryngol. 85, 85–89.

29.

Jimenez

D.F.

, Sundrani

, and Barone

C.M.

(1997). Posttraumatic anosmia in craniofacial trauma. J. Craniomaxillofac. Trauma., 3, 8–15.

30.

Zasler

N.D.

, McNeny

, and Heywood

P.G.

(1992). Rehabilitative management of olfactory and gustatory dysfunction following brain injury. J. Head Trauma Rehabil., 7, 66–75.

31.

Drummond

, Douglas

, and Olver

(2007). Anosmia after traumatic brain injury: A clinical update. Brain Impair. 8, 31–40.

32.

Miwa

, Furukawa

, Tsukatani

, Costanzo

R.M.

, DiNardo

L.J.

, and Reiter

E.R.

(2001). Impact of olfactory impairment on quality of life and disability. Arch. Otolaryngol. Head Neck Surg., 127, 497–503.

33.

Santos

D.V.

, Reiter

E.R.

, DiNardo

L.J.

, and Costanzo

R.M.

(2004). Hazardous events associated with impaired olfactory function. Arch. Otolaryngol. Head Neck Surg., 130, 317–319.

34.

Menon

D.K.

, Schwab

, Wright

D.W.

, and Maas

A.I.

(2010). Position statement: definition of traumatic brain injury. Arch. Phys. Med. Rehabil., 91, 1637–1640.

35.

Pink

(2011). Socio-economic Indexes for Areas (SEIFA). Australian Bureau of Statistics: Canberra, Australia.

36.

Teasdale

and Jennett

(1974). Assessment of coma and impaired consciousness. A practical scale. Lancet, 2, 81–84.

37.

Wechsler

(1999). Wechsler Abbreviated Scale of Intelligence. The Psychological Corporation: San Antonio, TX.

38.

Wechsler

(2005). Wechsler Intelligence Scale for Children—4th Edition Australian Standardisation Edition. Harcourt Assessment: Sydney, Australia.

39.

Roberts

, Howard

, Spittle

A.J.

, Brown

N.C.

, Anderson

P.J.

, and Doyle

L.W.

(2008). Rates of early intervention services in very preterm children with developmental disabilities at age 2 years. J. Paediatr. Child Health, 44, 276–280.

40.

Doty

R.L.

(1995). The smell identification test administration manual. Sensonics, Inc.: Haddon Heghts, NJ.

41.

Mackay-Sim

and Doty

R.L.

(2001). The University of Pennsylvania Smell Identification Test: Normative adjustment for Australian subjects. J. Otolaryngol. Soc. Aust., 4, 174–177.

42.

Volker

M.A.

(2006). Reporting effect size estimates in school psychology research. Psychol. Schools, 43, 653–672.