CONCUSSION IS INCREASINGLY being described as an epidemic in both the medical literature and popular press.1,2 Ontario is part of this trend as demonstrated by a 221% increase in workers' compensation claims between 2011 (955 claims) and 2016 (3067 claims).3 Evidence-based resources are available to guide best practice in the management of these cases4,5); however, implementation of coordinated care models is lacking and there continues to be large variation in the nature, quality, and timing of concussion services offered.6
The majority of adults who experience concussion recover within the first few weeks following injury, although 15% to 20% of patients continue to experience symptoms.5 Persistent concussion symptoms are defined as symptoms continuing beyond expected time frames (ie, >10–14 days in adults).5 In such cases, clinical assessment is recommended to identify specific primary and secondary pathologies that may be contributing to the symptoms. Assessment should include a comprehensive history, focused physical examination, and special tests where indicated.4,5 Early intervention including education, reassurance, and support for guided resumption of activities significantly reduces social morbidity and severity of postconcussion symptoms.7–10
Physician guidance in the return-to-work process may also have a significant impact on return-to-work outcomes. A systematic review synthesizing the best available evidence on return to work after concussion suggested most workers returned to work within 3 to 6 months and concussion was not a significant risk factor for long-term work disability.11. From that analysis, predictors of delayed return to work included lower level of education, nausea or vomiting on hospital admission, extracranial injuries, severe head/bodily pain after injury, and limited job independence and decision-making latitude.11 Specific to physician recommendations, in a randomized controlled trial in which the attending neurosurgeon recommended return to work either 3 or 7 days following the date of injury, the actual return-to-work time (median 7 days, mean 11.4 days) was considerably shorter than prior studies including a sample from another recent study in which the median return–to-work time was 16 days, with a mean time of 26.1 (subsample with no postconcussion syndrome at 12 months) or 146.4 days to return to work (subsample with mild postconcussion syndrome at 12 months).12 Both studies indicated 2 patterns of people with concussion that differ greatly—early return to work with full workload and less postconcussion syndrome symptoms, and later return to work with more postconcussion syndrome symptoms.13
In 2016, the Ontario Workplace Safety and Insurance Board (WSIB) implemented an evidence-based care model for concussion involving assessment and intervention for people with concussion who had not returned to work by 2 weeks following the date of injury. Consistent with the recognition that physicians have a key role in concussion diagnosis, management, and activity resumption planning, the program is delivered by physicians who received training in concussion care as part of their engagement in this program.14 The main aspects of the program are:
- detailed appropriate clinical assessment,
- patient education, and
- facilitation of early and safe return to work with input regarding the work situation from a return-to-work specialist.
Coordination and communication among health providers, insurers, and the workplace is considered best practice in disability management of people who have been injured or become ill at work.15 Accordingly, an additional program requirement is for telephone contact by the concussion program physician with the patient's primary care physician to discuss the findings and recommendations of the assessment.
Physician providers across the province with a variety of medical specialties including psychiatry, neurology, emergency medicine, and sports medicine were selected to deliver this service. The providers participated in targeted education delivered by recognized clinical experts in concussion from the fields of neurology, psychiatry, and occupational medicine. This specialized education was mandatory irrespective of experience or medical specialty and ensured a common foundation for program delivery.
The aim of this study was to evaluate the impact this concussion assessment program had on healthcare utilization and duration of time on loss of earnings benefits, a frequently utilized surrogate measure for recovery and return to work.
Study design and participants
Administrative data for the analysis were drawn from the WSIB of Ontario. The database includes demographics, injury, workplace type, return to work, and other information for all injured workers of Ontario. The cases were administratively identified by claim numbers. A quasiexperimental pre-/poststudy design with difference-in-difference (DID) modeling to approximate estimation of causal effects was used to evaluate the impact of introducing the new concussion assessment program. The cohorts of patients used for the analysis were derived from the preimplementation (injury date from January 1, 2014, to September 30, 2015) and postimplementation periods (injury date from March 1, 2016, to February 28, 2017). To enable outcome observation at 6 months postinjury, as well as healthcare utilization within 6 month postinjury, corresponding follow-up periods were used when building the pre- and postimplementation cohorts. Follow-up periods were the same (6 months) for pre- and postimplementation cohorts.
The case definition applied to identify study participants was based on nature of injury code and duration of lost time. Participants were identified by 1 of 3 specific natures of injury codes (intracranial injuries, unspecified, concussions, or intracranial injuries). The use of broader nature of injury codes was required to account for variability in coding practices for concussion. To ensure moderate and severe cases were not included in the analysis, cases requiring inpatient care were excluded. The case definition further narrowed the study population to those having at least 2 weeks of lost time as measured by loss-of-earnings benefits. The duration of lost time criterion was required because this was part of the program referral criterion, which was set to ensure patients with uncomplicated concussion who were going to improve within 2 weeks were not overmedicalized by receiving the intervention. Patients working at the time of injury were included in the analysis regardless of age (all age groups). The administrative data collected in the WSIB database do not include race or ethnicity information, so analysis by these variables was not possible.
The concussion assessment program was introduced to provide expert assessments by physicians with expertise in the assessment and management of concussions. The primary goals of the program were to provide patient education, to make appropriate treatment recommendations, and to facilitate better communication between the injured person's specialist and the primary healthcare provider, usually the family physician. Where cases required more in-depth diagnostic testing or where a multidisciplinary assessment was deemed required, appropriate triage was another integral aspect of the program. Prior to the introduction of this program, people with accepted claims for concussion were treated in the community and those demonstrating persistent symptoms over prolonged durations, generally greater than 3 months, were referred to a tertiary care center for further comprehensive multidisciplinary assessment. This program provided earlier assessment than the prior program with referral for assessment taking place when patients had not returned to work by 2 weeks following the date of injury.
Providers in the program are physicians who indicated they had a special interest in concussion care. Each physician provider also underwent additional training to ensure the care being provided was up to date, evidence based, and standardized across all providers. The training was based on the Ontario Neurotrauma Foundation (ONF) Guidelines, which provide specific detail on best practice in the assessment and management of concussion.4 The ONF is a nonprofit government-funded organization, which is a recognized leader in moving research to evidence-informed practices. The ONF Guideline development process, overseen by a governing committee, adheres to best practice in guideline development and involves a scientific committee, core and advisory working groups, and a consensus panel of experts consisting of clinicians, researchers, program and facility managers, and other stakeholders. The ONF Guidelines include patient resources and handouts. A quality management program ensured providers adhered to the ONF Guidelines.
The process flow of the concussion assessment program is depicted in Figure 1. Referral criteria to the program specify a focus on workers with greater than 2 weeks lost time following the date of injury with symptoms related to concussion being reported as one of the barriers to return to work. People referred to the program receive the following services: a minimum 45 minutes of direct contact with the assessing physician; a neurological assessment including clinical interview and objective evaluation; evaluation of risk factors for prolonged duration of symptoms and time off work; review of current medications; education regarding the typical course of recovery, symptoms, and appropriate physical/cognitive rest to promote safe return to activity and return to work. The physician then works with the patient to formulate a return-to-work plan, taking into account cognitive load/exertion requirements and the return-to-work specialist report that contains details on the usual nature of work and opportunities for accommodation. The recommendations, including symptom management (eg, fatigue, headache, and sleep hygiene), are then discussed with the patient and the primary healthcare provider. Up to 2 follow-up visits are included in the program to support the treatment process and provide education and guidance on the safe and healthy return to activity and work.
Telephone contact with the injured person's treating healthcare provider was the third component of the program. The purpose of the phone call was to enable to the physician provider to discuss the findings and recommendations of the assessment to the primary care provider, most often the family physician, to ensure continuity of care. A quality management program was used to assess and monitor implementation of the new program. The quality management program is performed by a multidisciplinary team that audits clinical reports, performs claim file reviews, and meets with providers to discuss the audit results using a Quality Management framework. The results of the Quality Management Program for the concussion assessment program indicated that providers were adhering to the requirements of the new program and the worker's primary health provider was being successfully contacted by telephone.
Subjects in observational studies are not randomly assigned to intervention and control groups. Confounding can occur if covariates are related to both the intervention assignment and the outcome. Consequently, there can be systematic differences between subjects in the intervention and control groups. In the presence of confounding, statistical approaches, such as propensity score matching, are required to account for potential confounding when estimating the effect of an intervention. Propensity score matching was used to match concussion cases in the postimplementation period (period 1) to concussion cases in the preimplementation period (period 0). A second round of propensity score matching was then done to match concussion cases from the first step to nonconcussion cases by corresponding period. This second step involved DID methodology, which is a research design used for estimating causal effects.16 DID estimation is a multilevel risk adjustment that controls for organizational duration shifts. This is necessary because organizational operational strategies (internal practices for claim management) and other external factors may have differed in the pre- to postimplementation periods and thus could bias the observed program results. By removing those effects, causal inference to pinpoint the true program effect on measured outcomes could be approximated. Thus, matching cases in the pre- and postimplementation periods controlled for potential differences in worker populations, while additional matching to nonconcussion cases controlled for potential external factors that might have influenced the program outcomes.
Propensity score-matched samples were risk-adjusted for established factors that predict work disability duration,17 specifically controlling for age, gender, injury mix (nature of injury, part of body risk groups), psychological services, healthcare services, first digit of source of injury grouping, industry mix (represents baseline risk in different industries such as construction, healthcare, etc, at a subindustry level), employer insurance type (collective or individual employer cost allocation), size of employer's organization, prior claims, wage, foreign language, and first digit of National Occupational Classification.
Categorical variables with too many levels to include (eg, industry mix with claim rate group) were binned into fewer levels to reduce the degrees of freedom, which hinders the estimation and to address the problem of some levels having too small a sample size. The risk of the outcome was calculated (proportion of cases on loss-of-earnings benefits at 6 months) in each rate group based on the whole study population. The same method as above was used for grouping nature of injury and part of body codes into injury mix groups. In the second level of propensity scoring, cases were exactly matched based on the injury mix group, which enabled the analysis to match more risky (in terms of cost and duration) than average concussion cases to other similar risk-level nonconcussion cases. Usual propensity score derivation using logistic regression on the remaining baseline covariates followed.
Incremental healthcare, loss-of-earnings costs, and duration rate outcomes were estimated on propensity score-matched samples. Loss-of-earnings and healthcare costs within 6 months from injury date were derived from the WSIB financial system, and costs were indexation/inflation adjusted. We also estimated incremental costs percent change and 3- and 6-month duration rate outcomes using DID estimation. Due to the complex nature of the analyses (multiple steps in modeling), we estimated DID risk difference is 95% confidence intervals (CIs) using bootstrapping.
Healthcare costs by service type (eg, assessment, diagnostic, and prescription drug) were compared on propensity score-matched samples to better understand the drivers of the observed outcomes. Healthcare spend categories were derived using healthcare billing codes.
As a potential explanatory factor for differences in observed outcomes due to the program implementation, timeliness of first concussion-specific healthcare assessment visit, as well as number of concussion treatment visits, was examined. Days from injury date to the first concussion healthcare assessment visit were calculated, and the t-test analysis was performed on propensity score-matched samples comparing pre- and postimplementation periods for concussion cases.
This research was based on an intervention introduced by the WSIB as a health service for people with work-related injuries in Ontario. The data analyzed were collected as part of standard administrative practice by the WSIB. Only deidentified administrative data were used for this analysis.
Characteristics for concussion patient cohorts pre- and postimplementation are described in Table 1. Similar baseline characteristics are observed and were confirmed by derived propensity scores.
Cost and duration analysis on propensity score-matched samples with DID estimation showed that the program resulted in a −14.4% (95% CI, −28.7% to −0.8%) decrease in healthcare costs and a marginal but nonsignificant increase in duration on loss-of-earnings benefit costs of 2.4% (−9.5% to 14.3%) (see Table 2). The bootstrap distribution for the total combined loss-of-earnings and healthcare costs DID estimate was −4.6% (95% CI, −16.1% to 7.2%).
Results of the analysis of healthcare costs are presented in Table 3. The greatest decrease occurred in healthcare assessment costs (−27.9%). This decrease largely reflects the transition from a system involving fewer but more expensive multidisciplinary assessments at a later date to a system where a greater number of lower cost assessments took place at an earlier date. Significant decreases were also observed in healthcare costs related to diagnostics (−25.7%), primarily imaging, and prescription drugs (−33.4%). For prescription drugs, while the percentage decrease was substantial, the dollar expenditure was small (average cost per claim decreased from $11.29 to $7.52), suggesting prescription drug use is not significant in this population.
The analysis of timeliness of the first concussion assessment program visit showed people in the program had specialized concussion assessment at an earlier date than people with concussion in the year prior to program implementation (mean 49.5 days vs 70.1 days, P < .0001) and required fewer healthcare visits (mean 15.0 vs 15.9, nonsignificant).
The principal finding of this study was that a care model for people with work-related concussion involving an evidence-based assessment by a single physician focused on patient education resulted in significantly decreased healthcare utilization as measured by average healthcare expenditure per claim and demonstrated similar effectiveness, based on time to return to work, compared with the prior usual care model in which only patients with persistent symptoms for many months were referred for time-consuming hospital-based multidisciplinary assessments. The results suggest that evidence-based assessment programs for adults with concussion that include the primary care physician where possible can help reduce unnecessary additional health interventions while maintaining similar recovery outcomes.
The findings of this study are in keeping with prior research demonstrating the effectiveness of early education for concussion patients8,18 and brief simple interventions involving assessment and education given soon after concussion being as helpful as more intensive treatments for most concussion patients.9,10 Paniak et al10 conducted a randomized controlled trial in which participants were randomly assigned to either a single session treatment within 3 weeks of injury versus a single session and multidisciplinary assessment. The single-session intervention aimed to legitimize participants' experience, and provide reassurance, and education on common complaints following concussion and suggestions on how to cope with common symptoms. The multidisciplinary assessment group received the single-session intervention along with neuropsychological and personality assessments and feedback, a consultation with a physical therapist, and further treatment-as-needed for concussion complaints. That study showed the brief educational intervention given soon after concussion appeared to be as helpful as more intensive treatment for most patients measured at 12 months postinjury.9 Based on the results of the current study, the findings of Paniak et al seem to be generalizable to workers with concussion and the result of this study further validate this conclusion using the metric time away from work as a proxy for recovery.
The study results showed healthcare costs per claim decreased on average 14.4% after introduction of the new assessment model. The most significant decrease in healthcare costs was for assessment costs. This largely reflects the fact that the prior program involved usual care in the community, not necessarily involving early education and appropriate triage, with patients with persistent symptoms then requiring referral for comprehensive multidisciplinary assessment. The findings therefore suggest the new program model, which provided earlier expert physician assessment with education, was less costly, perhaps because the assessment and treatment interventions were tailored to each patient's specific needs at an earlier stage in their recovery. Significant decreases were also observed in healthcare costs related to diagnostics. The finding that recovery duration, as measured by lost time from work, was similar pre/postimplementation suggests that the additional investigations and prescription medications in the prior model were either unnecessary or that the prolonged duration of time away from work led to worse symptoms or secondary conditions requiring additional investigation and treatment.
Healthcare utilization is an important outcome not only because it may be a proxy for symptom resolution—those with persistent symptoms would reasonably be expected to utilize healthcare resources more intensively—but also because unnecessary medical services can contribute to symptom severity and duration. Termed iatrogenic disability, this process is well recognized in the literature as an important and potentially modifiable risk factor for the development of postconcussion syndrome.19,20 Conversely, programs that facilitate appropriate use of healthcare resources, including diagnostic services and treatments, can thereby promote recovery and prevent iatrogenic disability. For this reason, the finding of reduced healthcare costs in this study is important, and particularly relevant in the current medical care landscape, which is increasingly focused on promotion of engaged conversations between clinicians and patients regarding unnecessary tests and treatments to facilitate smart and effective healthcare choices.21
Methodological strengths of this study are the large sample size and carefully controlled analysis using propensity score-matched samples with difference in difference estimation. The DID methods are particularly important for studies analyzing outcomes that could be influenced by organizational factors beyond the program intervention itself. By controlling for trends occurring at the same time as the concussion program implementation, the findings can be considered to have greater validity and that the results attributable only to the new program intervention.
A potential limitation of this study is that, given the case management process, only about a third of all people with a work-related concussion were treated through this program; however, this would most likely underestimate the program effect and bias the results toward the null since all everyone with a work-related concussion was included in the data analysis although not all received the intervention. This is similar to an intention to treat analysis. Another study limitation was the inability to specifically assess concussion symptoms, given available information in the claim file data. Instead, duration on benefits was used as an indicator of recovery. An additional limitation, common to studies using workers' compensation records, is that the available data are collected by the compensation organization and does not include costs that may be paid by other insurers or the public healthcare system.
Future research should focus on the qualitative aspects of the program capturing the perceived impact of early education from both the patient and treating healthcare provider's perspective, and the benefit of contact between the concussion program physician and the primary care physician.
Evidence indicates that early intervention following concussion, including education, reassurance, and support for guided resumption of activities, significantly reduces social morbidity and severity of postconcussion symptoms.7–9 The findings of this study align with prior reports in the literature demonstrating the benefit of early assessment and education for patients with concussion and extend these findings to a cohort of workers with occupationally related concussions. The findings of this study may help inform concussion care in other workers' compensation systems and may also be generalizable to other public and private insurer settings where adults with concussions are assessed and treated.
1. Buck PW. Mild traumatic brain injury: a silent epidemic in our practices. Health Soc Work. 2011;36(4):299–302.
2. Hall V. Tackling Canada's concussion
epidemic head on: “Patients are really still in the dark.” The National Post. April 22, 2016.
3. Workplace Safety and Insurance Board. By the Numbers: 2015 Workplace Safety and Insurance Board Statistical Report. Schedule 1 and Schedule 2. http://www.wsibstatistics.ca/
Accessed 13 May 2018.
4. Marshall S, Bayley M, McCullagh S, et al Clinical practice guidelines for mild traumatic brain injury and persistent symptoms. In: Guidelines for Concussion
& Persistent Symptoms. 2nd ed. Toronto, Canada: Ontario Neurotrauma Foundation.
5. McCrory P, Meeuwisse W, Dvořák J, et al Consensus statement on concussion
in sport—the 5th
international conference on concussion
in sport held in Berlin, October 2016. Br J Sports Med. 2017;51(11):838–847.
7. Wade DT, King NS, Wenden FJ, Crawford S, Caldwell FE. Routine follow up after head injury: a second randomized controlled trial. J Neurol Neurosurg Psychiatry. 1998;65(2):177–183.
8. Ponsford J, Willmott C, Rothwell A, et al Impact of early intervention on outcome following mild head injury in adults. J Neurol Neurosurg Psychiatry. 2002;73(3):330–332.
9. Paniak C, Toller-Lobe G, Durand A, Nagy J. A randomized trial of two treatments for mild traumatic brain injury. Brain Inj. 1998;12(12):1011–1023.
10. Paniak C, Toller-Lobe G, Reynolds S, Melnyk A, Nagy J. A randomized trial of two treatments for mild traumatic brain injury: 1-year follow-up. Brain Inj. 2000;14(3):219–226.
11. Cancelliere C, Kristman VL, Cassidy JD, et al Systematic review of return to work
after mild traumatic brain injury: results of the International Collaboration on Mild Traumatic Brain Injury Prognosis. Arch Phys Med Rehabil. 2014;95(3, suppl):S201–S209.
12. Losoi H, Silverberg N, Wäljas M, et al Recovery from mild traumatic brain injury in previously healthy adults. J Neurotrauma. 2016;33(8):766–776.
13. Studerus-Germann AM, Engel DC, Stienen MN, von Ow D, Hildebrandt G, Gautschi OP. Three versus seven days to return to work
after mild traumatic brain injury: a randomised parallel-group trial with neuropsychological assessment. Int J Neurosci. 2017;127(10):900–908.
14. Tator CH. Concussions and their consequences: current diagnosis, management and prevention. CMAJ. 2013;185(11):975–979.
15. Pransky G, Shaw W, Franche RL, Clarke A. Disability prevention and communication among workers, physicians, employers, and insurers—current models and opportunities for improvement. Disabil Rehabil. 2004;26(11):625–634.
16. Lechner M. The estimation of causal effects by difference-in-difference methods. Found Trends Econometrics. 2010;4(3):165–224.
17. Cancelliere C, Donovan J, Stochkendahl MJ, et al Factors affecting return to work
after injury or illness: best evidence synthesis of systematic reviews. Chiropr Man Therap. 2016;24(1):32.
18. Alves W, Macciocchi SN, Barth JT. Postconcussive symptoms after uncomplicated mild head injury. J Head Trauma Rehabilitat. 1993;8(3):48–59.
19. Rees PM. Contemporary issues in mild traumatic brain injury. Arch Phys Med Rehabil. 2003;84(12):1885–1894.
20. Silverberg ND, Iverson GL. Etiology of the post-concussion
syndrome: physiogenesis and psychogenesis revisited. NeuroRehabilitation. 2011;29(4):317–329.