This study demonstrates that the application of a decision support tool in the ED based on the CCHR coincided with a 20% reduction in the proportion of RACF residents undergoing CTB after presenting to ED with a fall. A non-significant trend towards increased diagnostic yield was observed. There were no differences in neurosurgical management or other measurable patient outcomes between the pre- and post-intervention groups.
A wide variation has been reported in the diagnostic yield of CT imaging for an acute ICH in older adults following a fall, ranging from 1.6–11.6%.7,8 A systematic review by de Wit et al. identified a pooled incidence of 5.2% from 11 studies which specifically looked at patients aged 65 years or above who presented to the ED following a fall.9 Although our study specifically selected only patients from RACF (and could potentially represent a more frail cohort at risk for ICH than those reported elsewhere) our findings appear comparable, with yields of 3.7% in the pre-implementation period and 5.6% during the post implementation period, with the latter possibly being attributable to the more targeted scanning strategy as a result of the intervention. A low baseline yield within this sub-population of ED attendees could, however, also be affected by a potentially greater tendency for RACF residents to be referred to hospital following a fall compared to patients who self-present from the community. This is an important consideration, particularly in our local Australian context, given the Aged Care Royal Commission’s emphasis on falls prevention10 and a recent public report from the Coroner’s Court in South Australia that encourages CTB scans in any RACF residents with a possible head strike who are on anticoagulation therapy11. It is likely that this directive, questioned in a recent publication by Green et al.12, has resulted in a reduced threshold for care providers in RACF to seek hospital assessment when one of their residents falls. Further research would be required to determine how these factors interrelate.
In our study, no patients in either group underwent neurosurgical intervention. These findings are consistent with similar studies where surgical intervention ranged from 0–5%.1,12, 13 Comparison of rates of medical management for patients presenting with a fall is more difficult, in part due to the heterogenous nature of possible interventions which can also depend upon other existing comorbidities. For example, anticoagulation cessation can be considered a medical intervention in this context but is of course only a potential intervention in those already taking such therapy. We identified a single study by Sartin et al.14 that described 29.4% of patients with a confirmed “acute injury” on imaging receiving medical intervention in the form of medication changes; however, this included patients aged 55 years and older with a GCS of 15 on presentation. Fifty percent of patients with a diagnosed ICH in our study received medical intervention, reflecting a much higher rate of intervention that is likely due to our focus on an older, frailer cohort. Cessation of anticoagulation or antiplatelet use was the most common type of medical management seen in our study, yet these medications are less likely to be present in a younger cohort.
A similar recently published quality improvement initiative was performed in Toronto, Canada by Masood at al.15 During initial root cause analysis, the authors also identified that the CCHR was applied to many patients with “minimal head injury”, rather than those with only “minor head injury”. Following the introduction of a comprehensive education campaign and a “modified Canadian CT Head Rule checklist” to overcome this issue, they observed a reduction in CTB rates in adults of all ages by 13.9% at 3 months and 8% at 16 months, with no increase in adverse outcomes. A point of difference in our study was our much narrower focus on older patients from RACF. The greater reduction in CTB scanning rates associated with our tool may be related to the natural tendency of clinicians to err on the side of caution and investigate more frequently with head imaging in this particular cohort, especially in light of the above-mentioned coroner recommendation. As such, CTB rationalisation with tool implementation in this group has potential for a relatively larger reduction. Our consideration in a change from baseline GCS rather than using a threshold of any GCS <15 in conjunction with incorporation of goals of care discussions may also have contributed to the greater reduction in CTB scanning rates that we observed.
Despite the well-established nature of the CCHR, there has been little exploration of its use explicitly in the older person. Many clinicians continue to interpret the rule as suggesting that all patients aged >65 require a CTB after a head strike without considering whether it was appropriate to apply the rule in the first place. In a retrospective cohort study, Fournier at al.16 demonstrated that increasing the high-risk age criteria from 65 years to 75 years was associated with an increase in the sensitivity of the CCHR. This could translate into a 25% reduction of CT scans in those aged 65–74 years without compromising diagnosis of “clinically important brain injuries”. Although their mechanism for reducing CTB in the elderly was different to ours, this further demonstrates that recalibrating clinicians to apply CCHR in a more select cohort, either by increasing the age definition of “high risk” or by tightening applicability criteria within those presenting from RACF following a fall appears to have potential to safely reduce CTB utilisation.
Whilst medical investigations can provide useful prognostic information and assistance with treatment decision making, 17,18 the indiscriminate use of low-yield tests has numerous associated healthcare and patient implications. A simple calculation based on an Australian Medicare rebate amount of $230.40 for each CTB 19 and the 20% observed reduction in scanning rate that we observed would equate to a saving of $40,550 per annum in our department, although it is likely that the real figure is much higher when the effect on ED patient flow and follow-up of incidental findings are also considered. 20,21 Primum non nocere (“first, do no harm”); also important is the patient experience and consideration of avoiding unnecessary or burdensome interventions in a vulnerable cohort. Appropriately limiting investigations has potential to improve quality of care provided via such means as avoiding chemical or physical restraints, reduction in ED length of stay, reduction of incidental findings on imaging and potential prevention of cascading investigations or hospital admission that may not be indicated and of little clinical benefit. These potential positive outcomes could also ultimately circumvent exposure to numerous risk factors that may lead to development of delirium in this susceptible cohort and subsequently prevent the significant associated morbidity.
This study followed a retrospective audit design that relied on interpretation of existing electronic documentation and did not allow for detailed matching of the two groups. Accuracy and availability of clinician documentation may influence study results and limitations related to missing data and incomplete adherence to the pathway during the post intervention period need to be acknowledged. In addition, our simple comparison of means between pre- and post-intervention groups does not account for possible pre-existing trends, changes in clinicians’ approaches to falls management or changes in clinicians’ CTB ordering practices. Nevertheless, our use of symmetrical time periods (12 months apart at the same institution) reduces the potential impact of such confounders.
Follow-up of patients with regard to missed pathology or delayed diagnosis was limited to Eastern Health hospitals or received correspondence and so may not capture all individuals who subsequently presented to other health networks or who may have subsequently died in the community. However, it is unlikely that a patient from an RACF would represent to a different health service in such a circumstance, or that Eastern Health would not be notified of such an incident.
Finally, our use of data from a single centre may limit the external applicability of these findings.