Abstract

Background: Injury represents 260,000 hospitalisations and $27 billion in healthcare costs each year in Canada. Evidence suggests that there is significant variation in the prevalence of hospital admissions among ED presentations between countries and providers but we lack data specific to injury admissions. We aimed to estimate the prevalence of potentially low-value injury admissions following injury in a Canadian provincial trauma system, identify diagnostic groups contributing most to low-value admissions and assess inter-hospital variation.
Methods: We conducted a retrospective multicenter cohort study based on all injury admissions in the Québec trauma system (2013-2018). Using literature and expert consultation, we developed criteria to identify potentially low-value injury admissions. We used a multilevel logistic regression model to evaluate inter-hospital variation in the prevalence of low-value injury admissions with intraclass correlation coefficients (ICC). We stratified our analyses by age (1-15; 16-64; 65-74; 75+ years).
Results: The prevalence of low-value injury admissions was 16% (n=19,163) among all patients, 26% (2136) in children, 11% (4695) in young adults and 19% (12,345) in older adults. Diagnostic groups contributing most to low-value admissions were mild traumatic brain injury in children (48% of low-value pediatric injury admissions; n=922), superficial injuries (14%, n=660) or minor spinal injuries (14%, n=634) in adults aged 16-64, and superficial injuries in adults aged 65+ (22%, n=2771). We observed strong inter-hospital variation in the prevalence of low-value injury admissions (ICC=37%).
Conclusion: One out of six hospital admissions following injury may be of low-value. Children with mild traumatic brain injury and adults with superficial injuries could be good targets for future research efforts seeking to reduce health care services overuse. Inter-hospital variation indicates there may be an opportunity to reduce low-value injury admissions with appropriate interventions targeting modifications in care processes.

What is known

Low-value care leads to inefficient use of resources and less favorable patient outcomes.
The literature suggests that unnecessary hospitalisations may be prevalent but we lack data on this problem for injury admissions.

What this study adds

We found high variability in the prevalence of potentially low-value injury admissions between hospitals in a trauma system, suggesting the presence of low-value care.
Results suggest that children with mild traumatic brain injury and adults with superficial injuries or minor spine injuries may constitute targets for future research aiming to assess benefits & harms and identify solutions to reduce unnecessary injury admissions.

Introduction

Injury represents 3.5 million visits to emergency rooms and 260,000 hospitalisations per year in Canada1. Furthermore, injury care costs Canadians $27 billion per year, more than heart and stroke disease combined1. Evidence suggests that low-value care, defined as the provision of medical services for which the potential for harm exceeds the potential for benefit , consumes around 30% of healthcare resources2, 3. Research has shown that there is considerable variation in the prevalence of hospital admissions among ED presentations between countries and providers4-6, suggesting the presence of low-value hospital admissions2. These admissions contribute to hospital overcrowding, increased costs, medical errors, mortality, morbidity and functional decline7-9. A recent scoping review and expert consultation survey identified hospital admission in specific injury populations as a potentially low-value practice10. However, we lack knowledge on the frequency of low-value injury admissions and on inter-provider variations.
The objectives of this study were to estimate the prevalence of potentially low-value injury admissions following injury, identify diagnostic groups that most contributed to low-value admissions, and assess inter-hospital variation.

Methods

Study Design & Population

Our multicenter, retrospective cohort study was based on the 59 designated trauma centers of the Québec trauma care continuum, an integrated trauma system covering a population of 8.5 million in a geographic area of approximately 1.7 million km2 that provides care for approximately 20,000 injured patients each year11. The system comprises 3 level I adults trauma centers and 2 pediatric trauma centers providing highly specialized care in metropolitan areas (Montréal and Québec city), 5 level II centers offering similar services in smaller urban areas (i.e. lower volume), 22 level III and 27 level IV centers generally in rural areas that stabilize major trauma cases then transfer them to high level care. We included all patients admitted with a primary diagnosis of injury from April 1, 2013 to March 31, 2018 to any trauma center in the system. Patients were followed from hospital admission to discharge.

Study Data

We extracted data from the Québec trauma registry, based on mandatory data collection for all hospital admissions for which injury was the primary motivation in all provincial trauma centers. Data is collected prospectively from patient charts in each center by trained medical coders. The registry is centralized and managed by the Québec Ministry of Health and Social Services. Data quality mechanisms include standardized training for medical coders, periodic validation of the data to correct incoherence and continuous facilitated information exchange between data coders, trauma coordinators and clinical experts12. Random re-abstraction of patient chart data suggests the trauma registry has 98% accuracy on patient demographics, vital status, injury codes, interventions and discharge destinations (data not published).

Selection of Criteria to Identify Low-Value Injury Admissions

We used recommendations for the design and content validation of health instruments to develop a tool to identify potentially low-value injury admissions from the Québec trauma registry13. First, a trauma registry specialist, a trauma researcher and a trauma clinical expert used published literature14-22 and their knowledge on trauma systems to select and define criteria consistent with patients not requiring hospitalisation (Supplemental Digital Content 1). For example, minor injury defined as a maximum Abbreviated Injury Scale (AIS) score ≤ 2 and the absence of interventions requiring hospitalisation, such as a surgery. Second, 8 experts (2 emergency physicians, 2 critical care physicians, a trauma surgeon, a general practitioner, a trauma care coordinator, and a trauma system manager) were individually consulted to validate the relevance and the definition of each criteria and were asked if any other criteria should be added. Based on these consultations, the list of interventions that did not require hospitalization was modified. Third, criteria were applied to trauma registry data and an age-stratified random sample of patient records were extracted and revised by the same experts to refine criteria. This last step was repeated three times until no further modifications were suggested. Criteria which were suggested but not retained in the final version were considered in sensitivity analyses (see section below). The final tool included 6 criteria (supplemental digital content 1).

Statistical Analysis

To identify diagnostic groups most contributing to low-value injury admissions, our steering committee grouped ICD primary diagnosis codes (supplemental digital content 2) and we stratified them by age group (1-15, 16-64, 65-74, 75+). We then described the following aspects of resource use among low-value admissions: i) therapeutic interventions in the ED and following admission (grouped as medication, reduction/immobilization, skin repair, other; supplemental digital content 1), ii) diagnostic imaging in the ED and following admission (none, X-ray, computed tomography by body region, magnetic resonance imaging, angiography/arteriography), iii) length of hospital stay (days from admission to discharge).
To evaluate inter-hospital variation in the prevalence of low-value injury admissions, we used a multilevel logistic regression model to generate intraclass correlation coefficients (ICC). Inter-hospital variation was considered to be weak, moderate and strong for ICC under 5%, between 5% and 20%, and over 20%, respectively23. We adjusted for patient case mix using age (1-4, 5-12, 13-15, 16-54, 55-64, 65-74, 75-84, 85+), sex, anatomic injury severity (New Injury Severity Score; 1-3, 4, 5-6, 8-11, ≥12)24, the Glasgow Coma Scale (GCS) score on arrival in the ED (≤13, 14, 15), injury mechanism (motor vehicle collision, fall from height, fall from own height, knife/firearm, other), number of injuries (1, 2, 3+), primary injury diagnosis and number of comorbidities (0, 1, 2, 3+). Analyses were conducted for the whole sample and according to age (1-15, 16-64, 65-74, 75+). We handled missing data on the GCS, often not evaluated in patients with minor, extracranial injury, using multiple imputation according to recent recommendations25.

Sensitivity analyses

We conducted sensitivity analyses for criteria that were mentioned by members of the steering committee during the development of criteria to identify low-value injury admissions but did not gain consensus. Thus, we successively removing patients in whom hospital admission may have been justified from the numerator: i) patients with a GCS of 13 or 14 on arrival, ii) patients who developed complications during their stay, iii) patients discharged home with services or to long-term care/nursing home, and iv) patients transferred-in from another hospital. For each sensitivity analysis, we compared the prevalence of low-value admissions and inter-hospital variation (ICC) to values from the main analysis.
All analyses were performed using Statistical Analysis Software (SAS Institute, Cary, NC, version 9.4). The study was approved by the research ethics committee of the CHU de Québec – Université Laval.

Results

Our study sample included 118,032 injury admissions, of which 19,163 (16%) were considered to be potentially low-value (Figure 1, Supplemental Digital Content 1). This proportion was 26%, 11%, 14% and 20% in children, young adults, adults 65 to 74 years of age and adults over 74 years of age, respectively. Overall, 64% of patients were aged 65 years or older, 39% were male, 75% were injured in a fall and 53% had isolated injuries (Supplemental Digital Content 3). Children had less severe injuries (59% with maximum AIS=1 compared to 40% in adults), more head injuries (60% compared to 20% in adults) and were more often transferred-in from another hospital (16% compared to 6% in adults).
The diagnostic groups contributing most to low-value injury admissions were mild traumatic brain injury in children (n=922; 48% of low-value pediatric injury admissions; Figure 2), superficial injuries (660; 14%) or minor spine injuries (634; 14%) in adults aged 16-64, and superficial injuries in adults aged 65+ (2771; 22%).
Out of the 19,163 patients with a potentially low-value injury admission, 5582 (29%) had a therapeutic intervention considered not to require hospital admission (Supplemental Digital Content 1; Figure 3) in the emergency department and 3471 (18%) patients had such an intervention after admission. Administration of medication was the most common intervention in every patient group, except patients with an extremity fracture, in whom reductions and immobilizations were more frequent. In addition, physiotherapy and social services consultations were used in 51% and 29% of patients, respectively. These numbers went up to 70% and 41% in patients aged 75 +.
Overall, 18,265 (95%) patients had at least one diagnostic imaging procedure (Table 2), of which over 90% were conducted in the ED. X-ray was used in over 90% of patients in every age group except children. Head CT was the predominant advanced imaging modality in every age group (38% of patients). Median length of stay varied from 1 day in children to 13 days in adults aged 75+ and varied across diagnostic groups (Table 1). Our population of potentially low-value injury admissions in the Québec trauma system accrued on average 36,642 hospital days per year. Among the 19,163 potentially low-value injury admissions, there were over 2,500 complications including 520 cases of delirium, 598 urinary tract infections, 50 stage II-IV decubitus ulcers and 98 hospital-acquired pneumonias.
We observed strong inter-hospital variations in the prevalence of potentially low-value injury admissions, with an ICC of 37% (Figure 4). When we stratified by age, inter-hospital variations remained strong for patients aged 74 + and were moderate for every other age group (Supplemental Digital Content 4). Even within designation levels, variation remained high with prevalences varying between 10% and 23% for level I and II trauma centers. However, highly specialized (level I/II) centers had lower average prevalences than smaller referral (level III/IV) centers.

Sensitivity analyses

When patients with a GCS of 13 or 14 or patients who developed an in-hospital complication were removed, prevalence and ICCs all changed by less than 10% (Supplemental Digital Content 5). The removal of patients transferred-in had an impact on prevalence in children (22% compared to 26%) as did the removal of patients who were discharged home with services or to long-term care or a nursing home (6% compared to 16%). However, all ICC remained over the threshold for strong inter-hospital variation.

Discussion

In this multicenter retrospective cohort study, 16% of all injury admissions in a Canadian trauma system were potentially low-value. Predominant diagnostic groups included mild traumatic brain injury in children, superficial injuries or minor spine injuries in adults aged 16-64, and superficial injuries in adults aged 75+. Prevalences varied from 0% to 34% between hospitals.
The important contribution of children with mild traumatic brain injury to potentially low-value injury admissions is consistent with other studies on children admitted for observation with minor head injury despite extremely low rates of missed injuries16-18, 26-31. Contributory factors documented in the literature include the lack of ability of children to communicate symptoms, perceived seriousness of the consequences of missed injuries, physician discomfort with prognostication in non-pediatric centers, pressure from parents, concerns about non-accidental injury, and observation to avoid exposure to ionizing radiation from CT32-36. However, 368 (40%) of children admitted for mild traumatic brain injury had a CT in the ED. Children represented only 2% (737 days) of hospital days per year accrued for potentially low-value injury admissions in our study population.
Adults aged 16-64 had the lowest prevalence of low-value admissions. They also had the most varied primary diagnoses, with none representing more than 14% of admissions. They had the highest frequency of motor vehicle collisions, which is widely used as a triage criteria because of its potential for serious injuries, but often leads to overtriage37. Intoxication could also explain potentially low-value admissions in these patients, since substance abuse is most prevalent in this group and intoxicated patients pose a challenge in terms of diagnostic evaluation, notably for traumatic brain injuries and spinal injuries38, 39. This group represented 11.4% (4,181 days) of hospital days per year accrued for potentially low-value injury admissions in our study population.
Patients aged 65 years and over comprised 64% of the low-value admissions in our sample. Factors contributing to admission in this population may be frailty, low mobility, lack of social support, cognitive conditions or comorbidities40-42. They may be admitted while social service evaluations are conducted and/or alternative care is secured. Research has shown that these patients are particularly complex to care for since many have important comorbidities, difficulties with mobility or communication, or develop delirium, all of which contribute to longer lengths of stay43. They are also more susceptible to hospital-related adverse events such as functional decline or nosocomial infections, which often lead to additional hospital days44, 45. Delayed discharge due to problems accessing post-acute care is also a problem in older adults46, 47. This age group represented 86.6% (31,724 days) of hospital days per year accrued for potentially low-value injury admissions in our study population.
We observed strong inter-hospital variation in the prevalence of potentially low-value injury admissions, but more specialized trauma centers, in general, had a lower prevalence than level III and IV centers. A potential explanation is that physicians working in centers with a higher volume of injury presentations have a better understanding of whether or not patients require hospital admission. This may also be due to bed availability, which could be less of an issue in rural hospitals due to low patient volume48.
Most patients did not receive any therapeutic intervention while others received interventions that can be administered in the ED (e.g. closed reductions, immobilisations or administrations of analgesics or antithrombotics). However, the majority of patients received a radiographic evaluation and/or a CT scan, over 90% of which were done before admission. This is consistent with estimates of overuse of imaging, notably for minor head injury34, 49.

Strengths and limitations

We had access to high-quality, current clinical data which, contrary to hospital discharge data, has detailed information on physiological parameters and anatomical injury severity. Furthermore, unlike many trauma registries that condition inclusion on injury severity, the Québec trauma registry includes all hospital admissions for injury in any designated trauma center. However, our study did not include patients admitted to non-designated hospitals, which are estimated to represent 25% of all injury admissions. Thus, we may have underestimated the prevalence of low-value injury admissions because non-designated hospitals could be less comfortable with discharging patients from the ED, even for minor injuries. Due to the lack of a validated definition for low-value injury admissions, we used broad filters based on literature and expert consultations and conducted extensive sensitivity analyses. Therefore, despite efforts to develop a clinically meaningful definition (including revision of a random stratified sample of patients included and excluded after each iteration of the algorithm), cases can only be interpreted as potentiallylow-value. We did not exclude patients based on comorbidities as we had no information on their severity and most patients over 65 had at least one comorbidity, which may be well controlled and unrelated to admission. Considering the retrospective nature of our data, our study is subject to potential measurement error and missing data (50% for the GCS). There may be measurement error among variables used to adjust inter-hospital comparisons (e.g. under-reporting of comorbidities). We may also have omitted potential confounders such as patient mobility, pain, socio-economic status and social support, because they were not available in the registry. This could lead to an overestimation of inter-hospital variation. In addition, the registry does not contain information on patients who were returned home from the ED so we could not compare patients admitted to those who were not. Finally, there is also a possibility that hospitals with a low prevalence of low-value injury admissions are undertriaging patients in which hospital admission was required. However, we consider this unlikely as highly specialized, high-volume centers in which undertriage is less likely, tended to have a lower prevalence than lower-level centers in our sample.

Impact and future research

This study sets the foundation for further research targeting improved value in injury care. This research should aim to determine whether low-value injury admissions could be reduced by interventions optimizing care in the ED or in the community and should consider healthcare professionals, patient and family perspectives. First, we will need to assess whether low-value admissions are potentially avoidable using case revision methodology. We would then need to identify modifiable contributing factors, for example discomfort with clinical examination and management of children with traumatic brain injury in non-pediatric centers and difficulty obtaining physiotherapy evaluation and social services consultations in the ED to determine if older adults can be safely discharged with available community services. Thereafter, interventions such as Audit & Feedback with a positive deviance approach and consensus recommendations could be used to modify care processes (e.g. inter-hospital transfer, virtual consultations, screening tools, community services). Older adults are of particular interest. On one hand, their complexity and frailty makes it harder to adequately modify practices to reduce care. On the other hand, they are a group in which such modifications could yield considerable gains both in terms of resources saved and improved outcomes since they have a high risk of functional decline and other hospital-related adverse events44, 45.

Conclusion

In the inclusive trauma system of the province of Québec, we estimated that one out of six hospital admissions following injury may be low-value. This represented on average 36,642 hospital days per year in the 59 designated trauma centers in the province of Québec alone. These resources may be better invested in ambulatory care. Children with mild traumatic brain injury, young adults with superficial injuries or minor spine injuries, and older adults with superficial injuries may be good targets for future research efforts seeking to reduce overuse. Strong variations between providers suggests that appropriate interventions have the potential to reduce low-value injury admissions following injury in Canada and thus improve patient outcomes and free up resources. We expect our results to be generalizable to other Canadian provinces. Future research should aim to identify modifiable determinants of low-value injury admissions and to develop potential solutions, which consider healthcare professional, patient and family perspectives.

References

1. Parachute Canada. The cost of injury in Canada (web) 2015 [Available from: https://parachute.ca/en/professional-resource/cost-of-injury-report/ (accessed 26 Mar 2020).
2. Brownlee S, Chalkidou K, Doust J, et al. Evidence for overuse of medical services around the world. The Lancet. 2017;390(10090):156-68.
3. Chassin MR, Galvin RW. The urgent need to improve health care quality. Institute of Medicine National Roundtable on Health Care Quality. JAMA. 1998;280(11):1000-5.
4. OECD. Geographic Variations in Health Care 2014 [Available from: https://www.oecd.org/els/health-systems/FOCUS-on-Geographic-Variations-in-Health-Care.pdf (accessed 11 feb 2019).
5. Kristiina Manderbacka MA, Markku Satokangas, Ilmo Keskimäki. Regional variation of avoidable hospitalisations in a universal health care system: a register-based cohort study from Finland 1996-2013. BMJ Open. 2019.
6. Purdy S, Huntley A. Predicting and preventing avoidable hospital admissions: a review. J R Coll Physicians Edinb. 2013;43(4):340-4.
7. Rafter N, Hickey A, Condell S, et al. Adverse events in healthcare: learning from mistakes. QJM. 2015;108(4):273-7.
8. Halfon P, Staines A, Burnand B. Adverse events related to hospital care: a retrospective medical records review in a Swiss hospital. Int J Qual Health Care. 2017;29(4):527-33.
9. Admi H, Shadmi E, Baruch H, Zisberg A. From research to reality: minimizing the effects of hospitalization on older adults. Rambam Maimonides Med J. 2015;6(2):e0017.
10. Moore L, Lauzier F, Tardif PA, et al. Low-value clinical practices in injury care: A scoping review and expert consultation survey. J Trauma Acute Care Surg. 2019;86(6):983-93.
11. INESSS. Continuum de services en traumatologie. In: sociaux Indeesees, editor. 2015.
12. Moore L LA, Sirois MJ, Amini R, Belcaïd A, Sampalis JS. Evaluating trauma center process performance in an integrated trauma system with registry data. J Emerg Trauma Shock. 2013;6(2):95–105.
13. Mokkink LB, Terwee CB, Patrick DL, et al. The COSMIN checklist for assessing the methodological quality of studies on measurement properties of health status measurement instruments: an international Delphi study. Qual Life Res. 2010;19(4):539-49.
14. Foks KA, Cnossen MC, Dippel DWJ, et al. Management of mild traumatic brain injury at the emergency department and hospital admission in Europe: A survey of 71 neurotrauma centers participating in the CENTER-TBI study. J Neurotrauma. 2017.
15. Saragiotto BT, Maher CG, Lin C-WC, et al. Canadian C-spine rule and the National Emergency X-Radiography Utilization Study (NEXUS) for detecting clinically important cervical spine injury following blunt trauma. Cochrane Database of Systematic Reviews. 2018.
16. Tavarez MM, Atabaki SM, Teach SJ. Acute evaluation of pediatric patients with minor traumatic brain injury. Curr Opin Pediatr. 2012;24(3):307-13.
17. Beaudin M, Saint-Vil D, Ouimet A, Mercier C, Crevier L. Clinical algorithm and resource use in the management of children with minor head trauma. J Pediatr Surg. 2007;42(5):849-52.
18. Blackwood BP, Bean JF, Sadecki-Lund C, et al. Observation for isolated traumatic skull fractures in the pediatric population: unnecessary and costly. J Pediatr Surg. 2016;51(4):654-8.
19. Joseph B, Pandit V, Haider AA, et al. Improving Hospital Quality and Costs in Nonoperative Traumatic Brain Injury: The Role of Acute Care Surgeons. JAMA Surg. 2015;150(9):866-72.
20. Biffl WL, Leppaniemi A. Management guidelines for penetrating abdominal trauma. World J Surg. 2015;39(6):1373-80.
21. Schonfeld D, Lee LK. Blunt abdominal trauma in children. Curr Opin Pediatr. 2012;24(3):314-8.
22. Kevric J, O’Reilly GM, Gocentas RA, et al. Management of haemodynamically stable patients with penetrating abdominal stab injuries: review of practice at an Australian major trauma centre. Eur J Trauma Emerg Surg. 2016;42(6):671-5.
23. Krogstad U, Hofoss D, Veenstra M, Gulbrandsen P, Hjortdahl P. Hospital quality improvement in context: a multilevel analysis of staff job evaluations. Qual Saf Health Care. 2005;14(6):438-42.
24. Turner O, Susan B, William L. A Modification of the Injury Severity Score That Both Improves Accuracy and Simplifies Scoring. The Journal of Trauma: Injury, Infection, and Critical Care. 1997;43.
25. White IR, Royston P, Wood AM. Multiple imputation using chained equations: Issues and guidance for practice. Stat Med. 2011;30(4):377-99.
26. Simma B, Lutschg J, Callahan JM. Mild head injury in pediatrics: algorithms for management in the ED and in young athletes. Am J Emerg Med. 2013;31(7):1133-8.
27. Coon ER, Young PC, Quinonez RA, et al. Update on Pediatric Overuse. Pediatrics. 2017;139(2).
28. Metzger RR, Smith J, Wells M, et al. Impact of newly adopted guidelines for management of children with isolated skull fracture. J Pediatr Surg. 2014;49(12):1856-60.
29. Powell EC, Atabaki SM, Wootton-Gorges S, et al. Isolated linear skull fractures in children with blunt head trauma. Pediatrics. 2015;135(4):e851-7.
30. Reuveni-Salzman A, Rosenthal G, Poznanski O, Shoshan Y, Benifla M. Evaluation of the necessity of hospitalization in children with an isolated linear skull fracture (ISF). Childs Nerv Syst. 2016;32(9):1669-74.
31. Rollins MD, Barnhart DC, Greenberg RA, et al. Neurologically intact children with an isolated skull fracture may be safely discharged after brief observation. J Pediatr Surg. 2011;46(7):1342-6.
32. Mannix R, Bourgeois FT, Schutzman SA, Bernstein A, Lee LK. Neuroimaging for pediatric head trauma: do patient and hospital characteristics influence who gets imaged? Acad Emerg Med. 2010;17(7):694-700.
33. Carlson AP, Ramirez P, Kennedy G, et al. Low rate of delayed deterioration requiring surgical treatment in patients transferred to a tertiary care center for mild traumatic brain injury. Neurosurg Focus 29 (5):E3. 2010.
34. Goodman TR, Mustafa A, Rowe E. Pediatric CT radiation exposure: where we were, and where we are now. Pediatr Radiol. 2019;49(4):469-78.
35. Wong AC, Kowalenko T, Roahen-Harrison S, et al. A Survey of Emergency Physicians’ Fear of Malpractice and Its Association With the Decision to Order Computed Tomography Scans for Children With Minor Head Trauma. Pediatric Emergency Care. 2011;27.
36. Paul AR, Adamo MA. Non-accidental trauma in pediatric patients: a review of epidemiology, pathophysiology, diagnosis and treatment. Transl Pediatr. 2014;3(3):195-207.
37. Stuke L, Duchesne J, Hunt J, et al. Mechanism of Injury Is Not a Predictor of Trauma Center Admission. The American surgeon. 2013;79:1149-53.
38. Bracken ME, Medzon R, Rathlev NK, Mower WR, Hoffman JR. Effect of intoxication among blunt trauma patients selected for head computed tomography scanning. Ann Emerg Med. 2007;49(1):45-51.
39. Jerome R. Hoffman, Allan B. Wolfson, Knox Todd, William R. Mower. Selective Cervical Spine Radiography in Blunt Trauma: Methodology of the National Emergency X-Radiography Utilization Study (NEXUS). Annals of Emergency Medicine. 1998;32:4.
40. Cynthia J. Brown, Rebecca J. Friedkin, Sharon K. Inouye. Prevalence and Outcomes of Low Mobility in Hospitalized Older Patients. American Geriatrics Society. 2004;VOL. 52, NO. 8.
41. Currie L. Fall and Injury Prevention. 2008.
42. Chaudhry SI, McAvay G, Chen S, et al. Risk factors for hospital admission among older persons with newly diagnosed heart failure: findings from the Cardiovascular Health Study. J Am Coll Cardiol. 2013;61(6):635-42.
43. Landefeld CS. Care of hospitalized older patients: opportunities for hospital-based physicians. J Hosp Med. 2006;1(1):42-7.
44. Graf C. Functional Decline in Hospitalized Older Adults. The American Journal of Nursing. 2006.
45. van Vliet M, Huisman M, Deeg DJH. Decreasing Hospital Length of Stay: Effects on Daily Functioning in Older Adults. J Am Geriatr Soc. 2017;65(6):1214-21.
46. Age UK. Age UK briefing - discharging older people from hospitals. 2016.
47. Christina R. Victor BA M Phil PhD Hon MFPHM, Judith Healy BA Dip Soc Stud MSW PhD, Anna Thomas BSc PhD, CQSW JSBMD. Older patients and delayed discharge from hospital. Health and Social Care in the Community. 2000.
48. Bond K, Ospina MB, Blitz S, et al. Frequency, Determinants and Impact of Overcrowding in Emergency Departments in Canada: A National Survey. Healthcare Quarterly. 2007;10.
49. Canadian Institute for Health Information. Unnecessary care in Canada . 2017.