SGEM#324: Blow Away, Blow Away, Blow Away – Spirometry for Discharging Older Patients with Rib Fractures

Date: March 17th, 2021 Guest Skeptic: Dr. Emil Ejersbo Iversen is an emergency medicine resident at the University Hospital of Zealand, Denmark. He currently serves as Vice-Chair of the Danish Society for Emergency Medicine and Chair of the Young Doctors in Emergency Medicine in Denmark. He has a passion for FOAMed and is the creator of the Danish EM platform www.akutmedicineren.dk. Reference: Schuster et al. Spirometry not pain level predicts outcomes in geriatric patients with isolated rib fractures. J Trauma Acute Care Surg. 2020 Case: A 74-year-old woman who suffered a fall earlier today presents to the emergency department (ED) and is found to have five rib fractures to her right thorax, but no other injury. She is otherwise well, and her vitals are stable, but she is in some pain. Recent guidelines recommend admitting the patient to the intensive care unit (ICU), but the patient is eager to return home to her husband who is also well, and whom she claims will be able to help her. Background: Rib fractures are a common injury among the older population and can potentially lead to life-threatening complications such as pneumonia, pneumothorax or decreased inspiratory capacity. Some recent guidelines recommend admitting patients older than 65 years of age with two or more with rib fractures to an intensive care unit (ICU) or other step-down monitored setting [1]. Currently, patients with three or more rib fractures are often admitted for analgesia and monitoring and subsequently discharged without complications. Recent retrospective studies have suggested that early spirometry may be a useful indicator of prognosis in patients with multiple rib fractures [2]. Identifying patients with a good prognosis that could be safely discharged home with analgesia could potentially avoid unnecessary hospitalization. This would likely lower healthcare costs and decrease the risk of hospital-acquired infections. Spirometry includes metrics such as forced vital capacity (FVC), peak expiratory flow (PEF), forced expiratory volume 1 second (FEV1), and negative inspiratory force (NIF). The PEF has not been demonstrated to be closely correlated with patient outcomes [3]. However, FVC has been shown to correlate with patient outcomes and length of stay (LOS) in patients who have multiple rib fractures [4-5] These studies were limited by their retrospective observational nature. Hand grip strength has also been used to measure overall frailty. GeriEM guru Chris Carpenter has done some work in this area over ten years ago. His team found grip strength was weakly correlated with frailty in older ED patients [6]. Future research should confirm this association and assess the correlation of grip strength with other measures of frailty. Multiple other authors have investigated this simple and inexpensive tool for predicting frailty [7-8]. Clinical Question: Can spirometry testing identify patients 60 years and older with at least three rib fractures who can safely be discharged home from the ED? Reference: Schuster et al. Spirometry not pain level predicts outcomes in geriatric patients with isolated rib fractures. J Trauma Acute Care Surg. 2020 Population: Patients 60 years of age and older admitted to hospital with at least three rib fractures within 24 hours of injury Exclusions: Injury occurred >24hrs before presentation, significant additional musculoskeletal injury or cognitive impairment and able to cooperate with testing Exposure: Spirometry measuring (FVC, FEV1 and NIF) Comparison: Hand grip strength and pain assessment (VAS) Outcome: Primary Outcomes: Discharge disposition and length of stay (LOS) Secondary Outcomes:Mortality, pneumonia, intubation, unplanned transfer to higher level of care and readmission (within 30 days) Authors’ Conclusions: “Spirometry measurements early in the hospital stay predict ultimate discharge home, and this may allow immediate or early discharge. The impact of pain control on pulmonary function requires further study.” Quality Checklist for A Prognostic Study: The study population included or focused on those in the ED? Yes The patients were representative of those with the problem? Yes The patients were sufficiently homogenous with respect to prognostic risk? Yes Objective and unbiased outcome criteria were used? Yes/No The follow-up was sufficiently long and complete? Yes/No The effect was large enough and precise enough to be clinically significant? Unsure Result: There were 346 patients over the age of 60 admitted to hospital with isolated rib fractures. Exclusion criteria was met in 260 patients. This resulted in a cohort of 86 patients with a mean age of 77 years and 50% female. Just over half (45/86) were admitted to the step-down unit, 19/86 (22%) were admitted to the ICU and 22/86 (26%) to the surgical floor. The mechanism of injury was a fall (54%), motor vehicle collision (45%) or motorcycle collision (1%). The median number of fractured ribs was five. Pneumothorax was present in 5% and hemothorax in 4%. One patient out of 86 died (1.2%). Key Results: Higher spirometry values and grip-strength were associated with early discharge from hospital Primary Outcomes: Discharge disposition and length of stay FEV1 Adjusted Odds Ratio (aOR) 1.03 (95% CI; 1.01 to 1.06) p = 0.001 Grip strength was also significantly associated with being discharged home FVC and NIF were not statistically significant Pain score was poorly predicative of length of stay Secondary Outcomes: There were a few patients that had some of the secondary outcomes of interest (n). This included mortality (1), pneumonia (2), intubation (1), unplanned transfer to higher level of care (3) and readmission (3) within 30 days 1. Selection Bias: All patients were screened except when an investigator was unavailable. We suspect that this was nights, weekends and holidays. No details were provided in the manuscript of how many of the 260/346 (75%) of the exclusions were due to this reason. This could have introduced some selection bias into the data. 2. Power: There was no formal power calculation done a priori. The authors say a “rough” estimate to find a difference between the two groups would have been about 400 patients. It is unclear exactly how they arrived at this number. They did estimate a complication rate of about 20% based on the Geriatric Trauma Outcome Score (GTOS) [9]. The GTOS is calculated by taking the patients age + (injury severity score x 2.5) + 22 (if given packed red blood cells by 24 hours). Post hoc power (PHP) calculations really shouldn’t be done. It is good practice to do a power calculation a priori to plan your research project. In contrast, doing a PHP calculation can be misleading. It is better just to look at the confidence interval around the point estimate. The results are the results and looking backwards with a PHP calculation does not help interpret the results. Thank you to Andrew Althouse from EpiTwitter for providing me with a number of citations discussing this issue. Hoenig and Heisey. The Abuse of Power: The Pervasive Fallacy of Power Calculations for Data Analysis [10] Althouse AD. Post Hoc Power: Not Empowering, Just Misleading [11] Dziak et al. The Interpretation of Statistical Power after the Data have been Gathered [12] 3. Blinding: The investigators in this study were not blinded and although spirometry is a seemingly objective measurement it does require significant instructing of the patient and the fact that the same investigator performed both the pain assessment and the spirometry could have introduced “coaching/spectrum bias” Spectrum Bias When we use the term “bias” it is referring to something that systematically moves us away from the “truth”. And by truth we mean the best point estimate of an observed effect size with a confidence interval around estimate. Can you tell us more about Coaching/spectrum bias?  Coaching bias could be conscious or unconscious. It could occur when patients who were thought to have higher levels of pain were not pushed as hard to go through with the spirometry. This coaching bias is a form of spectrum bias. Sensitivity depends on the spectrum of disease, while specificity depends on the spectrum of non-disease. So, you can falsely raise sensitivity if the cohort has lots of very sick people and specificity can look great if you have no sick patients in the cohort. The best resource on understanding the direction of bias in diagnostic test accuracy is by Kohn et al published in AEM [13]. 4. Rib Fractures: All but four patients had their injuries identified by CT scan. Rib fractures are often missed on initial CXR in up to 50% of cases and one rib fracture on CXR is associated with a high risk of multiple rib fractures [14]. Are we discharging patients with undiagnosed multiple rib fractures? This could introduce denominator neglect. How many people were seen and not suspected of having a fractured rib, had CXR and was read as normal, or had a CXR and only one rib fracture was seen? There is evidence of this issue in the manuscript. The authors did not screen 322 acutely injured patients who were discharged directly from the ED. Three were believed to have subacute rib fractures by imaging characteristics. 5. Follow-up: The final nerdy point is about the follow-up in this study. They had their secondary outcomes of pneumonia, readmission and mortality at 30 days. Very few patients were observed to have any of these outcomes. Perhaps this is because patients can deteriorate more gradually, and these outcomes may not be realized until after 30 days. It would have been nice to see a longer follow-up period to be more confident there were not delayed adverse events.