Unnecessary hospitalisation and investigation of low risk
patients presenting to hospital with chest pain
Aggarwal L, Perera M, Scott IA*
Department of Internal Medicine and
Medical Assessment and Planning Unit, Princess Alexandra Hospital, Brisbane
Objectives: Among patients with undifferentiated
chest pain admitted from emergency departments (EDs), less than 20% have acute
coronary syndrome (ACS) as final discharge diagnosis suggesting potentially
avoidable admissions. This study determined clinical characteristics, risk level,
investigations and outcomes of patients with chest pain admitted to a medical
assessment and planning unit (MAPU).
Method: Retrospective study of all consecutive patients
admitted to MAPU between February and June 2012 for evaluation of chest pain. Demographic
data, ECG results, cardiac biomarker levels, Thrombolysis in Myocardial
Infarction Score (TIMI score) and results of non-invasive cardiac investigations
and coronary angiography were analysed. Outcome measures comprised confirmed
ACS, all-cause death, and readmissions to hospital over six months follow-up.
Results: 321 patients were studied aged 58.5 +/-
SD 14.2 years with TIMI score of 1.8 (+/-1.7). Exercise stress test and
computerised tomography of the coronary arteries were performed in 16.5% and
13.7% of patients respectively, with a combined positivity rate for probable
coronary artery disease (CAD) of 3.5%. At discharge, 31(9.6%) patients were
diagnosed with ACS, of whom 25 (80.6%) were diagnosed in ED prior to MAPU admission.
Of 290 patients with initially negative investigations in ED, only six (2.0%)
were subsequently diagnosed with ACS during hospital stay. At 6 months
follow-up, only one patient (0.3%) represented with ACS and two (0.6%) died of
non-coronary causes. Among 165 patients (51.4% of total cohort) with TIMI score
of 0 or 1, only seven (4.2%) had confirmed ACS at discharge; none were
readmitted with ACS or died up to 6 months.
Conclusion: More than 9 out of 10 patients admitted
to MAPU for evaluation of chest pain after initially negative investigations in
ED have a very low risk of ACS, death or cardiac-related readmission at 6
months. Further in-patient testing has very low yield for CAD. Strategies are
needed to identify such patients and facilitate their early discharge.
Underuse
of risk assessment and overuse of CTPA in patients with suspected pulmonary
thromboembolism
Aggarwal L, Perera M, Scott IA*
Department of
Internal Medicine and Medical Assessment and Planning Unit,
Princess Alexandra
Hospital, Brisbane.
Objectives: Increasing use of computerised tomography
pulmonary angiography (CTPA) in patients with suspected pulmonary
thromboembolism (PTE) without commensurate improvement in clinical outcomes suggests
possible overuse. This study assessed the use and clinical utility of pre-test
clinical prediction rules and D-Dimer assays in ruling PTE in or out in
patients presenting to hospital with suspected PTE, and identifying those who
warrant CTPA.
Methods: All consecutive patients undergoing CTPA at a
tertiary hospital between August 1st and December 31st
2013 were studied retrospectively. Use of D-dimer assays and clinical
prediction rules for PTE were evaluated by review of clinical notes. For each
patient, a modified Wells score (mWS), revised Geneva score (rGS) and PISA
model were calculated retrospectively
and performance characteristics for PTE determined in reference to CTPA results.
Results for the mWS and D-dimer assays (when performed) were used to estimate
overuse of CTPA according to risk category.
Results: Of 344 patients undergoing CTPA, 53 (15.4%) were
diagnosed with PTE. Use of a pre-test
PTE prediction rule was documented in only 5.0% of patients. Of 269 low risk patients
(78.2% of total cohort) who had a calculated mWS ≤4, only 64 (23.8%) had a
D-Dimer assay performed, and only 30 (11.1%) had PTE on CTPA. Among 75 patients with a mWS >4, 23
(30.7%) had PTE on CTPA (p<0.001). Of all prediction rules, a high risk mWS
had the highest positive predictive value (31%) for PTE in this cohort; all
rules demonstrated similar negative predictive values for low risk scores (between
87% and 89%). After adjusting for 11% false negative rate for PTE in patients
with low risk mWS, avoidable overuse of CTPA was possible in up to 190 (55.2%)
patients.
Conclusion: Use of clinical
prediction rules and D-dimer use need to be systematised in emergency medicine
practice in facilitating more selective use of CTPA in suspected PTE.
The elephants in the
room – the role of cognitive biases in overdiagnosis and overtreatment
Scott
IA*, Soon J, Elshaug A, Lindner R.
*Director
of Internal Medicine and Clinical Epidemiology, Princess Alexandra Hospital;
Associate Professor of Medicine, University of Queensland, Brisbane
Background: Among the
many drivers of low value care, the role of cognitive biases in clinician
decision-making deserve greater attention. Strategies frequently adopted to
counter low value care – such as education programs, guidelines, and audit and
feedback – are based on rationalist principles (ie system 2 thinking). However,
clinical decision making often defaults to mindlines and heuristics (ie System
1 thinking) based on highly personal and internalised beliefs vulnerable to
cognitive bias.
Methods: A narrative
review of relevant literature combined with insights from authors’ experience
in leading overuse minimisation projects were employed in identifying several
biases and strategies for overcoming them. These will be illustrated using
clinical examples.
Results: Common
biases include: clinician regret of omission (or loss aversion), availability
heuristic, optimism and outcome bias, confirmation bias and framing effects,
innovation (or novelty) bias, endowment effects and sunken costs fallacy,
extrapolation bias, affect and authority bias, certainty and reassurance needs,
and groupthink.
Debiasing strategies
include: cognitive autopsies and huddles for sharing discomfort with
uncertainty and omission regret, ‘teachable moment’ narratives of patient harm
that invert availability heuristics, emphasising gains over losses in clinical
outcomes, reframing overuse messaging from negative to positive, presenting
avoidance of low value care as innovative renewal, offering alternative forms
of care to nullify endowment effects and sunken costs fallacy, building
equipoise for more research to counter extrapolation bias, using normalisation
of deviance and nudge techniques to preserve autonomy, and deploying shared
decision-making.
Conclusion: Cognitive
and psychosocial factors that influence clinician decision-making deserve more
attention in determining ways in which behavioural economics can be used to minimise
overuse of care.