Translation of the ADAPT Accelerated Diagnostic Protocol into clinical practice: Impact on hospital length of stay and admission rates for possible cardiac chest pain

WA Parsonage, S Ashover, T Milburn, W Skoien, J Greenslade, L Cullen - Royal Brisbane & Women's Hospital, Brisbane, Australia

IntroductionChest pain is the second most common single complaint in patientspresenting to Emergency Departments (EDs) in Australia. In 2014-15 chestpain accounted for 3.4% of ED presentations and 5.1% of hospitaladmissions. The most common serious cause of chest pain is acutecoronary syndrome (ACS), however up to 85% of patients presenting withchest pain do not have ACS.

Several accelerated diagnostic pathways (ADPs) that safely identifypatients who are at low risk of ACS have been derived but few have beenevaluated in practice. The ADAPT ADP was derived and published by ourgroup in 20121 and identifies low risk patients using a very simplealgorithm. The algorithm and the key findings of the ADAPT ADP aredetailed in Boxes 1 and 2.

The aims of the study were to test the feasibility of large scale translation ofthe ADAPT ADP into clinical practice and to measure the impact on healthservice delivery.

MethodAll government hospital EDs responsible for the care of adult patients andhaving access to laboratory based tests for cardiac troponin I wereapproached. Clinical pathways incorporating the ADAPT ADP wereintroduced into eligible hospitals through a structured process of clinicalservice redesign between May 2013 and September 2015. This wasimplemented by a small project team in collaboration with local clinicians.

A quasi-experimental observational design was used to evaluate the effectof implementing the ADP on parameters of patient flow. Patients presentingwith possible cardiac chest pain were identified from entry of relevantdiagnostic codes into the Emergency Department Information System(EDIS, Healthcare Group, CSC). After implementation of the ADP the EDISprompted staff to identify eligible ‘low risk’ patient using a single binaryquery (Yes/No). Where this was incomplete patients were considered to benot ‘low risk’. Primary diagnosis, arrival and discharge date/time anddischarge destination were extracted. Data linkage to inpatient hospitalrecords for admitted patients used the unique hospital identifier. Data wereextracted for the 12 months immediately prior to and after implementationof the ADP at each site and took place between May 2014 and November2015.

AnalysisA multilevel regression was used to compare the trends in hospital lengthof stay across time. Study period (pre/post implementation) was entered asa dichotomous variable to examine for a quantitative change in length ofstay after implementation of the ADP. Hospital was entered as a randomeffects parameter to account for differing baseline length of stay acrosshospitals.

Box 1: The ADAPT – ADP

TIMI Score = 0

cTnI <99th percentile at 0 and 2 hrs

Normal ECG at 0 and 2 hours

Before Intervention(n = 32,065)

After Intervention(n=36,133) p

Mean ED Length of Stay (95% CI)

291.0 min(259.1-326.9)

256.3 min(228.3-287.6) <0.01

Mean Hospital Length of Stay (95% CI)

57.5 hr(50.2-65.8)

44.0 hr(38.8-49.9) <0.01

Hospital Admission Rate (95% CI)

68.2 %(59.2-78.5)

52.2 %(42.3-64.7) <0.01

Analysis (continued)For the regression analyses, the post implementation period was capped at12 months to compare equal periods before and after the implementationof the ADP. Hospital length of stay was log transformed to ensure thatoutlying values did not obscure the results and back-transformedcoefficients were reported. A multilevel logistic regression was alsoconducted to compare trends in hospital admission over time. For thismodel, hospital admission was regressed on time, months afterimplementation and study period.

ResultsThe ADP was implemented across 16 hospitals and data reflect outcome of68,198 patients presenting with possible cardiac chest pain. 7,916 (21.9%of 36,133)) of patients were identified as ’low risk’ according to the ADPfollowing implementation.

Implementation of the ADP led to a significant reduction in ED length ofstay, hospital admission rate and total hospital length of stay (Table 1). Theimpact and sustainability of the ADP across time is illustrated for totalhospital length of stay in Chart 1.

LimitationsEvaluation of individual patient outcomes was beyond the scope of thestudy but the ADP implemented was consistent with the approach derivedby the widely cited ADAPT study. Clinical risk stratification was performedby local clinicians but not adjudicated centrally. However, the proportion ofpatients stratified as ‘low risk’ was consistent with the findings of theoriginal ADAPT derivation. All data was extracted from routinely collectedclinical/administrative data sets.

Conclusions

• The ADAPT ADP is clinically feasible and translates well intoclinical practice across a diverse range of hospital EDs

• The ADAPT ADP leads to substantial and sustainableimprovement in measures of patient flow including length of stayand hospital admission

• Implementation of the ADAPT ADP has the potential forconsiderable release of clinical capacity

References1. Than M et al. 2-Hour accelerated diagnostic protocol to assess patients with chestpain symptoms using contemporary troponins as the only biomarker: the ADAPT trial.J Am Coll Cardiol 2012;59:2091– 8.

Contact: w.parsonage@mac.comWeb: https://emergencycardiologygroup.wordpress.comConflict of Interest: None declared

Box 2: Key findings of the ADAPT Study

20% of 1975 patients ‘low risk’

The ADP had a 99.7(98.1-99.9)% sensitivity and 99.7(98.6-100)% negative predictive value for

MACE

Table 1 : Effect of implementation of the ADP on patient flow parameters

1416

1820

2224

Hos

pita

l len

gth

of s

tay

(hou

rs)

12 11 10 9 8 7 6 5 4 3 2 1 1 2 3 4 5 6 7 8 9 10 11 12Months before implementation of ACRE Months after implementation of ACRE

Months before intervention Months after intervention

Hospital Length of

Stay (Hours)

Chart 1 : Effect of implementation of the ADP on hospital length of stay by month