Title:A Flexible Simulation Platform to Quantify and Manage Emergency Department Crowding

Transcription

1 Author's response to reviews Title:A Flexible Simulation Platform to Quantify and Manage Emergency Department Crowding Authors: Joshua E Hurwitz (jehurwitz@ufl.edu) Jo Ann Lee (joann5@ufl.edu) Kenneth K Lopiano (klopiano@gmail.com) Scott A McKinley (scott.mckinley@ufl.edu) James Keesling (kees@ufl.edu) Joseph A Tyndall (tyndall@ufl.edu) Version:3Date:28 May 2014 Author's response to reviews: see over

2 Response to Referees Joshua E. Hurwitz, Jo Ann Lee, Kenneth K. Lopiano, Scott A. McKinley, James Keesling, Joseph A. Tyndall May 28, 2014 We would like to thank the reviewers for their thoughts and suggestions. After addressing the reviewers comments, we believe the ideas presented in the manuscript have been improved. In the text below, we offer our itemized responses to each of these comments. Referee 1 In paragraph 3 of Background: a reference is made to near - future ED crowding forecasts. What kind of time frame is this in reference to? Is near - future in reference to days, week, months or years? Are mathematical and computational models of relevance to day to day functioning or more longer term functioning of an ED? We realize that this was ambiguous. We ve added on a scale of hours to the text and believe this addresses the concern. Also, mathematical models are relevant at all the time scales the reviewer mentioned. We believe this is illustrated throughout the text and in the corresponding references. In paragraph under goals of investigation : it may be helpful to clarify that national is in reference to US ED s as this model cannot necessarily be used for international ED s. Added..., both in the United States to the manuscript. Paragraph 2 under patient flow model: Has physician history taking time been factored in? Only physical examination time is specifically mentioned. Changed to: A history is taken and a physical exam is then performed by a physician... Same paragraph: attributed to? What is the disposition to discharge exit time being 1

3 The disposition to-exit time for discharged patients includes the time it takes patients to receive discharge instructions and physically exit the ED. Paragraph 4 of simulation details: assumption is made that the proportion of each acuity group s arrivals is constant throughout the day and that arrival intensity functions can be easily adjusted to fit data from any emergency department. Given that arrival of patients at any given time in the real world is unpredictable how can the data be easily adjusted to fit any emergency department. Can you please clarify? There is a distribution of arrivals associated with each acuity for each hour of the day. For example, the number of ESI-3 patients who arrive between 12pm-1pm each day has a Poisson distribution with a certain rate parameter. The rate parameters for different hours are independent, as are the rate parameters for different acuities. This means that we can use arrival pattern data from any institution to generate arrival intensity functions that input the time of day and output the Poisson rate parameter. We do this by looking at the average number of arrivals for each acuity for each hour of the day. The simulation then feeds these intensity functions into a non-homogenous poisson process. Paragraph 1 under Explication: identifying site - specific causes of crowding discusses prolonged door-to-bed times in academic institutions. Is there data to support that? Added citation 35, a retrospective study examining the relationship between acuity (as determined by ESI) and length of stay. The study looked at data from a large, academic ED and found that ESI-3 patients had (on average) the longest length of stay of all acuities. Minor Essential Revisions Section Importance first sentence...improve efficiency, and organizational... : Consider breaking up into two sentences. Now reads: ED management face a variety of options when deciding how to improve efficiency, and seemingly straight-forward operational innovations can be rendered ineffective by counterintuitive patient flow dynamics [18, 23, 26]. Paragraph 4 under Discussion: What is (??) in reference to? Reference is to Table 1. The reference as been fixed. Paragraph 5 under Discussion: First sentence does not clearly clarify the point. Consider revising. Is it indented to be analysis of management through experiments? 2

4 Now reads: In this way, simulation of EDs does more than confirm management intuitions. Paragraph 5 under discussions, second last sentence: Consider taking the out when referring to conclusions from the ACEP. Removed the from the phrase. Reference 32: What are??? for? Now reference 28. Has been fixed. Referee 2 Title: Slightly problematic. In my opinion the simulation platform is not comprehensive and even less so as far as being site specific. Examples of being site specific would have to include topology or layout which this platform does not include. There are a few adjustable parameters associated with staff and beds but those are still fairly generic parameters. A more accurate title would be a preliminary simulation platform for qualitative assessment of emergency department overcrowding. Although actual numbers are presented in some of the results at best they provide trends and in most cases as expected. In addition to the present simulation being somewhat general, no simulation of agency of individuals or bursty arrivals is modeled. Topology and layout are important factors when assessing ED throughput. Our simulation accounts for this by incorporating the time it takes for a provider to move from room to room. This is modeled using time- and provider-dependent exponential random variables. This was previously not mentioned in the manuscript, but we have now added a paragraph to the Limitations section addressing this concern. Actual numbers are presented in all of our results, and we believe it speaks towards the validity of our simulation that it can replicate expected trends in ED operations. The strength of our simulation is that it is able to accurately quantify these trends. For example, knowing that additional resources decrease length of stay is not enough to evaluate whether adding those resources is worth it ED managers need to know how much. We have changed the title to A Flexible Simulation Platform to Quantify and Manage Emergency Department Crowding. With regard to agency of individuals, we have included patients tolerance for waiting before leaving without being seen. Providers also prioritize patients based on statedependent rules. 3

5 We do not explicitly consider bursty arrivals, though they are possible as a result of the stochasticity of the non-homogenous Poisson process. Should bursty arrivals occur due to phenomena such as bus arrivals or other outside influences, then this could be captured in the site-specific arrival rates. A final comment concerns the use of the term emergent on page 6. The simulation is more or less tuned to these outputs as the input parameters were statistical averages and the simulation is that of an ED. We have changed the wording to read outcomes of the simulation. 4