The Mount Sinai Experience

How One Urban Academic Center Successfully Implementing a Comprehensive ED Information System

Ours was once a department like many others – urban, academic, and overwhelmingly busy. Care in our department involved over two dozen different processes supported by separate, disparate systems: from triage and registration through laboratory, radiology, medication orders and review of previous medical records, to bed request or discharge instructions. Patients waited at each step of the process. Documentation was lacking, with many charts lost or insufficiently complete for proper reimbursement.

We looked to redesign ED throughput as an opportunity to fulfill our core mission to provide quality patient care, and also, as an academic institution, to effectively teach residents and promote research.

We settled on an electronic chart because it served so many key functions: documentation of the care delivered, communication to each other and to our colleagues on other services, official legal record for purposes of litigation, generation of facility charges and professional fees, and a source of data for clinical research. Compared to unstructured or template paper charts, or dictation systems, or scribes, only an electronic chart system could provide us with a completely legible record of the care given during the visit, with every data element stored in a searchable database, and the capability to give real-time decision support to improve clinical care and to optimize reimbursement.
We chose to seek a comprehensive EDIS that provided triage, patient tracking, electronic physician and nurse charting, electronic order entry, discharge instructions and prescription writing. While setting up a comprehensive EDIS requires more initial preparation, training and investment, we figured more functionality from the get-go would actually be worth it, and that a single comprehensive implementation and go-live date would ultimately be less disruptive than expanding a limited feature set slowly over many months.

One essential feature we agreed upon early was a single, universal log-in. We’ve witnessed the inefficiency and frustration of doctors carrying around separate passwords to log in to one system for lab results, another for charting, and a third for prescription and discharge documents – often on different, dedicated computers. We wanted every feature of our comprehensive EDIS available to every user on every computer in the ED.

Other essential characteristics included a user-friendly interface that facilitated the work of ED staff physicians and nurses; useful administrative data on throughput times, turn-around times, staff productivity, resource utilization and quality of care; and facilitation of data collection for research in the ED. It was also specified that the system must sufficiently improve documentation and billing to pay for itself within 18 months.

Having chosen this direction, we took a lot of steps to make sure our ED information system was implemented well. Because attempts to automate dysfunctional processes always seem to produce unsatisfactory results, we performed a workflow analysis and mapped out each step in patient flow through our ED, and redesigned our processes for IT optimization.

The emergency department is nothing like an inpatient unit, a doctor’s office or a clinic*. The pace, nature and organization of the work are unique. All activities are time sensitive. ED providers see multiple patients simultaneously and are constantly interrupted. Multiple providers must be able to access the ED chart simultaneously. Orders are often placed singly rather than in sets. An understanding of these workflow issues should be apparent in the design of an EDIS.

So we went to trade shows, reviewed the literature available at the time, and surveyed our colleagues at other institutions.

Then, we had structured vendor site visits. We asked doctors, nurses, registrars and administrators to evaluate the products using standardized scoring sheets, and brought in hospital IT personnel to take a look at our finalists and determine support needed for each. This screening period had the added affect of building consensus and achieving buy-in from our various stakeholders – because so many people had a hand in choosing features and final product, they were invested in making sure it worked. Buy-in was no doubt also helped because our EDIS champions were physicians and nurses, rather than something foisted upon the department by outside administrators or consultants.

Implementation was expensive, and had a steep learning curve. But when it was complete, our process redesign and paperless EDIS implementation enabled a single sign-on capability for access to all system applications and a streamlined, more efficient operation.

The EDIS provides triage, patient tracking, physician and nurse documentation, retrieval of charts from prior ED encounters, one-click access to more extensive historical hospital data from an enterprise data repository, computerized provider order entry, results review, discharge instructions and prescription writing. All data entered into the EDIS are time-stamped. Patient care documents from other departments or facilities are scanned into the patient’s electronic chart and are simultaneously viewable by all personnel caring for the patient.

Switching to a paperless ED has had a tremendous effect on our department – some effects are easy to measure, others less so. Our own data (some of this previously published, some of this in press) about ROI and implementation is, of course, unique to our busy urban academic center.

•When we compared a period over a year post-implementation to the days before our EDIS, the ED Length-Of-Stays for all patients decreased by 29%, from 6.7 hours pre-intervention to 4.8 post-intervention. The ED LOS for admitted patients decreased 35%, from 12.2 hours pre-intervention to 8.0 post-intervention

•Door-to-doctor time for all patients (triage to first doctor-patient contact) decreased 44%, from 1.2 hours to 0.7, while doctor to disposition time for all patients (first doctor-patient contact to disposition decision) also decreased 52%, from 3.6 hours to 1.7 hours. Disposition to discharge for admitted patients (boarding time) decreased 28%, from 6.8 hours to 4.9 hours.

•CT scan turn-around time decreased by 40% from 3.9 hours to 2.3, laboratory report TAT decreased from 2.0 hours to 1.4, and X-ray TAT decreased from 0.9 hours to 0.7.

•On the revenues side, average collections per patient rose 47.5% between pre-implementation and a sustained-effects period over a year later. Total charges rose 69.4% during the same period and total receipts rose 70.1%.

•End-of-month chart completion rates by attending physicians rose from 65% in 2003 to 95% in 2005, while lost or illegible charts decreased from 4,992 in 2003 to zero in 2005. The average professional evaluation and management (EM) levels (the five-point scale used for coding ED charts for billing purposes) rose from 3.17 during the pre-implementation period to 3.73 during the a period more than a year implementation.

•The number of charts meeting professional fee criteria for critical care billing increased from just one chart during the entire 15-month pre-implementation period to 1,614 charts in the more than a year afterward, and from 45 to 974 respectively for facility critical care billing.

•Despite an overall decline in the facility EM level, net facility receipts increased 60.9% between the pre-intervention period and a time over a year later.

As dramatic as our improvement was post-implementation, this data says nothing about improvement in patient safety or satisfaction, or improvements in QA and core measure monitoring, all of which have been noted but not as rigorously reviewed. Furthermore, as an academic institution, the EDIS has made it far easier for residents to review cases for educational or research purposes, which allows us to fulfill an important mission as a teaching hospital.

But perhaps most importantly, if you survey faculty who lived through the transition, they’ll all note that, as challenging as adoption of an EMR was, no one would go back.

*Because the ED is not a unit or a clinic, interpreting the growing literature about EMR adoption in various healthcare settings becomes more problematic. Many of the concerns ED docs have about EMR – more clerical duties in front of a computer, less face-time with patients, expensive and time-consuming training – have been measured and quantified in other settings. So, too, have the benefits in safety, increased revenue, core measure and QA compliance. If someone is waiting for a more precise cost-benefit analysis for EMR implementation for the unique environment of the ED, they’d do well to remember that each ED itself is unique, and what ultimately gets reported from one ED’s experience may not apply well to another.


  1. So, who would you recommend getting an antiviral? From prior influenza–symptoms within 48 hours of ED arrival, children and elders?

  2. Amesh Adalja, MD on

    The CDC’s position is that “Empiric antiviral treatment should be considered for confirmed, probable or suspected cases of swine influenza A (H1N1) virus infection.” Antivirals are most effective when given within 48 hrs. of symptom onset, but there is evidence of decreasing mortality and hospital stays with antiviral treatment started later than 48 hrs. I believe the highest priority should be treat those that hospitalized and at high-risk.

  3. Joe Howton MD on

    If supplies of oseltamivir become tight, we should remember that giving probenecid will block excretion, thereby boosting levels of oseltamivir. We could thereby extend the supply of oseltamivir to treat more patients.

  4. What is the evidence that Oseltamifivr decreases mortality from influenza? I thought that this year a lot of influenza a was resistant and you needed to add amantadine. Is this still the case, even for swine flu?.

  5. For health care edit docs flu symptoms positive testing rapid influ a what exact recommendation would you start considering resistence pattern and side affects and avality to meds?

    I am ed Md please state exact med dosage lenghth time understanding CDC will do some quess work for general pop. Also I don’t hold you responsible for what you would take and understand you are not the CDC
    Also why two agents? Tamiflu plus ?
    Relafen alone. Probenecid plus?

    I will need to continue working


  6. Here’s a link to a paper from CID regarding the decrease in mortality (OR 0.21) with use of oseltamivir:

    The CDC is recommending use of either amantadine and rimantidine with oseltamivir because there are still seasonal influenza A strains (H1N1) circulating that are all resistant to oseltamivir. Rapid tests will not distinguish which virus is present. The current swine influenza isolate is resistant to amantadine and rimantidine, but unless you can rule out seasonal H1N1 influenza, 2 drugs are needed.

    The dosage for adults with Tamiflu is 75mg orally twice daily for 5 days. For Relenza, dosing is two inhalations twice daily for 5 days.

  7. My understanding of the CDC recommendations is for “consideration” of treatment with anti-virals for confirmed or suspected cases. I doubt a consensus exists at this point on whether we should treat all-comers or only those who need hospitalization or who have higher risks of complications. Of concern, of course, if we treat everyone with an URI with these drugs, the risk of resistance and drug availability for those who truly need it is very high. My preference would be not to treat the uncomplicated/low risk case. Is this justified?

  8. Amesh Adalja, MD on

    I think it would be justified to not treat uncomplicated cases. In fact, the vast majority of the early US cases did not receive antiviral treatment and recovered uneventfully.

  9. Matthew Perl, M.D. on

    I called seven pharmacies in San Diego last night to find Tamiflu for a patient, including CVS, Sav-On, Rite-Aid and Costco; none had the medication, nor did they expect to be receiving any in the near future.

  10. Amesh Adalja, MD on

    If a patient is able to use the Relenza inhaler, that is an equally good option if Tamiflu is unavailable.

  11. What do you recommend for children under age 7? Tamiflu liquid is no longer available, and Relenza is not approved for this age. Thanks

  12. Are N-95s recommended for patients, or are regular surgical masks adequate? I know N-95s are best for HC workers and contacts due to aerosolization, but since the patient is coughing/sneezing larger particles, is a surgical mask OK?

  13. Amesh Adalja, MD on

    I am not sure of the best course of action when Tamiflu pills can’t be taken and Relenza is not a choice. I can’t find any good information on whether Tamiflu can be crushed, etc.

    Regarding infection control, the CDC recommends that patients should wear a surgical mask when outside their room.

  14. Matthew Perl, M.D. on

    We’ve haven’t had Relenza in San Diego at all this year. I called around for it in January, then again when trying to get Tamiflu yesterday.

  15. Chris Carpenter on

    The first US swine flu death was reported today (Texas toddler). Since the current flu epidemic is occurring concurrently with spring allergy season, can clinicians use any decision aids to differentiate swine flu from non-swine flu and/or allergic rhinitis? I know that fever is a distinguishing feature for flu vs. allergies, but I’m not sure of the sensitivity/specificity of fever nor am I confident when patients tell me they’ve not taken any anti-pyretics prior to my evaluation.

    In 2004 shortly after the SARS epidemic, Wash U EM did a Journal Club on several SARS Clinical Decision Rules which are archived online
    ( Is there any evidence to suggest these rules (or others) might be useful during the current swine flu epidemic? Were any descriptions of the diagnostic test performance of bedside signs/symptoms noted during the last swine flu epidemic in 1976?

    Thanks for an informative synopsis and reliable updates on this flu outbreak, Amesh.

  16. Mike Klevens, MD on

    I would like a comment from someone on the specificity and sensitivity of screening using nasal swabs. Our manufacturer claimed specificity is >95% and the manufacturer claimed sensitivity is 92%. I’ll bet this is in line with other rapid influenza testing. In a setting of low prevalence (i.e. currently in Missouri) this is a pretty crappy screening tool. So, a CDR such as the rules for SARS noted by Dr. Carpenter, would be helpful prior to testing everyone with a URI or viral syndrome. Comments?

  17. Amesh Adalja, MD on

    I don’t think there is any reliable way to distinguish swine influenza from seasonal influenza clinically. Some cases have reported the presence of nausea, vomiting, and diarrhea–a clinical feature not characteristic of seasonal flu.

  18. Chris Carpenter on

    Although I am not at all confident that this research can be extrapolated to swine flu, Wash U EM Journal Club had evaluated rapid point of care tests for Influenza in January 2007 ( with the following bottom line:

    QuickVue (one influenza POC test) is far superior (LR 28.2 for QuickVue vs. 3.8 for gestalt or 5.1 for Monto’s CDR). Clinical gestalt is improved when symptoms have been present for

  19. Chris Carpenter on

    Sorry — above response got cut off for some reason.

    Clinical gestalt is improved in adults when symptoms have been present for < 48 hours. QuickVue Influenza test is a useful POC test in children under age 5 with LR 126 and LR- 0.18 with excellent reproducibility (Kappa = 0.98) when obtained by trained research nurses in those with non-specific upper respiratory symptoms. Despite these admirable test characteristics, a positive QuickVue test does not appear to impact overall or test-specific diagnostic testing, antibiotic prescribing or appropriate antiviral use. Appropriate use of POC tests like QuickVue have the potential to reduce ED LOS and inappropriate testing/prescribing while relieving parental anxiety and maintaining up-to-date
    regional surveillance data, but further studies will be needed to assess the utility and acceptability of these possibilities at various health care settings.

  20. David Gray MD on

    Amesh, where and how can this viral strain be identified? Can any lab do it, or does a specimen (what type?) have to be sent somewhere, ie; CDC?

  21. Amesh Adalja, MD on

    Viral identification is done through your state health department laboratory. Some academic centers may be able to figure out that it is a probable case with PCR based diagnostics (i.e. finding out it is an untypable isolate).

  22. Brian Melito on

    In my reading of the studies to date, it seems that the current choices of antivirals are, at best, marginally effective. The 2007 Ontario study looked at a group with a median age of 77. There was reduction in mortality shown, but there were many reasons this data could not, and should not, be used in any way to make projections across the general population. My simple question is this: given the now extremely broad recommendations for use of these drugs during this period, how many people with “the flu” do I need to treat in order to save one individual? Thanks.

  23. Amesh Adalja, MD on

    I agree that CDC guidance on antiviral treatment is overly broad. I think the people that must be treated are those that are hospitalized and those with high-risk features (HIV , immunosuppressed, elderly, COPD, infants, etc). I am not sure there is enough enough evidence to say how many people need to be treated to save one life at this point. The vast majority of people will likely recover without antiviral treatment, so it should be prioritized to those at high risk of death and other serious complications.

  24. John Murphy on

    One of the major questions is the message to send the public regarding whether and how to access the health care system. We are trying an overriding message of “stay home but see us if you have risk factors.” It’s not a very catchy sound bite, and a little tough for the public to process. The upside with this approach is you might decrease clinic traffic, the risk is you might discourage a healthy 28 year-old from getting treatment who could develop a life-threatening complication…
    Also, we’re seeing a marked shortage of Tamiflu suspension, even in the stockpile meds. you might want to check with your pharmacies to see if they are willing to compound the capsules into the susp; instructions on Roche website.

  25. VacationPlans on

    I will be going on vacation with the family tomorrow to Orlando, FL. While I am less concerned, my wife is concerned. We have discussed cancelling the trip (tomorrow -w hich would cost a substantial amount of money ). We have two children a 1 year old and 2.5 year old, thus cancelling a trip is secondary concern, except that I really don’t want to do so where there are unreasonable risks. Specifically, I don’t want to cancel a vacation where the odds of any of us becoming fataly ill would be at the level of me dying of a heart attack prior to adding this comment.

    My thoughts are that because the number of fatalities in Mexico, where the disease originated has been relatively low (and does not appear to be expanding), that the mortalitiy risk factors are also very low. That being said the one confirmed death in the US was a toddler, which off course is of concern even if the fatality was anecdotal.

    Our pediatrition, went on to recommend to my wife that canceling the trip would be advisable because of the number of people one would come in contact with at a theme park. My response to that (from a non medical approach) is that wouldn’t we have started to see some signs by now if this were blowing up?)

    My feeling is that 300 confirmed cases worldwide, with the appearance of some stability at the source does not rise to the level of a serious concern.

    Lastly, in the even my kids develop a cold or flud like symptoms, what specifically should we look for. My feeling is that if we are vigilent in the event one of us gets sick, we can mitigate against the risk of a mortal event.

    I guess my questions is two fold:

    1.) Does what you know rise to the level of cancelling a trip.

    2.) If I take my chances, and react quickly with respect to symptoms, is it reasonable to assume that I will mitigate any serious risk of death.

    Best Regards
    Concerned Parent

  26. Logan Plaster on

    The following is an excerpt from today’s Q&A teleconference with Dr. Besser of the CDC:

    “Is it safe to fly? I ask that because today on the “Today” show, vice president Joe Biden said if it were his family members he would tell them not to fly at all, that it’s not safe.”

    “There’s a lot of things that we can do to try and reduce our risks. A lot of things people are doing on their own. For us in public health it’s important to say what things are evidence-based, what things can you do to put yourself at risk and what things can you do to reduce your risk. In terms of flight, if you have a fever, flu-like symptoms, you should not be getting on an airplane. That is part of being a responsible part of our community. You don’t want to put people at risk. I think flying is safe. Going on the subway is safe. People should go out and live their lives. There are some people who may not be comfortable doing that. As a public health community, we can put in context what the risk is. People are doing things to reduce their risk, hand washing, covering of the cough, avoiding ill people. And if we look to each other to be responsible and not get on airplanes and places when we’re sick, that makes everyone else safer.”

  27. concerned doc on

    You note that this week you have shown employees how important they are and their health is protected. How have you done that? If this virus mutates into a potentially more virulent and deadly one come the next flu season, what measures are in place to separate patient populations from the get-go such that other patients in the waiting area don’t get infected. What measures are in place such that the nurses, physicians, EMS workers and laboratory personnel and so forth are protected? I don’t believe these questions are adequately answered. In Mexico, the ICU in which one of the first patients died was infected. That is, all the other patients and 16 staff members also got the h1n1 virus. The hospital was closed. What measures are we taking to protect health care workers, as they are integral to protecting the health of the population?

  28. What is the projection in terms of the time the H1N1 virus may reach the Virgin Islands or Puerto Rico?

Leave A Reply