Healthcare Informatics: Analytics for Readmissions, Length of Stay, CMI, Core Measures, EBM



like never before in history health care in the United States is under pressure to improve its quality of service and at the same time reduce costs and extend access to care achieving these goals has many challenges but one thing is for sure they can't be achieved without information systems to provide a fact-based foundation upon which decision-making must rest and they should have by the Office of the National Coordinator to computerize patient data and make it shareable set the stage for making new levels of insight possible the advent of electronic medical records brings a wealth of new information to bear hello my name is John Frieda's and in just a moment I'll be showing you how to analyze healthcare performance using scalability experts new healthcare analytics platform this system is based on a relational data warehouse implemented using sequel server 2012 it uses a dimensional data model as described by ralph kimball in addition to the relational data warehouse the system makes use of high performance OLAP data cubes using sequel server analysis services data cubes help make more powerful interfaces for working with business intelligence data facilitating important interface features such as drillable hierarchies and graphical key performance indicators the data you'll be seeing is for impatience only I can't show you real customer data for confidentiality reasons so you'll be seeing simulated data none of the names patients or doctors refer to real people contact us if you'd like to analyze your data with a platform like this the system you're about to see has eight dashboards patient flows length of stay case mix index readmissions two core measures dashboards as well as procedure counts and admit source and discharge status analysis you may know that every patient who is admitted to a hospital will be assigned to a diagnostic related group or DRG based on his or her diagnosis and procedures as well as the severity of their case this is done by a program called a grouper there are approximately 750 DRGs each DRG is one of two types either medical or surgical and is further grouped into major diagnostic categories or MDC's you will see these elements throughout this demonstration of healthcare analytics every DRG has a relative weight assigned to it that indicates the amount of resources required to treat patients in that group as a compare to others for example a heart transplant without complication may have a value of say 14 a hand procedure for injury a 1 a normal newborn 0.15 the first dashboard we'll look at is the patient flow dashboard this dashboard deals with patient admits and discharges and patient census counts utilization information such as this is useful for computing occupancy rates planning and is often requested by the State Bureau of Health's statistics on this dashboard you'll see three data displays at the top is a bar graph showing how many patients were admitted and how many were discharged for each of the last 12 months the line graph beneath it is a monthly trendline that shows the inpatient days of care delivered over the same 12-month period of time inpatient days of care for a month is the sum of the daily patient census for that month the number of admits by medical condition bar chart at the bottom shows the number of patient admits for the given time frame broken out by major diagnostic category let's look at the top two displays these two displays respond to the control on the left of the dashboard titled major diagnostic category this control is called a slicer and it makes it possible to select a sub population of patients based on major diagnostic category or MDC shift and control clicking work so you can select any combination of MDC's you wish notice that when a selection is made both the bar graph and the line graph change to reflect the new data now let's look at the bar chart at the bottom major diagnostic categories are the chart categories the MDC's are numbered there are 25 of them the number of admits by medical condition chart starts at the MDC level and is drillable down to the diagnostic related group or DRG you can drill down by double-clicking on one of the bars drilling once brings you to the surgical and medical level doing it again gets you to the individual DRG drill up by right clicking on the bar label area and choose drill down drill up select the months you want using the month of admit slicer at the left at the very bottom of the dashboard are three slicers that you can use to sub select all three displays based on facility financial class or setting if you need other or different breakout factors please let us know now let's take a look at length of stay every inpatient stays in the hospital for a certain number of days and there are expectations concerning that length of stay based on the medical condition those expectations derive from statistics that are collected for each DRG by the Centers for Medicare and Medicaid Services the average length of stay is computed for each DRG and published annually hospitals one length of stays to be low for a variety of reasons if a given medical condition can be treated with a shorter stay it generally results in a lower total cost and many of the cases are paid on a fixed amount basis shorter stays increase hospital bed and resource availability also shorter stays mean a smaller risk of developing a hospital-acquired infection or other medical errors in many cases shorter lengths of stays are preferred by the patients and doing more recovery at home means the patient can spend more time with their families and loved ones the length of stay dashboard lets you check up on how you're doing this dashboard has two data presentations the top presentation which comes in two parts shows the number of cases total days of care delivered actual average length of stay target average length of stay and the percent difference between the target and the actual the target might be the national average or some other benchmark value chosen by management in this model the target is set to the national average but this can be set to reflect your company's goals the length of stay model calculates the target average length of stay value based on the mix of DRGs seen in the healthcare providers patient population and also calculates the observed actual length of stay based on the empirical data data-driven formatting is used to set the background color of the key metric you'll want to keep an eye on the percent difference between the target and actual green indicates a value that is comfortably in their desirable range yellow a marginal range and red undesirable this data is given for four quarters starting with the current quarter to date figures and working back three full borders the line beneath it is a monthly trendline that shows the percent difference between the target and the actual for each of the facilities as with patient flows there's a slicer that makes it possible to sub select populations of patients based on major diagnostic categories this makes it possible for you investigate where performance is good and where it's not so good MBC seven for instance is not doing as well as MDC eight if you explore your organization like this what do you think you might find could you use this information to identify pockets of excellence or problem areas now let's take a look at what's known as case mix index or CMI the case mix index is a measure of expected resource utilization more complex cases require greater amounts of resources and case mix gives a way to measure it the CFO of a healthcare company takes the CMI very seriously it is used in determining the hospital's budget if the hospital's actual CMI turns out to be less than what the finance department predicted the hospital may experience a loss in revenue and even seemingly small changes in CM I have a large effect on the hospital's bottom line I'll large CMI is a matter of prestige and may even be a driver of market share for a hospital here is how it's calculated identify your patient population get the DRG for each patient and add all the Associated relative weights now divide by the number of patients in your population a CMI can be computed for an entire Hospital based on its entire patient population or any group of patients scalability experts case-mix dashboard has two graphical data presentations in the top right corner of the page is a radar chart showing case-mix broken out by financial class for the three hospitals in the system each of the concentric Pentagon's represents a particular value of CMI and is labeled with that value points farther out towards the outer perimeter have a higher value read the values from the labels or position the cursor over the data point to get a pop up tip displaying the CMI the radar chart has a slicer for choosing the time frame beneath the radar chart is a bar chart showing CMI for the most recent three months grouped by financial class on this dashboard you have four slicers that apply to both graphical presentations major diagnostic category facility type of physician contract or employee and admission source use these slicers to define the patient population you are interested in and get an instant calculation of the case-mix index for it whenever you use the slicers to limit to a patient population of interest you will get a count of the number of patients included from the encounters selected table would you ever want to know which patients are included in your selection or which physicians are doing the work you can get this information easily just right-click on the number and go to additional actions where you will see the choices show admit and discharge information show all case summary information show patient information and show procedures choosing one of these causes the information to be downloaded to a spreadsheet note that even thousands of lines of data can be downloaded in a few seconds how would you like to use this information in your organization if your CMI drops could it be a sign of change in surgical or medical volumes if your hospital's CMI is lower than other hospitals in your area could it be a sign that your hospital is not capturing the complications and comorbidities associated with accounts and hired waited DRGs my tale ocmi denote DRG assignments that do not adequately reflect the resources used to treat Medicare patients you can only see these problems if you can see your CMI let's talk about readmissions a readmission occurs when a patient goes back into the hospital for the same or a similar condition within a prescribed period of time quality organizations such as the Joint Commission or Centers for Medicare and Medicaid Services have defined certain types of readmissions of particular interest 30 day readmission for acute myocardial infarction heart failure and pneumonia are three of them low readmission rates are viewed as a sign of higher quality medical performance this is because readmissions are often related to a problem in adequately resolved in a prior hospitalization or an unnecessary complication such as a hospital-acquired infection they can also be the result of inadequate management of the patient condition after discharge misunderstanding of the discharge instructions or lack of patient access to appropriate services or medications the hospital readmission reduction program established in 2012 by the Center for Medicare and Medicaid Services analyzes hospitals with high readmission rates let's go to the dashboard on top there are four 12-month trend lines displayed for the three readmission rates included in the hospital readmission reduction program other rates can be added to accommodate your needs there are slicers for facility type of physician and financial class beneath our two key performance indicator sections one based on diagnosis and the other on location of treatment key performance indicators let you see how you're doing at a glance using traffic light colors the indicators are drillable for example if you want to see how you're doing by individual a diagnosis code you can do so you significant regional variations in health care costs and efficacy have been soundly documented and is known all too well that there can be little or no correlation between spending and healthcare quality interest in increasing the value obtained from health care investments as stimulated efforts to develop the best science and apply it to health care delivery evidence-based medicine is an approach to medical practice that provides a framework for addressing healthcare policy challenges and helping the industry achieve its goals evidence-based medicine means using the best evidence to guide those decisions and is considered the gold standard in clinical practice core measures measure whether the best course of action as determined by established evidence-based medicine is being taken in the delivery of healthcare they come in sets by application area the acute myocardial infarction core measure set for example consists of eight quality measures the Joint Commission and Centers for Medicare and Medicaid Services collaborated to create this core measure set as well as others such as for heart failure and pneumonia performance in these measures factor in to accreditation and the results are publicly available top performers are given special mention scalability expert score measures dashboards will show you the degree to which the best course of action is being followed by your organization in this demonstration we will look at AMI core measures other areas would have similar displays there are separate dashboards for a facility perspective and a physician perspective now let's take a look the essential element of this dashboard is the eighth row report showing compliance for each of the eight core actions recommended for optimal care with heart attack patients statistics are compiled by month and the background color is set depending on the level of compliance there are slicers for selecting based on facility and type of physician beneath the reports are line graphs showing the trends for each of the core measures for 12-month period of time there is a second dashboard showing the same measures but broken out by physician these rows are sorted by the physician handling the greatest number of cases for the selected time frame on top the measures where a high value is desired are depicted with green bars the measures where a low value is desired with read these data are all fictionalized of course data from an actual firm would be far more interesting to look at there are other dashboards such as the admit source and discharge status display and procedure counts these comprise a corset many other displays can be developed depending on individual needs the calls of health care firms can only be met if there is quantitative information that shows decision-makers what is happening dimensional data models and business intelligence technology provide a foundation for that information in this presentation you saw several of scalability experts healthcare analytics platform dashboards you saw how business intelligence technology makes it possible to create interactive dashboards that reveal a performance in flexible ways thank you for watching contact us if you have questions or if you'd like to see your healthcare firm analyzed like this

4 Comments

  1. Hi John! You did an impressive presentation! Thanks for sharing this with us. I learned a lot. Please share more if you can. Noticed, you haven't posted any recent ones here. Did you open up any new channel?

  2. Hi John. Your video is very informative. Can you let me know how to download and get this tool running. I have some health care data that needs to be analysed.

  3. If you need data, this may help. I have not tried it, but take a look:
    https://www.hcup-us.ahrq.gov/db/nation/nis/nisdbdocumentation.jsp

  4. Hi John, I am really enjoying your series on healthcare dashboards. What is your source data for this? I would like to play around with building my own and am having a hard time finding a dummy dataset. Thank you!

Leave a Reply

Your email address will not be published. Required fields are marked *