iSixSigma

Hospital Los

Viewing 14 posts - 1 through 14 (of 14 total)
  • Author
    Posts
  • #25705

    Lorax
    Participant

    Folks,
     
    We’ve been finding LOS or squishing out an average so ALOS to be a problematic measure.
    There seems to be many, many things which can effect it and add variation to an already very noisy data stream – patient acuity, co morbidities, diagnosis(es)…
    Conservable days is looking like a cleaner and more usable number.
    Thoughts?
    Lorax
     

    0
    #62176

    DLW
    Participant

    In the discipline of Strategy Execution & Management, we always
    stress the importance of metrics being chosen such that they
    represent positive change. From that perspective, a good metric in
    your case may be REDUCTION in LOS or ALOS. You can express it
    either as number of days or as a percentage.If, for example, ALOS is 10 days and you want to reduce it to 8
    days, then your target for such a metric would be a REDUCTION of
    2 days, or 20%. When you have achieved an ALOS of 9 days, you will
    have realized 50% of your target (1 of the 2 days; 10 of the 20%).In this way, you can focus on some across-the-board target for
    improvement without having to consider all the variables that make
    up the base metric.It can take a little getting used to expressing metrics this way, but
    the huge advantage is that you consistently are focused on the
    degree of improvement, rather than on the base metric which in
    and of itself may be of the “smaller is better” type.
    DLW – BPEX

    0
    #62177

    DLW
    Participant

    One other thought:
    I would expect that LOS is a Systems-level metric; i.e., it results
    from all the various Process/Tactical-level things that roll up into
    it. As a result, LOS most likely is not a good choice for a Project
    metric. Instead, project metrics should be those things that can
    show improvement in a particular process such as Admissions,
    Diagnosis, Radiology, Discharge, etc.
    (You may want to view my recent post on the main site under “Six
    Sigma Project on Business Metric”. It has a bit more verbiage.)
    DLW – BPEX

    0
    #62178

    Lorax
    Participant

    Fair point. Just look at the deltas away from the institution’s ALOS.
    The trouble comes when you try to do a reality check and let’s say (cue made up numbers) the ALOS is 3 days.
    Most people will look at this and say something along the lines of the last patient they saw  was at 30 days and that 3 could not possibly be correct. Then they will ponder it for a while and end up with the thought that 3 could be right if you took into account all patients.
    The trouble is that there is an immense amount of variation around that 3. Some patients have a far higher LOS (ALC?) and some far lower.
    I’m thinking that it might be so noisy that it is of no use for process improvement and that something like Conservable Days, because it takes into consideration CMGs and patient complexity and all, makes a better indicator – its variance indicating the effect of good things done to the process and the effect of the not-so-good too.
     

    0
    #62179

    DLW
    Participant

    I would tend to agree with you. LOS is (I believe) a JCAHO metric, yet
    in my estimation it does not make an effective process-improvement
    metric for the reasons you cite. It belongs at the next-higher level,
    where it is the composite effect of many things below it.
    DLW – BPEX

    0
    #62186

    BritW
    Participant

    I agree that conservable days (we call them avoidable days) is a good measure for improvement, but anyone who collects it knows that the value you get is only as good as the reporting – just like medication variances.  We do collect it – but be a little wary of the #.
    LOS is not a TJC measure – it is used primarily as a comparison against the Medicare DRG structure – that pays most American Hospitals.  Let’s say you have a pnemonia DRG is 3.75 days but your pneumonia patients are statying 4.5 days.  The payment for the service will be for the 3.5 days so you provide .75 days of care free – costly. So the number is a valid one to examine.  It does have a lot of variables, but those can be segregated and identified.  You don’t have to know all of them to get the improvement ball rolling.
    My suggestion – compare your hospital LOS by DRG.  Find the ones that are most troublesome (pareto).  Take a few and break down the key items that make up LOS – you can use a team to identify the 5-10 most troublesome issues, then measure those instead of identifying all 100+ variables.  My guess is you know the issues already – e.g., isolation when it isn’t needed, doctors rounding too late, patients eating/drinking when they shouldn’t before a procedure, patient placement and discharge procedures, lab and radiology turnaround times, service availability over the weekend (ECHO dictation), etc.  Start with these.
    You can attach a hospital-wide $ value to every 1/10 day removed from ALOS – don’t discard it because there are too many variables – just deal with the variables you know are already the problem.  Every 1/10 day we save saves the hospital $220,000 per year.

    0
    #62203

    not a doc
    Participant

    I’ve just completed an ALOS project to reducing NICU stays.  It was a challenging project to complete due to the many variables stated in this thread, but it’s a very viable and measurable stat for any medical operation (hospital or insurance).  Be prepared to become very familiar with Logistic Regression tests.  Needless-to-say I wouldn’t shy away from the challenge.  It will likely take you a while to move the needle, but you’ll get significant dollars and a much improved process.  Hope that helps you. 

    0
    #62204

    Mundorff
    Member

    Rather than DRG, try APR-DRG if your hospital uses that system.  APR-DRG has another level, severity of illness, which can/should be used as an adjuster.  The APR-DRG documentation goes into more detail about that, but basically it is analogous to a chi-square where the actual value is compared to an expected value.  Or you can stratify by individual SOI.
    The new Medicare DRGs have more severity built into them, but they are still not as robust as APR-DRGs, particularly for pediatrics since Medicare is not a big payor in pediatric medicine.
    Our hospital system uses this methodology all the time.  At my individual hospital, a pediatric specialty hospital, we use it frequently.  Two projects I have recently worked on regarding LOS are appendectomy and viral respiratory illness.

    0
    #62206

    Robert Butler
    Participant

       When you say “Logistic Regression tests” do you mean running logistic regression?  If so, and assuming LOS is your outcome, why are you running logistic regression on something like LOS or ALOS?   The LOS data I’ve analyzed is usually measured in either days or hours – either way it’s a continuous enough measure that it is ameniable to standard regression methods.  When building the predictive model we do have categorical predictor variables such as gender or illness severity but these are X’s and are coded in the usual fashion.

    0
    #62207

    keyes
    Participant

    I support this form of the response, however, structuring along the theme of a Value Stream Map.
    LOS could be considered the overall Cycle Time and each project that is run to improve the Cycle Time would have its own specific metrics. I am not aware of too many projects/kaizen events tackling the overall Cycle Time as they would be trying to improve various steps along the Value Stream to have an ultimate effect on the overall Cycle Time.

    0
    #62208

    BritW
    Participant

    We used standard regression methods to develop predictor models based on top 10 DRGs.  I found no real need to get into logistic regression for the reasons you mentioned.

    0
    #62212

    not a doc
    Participant

    Regarding the comments why logistic regression – quite simple – we were measuring the days to birth outcome:  one discrete, one continuous (I won’t get into details other than that on a discussion board, if you don’t mind).  Rest assured – I ran the correct test for the data.  If regression would have worked I would have used that test – it’s easier. 
    Either way – the point being is that LOS or ALOS is definitely a viable metric.  It may be more difficult and you’ll never get 100% agreement on any question/issue/project idea, but it’s viable none-the-less. 

    0
    #62214

    Robert Butler
    Participant

        The reason for my question concerning the choice of logistic regression was not to find fault but just to clarify.  So it sound like the binary outcome was live birth yes/no and LOS was one of the X’s. If this is true then this differs from the original post where LOS is the outcome and you are looking for significant correlates that might increase/reduce that value. Just a word concerning terminology – if you are going to write this up for a journal logistic and OLS regression are not tests they are means of data analysis. 

    0
    #62215

    not a doc
    Participant

    Both binary and I had a nominal.  I just didn’t want to get too much in the details on a discussion board – it was a 2 year project and can’t really be discussed well like this (too much detail, you know).  Don’t want to get in semantics on test versus data analysis.  I’m just a simple BB who wants to share with the original poster that there are ways to measure LOS/ALOS.  Of course in any project there’s more than one way to skin a cat – that’s what makes it fun. 

    0
Viewing 14 posts - 1 through 14 (of 14 total)

The forum ‘Healthcare’ is closed to new topics and replies.