Why length of stay is a poor measure of hospital performance

Nearly every hospital measures and manages its performance based on inpatient length of stay (LOS). But LOS as an absolute measure really doesn't provide much useful information for managing a hospital. Here is why:

 

Lack of clear-cut root causes undermines accountability - Who is ultimately accountable for patient length of stay? The nurse? The doctor? The lab? The hospital administrator?

 

There are many reasons patients are in the hospital and, in an ideal world, objective clinical criteria would govern census. However, in the real world, there are statutory, socio-economic, familial, emotional, and even political drivers that affect a patient's status. It is nearly impossible to objectively determine root causes for excessive LOS because each patient's circumstances are so different, making root causes tough to classify and hard to analyze for better management.

 

In the absence of clear-cut root causes, no one member of the hospital staff can be held accountable for controlling LOS. As a result, lack of individual accountability makes everyone accountable; and, when everyone is accountable, no one is accountable.

 

Imprecise Measurement - Hospitals measure length of stay by the day. It doesn't matter to them whether a patient is admitted at 10 AM or 10 PM; any admission before midnight counts as a day. Since when does two hours constitute a day? Measuring length of stay this way is imprecise and dilutes the power of the metric. Wouldn't it make more sense to measure length of stay, with its attendant costs, on a rolling 24-hour basis?

 

That practice is already at work with observation patients. As an observation patient approaches the 24-hour target, doctors, nurses, and case managers are all actively managing the patient to either admit or discharge regardless the time of day. In this case, the target is met because there is a clear, unambiguous goal and patients on this 24-hour clock tend to be (decisioned? or have their status resolved?) discharged in 24 hours. LOS would be far more meaningful if it were less ambiguous and matched to specific conditions for better measurement, just like the observational patients.

 

Hospital performance can improve with a simple change in how LOS is calculated, which would also make it a better measurement on which to base management changes.

 

Conflicting goals - There are many conflicting goals in every hospital. (The conflict between hospital reimbursement and physician reimbursement is the stuff of legend.) While the process is certainly imperfect, the simple fact is that, hospitals are paid for utilization of their assets. So, up to a certain point, inpatient LOS is a good measurement for the hospital.

 

But there is an optimum length of stay that is as unique to each hospital as its geography, service model, patient mix, charge master, and contracts. Using one standard LOS to measure hospitals across a disparate universe just makes no sense (See example below).

 

Not Normalized - In benchmarking LOS across the nation, there is shocking disparity between the different regions of the USA. Using CMS data, we compared two hospitals that are nearly identical, one in Cleveland, OH, the other in Queens, NY. Both are short-term acute-care facilities, with 250 beds, proprietary corporation control, and a case mix index of 1.25.

 

When we compared the average LOS for DRG 127, Heart Failure & Shock (CHF), a perennial high volume diagnosis, here is what we found: The Ohio hospital saw 237 cases with an average LOS of 5.16 days. The New York hospital saw 150 cases with an average LOS of 7.86 days. What explains this difference? Who cares? CHF is CHF. There is no clinical reason for treatment requiring 2.5 more days in New York than in Ohio, but there may be numerous other reasons that cannot be determined by single-mindedly looking at length of stay.

 

Good measures are relevant measures - Good measures are clear, objective, assignable and relevant. They are not open to interpretation. (An automobile is going 55 MPH or it is not.) Good measures are actionable, which means targets can be set, actions assigned, and outcomes evaluated. Good measures can be used to initiate and sustain intervention or process improvement.

 

Hospital performance measured solely or primarily by LOS is meaningless. Want a simple, easy, and very effective way to measure hospital performance? Email me. <click here>

 

 

Note: Data used is from the Medicare Hospital Market Service Area File which is updated annually by CMS. The file includes Medicare discharges, patient days, and gross charges by ZIP code for each hospital. Data are based on 100% of all Medicare fee-for-service claims during a calendar year. Go to CMS.gov for more info.