Originally published in Becker’s Hospital Review, September 2021.
To reduce hospital length of stay (LOS) became a critical priority during COVID-19, and continues to be one as surges continue. Even in “peacetime”, however, capitated reimbursement levels and the need to decrease hospital-acquired conditions made LOS reduction key for health systems. While efforts to reduce hospital length of stay often focus on specific disease classes, there are general operational factors that are more straightforward to address.
The economic driver: As patient access is inversely proportional to length of stay, the impact of even an incremental change in can be quite significant. If a 300-bed hospital with an average LOS of four days could reduce the average LOS by 5%, or five hours, they could treat over 1,350 more patients per year. That’s also millions of additional dollars of income generated from the same fixed capacity.
The basic demand-supply problem: In most hospitals, daily demand for beds usually swells before the supply of right beds opens up. In other words, the arrival patterns of patients who need a bed are “out of phase” with the departure patterns of patients who are leaving the hospital. At most hospitals, morning surgeries and overnight ED arrivals create a need for beds in the morning, while typical hospital discharge processes make beds available later in the day.
Bed capacity must also be divided into “units,” as determined by the level of care and skilled staffing required. This adds up to a limited availability of inpatient beds on a recurring basis.
Seven ways to reduce length of stay in hospitals
In working with leading hospitals across the country, we have identified seven factors that, if addressed with precision and the right data and predictive analytics, can reduce length of stay. Several of these will require cultural change or “buy-in” from clinicians and staff, but the rewards of adopting these solutions will be clear to those involved.
- Rightsize unit capacity: Incorrectly assessing required unit size leads to a higher percentage of patients not being placed into their target unit, longer wait times to get patients out of the ED, and more complexity in decision making.
Solution: The correct number of required beds can be computed for a given unit by analyzing historical patient volumes by service. For example, medicine units accept general medicine patients in addition to the specialty patients for whom the unit is intended. To reduce length of stay hospitals can downsize specialty units and open a larger (“virtual”) general medicine unit. This virtual unit can still consist of several sizable subunits. This approach leads to simpler patient placement and makes it easier to co-locate patients assigned to a given clinical team.
- Ensure proper patient placement: Incorrectly placed patients end up at the wrong level of care and/or being treated by clinical staff who do not specialize in their condition. This can lead to slower recovery times and even clinical complications. Getting the right patient to the right bed at the right time can reduce LOS.
Solution: Use an algorithmic approach to patient placement to balance load across units and minimize the number of patients not being placed in target units.
- Smooth patient flow from the operating room (OR): Flow of patients from the OR is significant and can cause spikes in the i-patient census, but is more controllable than census contribution from the ED. Optimizing the elective surgery schedule with respect to recovery time yields a flatter in-patient census.
Solution: Forecasting the volume and case mix of surgeries together with the associated recovery time per case allows hospitals to apply lean manufacturing techniques, combined with an optimization engine to build surgery templates that avoid spikes in the downstream census.
- Use predictive discharge planning to focus case teams: At the end of a patient’s stay there are many avoidable delays in discharge. These delays could be avoided if case managers were alerted to the problem earlier in the patient’s stay.
Solution: Use a machine learning framework to identify key patient attributes that indicate possible difficulty in discharge. This risk model is used as an early warning signal to alert case managers to at-risk patients.
Note: Building a high-quality machine learning model requires a rich data set. Avoidable delays in discharge need to be accurately tagged in the EHR for a number of months to establish a viable training set.
- Reduce delays in admission from the ED: The process of admitting patients from the ED is time-intensive, highly variable and frequently leads to delays. Visibility tools can allow physicians to identify delays in the system and react accordingly.
Solution: Visibility into the ED and inpatient units allows for better decision making and accountability for physicians, RNs, transport staff and housekeeping. A summary view of ED activity allows users to quickly view the number of patients waiting for a bed, average boarding times, etc. A unit view displays the number of occupied, dirty, and available beds.
- Address provider workflow: Most hospitals employ a general approach to physician rounding with the activity usually occurring in the late morning or early afternoon. This timing can lead to a missed opportunity to discharge patients early in the morning.
Solution: Take simple measures such as maintaining a list of patients that can likely be discharged early and have a team round on these patients early the following morning.
- Make some discharge procedures outpatient: In some cases it is more cost-effective to schedule certain procedures as outpatient rather than inpatient, particularly if the patient is local. Solution: Identify candidate procedures and patients that can be easily transferred to outpatient. In some cases it may be more desirable to move the patient to a local hotel and report back in the morning for an outpatient procedure.
For more support to reduce hospital length of stay, including case studies, visit our inpatient beds Resource page.