The project team consisted of a wide variety of clinical and support personnel. Its members represented outcomes management, nursing, respiratory, professional practice, medical staff, and finance. The team selected the Six Sigma DMAIC process as its improvement method. Using Juran’s Pareto Analysis, VOCs, cause-effect diagrams, and FMEA, the team developed improvement strategies and was able to reach its goal of a 90% compliance rate with heart failure discharge instruction.
A not-for-profit healthcare system found that adherence to clinical quality ob- served metrics for inpatient heart failure discharge instruction (HF-1) was consistently below national standards. For FY 2006, the average observed rate of compliance was 45.3%. Noncompliance could result in penalties with reimbursements from the Centers for Medicare and Medicaid Services (CMS), additional costs because of the potential of harmful events, and a decrease in patient satisfaction.
The project Y was average length of stay (ALOS), measured in days for all adult inpatients coded with DRG 1272 (heart failure and shock). This included patients entering the facility through the Emergency Department (E.D.), direct admits from a physician, or patients arriving from another healthcare organization. The beginning boundary for the project was the time the admission order was logged. The ending boundary was when the patient was discharged from the bed and left the floor. Excluded from the project was the patient’s stay in the E.D., and observation patients (held
To better understand the current process, length of stay data were gathered and the process was characterized in terms of the major workflows. Over the preceding year, 57% of DRG 127 patients had a length of stay less than or equal to the target of 4.1 days. This had an associated baseline sigma level of 1.68 and cost of poor quality of $1,001,000 annually. A SIPOC high-level process map and detailed process maps were created for the following workflows: E.D., inpatient flow, floor arrival, critical care transfer, ongoing assessment, and discharge. This effort provided all team members with a deeper under- standing of the overall process.
After analyzing these process maps, the team brainstormed potential causes of extended length of stay and organized
them into possible cause categories. A cause-effect diagram was constructed for each possible cause category. Using the diagram, the team was able to further identify possible root causes. Subject matter experts organized these theories by common groupings. As a result 25 possible root causes were identi- fied. To narrow the group, the team prioritized the root causes based on the degree of expected impact on length of stay, and the degree of control the team had over them.
A detailed data-collection plan was created to document data sources, sample sizes, data analysis tools, and responsible parties for each of the possible root causes. In most cases, data were available in electronic logs, but new data had to be collected for others. Graphical analysis tools used included box plots, scatter plots, Pareto charts, and bar charts. In addition to descriptive statistics (average, median, standard deviation), statistical analysis tools including non- parametric hypothesis tests, regression, and Chi-square analysis were used. Some hypothesized root causes were:
- Inpatient holding process was not standarized
- Socially-related discharge needs assessments were not comprehensive
- Socially-related discharge needs were not identified early in admission
Rigorous analysis of the data revealed the vital few Xs driving extended length of stay. Some included:
- Inpatient holding process not standardized
- CHF standard orders were not used (no parameters)
- There was delay between the discharge order and the time the patient leaves the floor.
The team brainstormed possible solution strategies that would address each of the vital few Xs causing extended length of stay for congestive heart failure patients. Some included:
- Patient holding: Develop ways to get the patient out of the E.D. faster; improve and expedite care for patients that are held.
- CHF Standard Orders: Reduce variation in practices by developing an order set and interdisciplinary pathway and provide for the education of physicians and hospital staff in their use.
- Delay in DC orders to leave floor: Develop a better communication process in relationship to the anticipated discharge date and the needs starting at day one of admission.
Additional detailed solutions were developed to enable these strategies. These solutions were rated against
13 specific performance and business criteria using a Pugh Concept Selection Matrix. The selected solution was piloted over a four-week period. During the pilot, the team collected data on length of stay and key process variables to ensure individual components of the overall solution were properly implemented and effective.
The pilot was successful in reducing length of stay to an average of 2.6 days for patients with hospitalists attending, with 91% of patients discharged within 4.1 days of admit. The team documented process changes on the original process maps and developed an implementation plan to formally roll out the new process.
A control plan was developed to ensure the improvements and gains would be sustained over the long term. Key elements included the control subjects (length of stay, readmission rate, and proven Xs), measurements (sensor, frequency, sample size), and actions (criteria for taking action, responsibilities).
Results are being monitored as an ongoing activity. To date, the ALOS has been reduced 31%, from 5.18 days to 3.6 days and continues to drop towards the level shown possible in the pilot. Compared to the baseline of 57% of patients discharged within 4.1 days, more than 80% are now discharged within 4.1 days. Readmission rates are being monitored to ensure there is no increase.