Rose RohloffMany healthcare organizations are focusing on developing their business intelligence (BI) and analytics. The resulting information has value, however, only when it emanates from good data and drives ongoing positive, sustainable change. Regardless of the data tool, it is only as effective as how the tool is actually used.

This idea calls to mind an old anecdote about making a pot roast: 

A little girl asks her mother why she cuts inches off the ends of the pot roast before cooking. The mother responds, “Because that is the way my mother cooked and the way her mother cooked before her.” When the little girl asks her great-grandmother why, the venerable old woman replies, “My pans were always too small to fit a whole pot roast.”  

The story points out two distinct lessons: First, bad practices can be adopted if there is no understanding of the rational for them; and second, it’s important to adopt new methods for addressing changed environments, tools, and issues. The whole point of the story is that the mother and grandmother—the key decision makers—had no idea waste was even an issue; it never occurred to them to question the way they were performing tasks for an efficient, high-quality outcome. 

Applying these premises to the healthcare environment, there is a need to look closely and question the underlying assumptions for decision-making based on retrospective data, along with the rationale for existing practices and whether they still make sense in the current environment. Three primary elements are essential to ensure that this process results in positive, data-based change.   

Clean data. When organizations look at claims data with only a “10,000 foot" view, they run the risk of reacting to retrospective data that has not been analyzed for reliability at a very basic level. The bigger the data, the greater the risk that administration may become removed from departmental clinical and operational practices and, more important, the actual processes of data entry. Consider, for example, a case of hospital executives who want to assess claims data for patients with congestive heart failure (CHF). In this instance, data correlated at a department level point to insulin being the highest-volume medication given. Yet no patient given insulin has an assigned ICD code for diabetes as a comorbidity, suggesting the data process should be evaluated for coding inaccuracies, and the CHF priority should be assessed in conjunction with diabetes. 

New methodologies. Business practices created in the 1950s-60s may not address new healthcare environments. Times have changed and there is continued change on the horizon. Nursing workflow, for example, has evolved, and many hospitals staff historical three-eight-hour shifts without overlapping shifts to meet workflow volumes, especially when a high percentage admissions occur though the emergency department versus direct physician admits, multiple transfers occur between units during the day, and discharges and patient care change orders occur later in the afternoon. 

Another example is performing cost cutting by reducing FTEs to benchmarks instead of evaluating data with an eye to streamline unit admissions and instituting palliative care processes across inpatient, outpatient and community continuum:  to reduce care workload, length of stay (LOS), readmissions or to eliminate multiple reclassifications of patients from inpatient to outpatient and again to inpatient.

Training. To achieve sustained transformation, hospitals need to initially undertake in-depth analysis and BI creation to determine root causes of issues. Hospitals should consider outsourcing this process to ease the burden on hospital personnel so they can focus on change management. Then, decision-makers can be mentored regarding how data should be correlated across department silos and reporting structures, and how data sets should be analyzed to understand "whys." BI based only on historical Crystal Reports or static dashboards may not support meaningful analysis. As an example, administrators using case mix index (CMI) as a key performance indicator for patient levels and cost adjustment cannot begin to truly understand the actual correlation between costs, patient intensities, and quality care delivery if they do not also evaluate relative weights at a department level with patient acuity, staffing ratios, LOS, and outcomes.

Publication Date: Monday, March 03, 2014