At a Glance
Implementing an effective business intelligence (BI) system requires organizationwide preparation and education to allow for meaningful analysis of information. Hospital executives should take steps to ensure that:
- Staff entering data are proficient in how the data are to be used for decision making, and integration is based on clean data from primary sources of entry
- Managers have the business acumen required for effective data analysis
- Decision makers understand how multidimensional BI offers new ways of analysis that represent significant improvements over historical approaches using static reporting
Implementing a business intelligence system signals a hospital's readiness to embrace the future of data analysis for performance improvement, but the tool's true power is lost if the data entered are not reliable and decision makers use traditional practices for data analysis.
Hospitals can use business intelligence (BI) systems to improve quality of care, margins, employee and patient satisfaction, and operational and clinical efficiencies. Business intelligence tools give users the ability to correlate data elements for multidimensional macro- and microanalysis of information for effective strategic decision making. But a BI system is only as good as the data it contains and the skill of the analysts using it. To realize the full benefits of a BI tool, a hospital must first optimize the entering, integration, and analysis of data.
Data Reliability, Lineage, and Integration
Information from BI systems is only as good as the core data. Frequently, data are incorrectly entered because employees are using processes that have not evolved from legacy/sunsetted systems or the data have been extracted from systems with different data structures. In some instances, employees do not understand the importance of consistent data entry because they have not been instructed on how BI tools can impact the way data are used to make management decisions. Often, in such instances, the lack of instruction can lead to clerical staff populating the fields with default data when they do not really understand certain information.
Consider, for example, the case of a 350-bed community hospital that is experiencing operating room (OR) staff turnover. Current staff instruct new hires on the surgical documentation system use based on historical practices from use of an outdated, replaced system that required a work-around because of missing functionality. Drop-down selection lists have been added to over the years without maintaining standardization of the data sets. As a result, add-on surgical cases performed are being misclassified and entered with scheduled start times instead of being entered as converted add-on cases. This data entry skews analysis of late starts and case types making it impossible to accurately evaluate associated revenue.
By taking steps to ensure that case types are correctly assigned during data entry, hospital administrators can reliably determine lost revenue associated with add-on cases not performed because of late starts, as shown in the exhibit on page 101. Data can be viewed from an overall service down to individual surgeon or anesthesia level with the associated reasons directly contributing to the lost revenues. By ensuring consistent, accurate data entry at the primary source, administrators have the reliable BI data they need to work with surgeons and anesthesia to take the appropriate steps for enhancing efficiency and increase revenues.
In sum, to ensure valid data and integration, hospitals should take four broad steps.
Determine data lineage, flow, and integrity. It is important to establish that data are obtained and then shared from designated primary sources of entry. BI users can then be assured of having actionable information because accountability for accurate entry is maintained in one, primary system without possible conflicting data from various sources. Next, attention should be given to who will be using the data and the decisions they will be making. All staff-clerical through managerial-should be educated on how the data are to be used. Finally, IT and BI users should not flippantly dismiss bad numbers as a problem with the data without first performing microanalysis to determine whether practices are really that bad or bad data have been entered and need to be corrected before decisions can be made at a macro level.
Review processes for eliminating bad data input. With this step, the hospital should verify that data-entry processes have been optimized with the sunsetting or transition of software programs or with system upgrades. The hospital should ensure that staff are retrained with system changes and upgrades with a focus on decisions to be made with the data. Education policies for new hires need to be established to avoid staff being "trained as they go" on how data should be entered into systems.
Integrate and review data across the continuum of care and remove data silos. Primary data origins and flow should be mapped throughout the enterprise before being used in key performance indicators and important decision making. Executives need to review correlated data from multiple departments-for multiple components such as length of stay (LOS), cost, care delivery, and quality-rather than limiting the analysis to individual components being reviewed by different departments.
Assign ownership to clean up rather than simply scrub bad data. It is important to ascertain that all patients are properly coded-for example, all inpatients need DRGs assigned-and that chargemaster entries, past and present, are accurate (e.g., correct $0.00 charges in the system, or ensure the number of room/bed charges is consistent with the LOS). All systems should be evaluated to ensure standardization of drop-down selection lists, including historical data being analyzed, because in many organizations, lists have been added over the years without consideration of analytical data extraction.
Data Analysis: Instilling Business Acumen
Mnagers often intuitively know the issues affecting their departments, although they may lack the business experience to know how to quantify the issues with the right data elements to support a case for change.
Consider, for example, that a unit manager is faced with mandated staffing ratios that are not being supported by the acuity of patient conditions being seen, with patients undergoing vascular procedures representing the highest percentage of the census. Although a macro-level review by administration finds that vascular surgery is showing a positive operating margin for the facility, the manager knows there is a case management issue that has had an impact on the unit's staffing, but lacks the ability to take action.
With an improved understanding of how to correlate data in the BI tool, the unit manager views the operating margins and associated avoidable days by service line. The manager can also drill down to quantify which physicians have patients with an average length of stay (ALOS) that is substantially longer or shorter for the same DRG.
Avoidable days, and possible inappropriate admissions for one-day LOS, are linked with revenue that the manager can compare with acuity to isolate possible case management issues impacting department staffing. This knowledge empowers the manager to take action initiating discharge planning and/or palliative care with case management, at the start of the patient admissions, and tells the manager where to focus energies for evaluating the variances in the delivery of patient care for the same DRG.
The new amounts of data that come from introducing BI tools can be overwhelming. Managers may lack the training to effectively drill down into the data, and they may continue to base their decisions on historical practices. Managers' time is often consumed with "putting out fires," leaving little opportunity to efficiently analyze the information.
Consider, for example, the manager of a 20-room operating unit allocates surgical room times based upon block utilization percentages. Cardiovascular (CV) surgery currently has 80 percent block utilization and 23 percent of the case volume. Based on the hospital's historical precedent, CV service would receive preference of scheduled time.
The manager also notes that monthly case minutes and case costs are trending up. It is here that BI allows the manager to drill down effectively into the abundance of BI data. The manager now can determine that 46 of those cases had an average 57 minutes late start. Drilling down further, the manager finds that 20 of those cases were first cases of the day, and 17 were due to the surgeon being late or unavailable, causing a late ripple effect for all following cases.Drilling down further the manager can see that 40 percent of late case patients required double dosing of an antibiotic to maintain administration an hour before cut time, and 15 add-on cases could not be performed. The manager is now able to determine that CV service is not efficiently utilizing its 80 percent block time and is causing lost revenue.
Ultimately, to ensure that managers have the business acumen be able to perform effective data analyses using BI tools, hospitals will need to take the following steps.
Train decision makers to go beyond looking at just costs containment. Managers should receive in-depth training in how to analyze the correlations among processes, revenues, and drivers of cost and lost revenues for their areas.
Create dashboards and scorecards to help managers identify and prioritize the greatest opportunities for improved quality and profitability. Dashboards starting with top priorities for the area (e.g., DRGs by volume, quality outcomes, highest costs, readmits, and payer mix) can focus decision makers' energies so they can make better use of their time for analyzing issues according to greatest need or highest return of improvement.
Educate managers on macro to micro drill-down analysis. Managers should be trained to review the data with a focus on always answering the question "Why?"-which leads logically to a next level of analysis.
Adjust priorities as data turn into information. Sometimes drill-down analysis of new correlations of data shows decisions were being made based upon a symptom of an underlying problem that previously was not been able to be identified and now is a priority to solve additional issues.
Multifaceted Data Analysis: Transcending the Limitations of Historical Reporting
BI tools offer considerable advantages over historical use of static reports-that is, reports that do not allow complex drill downs and combination of data for cross-relational analysis. Users of BI tools can isolate inaccurate data, identify outliers, quickly view glaring issues to be addressed, and display trending simultaneously with point-in-time data. The capability of BI systems to present meaningful information can be severely limited, however, if they are designed only as an extension of historical reporting, where users try to match the BI numbers to their existing static reports. BI tools provide the opportunity for analysis of data in new, detailed ways that can help steer decision makers away from imprecise conclusions.
As a case example, imagine that a chief nursing officer (CNO) has historically benchmarked each unit's hours per patient day (HPPD) and evaluated staffing ratio reports calculated from pay-period aggregate data. Based on accumulated two-week calculations, the reports and the BI display reflect adequate ratios, although nurses are complaining of excess workload. The conclusions from using the BI based on historical reporting practices do not accurately reflect daily processes because the information is retrospective and accumulative for two weeks. Staffing ratios calculated from aggregated numbers over two weeks cannot reflect the staffing compared with patient-flow volumes throughout individual days.
The CNO is able to optimize her staffing using a redesigned BI that can provide a view of patient flow throughout the day (see the exhibit above)by proactively assigning staffing patterns to projected volumes with associated patient acuity. Daily ratios can then be aggregated for weekly, day-of-the-week, pay period, monthly, quarterly, and yearly measures along with being associated with CPPD, productivity with acuity, quality and outcomes, and profitability. Analyzing information with this new perspective enables unit managers to create new shifts to meet patient flow instead of straight eight- or 12-hour shifts, supporting volume fluxes from changing physician practices, reducing their CPPD, and improving nurse satisfaction and quality of care.
Opportunities for improvement may be lost if users quickly dismiss outliers as aberrant or insignificant data. Outliers may bring to light internal causes of inaccurate data (e.g., an implant cost being accidentally entered as $12,561,256 when it was $1,256), or reflect high-cost, inefficient clinical processes.
Consider, for example, the situation depicted in the exhibit above where BI has been used to detect an instance in which a physician has ordered two CT scans for the same patient and same condition in 24 hours. Being able to hone in on outlier activity helps isolate process issues causing unnecessarily high costs, or processes causing inaccurate data entry (e.g., improper coding of patient DRG).
BI tools enable managers to make decisions with multidimensional analysis by seeing a whole picture through individual components across business silos. Correlating individual data elements while incorporating trending with benchmarking enables executives to assess where their organization is operationally, compared with where they want it to be, and helps them determine how well they are progressing toward that goal.
As another example, imagine that administration and physicians want to review the highest volume surgical procedure, coronary artery bypass graft (CABG). Individual managers perform historical analysis of surgical cases at unit levels.
Administration reviews overall profit and loss of surgical services, the quality manager monitors quality indicators, and the operating room manager assesses case minutes and costs along with benchmarks. Original BI is created for individual departments. With reformatted dashboards, decision makers can now meaningfully review information across the spectrum of care-intra-operative, critical care, LOS, quality, and patient satisfaction-to assess performance with an enterprise view. The results of this analysis are shown in the exhibit on page 106.
By implementing a BI system that allows the hospital staff to assess current operations with trending, the organization can also begin to use the data as a baseline for modeling changes for strategic planning.
To ensure that BI tools are being used for analysis of data in a way that transcends the limitations of historical reporting, hospitals should take the following steps.
Educate management on the inherent differences between dynamic analysis using BI and static historical reporting. Managers should understand the specific ways BI dashboards offer the ability to expand analysis for more informed decision making. Educating managers on the capabilities of BI analysis should be part of a larger effort to increase awareness of the business of health care as it continues to evolve.
Use benchmarks and targets in BI tools for gap analysis. BI tools are more effective when used to isolate whether data indicate a trend or just isolated incidents. Such analyses can help BI users focus improvement efforts where the greatest opportunity lies. (For an example of a gap analysis, go to "Using BI for Trending Analysis.")
Model for the future. After using BI for the analysis of existing processes, the next step is to use the information for planning by modeling what the hospital can expect in the near and long-term future.
Putting Data to the Best Possible Use
BI provides the benefit of seeing the whole picture through the individual components across business silos. The ability to correlate data throughout the enterprise and incorporate trending with benchmarking enables hospitals to see where they are operationally, compared with where they want to be, thereby helping them to progress toward their goals.
As the healthcare industry undergoes dramatic changes, with the unfolding healthcare reform legislation, the nation's hospitals have recognized the importance of focusing on implementing electronic information systems. BI tools that are created and used effectively-with their ability to take clean, reliable data and provide innovative analysis-are powerful means for improving practices and promoting more-informed decision making.
Rose Rohloff is executive director, The PARC Group, LLC, Chicago (email@example.com).
Publication Date: Monday, May 02, 2011