Pharmaceutical cost containment and appropriate drug utilization are essential for achieving successful financial and clinical outcomes in today’s health care landscape. To those ends, automation now inundates the pharmacy and is infiltrating even more distant areas, such as outpatient pharmacies and operating rooms.
But automation is not the only means to an optimized pharmacy. Properly cultivated and utilized data can present a wealth of opportunities for driving costs down and improving patient outcomes. Fortunately, multiple data sources, databases, and products are available that are designed to assist pharmacy personnel in accomplishing these goals.
Two such data products exist at the University HealthSystem Consortium (UHC). Together they provide members with access to detailed information necessary to identify optimization opportunities. UHC is an alliance of the nation’s leading nonprofit academic medical centers. Formed in 1984 and based in Chicago, UHC fosters collaboration among its 120 academic medical centers and 310 affiliated hospital members through its renowned programs and services in the areas of comparative data and analytics, performance improvement, supply chain management, strategic research, and public policy. Mining and manipulating the information in these databases is one of the primary ways UHC is able to help its members achieve excellence in quality, safety, and cost-effectiveness.
Comparative Database Tools
The two databases that work in tandem at UHC are referred to as the Clinical Data Base/Resource Manager (CDB/RM). Both are comparative database products that house discharge and line-item data from UHC members and affiliate hospitals. The CDB allows organizations to compare clinical outcome performance both within a facility and against that of other hospitals in the consortium. The information in the CDB is primarily that which is found on a typical administrative UB-04 billing form and includes such items as:
UHC’s risk models provide expected length of stay (LOS), direct case cost, and mortality predictions for each inpatient discharge, allowing hospitals to see not only their own performance, but also to benchmark that performance against the outcomes expected based on the information available about the patient.
The RM database builds upon the patient-level data housed within the CDB. It provides comparative utilization information for select resource categories (eg, lab, imaging, pharmaceuticals, etc) by obtaining detailed line-item transaction information from participating member hospitals. This information can be extremely useful in understanding practice variation patterns among groups and in determining which resources are experiencing high utilization.
Using the RM database, members can also review the utilization patterns for their high-cost drugs through a comparison of the percent of a clinical population that received a particular medication, the duration of therapy, and an estimated drug cost per case that is based on the World Health Organization’s defined daily dose methodology (the assumed average maintenance dose per day for a drug used for its main indication in an adult).1 Together, these databases are flexible and powerful tools that support performance and operationalimprovement.
Using Data to Advance a Pharmacy Initiative
Taking a methodological approach to administrative data can help uncover opportunities to optimize resource utilization. The following is a sequential method for using administrative data to help advance pharmacy initiatives:
1. Identify a topic, question, problem, or hypothesis to analyze. The first step in a successful database analysis is to clearly and concisely formulate the topic or question you want to analyze. Often, this is the most difficult, yet most essential, step in the analytical process. When focusing on a topic, evaluate the time needed to complete the study and identify any limitations, along with the inclusion and exclusion criteria. In addition, define your clinical population through the effective use of existing coding systems.
Examples of appropriate topics and questions include:
2. Determine what data to collect and how. Having a template that outlines the necessary data elements can help provide structure and ensure that all relevant metrics are gathered. At this stage, it is important to determine if administrative data are, in fact, the most effective source of information, and whether supplemental metrics will be needed from other clinical decision support systems, such as the health record.
Administrative data—which are generated anytime a patient has an encounter with a provider or facility and reimbursement for a service is sought—consist of demographic information, assigned diagnoses, performed procedures, clinical resources used, payer information, physicians associated with the case, etc. Patient attributes found in the medical record, such as laboratory values, vital signs, height, and weight are usually not considered administrative, but if they can be accessed easily, they can provide substantial support to your analysis. To determine what information is currently available (whether it is internally collected and stored within a data warehouse or obtained from a third-party partner), talk with your decision support/information technology department.2
3. Familiarize yourself with the coding system to identify populations, diseases, procedures, etc. Administrative databases leverage coded material (typically from the medical records department) in patient files. Therefore, when working with these data, it is important to have a basic understanding of your organization’s coding practices. Currently, most of the information found in administrative databases in the US use International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) codes to document the diagnoses and procedures associated with a patient encounter, although health care providers are bracing to adopt the next and even more delineated version of these codes (ICD-10-CM) on October 1, 2015. Terms commonly used in these coding sets include:
A. Principal Diagnosis. This is the condition established to be chiefly responsible for the admission of a patient to the hospital for care. It may differ from the physician’s admitting diagnosis and may not be the most important or serious diagnosis identified during the hospital stay. Coders must follow explicit rules to accurately determine the principal diagnosis and must rely on what is documented by the treating physicians.
B. Other Diagnoses (Secondary Diagnoses). Secondary diagnoses consist of all conditions that coexist at the time of admission or that develop subsequently and affect patient care in terms of requiring at least one of the following:
Every secondary diagnosis must be assigned a Present on Admissions (POA) Indicator to designate if the condition or complication was present at the time the patient was admitted. The Centers for Medicare and Medicaid Services (CMS) use the POA indicator to adjust payment for certain preventable hospital-acquired conditions/complications.
C. Principal Procedure. A principal procedure is one performed for definitive treatment, rather than for diagnostic or exploratory purposes, or one that is necessary to address a complication. If two procedures appear to be principal, the one most related to the principal diagnosis is selected.
D. Significant Procedure. All other significant procedures should be reported. Procedures are deemed significant if they are:
E. MS-DRGs. Developed by CMS as a prospective payment reimbursement methodology for acute care inpatient hospital stays, MS-DRGs primarily use ICD-9-CM diagnosis and procedure codes and, in some cases, discharge status, to classify inpatients into clinically cohesive groups with similar resource consumption and LOS; they are classified as being medical or surgical. Medicare further identifies complications and comorbidities associated with MS-DRGs as either a Major Complication or Comorbidity (MCC) or a Complication or Comorbidity (CC). A complication is defined as a condition that arises after admission to the hospital that potentially can modify the course of the patient’s illness or the medical care required. Conversely, a comorbidity is a pre-existing condition.
Because coding rules and guidelines change over time, it is important to work with your facility’s coding office to ensure you are using the most up-to-date information on codes and MS-DRG assignments. Once you have a basic understanding of coding and have determined the topic and appropriate data to collect, it is time to consider how the database can be used to analyze drug utilization.
4. Determine if comparative data are needed. The effective utilization of both internal and external benchmark data enables an organization to elevate its performance. Internal benchmarking necessitates that a facility examine itself in order to identify local variation in practice.3 For example, a hospital can review and compare outcomes, resource utilization, and cost across similar physician specialties. The following are sample scenarios involving benchmarking:
Example Scenario 1: A review of the utilization of thrombin products (5,000-unit vials versus 20,000-unit vials) is proposed because of their high cost and anecdotal reports of variability in use. Available data allow for the breakdown of internal utilization by service line, MS-DRG, procedure, and physician. A complete analysis reveals that certain clinicians at your organization are using larger vials than their peers for similar cases. Thus, a recommendation is made to switch from 20,000-unit vials to 5,000-unit vials in an effort to reduce waste, which results in an estimated an-nual savings of $45,000. By reviewing internal data, organizations are able to highlight best performers and begin to address variation between clinicians who share the same culture and systems. The results can then serve as a baseline for comparisons with other health systems.
Certain performance improvement and/or cost-saving endeavors, such as those designed to gain buy-in from key stakeholders to successfully complete an initiative, may require only internal benchmark data. Nevertheless, for each initiative, it is necessary to determine if external comparative data will provide additional value. If not essential at the outset, external data may become vital later in the process to demonstrate to clinicians that their practice is uncommon or that it varies from other peer health systems.2 The use of external benchmarks allows you to compare and analyze inpatient and outpatient data from similar organizations to ascertain potential opportunities for clinical, operational, and financial improvement that may not have been discovered through analysis of internal variation alone.
Example Scenario 2: Your facility makes the decision to review potential medication utilization opportunities within your leukemia and lymphoma population (defined by select MS-DRGs). Comparison with all hospitals that submit data to a third party, and with a custom comparison group that takes into account variables including case volume, mortality rates, and LOS, reveals an increased use of multiple agents. You decide to focus initially on utilization of rasburicase because the results indicate utilization of that drug in 13.3% of cases in the former group versus in 4.27% of cases in the custom comparison group.
The transparent nature of the third-party benchmarking database enables you to reach out to colleagues to better understand their criteria of use and the associated outcomes. Based on conversations with key stakeholders, you determine that changes in utilization could result in estimated annual savings of $130,000. External benchmarking allows for an organization to identify and gain insight into best practices that they then can apply in the effort to reduce costs while enhancing clinical and operational performance through continuous quality improvement.
5. Confirm and summarize findings. It is important to keep in mind that limitations do apply to administrative data sets. While these data allow organizations to associate utilization of various resources (eg, thrombin, rasburicase, etc) within specific clinically defined populations, they may not provide the complete clinical picture. Therefore, once data have been generated, it is important to summarize findings and connect with key stakeholders to confirm the information.
After key findings have been identified and validated, it is important to summarize the overall meaning and impact on the organization as a whole. A final report for senior leadership should be developed that is succinct and easy to interpret. Where appropriate, recommendations should be supported with a thorough literature review or identified best practices; this will aid in the development of key implementation steps needed to move forward with the initiative being presented.2 See FIGURE 1 for highlights of key steps to take in order to translate data into actionable implementation steps.
Every day hospitals collect an abundance of information about patients and their treatment. Mining this data provides pharmacy with an excellent opportunity to spearhead improvement efforts and drive cost savings. Through the methodological use of administrative databases, as well as internal and external benchmarking, pharmacy can take a leadership role in the always ongoing effort to optimize clinical and financial outcomes.
Saloni Kapur Jain, PharmD, MPA, is the director of the pharmacy informatics program at UHC. Saloni received her Doctor of Pharmacy and Master of Public Administration degrees from Drake University and began her career at UHC as a fellow in 2008.
Arati Kurani, PharmD, BCPS, is an informatics pharmacist at UHC. She works primarily with member hospitals to mine the UHC clinical database to benchmark cost-effective and clinically appropriate utilization of medications and related outcomes to support a variety of projects. Arati received her Bachelor of Science degree in Information Systems from DePaul University and a Doctor of Pharmacy from Midwestern University. Following completion of a one-year pharmacy practice residency at Rush University, Arati obtained board certification as a pharmacotherapy specialist.
Michael J. Oinonen, PharmD, MPH, is senior director for the product technology group at UHC. His primary responsibilities are to lead and manage the operations for the Clinical Data Base, Resource Manager Data Base, and Core Measure regulatory reporting services. Michael received his BS in pharmacy and Doctor of Pharmacy from The Ohio State University. He also holds an MPH in health policy from the University of Illinois at Chicago. His post-doctorate training included a two-year, ACCP-recognized, Alexander M. Schmidt Fellowship at UHC.
Enter our Sweepstakes now for your chance to win the following prizes:
Just answer the following quick question for your chance to win:
Entries are limited to one entry per person in any active sweepstakes.
Thank you for your entry.