After reading this week?s chapters and the article below, address the following:

It has taken time for health care to embrace greater use of analytics. Imagine you are a newly hired manager of a brand-new clinic in the hospital. Decide either for or against the use of analytics and state your case as to why you either want to include analytics or wait until the clinic is more established.II, J.R. L. (2018). Performance Improvement in Hospitals and Health Systems (2nd Edition). Taylor & Francis. https://online.vitalsource.com/books/9781351584944

Introduction

Many management challenges in the healthcare industry can be addressed by analyzing readily available clinical and business data. Healthcare providers gather significant amounts of data on a routine basis, but do not make use of those data to analyze and understand organizational performance or to find ways to improve performance. Using simple analytical techniques, managers can identify patterns of resource utilization or associate payments with costs and find significant opportunities to improve profitability or quality of patient care.

Analytical Challenges

Kevin Rogers took his seat in his office as the new chief operating officer for Community Medical Center. As he started to peruse the reports presented to him daily, he noticed a summary of insurer payments and saw that there were some accounts paid for patients served by a newly recruited orthopedic surgeon. He noted that there appeared to be quite a bit of volume coming from this physician and the accounts appeared to be paid in a timely manner. So far, so good. Then he moved along to reviewing some invoices requiring approval for payment. Near the top of that stack, Kevin saw an invoice for implantable prosthetics, many of which were used on surgical cases by that newly recruited orthopedic surgeon. Out of curiosity, Kevin looked at some of the prices the hospital was charged for these items used as a part of surgeries performed by the new orthopedic surgeon. An unusual fact caught his attention?the invoice price for those orthopedic implants was about as much as the payment received for the surgeries he noted only a few moments before. ?How could this be?? he asked himself as a great state of concern and confusion built up inside him. He wondered, ?Is it possible that we could be losing money on much of the patient volume being generated by this new orthopedic surgeon? What could management be missing here?? Kevin pulled up the day?s operating room schedule and noticed that this newly recruited physician appeared prominently throughout the day in one operating room. Suddenly, Kevin started to feel a sense of dread.

It is Monday morning, and Tiffany James comes into her office as chief executive officer of High Country Health Plan. As she settled into the morning routine, she noted in her web news headline feed that a local physician had just been indicted for billing fraud by state Medicaid officials. She paused for a moment because that name sounded somewhat familiar to her. Opening her weekly claims payment report, she found that same physician?s name near the top of the list of largest claim payment amounts for the prior week. She suddenly started to wonder how much in those payments might be similarly fraudulent. ?How could it be that other insurers have discovered this and her plan had not?? she asked herself. She reached for her keyboard to send an e-mail to the health plan claims director to get more details.
It is Monday morning, and Tiffany James comes into her office as chief executive officer of High Country Health Plan. As she settled into the morning routine, she noted in her web news headline feed that a local physician had just been indicted for billing fraud by state Medicaid officials. She paused for a moment because that name sounded somewhat familiar to her. Opening her weekly claims payment report, she found that same physician?s name near the top of the list of largest claim payment amounts for the prior week. She suddenly started to wonder how much in those payments might be similarly fraudulent. ?How could it be that other insurers have discovered this and her plan had not?? she asked herself. She reached for her keyboard to send an e-mail to the health plan claims director to get more details.

These two scenarios point out common problems that face healthcare managers in today?s marketplace. In each case, an ongoing program of data analysis?asking questions about the utilization of resources and the efficiency and effectiveness of operational operations, and seeking ways to improve patient care?is becoming an essential tool in running a successful healthcare business. The challenges facing managers in this industry call for the ability to analyze data and understand what is happening in the business.

The healthcare system in the United States in the early twenty-first century might reasonably be labeled to be in a state of great disarray. Policy makers, healthcare providers, and government policy makers struggle with the conflicting priorities of providing good patient care at a reasonable price within a financing system that is sustainable for the broader economy. Considering that the U.S. healthcare system currently consumes nearly 18% of the nation?s economic output (Kaiser Family Foundation, 2017), the prospect of healthcare cannibalizing resources from other essential parts of the economy is a great policy concern. At the same time, reducing payments to providers to stem the rate of growth of healthcare in the economy poses a risk of taking resources away from meeting essential patient care needs. Thus, our current healthcare system faces several vexing dilemmas, primarily how to deliver care that meets generally accepted standards of quality, at a lower cost to the economy?

Fundamental to our healthcare system is the gathering of detailed documentation of services provided to patients for each occasion of service. The healthcare system for decades has gathered enormous volumes of data on patients and then systematically filed them away in archives. This store of useful data has remained essentially unused in understanding the business of healthcare for decades. Since passage of the Health Information Technology for Economic and Clinical Health (HITECH) Act in 2009, the industry has started to move toward gathering that patient care data in a form that can be analyzed, referred to as structured data (Hoyt and Yoshihashi, 2014). The advent of electronic health records (EHRs) has converted our old practice of creating data on paper (unstructured data) and relying on manual review of paper records to the new norm of documenting patient data using structured data. This gives managers in the industry greater resources to use in understanding what is happening in the organization and to find ways to improve performance.

But why bother to analyze performance in healthcare organizations? The passing of the Patient Protection and Affordable Care Act (ACA) in 2010 set the stage for healthcare providers (e.g., hospitals, physicians, and clinics) to be held to standards of demonstrating patient care quality, while doing so at reduced fees. Private insurers quickly followed suit. Payment reductions were further exacerbated by the implementation of quality penalties for hospital providers if certain levels of patient care process were not documented in the case of Medicare patients (Kaiser Family Foundation, 2013). Commercial insurers have also similarly implemented such quality process penalties. Further, as was illustrated in the opening scenario, tracking utilization of resources is essential in maintaining profitable operations. As described in the first scenario mentioned at the beginning of this chapter, being able to track the cost and volume of patient care resources is an essential management exercise in today?s environment of tightly constrained payments.

Using another example presented at the beginning of this chapter, a health plan could easily protect its resources from fraudulent or ineffective services by tracking resource utilization and perhaps profiling the utilization of resources by providers in its network. Being able to review payments for patterns of unusual behavior or to compare utilization among providers can help a health plan protect the integrity of its claims adjudication process and keep medical costs low for its customers.

The Value of Analytics in Healthcare

The focus of this chapter is therefore to introduce the reader to the types of analyses useful in today?s healthcare industry, to discuss the sources and types of data available, to give a brief synopsis of the types of analyses that can be useful to healthcare managers, to define performance metrics that can be tracked on an ongoing basis, and finally, to apply these analytic tools to other established performance improvement programs, such as dashboards and Six Sigma, and ensure compliance with federal value-based purchasing requirements.

Healthcare businesses can use the data they routinely gather to make necessary changes to operations that will improve profitability and patient care. This analysis is referred to as analytics. Analytics involves the use of available business data to understand operational performance, resource utilization, and costs of patient care, and using such information to support managers in making critical decisions about operations (Strome, 2013). While healthcare has been using computer systems for decades, it has only been in the last 10 years that provider organizations have gathered data specific to patient care in a computerized format that can be easily combined with financial data to gain much greater insight into organizational performance.

Types of Data in Healthcare Organizations

Healthcare organizations have a wealth of data available to them to conduct very productive and insightful analytic work. The centerpiece of this analytics ?toolbox? is the electronic medical record. The electronic medical record replaces much of the paper record keeping that has been used by healthcare providers for decades. However, the electronic medical record is not the only source of patient care data. Other applications that can be used in healthcare analytics include radiology information systems (RISs), picture archival and communication systems (PACSs), medication administration records (MARs), laboratory information systems (LISs), and pharmacy information systems (RxISs) (Glandon et al., 2013). The combination of the clinical documentation applications described here is the EHR. All these applications gather data specific to the care of the patient: the clinical signs and symptoms evidenced by the patient, plans for treatment, and evaluation of treatment outcomes?referred to as clinical data (Wager et al., 2013).

Essential to managing the business affairs of a healthcare organization is an entirely different set of applications aimed at operational functions, such as billing and collection, purchasing and inventory of clinical supplies, payroll, and claims adjudication and accounts payable systems. The patient demographics, cost of supply, and cost of labor data gathered by these applications are commonly referred to as administrative data (Wager et al., 2013).

Administrative and clinical data are both essential elements in healthcare analytics projects. The challenge is understanding what data elements reside in which area. Table 8.1 summarizes some common types of administrative and clinical data elements.
The Value of Analytics in Healthcare

The focus of this chapter is therefore to introduce the reader to the types of analyses useful in today?s healthcare industry, to discuss the sources and types of data available, to give a brief synopsis of the types of analyses that can be useful to healthcare managers, to define performance metrics that can be tracked on an ongoing basis, and finally, to apply these analytic tools to other established performance improvement programs, such as dashboards and Six Sigma, and ensure compliance with federal value-based purchasing requirements.

Healthcare businesses can use the data they routinely gather to make necessary changes to operations that will improve profitability and patient care. This analysis is referred to as analytics. Analytics involves the use of available business data to understand operational performance, resource utilization, and costs of patient care, and using such information to support managers in making critical decisions about operations (Strome, 2013). While healthcare has been using computer systems for decades, it has only been in the last 10 years that provider organizations have gathered data specific to patient care in a computerized format that can be easily combined with financial data to gain much greater insight into organizational performance.

Types of Data in Healthcare Organizations

Healthcare organizations have a wealth of data available to them to conduct very productive and insightful analytic work. The centerpiece of this analytics ?toolbox? is the electronic medical record. The electronic medical record replaces much of the paper record keeping that has been used by healthcare providers for decades. However, the electronic medical record is not the only source of patient care data. Other applications that can be used in healthcare analytics include radiology information systems (RISs), picture archival and communication systems (PACSs), medication administration records (MARs), laboratory information systems (LISs), and pharmacy information systems (RxISs) (Glandon et al., 2013). The combination of the clinical documentation applications described here is the EHR. All these applications gather data specific to the care of the patient: the clinical signs and symptoms evidenced by the patient, plans for treatment, and evaluation of treatment outcomes?referred to as clinical data (Wager et al., 2013).

Essential to managing the business affairs of a healthcare organization is an entirely different set of applications aimed at operational functions, such as billing and collection, purchasing and inventory of clinical supplies, payroll, and claims adjudication and accounts payable systems. The patient demographics, cost of supply, and cost of labor data gathered by these applications are commonly referred to as administrative data (Wager et al., 2013).

Administrative and clinical data are both essential elements in healthcare analytics projects. The challenge is understanding what data elements reside in which area. Table 8.1 summarizes some common types of administrative and clinical data elements.

The combination of clinical and administrative data is often needed to answer management questions such as, how much does it cost to take care of a hip replacement, or which physician has the best outcomes for invasive cardiology? Usually, the clinical and administrative data needed to answer these types of questions reside in different application databases. Trying to join the data from these disparate applications may be a challenge to completing an analytics project. However, most administrative and clinical applications are joined together through a common database, known as the master patient index (MPI). Within the MPI, patient demographics are usually captured in one place and shared with other applications?both clinical and administrative?used within a healthcare organization. The MPI captures patient name, other demographic information, and a unique patient identification number used within the organization. That identification number in the MPI can be used to join data for a specific patient or group of patients across administrative and clinical application databases. Figure 8.1 presents a high-level view of these myriad applications and their data-?sharing relationships.

From the health plan perspective, there certainly will be less clinical data gathered since the health plan primarily acts as a vehicle for financing payments to healthcare providers on behalf of patients. However, there is limited gathering of clinical data in a health plan for purposes of utilization management and coordination of care among provider groups, usually for patients with chronic conditions, such as high blood pressure or asthma. Administrative applications used by health plan organizations would be like those described earlier, but will focus primarily on claims adjudication systems that will gather significant amounts of data from claims for payment sent by healthcare providers to the insurer. In addition, the MPI used by providers would be replaced with a membership or enrollment database that would serve the same function of describing patient demographic data, but also include the dates between which a health plan would be obligated to pay for services on behalf of the patient.
The medical claim form normally used in the healthcare industry today includes basic clinical information that can help to guide analytics projects as well. Data from the medical claim form can be obtained from the patient accounting application in a provider organization or the claims adjudication system in the health plan entity. The data elements captured in the provider claim?either an institutional provider (such as a hospital, nursing facility, home healthcare, outpatient surgery center, or hospice) or professional provider?include the items listed in Table 8.2 (Centers for Medicare and Medicaid Services, 2017). Often, medical claim data can be a comprehensive source of useful data for many operational performance analyses.

Claim data can be a useful start to an analysis project. However, that data will have limited value in associating clinical signs or symptoms with actual treatments or resource decisions. It also may not assist in identifying the costs of resources used in treating a patient at a microlevel, such as the cost of implantable devices or the labor hours involved in one day of care in the intensive care unit. It is usually limited to identifying types of cases (e.g., by diagnosis or by insurer) or high-level comparisons of metrics by patient type, insurer, or treating physician. To associate clinical data with other types of data gathered in clinical applications, such as the EHR, database tools and some limited data manipulation skills will be needed.

Understanding and Managing Data

Administrative and clinical applications used in healthcare organizations rely on databases to compile and organize data for use within the application. A database is a collection of data organized in a series of tables, organized by type of data or subject, such as a listing of patients with date of birth or a table of medications used in a clinic. Each table is made up of multiple rows and columns, where each row is one record in the table and columns represent a variable. The databases of most contemporary health applications are organized in what is known as a normalized database form. A normalized database separates data into multiple different tables to minimize the duplication of items in the database. The normalized database relies on connections between tables using a common item, known as a key.

In this illustration, the patient, Fred Jones, was born on November 27, 1952; has Medicare insurance, policy number 40909Z; and was seen in the clinic on January 24, 2017, under encounter number 1583. The diagnosis was pancreatic cancer, and the service provided was an office visit for an established patient with a complex medical diagnosis. As you will note in Figure 8.3, getting all the information for the patient requires pulling elements from different tables in the EHR to obtain the variables that make up the details of this patient encounter. This normalized database helps information technology staff by minimizing the amount of storage space needed in computer hardware, but it presents a challenge to the manager wanting to analyze those data. The challenge arises from three different perspectives:

1.The manager must know how data are organized in the database to find all the various elements required to answer an analytical question.

2.The manager must understand the various keys used in the database to be able to join data elements from multiple tables in the database.

3.The manager must have a database tool that can find the desired rows in the database table and join them with rows from other tables to create information that can answer a question such as ?Which Medicare patients were seen in January with a diagnosis of pancreatic cancer???as was shown in this simple illustration.

Analytical Tools

The analytical tools that can be used to accomplish operational analyses can be as simple as a Microsoft Excel spreadsheet, where lists could be joined together using a simple combination of?=?VLOOKUP() functions to combine elements in two lists, such as that shown in Figure 8.4.

Note in Figure 8.4 how the patient ID in row 3, column A is used to find the patient encounter ID in the patient encounter table shaded in red. This approach using the Excel spreadsheet tool can be useful for simple analyses where joining few elements from small lists. Were this analysis to use much larger lists with more variables, the capacity of an Excel spreadsheet could be exceeded. That is where the analytical question may require a database query tool, such as Microsoft Access, Microsoft SQL Server, Oracle, or MySQL.

Database tools such as those mentioned above use a query to find ?specific rows in a database table and join them with rows from other tables using one or more database keys. Such queries are often completed using a common database programming language like Structured Query Language (SQL). SQL can create queries that pick specific variables from tables, joining them based on one or more keys. The example given earlier of Mr. Jones might be carried out in SQL using a query such as this:

SELECT ID, Firstname, Lastname, DOB FROM MasterPatientIndex INNER JOIN MasterPatientIndex.ID = PatientInsuranceDetail.ID, InsurerID FROM PatientInsuranceDetail, INNER JOIN MasterPatientIndex.ID = PatientEncounterList.ID, PatientEncounterDetail.Diagnosis INNER JOIN PatientEncounterList.EncounterID = PatientEncounterDetail.EncounterID, Diagnosis from PatientEncounterDetail

WHERE PatientInsuranceDetail.InsurerID = ?101? AND PatientEncounterDetail.Diagnosis = ?C25.3?;

This SQL query would have sought out the first name, last name, date of birth, insurer ID code, and diagnosis code for all patients having Medicare insurance and a diagnosis of pancreatic cancer, just as was shown at the top of Figure 8.4. Thus, a database tool such as SQL can provide a great deal of analytical power to managers seeking data from the complex databases used in today?s health applications. The details on application databases and the nuances of SQL programming needed to execute an analytical task like this are beyond the scope of this text. For more details on the composition of the application databases used in a healthcare organization, readers are encouraged to consult with their application vendor technical support teams. Many application vendors have help desk resources dedicated to providing managers with assistance on where to look for specific data elements or how to query application data tables. For more assistance on SQL query construction, the reader is encouraged to seek out an SQL programming resource, such as Sam?s Teach Yourself SQL in 24 Hours by Stephens et al. (2015).

Analyzing Data

Database tools allow the manager to filter and combine large tables to complete very sophisticated queries that can create a very specific data set usable for detailed analysis of a management question or issue (Madsen, 2012). However, it is important to know about the nature of the data when designing an analytical project?are the variables numeric, alpha characters, or special codes using a combination of letters and numbers? Successful healthcare analytics requires an understanding of how the data are organized and what they express.

A good example of this challenge comes in the use of diagnosis and procedure codes in the EHR. Diagnoses and inpatient hospital surgeries are described using a combination of letters and numbers in the International Classification of Diseases, of which the tenth edition is currently in use in the United States (known as ICD-10). This coding set provides a very detailed description of diagnostic findings, using a code that varies from three to five characters in length, with or without a decimal point included (Sinha, 2013). The ICD-10 code for myocardial infarction (without other descriptive terms) is I21.3. However, that diagnosis could be further subdivided into the location of the infarction injury in the heart, such as the anterior (I21.09), inferior (I21.19), or lateral (I21.29) aspect of the myocardium. Similarly, a malignant cancer diagnosis in the colon is generally described by the ICD-10 code C18, but can be further subdivided based on the location along the colon by the cecum (C18.0), ascending colon (C18.2), transverse colon (C18.4), or descending colon (C18.6).

In the ICD-10 coding set, numbers and letters are used interchangeably, where numerals act as another letter. Numerals in this code set cannot be used for calculation. The key for the manager is to have resources that can assist in understanding this detailed code set and provide guidance in selecting the correct codes for the analytical project. Additional resources and training in the use of the ICD-10 coding system can be found through the American Health Information Management Association (www.ahima.org).

Procedures (other than inpatient surgeries) or other diagnostic or treatment services are described using the Common Procedural Terminology (CPT) code set. A CPT code is made up of five characters, and like the ICD-10 diagnosis code, can include both letters and numerals. CPT is subdivided into three broad categories. Category I procedures are those performed by healthcare providers for patients and are subdivided into six general subcategories:

1.Evaluation and management (codes 99201?99499)?Usually physician services that involve assessing the patient?s medical history and current condition and rendering treatment or creating a treatment plan

2.Anesthesia (codes 00100?01999 and 99100?99140)?Specific to the type of anesthesia given and any additional services needed to safely administer anesthesia with other health conditions

3.Surgery (codes 10021?69990)?Provide a description of surgical procedures provided by the physician and the surgical facility and are organized by the area of the body where the surgery was performed;
4.Radiology (codes 70010?79999)?Describe any diagnostic imaging studies performed or any treatment procedures that used radiology services in addition (such as checking the placement of an implantable device using fluoroscopy)?this set of codes is organized by the type of imaging modality used, such as computerized tomography or ultrasound

5.Pathology and laboratory (80047?89398)?This range of codes describes diagnostic tests performed in the clinical laboratory, as well as any professional examination of patient tissue by a pathologist

6.Medicine (codes 90281?99199 and 99500?99607)?Describe all other nonsurgical treatment services rendered by a healthcare professional, and again are organized by the body system involved in the treatment (Sinha, 2013; MB&CC, n.d.)

Category II CPT codes are used to track services and quality of care measures and are made up of four digits with an F at the end. These codes are used in much the same way as a Category I code described above, but are used for statistical tracking of preventive health services that are not currently reimbursable by insurers. These codes are subdivided into nine groups used to describe patient quality of care metrics:

1.Composite measures (CPT codes 0001F?0015F)?Address conditions that may impact the entire patient, such as code 0002F for assessment of a patient?s tobacco use by smoking

2.Patient management (CPT codes 0500F?0575F)?Describe specific actions by a physician to manage an identified health condition, such as code 0556F for developing a plan of care to achieve control of elevated lipids in a patient

3.Patient history (CPT codes 1000F?1220F)?These codes further examine the presence of conditions reported by the patient in a medical history, such as code 1005F for evaluation of asthma symptoms

4.Physical examination (CPT codes 2000F?2050F)?This range of codes considers management of findings made during a physical exam, such as code 2044F, which documents a mental health assessment

5.Diagnostic and screening processes and results (CPT codes 3006F?3573F)?These codes are used to document review of the results of previously ordered diagnostic tests, an example of which is CPT code 3015F, which is used to document review of the results of a cervical cancer screening test

6.Therapeutic, preventive, or other interventions (CPT codes 4000F?4306F)?This range of CPT codes documents the performance of other health treatments or interventions, such as code 4062F to document the referral of a patient for psychotherapy

7.Follow-up or other outcomes (CPT codes 5005F?5100F)?These codes describe actions by a treating professional to act on the results of treatment or physical exam, as done with CPT code 5020F to document the communication of a patient?s treatment summary to another physician or health professional

8.Patient safety (CPT codes 6005F?6045F)?This set of codes describes actions taken by the treating provider to protect the patient?s safety during treatment, such as with CPT code 6040F to document the use of radiation protection devices during a diagnostic imaging test or treatment

9.Structural measures (CPT codes 7010F?7025F)?This set of CPT codes documents actions by a treating provider to conduct follow-up with the patient on screening results as would be done using CPT code 7020F for evaluating a patient for mammography follow-up (MB&CC, n.d.)
Category III CPT codes encompass those used for new technologies, services, or procedures that do not yet have a permanent CPT code assigned to them. This classification of CPT code is like the Category II setup in that it is a five-character code using four digits followed by the letter T. These codes are usually used for data collection and tracking the utilization of emerging technologies or services. These procedures are often involved with biomedical research and have not yet been approved for use as a Category I procedure code. Updates to the Category I codes are published every other year by the American Medical Association.

CPT codes are further detailed using a set of two-character (letter or numeral) modifiers that provide an additional description of the service. A modifier is an adjunct to a CPT code that provides a further description of the service or procedure. One example is use of the modifier code 50 on a surgical procedure to show that the procedure was performed on both the left and right sides of the body. Other modifiers frequently used are those to describe the primary procedure done during a surgical case where more than one procedure was performed on the patient (modifier 51) or if multiple surgeons were involved in the case (modifier 62).

When examining data from a facility (e.g., a hospital or surgical center) patient billing system, the analytical task may require identification of the facility department that provided a service to the patient. A facility revenue code can be used to identify that department. The revenue code is a four-digit numeric code that shows the department a service was provided in. A list of common revenue codes is shown in Table 8.3 (Centers for Medicare and Medicaid Services, 2017).
Many analytic questions may involve the use of medications, which can vary by manufacturer, name, and packaging (Langabeer and Helton, 2016). Attempting to screen medication data by name and then these other characteristics could be very challenging were it not for a standardized system of identifying patient medications?the National Drug Code (NDC). The NDC is a 10-digit unique identifier that allows an analysis to focus on specific characteristics of medications?manufacturer, drug name, and packaging type. The NDC is broken into three segments:

1.Manufacturer: Identified by a five-digit code, such as 666582 for Merck

2.Product: Identified by a three-digit code, such as 311 for Vytorin, 10 mg tablet

3.Packaging: Identified by a two-digit code, such as 31 for 30 tab