Health Analytics Discussion

Y Z[\ ]^ _̀ab _cde
fZ_a][ Zgh ]iZjk

lm

n opqrstouvw txy vzty
quw vu{ q|} stzvw~ o�y
quw u

�� ����� �����

� ���� ���� ���

�SS�WUOUR�QU�TSW��
QUV

OPQRSTPUVQTWXWQPO

�PVOS�Q�SOO���QXT
SW��QUV

� ����� � ����� � �����

� ���� ���� ���

����PRW�QXS��������U
Q�OP���UV

�RU�QOP��

�UV�P�QPR

WQ����P�Q�TUR�QXOP
���UV�P�QP O�UO��UVVPXTSW��QU

V
ST�SWQUQP����PRW�QX
TSW��QUVW

�RU����P�TSW��QUV�
OP���UV�P�OPR�XWQ����P�QOP��

�UV�P�QPR

V�QTPRU�TSW��QUV � ���� ���� ���
�S�QSRWTSW��QUV OUR�S��P�PRUV

TSW��QUV
ST�STPUVQTS��VP�PWW
TSW��QUV

�VU��TUR��UVVPXTSW
��QUV OS��Q�UROPV�PWQ R��PRW��POPQTS��WQT

SW��QUV�P�PW�WTSW��QUV WS�QTPR�ST�SOP���U
V�P�QPR

VU�PTPUVQTO ��� �� � ���� ���� ���
�PQTPW�U�SRQT OPR�PR�S��QX�S��QQ

S��WT���SO ����PRW�QXTSW��QUVWU
T��UOP����TR�WQTSW��QUV �UROU�SOO���QX�P�

PRUVTSW��QUV O ��� ��O ��� ��

“#$%&’ ()*+ ,-.
/ )-0123 .4 5*(-1+ 5.
6 -.-7)().1′ 885,
+ -1+ 5.
9 :;< =;< >?:@AB =
BC D����
����
������
������
������
��

� � ¡ ¢£¡ ¤¥¦§ �©̈ £¡ ª

“#$%&’ ()*+ ,-.
/ )-0123 .4 5*(-1+ 5.
6 -.-7)().1′ 885,
+ -1+ 5.
9 :;< =;< >?:@AB =
BC D?:@AB =
BC D?:@AB =
BC D?:@AB =
BC D?:@AB =
BC DHEALTHDATA ANALYSIS

TOOLKIT

A m e r i c a n H e a l t h I n f o r m a t i o n M a n a g e m e n t A s s o c i a t i o n

Copyright ©2014 by the American Health Information Management Association. All rights reserved.
Except as permitted under the Copyright Act of 1976, no part of this publication may be reproduced,
stored in a retrieval system, or transmitted, in any form or by any means, electronic, photocopying,
recording, or otherwise, without the prior written permission of AHIMA, 233 N. Michigan Ave.,
21st Fl., Chicago, IL, 60601 (https://secure.ahima.org/publications/reprint/index.aspx).

ISBN: 978-1-58426-442-2
AHIMA Product No.: ONB194014
AHIMA Staff:
Jessica Block, MA, Project Edito

r

Pamela Woolf, Director of Publications, AHIMA Press
Anne Zender, Senior Director, Periodical

s

Limit of Liability/Disclaimer of Warranty: This book is sold, as is, without warranty of any kind, either
express or implied. While every precaution has been taken in the preparation of this book, the publisher
and author assume no responsibility for errors or omissions. Neither is any liability assumed for damages
resulting from the use of the information or instructions contained herein. It is further stated that the
publisher and author are not responsible for any damage or loss to your data or your equipment that
results directly or indirectly from your use of this book.

The websites listed in this book were current and valid as of the date of publication. However, webpage
addresses and the information on them may change at any time. The user is encouraged to perform his
or her own general web searches to locate any site addresses listed here that are no longer valid.

CPT® is a registered trademark of the American Medical Association. All other copyrights and trademarks
mentioned in this book are the possession of their respective owners. AHIMA makes no claim of owner-
ship by mentioning products that contain such marks.

For more information about AHIMA Press publications, including updates,
visit ahima.org/education/press.

American Health Information Management Association
233 N. Michigan Ave., 21st Fl.
Chicago, Illinois 60601

HEALTH DATA ANALYSIS

TOOLKIT

AHIMA.ORG

HEALTH

DATA ANALYSIS TOOLKIT

TABLE OF

CONTENTS

Foreword …………………………………………………………………………………………………………………………………………… 4

Authors and Acknowledgments ……………………………………………………………………………………………………. 4

Introduction……………………………………………………………………………………………………………………………………….

5

Data Dictionary ………………………………………………………………………………………………………………………………… 7

Study Design and Report Request Form ……………………………………………………………………………………… 9

Data Capture Methods and Best Practices ……………………………………………………………………………… 10

Validating Data Outcomes ………………………………………………………………………………………………………….. 11

Presenting the Data ……………………………………………………………………………………………………………………… 13

Formulas and Statistics ……………………………………………………………………………………………………………….. 19

Sample Job Skills and Responsibilities ………………………………………………………………………………………. 28

Glossary of Terms ………………………………………………………………………………………………………………………….. 32

Annotated Bibliography ………………………………………………………………………………………………………………. 41

Case Study ………………………………………………………………………………………………………………………………………. 48

Appendix A: Data Dictionary Sample ………………………………………………………………………………………… 49

Appendix B: Study Report and Request Form …………………………………………………………………………. 50

Appendix C: Sample Data Definitions ……………………………………………………………………………………….. 52

Appendix D: Meaningful Use ………………………………………………………………………………………………………. 62

Appendix E: Clinical Quality Measures …………………………………………………………………………………….. 64

4 | AHIMA

HEALTH DATA ANALYSIS TOOLKIT

FOREWORD

This toolkit provides a variety of resource tools for healthcare professionals performing data analysis tasks, whether
they are reviewing and trending healthcare data for a healthcare entity, reporting on quality measures in a physi-
cian office, managing the enterprise master patient index, or working with other operational and financial data.
The toolkit begins by discussing the current healthcare initiatives and programs that are increasing the demand for
data analytics. From there, the resources focus on the acquisition of the data and the purpose of the data dictionary
and include resources from the point when information is requested via a sample report request form to how data
should be validated. Examples of good and bad data displays are discussed. The second half includes a listing of
common formulas and statistics, a glossary of terms, and an annotated bibliography used in data analysis. Finally,
the toolkit concludes with a sample case study illustrating how data are collected, analyzed, and transformed into
information for reporting.

AUTHORS
Julie Dooling, RHIA, CHDA

Pawan Goyal, MD, MHA, MS, PMP, FHIMSS,
CPHIMS, CBA

Linda Hyde, RHIA

Lesley Kadlec, MA, RHIA

Susan White, PhD, RHIA, CHDA

ACKNOWLEDGEMENTS
Michelle Custodio, RHIA, CCDS, CDIP

Jane DeSpiegelaere, MBA, RHIA, CCS, FAHIMA

Katherine Downing, MA, RHIA, CHPS, PMP

Kim Turtle Dudgeon, RHIT, CHTS, IS/TS, CMT

Angela Dinh Rose, MHA, RHIA, CHPS, FAHIMA

Carol F. Smith, MBA, RHIA, FAHIMA

Diana Warner, MS, RHIA, CHPS, FAHIMA

Lou Ann Wiedemann, MS, RHIA, FAHIMA, CDIP,
CHDA, CPEHR

ORIGINAL AUTHORS
June Bronnert, RHIA, CCS, CCS-P

Jill S. Clark, MBA, RHIA

Linda Hyde, RHIA

C. Jeanne Solberg, MA, RHIA

Susan White, PhD, RHIA, CHDA

Mark Wolin, MHSA

ORIGINAL ACKNOWLEDGEMENTS
Jane DeSpiegelaere-Wegner, MBA, RHIA, CCS,
FAHIMA

Angela K. Dinh, MHA, RHIA, CHPS

Kathy Giannangelo, MA, RHIA, CCS, FAHIMA

Syreeta Kinnard, RHIA

Wendy Scharber, RHIT, CTR

Michelle Shimmel, RHIT

Mary H. Stanfill, MBI, RHIA, CCS, CCS-P, FAHIMA

Diana Warner, MS, RHIA, CHPS

Lou Ann Wiedemann, MS, RHIA, FAHIMA, CPEHR

Kelli Wondra

5 | AHIMA

DATA ANALYSIS TOOLKIT

INTRODUCTION
As electronic health record (EHR) use advances, the available data elements continue to expand. Healthcare has
become a data-rich field. Various government initiatives, such as the transition to the International Classification
of Diseases 10th Revision Clinical Modification/Procedure Coding System (ICD-10-CM/PCS) and legislation
such as the American Recovery and Reinvestment Act and the Patient Protection and Affordable Care Act, further
enhance the necessity for organizations to analyze their data to make informed business decisions. This toolkit
provides a resource that facilitates and supports data analytics performed by health information management
(HIM) and other professionals at a foundational level.

VALUE-BASED PURCHASING

Healthcare payers are moving from payment for volume to pay for value and performance. Data analysis can help
healthcare organizations maximize their value as providers to patients and payers. The first phase of the CMS
value-based purchasing (VBP) program was implemented in October 2012. Hospitals paid via the inpatient
prospective payment system (IPPS) found two significant changes to their payment during FY 2013. First,
implementation of the VBP program reduced payment by 1 percent, but provided hospitals with the opportunity
to earn back that 1 percent reduction and more by performing well according to the CMS Total Performance Score
(TPS).1 Second, some hospitals faced reductions in payment due to the Readmissions Reduction Program.

Data mining to determine which VBP metric is causing a reduction in a hospital’s payment requires not only
knowledge of healthcare data analytics, but also knowledge regarding reimbursement and the business side of
healthcare. Quality measurement requires attention to data quality and validity. Health data analysts are needed to
design sampling plans for abstracted measures and specify data extract parameters for administrative data-driven
measures. All of these roles create new opportunities for HIM professionals that can bring both context and
content to the table in healthcare data analytics.

MEANINGFUL USE AND CLINICAL QUALITY MEASURES

In order for an eligible professional or hospital to attest to having met the objectives of the Medicare and Medicaid
EHR incentive programs, specific data must be collected on each provider and hospital for a specified time period,
and a percentage is calculated based upon a specific subset of patients. There are inclusion and exclusion criteria
for each objective. These incentive programs provide financial incentives for the “meaningful use” of certified
EHR technology to improve patient care. The ability to demonstrate “meaningful use” is driven by electronic data
capture, management, analysis, and reporting. Health data analysts play a key role in obtaining the necessary data
and analyzing it to ensure it is correct.

For more details on Meaningful Use, please see Appendix D.

In addition to meeting the core and menu objectives, eligible professionals, eligible hospitals, and critical access
hospitals are also required to report clinical quality measures. For more details, please see Appendix E.

1. Centers for Medicare and Medicaid. “Frequently Asked Questions, Hospital Value-Based Purchasing Program.” March 9, 2012.
http://www.cms.gov/Medicare/Quality-Initiatives-Patient-Assessment-Instruments/hospital-value-based-purchasing/Downloads/
FY-2013-Program-Frequently-Asked-Questions-about-Hospital-VBP-3-9-12 .

HEALTH DATA ANALYSIS TOOLKIT

6 | AHIMA

HOW DATA ANALYTICS ALIGNS WITH AHIMA STRATEGIC INITIATIVES

Within the healthcare environment, data is limitless. Whether it is clinical, administrative, financial, patient-
generated or other data, the need to manage the data efficiently is more important than ever.

As healthcare organizations (HCOs) continue to implement EHRs and new care delivery models such as account-
able care organizations (ACOs) flourish and grow, demands for accountability, reliability, and security must be met.

Now is the time for healthcare professionals to start creating and participating in information governance pro-
grams. In healthcare, the introduction of information governance represents something of a change, though for
many other industries information governance programs have been quite successful. Information governance is a
strategic initiative for AHIMA.

The goal of an information governance program is to ensure that all information resources support the business
goals of an organization. Health data is at the core of information governance. Having the knowledge to acquire,
manage, analyze, interpret, and transform this data into accurate, consistent, and timely information, while
balancing the strategic vision with day-to-day details, is a core responsibility of a health data analyst and is a key
factor in the success of information governance programs.2

WHAT IS DATA?

Data is plural of datum, which is the dates, numbers, ages, symbols, letters, and words that represent basic facts and
observations about people, processes, measurements, and conditions.3

Data could be qualitative or quantitative. (See the glossary of terms.) It could be collected and represented as an
individual or set of numbers, alphabets, symbols, pictures, voice, or video.

Data in its raw form is of limited value. It needs to be processed in a meaningful way to create information that
would be relevant to a situation. When the information is used for decision making, it leads to knowledge creation
for common understanding. Repeatable use of knowledge leads to development of best practices which, applied
over time to achieve organizational goals, leads to behavior change.

2. Thomas Gordon, Lynne. “Information Governance for the Health Care Industry: Now is the Time.” iHealthbeat, Feb. 3, 2014.
http://www.ihealthbeat.org/perspectives/2014/information-governance-for-the-health-care-industry-now-is-the-time

3. AHIMA. Pocket Glossary of Health Information Management and Technology, 4th ed. Chicago: AHIMA Press, 2014.

HEALTH DATA ANALYSIS TOOLKIT

7 | AHIMA

DATA DICTIONARY
A data dictionary is a descriptive list of names (also called “representations” or “displays”), definitions, and attri-
butes of data elements to be collected in an information system or database. The purpose of the data dictionary is
to standardize definitions and ensure consistency of use. It is a tool to aid in the standardization of data definitions.
It is not and should not be confused with other terms such as a “pick list” or “drop-down menu” within an EHR.
See Appendix A for a sample data dictionary.

A key focus of a data dictionary is to support and adopt more consistent use of data elements and terminology to
improve the use of data in reporting. A data dictionary promotes clearer understanding, helps users find infor-
mation, promotes more efficient use and reuse of information, and promotes better data management. The data
dictionary is a critical component of data governance.

According to the International Organization for Standardization:

The increased use of data processing and electronic data interchange heavily relies on accurate, reliable, con-
trollable, and verifiable data recorded in databases. One of the prerequisites for a correct and proper use and
interpretation of data is that both users and owners of data have a common understanding of the meaning
and descriptive characteristics (e.g., representation) of that data. To guarantee this shared view, a number of
basic attributes has to be defined.4

A dictionary describes the definitions or the expected meaning and acceptable representation of data for use
within a defined context of data elements within a data set. In addition to the name and definition of the data
element, the metadata may include other attributes or characteristics such as length of data element, data type
(character or numeric), data frequency (mandatory or not), allowable value and constraints, originating source
system, data owner, data entry date, and data termination date.

Data are often stored in many different databases and may be of variable quality. Inconsistent naming conventions,
inconsistent definitions, varying field length for the same data element, and/or varied element values all can lead to
significant problems, including poor data quality and misuse of data in reporting, among others. The following are
a few of many examples of inconsistent data throughout an organization:

1. Inconsistent naming conventions

• The date of the patient’s admission is referred to as the date of admission in the patient management (PM)
system, admit date in the fetal monitoring system, and admission date in the cardiology database.

• The unique patient identifier is referred to as a medical record number in the PM system, patient record
identifier in the operating room system, and “A” number (a moniker leftover from a legacy system from
25 years ago) and enterprise master patient identifier in the catheterization laboratory system.

2. Inconsistent definitions

• Admission, discharge, transfer (ADT) system: date of admission is the date on which an inpatient or day
surgery case admission occurs; in the trauma registry system, date of admission is the date on which the
trauma patient enters the operating room.

• The pediatric age is defined as age less than or equal to 13 in the PM system, whereas the pediatric disease
registry defines a pediatric age as less than the age of 18.

• In the bed board system, a nursing unit may be defined as 5W or 5 West. Within the scheduling system,
unique locations are defined as short procedure unit or SPU, such as X-ray or radiology, for example.

4. International Organization for Standardization, 2004. “Information Technology, Parts 1–6.” (2nd ed). www.iso.org/iso/home.html.

HEALTH DATA ANALYSIS TOOLKIT

8 | AHIMA

3. Varying field length for same data elemen

t

• The field length for a patient’s last name is 26 in the PM system, whereas the field length for a patient’s last
name is 15 in the cancer registry system.

• The medical record number in the PM system is 16 characters long, whereas the cancer registry system
maintains a length of 13 characters for the medical record number.

4. Varied element values

• The patient’s sex is captured as M, F, or U in the ADT system, whereas the patient’s sex is captured as
Male, Female, or Other in the peripheral vascular laboratory database.

Data dictionaries facilitate the work of the health data analyst by developing a common understanding of an
organization’s data quality. Therefore, although ability to edit the organization’s data dictionary should be limited
to system administrators, it is important that the dictionary be viewable for all those who use data to manage their
work, including but not limited to health data analysts. The dictionary is often organized as a table and should be
in a format that others can access. To review the varying formats, please see Appendix A or use the following links:

• Research Data Assistance Center

• Medicaid and CHIP Statistical Information System

• Google image search on “data dictionary example”

HEALTH DATA ANALYSIS TOOLKIT

9 | AHIMA

STUDY DESIGN AND REPORT REQUEST FORM
Reports can be requested for a variety of reasons; therefore, it is imperative to define the objectives and parameters
of the request—whether it is for a specific report, research study, or other type of analysis. The scope of the request
will determine the amount of detail needed to ensure that the request can be completed successfully and the
analysis will achieve the desired results. Before designing and running a requested report, analysts should
document and review the requirements with the report’s requester or end user. Doing so is critical to providing
accurate information and avoiding rework. Below are a list of guidelines to consider when developing a report
request form or designing a study.

1. Defining the study or report objectives

a. If multiyear data are requested, will trends be analyzed? If so, then the consistency and completeness
of the data for the entire period of the study must be considered.

b. If multisite data are requested, thresholds need to be set to determine the eligibility of each site for the
request. If hospital characteristics are important to the study (e.g., teaching, size, geographic location),
then distribution of sites and data according to these characteristics needs to be considered when
determining the final study population.

c. What requirements are needed to meet privacy and security requirements for requests involving
patient-level data? Will the data set needed for analysis qualify as a limited data set? Will institutional
review board approval be needed? If the data are not already de-identified, what method will be used
to mask protected health information data elements?

2. Identifying the correct study or report population

a. How will the population be identified for the request?

b. What are the sources of the data used to fulfill the request? If data are needed from multiple sources,
how will the data be linked or merged for the final analysis data set?

c. What types of coded data will be needed (e.g., ICD-10-CM, CPT, LOINC, SNOMED CT, RxNorm,
UB codes)?

d. Are there any known limitations in the data set that would affect the interpretation of the results of
the report or analysis?

3. Determining the calculations and statistical tests needed for the request

a. Will additional derived data fields be needed? What raw data will be needed to create the new
data fields?

b. What types of descriptive and inferential statistics will be used? Define the software program used
to create the statistics and which tests will be used.

4. Designing the presentation

a. How will the data be presented (charts, graphs, tables)?

b. What format will be used for data presentation (electronic media, hard copy)?

For more guidance on designing a data request report form, see Appendix B.

HEALTH DATA ANALYSIS TOOLKIT

10 | AHIMA

DATA CAPTURE METHODS AND BEST PRACTICES
With the advent of new electronic data capture methods such as mobile health apps, patient self monitoring,
patient portals, and health information exchange, there has been a dramatic increase in the ways that healthcare
organizations acquire patient data for use in the EHR.

The patient’s administrative and clinical data is captured at various sources inside and outside the healthcare
organization. The process is typically initiated by registering the patient. At the point of registration, data such
as the patient’s demographic and insurance information is captured. As the patient moves throughout the visit,
additional data is collected at each care site.

Capturing electronic discrete data elements can include the following: pre-defined or custom-built templates,
electronic forms with or without drop-down menus, use of bar coding technology, direct entry into free text
fields, front-end or back-end speech recognition with or without applied natural language processing, traditional
dictation, and transcription. Unstructured data is also captured such as handwritten notes and scanned images.

External data is captured and brought into the health record as well. One trend that is on the rise is patient-
generated data, generated when patients enter their data into the healthcare provider’s system via templates, drop-
down menus, or free text fields.

Once the data is collected, many tools may be used to enhance the capture process. Examples include the use of
optical character recognition (OCR) software to read and parse data, such as in the remittance advice process or
the use of natural language processing (NLP) or natural language understanding (NLU) in computer-assisted
coding (CAC) or transcription processes.

Data workflows vary depending upon the care setting. The workflow in physician practice is more streamlined
when compared to the hospital outpatient or inpatient processes. In all care settings, charges are captured through
a coding classification system such as ICD-9-CM/PCS, CPT, or HCPCS. The data is then transferred to a claim
form such as the UB-04 for inpatient or outpatient or the CMS-1500 for professional fees.

Best practices for EHR data capture should incorporate the following:5

• Consider what data needs to be captured and customize available tools to collect it.

• Evaluate the data and determine its placement in the record to determine what rules or procedures
need to be put in place to upload the information most efficiently and without errors.

• Collect the data in a standardized format using templates or discreet fields to make retrieval for
reporting easier.

• Routinely audit a sample of records that were collected using the data capture methods described above.

• Acquire primary and secondary data from existing internal or external data sources.

5. White, Susan. A Practical Approach to Analyzing Healthcare Data. Chicago: AHIMA Press, 2013.

HEALTH DATA ANALYSIS TOOLKIT

11 | AHIMA

VALIDATING DATA OUTCOMES
Validation of data can occur at many different points in the capturing, storing, managing, reporting, and analyzing
process. The validation steps outlined below focus on validating the outcomes during the reporting and perform-
ing of analysis tasks.

A. Data Extraction and Aggregation

1. Identify the source of all the data elements for the request.

2. When a data dictionary is available for the data elements being used, become familiar with how the raw
data were collected and identify any potential issues with the data. For example, if your analysis is based
only on Medicare patients, are you identifying these patients by using a data field that is complete and
up to date?

3. If data are being merged or linked across multiple sources, create a plan to ensure that the linkage is
correct. If there is a need to access laboratory results from the laboratory information system and link to
patients in the billing system, what method will be used to match the patient and encounter informa-
tion? Create a frequency of matched patients to determine that the match rates make sense (e.g., should
all patients in your analysis population have at least one laboratory test).

4. For classification systems used to aggregate data, verify that the codes selected are appropriate for the
time frame of the study. If the analysis period crosses times when the classification system has changed,
there may be codes that are valid for only part of the period. For example, ICD codes are updated every
October 1. If your study spans a calendar year, you may need to include codes that appear for discharges
only in the last quarter of the year that are consistent with the codes that were used in the first three
quarters of the year.

B.

Calculations and Statistics

1. When creating derived data fields, the analyst must determine a method to verify the results of the
calculations. Derived fields must be reviewed before performing any further statistical tests.

2. Create a plan for verifying statistics used in the request.

a. Do individual row percentages sum to 100?

b. Do individual row counts sum to the grand totals?

c. For complex data sets or statistical tests, determine a method to verify the results. Is there another
staff member who can review program code or output, or are there other means to determine that
tests were performed correctly?

d. Identify references, such as reports from other departments for similar types of data or literature
searches on the topic being analyzed that can be used to verify results. For example, if you are
analyzing costs per member per month (PMPM) stratified according to service level, check the
total PMPM against what is being reported by finance for the same or a similar period. If you are
studying a particular disease or condition, reviewing published literature or external databases for
outcomes such as mortality rates, length of stay (LOS), or readmissions can serve as a reference
point in validating the results.

HEALTH DATA ANALYSIS TOOLKIT

12 | AHIMA

C. Presentation of Data

1. Ensure data are represented in accordance with request. Options for presentation may include but are
not limited to:

a. Pie chart

b. Line graph

c. Bar graph

2. Verify that all labels on charts and graphs are clear and readable to the user.

3. Compare the graph back to the source data used to create the graph, and make sure that the graph
correctly represents the data.

4. Provide any additional information regarding inclusions or exclusions in a legend or other notation.

5. Verify all file names, figures, and tables against any documentation produced for the analysis.

Validating data outcomes when fulfilling a data request, including understanding the source of the data, verifying
the calculations needed, and selecting the proper presentation methods, is critical to performing data analysis.
Data display examples will be explored further in the following section.

HEALTH DATA ANALYSIS TOOLKIT

13 | AHIMA

PRESENTING THE DATA
This section will touch on data analysis tools and highlight examples of poor data display.

PIVOT TABLES

Pivot tables are an excellent Excel tool to summarize data according to categories. For example, they may be used
to summarize charges according to department, counts of coded data elements, etc. Pivot tables also provide
flexibility for the end user or analyst to organize and filter the data in various ways before finalizing the analysis.
Below is a small example of a pivot table totaling diagnosis codes according to sex:

Count of Dx

Dx Female Male Grand Total

4242 1 1

4254 1 1 2

4280 1 1

4290 1 1

5119 1 1

7455 1 1

78341 1 1

7861 1 1

Grand Total 7 3 10

The following resources offer instruction on creating Microsoft pivot tables:

Microsoft resources:

• http://office.microsoft.com/en-us/excel-help/pivottable-reports-101- HA001034632.aspx

• http://office.microsoft.com/en-us/excel-help/pivottable-i-get-started-with-pivottable- reports-in-excel-
2007-RZ010205886.aspx

Non-Microsoft resources:

• http://www.brighthub.com/computing/windows-platform/articles/27415.aspx

• http://www.youtube.com/watch?v=7zHLnUCtfUk

• http://www.youtube.com/watch?v=i67XK3qjL_w

HEALTH DATA ANALYSIS TOOLKIT

14 | AHIMA

CREATING A FREQUENCY TABLE

Frequency tables are another useful Excel tool to summarize a set of data, specifically by recording how often each
value (or set of values) occurs. As an added benefit, frequency tables can use percentages to further enhance the
results. Below are instructions for how to create a frequency data table by using raw diagnostic or procedure data.

1. Place all diagnostic or procedure data into column A in an Excel spreadsheet.

2. Click on the cell to the right of the first number in the column.

3. On the Formula menu, choose More Functions, Statistical and then COUNTIF.

4. For the Range, highlight the entire group of cells that contain your data.

5. For the Criteria, highlight the first number in the column. This places the count of the number of times this
code is present in the data into the cell where the pointer was located.

6. To copy this formula to count the remainder of the data, first enter a $ before the column letter and row
number in both cell addresses. For example, if the address looks like this: A2:A4000, make the range look
like this: $A$2:$A$4000 (shortcut: highlight entire formula [A2:A4000] and then select F4).

7. Next, copy this formula into the cell just to the right of each number to be counted (in this case, from A3
though A4000). This formula counts how many times each code is present, but the count is listed as the
result of a formula.

8. To translate this to a value, copy all of the formula results (highlight and Control C). Then, use Paste
Special from the Home menu and click on Values, which copies the formula results into the same location
but as actual numbers that can be sorted or calculated.

9. Highlight the entire spreadsheet by clicking on the upper left corner. On the Data tab, in the Data Tools
group, click on Remove Duplicates. Uncheck everything but column A in the selection window and
click OK.

This procedure results in a listing of each code present in the data and the corresponding count from the original
data. Below is a small example of a frequency table created for diagnosis codes. You can sort these entries by either
column A for number order or column B for frequency order.

Dx frequenCy

0092 1

135 1

243 4

311 3

319 12

0380 1

0382 1

0388 3

0389 10

OTHER RESOURCES

For more information on data analysis tools, such as statistical calculators, please refer to the “Calculations and
Statistics” section of the Annotated Bibliography on page 41.

HEALTH DATA ANALYSIS TOOLKIT

15 | AHIMA

EXAMPLES OF POOR DATA PRESENTATION AND SUGGESTED ALTERNATIVES

Numbers are factual data. However, when displayed incorrectly, they can be deceiving. Incomplete or misrep-
resented data may leave the reader with unanswered questions; therefore, healthcare data analysts must choose
graphic representations that are appropriate to the information they are presenting. They should follow established
guidelines for data visualization, such as using pie charts only for data that add up to a meaningful total. When
displaying counts according to category, the categories being compared should be the same size; the size of the
image displayed should be related to the corresponding data value; and data values must match other components
of the graphic display. In the examples that follow, poor practice has been identified and guidance provided for
improved data display.

ExamplE 1

Pie charts should be used only for data that add up to a meaningful total. In this first example, the sections do not.

73.8% 61.7%

66.7% 64.7%

62.2%

Education Level of Attendees at Conferenc

e

GED

High School

Bachelor’s

Master’s

PhD/MD

In comparison, the component parts of the pie chart below total 100 percent of the whole.

38%

1

1%

19%

31%

1%
Education Level of Attendees at Conference
GED
High School
Bachelor’s
Master’s
PhD/MD

HEALTH DATA ANALYSIS TOOLKIT

16 | AHIMA

ExamplE 2

In this example, time is displayed backward on the x-axis, and the three-dimensional effect is difficult to interpret:
does the scale match the front or the back of the bars? Furthermore, the high values on the community college
scale minimize the important variation in the other two series.

Public Universities Community Colleges Private Institutions

700,000

600,000

500,000

400,000

300,000

200,000

100,000

0
2000 1999 1998 1997 1996 1995

Total 12-Month Headcounts

Below is a sample of how the same data could be displayed differently, omitting community colleges.

Public Universities Private Institutions

210,000

200,000

190,000

1

80,000

1995 1996 1997 1998 1999 2000

Fall Headcount Enrollment, Illinois Public Universities and
Private Institutions: 1995–2000

HEALTH DATA ANALYSIS TOOLKIT

17 | AHIMA

ExamplE 3

In this example, the bin width is not consistent across the samples (1 to 10 versus 10 to 24).

Not Yet Paperless
The majority of office workers print up to 24 pages a day.

40%
38%

12%
10%

1 – 10 10 – 24 25 – 49 50+

When displaying counts by category, the categories being compared should be the same size.

ExamplE 4

Here, the proportionality is off—the marker representing “easel paper” is not half the length of the marking
representing “pen/pencil/paper.”

Pens Still Popular
More than half of workers take meeting notes with pen and paper.

Pen/Pencil/Paper

Erasable Marker Board

Laptop

Easel Paper

53%

35%

27%

23%

HEALTH DATA ANALYSIS TOOLKIT

18 | AHIMA

Health data analysts must always ensure the size of the image displayed is related to the corresponding data value.
For example, if the symbols are different widths, then the width should have an interpretation. In the example,
even if the symbols had the same width, the length of the pencil representing “easel paper” should be half as long
as the “pen/pencil/paper” symbol.

PREDICTIVE MODELING

Predictive modeling applies statistical techniques to determine the likelihood of certain events occurring
together.6 Statistical methods are applied to historical data to “learn” the patterns in the data. These patterns are
used to create models of what is most likely to occur.

Predictive modeling is used by credit card issuers to determine if transactions are likely fraudulent. Customers
who receive a phone call from their credit card company verifying that they authorized a transaction were the
subjects of a predictive model.

For example, a customer’s typical credit card transaction is $100. The credit card issuer notices that the customer
submitted three $5,000 transactions in one day. Given the customer’s history and the credit card issuer’s historical
data regarding fraudulent transactions, those transactions look suspicious.

The credit card company may then put a hold on the card and call to verify that the customer really did authorize
the suspect transactions. The triggers that tell the credit card company when to suspect a fraud issue are created
via predictive modeling techniques.

Predictive modeling techniques use multiple data sources. Data such as the provider’s claim history, the patient’s
demographics and health status, the services included on the claim, and the attributes associated with previously
identified fraudulent claims may all be used to develop a statistical model.

Statistical techniques used to create the model may include logistic regression, cluster analysis, or decision trees.
All of these statistical techniques allow the user to combine multivariate historical data into a model that may be
used to assess the probability or likelihood that current claims are fraudulent.

In logistical regression, the likelihood that a claim is fraudulent is estimated based on a series of historical data.
In cluster analysis, historical data are used to build a model that will measure the “distance” of a claim from the
typical claims submitted by that provider or for that type of service. Decision trees use a series of screens or yes/no
questions to determine the probability that a claim is valid.

The output of each of these methods is the probability of a claim’s validity that is expressed as a score.

The claim score is typically structured so that it is directly related to the probability that a claim is in error. A high
score may indicate a high probability that a claim is not legitimate. If the score meets a criteria (either above or
below a cutoff value), then it is identified as a potential error.

The criteria or cutoff value may be used to tune the model to control the sensitivity and specificity of the model.
If the cutoff is too extreme, then the model may not be sensitive enough and will allow fraudulent claims to be
paid. If the cutoff is not extreme enough, then the model may not be specific enough and identify a large number
of false positives.

In the healthcare setting, clinicians can use predictive modeling to improve patient care—one example would be
to use predictive modeling to prevent readmission. For example, if you know the patient has a certain medical
condition and lives alone, you might predict that they are a higher risk for readmission, but through prediction,
the clinician can take preventive action and possibly mitigate the readmission risk.

6. White, Susan. “Predictive Modeling 101.” Journal of AHIMA vol. 82, no. 9 (September 2011): 46–47.

HEALTH DATA ANALYSIS TOOLKIT

19 | AHIMA

FORMULAS AND STATISTICS
Statistical analysis typically is segmented into two areas: descriptive and inferential. Descriptive statistics, as the
name would imply, are used to describe the characteristics of a set of data. They include measures of central
tendency and measures of variation or dispersion. Inferential statistics are used to make decisions on the basis
of sampled data, and they include measures such as confidence intervals (CIs) and hypothesis testing. Both are
discussed in more detail in this section.

Descriptive Statistics

MEASURES OF CENTRAL TENDENCY

These statistics measure the center of a distribution. They are well suited for describing the typical value of a
particular data element.

MEAN

Also known as:
Arithmetic average, arithmetic mean, avera

ge

Notation:
x = n

i = 1xi
n∑

or x-bar is the typical symbol used for the mean.

Properties:
The mean is the most common measure of central tendency. It is appropriate to use for continuous variables
(charges, LOS, systolic blood pressure). The mean is not appropriate for use with nominal variables (categories in
which order does not convey information, such as sex, race, or CPT codes). The mean can be influenced by outliers
and may not be the best choice of statistic for a heavily skewed variable.

Calculation:
The mean is the sum of the observations divided by the number of observations

x = n
i = 1xi
n∑

where xi is the ith observation and n is the number of observations to be averaged. Example:

The average of the observations 2, 6, 9, 1 is x = = 4.54
(2 + 6 + 9 + 1) .

Excel Command:
=average(a1..a4), where a1..a4 refers to the data cells

HEALTH DATA ANALYSIS TOOLKIT

20 | AHIMA

MEDIAN

Also known as:
50th percentile, 2nd quartile, middle value

Notation:
There is no standard notation for the median. The symbol or x-tilde is sometimes used.

Properties:
The median is less influenced by outlier observations than is the mean.

Calculation:
To calculate the median of a series of observations, sort them from smallest to largest and choose the middle value
as the median. If there is an even number of observations, then the median is the average of the two middle values.

Example:
To find the median of the observations 2, 6, 9, 1, first order the values (1, 2, 6, 9). There are an even number of
observations (4); therefore, we average the two middle values to calculate the median:

x = = 42
(2 + 6)

Excel Command:
=median(a1..a4) or =quartile(a1..a4,2), where a1..a4 refers to the data cells

Note that Excel and other software packages may each use different algorithms to estimate the median, so the
results for an even numbered set of values may not match the hand-calculated median.

MODE

Also known as:
Value with the highest frequency or most “popular” value

Notation:
There is no standard notation for the mode.

Properties:
The mode is appropriate for use with categorical variables (sex, diagnosis-related group [DRG], etc.). The mode
should not be used to describe the center of the distribution of a continuous variable (charges, time, etc.).

Calculation:
To calculate the mode of a set of observations, count the frequency of each value and select the value with the
highest frequency. There may be multiple values of the mode if two or more values have the same (maximum)
frequency. If there are two modes, the distribution is called “bimodal.”

Example:
To find the mode of the observations 2, 6, 9, 1, 2, note that each value except 2 occurs once. The value 2 occurs
twice and is, therefore, the mode.

Excel Command:
=mode(a1..a4), where a1..a4 refers to the data cells

x = = 42
(2 + 6)

HEALTH DATA ANALYSIS TOOLKIT

21 | AHIMA

PERCENTILES

Also known as:
25th percentile = first quartile; 50th percentile = second quartile or median; 75th percentile = third quartile.
The minimum and the maximum are special cases of the percentile.

Notation:
pk, where k is the k

th percentile. For instance, p5 = 5
th percentile.

Properties:
Percentiles are most appropriate for continuous variables (ratio or interval). The minimum and maximum
obviously are influenced heavily by outliers. Quartiles and midpercentiles are not heavily influenced by outliers.
Quartiles often are used in benchmarking financial data.

Calculation:
To calculate percentiles of a distribution, you should first order the values. The kth percentile is a value, pk, such
that most (100k)% of the measurements are less than this value and most 100(1 – k)% are greater. First calculate
kN/100, where k is the kth percentile and N is the number of observations. If kN/100 is not an integer, then round
up to the next integer. The kth percentile is the kth ordered observation. Computer programs such as Excel and
SPSS may use different algorithms to estimate percentiles. Hand calculations likely will give a slightly different
answer.

Example:
To find the 70th percentile or p70 of the observations 2, 6, 9, 1, 2, first calculate kN/100 where k is the k

th percen-
tile and N is the number of observations (70*5/100 = 3.5). Round up to an integer: 4. The kth percentile is the
kth ordered observation; thus the 70th percentile of 1, 2, 2, 6, 9 is p70 = 6.

Excel Command:
=percentile(a1..a4, k), where a1..a4 refers to the data cells and k is the percentile desired

HEALTH DATA ANALYSIS TOOLKIT

22 | AHIMA

GEOMETRIC MEAN

Also known as:
GM (as in GMLOS or geometric mean length of stay as reported by Centers for Medicare and Medicaid
Services [CMS])

Notation:
pk , where k is the k

th percentile. For instance, p5 = 5
th percentile.

Properties:
In the geometric mean, multiplication is used to summarize the variables. The geometric mean is appropriate to
use for positive continuous variables. The geometric mean is always smaller than the arithmetic mean or average.
The geometric mean is less influenced by large positive outliers than is the arithmetic mean; therefore, if a distribu-
tion has a long tail or is positively skewed, the geometric mean is a good measure of the center of the distribution.
It is fairly common in the finance industry. CMS uses the geometric mean to summarize lengths of stay according
to Medicare severity-adjusted DRG (MS-DRG). Lengths of stay are positive variables and tend to have a long tail
(a few patients have very long stays).

Calculation:
The geometric mean is the nth root of a series of n observations. The formula for the GM is GM = xi

nn
i = 1√ ∏ ,

where n is the number of observations, or equivalently GM = xi
n
i = 1( )∏

1/n, where n is the number of observations.

(The 1/n power and the nth root are mathematically equivalent.)

Example:
To find the GM of the observations 2, 6, 9, 1, 2:

2x6x9x1x2 = √
5 216 = 2.93√

5

Excel Command:
=geomean(a1..a4, k), where a1..a4 refers to the data cells and k is the percentile desired

HEALTH DATA ANALYSIS TOOLKIT

23 | AHIMA

Measures of Variation or Dispersion

SAMPLE OR POPULATION VARIANCE

Also known as:
Most reported variance values are based on a sample and therefore should be referred to as the “sample variance.”
In practice, the shorthand “variance” typically is used.

Notation:
Population variance: σ 2
Sample variance: s2

Properties:
The population variance is the average of the squared deviations from the population mean. Since the values of
any variable rarely are known for the entire population, the sample variance typically is used. The deviation (or
difference) between each value and the mean is squared so that the “typical” deviation can be summarized regard-
less of the sign (or direction) of the deviation. The sum of the deviations from the sample mean is actually zero
for any sample of observations, so the “squaring” allows the typical deviation to be quantified. The sample variance
is somewhat influenced by outliers. The variance is an appropriate measure of spread for continuous variables
(interval or ratio).

Calculation:
s2 = i = 1

n (xi – x)2∑ n – 1 `
where x = n

i = 1xi
n∑

is the sample mean and n is the number of observations.

Example:
To find the variance of the observations 2, 6, 9, 1, first calculate the sample mean, which is 4.5 (see above).

s2 = = = 13.67 (2 – 4.5)
2 + (6 – 4.5)2 + (9 – 4.5)2 + (1 – 4.5)2

4 – 1
41
3

Excel Command:
Population variance (denominator is n instead of n – 1):
=varp(a1..a4, k), where a1..a4 refers to the data cells

Sample variance:
=var(a1..a4, k), where a1..a4 refers to the data cells

HEALTH DATA ANALYSIS TOOLKIT

24 | AHIMA

STANDARD DEVIATION

Also known as:
Most reported standard deviation values are based on a sample and therefore should be referred to as the “sample
standard deviation.” In practice, the shorthand “standard deviation” typically is used.

Notation:
Population standard deviation: σ
Sample standard deviation: s

Properties:
The sample standard deviation is the square root of the sample variance. The units of measure of the variance is
“squared units” (squared days, square inches, etc.) and is therefore less intuitive to apply than is the standard devia-
tion. The standard deviation (sample or population) has the same unit of measure as the original observations. The
sample standard deviation is somewhat influenced by outliers. The standard deviation is an appropriate measure of
spread for continuous variables (interval or ratio).

Calculation:
s2 = s2√

where s2 is the sample standard deviation.

Example:
To find the variance of the observations 2, 6, 9, 1, first calculate the sample mean, which is 4.5 (see above).

s = 13.67 = 3.70√
Excel Command:
Population standard deviation:
=stdevp(a1..a4, k), where a1..a4 refers to the data cells

Sample standard deviation:
=stdev(a1..a4, k), where a1..a4 refers to the data cells

HEALTH DATA ANALYSIS TOOLKIT

25 | AHIMA

RANGE

Also known as:
Range of values, spans

Notation:
There is no standard notation for the range.

Properties:
The range is the maximum (largest) value minus the minimum (smallest) value of a series of data points.
The range is heavily influenced by outliers. The range is an appropriate measure of spread for most variable types
(continuous and discrete). It is not appropriate for use with nominal data where order does not have a meaning
(minimum and maximum would not have a meaning for that variable type either).

Calculation:
Range = maximum – minimum

Example:
To find the range of the observations 2, 6, 9, 1, first order the observations from smallest to largest: 1, 2, 6, 9.
Note that the minimum value is 1 and the maximum value is 9. The range is, therefore, 9 – 1 = 8.

Excel Command:
There is no Excel command to calculate the range directly. You must use the min and max functions:
=max(a1..a4, k) – min(a1..a4, k), where a1..a4 refers to the data cells.

STATISTICAL INFERENCE

“Statistical inference is the process of making conclusions using data that is subject to random variation, for
example, observational errors or sampling variation.”7

“More substantially, the terms statistical inference, statistical induction and inferential statistics are used to
describe systems of procedures that can be used to draw conclusions from datasets arising from systems affected
by random variation. Initial requirements of such a system of procedures for inference and induction are that the
system should produce reasonable answers when applied to well-defined situations and that it should be general
enough to be applied across a range of situations.” 8

These definitions cover most of the aspects of statistical inference, which is basically a set of statistical methods
that is used to make decisions or conclusions on the basis of sample data.

That decision typically has a probability of error associated with it. The two most common types of statistical
inference used in practice are confidence intervals and hypothesis tests.

7. Upton, Graham., and Ian Cook. Oxford Dictionary of Statistics. New York: Oxford University Press, 2008.
8. Dodge, Yadolah, ed. The Oxford Dictionary of Statistical Terms. New York: Oxford University Press, 2003.

HEALTH DATA ANALYSIS TOOLKIT

26 | AHIMA

Confidence Intervals
A confidence interval (CI) is a range of values that has a set probability or confidence level of containing the
population value of the statistic of interest. CIs are encountered quite often in practice. Opinion polls are reported
in the news with an associated +/– value. For instance, a 2010 Gallup poll reported that Democrats held an
advantage over Republicans of 48 percent to 44 percent in a generic ballot poll. Each Gallup poll includes a section
titled “Survey Methods.” For this survey, Gallup reports:

Results are based on telephone interviews conducted as part of Gallup Daily tracking survey July 19–25, 2010,
with a random sample of 1,633 registered voters, aged 18 and older, living in all 50 U.S. states and the District
of Columbia, selected using random-digit-dial sampling.

For results based on the total sample of registered voters, one can say with 95% confidence that the maximum
margin of sampling error is ±4 percentage points.9

In effect, Gallup is 95 percent certain that the percentage for Democrats is between 44 percent and 52 percent and
the percentage favoring Republicans is between 40 percent and 48 percent. When stated in this manner, it is clear
that this is a “statistical tie.”

The CI is the probability of the interval covering the true value. In our example, Gallup set the confidence level at
95 percent for the percentage of voters favoring each party.

The width of the CI (the portion after the +/–) typically is referred to as the “precision.” A narrower interval or
smaller +/– value is more precise. The width of a CI depends on:

1. Sample size: increased sample size will result in increased precision.

2. Standard deviation of the variable: smaller standard deviation results in increased precision.

3. Decreased confidence level: decreased CI results in increased precision.

The formula to determine a CI depends on the statistic used as the estimator (percentage, arithmetic mean, etc.).
The basic form of a CI is typically:

A(1–a)% Confidence Interval = (estimator) ± (critical value) × (standard error of estimator)

The critical value depends on the distribution of the estimator. If the arithmetic mean is used as the estimator,
then the critical value is based on the normal or t-distribution. A larger confidence level typically results in a larger
critical value and a wider interval.

More details on this topic can be found in most introductory statistics textbooks or via a quick Internet search.

9. Newport, Frank. “Democrats Maintain Advantage on Generic Ballot, 48% to 44%.” Gallup Politics, July 26, 2010.
http://www.gallup.com/poll/141557/Democrats-Maintain-Advantage-Generic-Ballot.aspx.

HEALTH DATA ANALYSIS TOOLKIT

27 | AHIMA

Hypothesis Testing
In hypothesis testing, the analyst is trying to make a decision, on the basis of sample data, between two hypotheses:
the null hypothesis and the alternative hypothesis. The null hypothesis is typically the “status quo” and requires no
action. The alternative hypothesis is sometimes called the “research hypothesis” and typically requires some action.

A test statistic is calculated to determine if the null hypothesis should be rejected or not on the basis of the
evidence presented in the sample data, which is best illustrated with a simple example. Suppose the state average
LOS for acute myocardial infarction is 4.2 days. The marketing department of Major Hospital would like to claim
that their LOS is shorter than the state average. From a 10-patient sample during the last year, an average LOS
of 3.5 days was calculated with a standard deviation of 1.1 days. A simple hypothesis test can help the marketing
director understand if the data support that claim. The null hypothesis (Ho) is that the LOS at Major Hospital is at
least as long as 4.2 days (state average), and the alternative hypothesis is that the LOS at Major Hospital is less than
4.2 days.

This situation calls for a t-test to be performed. After performing the t-test, compare the value of the test statistic
to its distribution to determine the P value. The P value is the probability of obtaining a test statistic that large by
chance. Alternatively, the P value is the probability of being wrong in concluding that the null hypothesis is false.
In our example, the P value is the probability of being incorrect when claiming Major Hospital’s LOS for patients
with acute myocardial infarction is lower than the state average on the basis of the sample data presented.

The P value for this sample is 0.04, or 4 percent. Thus, if the marketing director is willing to take a 4 percent chance
of making an incorrect conclusion, she can go forward and make the claim that Major Hospital’s LOS is shorter.

Often an alpha level is set before performing a hypothesis test. The alpha level should be based on the cost of
incorrectly rejecting the null hypothesis. In clinical studies, the alpha level is typically set low (1 percent or 5 
percent). Since the P value is the probability of incorrectly rejecting the null hypothesis, the null hypothesis
should be rejected if the P value is smaller than the alpha level.

More details on this topic can be found in most introductory statistics textbooks or via a quick Internet search.

HEALTH DATA ANALYSIS TOOLKIT

28 | AHIMA

SAMPLE JOB SKILLS AND RESPONSIBILITIES
This section discusses potential job skills and responsibilities required by a health data analyst at entry, middle,
and senior levels. Please note, this is not an all-inclusive listing; rather, it is a guide for current professionals in
these roles or those with an interest in further career opportunities.

ENTRY LEVEL

Education:
BA/BS (preferably in health information management, health services research, or health administration)

Experience:
Prior healthcare experience required. Previous health data analyst and report writing experience preferred.

Skills and Abilities:
• Has basic understanding of coding systems (CPT, DRG, ICD-9-CM, ICD-10-CM/PCS,

National Drug Code [NDC])

• Demonstrates strong verbal and written communication skills

• Demonstrates excellent organizational and time management skills

• Exhibits keen attention to detail and problem-solving skills

• Knows Microsoft Office, especially Excel and Access

Preferred: RHIA, CHDA, master’s degree (health administration, health informatics, or similar), knowledge of
Structured Query Language (SQL)

Typical Job Description:

Responsibilities
Daily Operations

• Identify data problem areas and conduct research to determine best course of action

• Analyze and solve issues with legacy, current, and planned systems as they relate to the integration and
management of patient data (e.g., review for accuracy in record merge and unmerge processes)

• Analyze reports of data duplicates or other errors to provide ongoing appropriate interdepartmental
communication and monthly or daily data reports (e.g., related to the enterprise master patient index [EMPI])

• Monitor metadata for process improvement opportunities (e.g., monitoring orders for successful
computerized physician order entry (CPOE) implementation)

• Identify, analyze, and interpret trends or patterns in complex data sets

• Monitor data dictionary statistics

Data Capture

• In collaboration with others, develop and maintain databases and data systems necessary for projects
and department functions

• Acquire and abstract primary or secondary data from existing internal or external data sources

• In collaboration with others, develop and implement data collection systems and other strategies
that optimize statistical efficiency and data qual

ity

• Perform data entry, either manually or using scanning technology, when needed or required

HEALTH DATA ANALYSIS TOOLKIT

29 | AHIMA

Data Reporting

• In collaboration with others, interpret data and develop recommendations on the basis of findings

• Develop graphs, reports, and presentations of project results, trends, data mining

• Perform basic statistical analyses for projects and reports

• Create and present quality dashboards

• Generate routine and/or ad hoc reports

MID-LEVEL

Education:
Master’s degree in health administration, public health, health informatics, or similar

Experience:
• Minimum of five years of experience in data management and analysis in a healthcare‚ managed care, or

insurance setting required; strong data manipulation techniques and proficiencies required; analytical expe-
rience developing recommendations from claims data analysis

• Strong quality control ethic, ability to assess accuracy of output at report level, as well as investigate potential
issues at a micro level

• Firm understanding of the nuances of the managed care business structure and the ability to apply this
knowledge to analytical projects

• Strong problem solving and analytical skills

Skills and Abilities:
• Strong verbal and written communication skills with experience interacting with senior management

• Ability to present complex information in an understandable and compelling manner

• Project management experience

• Proficiency in Microsoft Word, Excel, Access, and PowerPoint

• Experience using SAS, SPSS, or other statistical package is desirable for analyzing large data sets

• Programming skills preferred, adept at queries and report writing, knowledge of SQL

• Knowledge of statistics, at least to the degree necessary to communicate easily with statisticians

• Experience in data mining techniques and procedures and knowing when their use is appropriate

• Knowledge of coding classification systems and terminologies

• Understanding of risk adjustment models

• Familiarity with outcome measurements

• Preferred: CHDA, project management certification, budgeting experience, understanding of
database design

HEALTH DATA ANALYSIS TOOLKIT

30 | AHIMA

Typical Job Description:
• Work collaboratively with data and reporting and the database administrator to help produce effective pro-

duction management and utilization management reports in support of performance management related
to utilization, cost, and risk with the various health plan data; monitor data integrity and quality of reports
on a monthly basis

• Work collaboratively with data and reporting in monitoring financial performance in each health plan

• Develop and maintain claims audit reporting and processes

• Develop and maintain contract models in support of contract negotiations with health plans

• Develop, implement, and enhance evaluation and measurement models for the quality, data and reporting,
and data warehouse department programs, projects, and initiatives for maximum effectiveness

• Recommend improvements to processes, programs, and initiatives by using analytical skills and a variety of
reporting tools

• Determine the most appropriate approach for internal and external report design, production, and distribu-
tion, specific to the relevant audience

SENIOR LEVEL

Education:
Master’s degree in health administration, public health, health informatics, or similar

Experience:
• Minimum of five years of experience in healthcare analysis required, preferably within a large tertiary care

hospital, academic medical center, or other large medical institution; background in quality improvement,
health statistics, health services research, or healthcare outcomes research strongly preferred

• Two to four years of experience developing SQL or Procedural Language (PL)/SQL programs, preferably
with Oracle Database, and developing reports by using Crystal Reports (Business Objects) or similar presen-
tation tools

• Experience leading system-wide improvement projects in a matrix environment

• Minimum of three years of demonstrably successful managerial experience

Skills and Abilities:
• Strong verbal and written communication skills, with experience interacting with senior management and

medical leadership

• Ability to present complex information in an understandable and compelling manner

• Understanding of healthcare finance and claims systems

• Ability to foster cooperation in a highly charged political environment

• Knowledge of No-SQL and/or Hadoop for large data sets

• Knowledge of SAS and SPSS

• CHDA preferred

HEALTH DATA ANALYSIS TOOLKIT

31 | AHIMA

Typical Job Description:
• Understand and address the information needs of governance, leadership, and staff to support continuous

improvement of patient care processes and outcomes

• Lead and manage efforts to enhance the strategic use of data and analytic tools to improve clinical care
processes and outcomes continuously

• Work to ensure the dissemination of accurate, reliable, timely, accessible, actionable information (data
analysis) to help leaders and staff actively identify and address opportunities to improve patient care and
related processes

• Work actively with information technology to select and/or develop tools to enable facility governance
and leadership to monitor the progress of quality, patient safety, service, and related metrics continuously
throughout the system

• Engage and collaborate with information technology and senior leadership to create and maintain a succinct
report (e.g., dashboard), as well as a balanced set of system assessment measures, that conveys status and
direction of key system-wide quality and patient safety initiatives for the trustee quality and safety commit-
tee and senior management; present this information regularly to the quality and safety committee of the
board to ensure understanding of information contained therein

• Actively support the efforts of divisions, departments, programs, and clinical units to identify, obtain, and
actively use quantitative information needed to support clinical quality monitoring and improvement
activities

• Function as an advisor and technical resource regarding the use of data in clinical quality improvement
activities

• Lead analysis of outcomes and resource utilization for specific patient populations as necessary

• Lead efforts to implement state-of-the-art quality improvement analytical tools (i.e., statistical
process control)

• Play an active role, including leadership, where appropriate, on teams addressing system-wide clinical quality
improvement opportunities

REFERENCE

AHIMA. “AHIMA Strategic Plan.” 2013.
http://library.ahima.org/xpedio/groups/public/documents/ahima/bok1_050165 .

ADDITIONAL RESOURCES

Agency for Healthcare Research and Quality. ”Data Sources Available from AHRQ.”
http://www.ahrq.gov/research/data/dataresources/.

Healthdata.gov. “CMS Medicare and Medicaid EHR Incentive Program, electronic
health record products used for attestation.” http://www.healthdata.gov/data/dataset/
cms-medicare-and-medicaid-ehr-incentive-program-electronic-health-record-products-used.

National Center for the Analysis of Healthcare Data. “Map Samples.” http://ncahd.org/map-samples.php.

HEALTH DATA ANALYSIS TOOLKIT

32 | AHIMA

GLOSSARY OF TERMS

CONTENTS

Calculations and Statistics …………………………………………………………………………………………………………32

Classifications and Terminologies …………………………………………………………………………………………….34

Clinical Quality Measurement

…………………………………………………………………………………………………..35

Database Terms

. …………………………………………………………………………………………………………………………..36

Finance and Reimbursement Terms

…………………………………………………………………………………………37

General

…………………………………………………………………………………………………………………………………………..39

Calculations and Statistics

Alpha level: The probability of making a type I error. The
alpha level should be based on the cost of incorrectly
rejecting the null hypothesis. In clinical studies, the alpha
level is typically set low (1 percent or 5 percent)

Analysis of variance (ANOVA): Test used to determine the
differences among two or more means

Average length of stay (ALOS): The mean length of stay for
hospital inpatients discharged during a given period

Bar chart: A graphic technique used to display frequency
distribution data that fall into categories

Chi-square test: A statistical calculation used to determine
whether proportions in a randomly drawn sample are
significantly different from the underlying or theoretical
population proportions

Confidence interval (CI): An interval that has a certain
probability (confidence level) of including the true value
of a population parameter. CIs may be calculated for pop-
ulation means, proportions, and standard deviations, etc.

Correlation: The existence and degree of linear relationship
among factors

Dependent variable: A measurable variable in a research
study that depends on an independent variable

Derived attribute: An attribute whose value is based on the
value of other attributes (e.g., current date minus date of
birth yields the derived attribute age)

Descriptive statistics: A set of statistical techniques used to
describe data such as means, frequency distributions, and
standard deviations; statistical information that describes
the characteristics of a specific group or a population

F test: The ratio of the between-group variance to the within-
group variance in the ANOVA procedure. If the F ratio is
statistically significant (the F value equals or exceeds the
critical value of F), the observed differences between the
group means of the independent variables under study
will be significantly different from each other. See also
analysis of variance (ANOVA)

Frequency distribution: A table or graph that displays the
number of times (frequency) a particular observation
occurs

Geometric mean length of stay (GMLOS): Statistically
adjusted value of all cases of a given DRG, allowing for
the outliers, transfer cases, and negative outlier cases that
normally would skew the data. The GMLOS is used to
compute hospital reimbursement for transfer cases

Hypothesis: A statement that describes a research question in
measurable terms

Independent variable: An antecedent factor that researchers
manipulate directly

Inferential statistics: Statistics that are used to draw conclu-
sions regarding a population parameter on the basis of a
sample

Length of stay (LOS): The total number of patient days for an
inpatient episode, calculated by subtracting the date of admis-
sion from the date of discharge. If the admission and discharge
are on the same day, the LOS generally is set to one day

Line graph: A graphic technique used to illustrate the rela-
tionship between continuous measurements; it consists of
a line drawn to connect a series of points on an arithmetic
scale and often is used to display time trends

HEALTH DATA ANALYSIS TOOLKIT

33 | AHIMA

Mean: A measure of central tendency that is determined by
calculating the arithmetic average of the observations in a
frequency distribution

Measures of central tendency: The typical or average num-
bers that are descriptive of the entire collection of data for
a specific population

Median: A measure of central tendency that shows the mid-
point of a frequency distribution when the observations
have been arranged in order from lowest to highest

Mode: A measure of central tendency that consists of the
most frequent observation in a frequency distribution

Normal distribution: A theoretical family of continuous
frequency distributions characterized by a symmetric
bell-shaped curve, with an equal mean, median, and mode;
any standard deviation; and half of the observations above
the mean and half below it

Null hypothesis: A hypothesis that states there is no associa-
tion between the independent and dependent variables in
a research study. The null hypothesis often represents the
status quo or a state of no statistical difference in a study

Pie chart: A graphic technique in which the proportions of
a category are displayed as portions of a circle (like pieces
of a pie)

P value: The probability that the observed difference could
have been obtained by chance alone, given random
variation and a single test of the null hypothesis

Qualitative analysis: In healthcare data, determining that the
data accurately portray the care that was administered and
that the content is correct

Quantitative analysis: In healthcare data, analyzing aggre-
gate data for patterns. Typically, quantitative analysis
requires numeric data for calculations

Range: A measure of variability that is the difference between
the smallest and largest observations in a frequency
distribution

Rank: Denotes a score’s position in a group relative to other
scores that have been organized in order of magnitude

Rate: A measure used to compare an event across time; it is a
comparison of the number of times an event did happen
(numerator) with the number of times an event could
have happened (denominator). See also ratio

Ratio: A calculation performed by dividing one quantity by
another. It is also a general term that can include a number of
specific measures such as proportion, percentage, and rate

Regression analysis: Statistical technique that uses an
independent variable to predict the value of a dependent
variable. In the inpatient psychiatric facility prospective
payment system (IPF PPS), patient demographics and
length of stay (independent variables) were used to predict
cost of care (dependent variable)

Sample: A set of elements drawn from and analyzed to
estimate the characteristics of a population

Sample size: The number of subjects needed in a study to
represent a population

Scatter diagram: A graph that visually displays relationships
among factors

Standard deviation: A measure of variability that describes
the deviation from the mean of a frequency distribution in
the original units of measurement—the square root of the
variance

Stratified random sampling: The process of selecting the
same percentages of subjects for a study sample as they
exist in the subgroups (strata) of the population

Systematic sampling: The process of selecting a sample of
subjects for a study by drawing every nth unit on a list

t-distribution: The t-distribution is a probability distribution
that is bell shaped and centered at zero, much like the
standard normal distribution. The distribution is defined
by the degrees of freedom, which determines the spread
(or width) of the distribution. The t-distribution typically
is used to determine the significance of a t-test of hypothe-
ses regarding a population mean

t-test: Assesses whether the means of two groups are statisti-
cally different from each other; appropriate when compar-
ing the means of two groups

Trend: A long-term movement in an ordered series, say a
time series, which may be regarded, together with the
oscillation and random component, as generating the
observed values1

Type I error: An error in which the researcher erroneously
rejects the null hypothesis when it is true

Type II error: An error in which the researcher erroneously
fails to reject the null hypothesis when it is false

Variable: A factor or quantity capable of assuming any of a
set of values

Variance: A measure of variability that gives the average
of the squared deviations from the mean—the difference
between the planned or expected value and the actual value

1. Dodge, Yadolah, ed. The Oxford Dictionary of Statistical Terms. New York: Oxford University Press, 2003.

HEALTH DATA ANALYSIS TOOLKIT

34 | AHIMA

Classifications and Terminologies

Classification: A clinical vocabulary, terminology, or nomen-
clature that lists words or phrases with their meanings;
provides for the proper use of clinical words as names
or symbols; and facilitates mapping standardized terms
to broader classifications for administrative, regulatory,
oversight, and fiscal requirements

Current Procedural Terminology (CPT): A coding system
used to report medical, surgical and diagnostic services
provided in the outpatient and provider settings. Created
and maintained by the American Medical Association.

Diagnosis-related group (DRG): A prospective payment
mechanism for hospital inpatients in which diseases are
placed into groups because related diseases and treatments
tend to consume similar amounts of healthcare resources
and incur similar amounts of cost; in the Medicare and
Medicaid programs, one of the more than 700 diagnostic
classifications in which cases demonstrate similar resource
consumption and length-of-stay patterns. Prior to October 1,
2007, DRGs were not severity adjusted and were referred to
as “CMS-DRGs.” After October 1, 2007, DRGs are referred to
as “Medicare severity-adjusted DRGs,” or “MS-DRGs.”

DRG grouper: A computer program that assigns inpatient cases
to DRGs and determines the Medicare reimbursement rate

Evaluation and management (E/M) codes: Codes con-
tained within CPT, developed by the American Medical
Association, used for documenting physician services for
billing purposes. These codes represent the level of service
performed by the physician

Healthcare Common Procedure Coding System (HCPCS):
A classification system to identify healthcare procedures,
equipment, and supplies for claim submission purposes.
There are two levels: level I CPT codes, developed by the
American Medical Association, and level II codes for
equipment, supplies, and services not covered by CPT
codes, as well as modifiers that can be used with either
level of codes, developed by CMS

International Classification of Diseases 9th Revision
Clinical Modification (ICD-9-CM): Coding and classi-
fication system used to report diagnoses in all healthcare
settings and inpatient procedures and services

International Classification of Diseases 10th Revision
Clinical Modification (ICD-10-CM): Coding and classi-
fication system used to report diagnoses in all healthcare
settings and inpatient procedures and services

Logical Observation Identifiers Names and Codes
(LOINC): A database protocol aimed at standardizing
laboratory and clinical codes for use in clinical care,
outcomes management, and research. LOINC is currently
required by CMS for identifying and reporting laboratory
tests under Meaningful Use. Developed and maintained
by the Regenstrief Institute for Health Care. Regenstrief
and the International Health Terminology Standards
Organisation (IHTSDO) entered into an agreement in
2013 to develop a linkage between LOINC and SNOMED

Major complication or comorbidity: A complication or
comorbidity that increases the severity of a patient’s
condition and allows for assignment of an inpatient case
to the highest-weighted DRG in the MS-DRG system

Major diagnostic category (MDC): Under DRGs, one of 25 cat-
egories based on single- or multiple-organ systems into which
all diseases and disorders relating to that system are classified

Medicare severity-adjusted DRG (MS-DRG): See diagnosis-
related group (DRG)

National Drug Code (NDC): Codes used to describe a drug
on the basis of the manufacturer, the drug name, and the
package size and type

RxNorm: Provides normalized names for clinical drugs
and links those names to many of the drug vocabularies
commonly used in pharmacy management and drug
interaction software.These systems provide a link to the
RxNorm identifier using the NDC. RxNorm is currently
required by CMS for reporting pharmacy data under
Meaningful Use. It is maintained by the National Library
of Medicine (NLM)

Systematized Nomenclature of Medicine—Clinical Terms
(SNOMED CT): A comprehensive clinical terminology, orig-
inally created by the College of American Pathologists and,
as of April 2007, owned, maintained, and distributed by the
International Health Terminology Standards Development
Organisation (IHTSDO), a not-for-profit association in
Denmark. The College of American Pathologists continues
to support SNOMED CT operations under contract to the
IHTSDO and provides SNOMED-related products and
services as a licensee of the terminology

Therapeutic classification codes: A classification system that
groups drugs into classes on the basis of their pharma-
cologic uses, such as antihypertensives and antibiotics.
The AHFS Pharmacologic-Therapeutic Classification was
developed and is maintained by the American Society of
Health-System Pharmacists (ASHP)

HEALTH DATA ANALYSIS TOOLKIT

35 | AHIMA

Clinical Document Architecture (CDA): An HL7 document
markup standard that specifies the structure and seman-
tics of “clinical documents” for the purpose of exchange
between healthcare providers and patients.

Clinical Quality Measure (CQM): Tools to help measure
and track the quality of healthcare services provided by
physicians, nurses, hospitals, and others in our health care
system. CQMs are required as part of Meaningful Use
requirements for the Medicare and Medicaid Electronic
Health Record (EHR) incentive programs

Continuity of Care Document (CCD): A joint effort of
HL7 International and ASTM to foster interoperability
of clinical data by allowing physicians to send electronic
medical information to other providers without loss of
meaning and enabling improvement of patient care. The
CCD establishes a rich set of templates representing
the typical sections of a summary record, and expresses
these templates as constraints on Clinical Document
Architecture (CDA)

eMeasure: A health quality measure formatted using the
HL7 Health Quality Measure Format (HQMF) standard
specification

Health Quality Measure Format (HQMF): A standard for
representing health quality measures as an electronic doc-
ument. HQMF formally defines a quality measure (data
elements, logic, definitions, etc.) to support consistent and
unambiguous interpretation. Quality measure developers
can encode their measures in this format so that they can
be consumed by provider organizations, which will then
be able to use the formal definitions to, for instance, query
their EHR system

Measure Authoring Tool (MAT): A publicly available, web-
based tool for measure developers to author electronic
Clinical Quality Measures (eCQMs) using the Quality
Data Model (QDM). Originally developed by the National
Quality Forum under contract from CMS, as of January
2013, CMS assumed ownership of the tool

Quality data element: Data criteria within a QDM eMeasure
composed of Quality Data Type, Quality Data Attribute,
and Value Set

Quality Data Model (QDM): An “information model” that
clearly defines concepts used in quality measures and
clinical care and is intended to enable automation of struc-
tured data capture in electronic health record (EHR) and
other electronic data sources. Developed and maintained
by the National Quality Forum

Quality Reporting Document Architecture (QRDA): A
document format that provides a standard structure with
which to report quality measure data to organizations
that will analyze and interpret the data. The Office of
the National Coordinator for Health IT (ONC) adopted
QRDA Category I (patient level) and QRDA Category III
(aggregate) data submission approaches for Meaningful
Use Stage 2 reporting.

QRDA Category I: individual patient-level report contain-
ing data for one or more quality measures

QRDA Category III: an aggregate level report containing
calculated summary data for one or more measures

Value Set: One or more coded values used to filter patient
populations for a specific data criterion (e.g., race)

Clinical Quality Measurement

HEALTH DATA ANALYSIS TOOLKIT

36 | AHIMA

Clinical data repository (CDR): A central database that
focuses on clinical information

Data dictionary: A descriptive list of the data elements to
be collected in an information system or database, the
purpose of which is to ensure consistency of terminology

Data element: An individual fact or measurement that is the
smallest unique subset of a database

Data mart: A well-organized, user-centered, searchable
database system that usually draws information from a
data warehouse to meet the specific needs of users

Data mining: The process of extracting information from
a database and then quantifying and filtering discrete,
structured data

Data repository: An open-structured database that is not
dedicated to the software of any particular vendor or data
supplier, in which data from diverse sources are stored
so that an integrated, multidisciplinary view of the data
can be achieved; also called a “central data repository” or,
when related specifically to healthcare data, a “clinical data
repository”

Data warehouse: A database that makes it possible to access
information from multiple databases and combine the
results into a single query and reporting interface. See also
clinical data repository

Database: An organized collection of data, text, references,
or pictures in a standardized format, typically stored in a
computer system for multiple applications

Decision support system (DSS): A computer-based system
that gathers data from a variety of sources and assists in
providing structure to the data by using various analytical
models and visual tools to facilitate and improve the
ultimate outcome in decision-making tasks associated
with nonroutine and nonrepetitive problems

Edit: A condition that must be satisfied before a computer
system can accept data

Interface engine: A computer program that isolates the task
of transferring data from one database to another

Metadata: Data about data that describe a specific item’s
content

Query: The process of making a logical inquiry or request
from a database

Structured Query Language (SQL): A fourth-generation
computer language that includes both data definition
language and data manipulation language components
and is used to create and manipulate relational databases

Database Terms

HEALTH DATA ANALYSIS TOOLKIT

37 | AHIMA

Actual charge: 1. A physician’s actual fee for service at the
time an insurance claim is submitted to an insurance
company, a government payer, or a health maintenance
organization; it may differ from the allowable charge 2.
The amount a provider actually bills a patient, which may
differ from the allowable charge

Allowable charge: Average or maximum amount a third-
party payer will reimburse providers for a service

Ambulatory payment classification (APC) system: The
prospective payment system used since 2000 for reim-
bursement of hospitals for outpatient services provided to
Medicare and Medicaid beneficiaries

Case mix: A description of a patient population on the basis
of any number of specific characteristics, including age,
sex, type of insurance, diagnosis, risk factors, treatment
received, and resources used

Case-mix index (CMI): The average relative weight of all
cases treated at a given facility or by a given physician,
which reflects the resource intensity or clinical severity
of a specific group in relation to the other groups in the
classification system; it is calculated by dividing the sum
of the weights of DRGs for patients discharged during a
given period by the total number of patients discharged

Claim: An itemized statement of healthcare services and
their costs provided by the hospital, physician office,
or other healthcare provider; it is submitted for reim-
bursement to the healthcare insurance plan by either the
insured party or the provider

CMS-1500: A Medicare claim form used to bill third-party
payers for provider services (e.g., physician office visits)

Cost report: A report that analyzes the direct and indirect
costs of providing care to Medicare patients; it is required
from providers annually for the Medicare program to
make a proper determination of amounts payable to
providers under its provisions

Explanation of benefits (EOB): A statement issued to the
insured and the healthcare provider by an insurer to explain
the services provided, amounts billed, and payments made
by a health plan. See also remittance advice (RA)

Financial data: The data collected for the purpose of manag-
ing the assets of a business (e.g., a healthcare organization,
a product line); in healthcare, data derived from the
charge-generation documentation associated with the
activities of care and then aggregated according to specific
customer grouping for financial analysis

Geographic adjustment factor (GAF): Adjustment to the
national standardized Medicare fee schedule relative value
components used to account for differences in the cost of
practicing medicine in different geographic areas of the
country

Geographic practice cost index (GPCI): An index devel-
oped by CMS to measure the differences in resource
costs among fee schedule areas compared to the national
average in the three components of the relative value unit
(RVU)—physician work, practice expenses, and malprac-
tice coverage; a separate GPCI exists for each element of
the RVU, and they are used to adjust the RVUs, which are
national averages, to reflect local costs

Health plan: An entity that provides or pays for the cost of
medical care, including a group health plan, a health insur-
ance issuer, a health maintenance organization, or any
welfare benefits plan such as Medicare, Medicaid, Civilian
Health and Medical Program of Uniformed Services, and
Indian Health Service

Hospital outpatient prospective payment system (HOPPS):
See outpatient prospective payment system (OPPS)

Incurred but not reported (IBNR): One of the basic con-
cepts of managing capitation contracts is estimating and
accruing any IBNR expenses and liabilities

Inpatient prospective payment system (IPPS): A type of
reimbursement system that is based on present payment
levels rather than actual charges billed after the service
has been provided; specifically, one of several Medicare
reimbursement systems based on predetermined payment
rates or periods and linked to the anticipated intensity of
services delivered, as well as the beneficiary’s condition

Medicare Administrative Contractor (MAC): An entity that
processes all Medicare claims for both Part A and Part B
by using the same common working file

Medicare code editor (MCE): A software program that
detects and reports errors in the coding claims data
submitted to the Medicare program

Medicare Provider Analysis and Review (MEDPAR) file: A
file that contains data from Medicare claims for services
provided to beneficiaries admitted to Medicare-certified
inpatient hospitals and skilled nursing facilities

National Provider Identifier (NPI): A 10-digit number that
uniquely identifies a healthcare provider

Finance and Reimbursement Terms

HEALTH DATA ANALYSIS TOOLKIT

38 | AHIMA

Outpatient code editor (OCE): A software program linked
to the National Correct Coding Initiative that applies a set
of logical rules to determine whether various combina-
tions of codes are correct and appropriately represent the
services provided

Outpatient prospective payment system (OPPS): The
Medicare prospective payment system is used for hospital-
based outpatient services and procedures and is predicated
on the assignment of APCs

Outpatient service-mix index (OSMI): The sum of the
weights of ambulatory payment classification groups for
patients treated during a given period divided by the total
volume of patients treated

Outpatient visit: A patient’s visit to one or more units located
in the ambulatory services area (clinic or physician’s office)
of an acute care hospital

Per member per month (PMPM): A common method to
express healthcare costs on the basis of a single member in
a month

Prospective payment system (PPS): A type of reimburse-
ment system that is based on preset payment levels
rather than actual charges billed after the service has
been provided— specifically, one of several Medicare
reimbursement systems based on predetermined payment
rates or periods and linked to the anticipated intensity of
services delivered, as well as the beneficiary’s condition

Recovery audit contractor (RAC): Entity operating on
behalf of CMS to identify improper payments made
on claims of healthcare services provided to Medicare
beneficiaries. Improper payments may be overpayments or
underpayments and may be based on coding or medical
necessity policies

Relative value unit (RVU): A measurement that represents
the value of practice expense, malpractice expense, and
physician work involved in providing a specific profes-
sional medical service in relation to the value of other
medical services

Remittance advice (RA): Report sent by third-party payer
that outlines claim rejections, denials, and payments to the
facility. See also explanation of benefits (EOB)

Resource-Based Relative Value Scale (RBRVS): A Medicare
reimbursement system to compensate physicians accord-
ing to a fee schedule predicated on weights assigned on the
basis of the resources required to provide the services

Revenue: The charges generated from providing healthcare
services—earned and measurable income

Revenue code: A four-digit number in the chargemaster that
totals all items and their charges for printing on the form
used for Medicare billing

Revenue cycle: The process of how patient financial and
health information moves into, through, and out of the
healthcare facility, culminating with the facility receiving
reimbursement for services provided—the regularly
repeating set of events that produces revenue

Service-mix index (SMI): Describes the outpatient popula-
tion in a single number; it is determined by multiplying
the number of cases times the relative weight for the
particular ambulatory payment classification (APC),
summing the relative weights, and dividing by the total
number of cases

Uniform Bill-04 (UB-04): The single standardized Medicare
form for uniform billing, implemented in 2007 for hospital
inpatients and outpatients; this form is used by the major
third-party payers and most hospitals

HEALTH DATA ANALYSIS TOOLKIT

39 | AHIMA

Accuracy: A characteristic of data that are free from sig-
nificant error and are up-to-date and representative of
relevant facts

Administrative data: Data created or collected through the
registration and billing process

Aggregate data: Data extracted from individual health
records and combined to form de-identified information
about groups of patients, which can be compared and
analyzed

Benchmarking: An analysis process based on comparison;
a comparison of performance against a standard point of
excellence, either within the organization (e.g., from year
to year) or among organizations on specified variables (for
example, length of stay or cost per DRG)

Categorical data: Types of data (nominal, ordinal) that
represent values or observations that can be sorted into a
category

Clinical data: Data captured during the process of diagnosis
and treatment

Coded data: Data that are translated into a standard nomen-
clature of classification so that they can be aggregated,
analyzed, and compared

Coding: The process of assigning numeric or alphanumeric
representations to clinical documentation

Comorbidity: Pre-existing condition that, because of its
presence with a specific diagnosis, causes an increase in
length of stay by at least one day in approximately 75% of
the cases (as in complication and comorbidity)

Complication: A medical condition that arises during an
inpatient hospitalization (e.g., a postoperative wound
infection)

Data: The dates, numbers, images, symbols, letters, and words
that represent basic facts and observations about people,
processes, measurements, and conditions

Data analysis: The process of looking at and summarizing
data with the intent to extract useful information and
develop conclusions

Data capture: The process of recording healthcare-related
data in a health record system or clinical database

Data comparability: The standardization of vocabulary such
that the meaning of a single term is the same each time the
term is used. Data comparability produces consistency of
information derived from those data

Data integrity: The extent to which healthcare data are
complete, accurate, consistent, and timely

Data set: A list of recommended data elements with uniform
definitions that are relevant for a particular use

Data stewardship: Responsibilities and accountabilities
associated with managing, collecting, viewing, storing,
sharing, disclosing, or otherwise making use of personal
health information

Demographic information: Information used to identify an
individual, such as name, address, sex, and age

Diagnostic data: Data obtained when diagnoses or the
reasons for visit are coded

Discrete data: Data that represent separate and distinct
values or observations (i.e., data that contain only finite
numbers and have only specified values)

Grouper: A computer software program that automatically
assigns prospective payment groups on the basis of clinical
codes

Health information management (HIM) professional: An
individual who has received professional training at the
associate or baccalaureate degree level in the management
of health data and flow of information throughout the
healthcare delivery system

Healthcare Cost and Utilization Project (HCUP): A group
of healthcare databases and related software tools devel-
oped through collaboration by the federal government,
state governments, and industry to create a national
information resource for patient-level healthcare data

Healthcare Effectiveness Data and Information Set
(HEDIS): A data set used by health plans to collect data
about the quality of care and service they provide

Hospital-acquired condition (HAC): Condition that is not
present at the time of inpatient admission, resulting in
the assignment of the case to a DRG that has a higher
payment when the condition is present as a secondary
diagnosis and could reasonably have been prevented by
following evidence-based guidelines

Inpatient: A patient who is provided with room, board, and
continuous general nursing services in an area of an acute
care facility where patients generally stay at least overnight

Leapfrog Group: Organization that promotes healthcare safety
by giving consumers the information they need to make
better-informed choices about the hospitals they choose

General

HEALTH DATA ANALYSIS TOOLKIT

40 | AHIMA

Map: The process of linking the content of one classification
or vocabulary system to another

Medical Group Management Association (MGMA): A
professional association for executives within physician
practices

Outpatient: A patient who receives ambulatory care services
in a hospital-based clinic or department

Present on admission (POA) indicator: A code that
describes whether the patient’s diagnosis was present at
the time of admission to the hospital

Primary data source: A record developed by healthcare
professionals in the process of providing patient care. See
also secondary data source

Principal diagnosis: The reason established after study to
be chiefly responsible for occasioning the admission of
the patient to the hospital for care. See also secondary
diagnosis

Procedural data: Data obtained when procedures are coded
by using a procedural classification and nomenclature
system

Secondary data source: Data derived from the primary
patient record, such as the data contained in an index or a
database. See also primary data source

Secondary diagnosis: A statement of conditions coexisting
during a hospital episode that affects the treatment
received or the length of stay. See also principal diagnosis

Table: An organized arrangement of data, usually in columns
and rows

HEALTH DATA ANALYSIS TOOLKIT

41 | AHIMA

ANNOTATED BIBLIOGRAPHY
This is a listing of commonly used Web sites, online courses, publications, and other helpful resources that provide
related learning. It is not meant to be all-inclusive.

CONTENTS

Calculations and Statistics ……………………………………………………………………………………………………………………….41

Classifications and Terminologies ………………………………………………………………………………………………………….42

Database Terms. ………………………………………………………………………………………………………………………………………..43

Finance and Reimbursement Terms ………………………………………………………………………………………………………44

General ………………………………………………………………………………………………………………………………………………………..45

Certification ……………………………………………………………………………………………………………………………………………….47

Calculations and Statistics

Calculating and Reporting Healthcare Statistics, 4th Edition
By Loretta Horton, 2012. An excellent reference for any healthcare professional interested in acquiring knowledge
and sharpening skills in this area. Explanations of why and how each kind of statistic is calculated and used are
reinforced with exercises to practice compiling inpatient service days, average length of stay and occupancy,
mortality rates, and other important data. Data presentation, inferential statistics, and basic research principles are
also covered.

Free Statistical Calculator Software
This page contains links to free software packages that you can download and install on your computer for stand-
alone computing. http://statpages.org/javasta2.html

Healthcare Data and SAS
By Marge Scerbo, 2001. A how-to guide for SAS analysts working with health plan data.

Pennsylvania Health Care Cost Containment Council (PHC4)
PHC4 was one of the first organizations to release risk-adjusted comparative healthcare statistics.
http://www.phc4.org/

SAS
SAS is a vendor of easy-to-use business analytics software and services.
http://www.sas.com/

SPSS
IBM SPSS Statistics is a comprehensive, easy-to-use set of predictive analytic tools for business users, analysts, and
statistical programmers. http://www.spss.com/

Statistical Applications for Health Information Management
By Carole Osborn, 2006. Copublished with Jones & Bartlett, this book covers the basic biostatistics, descriptive
statistics, and inferential statistics that are unique to HIM.

HEALTH DATA ANALYSIS TOOLKIT

42 | AHIMA

Classifications and Terminologies

AHIMA
AHIMA. “Data Standards, Data Quality, and Interoperability(Updated). Appendix A: Data Standards Resource.”
Journal of AHIMA 84, no. 11 (November-December 2013): Web extra. http://library.ahima.org/

American Medical Association
CPT codes http://www.ama-assn.org/

Centers for Medicare and Medicaid Services (CMS)
ICD-9-CM codes http://www.cms.hhs.gov/ICD9ProviderDiagnosticCodes/01_overview.asp
ICD-10-CM codes http://www.cms.gov/ICD10/

Healthcare Code Sets, Clinical Terminologies, and Classification Systems
By Kathy Giannangelo, 2010. Describes the latest developments in the growing field of health informatics and
multiple terminologies, vocabularies, code sets, and classification systems.

Healthcare Common Procedure Coding System (HCPCS) codes
http://www.cms.gov/HCPCSReleaseCodeSets/

International Health Terminology Standards Development Organisation (IHTSDO)
Systematized Nomenclature of Medicine-Clinical Terms (SNOMED CT)
http://www.ihtsdo.org/snomed-ct/

LOINC
LOINC laboratory codes
http://loinc.org/

National Drug Code Database
http://www.accessdata.fda.gov/scripts/cder/ndc/default.cfm

Therapeutic Classification Codes
http://www.ahfsdruginformation.com/pt-classification-system.aspx

HEALTH DATA ANALYSIS TOOLKIT

43 | AHIMA

Database Terms
ActiveReports
ActiveReports is a popular report writer. This site has good tips on writing effective reports.
http://www.componentone.com/SuperProducts/ActiveReports

“Beginner’s Guide to Databases”
By Charles Nadeau. This article takes a case study through the database design from the very beginning—not a
healthcare application, but a good discussion of the basics.
http://www.adobe.com/devnet/dreamweaver/articles/beginners_databases.html

“Database Design Basics”
This article gives an excellent overview of database design considerations. It is biased toward the features and
functionality of Microsoft Access, but most of the concepts transfer to any database tool.
http://office.microsoft.com/en-us/access/HA012242471033.aspx

“Database Design from the Ground Up”
By Mike Chapple. This article is more advanced and discusses topics beyond the scope of this course. There are a
number of good links to secondary articles.
http://databases.about.com/cs/specificproducts/a/designmenu.htm

“Data Dictionary”
This Wikipedia reference has a good description of a data dictionary and excellent references.
http://en.wikipedia.org/wiki/Data_dictionary

“Difference Between Reliability and Validity”