HEALTHCARE COST AND UTLIZATION PROJECT – HCUP
A FEDERAL-STATE-INDUSTRY PARTNERSHIP IN HEALTH DATA
Sponsored by the Agency for Healthcare Research and Quality

 

 

INTRODUCTION TO

THE HCUP NATIONWIDE READMISSIONS DATABASE (NRD)

2013

 

 

These pages provide introductory-level information about the NRD.   For full documentation and notification of changes, visit the HCUP User Support (HCUP-US) website at http://www.hcup-us.ahrq.gov.

 

November 2015

 

Agency for Healthcare Research and Quality
Healthcare Cost and Utilization Project (HCUP)

Phone: (866) 290-HCUP (4287)
E-mail: hcup@ahrq.gov
website: http://www.hcup-us.ahrq.gov

 

NRD Data and Documentation Distributed by:
HCUP Central Distributor
Phone: (866) 556-4287 (toll-free)
Fax: (866) 792-5313
E-mail: HCUPDistributor@ahrq.gov

Table of Contents

Skip Table of Contents

• SUMMARY OF DATA USE LIMITATIONS
• HCUP CONTACT INFORMATION
• UNDERSTANDING THE NRD
• ABSTRACT
• INTRODUCTION TO THE READMISSIONS DATABASE (NRD)
  • Overview of NRD Data
    • States Included in the NRD
    • Hospitals Included in the NRD
    • Discharges Included in the NRD
    • Summary of Hospitals and Discharges Included in the NRD
    • State-Specific Restrictions
  • File Structure of the NRD
    • NRD Data Elements
  • Getting Started
    • Copying and Decompressing the Comma-Delimitated Files
    • Downloading and Running the Load Programs
    • NRD Documentation
    • HCUP Online Tutorials
    • HCUP Methods Series Reports on Readmission Methodology
• SAMPLING DESIGN OF THE NRD
  • Target Universe
  • Sampling Frame
  • Discharge Weights
    • Post-Stratification for Weighting
    • Weighting
  • Final NRD Sample
  • Limitations of the NRD
• HOW TO USE THE NRD FOR READMISSION ANALYSES
  • NRD Data Elements Critical to Tracking a Patient and Determining the Time Between Admissions
    • Patient Linkage Number (NRD_VisitLink)
    • Time Between Admissions (NRD_DaysToEvent and LOS)
  • Defining Readmissions
    • Defining the Index Event
    • Specifying the Criteria for a Readmission
    • Selecting the Time Period for Revisits
    • Reporting Rates of Readmission
  • Calculating Nationally Weighted Estimates
  • Variance Calculations
  • Computer Software for Weighted and Variance Calculations
• NRD READMISSION RATES REPORTED ON THE HCUPNET WEBSITE
• APPENDIX A: NRD INTRODUCTORY INFORMATION
  • Table A.1. HCUP Partners Participating in the 2013 NRD
  • Figure A.1. HCUP States Participating in the 2013 NRD
  • Table A.2. Percentage of SID Discharges in the NRD by Type of Discharge
  • Table A.3. Summary of NRD States, Hospitals, and Inpatient Stays
  • Table A.4. NRD-Related Reports and Database Documentation Available on HCUP-US
  • Table A.5. Range of Discharge Weights by Patient Age and Sex
• APPENDIX B: HANLDING OF TRANSFERS AND SAME-DAY STAYS
  • Defining Transfers and Same-Day Events
  • Creating a Combined Transfer Record
• APPENDIX C: STATE-SPECIFIC RESTRICTIONS
  • Table C.1. State-Specific Restrictions
• APPENDIX D: NRD DATA ELEMENTS AND CODES
  • Table D.1. Data Elements in the NRD Core File
  • Table D.2. Data Elements in the NRD Severity File
  • Table D.3. Data Elements in the NRD Diagnosis and Procedure Groups File
  • Table D.4. Data Elements in the NRD Hospital File
• APPENDIX E: EVALUATION OF THE DIFFERENCE IN READMISSION RATES CAUSED BY HCUP PATIENT LINKAGE NUMBERS BEING SPECIFIC TO A STATE
  • Table E.1. Ten Conditions with the Largest Number of Index Events, Restricting Readmissions to Within State and Across All States
  • Table E.2. Ten Conditions with the Largest Difference in Readmission Rates, Restricting Readmissions to Within State and Across All States
  • Table E.3. Readmission Rates Restricting Readmissions to Within State and Across All States, Medicare Standard Analytic File, 2011

HCUP NATIONWIDE READMISSIONS DATABASE (NRD) SUMMARY OF DATA USE LIMITATIONS

***** REMINDER *****

All users of the NRD must take the on-line HCUP Data Use Agreement (DUA) training course, and read and sign a Data Use Agreement. † Authorized users of HCUP data agree to the following restrictions: ‡ Will not use the data for any purpose other than research or aggregate statistical reporting. Will not re-release any data to unauthorized users. Will not redistribute HCUP data by posting on any Website or other publically-accessible online repository Will not identify or attempt to identify any individual, including by the use of vulnerability analysis or penetration testing. Methods that could be used to identify individuals directly or indirectly shall not be disclosed or published. Will not publish information that could identify individual establishments (e.g., hospitals) and will not contact establishments. Will not use the data concerning individual establishments for commercial or competitive purposes involving those establishments, and will not use the data to determine rights, benefits, or privileges of individual establishments. Will not use data elements from the proprietary severity adjustment software packages (3M APR-DRGs, HSS APS-DRGs, and Thomson Reuters Disease Staging) for any commercial purpose or to disassemble, decompile, or otherwise reverse engineer the proprietary software. Will acknowledge in reports that data from the "Healthcare Cost and Utilization Project (HCUP)" were used, including names of the specific databases used for analysis. Will acknowledge that risk of individual identification of persons is increased when observations (i.e., individual discharge records) in any given cell of tabulated data is less than or equal to 10. Any violation of the limitations in the Data Use Agreement is punishable under Federal law by a fine of up to $10,000 and up to 5 years in prison. Violations may also be subject to penalties under State statutes.

† The on-line Data Use Agreement training session and the Data Use Agreement are available on the HCUP User Support (HCUP-US) website at http://www.hcup-us.ahrq.gov. ‡ Specific provisions are detailed in the Data Use Agreement for Nationwide Databases.

Return to Introduction

HCUP CONTACT INFORMATION

All HCUP data users, including data purchasers and collaborators, must complete the online HCUP Data Use Agreement (DUA) Training Tool, and read and sign the HCUP Data Use Agreement. Proof of training completion and signed Data Use Agreements must be submitted to the HCUP Central Distributor as described below.

The on-line DUA training course is available at: http://www.hcup-us.ahrq.gov/tech_assist/dua.jsp.

The HCUP Nationwide Data Use Agreement is available on the AHRQ-sponsored HCUP User Support (HCUP-US) website at: http://www.hcup-us.ahrq.gov

HCUP Central Distributor

Data purchasers will be required to provide their DUA training completion code and will execute their DUAs electronically as a part of the online ordering process. The DUAs and training certificates for collaborators and others with access to HCUP data should be submitted directly to the HCUP Central Distributor using the contact information below.

The HCUP Central Distributor can also help with questions concerning HCUP database purchases, your current order, training certificate codes, or invoices, if your questions are not covered in the Purchasing FAQs on the HCUP Central Distributor website.

Purchasing FAQs:
https://www.distributor.hcup-us.ahrq.gov/Purchasing-Frequently-Asked-Questions.aspx

Phone: 866-556-HCUP (4287) (toll free)
Email: HCUPDistributor@AHRQ.gov
Fax: 866-792-5313 (toll free in the United States)

Mailing address:
HCUP Central Distributor
Social & Scientific Systems, Inc.
8757 Georgia Ave, 12th Floor
Silver Spring, MD 20910

HCUP User Support:

Information about the content of the HCUP databases is available on the HCUP User Support (HCUP-US) website (http://www.hcup-us.ahrq.gov). If you have questions about using the HCUP databases, software tools, supplemental files, and other HCUP products, please review the HCUP Frequently Asked Questions or contact HCUP User Support:

HCUP FAQs:
http://www.hcup-us.ahrq.gov/tech_assist/faq.jsp

Phone: 866-290-HCUP (4287) (toll free)
Email: hcup@ahrq.gov

WHAT IS THE NATIONWIDE READMISSIONS DATABASE (NRD)?

The Nationwide Readmissions Database (NRD) is a database of all-payer hospital inpatient stays that can be used to examine national estimates of readmissions. Outcomes of interest include national readmission rates, reasons for returning to the hospital for care, and the hospital costs for discharges with and without readmissions. The NRD addresses a large gap in healthcare data—the lack of nationally representative information on hospital readmissions for all types of payers and the uninsured. The NRD is designed to be flexible to various types of analyses of readmissions. Criteria to determine the relationship between multiple hospital admissions for an individual patient in a calendar year is left to the analyst using the NRD. The NRD is drawn from HCUP State Inpatient Databases (SID) containing verified patient linkage numbers that can be used to track a person across hospitals within a State, while adhering to strict privacy guidelines. Unweighted, the NRD contains data from approximately 14 million discharges each year. Weighted, it estimates roughly 36 million discharges in the United States. There are 21 HCUP Partner States that contributed to the 2013 NRD: Arkansas, California, Florida, Georgia, Hawaii, Iowa, Louisiana, Massachusetts, Missouri, Nebraska, New Mexico, Nevada, New York, South Carolina, South Dakota, Tennessee, Utah, Virginia, Vermont, Washington, and Wisconsin. These States are geographically dispersed and account for 49.3 percent of the total U.S. resident population and 49.1 percent of all U.S. hospitalizations. The NRD is a publicly available database that can be purchased through the HCUP Central Distributor. The NRD is available only for data year 2013. Users must complete the HCUP Data Use Agreement Training Course prior to receiving the data.

 

Return to Introduction

 

UNDERSTANDING THE NRD

This document, Introduction to the NRD, 2013, summarizes the content of the NRD and describes the development of the 2013 NRD sample and weights. Important considerations for data analysis are highlighted and references to further resources are provided. In-depth documentation for the NRD is available on the HCUP User Support (HCUP-US) website (http://www.hcup-us.ahrq.gov/db/nation/nrd/nrddbdocumentation.jsp). Please refer to detailed documentation before using the data.

HEALTHCARE COST AND UTILIZATION PROJECT — HCUP
A FEDERAL-STATE-INDUSTRY PARTNERSHIP IN HEALTH DATA
Sponsored by the Agency for Healthcare Research and Quality

HCUP NATIONWIDE READMISSIONS DATABASE (NRD)

ABSTRACT

The Nationwide Readmissions Database (NRD) is part of the Healthcare Cost and Utilization Project (HCUP) that is sponsored by the Agency for Healthcare Research and Quality (AHRQ). The NRD addresses a large gap in healthcare data - the lack of nationally representative information on hospital readmissions for all types of payers and the uninsured. The NRD was created to enable analyses of national readmission rates and to support public health professionals, administrators, policymakers, and clinicians in their decision making.

The NRD is drawn from HCUP State Inpatient Databases (SID) that contain reliable, verified patient linkage numbers that can be used to track a person across hospitals within a State, while adhering to strict privacy guidelines. The 2013 NRD is constructed from 21 SID. These States are geographically dispersed and account for 49.3 percent of the total U.S. resident population and 49.1 percent of all U.S. hospitalizations. See Appendix A, Table 1 for a list of data organizations participating in the NRD.

The NRD includes community hospitals, excluding rehabilitation or long-term acute care hospitals. All discharges from the SID are included except the following:

• Discharges from patients with an age of 0 form about 60 percent of the SID because patient linkage numbers are inconsistently reported for this age
• Discharges with missing patient linkage numbers
• Discharges with questionable patient linkage numbers, defined as 20 or more discharges in a year, hospitalized after discharged dead, and overlapping inpatient stays
• Discharges from hospitals with more than 50 percent of their total discharges excluded for any of the above causes, because patients treated at these hospitals may not be reliably tracked over time.

After exclusions, the 2013 NRD contains about 85 percent of SID discharges from the participating states. Unweighted, the NRD contains approximately 14 million discharges each year. Weighted, it estimates roughly 36 million discharges in the United States.

The NRD is designed to be flexible to various types of analyses of national readmissions for all types of payers and the uninsured. The criteria to determine the relationship between multiple hospital admissions for an individual patient is left to the analyst using the NRD. Outcomes of interest include national readmission rates, reasons for returning to the hospital for care, and the hospital costs for discharges with and without readmissions. The NRD is not designed to support regional, State-, or hospital-specific readmission analyses.

Access to the NRD is open to users who sign a Nationwide Databases Data Use Agreement (DUA). Uses are limited to research and aggregate statistical reporting. For more information on the NRD, visit the AHRQ-sponsored HCUP User Support (HCUP-US) website at http://www.hcup-us.ahrq.gov.

 

Return to Introduction

 

INTRODUCTION TO THE NATIONWIDE READMISSIONS DATABASE (NRD)

 

Overview of NRD Data

The Healthcare Cost and Utilization Project (HCUP) Nationwide Readmissions Database (NRD) was created to enable analyses of national readmission rates and to support public health professionals, administrators, policymakers, and clinicians in their decision making. Reducing hospital readmission rates is a key strategy for improving the quality of healthcare while containing cost. The NRD is designed to be flexible to various types of readmission analyses. The database includes discharges for patients with and without repeat hospital visits in a year and those who have died in the hospital. Repeat visits may or may not be related. The criteria to determine the relationship between hospital admissions is left to the analyst using the NRD.

States Included in the NRD

In their SID, some HCUP Partner organizations provided synthetic patient linkage numbers that can be used to track patients within and across hospitals in a particular State. Unique combinations of the patient linkage number and the patient's date of birth and sex constituted a verified patient linkage number.

States selected for the NRD have verified patient linkage numbers on at least 90 percent of adult discharges. From these States, the NRD included discharges from patients aged 1 year and older. Discharges aged 0 were only retained from States that had verified patient linkage numbers on more than 90 percent of discharges aged 0. Appendix A, Table 1 identifies the statewide data organizations that contribute to the NRD. Appendix A, Figure 1 displays the geographic distribution of the 21 HCUP Partner organizations participating in the 2013 NRD. Based on U.S. Census Bureau data, the 2013 NRD accounts for 49.3 percent of the U.S. population and 49.1 percent of U.S. hospitalizations reported in the American Hospital Association (AHA) Annual Survey Database.

Hospitals Included in the NRD

The SID contains inpatient discharges for all hospitals provided by the HCUP Partners (e.g., community, specialty, and Federal hospitals). The American Hospital Association (AHA) defines community hospitals as "all non-Federal, short-term, general, and other specialty hospitals, excluding hospital units of institutions." Specialty hospitals included in the AHA definition of community hospitals are obstetrics-gynecology, ear-nose-throat, short-term rehabilitation, orthopedic, pediatric institutions, and long-term acute care (LTAC) facilities. Also included are public hospitals and academic medical centers.

The NRD is limited to data from community hospitals that are not rehabilitation or LTAC facilities. Noncommunity hospitals were excluded because of inconsistent capture of data across HCUP States. We excluded rehabilitation or LTAC hospitals because they treat a unique patient population that has longer stays and higher costs. Information on the percentage of SID discharges excluded by type of exclusion is provided in Appendix A, Table 2. Details on the number of hospitals in the NRD are provided in Appendix A, Table 3.

Discharges Included in the NRD

All SID discharges from selected States and hospitals were included, with a few exceptions. Records with missing or unverified patient linkage numbers were excluded from the NRD. All discharges aged 0 from SID records with verified patient linkage numbers on less than 90 percent of discharges were excluded.

Another concern was verified patient linkage numbers that did not appear to track an individual within the year for the following reasons:

1. Extraordinary utilization in the year, defined as 20 or more admissions in a calendar year
2. Multiple discharges for the same identifier that showed the patient discharged dead from one admission and then admitted at a later date in the year
3. Overlapping hospitalizations for the same patient linkage number at the same or different hospitals.

Discharges for patient linkage numbers considered questionable because of these three criteria were excluded from the NRD.

Discharge-specific exclusions such as the removal of unverified, missing, or questionable patient linkage numbers impacted individual hospitals if they had more than 50 percent of their annual discharges excluded. These hospitals were not good candidates for a readmission analysis because too many of their discharges could not be tracked over time, and were therefore excluded from the NRD. Information on the percentage of SID discharges excluded by type of exclusion is provided in Appendix A, Table 2.

There were no exclusions for certain types of patients or clinical conditions. Therefore, we included conditions such as childbirth that generally do not result in a readmission. We retained discharges that resulted in an in-hospital death because these were possible readmission records. We also included discharges for residents and nonresidents of the State in which they were treated. Although most patients seek treatment at hospitals in their State of residence, there are occasions when patients are treated at hospitals in another State. Hospitals that specialize in a certain types of care may attract patients from across the United States. In addition, hospitals near State borders frequently treat patients that reside in neighboring States. The data element RESIDENT identifies a discharge as a resident of the State in which he or she received hospital care. Details on the number of discharges in the NRD are provided in Appendix A, Table 3.

Hospital administrative databases like the NRD and SID are "discharge-level" files, meaning that each record represents one discharge abstract from an inpatient stay. If a patient visits the hospital multiple times in a given year, the SID includes separate records for each inpatient stay. In addition, if a patient is transferred between hospitals within the State, the SID contain two discharge records, one record from the first hospitalization and a second record from the latter hospitalization.¹ Readmission analyses do not usually allow the hospitalization at the receiving hospital to be counted as a readmission. To eliminate this possibility, we collapsed the pairs of records representing a transfer into a single "combined" record in the NRD and removed the original separate discharge records from the NRD. We defined transfer records as having all of the following characteristics:

• Discharge date of the first inpatient stay equaled the admission date of a subsequent inpatient stay²
• The first record had a discharge disposition of transfer to an acute care hospital
• The second record was from a different hospital and had an admission source indicating a transfer.

We defined a discharge as a same-day stay if the discharge date for one inpatient stay was the same as the admission date of a second stay for the same patient (same as transfers), but there was no indication of a transfer by the discharge disposition or admission source. Same-day stays may or may not have involved different hospitals. Same-day stays may indicate that a patient was discharged too soon and then needed to be returned to the hospital on the same day. However, it was also possible that these were transfer records with an incorrect or missing discharge disposition and admission source. A more detailed description of the methodology used to identify and combine transfer and same-day stays is provided in Appendix B.

 

Return to Introduction

Summary of Hospitals and Discharges Included in the NRD

The NRD includes community hospitals and excludes the following types of hospitals:

• Noncommunity hospitals because of inconsistent capture of data across HCUP States
• Community hospitals that are either rehabilitation or LTAC hospitals because these hospitals treat a unique patient population that has longer stays and higher costs.

All discharges from included hospitals were retained in the NRD with the following exceptions:

• Discharges from patients with an age of 0 from SID with verified patient linkage numbers on less than 90 percent of discharges for this age group
• Discharges with missing or unverified patient linkage numbers
• Discharges with questionable patient linkage numbers, defined as 20 or more discharges in a year, hospitalized after discharged dead, and overlapping inpatient stays
• Discharges from hospitals with more than 50 percent of their total discharges excluded for any of the above causes.

Information on the percentage of SID discharges excluded by type of exclusion is provided in Appendix A, Table 2. There were no exclusions for certain types of patients or clinical conditions. Discharges that resulted in an in-hospital death were included because these were possible readmission records. Discharges for residents and nonresidents of the State in which they were treated were also included. In addition, pairs of discharge records representing transfers (i.e., one discharge record from the sending hospital and one discharge record from the receiving hospital) were collapsed into a single record so that the hospitalization at the receiving hospital could not be counted as a readmission. Details on the number of States, hospitals, and discharges in the NRD are provided in Appendix A, Table 3.

State-Specific Restrictions

Some sources that contributed data to the NRD imposed restrictions on the release of certain data elements or discharges that could be included in the database. In addition, because of confidentiality laws, some data sources were prohibited from providing HCUP with discharge records that indicated specific medical conditions, such as HIV/AIDS or behavioral health. Detailed information on these State-specific restrictions is available in Appendix C.

File Structure of the NRD

The NRD includes three discharge-level files and one hospital-level file:

• Discharge-level files
  • Core File: Contains data elements critical to readmission analyses
  • Severity file: Contains additional data elements to aid in identifying the severity of the condition for a specific discharge (e.g., comorbidity flags, 3M All-Patient Refined Diagnosis-Related Group [APR-DRG] value, risk of mortality, and severity)
  • Diagnosis and Procedure Groups File: Contains additional information on the diagnoses (e.g., chronic condition indicators) and procedures (e.g. procedure class)
• Hospital-level File: Contains information on hospital characteristics.

NRD Data Elements

The coding of data elements in the NRD is consistent with other HCUP databases. The following three objectives guided the definition of data elements in all HCUP databases:

• Ensure usability without extensive editing by analysts.
• Retain the largest amount of information available from the original sources, while still maintaining consistency among sources.
• Structure the information for efficient storage, manipulation, and analysis.

More information on the coding of HCUP data elements is available on the HCUP User Support (HCUP-US) website (http://www.hcup-us.ahrq.gov/db/coding.jsp).

The NRD contains more than 100 clinical and non-clinical variables provided in a hospital discharge abstract, such as:

• Discharge month, quarter, and year³
• ICD-9-CM diagnoses with external cause of injury codes
• ICD-9-CM procedures
• Diagnosis and procedure classifications variables such as the Clinical Classifications Software (CCS) category, Chronic Condition Indicator (CCI), and procedure class
• Patient demographics (e.g., sex, age, income quartile based on the median household income of the patient's ZIP Code, and urban/rural location of the patient's residence)
• Expected source of payment (e.g., Medicare, Medicaid, private insurance, self-pay, and other insurance types)
• Total hospital charges and hospital cost⁴
• Discharge weights for generating national estimates.
• Variables essential to readmission analyses:
  • Verified patient linkage number (variable NRD_VisitLink) that can be used to identify discharges belonging to the same patient
  • Timing between admissions for a patient (variable NRD_DaysToEvent)
  • Length of inpatient stay in days (variable LOS)
  • Identification of transfers, same-day stays, and combined transfer records (variables SAMEDAYEVENT and REHABtransfer)
  • Identification of the patient as a resident of the State in which he or she received hospital care (variable RESIDENT).

 

Return to Introduction

 

Appendix D identifies the data elements in each NRD file:

• Table 1 for the NRD Core File
• Table 2 for the NRD Severity File
• Table 3 for the NRD Diagnosis and Procedure Groups File
• Table 4 for the Hospital File.

The tables in Appendix D provide summary documentation for the data. Please refer to the NRD documentation located on the HCUP-US website (http://www.hcup-us.ahrq.gov) for comprehensive information about data elements and the files.

Getting Started

The NRD is an extremely large database that requires sophisticated statistical software for analysis. The following computer properties are needed in order to load and analyze the NRD data:

• A DVD drive
• A hard drive with 40 gigabytes (GB) of available space
• A third-party zip utility such as ZIP Reader, 7-Zip, or WinZip®
• SAS®, SPSS®, Stata® or similar analysis software.

The total size of the comma-separated version (CSV) of the NRD is 10 GB. The NRD files loaded into SAS are about 12 GB. In SAS, the largest use of space typically occurs during a sort, which requires work space about three times the size of the file. Thus, the NRD files would require at least 36 GB of available workspace to perform a sort. Most SAS data steps will require twice the storage of the file, so that both the input and output files can coexist. The NRD files loaded into SPSS are under 23 GB. Because Stata loads the entire file into memory, it may not be possible to load every data element in the NRD Core file into Stata. Stata users will need to maximize memory and use the "_skip" option to select a subset of variables. More details are provided in the Stata load programs.

With a file of this size and without careful planning, space could easily become a problem in a multi-step program. It is not unusual to have several versions of a file marking different steps while preparing it for analysis, and there may be more versions for the actual analyses. Therefore, the amount of space required could escalate rapidly.

Copying and Decompressing the Comma-Delimitated Files

The NRD is distributed as CSV files compressed with SecureZIP® from PKWARE. To copy and decompress the NRD files from the DVD, the following steps are outlined:

1. Create a directory for the NRD on the hard drive.
2. Unzip each compressed file from the DVD, saving it into a new directory using a third-party zip utility such as ZIP Reader, WinZip, or 7-Zip. You will be prompted to enter the encryption password (sent separately by e-mail) to decrypt each file. Please note that attempts to unzip files using Windows will produce an error message similar to the following: "Compressed (Zipped) Folders Error. The password you typed is not correct. Try typing it again." The solution is to use a third-party zip utility such as ZIP Reader, 7-Zip, or WinZip rather than the built-in Windows Explorer function to open the archive. ZIP Reader may be downloaded for free at https://www.pkware.com/zip-reader. 7-Zip can be downloaded for free at http://www.7-zip.org/. Evaluation versions of WinZip may be downloaded from the WinZip website at www.winzip.com.

Downloading and Running the Load Programs

Programs to load the data into SAS, SPSS, or Stata, are available on the HCUP User Support website (HCUP-US). These steps are used to download and run the load programs:

1. Go to the NRD Database Documentation page on HCUP-US at http://www.hcup-us.ahrq.gov/db/nation/nrd/nrddbdocumentation.jsp.
2. Go to the "Load Programs" section on this page.
3. Click on "SAS Load Programs", "SPSS Load Programs", or "Stata Load Programs" to go to the corresponding Load Programs page.
4. Select and download the load programs. They are specific to the data year. For example, the load program for the 2013 NRD Core file is linked to "Core File" under "2013 NRD". Save the load programs into the same directory as the NRD CSV files on the computer.
5. Edit and run the load programs as appropriate for the environment, to load and save the analysis files. For example, add directory paths for the input and output files as needed.

NRD Documentation

Comprehensive documentation for the NRD files is available on the HCUP-US website (http://www.hcup-us.ahrq.gov/db/nation/nrd/nrddbdocumentation.jsp). Users of the NRD can access complete file documentation, including variable notes, file layouts, and summary statistics. Similarly, data users can download SAS, SPSS, and Stata load programs. These important resources help the user understand the structure and content of the NRD and aid in using the database.

To locate the NRD documentation on HCUP-US, the user is instructed to:

• Choose "Databases" from the home page (http://www.hcup-us.ahrq.gov).
• Select the section labeled "Nationwide Readmissions Database (NRD)."

Appendix A, Table 4 details the comprehensive NRD documentation available on HCUP-US.

 

Return to Introduction

 

HCUP Online Tutorials

For additional assistance, AHRQ has created the HCUP Online Tutorial Series, a series of free, interactive courses that provide training on technical methods for conducting research with HCUP data. Topics include an HCUP Overview Course and these tutorials that are helpful to NRD data users:

• The Nationwide Readmissions Database Tutorial provides a description of the NRD database design and a detailed example on how to use the NRD to estimate national readmission rates.
• The Load and Check HCUP Data tutorial provides instructions on how to unzip (decompress) HCUP data, save it on a computer, and load the data into a standard statistical software package. This tutorial also describes how to verify that the data have loaded correctly.

Other tutorials about the design or use of other HCUP databases also are available. The Online Tutorial Series is located on the HCUP-US website at http://hcup-us.ahrq.gov/tech_assist/tutorials.jsp.

HCUP Methods Series Reports on Readmission Methodology

HCUP has created two Methods Series reports that provide additional information readmissions:

• Overview of Key Readmission Measures and Methods (Report #2012-04). This report describes 12 key measures of hospital readmissions from different agencies or companies. The 12 measures were selected for inclusion because they are: (1) endorsed by the National Quality Forum (NQF), (2) currently used by AHRQ in various initiatives, or (3) otherwise well-known and used by researchers.
• Methodological Issues when Studying Readmissions and Revisits using Hospital Administrative Data (Report #2011-01). This report discusses the decisions that need to be addressed as an analyst designs a study of patients with sequential acute care hospital visits that is based on hospital administrative data. Topics include considerations in preparing the administrative database for the analysis, defining repeat hospital visits, and reporting results.

 

Return to Introduction

 

SAMPLING DESIGN OF THE NRD

The NRD was built to facilitate analyses of both all-cause and condition-specific readmissions. National estimates can be created by applying weighting and stratification methods.

Target Universe

The target universe was limited to inpatient discharges treated at community hospitals in the United States that were not rehabilitation or LTAC facilities. Information on the target universe was available from the American Hospital Association (AHA) Annual Survey of Hospitals. The AHA Survey includes information on the number of inpatient discharges and hospital characteristics such as ownership, number of beds, and location

Sampling Frame

The sampling frame for the NRD was limited to discharges for patients treated at community hospitals in the NRD States that were not rehabilitation or LTAC facilities. All of the discharges in the sampling frame were included making the NRD a sample of convenience. Sampling discharges or hospitals was not recommended because the sample needed to balance the database's ability to estimate readmissions for common conditions such as chronic illnesses with the ability to estimate readmissions for rare diseases such as sickle cell anemia. Developing the database using a 100 percent sample allows researchers to study both all-cause and condition-specific readmissions.

Discharge Weights

This section explains the need for post-stratification for weighting the sampling frame to the target universe and the weighting strategy. We use the term post-stratification because the stratification was performed after sampling. Discharge weights for national estimates were developed using the target universe as the standard. Although we calculated discharge-level weights for the NRD, we did not calculate hospital-level weights. The NRD is not designed to support hospital-specific analyses. The NRD was a 100 percent sample of discharges, not hospitals; hospital weights were not applicable.

Post-Stratification for Weighting

Post-stratification for the purpose of weighting allowed us to compensate for any over- or under-represented types of hospitals and discharges in the sampling frame (the NRD) with respect to the distribution in the target universe (AHA data). The NRD was post-stratified by hospital and patient characteristics. We knew from the National (Nationwide) Inpatient Sample (NIS) design that the following hospital characteristics explained significant differences in inpatient outcomes: census region, urban/rural location, hospital teaching status, size of the hospital defined by the number of beds, and hospital control.⁵

We had excluded discharges with unverified and missing patient linkage numbers because these patients cannot be tracked across time. In an examination of the distribution of patient age and sex between discharges with verified and unverified/missing patient linkage numbers, we determined that the majority of the unverified discharges (52.7 percent) were females aged 18-44 years old. In addition, we compared 30-day all-cause readmission rates in 2011 by age-sex categories across States and saw variation between SID with a high percentage of verified patient linkage numbers and States with a lower percentage of verified patient linkage numbers. These analyses demonstrated that there were differences between discharges with and without patient linkage numbers by sex and age. Therefore, the NRD was also post-stratified by sex and five age groups (0, 1-17, 18-44, 45-64, and 65 and older).

 

Return to Introduction

 

Weighting

We based the discharge counts by stratum for the target universe totals on all SID discharges from all HCUP Partners, unless there were missing hospitals. If there were hospitals missing from the stratum according to the AHA, then the target universe total included SID discharges for all available hospitals plus the AHA discharge counts for the missing hospitals. This approach was consistent with the NIS and took advantage of the fact that the SID included over 95 percent of discharges from community hospitals that are not rehabilitation or LTAC hospitals in the United States. Discharge counts for the sampling frame were based on the NRD discharges (after the exclusion of States, hospitals, and discharges). To determine discharge-level weights, we summarized the number of discharges for the target universe and sampling frame by stratum defined by hospital characteristics (census region, urban/rural location, hospital teaching status, size of the hospital defined by the number of beds, and hospital control) and patient characteristics (sex and five age groups [0, 1-17, 18-44, 45-64, and 65 and older]). Within each stratum, s, each NRD inpatient admission received a weight:

DISCWT_i,j = Ns(universe)_i,j ÷ Ns(sample)_i,j

where Ns(universe)_i,j represents the number of inpatient admissions at community hospitals that were not a rehabilitation or LTAC hospital in the universe within stratum s for sex i and age group j; Ns(sample)_i,j is the number of inpatient admissions in the sampling frame for sex i and age group j. Age group j included ages 0, 1-17, 18-44, 45-64, 65 and older. Therefore, each discharge's weight (DISCWT_i,j) is equal to the number of inpatient admissions it represents in stratum s for sex i and age group j during that year.

To improve reliability of the age distribution of the SID discharges, we collapsed the strata prior to the weight calculations such that we included at least two SID hospitals and at least 100 discharges from the SID in each stratum. In addition, we collapsed the strata to include at least two hospitals in the sample frame. We first collapsed the stratum across control/ownership, combining either the two private designations or all three types of control (public, private not-for-profit, and private for-profit). If the stratum combined across control still lacked a sufficient number of hospitals or discharges, then the location category was collapsed. Small and large metropolitan areas are combined or micropolitan and rural areas combined. Lastly, if the stratum still lacked a sufficient number of hospitals or discharges, then the bed-size category was collapsed with large and medium hospitals combined. There was no collapsing of strata across region or teaching status. In addition, we adjusted weights if a hospital was missing data for one or more quarters in the year. The range of weights by age and sex are provided in Appendix A, Table 5.

 

Return to Introduction

 

Final NRD Sample

The NRD is an annual file constructed using one calendar year of discharge data. Included discharges were treated at community hospitals (excluding rehabilitation or LTAC hospitals) for which the majority of their discharges had patient linkage numbers that were verified and not questionable. Discharge weights were calculated using post-stratification on hospital characteristics (census region, urban-rural location, teaching status, bed size, and hospital control) and patient characteristics (sex and five age groups [0, 1-17, 18-44, 45-64, and 65 and older]). The target universe of inpatient discharges in the United States was estimated for each stratum using SID total discharges augmented by AHA discharge counts when hospitals were not reported in the SID. Details on the number of States, hospitals, and discharges in the NRD are provided in Appendix A, Table 3. The range of discharge weights by age and sex are provided in Appendix A, Table 5. The NRD is designed to be flexible to various types of analyses of national readmissions. The NRD is not designed to support regional, State- or hospital-specific readmission analyses.

Limitations of the NRD

The NRD contains over 14 million inpatient discharge records and over 100 clinical and non-clinical data elements. A multitude of research studies can be conducted with the data, but there are some limitations.

• Limitations on Studying Pediatric Readmissions: The NRD includes patients who were younger than 1 year from only 9 of the 21 SID in 2013. The weights for pediatric discharges were often higher than adult discharges. The range of discharge weights by age and sex are provided in Appendix A, Table 5.
• Limitations from Using One Year of Discharge Data: The NRD are annual files containing inpatient records for patients discharged in a calendar year. The files included patients admitted in the prior year and discharged in the current year, while excluding patients admitted to a hospital in the current year but discharged in the next year. Therefore, 30- or 60-day readmissions for patients admitted in the latter part of the year may not be captured if the subsequent admission crossed into the next year. In addition, one year of discharge data is probably an insufficient length of time for examining readmissions that are more than 90 days apart.
• Limitations from Using State-Specific Identifiers Patients who were hospitalized in one State and readmitted or transferred to a hospital in another State cannot be tracked in the NRD, because each of the SID uses different coding for their patient linkage numbers. The NRD includes nonresident patients because we want to retain discharges at border hospitals that provided care for patients in their community, even though that community happened to cross State borders. To understand the impact on readmission rates caused by using State-specific patient linkage numbers, we performed a sensitivity analysis using the 2011 Medicare Standard Analytic File (SAF) that allowed admissions for patients covered by Medicare fee-for-service (FFS) plans to be tracked across all States.
  
  The analysis demonstrated that condition-specific readmission rates were higher if a patient could be tracked across all States, but the percentage increase was less than 5 percent for most of the CCS categories. This analysis was limited to the Medicare FFS population because of the data source. The Medicare (FFS and managed care) population accounted for about 40 percent of all inpatient discharges in 2011,⁶ and previous research indicates that the Medicare population has higher readmission rates than populations covered by for other payers.⁷ Conditions often associated with younger adults, such as pregnancy, were included in the Medicare estimates because about 20 percent of Medicare discharges are younger than 65 years.⁸ Medicare patients younger than 65 include people who are disabled or who have been diagnosed with end-stage renal disease or amyotrophic lateral sclerosis (ALS).
  
  Given the volume and severity of illness for Medicare patients, the estimates for the increase in the condition-specific readmission rates using the Medicare data provide a reasonable upper boundary on the impact. The details of this analysis are provided in Appendix E.

 

Return to Introduction

 

HOW TO USE THE NRD FOR READMISSION ANALYSES

This section provides a brief synopsis of special considerations for using the NRD. For more details, refer to the comprehensive documentation on the HCUP-US website (http://www.hcup-us.ahrq.gov/db/nation/nrd/nrddbdocumentation.jsp).

All persons using the NRD (whether or not they are the original recipient of the data) must complete the online Data Use Agreement Training Tool available on the HCUP-US website (http://www.hcup-us.ahrq.gov/tech_assist/dua.jsp) and then read and sign a Data Use Agreement. A copy of the signed Data Use Agreements must be sent to the HCUP Central Distributor. See page 2 of this document for the mailing address.

NRD Data Elements Critical to Tracking a Patient and Determining the Time Between Admissions

For any readmission analysis of inpatient stays, three HCUP data elements are critical to tracking a patient and determining the time between admissions: NRD_VisitLink, NRD_DaysToEvent, and LOS (length of stay).

Patient Linkage Number (NRD_VisitLink)

NRD_VisitLink is the linkage variable for all inpatient stays associated with a unique patient. All discharges in the NRD include a value for NRD_VisitLink. The value was assigned based on a unique combination of the synthetic patient linkage number provided by the HCUP Partner, date of birth, and sex. No verified patient linkage number was assigned if any one of the three pieces of information was missing. Because of discharge-level exclusions, NRD_VisitLink is always coded on records in the NRD.

Although the term patient linkage number is used to describe the information in the NRD data element NRD_VisitLink, the values are not recognizable as specific patient information. NRD_VisitLink does not include the values of the encrypted person's social security number, date of birth, or sex.

Time Between Admissions (NRD_DaysToEvent and LOS)

NRD_DaysToEvent is the number of days from a randomly chosen "start date" to the admission date for each patient's discharge. The actual admission and discharge dates could not be included on the NRD because they were considered highly sensitive information according to Health Insurance Portability and Accountability Act (HIPAA) guidelines. The coding scheme for NRD_DaysToEvent was designed to adhere to these strict privacy guidelines and protect patient confidentiality.

Each verified patient linkage number (NRD_VisitLink) was assigned a unique start date that was used to calculate NRD_DaysToEvent for all visits associated with that NRD_VisitLink value. The variable NRD_DaysToEvent was the difference between the visit's admission date and the start date associated with the NRD_VisitLink. NRD_DaysToEvent was reported as missing if the admission date was unavailable.

For readmission analyses, determining the number of days between the end of one admission and the start of the next admission is critical. We did not include any single data element specific to this timing difference in the NRD because the calculation is dependent on which two discharges are of interest for the readmission study. For example, a study of readmissions for diabetes might only consider the number of days between two diabetes discharges, whereas a study of post-surgery infections might consider any discharge in 30 days.

Because NRD_DaysToEvent was based on the admission date, the calculation of days needs to be the difference of NRD_DaysToEvent between two selected discharges for a unique verified patient linkage number (NRD_VisitLink), adjusted for the length of stay. Consider the following example:

• A patient with congestive heart failure has a 3-day hospital admission on 1/10/2013 and another admission on 1/25/2013.
• The NRD_DaysToEvent value is 1109 for the 1/10/2013 admission, and the NRD_DaysToEvent value is 1124 for the 1/25/2013 visit.
• The number of days between the start of each admission is 15 days (1124 - 1109 = 15) because NRD_DaysToEvent is based on the admission date.
• The number of days between the admissions (from discharge date of the first admission to the start of the second admission) is 12 days (1124 - 1109 - 3 = 12) because the patient had a 3-day LOS.

If NRD_DaysToEvent or LOS was missing, then it was not possible to determine the number of days to a subsequent admission. We considered removing the discharges with missing NRD_DaysToEvent and LOS from the NRD, but these data elements were very rarely missing. LOS was only critical if it was missing on the first admission in a series. If the admission was the second in the series, then LOS was not pertinent.

The lowest value of NRD_DaysToEvent is the earliest inpatient stay in the year for a patient. It is important to remember that if patient A has a value of 605 for NRD_DaysToEvent and patient B has a value of 300 for NRD_DaysToEvent, patient B's hospital stay did not necessarily take place prior to patient A's stay. In fact, Patient B's NRD_DaysToEvent value has no relation to Patient A's NRD_DaysToEvent value. Because of the use of a random start date in the calculation of NRD_DaysToEvent, the value of NRD_DaysToEvent cannot be compared across patients.

Additional information about the HCUP revisit variables is available on the HCUP User Support website (http://www.hcup-us.ahrq.gov/toolssoftware/revisit/revisit.jsp).

Defining Readmissions

The NRD was designed to support many different types of readmission analyses. Analysts can use the information contained in the NRD to define the index event and readmission specific to their topic of interest. Common terminology is first defined:

• Index event - the starting point for analyzing repeat hospital visits
• Readmission - a subsequent inpatient admission within a specified time period; readmission may be for a specific cause or any cause.

The NRD was designed to support many different types of readmission analyses. Analysts can use the information contained in the NRD to define the index event and readmission specific to their topic of interest. We discuss the following analytic considerations for defining index events and readmissions:

• Defining the index event
• Specifying the criteria for a readmission
• Selecting the appropriate time period to qualify the readmission
• Reporting readmission rates.

 

Return to Introduction

 

Defining the Index Event

The index event is typically defined by a combination of clinical criteria. Inclusion and exclusion criteria should be used to define an index event indicator that identifies NRD discharges as an index event specific to the analysis of interest. The NRD did not include a variable for index events because they are specific to each analysis. The NRD included the information necessary to define different types of index events.

Criteria can include, but are not limited to, age of the patient and specific diagnoses and/or procedures. The NRD contained various data elements that can be used for inclusion criteria:

• ICD-9-CM diagnosis and procedure codes
• CCS categories
• Diagnosis-related group (DRG) and major diagnostic category (MDC)
• 3M All-Patient Refined DRG.

Possible exclusion criteria include the following:

• Index events in which the patient died in the hospital, because there is no risk of readmission.
• Patients with complicating comorbidities such as cancer or an immunocompromised state, because these conditions would greatly increase the risk of readmission.
• Transfers and same-day stays that were combined, because they represent a more complex type of care (NRD data element SAMEDAYEVENT not equal to 0).
• Discharge month for when the index event occurred, because there are limitations when using a calendar-year file defined by discharge date.

The annual NRD files include inpatient stays that were discharged in that data year. For example, the 2013 NRD includes admissions that began in 2012 and were discharged in 2013. In contrast, we did not include admissions that began later in 2013 and extended into 2014 in the 2013 NRD. About one percent of discharges in the NRD started in the previous year; therefore, we expected that we were missing about one percent of admissions that started in 2013 and were subsequently discharged in 2014. This will be true for each year of the NRD.

Deciding which months should be excluded when qualifying an index event depends on the time that will be allowed for a readmission. For example, if studying 30-day readmissions, the index event might be limited to those occurring in the discharge months of January through November. That allows the month of December for 30 days of follow-up. Although it would be advantageous to be able to select a more specific date for a cut-off, patient confidentially concerns limited the available information on the admission and discharge dates to discharge month (data element DMONTH) and discharge year (data element YEAR).

Specifying the Criteria for a Readmission

Readmission analyses tend to consider one of the following: any subsequent admission regardless of cause, any subsequent admission that does not involve trauma, or any subsequent admission only if the event is "related" to the index event. In addition, a study may consider all readmissions within a time period or just the first readmission. The selection of criteria can dramatically change results. More information on how the results can change is available in an HCUP Method Series report on Methodological Issues when Studying Readmissions and Revisits Using Hospital Administrative Data.⁹

The NRD includes a number of different diagnosis and procedure-related variables that can be used to examine why a patient returned for hospital care. The NRD does not identify any discharge as a readmission; instead, we include the information necessary to select the appropriate readmission discharges in the NRD. Inclusion and exclusion criteria should be used to define a readmission indicator that identifies NRD discharges as readmissions specific to the analysis of interest.

Selecting the Time Period for Revisits

When determining an appropriate time period for the readmission, considerations include selecting a time that encompasses the same risk of exposure to all patients, seasonality of the disease, and possible external factors. Shorter time frames (7 or 14 days) are often used to make events attributable to hospital acute care; longer time frames may reflect differences in ambulatory care and/or coordination of care.

Reporting Rates of Readmission

Although the definition of readmission rate—number of readmissions divided by number of cases followed— seems simple, our research into readmission rates showed no standard definition. In some cases, the unit of observation was a patient; in others, the unit of observation was index events, and individual patients were counted more than once. Some studies focused on the first readmission following an index event, whereas others counted all readmissions. The definitions of the readmission rate were specific to the purpose of the analyses.

Severity or risk adjustment may also be beneficial when comparing readmission rates across geographical regions, hospital types, or different patient populations. A simple risk adjustment would include the age and sex of the patient. A more complex adjustment might also include comorbidites, severity classified by the 3M All-Patient Refined DRG severity score, patient income quartile, or any other factor that could considerably increase or decrease the risk of subsequent hospital care. The NRD included variables to support these types of analyses in the Core, Severity, and Diagnosis and Procedure Groups files.

Calculating Nationally Weighted Estimates

An analyst using the NRD must use the discharge-level weight DISCWT to produce national estimates. Weighted statistics estimate discharges treated at community hospitals (excluding rehabilitation and LTAC facilities) in the United States.

Variance Calculations

It may be important for researchers to calculate a measure of precision for national estimates based on the NRD. Variance estimates must take into account both the sampling design and the form of the statistic. It is important to understand that the NRD is a sample of convenience from the SID and not a sample of hospitals or discharges. Standard error calculations should take into account the stratified sample (data element NRD_STRATUM) and hospitals defining the clusters (data element HOSP_NRD). One resource for understanding the issues surrounding variance calculations for the NRD is the HCUP Methods Series report #2010-05 Inferences with HCUP State Databases Final Report on the HCUP-US website (http://www.hcup-us.ahrq.gov/).

To accurately calculate variances from the NRD, appropriate statistical software and techniques must be used. A multitude of statistics can be estimated from the NRD data. Several computer programs that calculate statistics and their variances from sample survey data are listed in the next section. Some of these programs use general methods of variance calculations (e.g., the jackknife and balanced half-sample replications) that take into account the sampling design. However, it may be desirable to calculate variances using formulas specifically developed for certain statistics.

These variance calculations are based on finite-sample theory, which is an appropriate method for obtaining cross-sectional, nationwide estimates of outcomes. According to finite-sample theory, the intent of the estimation process is to obtain estimates that are precise representations of the nationwide population at a specific point in time. In the context of the NRD, any estimates that attempt to accurately describe readmissions during a specific year should be governed by finite-sample theory. Examples would be estimates of expenditure and utilization patterns.

Computer Software for Weighted and Variance Calculations

The NRD discharge weights are needed to calculate national estimates of readmission counts and rates. In most cases, computer programs are readily available to perform these calculations. Several statistical programming packages allow weighted analyses¹⁰ For example, nearly all SAS procedures incorporate weights. In addition, several statistical analysis programs have been developed to specifically calculate statistics and their standard errors from survey data. Version 8 or later of SAS contains procedures (PROC SURVEYMEANS and PROC SURVEYREG) for calculating statistics based on specific sampling designs. Stata and SUDAAN are two other common statistical software packages that perform calculations for numerous statistics arising from the stratified, single-stage cluster sampling design.

NRD READMISSION RATES REPORTED ON THE HCUPNET WEBSITE

Readmission rates generated from the NRD are available on HCUPnet, a free online query system based on data from the HCUP (https://datatools.ahrq.gov/hcupnet). We define in this section how the readmission rates are calculated for HCUPnet to provide an example of how the NRD data elements might be used to define an index event, 30-day readmission, and readmission rates. Other types of readmission analyses are possible with the NRD; this is just one of many possible applications.

For the readmission rates on HCUPnet, we defined an index event as follows:

• Patient was discharged between January and November (1 <= DMONTH <= 11).
• Patient was discharged alive (DIED = 0).
• Length of stay was non-missing (LOS >= 0).
• Discharge was for patient aged 1 year or older (AGE > 0).
• Patient may be a nonresident of the State (any value of RESIDENT).
• A patient is allowed to have multiple index events, regardless of how far apart.

For example, if a patient was discharged alive with a nonmissing length of stay on January 10, January 20, January 26, and March 30, all four discharges would qualify as index admissions.

For the readmission rates on HCUPnet, we defined readmissions as follows:

• The first discharge for a patient was within 30 days of an index event.
• Discharge occurred between January and December (1 <= DMONTH <= 12).
• Discharge may be to the same or a different hospital (HOSP_NRD) and may result in a death.

On HCUPnet, we defined the readmission rates as the percentage of index admissions that had at least one readmission within 30 days.

• Numerator = total number of index events that had at least one subsequent hospital admission within 30 days.
• Denominator = total number of index events between January and November.
• Rate = numerator / denominator * 100.

Rates were not risk adjusted.

Consider an example of the 30-day, all-cause readmission rate for any diagnosis for a patient discharged alive on January 10, January 20, January 26, and March 30. Each admission is considered an index.

• January 10 is the first index admission.
• January 20 qualifies as a 30-day readmission for the January 10th index. It is also an index admission.
• January 26 qualifies as a 30-day readmission for the January 20th index. It is also an index admission.
• March 30 is an index admission, but it does not qualify as a readmission because it does not fall within 30 days of another index.

The 30-day readmission rate is 50 percent, because there are two 30-day readmissions for the four index admissions.

HCUPnet can be used to query 30-day readmission rates by the following:

• Any diagnosis (no specific selection criteria on diagnosis or procedure)
• Principal diagnosis CCS (using the HCUP data element DXCCS1)
• First-listed procedure CCS (using the HCUP data element PRCCS1)
• MDC (using the HCUP data element MDC)
• DRG (using the HCUP data element DRGnoPOA that does not consider the present on admission indicator for assignment).

HCUPnet reports readmission counts, rates and costs stratified by the following characteristics of the index stay:

• Age group is based on the HCUP data element AGE.
• Sex is based on the HCUP data element FEMALE.
• Payer is assigned using the primary and secondary expected payer (HCUP data elements PAY1 and PAY2). If the primary or secondary expected payer indicates Medicare, then the payer category is assigned to Medicare. This categorization includes patients who are dually-eligible for Medicare and Medicaid under Medicare. If not Medicare and the primary or secondary expected payer indicates Medicaid, then the payer category is Medicaid. If not Medicare or Medicaid and the primary or secondary expected payer indicates private insurance, then the payer category is Private. If not Medicare, Medicaid, or Private and the primary expected payer indicates self-pay or no charge, then the payer category is Uninsured. Stays for other types of payers are not reported on HCUPnet because this is a mixed payer group with small numbers. The expected secondary payer data element PAY2 is not available on the NRD.
• Income level is based on the HCUP data element ZIPINC_QRTL for the national quartile of the median household income for the patient's ZIP Code.
• Location is based on the HCUP data element PL_UR_CAT4 for the location of the patient's residence according to the Urban Influence Code (UIC) designation. Urban includes large and small metropolitan areas with all other areas categorized as rural. The data element PL_UR_CAT4 is not available on the NRD. The data element on the NRD for patient location is PL_NCHS, a six-category urban-rural classification scheme for U.S. counties developed by the National Center for Health Statistics (NCHS).

 

Return to Introduction

 

APPENDIX A: NRD INTRODUCTORY INFORMATION

Table A.1. HCUP Partners Participating in the 2013 NRD

State	HCUP Data Source
Arkansas	Arkansas Department of Health
California	California Office of Statewide Health Planning and Development
Florida	Florida Agency for Health Care Administration
Georgia	Georgia Hospital Association
Hawaii	Hawaii Health Information Corporation
Iowa	Iowa Hospital Association
Louisiana	Louisiana Department of Health and Hospitals
Massachusetts	Massachusetts Center for Health Information and Analysis
Missouri	Missouri Hospital Industry Data Institute
Nebraska	Nebraska Hospital Association
New Mexico	New Mexico Department of Health
Nevada	Nevada Department of Health and Human Services
New York	New York State Department of Health
South Carolina	South Carolina Revenue and Fiscal Affairs Office
South Dakota	South Dakota Association of Healthcare Organizations
Tennessee	Tennessee Hospital Association
Utah	Utah Department of Health
Virginia	Virginia Health Information
Vermont	Vermont Association of Hospitals and Health Systems
Washington	Washington State Department of Health
Wisconsin	Wisconsin Department of Health Services

 

Return to Introduction

 

Figure A.1. HCUP States Participating in the 2013 NRD

text version

Table A.2. Percentage of SID Discharges in the NRD by Type of Discharge

Type of Discharge	Percentage of SID Discharges, 2013
Included in the NRD	84.72
Excluded from the NRD	15.28
Hospital-level exclusions
Noncommunity hospitals	2.54
Rehabilitation or LTAC hospitals	0.27
Discharge-level exclusions
Discharges from patients with an age of 0 (from 12 of 21 SID)	7.56
Discharges with missing or unverified patient linkage numbers	4.13
Questionable patient linkage numbers: same patient linkage number on 20 or more discharges	0.18
Questionable patient linkage numbers: patient is hospitalized after discharged dead	0.02
Questionable patient linkage numbers: overlapping inpatient stays	0.46
Discharges from hospitals with more than 50 percent of their total discharges excluded for any of the above causes	0.13

Table A.3. Summary of NRD States, Hospitals, and Inpatient Stays

Year	States	Number of States for Discharges Aged 1 and Older	Number of States for Discharges Aged 0	Number of Hospitals	Number of Discharges in the NRD, Unweighted	Number of Discharges in the NRD, Weighted
2013	AR, CA, FL, GA, HI, IA, LA, MA, MO, NE, NM, NV, NY, SC, SD,TN, UT, VA, VT, WA, WI	21	9	2,006	14,325,172	35,580,348

 

Return to Introduction

 

Table A.4. NRD-Related Reports and Database Documentation Available on HCUP-US

Description of the NRD Files Introduction to the NRD, 2013 (this document) and prior years HCUP Quality Control Procedures - describes procedures used to assess data quality File Specifications - details data file names, number of records, record length, and record layout Restrictions on the Use of the NRD Data Use Agreement for the HCUP Nationwide Databases Requirements for publishing with HCUP data Load Programs SAS Load Programs SPSS Load Programs Stata Load Programs Description of Data Elements in the NRD Description of Data Elements - details uniform coding and State-specific idiosyncrasies Availability of Data Elements by Year - lists which data elements are available each year Additional Resources for NRD Data Elements Summary Statistics - lists means and frequencies on nearly all data elements HCUP Coding Practices - describes how HCUP data elements are coded HCUP Hospital Identifiers - explains data elements that characterize individual hospitals

HCUP Tools: Labels and Formats Overview of Clinical Classifications Software (CCS) Format programs to create value labels DRG formats HCUP formats HCUP diagnoses and procedure groups, including CCS categories ICD-9-CM formats ICD-10-CM format NRD-Related Reports HCUP reports related to readmissions analyses

Table A.5. Range of Discharge Weights by Patient Age and Sex

Patient Age and Sex	Discharge Weight, 2013
	Minimum	Average Across Stata	Maximum
Males
Age 0	1.07	7.53	31.90
Age 0-17	1.04	3.89	18.48
Age 18-44	1.13	2.88	11.33
Age 45-64	1.10	2.83	10.58
Age 65 and older	1.07	2.83	11.81
Females
Age 0	1.07	7.66	35.30
Age 0-17	1.04	3.55	11.28
Age 18-44	1.06	2.67	7.53
Age 45-64	1.09	2.78	9.83
Age 65 and older	1.06	2.83	11.68

 

Return to Introduction

 

APPENDIX B: HANLDING OF TRANSFERS AND SAME-DAY STAYS

Hospital administrative databases like the NRD and SID are "discharge-level" files, meaning that each record represents one discharge abstract from an inpatient stay. If a patient visits the hospital multiple times in a given year, the SID includes separate records for each inpatient stay. In addition, if a patient is transferred between hospitals within the State, the SID contain two discharge records, one record from the first hospitalization and a second record from the latter hospitalization.¹¹

Defining Transfers and Same-Day Events

Readmission analyses do not usually allow the hospitalization at the receiving hospital to be counted as a readmission. To eliminate this possibility, we collapsed the pairs of records representing a transfer into a single "combined" record in the NRD and removed the original separate discharge records from the NRD. We defined transfer records as having all of the following characteristics:

• Discharge date of the first inpatient stay equaled the admission date of a subsequent inpatient stay. ¹²
• The first record had a discharge disposition of transfer to an acute care hospital.
• The second record was from a different hospital and had an admission source indicating a transfer.

We defined a discharge as a same-day stay if the discharge date for one inpatient stay was the same as the admission date of a second stay for the same patient (same as transfers), but there was no indication of a transfer by the discharge disposition or admission source. Same-day stays may or may not have involved different hospitals. Same-day stays may indicate that a patient was discharged too soon and then needed to be returned to the hospital on the same day. However, it was also possible that these were transfer records with an incorrect or missing discharge disposition and admission source.

We collapsed records that were part of transfers and same-day stays into a single combined record. These combined records account for about three percent of records in the NRD and are identified by the data element SAMEDAYEVENT. The value of SAMEDAYEVENT is defined as follows:

• Transfer involving two discharges from different hospitals (value): Discharge date of one admission equaled the admission date on another record for the same patient. There was a discharge disposition of transfer to an acute care hospital on the first record and an admission source of transfer (in) on the subsequent record; two different hospitals were involved.
• Same-day stay involving two discharges at different hospitals (value 2): The discharge date on the first inpatient stay equaled the admission date on a second inpatient stay; two different hospitals were involved, but the coding of either the discharge disposition or admission source did not indicate a transfer.
• Same-day stays involving two discharges at the same hospital (value 3): The discharge date on the first inpatient stay equaled the admission date on a second inpatient stay; the hospital was the same on both records.
• Same-day stay involving three or more discharges, same or different hospitals (value 4): Multiple records indicating a combination of transfers and other same-day events seem odd, but they do rarely occur in administrative data.
• Not a transfer or other same-day stay (value 0). About 97 percent of the records in the NRD did not involve transfers or same-day stays.

Please note that if a patient had a discharge disposition of transfer and an admission source that also indicated a transfer, but the discharge date of the first stay did not equal the admission date of the second stay (e.g., the patient was admitted the next day because the transfer occurred at night), the two records are not considered a transfer in the NRD.

Creating a Combined Transfer Record

Combining information across transfer and same-day stay records required specific rules for how to handle different types of information on the pairs of records. We first ordered the pairs of records by earliest occurrence in the year. The different scenarios described below detail how we combined different types of information:

• Use first. For information that pertains to day of admission (e.g., admission month, admission source, NRD_DaysToEvent), the combined transfer record used the information from the first record in the pair.
• Use last record. For information that pertains to the end of the time in the hospital (e.g., discharge disposition, expected payer, the Diagnosis Related group [DRG]), the combined transfer record used the information from the latter record in the pair.
• Summarize. For information that needed to reflect both stays, the combined transfer record summarized the information. For example, length of stay and total charge were summed across the pair of records.
• Diagnoses and related variables. Each record included arrays of International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) diagnoses and related variables. On the assumption that the diagnoses on the latter record would reflect the final determination of diagnoses after complete treatment, we retained the diagnoses reported on the latter record at the beginning of the diagnosis array (including the principal diagnosis). We added the diagnoses from the first record to the end of the diagnosis array. The NRD is limited to a maximum of 25 diagnosis codes, so no more than 25 total diagnoses were retained on a combined transfer record. The same scheme was used for all diagnosis-related variables also contained in arrays.
  
  • Exception to this rule. If the second part of the transfer had a principal diagnosis of rehabilitation, medical evaluation, or other types of aftercare, using the principal diagnosis from the second part of the stay on the combined record masked the real reason for the admission. If the latter record had any one of the following Clinical Classifications Software (CCS) categories as the principal diagnosis, the assignment of diagnoses and related variables was modified to give precedence to the information on the first inpatient stay with the information from the latter stay added to the end of the associated arrays:
    • CCS 254: Rehabilitation care; fitting of prostheses; and adjustment of devices
    • CCS 255: Administrative/social admission
    • CCS 256: Medical examination/evaluation
    • CCS 257: Other aftercare
    • CCS 258: Other screening for suspected conditions (not mental disorders or infectious disease)
    • CCS 259: Residual codes; unclassified.
    These types of transfer records are flagged by the data element REHABtransfer (value 1).
• Procedures and related variables. Each record included arrays of ICD-9-CM procedure codes and related variables. Although procedure codes were not necessarily reported in the order in which they were performed, we decided to retain the procedures reported on the first record at the beginning of the procedure array. The procedures from the latter record were added to the end of the procedure array. The NRD is limited to a maximum of 15 procedure codes, so no more than 15 total procedures were retained on a combined transfer record. The same scheme was used for all procedure-related variables that were also contained in arrays. The day of procedure on the latter record was adjusted for the length of stay on the first record, so the days reflected the time from the beginning of the first stay.
• Hospital identifiers and characteristics. The hospital identifiers and characteristics from the latter stay were retained on the combined transfer record. This assumed that the latter hospital was primarily responsible for the care.
• Patient characteristics. The patient characteristics (e.g., age, sex, income quartile) from the first stay were retained on the combined transfer record.

 

Return to Introduction

 

APPENDIX C: STATE-SPECIFIC RESTRICTIONS

The table below enumerates the types of restrictions applied to the Nationwide Readmissions Database. Restrictions include the following types:

• Confidentiality of records
  • Restricted release of age in years
• Missing discharges for specific populations of patients.

For each restriction type the data sources are listed alphabetically by State. Only data sources that have restrictions are included. Data sources that do not have restrictions are not included.

Table C.1. State-Specific Restrictions

Confidentiality of Records
Restricted release of patient's age in years to ensure patient confidentiality: Age (AGE) values greater than 90 are set to age 90 in the NRD.
Missing Discharges for Specific Populations of Patients
The following data sources may be missing discharge records for specific populations of patients: At least one Partner prohibits the release of discharge records for patients with HIV diagnoses. At least one Partner prohibits the release of behavioral health including chemical dependency care or psychiatric care discharges. At least one Partner prohibits the release of inpatient records that were discharged from Alternative Level of Care units of the hospital (e.g., skilled nursing or swing bed units). At least one Partner prohibits the release of abortion discharges.

 

Return to Introduction

 

APPENDIX D: NRD DATA ELEMENTS AND CODES

Table D.1. Data Elements in the NRD Core File

Category	Data Element Name	Description
Admission/Discharge	AWEEKEND	Admission on weekend: (0) admission on Monday-Friday, (1) admission on Saturday-Sunday
	DIED	Indicates in-hospital death: (0) did not die during hospitalization, (1) died during hospitalization
	DISPUNIFORM	Disposition of patient, uniform coding: (1) routine, (2) transfer to short term hospital, (5) other transfers, including skilled nursing facility, intermediate care, and another type of facility, (6) home healthcare, (7) against medical advice, (20) died in hospital, (99) discharged alive, destination unknown
	DQTR	Coded: (1) Jan-Mar, (2) Apr-Jun, (3) Jul-Sep, (4) Oct-Dec
	ELECTIVE	Indicates elective admission: (1) elective, (0) nonelective admission
	HCUP_ED	Indicator that discharge record includes evidence of emergency department (ED) services: (0) record does not meet any HCUP ED criteria, (1) ED revenue code was on SID record, (2) ED charge reported on SID record, (3) ED CPT procedure code on SID record, (4) other indication of ED services
	YEAR	Discharge year
Clinical Information	DRG	DRG in use on discharge date
	DRG_NoPOA	DRG assignment made without the use of the present on admission flags for the diagnoses
	DRGVER	Grouper version in use on discharge date
	DX1-DX25	Diagnoses, principal and secondary (ICD-9-CM)
	DXCCS1-DXCCS25	CCS category for all diagnoses
	E_CCS1-E_CCS4	CCS category for all External cause of injury codes
	ECODE1-ECODE4	External causes of injury codes (ICD-9-CM)
	MDC	MDC in use on discharge date
	MDC_NoPOA	MDC assignment made without the use of the present on admission flags for the diagnoses
	NCHRONIC	Number of chronic conditions
	NDX	Number of diagnoses coded
	NECODE	Number of external causes of injury codes coded
	NPR	Number of procedures coded
	ORPROC	Major operating room procedure indicator: (1) major operating room procedure reported on discharge record, (0) no major operating room procedure reported on discharge record
	PR1-PR15	Procedures, principal and secondary (ICD-9-CM)
	PRCCS1-PRCCS15	CCS category for all procedures
NRD Identifiers	HOSP_NRD	NRD hospital identifier specific to the NRD and is not linkable to any other HCUP or external databases
	KEY_NRD	NRD record identifier specific to the NRD and not linkable to any other HCUP or external databases
Patient Demographics	AGE	Age in years coded 0-90 years; any age greater than 90 was set to 90. Missing age was imputed using other records with the same patient linkage number. Less than 1000 discharges (0.007 percent) had the age imputed.
	FEMALE	Indicates sex: (0) male, (1) female. Age was imputed for other records with the same patient linkage number of missing. Missing sex was imputed using other records with the same patient linkage number. Less than 10 discharges had the sex imputed.
	PAY1	Expected primary payer, uniform: (1) Medicare, (2) Medicaid, (3) private insurance, (4) self-pay, (5) no charge, (6) other
	PL_NCHS	Patient location: National Center for Health Statistics (NCHS) urban-rural classification scheme for U.S. counties: (1) "Central" counties of metro areas of >=1 million population, (2) "Fringe" counties of metro areas of >=1 million population, (3) Counties in metro areas of 250,000-999,999 population, (4) Counties in metro areas of 50,000-249,999 population, (5) Micropolitan counties, (6) Not metropolitan or micropolitan counties
	ZIPINC_QRTL	Median household income quartiles for patient's ZIP Code. For 2013, the median income quartiles are defined as: (1) $1 - $37,999; (2) $38,000 - $47,999; (3) $48,000 - $63,999; and (4) $64,000 or more.
Readmission Specific	DMONTH	Discharge month coded from (1) January to (12) December
	NRD_DAYSTOEVENT	Count from "start date" to admission date coded differently for each value of NRD_VisitLink
	NRD_VISITLINK	Patient linkage number specific to the NRD and not linkable to any other HCUP or external databases
	REHABTRANSFER	A combined record involving transfer to rehabilitation, evaluation, or other aftercare: (1) yes, (0) no
	RESIDENT	Identifies patient as a resident of the State in which he or she received hospital care: (1) resident, (0) nonresident
	SAMEDAYEVENT	Identifies records that were combined from transfer or same-day stay pairs of records: (0) not a combined transfer or other same-day stay record, (1) combined transfer involving two discharges from different hospitals, (2) combined same-day stay involving two discharges at different hospitals, (3) combined same-day stay involving two discharges at the same hospital, (4) combined same-day stay involving three or more discharges at same or different hospitals
Resource Use	LOS	Length of stay, edited
	TOTCHG	Total charges, edited
Weighting	DISCWT	NRD discharge weight to be used for calculating national estimates
	NRD_STRATUM	NRD stratum for post-stratification based on geographic region, urban/rural location, teaching status, size of hospital based on number of beds, and control/ownership. For the confidentiality of hospitals and States, the NRD_STARTUM variable was randomly assigned into sequential numbers starting at 1.

 

Return to Introduction

 

Table D.2. Data Elements in the NRD Severity File

Category	Data Element Name	Description
3M APR-DRG	APRDRG	3M All Patient Refined DRG
	APRDRG_Risk_Mortality	3M All Patient Refined DRG: Risk of Mortality Subclass: (0) No class specified, (1) Minor likelihood of dying, (2) Moderate likelihood of dying, (3) Major likelihood of dying, (4) Extreme likelihood of dying
	APRDRG_Severity	3M All Patient Refined DRG: Severity of Illness Subclass: (0) No class specified, (1) Minor loss of function (includes cases with no comorbidity or complications), (2) Moderate loss of function, (3) Major loss of function, (4) Extreme loss of function
AHRQ Comorbidity Measures	CM_AIDS	AHRQ comorbidity measure - Acquired immune deficiency syndrome: (1) comorbidity present, (0) comorbidity not present
	CM_ALCOHOL	AHRQ comorbidity measure - Alcohol abuse: (1) comorbidity present, (0) comorbidity not present
	CM_ANEMDEF	AHRQ comorbidity measure - Deficiency anemias: (1) comorbidity present, (0) comorbidity not present
	CM_ARTH	AHRQ comorbidity measure - Rheumatoid arthritis/collagen vascular diseases: (1) comorbidity present, (0) comorbidity not present
	CM_BLDLOSS	AHRQ comorbidity measure - Chronic blood loss anemia: (1) comorbidity present, (0) comorbidity not present
	CM_CHF	AHRQ comorbidity measure - Congestive heart failure: (1) comorbidity present, (0) comorbidity not present
	CM_CHRNLUNG	AHRQ comorbidity measure - Chronic pulmonary disease: (1) comorbidity present, (0) comorbidity not present
	CM_COAG	AHRQ comorbidity measure - Coagulopath: (1) comorbidity present, (0) comorbidity not present
	CM_DEPRESS	AHRQ comorbidity measure - Depression: (1) comorbidity present, (0) comorbidity not present
	CM_DM	AHRQ comorbidity measure - Diabetes, uncomplicated: (1) comorbidity present, (0) comorbidity not present
	CM_DMCX	AHRQ comorbidity measure - Diabetes with chronic complications: (1) comorbidity present, (0) comorbidity not present
	CM_DRUG	AHRQ comorbidity measure - Drug abuse: (1) comorbidity present, (0) comorbidity not present
	CM_HTN_C	AHRQ comorbidity measure - Hypertension, uncomplicated and complicated: (1) comorbidity present, (0) comorbidity not present
	CM_HYPOTHY	AHRQ comorbidity measure - Hypothyroidism: (1) comorbidity present, (0) comorbidity not present
	CM_LIVER	AHRQ comorbidity measure - Liver disease: (1) comorbidity present, (0) comorbidity not present
	CM_LYMPH	AHRQ comorbidity measure - Lymphoma: (1) comorbidity present, (0) comorbidity not present
	CM_LYTES	AHRQ comorbidity measure - Fluid and electrolyte disorders: (1) comorbidity present, (0) comorbidity not present
	CM_METS	AHRQ comorbidity measure - Metastatic cancer: (1) comorbidity present, (0) comorbidity not present
	CM_NEURO	AHRQ comorbidity measure - Other neurological disorders: (1) comorbidity present, (0) comorbidity not present
	CM_OBESE	AHRQ comorbidity measure - Obesity: (1) comorbidity present, (0) comorbidity not present
	CM_PARA	AHRQ comorbidity measure - Paralysis: (1) comorbidity present, (0) comorbidity not present
	CM_PERIVASC	AHRQ comorbidity measure - Peripheral vascular disorders: (1) comorbidity present, (0) comorbidity not present
	CM_PSYCH	AHRQ comorbidity measure - Psychoses: (1) comorbidity present, (0) comorbidity not present
	CM_PULMCIRC	AHRQ comorbidity measure - Pulmonary circulation disorders: (1) comorbidity present, (0) comorbidity not present
	CM_RENLFAIL	AHRQ comorbidity measure - Renal failure: (1) comorbidity present, (0) comorbidity not present
	CM_TUMOR	AHRQ comorbidity measure - Solid tumor without metastasis: (1) comorbidity present, (0) comorbidity not present
	CM_ULCER	AHRQ comorbidity measure - Peptic ulcer disease excluding bleeding: (1) comorbidity present, (0) comorbidity not present
	CM_VALVE	AHRQ comorbidity measure - Valvular disease: (1) comorbidity present, (0) comorbidity not present
	CM_WGHTLOSS	AHRQ comorbidity measure - Weight loss: (1) comorbidity present, (0) comorbidity not present
NRD Identifiers	HOSP_NRD	NRD hospital identifier specific to the NRD and is not linkable to any other HCUP or external databases
	KEY_NRD	NRD record identifier specific to the NRD and not linkable to any other HCUP or external databases

 

Return to Introduction

 

Table D.3. Data Elements in the NRD Diagnosis and Procedure Groups File

Category	Data Element Name	Description
Clinical Information	CHRON1-CHRON25	Chronic condition indicator for all diagnoses: (0) non-chronic condition, (1) chronic condition
	CHRONB1-CHRONB25	Chronic Condition Indicators - body system for all diagnoses
	DXMCCS1	Multi-level CCS category for principal diagnosis
	E_MCCS1	Multi-level CCS category for first-listed E code
	PCLASS1-PCLASS15	Procedure class for all ICD-9-CM procedures: (1) Minor Diagnostic, (2) Minor Therapeutic, (3) Major Diagnostic, (4) Major Therapeutic
	PRMCCS1	Multi-level CCS category for first-listed procedure
NRD Identifiers	HOSP_NRD	NRD hospital identifier specific to the NRD and is not linkable to any other HCUP or external databases
	KEY_NRD	NRD record identifier specific to the NRD and not linkable to any other HCUP or external databases

 

Return to Introduction

 

Table D.4. Data Elements in the NRD Hospital File

Category	Data Element Name	Description
Admission/ Discharge	YEAR	Discharge year
Hospital Information	H_CONTROL	Control/ownership of hospital: (1) government, nonfederal [public], (2) private, not-for-profit [voluntary], (3) private, investor-owned [proprietary]
	HOSP_BEDSIZE	Size of hospital based on the number of beds: (1) small, (2) medium, (3) large. The categories are defined using region of the U.S., the urban-rural designation of the hospital, in addition to the teaching status.
	HOSP_UR_TEACH	Teaching status of hospital: (0) metropolitan non-teaching, (1) metropolitan teaching, (2) non-metropolitan
	HOSP_URCAT4	Hospital urban-rural location: (1) large metropolitan areas with at least 1 million residents, (2) small metropolitan areas with less than 1 million residents, (3) micropolitan areas, (4) not metropolitan or micropolitan, (8) metropolitan, collapsed category of large and small metropolitan, (9) non-metropolitan, collapsed category of micropolitan and rural
	NRD_STRATUM	NRD stratum for post-stratification based on geographic region, urban/rural location, teaching status, bed size, and control. Values range from 1 to 89. Region is not identified.
NRD Identifiers	HOSP_NRD	NRD hospital identifier specific to the NRD and is not linkable to any other HCUP or external databases
Weighting	N_DISC_U	Number of discharges in the tart universe in the stratum
	N_HOSP_U	Number of hospitals in the target universe in the stratum
	S_DISC_U	Number of NRD discharges in the stratum
	S_HOSP_U	Number of NRD hospitals in the stratum
	TOTAL_DISC	Total number of discharges for this hospital in the NRD

 

Return to Introduction

 

APPENDIX E: EVALUATION OF THE DIFFERENCE IN READMISSION RATES CAUSED BY HCUP PATIENT LINKAGE NUMBERS BEING SPECIFIC TO A STATE

The HCUP revisit variables (NRD_VisitLink and NRD_DaysToEvent) that can be used to track patients across hospitalizations were created from patient linkage numbers provided by the HCUP Partners. These identifiers can only track patients across hospitals in a single State. Consider the following two illustrative examples:

1. If a person is hospitalized in February in New York and then hospitalized in April to the same hospital or to a different hospital in New York, the HCUP revisit variables for the New York SID will be able to track the patient across the two visits.
2. If a person is hospitalized in February in New York and then hospitalized in April to a hospital in Florida, the HCUP revisit variables for the New York SID will not track the patient between New York and Florida.

The HCUP revisit variables are specific to tracking patients hospitalized in a State, regardless of whether the patient is a resident of the State. This limitation in the HCUP revisit variables causes the readmission rates to be artificially low. Consider the following example:

• Patient A has two index events in California and was readmitted one time in California within 30 days.
• Patient B has three index events in California, but was in Florida when readmitted within 30 days.

The true readmission rate across these two patients is 0.40 (0.40 = 2 readmissions / 5 index events). Using only California data, the readmission rate is 0.20 (0.20 = 1 readmissions / 5 index events). Using California and Florida data combined, the readmission rate is 0.17 (0.17 = 1 readmissions / 6 index events; there are 6 index events because the readmission for patient B in Florida gets counted as an index because it cannot be tied to the hospitalization in California).

We used the 2011 Medicare Standard Analytic File (SAF) to examine the impact on readmission rates caused by having State-specific patient linkage numbers. The SAF include patient linkage numbers that follow Medicare Fee-For-Service (FFS) patients across States; therefore, they do not have the same limitations as the HCUP data. We calculated condition-specific readmission rates in two ways. The index event for both readmission rates was allowed to include resident and nonresident discharges, similar to the NRD. The index events were grouped by the AHRQ Clinical Classifications Software (CCS) category for the principal diagnosis. For the first set of readmission rates, we required that a 30-day readmission for any cause occur in the same State as the index event, similar to the NRD. For the second set of readmission rates, we allowed the 30-day readmission for any cause to occur in any State. We limited the comparison of readmission rates to CCS categories with at least 100 index events. There were 247 CCS categories and the number of index events ranged from 101 to 468,709.

The analysis of readmission rates using the 2011 Medicare FFS data demonstrated that condition-specific readmission rates were higher if a patient could be tracked across all States, but that the percentage increase was less than 5 percent for most of the CCS categories. This analysis was limited in that it focused on the Medicare FFS population. The Medicare population accounted for 40 percent of all inpatient discharges in 2011,¹ and previous research indicates that this population has higher readmission rates than discharges for other payers.² Conditions often associated with younger adults, such as pregnancy, were included in the Medicare estimates because 20 percent of Medicare discharges are under the age of 65.³ Medicare patients under the age of 65 include people who are disabled or who have been diagnosed with end-stage renal disease or amyotrophic lateral sclerosis (ALS). Given the volume and severity of illness for Medicare patients, the estimates for the increase in the condition-specific readmission rates using the Medicare data provided a reasonable upper bound on the impact. The following three tables provide more detail on specific changes in readmission rates.

Table E.1 lists the percentage increase between the two types of readmission rates for the 20 conditions with the largest number of index events. These 20 conditions represent 54 percent of the index events in the analysis. The increase in the readmission rates when we capture readmission in other States ranged from 1.9 percent for urinary tract infections to 3.7 percent for chest pain.

 

Return to Introduction

 

Table E.1. Ten Conditions with the Largest Number of Index Events, Restricting Readmissions to Within State and Across All States

Clinical Classifications Software Principal Diagnosis Category for the Index Event	Number of Index Events	30-Day Readmission Rate per 100 Index Events		Percentage Increase in the Readmission Rates When Patients Were Followed Across States
		Readmissions Limited to the Same State as the Index Event	Readmissions Considered from Any State
108: Congestive heart failure; nonhypertensive	468,709	25.05	25.70	2.6
122: Pneumonia (except that caused by tuberculosis or sexually transmitted disease)	443,388	18.25	18.69	2.4
2: Septicemia (except in labor)	405,898	21.42	21.87	2.1
203: Osteoarthritis	363,295	5.57	5.70	2.2
127: Chronic obstructive pulmonary disease and bronchiectasis	356,562	21.99	22.40	1.9
106: Cardiac dysrhythmias	345,225	16.73	17.27	3.3
159: Urinary tract infections	277,674	18.31	18.66	1.9
237: Complication of device; implant or graft	274,669	21.23	21.83	2.8
109: Acute cerebrovascular disease	232,510	14.73	15.23	3.4
157: Acute and unspecified renal failure	228,134	22.04	22.60	2.5
101: Coronary atherosclerosis and other heart disease	220,481	15.44	15.99	3.5
100: Acute myocardial infarction	205,949	19.66	20.41	3.8
55: Fluid and electrolyte disorders	192,425	20.33	20.81	2.4
205: Spondylosis; intervertebral disc disorders; other back problems	178,614	10.62	10.93	2.9
197: Skin and subcutaneous tissue infections	178,097	16.09	16.45	2.2
226: Fracture of neck of femur (hip)	172,566	13.89	14.19	2.2
50: Diabetes mellitus with complications	159,421	23.81	24.31	2.1
238: Complications of surgical procedures or medical care	155,584	20.50	21.08	2.9
153: Gastrointestinal hemorrhage	155,410	18.75	19.27	2.8
102: Nonspecific chest pain	148,512	13.50	14.00	3.7

Table E.2 lists the 10 conditions with the largest percentage differences between the two types of readmission rates. We expect the readmission rates using discharges across all States to be higher than the readmissions restricted to the same State as the index event. Only three CCS categories had an increase of more than 10 percent in the two readmission rates, and these CCS categories had a very small numbers of index events. The other seven CCS categories had a percentage increase between 6.6 percent and 7.9 percent and also had relatively small numbers of index events.

 

Return to Introduction

 

Table E.2. Ten Conditions with the Largest Difference in Readmission Rates, Restricting Readmissions to Within State and Across All States

Clinical Classifications Software Principal Diagnosis Category for the Index Event	Number of Index Events	30-Day Readmission Rate per 100 Index Events		Percentage Increase in the Readmission Rates When Patients Were Followed Across States
		Readmissions Limited to the Same State as the Index Event	Readmissions Considered from Any State
188: Fetopelvic disproportion; obstruction	153	4.58	5.23	14.3
20: Cancer; other respiratory and intrathoracic	952	17.65	19.85	12.5
655: Mental disorders usually diagnosed in infancy, childhood, or adolescence	115	15.65	17.39	11.1
227: Spinal cord injury	2,420	17.69	19.09	7.9
185: Prolonged pregnancy	665	2.11	2.26	7.2
77: Encephalitis (except that caused by tuberculosis or sexually transmitted disease)	1,945	17.22	18.46	7.2
240: Burns	4,110	16.40	17.54	7.0
213: Cardiac and circulatory congenital anomalies	2,502	13.75	14.71	7.0
670: Miscellaneous disorders	3,592	13.34	14.23	6.7
96: Heart valve disorders	47,909	21.73	23.16	6.6

Table E.3 has the complete listing of readmissions rates for the 247 CCS categories. The majority of CCS categories (221 of the 247, 89.5 percent) had a percentage increase between the two types of readmission rates of less than 5 percent.

Table E.3. Readmission Rates Restricting Readmissions to Within State and Across All States, Medicare Standard Analytic File, 2011

Clinical Classifications Software Principal Diagnosis Category for the Index Event	Number of Index Events	30-Day Readmission Rate per 100 Index Events		Percentage Increase in the Readmission Rates When Patients Were Followed Across States
		Readmissions Limited to the Same State as the Index Event	Readmissions Considered from Any State
1: Tuberculosis	955	21.99	22.72	3.3
2: Septicemia (except in labor)	405,898	21.42	31.87	2.1
3: Bacterial infection; unspecified site	1,625	18.22	18.71	2.7
4: Mycoses	10,755	27.56	28.10	2.0
5: HIV infection	7,495	26.99	27.66	2.5
6: Hepatitis	10,121	35.23	36.29	3.0
7: Viral infection	14,033	16.01	16.46	2.8
8: Other infections; including parasitic	4,117	13.46	14.09	4.7
9: Sexually transmitted infections (not HIV or hepatitis)	699	14.16	14.45	2.0
10: Immunizations and screening for infectious disease	149	14.09	14.09	60.0
11: Cancer of head and neck	8,915	20.43	21.40	0.0
12: Cancer of esophagus	3,956	26.69	27.68	3.7
13: Cancer of stomach	7,171	23.79	24.68	3.8
14: Cancer of colon	37,987	17.16	17.53	2.2
15: Cancer of rectum and anus	12,496	21.26	21.68	2.0
16: Cancer of liver and intrahepatic bile duct	5,935	24.08	25.14	4.4
17: Cancer of pancreas	12,641	24.41	25.74	5.5
18: Cancer of other GI organs; peritoneum	7,043	24.00	25.27	5.3
19: Cancer of bronchus; lung	51,885	20.46	21.12	3.2
20: Cancer; other respiratory and intrathoracic	952	17.65	19.85	12.5
21: Cancer of bone and connective tissue	3,007	20.02	20.82	4.0
22: Melanomas of skin	989	13.25	13.65	3.0
23: Other non-epithelial cancer of skin	2,483	15.99	16.31	2.0
24: Cancer of breast	18,788	9.27	9.41	1.5
25: Cancer of uterus	10,304	11.81	12.35	4.5
26: Cancer of cervix	1,616	20.92	21.78	4.1
27: Cancer of ovary	6,450	21.83	22.71	4.0
28: Cancer of other female genital organs	2,155	15.82	16.15	2.0
29: Cancer of prostate	24,323	6.93	7.13	3.0
31: Cancer of other male genital organs	308	22.73	23.38	2.9
32: Cancer of bladder	14,780	22.18	23.03	3.8
33: Cancer of kidney and renal pelvis	14,717	13.23	18.84	4.6
34: Cancer of other urinary organs	1,716	16.14	16.96	5.1
35: Cancer of brain and nervous system	7,670	20.38	21.51	5.6
36: Cancer of thyroid	3,775	9.35	9.48	1.4
37: Hodgkin's disease	584	32.36	33.39	3.2
38: Non-Hodgkin's lymphoma	12,102	34.35	35.44	3.2
39: Leukemias	8,871	33.62	34.86	3.7
40: Multiple myeloma	5,748	28.97	29.91	3.2
41: Cancer; other and unspecified primary	1,514	16.84	17.90	6.3
42: Secondary malignancies	65,482	23.08	23.97	3.8
43: Malignant neoplasm without specification of site	2,667	24.03	24.82	3.3
44: Neoplasms of unspecified nature or uncertain behavior	14,836	25.74	26.54	3.1
45: Maintenance chemotherapy; radiotherapy	25,281	61.29	61.98	1.1
46: Benign neoplasm of uterus	4,909	7.29	7.37	1.1
47: Other and unspecified benign neoplasm	34,259	12.74	13.18	3.4
48: Thyroid disorders	9,288	10.81	11.00	1.8
49: Diabetes mellitus without complication	3,189	16.09	16.31	1.4
50: Diabetes mellitus with complications	159,421	23.81	24.31	2.1
51: Other endocrine disorders	23,041	20.09	20.52	2.1
52: Nutritional deficiencies	4,402	24.49	24.97	1.9
53: Disorders of lipid metabolism	147	14.29	14.29	0.0
54: Gout and other crystal arthropathies	9,186	17.46	17.92	2.6
55: Fluid and electrolyte disorders	192,425	20.33	20.81	2.4
56: Cystic fibrosis	1,507	20.31	20.77	2.3
57: Immunity disorders	306	28.76	30.07	4.5
58: Other nutritional; endocrine; and metabolic disorders	33,388	16.81	17.25	2.6
59: Deficiency and other anemia	92,656	23.19	23.74	2.4
60: Acute posthemorrhagic anemia	16,095	21.09	21.66	2.7
61: Sickle cell anemia	15,360	34.26	35.22	2.8
62: Coagulation and hemorrhagic disorders	9,731	27.73	28.53	2.9
63: Diseases of white blood cells	17,437	26.37	27.06	2.6
64: Other hematologic conditions	1,256	22.05	22.93	4.0
76: Meningitis (except that caused by tuberculosis or sexually transmitted disease)	3,339	16.17	17.07	5.6
77: Encephalitis (except that caused by tuberculosis or sexually transmitted disease)	1,945	17.22	18.46	7.2
78: Other CNS infection and poliomyelitis	2,052	22.86	23.64	3.4
79: Parkinson's disease	6,647	13.25	13.44	1.4
80: Multiple sclerosis	6,730	13.80	14.28	3.4
81: Other hereditary and degenerative nervous system conditions	17,788	16.29	16.83	3.3
82: Paralysis	1,732	17.21	17.55	2.0
83: Epilepsy; convulsions	70,843	15.62	16.10	3.1
84: Headache; including migraine	12,433	13.04	13.56	4.0
85: Coma; stupor; and brain damage	7,730	16.96	17.50	3.2
87: Retinal detachments; defects; vascular occlusion; and retinopathy	1,091	9.90	10.36	4.6
88: Glaucoma	274	13.14	13.50	2.8
89: Blindness and vision defects	1,943	12.30	13.02	5.9
90: Inflammation; infection of eye (except that caused by tuberculosis or sexually transmitted disease)	3,104	13.24	13.85	4.6
91: Other eye disorders	1,835	11.55	12.15	5.2
92: Otitis media and related conditions	1,410	11.49	12.13	5.6
93: Conditions associated with dizziness or vertigo	29,165	8.27	8.53	3.2
94: Other ear and sense organ disorders	1,672	13.22	13.64	3.2
95: Other nervous system disorders	88,536	18.37	18.94	3.1
96: Heart valve disorders	47,909	21.73	23.16	6.6
97: Peri-; endo-; and myocarditis; cardiomyopathy (except that caused by tuberculosis or sexually transmitted disease)	19,317	21.89	22.71	3.8
98: Essential hypertension	21,797	11.97	12.30	2.8
99: Hypertension with complications and secondary hypertension	101,943	24.08	24.58	2.1
100: Acute myocardial infarction	205,949	19.66	20.41	3.8
101: Coronary atherosclerosis and other heart disease	220,481	15.44	15.99	3.5
102: Nonspecific chest pain	148,512	13.50	14.00	3.7
103: Pulmonary heart disease	67,061	16.71	17.20	2.9
104: Other and ill-defined heart disease	3,266	13.38	14.15	5.7
105: Conduction disorders	29,215	11.42	11.86	3.9
106: Cardiac dysrhythmias	345,225	16.73	17.27	3.3
107: Cardiac arrest and ventricular fibrillation	3,236	18.88	19.96	5.7
108: Congestive heart failure; nonhypertensive	468,709	25.05	25.70	2.6
109: Acute cerebrovascular disease	232,510	14.73	15.23	3.4
110: Occlusion or stenosis of precerebral arteries	59,225	10.67	10.97	2.8
111: Other and ill-defined cerebrovascular disease	7,927	11.30	11.86	4.9
112: Transient cerebral ischemia	80,396	11.31	11.66	3.1
113: Late effects of cerebrovascular disease	8,073	15.88	16.31	2.7
114: Peripheral and visceral atherosclerosis	77,210	19.15	19.65	2.6
115: Aortic; peripheral; and visceral artery aneurysms	35,223	15.16	15.95	5.2
116: Aortic and peripheral arterial embolism or thrombosis	11,656	22.50	23.07	2.6
117: Other circulatory disease	61,972	17.52	17.98	2.6
118: Phlebitis; thrombophlebitis and thromboembolism	67,493	16.60	17.07	2.8
119: Varicose veins of lower extremity	1,061	16.21	16.68	2.9
120: Hemorrhoids	12,573	17.41	17.74	1.9
121: Other diseases of veins and lymphatics	9,833	20.34	21.06	3.5
122: Pneumonia (except that caused by tuberculosis or sexually transmitted disease)	443,388	18.25	18.69	2.4
123: Influenza	11,775	11.77	12.03	2.2
124: Acute and chronic tonsillitis	809	7.17	7.54	5.2
125: Acute bronchitis	26,446	13.00	13.32	2.4
126: Other upper respiratory infections	9,847	13.01	13.48	3.6
127: Chronic obstructive pulmonary disease and bronchiectasis	356,562	21.99	22.40	1.9
128: Asthma	85,354	18.61	18.99	2.0
129: Aspiration pneumonitis; food/vomitus	95,063	20.93	21.22	1.4
130: Pleurisy; pneumothorax; pulmonary collapse	37,794	25.02	25.81	3.1
131: Respiratory failure; insufficiency; arrest (adult)	136,502	24.58	25.19	2.5
132: Lung disease due to external agents	2,365	20.25	21.02	3.8
133: Other lower respiratory disease	44,901	18.94	19.54	3.1
134: Other upper respiratory disease	12,106	18.43	19.09	3.6
135: Intestinal infection	86,677	22.59	23.04	2.0
136: Disorders of teeth and jaw	3,436	12.34	12.75	3.3
137: Diseases of mouth; excluding dental	5,261	14.94	15.32	2.5
138: Esophageal disorders	41,912	16.29	16.70	2.5
139: Gastroduodenal ulcer (except hemorrhage)	12,983	17.04	17.62	3.4
140: Gastritis and duodenitis	34,213	19.14	19.57	2.2
141: Other disorders of stomach and duodenum	26,484	24.64	25.32	2.8
142: Appendicitis and other appendiceal conditions	18,638	9.55	9.84	3.0
143: Abdominal hernia	62,904	13.38	13.74	2.6
144: Regional enteritis and ulcerative colitis	16,244	23.15	24.02	3.7
145: Intestinal obstruction without hernia	131,147	17.46	17.93	2.7
146: Diverticulosis and diverticulitis	107,020	14.55	14.91	2.5
147: Anal and rectal conditions	13,837	16.30	16.68	2.3
148: Peritonitis and intestinal abscess	8,092	27.39	28.28	3.2
149: Biliary tract disease	107,986	14.87	15.38	3.4
151: Other liver diseases	36,907	32.88	34.08	3.7
152: Pancreatic disorders (not diabetes)	66,802	19.80	20.58	4.0
153: Gastrointestinal hemorrhage	155,410	18.75	19.27	2.8
154: Noninfectious gastroenteritis	41,796	16.12	16.57	2.8
155: Other gastrointestinal disorders	73,673	20.39	20.92	2.6
156: Nephritis; nephrosis; renal sclerosis	1,488	25.00	25.54	2.2
157: Acute and unspecified renal failure	228,134	22.04	22.60	2.5
158: Chronic kidney disease	8,075	26.13	26.81	2.6
159: Urinary tract infections	277,674	18.31	18.66	1.9
160: Calculus of urinary tract	32,158	13.21	13.53	2.4
161: Other diseases of kidney and ureters	12,302	19.05	19.64	3.1
162: Other diseases of bladder and urethra	8,649	18.28	18.582	3.0
163: Genitourinary symptoms and ill-defined conditions	15,706	20.16	20.69	2.6
164: Hyperplasia of prostate	23,248	11.51	11.75	2.1
165: Inflammatory conditions of male genital organs	6,873	13.90	14.22	2.3
166: Other male genital disorders	2,842	15.94	16.50	3.5
167: Nonmalignant breast conditions	2,997	13.38	13.65	2.0
168: Inflammatory diseases of female pelvic organs	3,024	15.18	15.68	3.3
169: Endometriosis	940	7.98	8.19	2.7
170: Prolapse of female genital organs	25,446	3.53	3.66	3.7
171: Menstrual disorders	3,085	9.17	9.37	2.1
172: Ovarian cyst	2,420	10.29	10.50	2.0
173: Menopausal disorders	2,132	12.48	12.81	2.6
175: Other female genital disorders	7,410	11.86	12.27	3.4
177: Spontaneous abortion	119	6.72	6.72	0.0
180: Ectopic pregnancy	159	8.18	8.18	0.0
181: Other complications of pregnancy	3,919	20.31	20.80	2.4
182: Hemorrhage during pregnancy; abruptio placenta; placenta previa	266	17.67	18.80	26.4
183: Hypertension complicating pregnancy; childbirth and the puerperium	1,544	16.78	16.90	0.8
184: Early or threatened labor	1,068	23.69	24.06	1.6
185: Prolonged pregnancy	665	2.11	2.26	7.2
186: Diabetes or abnormal glucose tolerance complicating pregnancy; childbirth; or the puerperium	822	19.34	19.83	2.5
187: Malposition; malpresentation	414	4.59	4.59	2.0
188: Fetopelvic disproportion; obstruction	153	4.58	5.23	14.3
189: Previous C-section	2,229	3.99	4.08	2.3
190: Fetal distress and abnormal forces of labor	584	4.62	4.62	0.0
191: Polyhydramnios and other problems of amniotic cavity	821	8.53	8.77	2.9
192: Umbilical cord complication	421	1.43	1.43	0.0
193: OB-related trauma to perineum and vulva	1,159	1.12	1.12	0.0
195: Other complications of birth; puerperium affecting management of mother	2,716	8.51	8.62	1.3
196: Normal pregnancy and/or delivery	450	1.56	1.56	0.0
197: Skin and subcutaneous tissue infections	178,097	16.09	16.45	2.2
198: Other inflammatory condition of skin	2,901	21.06	21.68	2.9
199: Chronic ulcer of skin	25,685	21.13	21.62	2.3
200: Other skin disorders	2,589	17.69	18.00	1.7
201: Infective arthritis and osteomyelitis (except that caused by tuberculosis or sexually transmitted disease)	24,731	19.24	19.77	2.8
202: Rheumatoid arthritis and related disease	5,487	12.96	13.30	2.7
203: Osteoarthritis	363,295	5.57	5.70	2.2
204: Other non-traumatic joint disorders	16,187	11.90	12.27	3.1
205: Spondylosis; intervertebral disc disorders; other back problems	178,614	10.62	10.93	2.9
206: Osteoporosis	146	21.23	21.92	3.2
207: Pathological fracture	31,718	19.03	19.46	2.3
208: Acquired foot deformities	1,705	7.57	7.68	1.5
209: Other acquired deformities	17,245	9.92	10.22	3.0
210: Systemic lupus erythematosus and connective tissue disorders	3,920	26.79	27.37	2.2
211: Other connective tissue disease	47,933	13.59	14.00	3.0
212: Other bone disease and musculoskeletal deformities	21,751	12.33	12.63	2.5
213: Cardiac and circulatory congenital anomalies	2,502	13.75	14.71	7.0
214: Digestive congenital anomalies	713	15.71	16.69	6.3
215: Genitourinary congenital anomalies	1,201	16.24	16.57	2.1
216: Nervous system congenital anomalies	397	17.88	18.39	2.8
217: Other congenital anomalies	4,654	8.10	8.36	3.2
225: Joint disorders and dislocations; trauma-related	6,439	11.14	11.34	1.8
226: Fracture of neck of femur (hip)	172,566	13.89	14.19	2.2
227: Spinal cord injury	2,420	17.69	19.09	7.9
228: Skull and face fractures	7,091	11.90	12.38	4.0
229: Fracture of upper limb	43,161	11.92	12.33	3.5
230: Fracture of lower limb	61,816	13.59	13.96	2.7
231: Other fractures	87,204	14.05	14.50	3.2
232: Sprains and strains	9,057	11.52	11.91	3.4
233: Intracranial injury	55,231	16.40	17.16	4.6
234: Crushing injury or internal injury	13,358	15.29	15.98	4.5
235: Open wounds of head; neck; and trunk	5,616	12.29	12.84	4.5
236: Open wounds of extremities	5,507	13.78	14.11	2.4
237: Complication of device; implant or graft	274,669	21.23	21.83	2.8
238: Complications of surgical procedures or medical care	155,584	20.50	21.08	2.9
239: Superficial injury; contusion	16,861	15.88	16.23	2.2
240: Burns	4,110	16.40	17.54	7.0
241: Poisoning by psychotropic agents	14,978	12.82	13.27	3.5
242: Poisoning by other medications and drugs	27,934	15.48	15.99	3.3
243: Poisoning by nonmedicinal substances	2,659	11.51	12.19	5.9
244: Other injuries and conditions due to external causes	19,846	15.79	16.34	3.5
245: Syncope	95,001	11.90	12.29	3.2
246: Fever of unknown origin	17,447	19.54	20.14	3.1
247: Lymphadenitis	1,297	18.50	19.20	3.8
248: Gangrene	17,575	33.79	34.34	1.6
249: Shock	554	23.11	24.19	4.7
250: Nausea and vomiting	16,042	22.97	23.48	2.2
251: Abdominal pain	36,919	20.14	20.89	3.7
252: Malaise and fatigue	14,952	16.77	17.19	2.5
253: Allergic reactions	5,477	15.45	15.94	3.2
254: Rehabilitation care; fitting of prostheses; and adjustment of devices	1,558	11.04	11.10	0.6
256: Medical examination/evaluation	101	22.77	23.76	4.3
257: Other aftercare	1,889	14.08	14.45	2.6
258: Other screening for suspected conditions (not mental disorders or infectious disease)	522	17.43	17.63	1.1
259: Residual codes; unclassified	43,683	18.57	19.14	3.0
650: Adjustment disorders	1,361	13.45	13.89	3.3
651: Anxiety disorders	4,677	15.72	16.02	1.9
652: Attention-deficit	245	19.59	20.00	2.1
653: Delirium	37,650	14.28	14.72	3.1
654: Developmental disorders	613	19.09	19.41	1.7
655: Disorders usually diagnosed in infancy	115	15.65	17.39	11.1
656: Impulse control disorders	417	17.03	17.51	2.8
657: Mood disorders	45,648	18.81	19.38	3.0
658: Personality disorders	431	25.52	26.45	3.6
659: Schizophrenia and other psychotic disorders	47,114	22.04	22.37	1.5
660: Alcohol-related disorders	30,228	20.72	21.24	2.5
661: Substance-related disorders	32,424	18.68	19.28	3.2
662: Suicide and intentional self-inflicted injury	240	16.67	17.50	5.0
663: Screening and history of mental health and substance abuse codes	11,302	28.35	29.03	2.4
670: Miscellaneous disorders	3,592	13.34	14.23	6.7

 

Return to Introduction

 

¹ If the patient is transferred to an out-of-state hospital, the subsequent discharge would not be included in the NRD because the HCUP patient linkage numbers only can follow a patient within a State.
² Although the text refers to using the discharge and admission date, in reality, we used the NRD data element NRD_DaysToEvent and LOS to identify the sequential order of inpatient records and whether they stopped and started on the same day.
³ Admission and discharge dates were not available because of patient confidentiality restrictions.
⁴ Total hospital cost must be added to the NRD using the HCUP supplemental Cost-to-Charge Ratio files.
⁵ Changes in the NIS Sampling and Weighting Strategy for 1998. ONLINE January 18, 2002. Available at http://www.hcup-us.ahrq.gov/db/nation/nis/reports/Changes_in_NIS_Design_1998.pdf. Accessed September 15, 2011.
⁶ HCUPnet query on the expected primary payer for the 2011 Nationwide Inpatient Sample. Accessed December 9, 2014.
⁷ Wier LM, Barrett ML, Steiner C, Jiang HJ. All-Cause Readmissions by Payer and Age, 2008. HCUP Statistical Brief No. 115. June 2011. Rockville, MD: Agency for Healthcare Research and Quality. http://www.hcup-us.ahrq.gov/reports/statbriefs/sb115.pdf.
⁸ HCUPnet query on the expected primary payer and age for the 2011Nationwide Inpatient Sample. Accessed December 9, 2014.
⁹ Barrett M, Steiner C, Andrews R, Kassed C, Nagamine M. Methodological Issues when Studying Readmissions and Revisits Using Hospital Administrative Data. HCUP Methods Series Report No. 2011-01. Online March 9, 2011. Rockville, MD: U.S. Agency for Healthcare Research and Quality. http://www.hcup-us.ahrq.gov/reports/methods/2011_01.pdf.
¹⁰ Carlson BL, Johnson AE, Cohen SB. An evaluation of the use of personal computers for variance estimation with complex survey data. J Off Statistics. 1993;9(4):795-814.
¹¹ If the patient is transferred to an out-of-State hospital, the subsequent discharge would not be included in the NRD because the HCUP patient linkage numbers only can follow a patient within a State.
¹² Although the text refers to using the discharge and admission date, in reality, we used the NRD data element NRD_DaysToEvent and LOS to identify the sequential order of inpatient records and whether they stopped and started on the same day.

Appendix E Footnotes

¹ HCUPnet query on the expected primary payer for the 2011 Nationwide Inpatient Sample. Accessed December 9, 2014.
² Wier LM, Barrett ML, Steiner C, Jiang HJ. All-Cause Readmissions by Payer and Age, 2008. HCUP Statistical Brief #115. June 2011. Rockville, MD: Agency for Healthcare Research and Quality.
³ HCUPnet query on the expected primary payer and age for the 2011 Nationwide Inpatient Sample. Accessed December 9, 2014.

 

Return to Introduction