Post on 12-Jan-2016
National Hospital Discharge Survey(NHDS)
National Survey of Ambulatory Surgery
(NSAS)
Centers for Disease Control and Prevention
Session OverviewNHDS and NSAS: Overview
Bob Pokras
Analytic Issues
Jean Kozak, Ph.D.
Examples of Research
Marni Hall, Ph.D.
Accessing Data
Maria Owings, Ph.D.
New Directions
N H A N E S D V S N H IS N H C S
N a tio n a l C e n te r fo r H e a lth S ta tis t ics
National Health Care Surveys• Visits to
– Doctors’ offices (NAMCS)– Emergency rooms (NHAMCS)– Outpatient departments (NHAMCS)
• Inpatients (NHDS)• Ambulatory surgery (NSAS) (1994-96)
• Long term care– Nursing homes (NNHS)– Home health care (NHHCS)– Hospices (NHHCS)
Handout
Internet Resources: Hospital Discharge and Ambulatory Surgery Data
For an email of this table of hotlinks, write to: NHDS@cdc.gov
Survey Years
• NHDS: Annually 1965-present– Latest data available – 2000– 2001 will be available this Winter
• NSAS: Annually 1994-1996
Survey Design and Operations
• NCHS Publications
– NSAS – Series 1 No. 37
– NHDS – Series 1 No. 39
Survey Design
• Similar designs and methods
• National probability samples– Short stay non-Federal hospitals
(NHDS/NSAS)– Freestanding ambulatory surgery centers
(NSAS)
Survey Design
• Three stage design
• PSU
• Facility
• Discharge/visit
Facility Sample Size
• 525 NHDS hospitals
• 751 NSAS facilities– 418 Hospitals– 333 Freestanding surgery centers
Response Rates
• NHDS – over 90 percent 300,000 sampled discharges per year
• NSAS -- 80 percent for hospitals -- 70 percent for FSASC
120,000 sampled visits per year
Data Collection
• NHDS– Manual; 60%– Automated; 40%
• NSAS– All manual
Manual Data Collection
• NCHS – Statistical Design
• Census Bureau – Field Work
• ASI – Coding and Data Entry
Automated Data Collection
• Purchase files– States– Commercial firms– Individual hospitals
Data Collection
• NCHS
– Editing
– Estimation
Estimation
• Weight
– Inverse of the probability of selection
– Adjustments for non-response
– Population weighting ratio adjustment
Variables on
Public Use Data Files
Patient Data
• Age
• Sex
• Race
• Expected source of payment
• Discharge status
• Marital status
Facility Characteristics
• Geographic region
• Bed size (NHDS)
• Ownership (NHDS)
• Hospital vs. Freestanding (NSAS)
Medical Data
• Diagnoses and procedures
• International Classification of Diseases, 9th Revision, Clinical Modification (ICD-9-CM)
Additional Variables
• NHDS– Days of care– Month of admission/discharge– DRG
• NSAS– Month of visit– Type of anesthesia– Anesthesia provider
– WEIGHT
New Variables for NHDS
• Available for Year 2001 NHDS– Source of Admission– Type of Admission
Source of Admission
• Physician Referral• Clinical Referral• HMO Referral• Transfer from a Hospital• Transfer from Skilled Nursing Facility• Transfer from other health facility• Emergency Room• Court/Law Enforcement• Other• Not Available
Type of Admission
• Emergent
• Urgent
• Elective
• Newborn
• Not available/unknown
Summary
• Source of the data– Design– Methods
• Variables
Analytic Issues
(things you need to know about NHDS data)
Lola Jean Kozak, Ph.D.
Topics
• Utilization measures
• Populations
• Medical coding system
• Statistical issues
NHDS Provides Data on
Hospitalizations
Not People
Measures Include:
• Discharges
• Days of care
• Average length of stay
• Diagnoses
• Surgeries/procedures
Discharges
• Include deaths
• Include transfers to other hospitals or long-term care facilities
• Do not usually include newborn infants
Days of Care
• Total number of days discharged patients spend in the hospital
• All stays are counted as at least 1 day
• The admission day is counted, but not the discharge day
Average Length of Stay
• Calculated by dividing the number of days of care by the number of discharges
• May want to examine length of stay distributions
Length of stay for women with deliveries: 1995 and 2000
0
500
1,000
1,500
2,000
1 day 2 days 3-4 days 5-7 days 8+ days
1995 2000
Discharges in thousands
Diagnoses
• Disease, injury or other reason for hospitalization
• Coded according to US adaptations of the International Classification of Diseases
Diagnoses
• Principal diagnosis: chiefly responsible for hospitalization
• First-listed diagnosis: principal if specified, otherwise one listed first
Diagnoses
• All-listed: total number of times diagnoses appears on record
• Any-listed: discharges with diagnosis in any
position on record
Hospital discharges with fractures, 2000
Principal or first listed
Any listed All listed
982,000
1,226,000
1,542,000
Surgery/Procedures
• Surgical (appendectomy)• Diagnostic (spinal tap)• Therapeutic (chemotherapy) procedures
• Coded according to US adaptations of the International Classification of Diseases
NHDS Provides Data on
Inpatient Procedures
Not Total Procedures
Procedures mainly performed in inpatient settings, 1996
0 250 500 750
Ambulatory Inpatient
Cesarean section
Hysterectomy
Coronary artery bypass graft
Appendectomy
Number in thousands
Procedures mainly performed in ambulatory settings, 1996
0 500 1,000 1,500 2,000 2,500
Ambulatory Inpatient
Insertion of lens
Endoscopy oflarge intestine
D & C
Arthroscopy Of knee
Number in thousands
Population for Rates
• Mid-year population estimates from the U.S. Bureau of the Census
• Civilian resident population
• Adjustments for underenumeration
Versions of the International Classification of Diseases
• 8th revision used 1970-78
• 9th revision used 1979-2002
• 10th revision for use in future
8th Revision
• Some codes different than in 9th Revision
• Did not use E-codes
• Made modifications in coding to accommodate available data
9th Revision
• Addenda added annually since 1986
• Codes added, deleted, expanded, and revised
• Lists of changes available in annual summary reports, file documentation
Weights
• Must use weighted data to obtain unbiased national estimates.
• Each record has a weight
• Sum the weights of the records
Reliable Estimates
• Are based on 30 records or more*
• And have a relative standard error of 30 percent or less
*Use estimates based on 30-59 records with caution
Standard Errors
• Some standard errors are in Advance Data summaries
• Generalized error curves are in the Series 13 Annual Summaries and data documentation
• Use SUDAAN for specific standard errors - need access to confidential data
Examples of Research Using the National Hospital Discharge
Survey
Marni Hall, Ph.D.
Hospital Transfers to Long Term Care Facilities in the 1990’s
Lola Jean Kozak, Ph.D.
Long-Term Care Interface – June 2002
Transfers to long-term care,1990-1999
0
0.5
1
1.5
2
2.5
3
1990 1991 1992 1993 1994 1995 1996 1997 1998 1999
Num
ber
in M
illi
ons
1.6 Million
2.8 Million
Average hospital stay for long-term transfers, 1990-1999
0
2
4
6
8
10
12
14
1990 1991 1992 1993 1994 1995 1996 1997 1998 1999
12.8 days
8.3 days
Hospital discharges transferred to long-term care institutions by length of stay,
1990-1999
0 200 400 600 800 1000 1200
0-2 days
3-5 days
6-7 days
8+ days
19991990
Hos
pita
l Sta
y
Transfers in Thousands
Long-term transfers by first-listed diagnoses
0100300600 600300100
1990 1999
Circulatory
Respiratory
Injury & poisoning
Digestive
Musculoskeletal
Endocrine, metabolic
Genitourinary
Neoplasms
Infectious & parasitic
Mental disorders
Number in thousands
Hospital Transfers to Long Term Care Institutions study:
• Trend data – 10 years
• Changing roles of hospitals and nursing homes
• Assessment of the effects of Medicare policy changes
• Post-acute care in nursing homes substituting for end of hospital stay
Trends in Avoidable Hospitalization: United States,
1980-1998
Lola Jean Kozak, Ph.D.
Margaret J. Hall, Ph.D.
and Maria F. Owings Ph.D.
Avoidable hospitalization diagnoses
• Selected by a panel of physicians
• Can often be prevented, controlled, or managed over time without the need for hospitalization if the patient receives timely and appropriate ambulatory care
• Used as indicators of access and the adequacy of ambulatory care
Diagnoses studied in avoidable hospitalization study
• Pneumonia• Congestive heart failure• Asthma• Cellulitis• Perforated or bleeding ulcer• Pyelonephritis• Diabetes with ketoacidosis or coma• Ruptured appendix• Malignant hypertension• Hypokalemia• Immunizable conditions• Gangrene
Avoidable hospitalizations
1980 1998
• # of discharges 2,200,000 3,700,000 million
• Rate per 1,000
population 99.2 133.8
Trend in rate of avoidable hospitalizations
0
60
120
180
1980 1984 1988 1992 1996 1998
Rat
e p
er 1
0,00
0
99.2
133.8
Trend in rate of other hospitalizations
0
60
120
180
1980 1984 1988 1992 1996 1998
Rat
e p
er 1
,000
157.8
103.1
Trend in rate of avoidable hospitalizations
0
100
200
300
400
500
600
700
1980 1984 1988 1992 1996 1998
Rat
e p
er 1
0,00
0
Under 65
Over 65
364.6
573.5
65 71.1
Trend in rate of avoidable hospitalizations for those over 65
0
100
200
300
400
500
600
700
1980 1984 1988 1992 1996 1998
Rat
e p
er 1
0,00
0
Whites
Blacks
352.8
564
325.2
450
Trend in rate of avoidable hospitalizations for those under 65
40
60
80
100
120
1980 1984 1988 1992 1996 1998 1990 1998
Rat
e p
er 1
0,00
0
Blacks
Whites
92.5
113.5
53.849.1
Avoidable hospitalization study:
• Trend data – 20 years• Avoidable hospitalization conditions as defined by
the literature• Measured access to care over time• Identified disparities between elderly/nonelderly
and white/black and identified those who should be targeted for intervention
• Used as a model for additional research funded by Center for Medicare and Medicaid Services
Pneumonia hospital discharge rate for the elderly
0
50
100
150
200
250
19801984
19881992
19962000
Rat
e p
er
10,0
00
65 and over
125.3
221.2
Disparities in the Rate of Hospitalization for Pneumonia Patients in Rural and
Urban Areas
Maria F. Owings, Ph.D.
Margaret J. Hall, Ph.D.
Study Objectives
• To compare urban and rural patients hospitalized for pneumonia based on
– Patient characteristics
– County characteristics, including health services availability and socioeconomic status (SES)
Disparity in Urban/Rural Pneumonia Hospitalizations, 2000
0
20
40
60
80
37.5
78.0
Rat
e pe
r 10
,000
pop
ulat
ion
Rural Urban
Indicators of Severity of Illness Urban Rural
Average # diagnoses 5.1 4.9
% with serious comorbidities 36% 40%
Average # seriouscomorbidities 1.2 1.3
1
1
Significant Difference
Urban Rural
Average age 59 641
Average length of 6.2 5.3
1
stay (days)
Routine Discharge 70% 65%
1
Significant difference
Indicators of Severity of Illness
0
5
10
15
20
25
Urban Rural
% w/some college Unemployment rate
Education, Unemployment and Poverty
% in poverty
0
5
10
15
20
25
Urban Rural
Active MD’s Hospital Beds
MD/Hospital Availability
Rate per 1,000 elderly
HMO Penetration
0
1
2
3
Urban RuralRate per 1,000 elderly
What policies could reduce avoidable hospitalizations?
• Promotion of rural managed care
• Programs which attract/keep rural MD’s
• More affordable, accessible outpatient health care
• More health education / outreach programs - e.g. smoking cessation, influenza / pneumonia shots
Urban/rural pneumonia hospitalization study:
• Urban/rural indicators
• NHDS merged with Area Resource File (ARF) data
• Severity of illness indicators using NHDS data
• Policy recommendations
Medical Care Expenditures for Hypertension, Its Complications,
and Its Comorbidities
Thomas A. Hodgson, Ph.D. , NCHS
Liming Cai, Ph.D., NOVA Research Co.
Estimated the economic burden of hypertension using utilization for:
• First-listed hypertension
• Cardiovascular complications
• Unrelated conditions for which hypertensives are at greater risk
• Comorbidities, i.e. secondary diagnoses
Data from the Centers for Medicare and Medicaid Services (CMS):
• Personal Health Expenditures• Part B Data
Data from the Agency for Healthcare Research and Quality (AHRQ):
• National Medical Expenditure Survey
Data from the National Center for Health Statistics:
• National Hospital Discharge Survey
• National Ambulatory Medical Care Survey
• National Hospital Ambulatory Medical Care Survey
• National Home and Hospice Care Survey
• National Nursing Home Survey
• National Health Interview Survey
Data on hospital costs were calculated using these data:
• National Hospital Discharge Survey data on the number of inpatient days
• National Medical Expenditure Survey data on the average facility charge per hospital inpatient day
$ 4.2 billion - diagnosis of hypertension
$ 17.1 billion - cardiovascular complications
$ 24.2 billion - other diagnoses
___________________________________
$45.5 billion – total hospital expenditures attributed to
hypertension
Total hospital expenditures attributed to hypertension
Expenditures for hypertension, 1998
Hospital care 42%
Physician Services 26%
Prescription Drugs 17%
Nursing home 12%
Home health care 4%
$ 22.8 billion - diagnosis of hypertension
$ 29.7 billion - cardiovascular complications
$ 56.4 billion - other diagnoses
___________________________________
$108.8 billion - total expenditures attributed to hypertension
Total expenditures attributed to hypertension
Expenditures for hypertension study:
• Example of how cost data can be combined with utilization data
• Hospital care studied as part of entire spectrum of health services – how it fits into the total picture
• Uses multiple national data sources• Regression analyses• Provides data for cost benefit analysis
For more information see our Internet Resources handout
Accessing Data from NHDS and NSAS
Maria Owings, Ph.D.
Centers for Disease Control and Prevention
Sources of Available Data
• Publications, including annual reports– Downloadable from the Internet
– Data years 1985 through 2000– Order and purchase – years before 1993
• Data tables on selected topics –viewed or downloaded from Internet
• Public-use data files for DO-IT-YOURSELF analysis– Downloadable from the Internet – On CD-ROM
• ICD-9-CM – to assist in using medical data
Quick and Easy Access to NHDS and NSAS Data
• Telephone the Hospital Care Statistics Branch: 301-458-4321
• Send an email to: NHDS@cdc.gov
• Go to the NCHS website on the World Wide Web: www.cdc.gov/nchs
Annual Publications
• ADVANCE DATA on Vital and Health Statistics reports provide early release of NHDS data – Very general and usually short
• Series 13 Reports provide more specific statistics on hospital utilization– Are more comprehensive and contain
detailed tables of diagnoses and procedures
Recent Annual Publications
• 2000 NHDS Advance Data : http://www.cdc.gov/nchs/data/ad/ad329.pdf
• 1999 NHDS Annual Summary : http://www.cdc.gov/nchs/data/series/sr_13/sr13_151.pdf –Includes estimates of diagnoses and
procedures by detailed ICD-9-CM code number
What to Know to Access Data and Pubs on the WWW
• Publications and data tables are in Adobe Acrobat PDF format.
• Require use of the free Adobe Acrobat Reader software, available for download at www.adobe.com
Where to Find NHDS and NSAS Data and Pubs on the WWW
http://www.cdc.gov/nchs/about/major/hdasd/listpubs.htm • Lists annual pubs (back to 1990 only) and special topic
reports by name and numberhttp://www.cdc.gov/nchs/products.htm
• Provides links by topic area for all NCHS products (not just NHDS & NSAS), including
– Data Warehouse (for microdata and tabulations)– Published Reports (by type, e.g. Advance Data, Series
13 Vital and Health Statistics, etc)--ADs and Series 13 for pre-1990 years
NHDS & NSAS Homepage• http://www.cdc.gov/nchs/about/major/hdasd/nhds.htm provides
links to all aspects of survey design, data, and dissemination, including
– Survey Methodology and Data Collection
– Publications and Journal Articles
– Public Use Data Files (microdata)
– Special Reports
– NCHS Health E-Stats
– Data Highlights & Selected Tables on topics such as hospital discharges among females with deliveries, HIV inpatients, newborn infants, and hospital inpatient deaths, and ambulatory surgery utilization
Public-Use Files Available on the Internet
• Data and documentation available for free from the NCHS website
–NHDS: 1996 through 2000
–NSAS: 1994, 1995, 1996
• These are “raw” ASCII data that require the use of statistical software packages, such as SAS, SPSS, Stata, etc.
What to Know to Access Public-Use Files on the WWW
•Downloadable public-use data files are “zipped” for a speedier download.
•“Unzip” these files with
•WinZip at http://www.winzip.com/
• PKunzip at http://www.pkware.com/
•Data documentation are available either as text files or PDF files.
Public-Use Files Available onCD-ROM
• Two separate multi-year files containing
– 1979-2000 data years (ICD-9-CM coding)
– 1970-1978 data years (ICD-8 coding)
• Single year files for 1990, 1994 to 2000
• IMPORTANT: DRGs are available on single year files only. Multi-year files do NOT have DRGs.
How to Get PU Files on CD-ROM
• Can be obtained at no cost from NCHS • Division of Data Services: 301-458-INFO• Hospital Care Statistics Branch: 301-458-4321
• Or ordered from National Technical Information Service (NTIS)• by phone at 1-800-553-6847 or (703) 605-6000 • online at http://www.ntis.gov/
• Annual files for single years prior to 1994 can be ordered from NTIS, not directly from NCHS
ICD-9-CM
• For full-text, addenda, and conversion tables of ICD-9-CM, see www.cdc.gov/nchs/icd9.htm
• Full-text ICD-9-CM documents are RTF (Rich Text Format) files and can be handled with any word processing package.
• Addenda and conversion tables are PDF documents.
Restricted Data in NHDS
• HCSB maintains confidential information in files which are restricted from unauthorized use
• These data are available to researchers through the NCHS Research Data Center (RDC)
• http://www.cdc.gov/nchs/r&d/rdc.htm
Through the RDCResearchers Can Use:
• Confidential files for NHDS and NSAS variance estimation;
• NHDS and NSAS analytic files that have been linked with outside data sources
NCHS Research Data Center
• Located within NCHS facilities in Hyattsville, MD
• Requires preapproval of research projects by an internal proposal review committee
• Subjects analytic results to disclosure limitation review and clearance
• Provides different modes of data access for approved research projects
Confidential Variables Available Only on Restricted Files
• ZIPCODE for residence of discharged patient• ZIPCODE for hospital• STATE/COUNTY FIPS CODE for both patient and
hospital• AHA ID for hospital• DESIGN VARIABLES needed to run SUDAAN to
obtain variances of complex NHDS statistics********************************************
• NOTE: Patient name and address are NOT collected in the NHDS
Restricted Data Needed by SUDAAN for NHDS and NSAS Applications:
• Variables corresponding to design stages for sampling and stratification
• Population counts at each sampling stage
• Type of sampling performed at each stage
SUDAAN Software
• Incorporates design-related variables unique to each survey
• Utilizes sampling weights of discharges and visits that reflect unequal probabilities of selection
• Produces sampling errors for NHDS and NSAS estimates that take into account the complexity of the survey design
NHDS Linked Files
• NHDS + American Hospital Association (AHA)
• NHDS + Area Resource File (ARF)
• Linkage is with contextual NOT personal / demographic information
• Contextual data include
– Hospital characteristics, services (AHA)
– County level information (ARF)
American Hospital Association (AHA) Database
• Hospital-specific data on over 6,200 hospitals and health care systems
• More than 600 data items collected with the assistance of State and Metropolitan Hospital Associations
Types of Variables in AHA
• Organizational Structure• Staffing• Utilization• Facilities and Services• Financial• Geographic codes• Approval and Accreditation Codes
Area Resource File(ARF)
• County-specific health resources information system designed to aid research on the health care delivery system and factors that may impact health status and health care in the U.S.
• Contains more than 7,000 variables from over 50 different source files for each county.
General Categories of Variables in the ARF
• Health facilities
• Health professions
• Health care utilization
• Morbidity and mortality measures
• County economic activity
• Socioeconomic and environmental variables
Beyond 20/20 Browser
• http://www.cdc.gov/nchs/about/otheract/aging/howto.htm#browser2
• Database providing up-to-date information on national trends and key variables that depict the health status of older Americans
• Data for persons 45 years old and over by sex and race
Listserv
• http://www.cdc.gov/nchs/about/major/hdasd/nhdslistserv.htm
• Provides current information about new data releases and publications
• Subscribers can post messages to other members and exchange information
How to Subscribe to HDAS Listserv
• In the body of an email message (leaving the subject line blank), type:
• subscribe hdas-data your name
• Send this message to:
• listserv@cdc.gov
NHDS or NSAS Questions?
Phone: 301-458-4321
Fax: 301-458-4032
Email: NHDS@cdc.gov
New Directions
• Beyond 20/20
• Public use variance file
New Directions
• Clinical data
• Evaluation of drugs in the NHDS
• Two phase contract– Phase I – Research– Phase II – Field test