1. Understanding and Using NAMCS and NHAMCS Data
Data Tools and Basic Programming Techniques
Donald Cherry
Ambulatory and Hospital Care Statistics Branch
Division of Health Care Statistics

2. Overview Some important features of NAMCS & NHAMCS File structure SETS
Exercises using SAS Proc Surveyfreq/Proc Surveymeans, SUDAAN, STATA
Downloading data & creating a SAS dataset
Simple frequencies with/without standard errors
Creating a new variable-Asthma
Visit rates for asthma-male/female
Total number of digestive write-in
procedures
Time spent with physician
Considerations
Summary
The presentation will be structured as follows…
I will talk a little about the type of data and how it is structured.
We will run the following exercises using SAS, SUDAAN, & STATA including—read list
I will talk about some considerations when using the data & the type of statistics generated
And finally, I will make some summary comments.

3. NAMCS and NHAMCS National Ambulatory Medical Care Survey (NAMCS)
Visits to nonfederal, office-based physicians
CHC’s sampled beginning in 2006
National Hospital Ambulatory Medical Care Survey (NHAMCS)
Visits to hospital outpatient and emergency departments
1) The NAMCS includes 112 PSUs (counties), Nonfederally employed, office-based physicians stratified by specialty.
2) About 30 visits per doctor over a randomly selected 1-week period.
3) For 2007 & 2008, the Disease and Prevention and Health Promotion (NCCDPHP) sponsored the inclusion of an additional 150 primary care physicians (general/family practice, internal medicine, and obstetrics/gynecology), 4)Also, the National Cancer Institute sponsored a supplementary sample of 200 oncologists.
5) The NAMCS sample design typically includes too few community health center (CHC) physicians for the estimates to be reliably presented. In order to improve the precision of CHC physician estimates, starting in 2006, a dual-sampling procedure was used to select CHC physicians and other providers.
1) For the NHAMCS, we also use the same 112 PSUs (counties), Panel of 600 non-Federal, general or short stay hospitals clinics (OPDs) and emergency service areas (EDs).
2) About 200 visits per OPD, and 100 visits per ED over random 4-week period.

4. NAMCS Sample Design Three stage design 112 PSUs
Physician practices within PSUs
Patient visits within practices
One-week reporting period
About 30 visits per doctor are typically sampled
For 2006—3,350 doctors sampled
104 CHC’s sampled & physician visits included in sample
Total visits 29,392

5. Scope of the NAMCS
Basic unit of sampling is the physician-patient visit
In scope visits:
Must occur in physician’s office
Must be for medical purposes
Administrative visits not sampled
House calls, s, phone calls not sampled

6. Scope of the NAMCS (cont.)
Physicians must be:
Classified by AMA or AOA as primarily engaged in office-based patient care
nonfederally employed
not in anesthesiology, radiology, or pathology
64 percent unweighted response rate in 2006
CHC’s are Federally Qualified or “look alike”

7. NHAMCS Sample Design Multistage probability design
First stage sample of 112 PSUs
Hospitals within PSUs
Clinics within OPDs, Emergency Service Area (ESA) within EDs
Patient visits within clinics, ESAs
4-week reporting period
382 hospitals sampled in 2006; 35,849 ED visits and 35,105 OPD visits

8. Scope of the NHAMCS Basic unit of sampling is patient visit
Emergency and outpatient departments of noninstitutional general and short-stay hospitals
Not Federal, military, or Veterans Administration facilities
Located in 50 states and D.C.

9. Sample Weight
Each NAMCS record contains a single weight, which we call Patient Visit Weight
Same is true for OPD records and ED records
This weight is used for both visits and drug/procedure mentions

10. Data Items Patient characteristics Age, sex, race, ethnicity
Visit characteristics
Source of payment, continuity of care, reason for visit, diagnosis, treatment
Provider characteristics
Physician specialty, hospital ownership…
MULTUM drug characteristics added in 2006

11. Coding Systems Used Reason for Visit Classification (NCHS)
ICD-9-CM for diagnoses, causes of injury and procedures
Drug Classification System-MULTUM

12. File Structure http://www.cdc.gov/nchs/about/major/ahcd/ahcd1.htm
Download data and layout from website
Flat ASCII files for each setting and year:
NAMCS:
NHAMCS:
STATA files on Web:
NAMCS:
NHAMCS:

13. Creating a usable STATA dataset
Two options:
Use the self-extracting file in STATA folder to open a complete dataset for the NAMCS, NHAMCS-ED, & NHAMCS-OPD
Use the DO file (*.do) and the dictionary file (*.dct) along with the flat data file (*.exe) to create a dataset
StatTransfer

14. Organizational structure-NAMCS data
Provider information: MD vs. DO, Specialty,
Practice information: type of office setting, solo vs. non-solo, employment status of physician, ownership of practice, lab testing
Geographic info: region of country, MSA vs. non-MSA

15. SETS-Statistical Export and Tabulation System

16. Hands-on Exercises STATA Users
Double-click: My Computer\Local Disk C:\DUC_08
Open STATA
In the command window type:
Set mem 1000m
Set matsize 5000
Under the “File” icon-double-click namcs05.dta
Under “New Do File Editor”-double-click: STATA exercises.do
SAS/SUDAAN Users
Double-click: My Computer\Local Disk C:\DUC_08
Double-click: Final Exercises

17. Visit rate estimates Female population=800 Calculation* New variable
Phycode
Sex
Patwt
(Patwt/Pop)*100
Sexwt
1401 1
100
(100/800)*100
12.5
1820
300
(300/800)*100
37.5
1001 50
(50/800)*100
6.25
500
120
(120/800)*100 15
71.25 visits per 100 persons
In creating a visit rate (# of visits per 100 persons) what we actually do is modify the original weight.
For example, let’s say we have 4 visits to 4 individual physicians…
All the physicians have different patient weights.
The new calculation simply divides the patient wt by the population (which is a constant) to get our new Sexwt
When we sum the individual sex
Sample size=4
Visits=570
*Note: Rate=est/pop=Σ patwt/pop=1/pop*Σ patwt.

18. Calculating Total Number of Write-in Procedures
Record
Proc1
Proc2
Proc3
Proc4
Proc5
Proc6
Proc7
Proc8
Totproc 1
1911
0000 2
2182
2186 3
5490 4 5
8192
8200
Note: 0000=No procedure recorded.

19. Data Considerations

20. NAMCS vs. NHAMCS
Consider what types of settings are best for a particular analysis
Persons of color are more likely to visit OPD’s and ED’s than physician offices
Persons in some age groups make disproportionately larger shares of visits to ED’s than offices and OPD’s

21. Procedures Program Categorical Variables Continuous Variables SAS
PROC SURVEYFREQ
PROC SURVEYMEANS
STATA
SVY: TAB
SVY: MEAN
SUDAAN
PROC CROSSTAB
PROC DESCRIPT

22. How Good are the Estimates?
Depends … In general, OPD estimates tend to be somewhat less reliable than NAMCS and ED.
Since 1999, our Advance Data reports include standard errors in every table so it is easy to compute confidence intervals around the estimates.

23. RSE improves incrementally with the number of years combined
RSE = SE/x
RSE for percent of visits by persons less than 21 years of age with diabetes
1999 RSE = .08/.18 = .44 (44%)
1998 & 1999 RSE = .06/.18 = .33 (33%)
1998, 1999, & 2000 RSE = .05/.21 = .24 (24%)

24. Some User Considerations
NAMCS/NHAMCS sample visits, not patients
No estimates of incidence or prevalence
No state-level estimates
May capture different types of care for solo vs. group practice physicians
Data is only as good as what is documented in the medical record

25. If nothing else, remember…The Public Use Data File Documentation is YOUR FRIEND!
Each booklet includes:
A description of the survey
Record format
Marginal data (summaries)
Various definitions
Reason for Visit classification codes
Medication & generic names
Therapeutic classes