deriving practice-level estimates from physician-level surveys catharine w. burt, edd and esther...

21
Deriving practice-level Deriving practice-level estimates from physician- estimates from physician- level surveys level surveys Catharine W. Burt , EdD and Esther Hing, MPH. Chief, Ambulatory Care Statistics Branch Session 32 June 20, 2007 ICES III, Montreal, Canada U.S. DEPARTMENT OF HEALTH AND HUMAN SERVICES Centers for Disease Control and Prevention National Center for Health Statistics

Upload: todd-may

Post on 13-Jan-2016

215 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Deriving practice-level estimates from physician-level surveys Catharine W. Burt, EdD and Esther Hing, MPH. Chief, Ambulatory Care Statistics Branch Session

Deriving practice-level Deriving practice-level estimates from physician-level estimates from physician-level

surveyssurveys Catharine W. Burt , EdD and Esther Hing,

MPH.Chief, Ambulatory Care Statistics Branch

Session 32June 20, 2007

ICES III, Montreal, Canada

U.S. DEPARTMENT OF HEALTH AND HUMAN SERVICESCenters for Disease Control and PreventionNational Center for Health Statistics

Page 2: Deriving practice-level estimates from physician-level surveys Catharine W. Burt, EdD and Esther Hing, MPH. Chief, Ambulatory Care Statistics Branch Session

TopicsTopics

IntroductionIntroduction Multiplicity theoryMultiplicity theory Re-weighting methodsRe-weighting methods Application to NAMCSApplication to NAMCS AssumptionsAssumptions Analytical exampleAnalytical example LimitationsLimitations

Page 3: Deriving practice-level estimates from physician-level surveys Catharine W. Burt, EdD and Esther Hing, MPH. Chief, Ambulatory Care Statistics Branch Session
Page 4: Deriving practice-level estimates from physician-level surveys Catharine W. Burt, EdD and Esther Hing, MPH. Chief, Ambulatory Care Statistics Branch Session

Multiplicity theoryMultiplicity theory

Multiplicity occurs when the same observation Multiplicity occurs when the same observation unit can be counted multiple times among the unit can be counted multiple times among the selection units selection units eg., same patient is counted in multiple records of eg., same patient is counted in multiple records of

visits/discharges or same medical practice is counted visits/discharges or same medical practice is counted in records of multiple physiciansin records of multiple physicians

Using principles of network sampling, you can Using principles of network sampling, you can adjust weights to estimate the observations of adjust weights to estimate the observations of interest rather than the selection unitsinterest rather than the selection units

Page 5: Deriving practice-level estimates from physician-level surveys Catharine W. Burt, EdD and Esther Hing, MPH. Chief, Ambulatory Care Statistics Branch Session

Desired observation units

Survey selection units

Page 6: Deriving practice-level estimates from physician-level surveys Catharine W. Burt, EdD and Esther Hing, MPH. Chief, Ambulatory Care Statistics Branch Session

Greek for the GeeksGreek for the Geeks

XX i

Mi

j

jj

M

i

N i1

( )

i

X i Xj j( ) X ij ( ) 0

= the selection probability of physician i (i = 1, …, N) and

if physician i is affiliated with practice j, and

if physician i is not affiliated with practice j.

Page 7: Deriving practice-level estimates from physician-level surveys Catharine W. Burt, EdD and Esther Hing, MPH. Chief, Ambulatory Care Statistics Branch Session

Weight adjustment to estimate Weight adjustment to estimate XX

Observation weight

= selection weight / M

where M is the multiplicity information for the selection unit

Page 8: Deriving practice-level estimates from physician-level surveys Catharine W. Burt, EdD and Esther Hing, MPH. Chief, Ambulatory Care Statistics Branch Session

Re-weighting methodologyRe-weighting methodology

Assumptions and definitionsAssumptions and definitions Use multiplicity information from the physician Use multiplicity information from the physician

data to adjust physician-level estimates into data to adjust physician-level estimates into practice-level estimatespractice-level estimates

Dividing the physician sampling weight by Dividing the physician sampling weight by number of physicians in the practice provides number of physicians in the practice provides a measure of practicesa measure of practices

Page 9: Deriving practice-level estimates from physician-level surveys Catharine W. Burt, EdD and Esther Hing, MPH. Chief, Ambulatory Care Statistics Branch Session

Physicians Physicians ►► practices example… practices example…

Samples of physician records in medical Samples of physician records in medical practices practices

Physician data have the same practice Physician data have the same practice included in multiple observations.included in multiple observations.

If we knew how many physicians were in If we knew how many physicians were in the same practice as the sampled the same practice as the sampled physicians, then we can adjust the physicians, then we can adjust the estimator to account for the multiplicity.estimator to account for the multiplicity.

Page 10: Deriving practice-level estimates from physician-level surveys Catharine W. Burt, EdD and Esther Hing, MPH. Chief, Ambulatory Care Statistics Branch Session

Application to NAMCSApplication to NAMCS

National Ambulatory Medical Care SurveyNational Ambulatory Medical Care Survey Annual survey of 3,000 nationally representative Annual survey of 3,000 nationally representative

office-based physicians in patient careoffice-based physicians in patient care Excludes radiologists, anesthesiologists, and Excludes radiologists, anesthesiologists, and

pathologists and federally-employed physicianspathologists and federally-employed physicians Face-to-face induction interview asks physicians Face-to-face induction interview asks physicians

questions about his/her office practicequestions about his/her office practice Records are weighted by the inverse of the Records are weighted by the inverse of the

probability of selection, adjusted for nonresponse probability of selection, adjusted for nonresponse (~60% RR), with a calibration ratio to annual totals(~60% RR), with a calibration ratio to annual totals

Page 11: Deriving practice-level estimates from physician-level surveys Catharine W. Burt, EdD and Esther Hing, MPH. Chief, Ambulatory Care Statistics Branch Session

Induction interview contentInduction interview content

Number of locationsNumber of locations Number of other physicians Number of other physicians OwnershipOwnership Type of officeType of office

• Private, clinic, HMO, faculty practice plan, etcPrivate, clinic, HMO, faculty practice plan, etc

EMR adoptionEMR adoption Revenue sourcesRevenue sources

Page 12: Deriving practice-level estimates from physician-level surveys Catharine W. Burt, EdD and Esther Hing, MPH. Chief, Ambulatory Care Statistics Branch Session

AssumptionsAssumptions

Used the first location reportedUsed the first location reported Assumes practice information provided by Assumes practice information provided by

sample physician is a constant for the practicesample physician is a constant for the practice Does not account for multiplicity of practices Does not account for multiplicity of practices

within a physicianwithin a physician• i.e., Ignores the fact that some physicians are i.e., Ignores the fact that some physicians are

affiliated with multiple practices (about 1% of affiliated with multiple practices (about 1% of physicians) physicians)

Page 13: Deriving practice-level estimates from physician-level surveys Catharine W. Burt, EdD and Esther Hing, MPH. Chief, Ambulatory Care Statistics Branch Session

3 medical practices with a total of 7 physicians

Solo practice Partner practice Group practice

Page 14: Deriving practice-level estimates from physician-level surveys Catharine W. Burt, EdD and Esther Hing, MPH. Chief, Ambulatory Care Statistics Branch Session

4/72/71/7

Probability of selecting a practice

Page 15: Deriving practice-level estimates from physician-level surveys Catharine W. Burt, EdD and Esther Hing, MPH. Chief, Ambulatory Care Statistics Branch Session

4/72/71/7

1 .5 .25

Multiplicity factor

Page 16: Deriving practice-level estimates from physician-level surveys Catharine W. Burt, EdD and Esther Hing, MPH. Chief, Ambulatory Care Statistics Branch Session

Multiplicity informationMultiplicity information

How many other physicians practice with How many other physicians practice with you at this location?you at this location?

M= 1+ # of other physiciansM= 1+ # of other physicians

Practice weight = physician weight / MPractice weight = physician weight / M

Page 17: Deriving practice-level estimates from physician-level surveys Catharine W. Burt, EdD and Esther Hing, MPH. Chief, Ambulatory Care Statistics Branch Session

Re-weighting exampleRe-weighting example

Practice Practice sizesize

Physician Physician weightweight

Multiplicity Multiplicity adjustmentadjustment

PracticePractice

weightweight

solosolo 1010 11 1010

partnerpartner 2020 .5.5 1010

33 4040 .333.333 13.313.3

44 4040 .25.25 1010

Sum = 110 physicians 43 practices

Page 18: Deriving practice-level estimates from physician-level surveys Catharine W. Burt, EdD and Esther Hing, MPH. Chief, Ambulatory Care Statistics Branch Session

Practice weight = Practice weight = physician weight / practice sizephysician weight / practice size

physician weight physician weight → 311,200 physicians→ 311,200 physicians ± 8,000± 8,000

practice weight → 161,200 practicespractice weight → 161,200 practices ± 5,300± 5,300

Page 19: Deriving practice-level estimates from physician-level surveys Catharine W. Burt, EdD and Esther Hing, MPH. Chief, Ambulatory Care Statistics Branch Session

Percent distribution of office-based medical Percent distribution of office-based medical physicians and practices by sizephysicians and practices by size

36

12

27

1511

69

1114

41

0

10

20

30

40

50

60

70Percent

Solo Partner '3-5 '6-10 11+ Solo Partner '3-5 '6-10 11+

Physicians Practices

Page 20: Deriving practice-level estimates from physician-level surveys Catharine W. Burt, EdD and Esther Hing, MPH. Chief, Ambulatory Care Statistics Branch Session

Computerized administrative and clinical support systems

69.2

15.06.5

74.2

19.0

9.2

0

10

20

30

40

50

60

70

80

Uses electronic billing Uses EMR Uses CPOE

Practices Physicians

Page 21: Deriving practice-level estimates from physician-level surveys Catharine W. Burt, EdD and Esther Hing, MPH. Chief, Ambulatory Care Statistics Branch Session

Limitations of NAMCS data…Limitations of NAMCS data…

GoodGood National estimates of National estimates of

practicespractices Characteristics that Characteristics that

are common among are common among physiciansphysicians

BadBad Characterizing Characterizing

practicespractices Underestimates larger Underestimates larger

practicespractices Be careful how you Be careful how you

define sizedefine size• First-listed locationFirst-listed location• Location with most Location with most

visitsvisits• Location of the visitLocation of the visit