Chronic Disease Surveillance using Administrative Data

1
Chronic Disease Surveillance using
Administrative Data
Lisa M. Lix, PhD Souradet Shaw, MA
MANITOBA CENTRE FOR HEALTH POLICY University of
Manitoba, Canada
2
Lecture Outline

• What is chronic disease surveillance?
• Why use administrative data for chronic disease
  surveillance?
• Constructing case definitions
• Validating case definitions
• An example Diabetes
• Conclusions
• References

3
What is Chronic Disease Surveillance?

• Chronic diseases are not prevented by vaccines
  or generally cured by medication, nor do they
  just disappear. To a large degree, the major
  chronic disease killersare an extension of what
  people do, or not do, as they go about the
  business of daily living. (CDC, 2004)

4

• Surveillance is
• the ongoing systematic collection, analysis,
  and interpretation of outcome-specific data for
  use in planning, implementing, and evaluating
  public health practice (Thacker Berkelman,
  1988).
• Chronic disease surveillance involves activities
  related to the ongoing monitoring or tracking of
  chronic diseases.

5
Why Use Administrative Data for Chronic Disease
Surveillance?

• Administrative data are usually collected by
  government for some administrative purpose (e.g.,
  paying doctors or hospitals), but not primarily
  for research or surveillance.

6
Advantages of Using Administrative Data for
Surveillance

• The databases are often population based, so
  important population subgroups are not missed.
• Comparisons between disease cases and non-cases.
• Trends over time can often be monitored.

7
Limitations of Other Data Sources

• Vital statistics data
• Clinical registries
• Survey data

8
Limitations of Administrative Data

• Administrative data are collected for purposes
  of health system management and provider payment,
  and not for chronic disease surveillance. Thus,
  it is important to assess their validity for
  surveillance

9
Constructing Case Definitions

• Diagnoses/Treatment
• - Diagnosis codes
• International Classification of Diseases (ICD)
• to identify diagnosed cases
• Prescription drugs
• Anatomic, Therapeutic, Chemical (ATC) codes
• to identify treated cases of chronic disease

10
Validating Case Definitions

• Validation data source
• Measures of validity

11
Potential Validation Data Sources

• Population-based survey data
• Chart review

12
Measures of Validity

• Case definition validation measures
• Kappa statistic (?)
• Sensitivity
• Specificity
• Positive predicted value (PPV)
• Negative predicted value (NPV)

13
Calculation of Validation Indices for Chronic
Disease Case Definitions

• Validation Data
• Administrative
• Data
• Sensitivity A/(AC)100
• Specificity D/(BD)100
• PPV A/(AB)100
• NPV D/(CD)100

Has Disease Does Not Have Disease
Has Disease A B
Does Not Have Disease C D
14
An Example Diabetes

• Case Definitions
• ICD-9-CM code 250 was used to identify diabetes
  cases in hospital and medical data.
• ATC code A10 (drugs used in diabetes) was used to
  identify diabetes cases in prescription drug
  data.

15
Validating Diabetes Case Definitions

• Data from Canadian Community Health Survey
  (CCHS), Cycle 1.1, collected between September
  2000 and November 2001 were used for the
  validation.
• 18 diabetes case definitions were tested.

16
Validation Results
Table 1 Estimates of agreement, sensitivity, specificity, and predictive values for diabetes case definitions Table 1 Estimates of agreement, sensitivity, specificity, and predictive values for diabetes case definitions Table 1 Estimates of agreement, sensitivity, specificity, and predictive values for diabetes case definitions Table 1 Estimates of agreement, sensitivity, specificity, and predictive values for diabetes case definitions Table 1 Estimates of agreement, sensitivity, specificity, and predictive values for diabetes case definitions Table 1 Estimates of agreement, sensitivity, specificity, and predictive values for diabetes case definitions Table 1 Estimates of agreement, sensitivity, specificity, and predictive values for diabetes case definitions
Years Case Definition Kappa Sens () Spec () PPV () NPV ()
1 a 1H or 1P 0.77 76.9 98.7 79.2 98.5
b 1H or 2P 0.73 63.2 99.5 89.5 97.7
c 1H or 1P or 1Rx 0.81 85.8 98.6 79.4 99.1
2 d 1H or 1P 0.78 85.2 98.1 74.0 99.0
e 1H or 2P 0.82 79.5 99.3 87.9 98.7
f 1H or 1P or 1Rx 0.80 89.6 97.9 73.7 99.3
3 g 1H or 1P 0.75 87.8 97.4 68.7 99.2
h 1H or 2P 0.83 84.9 99.0 83.9 99.0
i 1H or 1P or 1Rx 0.76 90.5 97.3 68.2 99.4
Data are for fiscal years 2000/01 2002/03
17
Estimating Diabetes Prevalence

• Cross-sectional and longitudinal prevalence can
  be estimated using administrative data.

18
Manitoba Prevalence Estimates
Table 2 Crude prevalence estimates for diabetes case definitions for Manitoba, Canada Table 2 Crude prevalence estimates for diabetes case definitions for Manitoba, Canada Table 2 Crude prevalence estimates for diabetes case definitions for Manitoba, Canada
Years Case Definition Prevalence Estimate ()
1 a 1H or 1P 5.8
b 1H or 2P 4.4
c 1H or 1P or 1Rx 6.5
2 d 1H or 1P 7.1
e 1H or 2P 4.6
f 1H or 1P or 1Rx 7.5
3 g 1H or 1P 7.9
h 1H or 2P 6.3
i 1H or 1P or 1Rx 8.2
Data are for fiscal years 2000/01 2002/03
19
Conclusions

• Administrative data appear to be a valid tool for
  identifying diabetes cases.
• No case definition is the best there is
  usually a trade-off between choosing a sensitive
  or specific case definition.

20
Conclusions, contd

• There are advantages to using administrative data
  for chronic disease surveillance, including easy
  access in most jurisdictions.

21
References

• Blanchard J.F., Ludwig S., Wajda A., Dean H.,
  Anderson K., Kendall O., Depew N. Incidence and
  prevalence of diabetes in Manitoba, 1986-1991.
  Diabetes Care 199619807-811.
• CDC. The Burden of Chronic Diseases and Their
  Risk Factors National and State Perspectives
  2004. Atlanta Department of Health and Human
  Services 2004. Available at http//www.cdc.gov/n
  ccdphp/burdenbook2004.
• Chronic Disease Prevention Alliance of Canada
  (CDPAC). The Case for Change. Available from
  http//www.cdpac.ca/content/case_for_change.asp.
  Accessed on January 19, 2006.
• Cricelli C., Mazzaglia G., Samani F., Marchi M.,
  Sabatini A., Nardi R., Ventriglia G., Caputi A.P.
  Prevalence estimates for chronic diseases in
  Italy exploring the differences between
  self-report and primary care databases. J Pub
  Health Med 200325254-257.

22
7. References, contd

• 5. Hux J.E., Ivis F., Flintoft V., Bica A.
  Diabetes in Ontario determination of prevalence
  and incidence using a validated administrative
  data algorithm . Diabetes Care 200225512-516.
• 6. Kue-Young T. 2005. Population Health
  Concepts and Methods. 2nd Ed. Oxford University
  Press, New York.
• 7. Lix L., Yogendran M., Burchill C., Metge
  C., McKeen N., Bond R. Defining and Validating
  Chronic Diseases An Administrative Data
  Approach. Winnipeg, MB Manitoba Centre for
  Health Policy, 2006.

• Maskarinec G. Diabetes in Hawaii estimating
  prevalence from insurance claims data. Am J Pub
  Health 1997871717-1720.
• Powell K.E., Diseker R.A. III, Presley R.J.,
  Tolsma D., Harris S., Mertz K.J., Viel K., Conn
  D.I., McClellan W. Administrative data as a tool
  for arthritis surveillance estimating prevalence
  and utilization of services. J Pub Health Manag
  Pract 20039291-298

23
7. References, contd

• 10. Rector Rector T.S, Wickstrom S.L., Shah M,
  Thomas Greenlee N., Rheault P., Rogowski J. et
  al. Specificity and sensitivity of claims-based
  algorithms for identifying members of Medicare
  plus Choice health plans that have chronic
  medical conditions. HSR 2004391839-1861.
• Robinson J.R., Young T.K., Roos L.L., Gelskey
  D.E. Estimating the burden of disease. Comparing
  administrative data and self-reports. Med
  Care199735932-947.
• Thacker S.B., Berkelman R.L. Public health
  surveillance in the United States. Epidemiol Rev
  198810164-190.
• WHO. Preventing Chronic Diseases A Vital
  Investment. http//www.who.int/chp/chronic_disease
  _report/contents/en/index.html
• Shultz S.E., Kopec J.A. Impact of chronic
  conditions. Health Reports 20031441-53.

24
8. Acknowledgements

• This presentation is based on a Manitoba Centre
  for Health Policy (MCHP) report, Defining and
  Validating Chronic Diseases An Administrative
  Data Approach, published in 2006 (Manitoba
  Health project 2004/05-01).