Supplementing Community Public Health Surveillance With Data From Electronic Healthcare Claims October 2002 Thomas Balzer, Ph.D. Chief Scientific Officer,

Supplementing Community Public Health Surveillance

With Data From Electronic Healthcare Claims October 2002

Thomas Balzer, Ph.D.Chief Scientific Officer, Verispan

Who is Verispan?

• Scott-Levin• SMG Marketing Group• Synergy Healthcare• Amaxis

• Kelly-Waldron• data sources

Public Health Surveillance

“Surveillance: Continuous analysis, interpretation, and feedback of systematically collected data, generally using methods distinguished by their practicality, uniformity, and rapidity rather than by accuracy or completeness.”

John M. Last, A Dictionary of Epidemiology, 3rd Edition

Simple (in concept)

Stable (in operation)

Acceptable (to providers)

Standardized high-quality data

Timely (in reporting healthcare events)

Representative (of all areas)

Sensitive (to outbreaks & other changes over time by applying traditional & non-traditional approaches to surveillance)

Flexible (to changing surveillance needs)

4

Critical Attributes ofPublic Health Surveillance Systems

Verispan Data Warehouse:“Practicality, Uniformity, Rapidity”

The largest, broadest real-time and longitudinal sample of patient-

centric pharmacy and medical transactions in the world.

The Verispan Data Warehouse Began in 1998“A Simple & Stable System Already Working”

50,000+ Pharmacies

640,000+ Unique Subscribers

5+ Million Claims Loaded Daily

100+ Million Unique Patients

1.7 Billion Annual Rx or Mx Claims

5 Billion Claims - total in warehouse

Verispan PatientData Base

Verispan PatientData Base

Pharmacies &Prescription

Services

Commercial(PBM, HMO)

Government(Medicare/Medicaid)

Hospitals & Facilities

Physicians

BCBS

Daily Claims VolumeDaily Claims Volume

HXHX

MXMX

RXRX

Payors

Health Care Clearinghouse

Verispan Harnesses Routine Billing Practices“Acceptable to Providers”

De-Identification

Medical, Hospital, and Pharmacy Data are Available“Verispan Has Standardized, High-Quality Data”

Pharmacy Data Medical Data

Jan ‘98 - to date July ‘98 - to date

RX Pharmacy Data(NCPDP)

RX Pharmacy Data(NCPDP)

Patient ID

Patient Age & Gender

Date Written

Date Filled

NDC Code

Quantity Dispensed

Days Supply

Refill Flag

Prescribing Physician

Pharmacy

Payor Type

Patient ID

Patient Age & Gender

Date Written

Date Filled

NDC Code

Quantity Dispensed

Days Supply

Refill Flag

Prescribing Physician

Pharmacy

Payor Type

MX Provider Data (HCFA 1500)

MX Provider Data (HCFA 1500)

Patient ID

Patient Age & Gender

Diagnosis Codes (ICD9)

Procedure Codes (CPT)

Service Dates

Physician/Provider ID

Location of Care

Payor Type

Patient ID

Patient Age & Gender

Diagnosis Codes (ICD9)

Procedure Codes (CPT)

Service Dates

Physician/Provider ID

Location of Care

Payor Type

HX Facility Data (UB-92)

HX Facility Data (UB-92)

Patient ID

Patient Age & Gender

Diagnosis Codes (ICD9)

Procedure Codes (CPT)

DRG

Admit Date

Discharge Date

Physician/Provider ID

Location of Care

Payor Type

Patient ID

Patient Age & Gender

Diagnosis Codes (ICD9)

Procedure Codes (CPT)

DRG

Admit Date

Discharge Date

Physician/Provider ID

Location of Care

Payor Type

Providers are Motivated to File Timely Electronic Claims“The Verispan Data Warehouse Is Updated Daily

Lag Days for enteric illness among children, 2000 - 2001

0

50

100

150

200

250

000 007 014 021 028 035 042 049 056 063 070 077 084

Diarrhea Enteritis Infectious_Diarrhea

First_Claim Y Patient_State OH PROVIDER_COUNTY Hamilton AGE_GROUP_10YR 00 to 09 Place_of_Visit A_OFFICE

Count of PATIENT_ID

REPORTING_LAG_DAYS

DISEASE

Lag Days = Processing Date minus Service Date

Excellent Geographic Distribution“All Areas of the U.S. Are Represented”

Every StateEvery MSAEvery 3 Digit Zip Code

Examples of Outbreak Detection Using Non-traditional Approaches

To Surveillance (NTAS)

For Additional Information:[email protected], Fax 919-998-7263

CDC Outbreak Detection Challenge

CDC developed three case studies for evaluating supplementary data bases for their ability to identify outbreaks. CDC provided only limited information about these known 2001 outbreaks:

Case 1: Shigella sonnei gastroenteritis in Ohio Case 2: Neisseria meningitides meningitis in Ohio in

school-age children Case 3: Histoplasma capsulatum in multiple states

among travelers to Acapulco

Only Verispan rose to the challenge of identifying the outbreak footprints in existing data bases.

A Traditional, Diagnosis-based Approach to Detecting an Enteric Illness Outbreak in Children, 2001

SHIGELLOSIS

0

5

10

2000

_01

2000

_03

2000

_05

2000

_07

2000

_09

2000

_11

2000

_13

2000

_15

2000

_17

2000

_19

2000

_21

2000

_23

2000

_25

2000

_27

2000

_29

2000

_31

2000

_33

2000

_35

2000

_37

2000

_39

2000

_41

2000

_43

2000

_45

2000

_47

2000

_49

2000

_51

2001

_01

2001

_03

2001

_05

2001

_07

2001

_09

2001

_11

2001

_13

2001

_15

2001

_17

2001

_19

2001

_21

2001

_23

2001

_25

2001

_27

2001

_29

2001

_31

2001

_33

2001

_35

2001

_37

2001

_39

2001

_41

2001

_43

2001

_45

2001

_47

2001

_49

2001

_51

SHIGELLOSIS

First_Claim Y Patient_State OH PROVIDER_COUNTYHamilton AGE_GROUP_10YR 00 to 09

Count of PATIENT_ID

SERVICE_EPI_WEEK

DISEASE

A Non-Traditional Approach to Outbreak DetectionUsing Surveillance of Enteric Syndromes

0

10

20

30

40

50

60

2000

_01

2000

_03

2000

_05

2000

_07

2000

_09

2000

_11

2000

_13

2000

_15

2000

_17

2000

_19

2000

_21

2000

_23

2000

_25

2000

_27

2000

_29

2000

_31

2000

_33

2000

_35

2000

_37

2000

_39

2000

_41

2000

_43

2000

_45

2000

_47

2000

_49

2000

_51

2001

_01

2001

_03

2001

_05

2001

_07

2001

_09

2001

_11

2001

_13

2001

_15

2001

_17

2001

_19

2001

_21

2001

_23

2001

_25

2001

_27

2001

_29

2001

_31

2001

_33

2001

_35

2001

_37

2001

_39

2001

_41

2001

_43

2001

_45

2001

_47

2001

_49

2001

_51

Diarrhea Enteritis Infectious_Diarrhea

First_Claim Y Patient_State OH PROVIDER_COUNTYHamilton AGE_GROUP_10YR 00 to 09

Count of PATIENT_ID

SERVICE_EPI_WEEK

DISEASE

Two Previously Unknown Outbreaks The Outbreak

A Traditional, Diagnosis-based Approach to Detecting a Community Meningitis Outbreak, 2001

0

1

2

3

4

5

2000

_01

2000

_03

2000

_05

2000

_07

2000

_09

2000

_11

2000

_13

2000

_15

2000

_17

2000

_19

2000

_21

2000

_23

2000

_25

2000

_27

2000

_29

2000

_31

2000

_33

2000

_35

2000

_37

2000

_39

2000

_41

2000

_43

2000

_45

2000

_47

2000

_49

2000

_51

2001

_01

2001

_03

2001

_05

2001

_07

2001

_09

2001

_11

2001

_13

2001

_15

2001

_17

2001

_19

2001

_21

2001

_23

2001

_25

2001

_27

2001

_29

2001

_31

2001

_33

2001

_35

2001

_37

2001

_39

2001

_41

2001

_43

2001

_45

2001

_47

2001

_49

2001

_51

Cuyahoga - 00 to 09 Cuyahoga - 10 to 19 Cuyahoga - 20 to 29 Hancock - 20 to 29 Summit - 10 to 19

First_Claim Y Patient_State (All) DISEASE Meningococcal infection Place_of_Visit (All)

Count of PATIENT_ID

SERVICE_EPI_WEEK

PROVIDER_COUNTY

AGE_GROUP_10YR

The Outbreak

A Non-Traditional Approach to Outbreak DetectionUsing Surveillance of Vaccination Procedures

Meningococcal Vaccination

0

10

20

30

40

50

60

70

2000

_01

2000

_03

2000

_05

2000

_07

2000

_09

2000

_11

2000

_13

2000

_15

2000

_17

2000

_19

2000

_21

2000

_23

2000

_25

2000

_27

2000

_29

2000

_31

2000

_33

2000

_35

2000

_37

2000

_39

2000

_41

2000

_43

2000

_45

2000

_47

2000

_49

2000

_51

2001

_01

2001

_04

2001

_06

2001

_08

2001

_10

2001

_12

2001

_14

2001

_17

2001

_19

2001

_21

2001

_23

2001

_25

2001

_27

2001

_29

2001

_31

2001

_33

2001

_35

2001

_37

2001

_39

2001

_41

2001

_43

2001

_45

2001

_47

2001

_49

2001

_51

10 to 19 - Cuyahoga 10 to 19 - Lake 10 to 19 - Mahoning 10 to 19 - Portage 10 to 19 - Stark 10 to 19 - Summit

20 to 29 - Cuyahoga 20 to 29 - Lake 20 to 29 - Mahoning 20 to 29 - Stark 20 to 29 - Summit

First_Claim Y Patient_State OH Proc Code 90733 Place_of_Visit (All)

Count of PATIENT_ID

PROCESS_EPI_WEEK

AGE_GROUP_10YR

PROVIDER_COUNTY

Unexpected Vaccination Pattern From the OutbreakExpected Vaccination

Pattern in Students Entering College

A Traditional, Diagnosis-based Approach to Detecting a National Histoplasmosis Outbreak, 2001 -

0

2

4

6

8

10

2000

_02

2000

_04

2000

_06

2000

_08

2000

_10

2000

_12

2000

_14

2000

_16

2000

_18

2000

_20

2000

_22

2000

_24

2000

_26

2000

_28

2000

_30

2000

_32

2000

_34

2000

_36

2000

_38

2000

_40

2000

_42

2000

_44

2000

_46

2000

_48

2000

_50

2000

_52

2001

_02

2001

_04

2001

_06

2001

_08

2001

_10

2001

_12

2001

_14

2001

_16

2001

_18

2001

_20

2001

_22

2001

_24

2001

_26

2001

_28

2001

_30

2001

_32

2001

_34

2001

_36

2001

_38

2001

_40

2001

_42

2001

_44

2001

_46

2001

_48

2001

_50

2001

_52

(bla

nk)

AZ CT DE FL GA MA ME MI MN NC NE NJ NY OK OR PA SC SD TX VA WI WV

First_Claim Y Histo_Endemic (blank) DISEASE Histoplasmosis AGE_GROUP_5YR (All)

Count of PATIENT_ID

PROCESS_EPI_WEEK2

Patient_State

PA, NY, NJ, MI, TX, DE, MN, NE

A Non-Traditional Approach to Outbreak DetectionUsing Surveillance of Ketoconazole Prescriptions

0

5

10

15

20

25

2000

_02

2000

_06

2000

_10

2000

_14

2000

_18

2000

_22

2000

_26

2000

_30

2000

_34

2000

_38

2000

_42

2000

_46

2000

_50

2001

_02

2001

_06

2001

_10

2001

_14

2001

_18

2001

_22

2001

_26

2001

_30

2001

_34

2001

_38

2001

_42

2001

_46

2001

_50

18 19 20 21 22 23 24

First RX Y HISTO_ENDEMIC(blank) NDC 51672402606 DRUGNAME KETOCONAZOLE PHARMACY_STATE (All)

Count of PATIENT_ID

SERVICE_EPI_WEEK2

AGE_IN_YEARS

Prescription “Footprint” of the Outbreak

Source: Outbreak of Escherichia coli O157:H7 and Campylobacter among attendees of the Washington

County Fair - New York, 1999 (MMWR 48(36); 803)

*Based on ICD•9•CM codes: 008.00, 008.04, 008.43, 009, 283.11, 787.91 [Values are raw and unadjusted]

6

1155

11

10

9

656

10

13

5

5

18

4

9

14

5

29

29

14

5

16

17

3

9

14

7

15

13

6

12

106

0

10

20

30

40

50

60

70

80

Cas

es

(b

y E

PI-

wee

k o

f h

ea

lth

ca

re v

isit

)

30 31(Aug)

32 33 34 35(Sep)

36 37 38 39 40(Oct)

County Fair E. coli 0157:H7 Outbreak*, NY State, 1999 Using the Informatics Mx Data Base

Wash. Co. Saratoga Co. Rensselaer Co. Warren Co.

Combining Traditional and Non-Traditional Approaches to Detect Outbreaks”

19

Examples of Unique Daily Public Health Reports of Syndromes, Rxs, and Reportable and Non-

Reportable Conditions That Are Available From Verispan Through the Web

For Additional Information:[email protected], Fax 919-998-7263

Syndromic Surveillance: Influenza-Like-IllnessAlbany, NY, MSA, Jan 01 to Aug 02

Syndromic Surveillance: SepticemiaFairfield County, CT, Jan 01 to Aug 02

Syndromic Surveillance: SepticemiaGeographic View

Syndromic Surveillance: Enteric IllnessHarrisburg, PA, MSA, Jan 01 to Aug 02

Prescription Surveillance: Anti-Influenza DrugsNew York City, Jan 01 to Aug 02

Surveillance of Non-Reportable Infectious Diseases:Influenza, Pittsburgh, MSA, Jan 01 to Aug 02

Surveillance of Reportable Infectious Diseases:Lyme Disease, CT, Jan 01 to Aug 02

Lyme Disease as Tracked by States & CDC in 2000 “Verispan Data Are Sensitive to Reportable Diseases”

Cumulative Cases Reported to CDC from State Health Departments: 2000

Cumulative Cases Reported to CDC from State Health Departments: 2000

Color Code Key:

# of Cases Color

3000+

100-2999

20-99

1-19

No Cases

Color Code Key:

# of Cases Color

3000+

100-2999

20-99

1-19

No Cases

2

7

104

0 0

0

0

0

0

0

0

3211

291393

89

13

28

34146

46

17

50

1,276

4,027

3984

1,098

34

45

2

4

17

4

4

1

36

1

11

9

3

4

9

15

4MD 559DC 11

DE 142NJ 1,467

CT 2,550RI 590

Lyme Disease:

ICD-9-CM: 088.81

Lyme Disease:

ICD-9-CM: 088.81

CDC preliminary case count: n = 13,309 (MMWR 49 [52])

28

Cumulative Cases Reported in Informatics Data: 2000Cumulative Cases Reported in Informatics Data: 2000

Lyme Disease:

ICD-9-CM: 088.81

Lyme Disease:

ICD-9-CM: 088.81

Color Code Key:

# of Cases Color

3000+

100-2999

20-99

1-19

No Cases

Color Code Key:

# of Cases Color

3000+

100-2999

20-99

1-19

No Cases

24

171

1

38316

2

0

2

18

2

10

40

209

2

7

38

29

113

14

40

29

190

332122

79

131

214

82

29 15

205

12760

238

469

112761

8,527

3466

56

411

MD 1,376DC 30

DE 603NJ 5,554

CT 6,268RI 56

0

Quintiles case count: n = 27,184

CDC preliminary case count: n = 13,309 (MMWR 49 [52])

Lyme Disease as Tracked by Quintiles, 2000“Quintiles Data Are Sensitive to Reportable Diseases”

29

Evaluate Community Responses to Emergencies“Verispan Data Are Flexible at the Local Level”

Cipro Daily Variation - WTC Area (60-Mile Radius)

-100%

0%

100%

200%

300%

400%

500%

08/0

1/20

01

08/0

8/20

01

08/1

5/20

01

08/2

2/20

01

08/2

9/20

01

09/0

5/20

01

09/1

2/20

01

09/1

9/20

01

09/2

6/20

01

10/0

3/20

01

10/1

0/20

01

10/1

7/20

01Va

ria

nc

e F

rom

Ex

pe

cte

d A

cti

vit

y

Departure from Expectation Lower Limit Upper Limit

30

9/11

10/12 - Anthrax

Generate Hypotheses for Further Study“Verispan Data Are Flexible at the Local Level”

Source: Verispan Mx Database

Asthma Diagnoses Daily Variation - NYC

-100%

-80%

-60%

-40%

-20%

0%

20%

40%

60%

80%

100%

08/0

7/20

01

08/1

0/20

01

08/1

3/20

01

08/1

6/20

01

08/1

9/20

01

08/2

2/20

01

08/2

5/20

01

08/2

8/20

01

08/3

1/20

01

09/0

3/20

01

09/0

6/20

01

09/0

9/20

01

09/1

2/20

01

09/1

5/20

01

09/1

8/20

01

09/2

1/20

01

09/2

4/20

01

09/2

7/20

01

Var

ian

ce F

rom

Exp

ecte

d A

ctiv

ity

Asthma Visits 95% Lower Conf 95% Upper Conf

Asthma Diagnoses Daily Variation - NYC

-100%

-80%

-60%

-40%

-20%

0%

20%

40%

60%

80%

100%08

/07/

2001

08/1

0/20

01

08/1

3/20

01

08/1

6/20

01

08/1

9/20

01

08/2

2/20

01

08/2

5/20

01

08/2

8/20

01

08/3

1/20

01

09/0

3/20

01

09/0

6/20

01

09/0

9/20

01

09/1

2/20

01

09/1

5/20

01

09/1

8/20

01

09/2

1/20

01

09/2

4/20

01

09/2

7/20

01

Var

ian

ce F

rom

Exp

ecte

d A

ctiv

ity

Asthma Visits 95% Lower Conf 95% Upper Conf

9/11

31

Oxycontin versus Control Set- Selected Counties5 Week Moving Average

0.00

0.01

0.02

0.03

0.04

0.05

0.06

0.07

0.08

0.09

0.10

2000

_01

2000

_04

2000

_07

2000

_10

2000

_13

2000

_16

2000

_19

2000

_22

2000

_25

2000

_28

2000

_31

2000

_34

2000

_37

2000

_40

2000

_43

2000

_46

2000

_49

2000

_52

2001

_03

2001

_06

2001

_09

2001

_12

2001

_15

2001

_18

2001

_21

2001

_24

2001

_27

2001

_30

2001

_33

2001

_36

2001

_39

2001

_42

2001

_45

2001

_48

2001

_51

Week

Ind

ex v

ersu

s C

on

tro

l Set

Miami-Dade, Florida Cook, Illinois Middlesex, Massachusetts Queens, New York

Control Group: All Prilosec and Amaryl NDC’s

Other Surveillance Opportunities

Verispan’ Unique Factors

Proven technology and systems in use for several years

Extensive database of over 100M de-identified patients

Prescription / Medical data integration processes

Daily receipt of ~5 million health claims

Data modeling and statistical strengths

Access to neural networking technology for detection

Existing broadcast technology for alert messages

33