Slide 1

Incorporation of Laboratory Results into Statewide Hospital Data

Todd Seto, M.D. Principal Investigator
Jill Miyamura, Ph.D. Co-Principal Investigator
Christine Reuschel, Project Manager

Slide 2

RATIONALE FOR ADDING CLINICAL DATA

• Adding limited clinical data elements improves quality assessment using administrative data
• Important, cost-effective additions
  • POA
  • Lab values on admission
  • (maybe) vital signs

Slide 3

HAWAII HEALTH INFORMATION CORPORATION and Hawaii Hospitals

Map of Hawaii showing location of of 24 acute care (non-federal hospitals) named on the image.

Slide 4

LARGE ADMINISTRATIVE DATABASES Administrative, or Billing Data

Image of a UB-92 (UB-04) Billing Form

Demographics (age, sex)
Diagnoses & procedures ICD-9-CM, DRG)
POA
Expected payer
Length of stay
Patient disposition
Admission source & type
Admission month
Charges

Slide 5

PREDICTIVE ACCURACY
Adding Laboratory

Line chart that demonstrates the potential value of adding lab data to risk stratification model, in this case, selected post-op complications. Shows improved predictive power as information is added. Can see that line levels out in terms of additional value beyond lab P O A. The additional clinical data collected from medical record had little additional value.

Slide 6

ADMISSION LABS

Category	Lab Tests
Chemistry	Albumin	Creatinine	Sodium
	Alkaline phosphatase	Glucose	Troponin I
	Blood urea nitrogen (BUN)	Gamma glutamyl transferase	SGOT
	Calicum	Potassium	SGPT
	Chloride	Phosphate	Bilirubin
	Creatine kinase-MB	BNP
Blood Gas	pO2	pH	Bicarbonate
	pCO2	Base Excess
Hematology	Hemoglobin	Partial thromboplastin time (PTT)	Platelet Count
			Prothrombin Time (PT)
	Hematocrit	INR	White blood count (WBC)
Microbiology	Blood Culture	Urine Culture	Sputum Culture

Slide 7

Hawaii Hospitals

Map of Hawaii showing location of 16 named medical centers and hospitals. Text says: Build infrastructure for CER, add laboratory data, extend MPI to newborns and all new records.

Slide 8

DATA TRANSFORMATION
HHIC Lab Data Flow

Diagram of four key components of data flow and transformation of HHIC lab data. Component 1, Hospital and lab engagement. Component 2, Quality review and standardization. Component 3, Linking lab records to hospitals. And Component 4, Risk adjustment enhancement.

Slide 9

Key Data Sources - As Conceived

Diagram showing two centralized labs, DSL and CLH, and HHIC. Speaker's notes: As initially conceived and as described in our grant proposal, Hawaii has two key centralized laboratories and serves the majority of hospitals in Hawaii.

Slide 10

Data from two hospitals going into the DSL lab database and data from 17 hospitals going into the CLH lab databases. Speaker's notes: Diagnostic Services Laboratories serves 2 hospitals and Clinical Labs Hawaii serves 17. Our plan was to leverage this unique feature to simplify the number of programs/interfaces that would be built to transmit data to us.

Slide 11

Key Data Sources - As Conceived

Data from two hospitals going into the DSL lab database and data from 17 hospitals going into the CLH lab databases, with arrows pointing from DSL and CLH to the HHIC icon. Speaker's notes: Our plan was to leverage this unique feature to simplify the number of programs/interfaces that would be built to transmit data to us. Rather than 19, we would need only 2.

Slide 12

Key Data Sources - As Conceived

Diagram showing data from 2 hospitals going into the DSL lab database and data from 17 hospitals going into the CLH lab database, with arrows pointing from DSL and CLH to the HHIC icon. Five hospital databases surround the HHIC icon, with arrows pointing from them to HHIC. Speaker's notes: Leaving 5 independent hospital labs. Thus 7 interfaces and providers to work with to receive data.

Slide 13

REALITY!!

Data from 2 hospitals going into the DSL lab database and from 17 hospitals data going into the CLH lab databases, with arrows pointing from DSL and CLH to the HHIC icon. Twelve hospital databases surround the HHIC icon, with arrows pointing from them to HHIC. Speaker's notes: Surprise number one of many: We learned early, when we began interviewing our lab directors, that 7 of the 17 performed their own blood gases. While the centralized labs still provided all other 27 labs for these hospitals, we needed to establish additional interfaces to receive the 5 blood gases.

Slide 14

DATA STANDARDS

Established Data Specifications

Lab Data Transmission — Data Elements
*Sending Facility	Receiving Application
*Account Number	Create Date/Time
*Medical Record Number	Patient Class
*DOB	Hospital Test (Order)
Gender	Hospital Test Result (LOINC)
*SSN	Observation Date/Time
*Patient First Name	Results Report/Status Change
*Patient Last Name	Results Status
*Patient Middle Initial	Observation Value
*Admission Date/Time	Unit of Measure
*Discharge Date/Time	Reference Range
Order Physician First Name	Abnormal Flag
Order Physician Last Name	Observation Result Status
Order Physician Middle Initial	Comments
Physician Identifier
* Variables used for linking Lab data to HHIC Inpatient discharge data

Slide 15

DATA QUALITY/STANDARDIZATION Too much, too little, what is this???

• Too Much
  • ALL lab data received vs. 32 requested labs
• Too Little
  • Limited or Missing Key Demographic Linking Variables
  • Missing Laboratory Data
• What is this?
  • Logical Observation Identifiers Names and Codes(LOINC)
  • Facility ID Standardization

Slide 16

3. DATA LINKING

Data Variable
Sending Facility
Account Number
Medical Record Number
Date of Birth
Gender
Patient First Name (up to first space if multiple names were present)
Patient Last Name
Date of Admission/Lab Observation Date
SSN

Slide 17

LINKING LAB DATA TO HOSPITAL RECORDS

• Preliminary Linking
  • Based on Account Number and Facility Code
• Deterministic/Probabilistic Linking
  • Multiple linking iterations performed to link the inpatient discharge data to one or more lab records based on 9 key variables
  • Linking Iterations "customized" due to variances on key linking variables for multiple hospitals

Slide 18

OVERALL

Inpatient Discharges Linked to One or more Lab Records

30,668,969
Total Lab tests received
CY2008-2011

9,450,293
Excluded
Not one of the 32;
Outside study period
Missing identifiers; test value,
Not IP lab record
Duplicate)

21,218,676
Final count linked to Inpatient

86.3%
Hospital Records linked to one or more lab records

Slide 19

DISCHARGES WITH NO LABS

Reason for Hospitalization	Discharges without Labs	Total Discharges	Percent
Newborns	31,656	66,051	47.9%
Mental Diseases and Disorders	6,808	15,317	44.4%
Diseases and Disorders of the Musculoskeletal System and Connective Tissue	3,982	32,224	12.4%
Pregnancy, Childbirth and the Puerperium	2,303	69,764	3.3%
Diseases and Disorders of the Respiratory System	2,159	37,700	5.7%
Miscellaneous Reasons	14,500	229,700	6.3%
Overall:	61,408	450,756	13.6%

Slide 20

CHALLENGES TO IMPLEMENTATION

• Competing Initiatives
  • Hawaii HIE
  • EMR implementations!!
  • Meaningful Use (friend and foe)
  • ACA signed into law March 23, 2010
  • Value Based Purchasing

Slide 21

TODAY IS A BETTER DAY!

• Technology standard
  • LOINC
  • HL7 interface
  • EMRs
• Technology lowers cost
• Technology increases feasibility of linking clinical with administrative claims

Slide 22

Costs

• Software investment
  • HL7 interface
  • $0 (open source)
  • Commercial solution cost more
• New data feed
  • Error records need resolved
• Hardware - adequate disc space etc.
• Marginal additional cost after setup

Slide 23

MAHALO!

• AHRQ (Grant # 1R01HS019990-01)
• Hospitals
• Clinical Labs Hawaii
• Diagnostic Lab Services
• HHIC Team
  • Christine Reuschel, Project Manager
  • Mike Murata, Data Warehouse Manager
  • Lana Kadooka, Senior Developer
  • Billy Lin, DBA/Business Analyst
  • Jean Kailiawa, Data Manager