As rich administrative and clinical databases are created and expanded, and computing resources become increasingly powerful and inexpensive, both the desire and feasibility of linking data sources is gaining traction as a tool for research and policy making. Commonly linked data sources include laboratory records, pharmacy records, vital records, and emergency medical services records.

There are challenges involved in linking data from different sources, particularly that of disparate system-specific identifiers as well as maintaining respondent privacy and confidentiality. Notwithstanding these hurdles, state-of-the-art data linkage strategies—both uniquely developed and adapted from other sources—have been employed by grant participants. Documents developed by the grantees to aid in the process of collecting, managing, and linking the disparate data types are presented below and are organized by the data source that grantees used in their data linkage project.

• Laboratory Data
  
• Pharmacy Data
  
• Emergency Medical Services (EMS) Data
  
• Vital Statistics Data
  
• Data Standards: HL7, LOINC, ASCII

Laboratory test results can enrich administrative data by identifying not only the existence of disease, but also its severity. In addition, associating laboratory results with administrative data can identify patient co-morbidities that would remain unknown by using standard data alone. Not only do laboratory data allow for more nuanced tracking of clinical outcomes, they also provide the ability to stratify the patient population on the basis of their clinical risk. High-risk individuals can then be targeted for disease management interventions. Linking laboratory data to administrative data also provides more accurate risk adjustments and better measures of patient safety events and healthcare outcomes. Minnesota, Hawaii, and New York added laboratory data elements to hospital discharge databases.

Lab Data Acquisition and Transmission (PDF file, 206 KB; HTML)

Presentation that provides an overview of the data transmission process and requirements. (Developed by the Minnesota grantee)

Instructions for Extracting and Submitting Lab Data (PDF file, 832 KB; HTML)

Overview of the data collection process and details of required data format. (Developed by the New York grantee)

Potential Solutions to Lab Data Extraction (PDF file, 194 KB; HTML)

The process of extracting laboratory test information from an organization's database can require considerable cooperation between IT personnel, healthcare leadership, and data analysts. This document carefully describes each data selection step and identifies potential difficulties and solutions to each. (Developed by the New York grantee)

Participating Hospital Questionnaire (PDF file, 340 KB; HTML)

Inquires about content and format of hospital lab and pharmacy data and how the hospital plans to send it to the state data organization. (Developed by the Minnesota grantee)

Lab Tests Desired for AHRQ Project (XLS file, 33 KB; HTML)

Spreadsheet that lists lab test data requested from hospitals participating in the MN project. (Developed by the Minnesota grantee)

Coronary Procedures Study: An Outline for Labs (PDF file, 575 KB; HTML)

Document that provides a quality assurance protocol for setting up data transmission. It identifies naming conventions, folder structures, file formatting options, and extraction and submission protocols. (Developed by the New York grantee)

Lab Data Requirements (PDF file, 14 KB; HTML)

Guides hospitals in understanding, formatting, and applying inclusion criteria for each data element to be reported. (Developed by the Minnesota grantee)

Enhancing Hawaii Hospital Information Content: Data Acquisition (PDF file, 1.5 MB; HTML)

Document that details the grantee’s process for acquiring laboratory data and includes implementation guidelines as well as appendices containing data specifications for different data standards such as HL7 and ASCII. (Developed by the Hawaii grantee)

Enhancing Hawaii Hospital Information Content: High-Level Analysis of Transmitted Laboratory Data (PDF file, 585 KB; HTML)

Document that details the grantee’s experience with standardizing their laboratory data feeds, dealing with missing data, and other data quality issues. (Developed by the Hawaii grantee)

Enhancing Hawaii Hospital Information Content: Linking Lab Data to Inpatient Discharge Data (PDF file, 550 KB; HTML)

Document that details the grantee’s experience linking 32 specific laboratory test results from 2008-2011 with Hawaii Health Information Corporation’s (HHIC) hospital discharge data. (Developed by the Hawaii grantee)

HL7 Integration Engine Vendor Evaluation Template (PDF file, 397 KB; HTML)

This template can be used by organizations when searching for and comparing the qualifications of vendors to implement data infrastructure changes. The organization’s leadership need to ensure that the vendor has a clear understanding of the project; a sound, evidence- and experience-based approach; and the proper personnel and organizational qualifications to conduct the task. (Developed by the Hawaii grantee)

Pharmacy claims data provide an accurate and longitudinal record of a patient’s medication regimen. Successful incorporation of pharmacy data into clinically enhanced administrative claims databases permits clinicians, hospital personnel, and researchers access to detailed information regarding drug therapies, and dosage and timing of pharmacologic agents administered to hospitalized patients. Not only does this allow for tracking a patient’s compliance with their medication regimen, but also supports studies of the comparative effectiveness of different therapies.

The Minnesota grant linked pharmacy records to hospital discharge records.

Data Acquisition and Transmission: Pharmacy Data (PDF file, 113 KB; HTML)

Presentation for the Acute Heart Failure project. (Developed by the Minnesota grantee)

Phase 1 Pharmacy Data Collection (PDF file, 194 KB; HTML)

Overview of the hospital-, patient- and drug-level data to be collected for the Acute Heart Failure project. (Developed by the Minnesota grantee)

Pharmacy Data Specifications and Formats (PDF file, 122 KB; HTML)

Detailed formatting guidelines for transmission of pharmacy data. (Developed by the Minnesota grantee)

Being able to link clinically rich prehospital data from emergency services databases to clinical and administrative data will enable researchers to answer questions about the comparative effectiveness of prehospital and hospital care, service quality, treatments, and the organization of emergency care.

Prehospital and hospital data each have their unique strengths. While hospital datasets generally contain more detailed measures of patient demographics, comorbidities, and other clinical characteristics; prehospital data contain information about the initial onset of symptoms, and whether there were any critical procedures performed on the patient en route to the hospital. Linkage of prehospital and hospital data could help create a robust database allowing clinicians, other hospital personnel, and/or researchers to track a patient’s trajectory from the advent of the medical emergency to post-discharge from the hospital.

Linking prehospital data to clinical data can also help emergency services personnel track whether the emergency care rendered in the field resulted in poor outcomes post-hospitalization. Having access to this information allows for a comprehensive review of the initial care rendered in the field, and for EMS training, administrative, and quality management staff to analyze their procedures and make the necessary improvements.

Linking EMS records to hospital admissions is a key feature of the project under the New Jersey grant.

Report from the Rutgers Center for State Health Policy

Vital statistics data, including infant and adult birth and death data, can provide valuable insights into population health when linked to clinical data. Vital statistics are the primary data on mortality but do not provide any information on morbidity, health behaviors, injuries, or healthcare usage which is available from clinical data. Linkages to clinical data can help in follow-up studies of cohorts or other populations to determine health outcomes.

Some of the ways these linked data can be used are to track at-risk infants, guide immunization efforts, and provide feedback on institutions or regions with high rates of birth defects. Birth data can be linked to the maternal discharge record, enabling healthcare providers to work toward improving maternal outcomes. Linked data can be used as a measure of patient outcomes based on hospital mortality rates, and also allow for assessment of readmission since failure to account for out-of-hospital deaths may underestimate hospital readmission rates.

Vital Records were linked to hospital discharge data as part of the projects under the Florida and Minnesota grants.

Matching Patients across Institutions without Definitive Patient Identifiers (PDF file, 591 KB; HTML)

Discussion of death certificate matching using ZIP codes. (Developed by the Minnesota grantee)

Linking Maternal and Child Health Data to Create a Comprehensive Longitudinal Dataset: The Florida Experience (PDF file, 502 KB; HTML)

Presentation that illustrates issues involved in linking maternal and child health data, provides a comparison of the capabilities of various software data linking products, and discusses a customized SAS macro that is being used for linking. (Developed by the Florida grantee)

Health Level 7 (HL7) International is one of several American National Standards Institute (ANSI)-accredited Standards Developing Organizations operating in the healthcare arena, and is dedicated to developing standards for the exchange of electronic health information. The organization’s goal is to improve the interoperability of software applications used by the healthcare industry.

In the field of health information technology (HIT), Logical Observation Identifiers Names and Codes (LOINC®) provide a universal standard for identification of laboratory observations. Prior to the establishment of LOINC, multiple laboratories used different codes for different test observations and this lack of standardization hindered the development of clinical repositories and research databases. The advent of LOINC has improved the process of exchanging laboratory data.

ASCII (American Standard Code for Information Interchange) is a character-encoding scheme that allows letters, numbers, punctuation, and other characters to be stored in computer files. Many healthcare organizations store and transmit their data in ASCII format.

HL7 Implementation Guide (PDF file, 830 KB; HTML)

Specifications for laboratory observation reporting using HL7. (Developed by the Hawaii grantee)

HL7 Drug Order Examples (PDF file, 163 KB; HTML)

Samples of minimum required HL7 segments for drug order data. (Developed by the Minnesota grantee)

ASCII Implementation Guide (PDF file, 197 KB; HTML)

Specifications for laboratory observation reporting using ASCII file format. (Developed by the Hawaii grantee)

Summary of Required Laboratory Tests and LOINC (PDF file, 257 KB; HTML)

Document that contains a list of all required tests and requests that the user identify the vendor or internal hospital operation responsible for reporting the results of each test. (Developed by the Hawaii grantee)

LOINC Overview and Instructions (PDF file, 102 KB; HTML)

Document that provides an instruction manual for specifying LOINC mapping. The document includes a background description of LOINC, instructions for submission, frequently asked questions, and sample data in a LOINC Code Worksheet. (Developed by the Minnesota grantee)

LOINC Worksheet (XLS file, 50 KB; HTML)

Spreadsheet that lists common lab tests and asks the hospital to provide specimen type, units, normal range, etc. for each test. Instructions provided in LOINC Overview and Instructions. (Developed by the Minnesota grantee)

Logical Observation Identifiers Names and Codes (LOINC)

Additional Tools on Data Standards from the Laboratory Data Toolkit on Enhancing the Clinical Content of Administrative Data