- - 1

advertisement
-1-
1.
INTRODUCTION
There is a persistent divide in the U.S. health care system between
what is known to constitute best practice and the care that is actually
delivered. Despite expenditures in excess of 13% of the gross domestic
product (GDP), only half of the U.S. population receives recommended
preventive care and only 70% of the recommended services are delivered
when patients seek care for acute conditions (Schuster, McGlynn et al.
1998).
A variety of approaches, ranging from government regulation to
market competition, have been proposed to improve the quality of health
care. Although no one approach is likely to be sufficient, any quality
improvement effort requires information about how well the health care
system is performing, where performance is deficient, and what drives
better performance.
If government regulation is to improve the quality of health care
through activities such as licensure or the setting of standards, then
policy makers should know what components of the health care system are
most deficient and how the implementation of new policies and standards
would affect the quality of delivered health care.
Similarly, if those
who pay for health care and reimburse providers are to influence the
quality of care for the better, they must have information on how
quality is affected by payment (Coye 2001).
Health care providers also
need performance information for quality improvement efforts in order to
identify processes that can be changed to improve care (Berwick 1989;
Blumenthal and Kilo 1998).
For market mechanisms to improve quality,
health care purchasers, including insurance companies, employers,
government bodies and patients, must have reliable quality of care
information to incorporate into their decisions about the health plans
with which they should contract and the providers from whom they should
seek care (Enthoven 1993; Hibbard and Jewett 1997; Davidson and
Restuccia 1998).
In sum, performance information is a key component to
any quality improvement mechanism.
-2-
Although performance information is necessary for quality
improvement, such information is limited (McGlynn and Brook 2001).
The
performance of the health care system is not routinely monitored.
Moreover, no systems are in place to ensure that best practices are
routinely implemented or to identify when basic standards are not being
met.
Current quality measurement and improvement projects frequently
fail to contribute to generalized knowledge of what changes are needed
to improve quality because they are done in isolation (e.g., in one
state, health plan, or hospital) and findings are not reported
typically.
Assembling accurate and broad-based information on the quality of
care is costly because much of the necessary data reside in paper
medical records, electronic data are not readily linked, and patient
input is not routinely obtained.
There would be enormous value in
finding reliable ways to measure the quality of care in a routine and
affordable manner.
For that reason, this dissertation explores the
potential uses of claims data for quality measurement.
DATA SOURCES AND QUALITY MEASUREMENT
Medical records are frequently considered the gold standard data
source to measure the technical quality1 of health care because they
contain detailed clinical information (Fowles, Fowler et al. 1997;
Steinwachs, Stuart et al. 1998).
Clinical data include diagnoses,
treatments, patient risk factors, and the clinical outcomes of care.
However, high costs and restricted access to medical records limit
widespread quality measurement with medical records.
Claims databases are an alternative data source.
Claims data are
generated as a result of patient encounters with the health care system
and are collected primarily for billing purposes.
Claims files may
contain data on demographics, diagnoses, delivered services, and
dispensed medications.
Although claims data have less clinical
information than medical records, they are widely available in
electronic form and are relatively inexpensive to obtain and use for
___________
1 Technical quality is defined by whether patients receive the
health care services known to improve health outcomes.
-3-
quality measurement (Lohr 1990; Dresser, Feingold et al. 1997;
Steinwachs, Stuart et al. 1998).
Claims data have been used to study
several dimensions of health care performance, including variations in
medical practice across geographic areas, patient access to and
utilization of health services, clinical outcomes, and whether
appropriate services are delivered (Wennberg and Gittelsohn 1973;
Lozano, Connell et al. 1995; Johantgen, Elixhauser et al. 1998; Wilt,
Cowper et al. 1999; Lo Sasso and Freund 2000; Weingart, Iezzoni et al.
2000; Fortney, Borowsky et al. 2002).
Using a combination of data sources, including claims and medical
records data, there are measurement activities that are national in
scope.
For example, the National Committee for Quality Assurance (NCQA)
sponsors and maintains the Health Plan Employer Data and Information Set
(HEDIS).
HEDIS includes 26 standardized measures of technical quality
that are primarily used to compare the performance of HMOs.
Both
medical records and claims data are used to assess HEDIS measures.
Other types of health plans, such as PPOs, are beginning to report their
HEDIS performance.
However, NCQA coordinates the collection and
dissemination of information about HEDIS performance for HMOs only –
there is not a similar coordinating activity for other types of health
plans.
This leaves a significant gap in information about the quality
of health plans because less than 25% of the commercial population is
enrolled in HMOs (Gabel, Levitt et al. 2001).
The Centers for Medicare
and Medicaid Services (CMS) and its contractors, Quality Improvement
Organizations (QIOs)2, have developed standardized measures using
claims, medical record, and survey data for six conditions (Jencks,
Cuerdon et al. 2000; Jencks, Huff et al. 2003).
However, these measures
are primarily used to monitor the quality of care provided to Medicare
beneficiaries.
Although the measurement efforts of NCQA and CMS provide
significant information about quality, they represent just a small
selection of the health care services that are known to improve
outcomes.
Each organization uses fewer than 25 measures of clinical
___________
2 QIOs were formerly known as Peer Review Organizations (PROs).
-4-
quality.
These efforts are also limited to measuring quality within
HMOs and the Medicare system.
Since quality of care information is a
fundamental component to any health care quality improvement effort, it
is important to find ways to routinely monitor and report on the
performance of the entire health care system.
Given the wide
availability and relative cost advantage of claims data, it is
worthwhile to understand whether they can be used to (a) measure a
broader range of technical quality, (b) measure quality in a comparable
fashion across all types of health care delivery systems, and (c)
facilitate measurement at the level of individual physicians.
As
decisions are made about future investments in enhancing existing claims
systems or incorporating clinical information it would be useful to know
which additions would be most beneficial from the perspective of quality
measurement and improvement.
PURPOSE OF THE DISSERTATION
This dissertation will analyze a fundamental set of issues for
quality measurement, including:
(1) What dimensions of technical quality can be measured with claims
data?
(2)
If incremental steps were taken to supplement the information in
claims data, how would the capacity of electronic data for quality
measurement increase?
(3)
How well do quality assessments with claims data agree with those
from medical records and what factors contribute to better or
worse agreement?
This dissertation contains two analyses concerning the strengths and
limitations of quality measurement with claims data.
Analysis 1:
An analysis of the dimensions of quality that can be
evaluated with claims data and the types of information that would
increase the capacity for quality measurement
This analysis reviews more than 550 quality of care indicators to
characterize the dimensions of quality that can be measured with claims
data.
The analysis also reveals the types of information that limit the
-5-
quality measurement capacity of claims data because they are not
typically available from claims files and estimates the potential
enhancement of quality measurement if these types of information were
available.
Analysis 2: An analysis of agreement between claims and medical
records data
This analysis is unique because it uses numerous quality of care
indicators to assess agreement about performance between claims and
medical records.
Agreement about (a) who is eligible for the indicated
care and (b) who receives the care are analyzed for patient-indicator
pairs.
Logistic regression is used to identify the characteristics of
quality of care indicators that are associated with better or worse
agreement.
Performance rates using claims data are compared to the
corresponding results based on medical records data.
ORGANIZATION OF THE DISSERTATION
The next chapter reviews literature on how quality of care is
conceptualized and measured.
Chapter 2 also reviews what is known about
the availability and accuracy of claims data and its current
applications in quality measurement.
The analysis in Chapter 3 reviews
553 quality of care indicators to describe the types of information
required for quality measurement and the dimensions of quality that
could be evaluated with claims data.
The types of data elements that
are currently missing from most claims files, but whose presence could
contribute significantly to quality of care measurement, are also
described in Chapter 3.
Chapter 4 suggests several ways that agreement
between claims and medical records data can be assessed.
A conceptual
framework of what predicts agreement about who is eligible for indicated
care and who receives the care is provided, and logistic regression is
used to analyze the factors associated with agreement.
The analysis in
Chapter 4 concludes with a comparison of the results of quality
measurement using claims and medical records data.
Chapter 5 summarizes
how claims data could be used more effectively for quality measurement
and addresses the applications of health information beyond quality
measurement.
Download