measuring outcomes: a methodological challenge m9205 october 24, 2000
TRANSCRIPT
Measuring Outcomes: a methodological challenge
M9205 October 24, 2000
Columbia University School of Nursing M6920, Fall, 2000
Dummy variable
variable in a regression equation with a finite number of values so that different categories of a nominal variable can be identified.
also called an indicator variable Examples
1=treatment0=no treatment
1=female2=male
1=female2=male
1=residence in West 0=residence is Central-1=residence is East
Columbia University School of Nursing M6920, Fall, 2000
Identification of the time frame
Immediately following the intervention
3-6 months post intervention 5 years Lifetime
Columbia University School of Nursing M6920, Fall, 2000
Selection of objective measures
Use of services Return to work Life expectancy
Columbia University School of Nursing M6920, Fall, 2000
Selection of subjective measures
Provider assessment Quality of life or other patient
reports Observer reports
Columbia University School of Nursing M6920, Fall, 2000
Development of data
Access to multiple sources Developing a cohort Retrospective cohort identification
Columbia University School of Nursing M6920, Fall, 2000
HEDIS as an example of the issues
A set of 75 performance measures for comparison of HMOs
Designed for purchasers of care plans
Columbia University School of Nursing M6920, Fall, 2000
Measures should be
relevant scientifically sound feasible
Columbia University School of Nursing M6920, Fall, 2000
Sample HEDIS Measures
Effectiveness of care• beta blocker
treatments after heart attack
• childhood immunization status
Access to care• availability of
primary care providers
• annual dental visit Cost of Care
• rate trends
Columbia University School of Nursing M6920, Fall, 2000
Measures, cont.
Informed care choices• language/
translation services
Descriptive information• provider
compensation• pediatric mental
health network• family planing
Columbia University School of Nursing M6920, Fall, 2000
A nursing perspective on the challenges*
increasing nursing involvement in quality initiatives
identifying strategies that are effective
changing the clinical framework demonstrating nursing’s
contributions *Jones KR et al 1997 Policy Issues Associated with analyzing outcomes of care Image 29:3 (261-7)
Columbia University School of Nursing M6920, Fall, 2000
ANA Report Card Goals
Test relationships between nurse staffing and specific outcome indicators
Assess the feasibility of capturing information necessary to develop measures with acceptable reliability and validity
Columbia University School of Nursing M6920, Fall, 2000
ANA Methodology
quantify nurse staffing at a sample of hospitals
quantify patient incidents and length of stay at same hospitals
measure relationship between these two sets of variables
Columbia University School of Nursing M6920, Fall, 2000
ANA Data challenges
For two states, had to allocate total numbers of hours worked to cost centers
California proportion of hours per cost center used as base of NY and Mass allocations
Columbia University School of Nursing M6920, Fall, 2000
ANA Data challenges, cont.
Nursing intensity weights used as developed in NY
dimensions include • assessment, • planing, • physical needs, • medical needs, • socioemotional support, • teaching
Columbia University School of Nursing M6920, Fall, 2000
ANA Quality indicators for acute care settings
Outcomes• Patient satisfaction
with nursing care, pain management, patient education
• Patient injury rate• Nosocomial infection
rate
Process• maintenance of skin
integrity• nursing staff
satisfaction Structure
• mix of RN, LPN, unlicensed staff
• total care hours per patient
Columbia University School of Nursing M6920, Fall, 2000
Blegan, Goode, Reed Nurse Staffing
Objective: describe at the level of the nursing care unit the relationships among total hours of care, RN skill mix and adverse outcomes
Methods: corellational and multivariate analyses controlling for patient acuity
Columbia University School of Nursing M6920, Fall, 2000
Measures selected included rates of
medication errors (from incident reports)
patient falls (from incident reports) skin breakdown patient and family complaints infections deaths
Columbia University School of Nursing M6920, Fall, 2000
Nurse staffing variables
hours of care per patient day from all nursing personnel
hours of care provided by RNs
Columbia University School of Nursing M6920, Fall, 2000
Key analysis
“total hours of patient care. . .was associated with higher rates of decubiti, complaints and death. . . .Given the high correlation between acuity and total nursing care hours, the interpretation of these coefficients must be done with care” (p. 49)
Columbia University School of Nursing M6920, Fall, 2000
Key analysis
“Most of the previous research in this area was multiinstitutional and generalizeable but suffered from an accompanying lack of detail. The results of this project are more detailed and specific but less generalizable.” (p. 49)
Columbia University School of Nursing M6920, Fall, 2000
Chen Top 100 Hospitals
Issue: do Medicare patients with acute MI admitted to a “top 100” hospital receive better care or have better outcomes than patients treated in other hospitals?
Columbia University School of Nursing M6920, Fall, 2000
Quality of care measures
use of aspirin during hospitalization
reperfusion therapy in-hospital use of beta blockers
Columbia University School of Nursing M6920, Fall, 2000
Findings
top 100 hospitals more likely to have higher AMI volume, be private for-profit, have on-site facilities for procedures
major differences were in length of stay and cost per AMI admission ($1,014 to $1,855 lower per admission; highly correlated with LOS)
Columbia University School of Nursing M6920, Fall, 2000
Comment
goal of ‘top 100’ was to identify superior financial management, operations and clinical practices.
it may, instead, identify lower LOS and cost for same outcomes
Columbia University School of Nursing M6920, Fall, 2000
Frank et al, Treating Depression
Why care? 1/2 of private insurance for MH is on depression
Defining value and benefits Critical of either traditional cost
effectiveness analysis or supply-demand analysis
Columbia University School of Nursing M6920, Fall, 2000
Frank: Definitions
system effectiveness: sum of all effects produced by health care in a system, including those persons treated by various methods and those not treated at all
system cost: sum of all direct treatment costs
Systems cost effectiveness is the ration of system effects to system costs
Columbia University School of Nursing M6920, Fall, 2000
Data identification
identifying persons and episodes: claims data
treatment effects: clinical research on efficacy and effectiveness
Columbia University School of Nursing M6920, Fall, 2000
Findings
cost for highest expected outcomes was $1,059/case
20% of all spending was in the range expected to equal no treatment.
Columbia University School of Nursing M6920, Fall, 2000
Kelleher
Concurrent, prospective study Consecutive admissions Resident assessment of variables Outcome of interest: length of stay Scales created for the study
Columbia University School of Nursing M6920, Fall, 2000
Kelleher
Regression analysis to identify explanatory power
Severity has been common indicator and does help
adding difficulty increases explanatory power
peak values and fluctuation strongly predictive
Columbia University School of Nursing M6920, Fall, 2000
Borzekowski
Outcome of interest: response to anti-violence PSA’s
Interest, understanding, credibility, effect
Sample size? location? Low but suggestive significance
Columbia University School of Nursing M6920, Fall, 2000
Sochalski et al
Research review Proposal of an agenda Focus on nursing staff patterns