Evidence Based Practice into Clinical Practice
Copyright © 2011 Wolters Kluwer Health | Lippincott Williams & Wilkins
Chapter 10
The Role of Outcomes and Quality Improvement in Enhancing and Evaluating Practice Changes
*
Copyright © 2015 Wolters Kluwer • All Rights Reserved
Evidence-Based Quality Improvement (EBQI) and Outcomes Management (OM)
- EBQI: Systematic and continuous actions that lead to improvement in health services and the health status or health outcomes of targeted patient groups (U.S. Department of Health and Human Services, 2011)
- OM: “Technology of patient experience designed to help patients, payers, and providers make rational medical care-related choices based on better insight into the effect of these choices on patient life” (Ellwood, 1988, p. 1549)
*
Copyright © 2015 Wolters Kluwer • All Rights Reserved
Basic Principles Supporting Outcomes Management
- Emphasizing practice standards that providers can use to select interventions
- Measuring patient functional status, well-being, and disease-specific clinical outcomes
- Pooling outcome data on a massive scale
- Analyzing and disseminating outcomes, in relation to the interventions used, to appropriate decision makers and stakeholders (Ellwood, 1988)
*
Copyright © 2015 Wolters Kluwer • All Rights Reserved
Health Outcomes Institute’s Outcomes Management Model
First model to provide steps to guide measurement of the impact of new interventions on improving healthcare outcomes
First two phases of the model:
- Define clinical problem, including structure/process contributors and descriptive and confounding variables; identify desired outcomes and related measures; build database; measure baseline
- Compare appraised evidence with traditional practice; engage stakeholders; negotiate adoption of new practice; develop methods to support new practice; adopt new standard
*
Copyright © 2015 Wolters Kluwer • All Rights Reserved
Health Outcomes Institute’s Outcomes Management Model—(cont.)
Last two phases of the model:
3. Educate all stakeholders; assure that role models and resources are available for troubleshooting processes; monitor reliability and stability of measures and refine as needed; finalize the refined process and measurement methods; begin data collection
4. Close first data collection cycle; analyze results and disseminate to stakeholders; identify opportunities for further improvement (return to phase 2 to begin refinement of improvement)
*
Copyright © 2015 Wolters Kluwer • All Rights Reserved
Question
Is the following statement true or false?
The Health Outcomes Institute’s Outcomes Management Model provides a four-step process for the critical appraisal of evidence.
*
Copyright © 2015 Wolters Kluwer • All Rights Reserved
Answer
False
Rationale: The Health Outcomes Institute’s Outcomes Management Model delineates a process that can be used to guide measurement of the impact of new interventions on improving healthcare outcomes. It does not provide a specific process for critical appraisal of the literature.
*
Copyright © 2015 Wolters Kluwer • All Rights Reserved
Sources of Internal Data for EBQI
| Internal Data Resources | Type/Source of Data |
| Quality management department | Incident reports, patient satisfaction scores, data collected for regulatory or accreditation bodies |
| Finance department | Charges for tests, medications, equipment, or supplies; patient days; readmission rates; patient demographics; patient diagnosis coding (MS-DRG, ICD-9/10) |
| Human resources | Staff turnover and education levels; hours by pay/labor category; contract labor use; provider skill mix; staffing ratios |
*
Copyright © 2015 Wolters Kluwer • All Rights Reserved
Sources of Internal Data for EBQI—(cont.)
| Internal Data Resources | Type/Source of Data |
| Clinical systems | Will vary with system—at minimum typically diagnostic test results and pharmacy data |
| Administration | Patient complaints |
| Electronic health record | Patient-level information captured through documentation of clinical care |
*
Copyright © 2015 Wolters Kluwer • All Rights Reserved
Question
Which of the following two sources would be the most likely to house the data needed to measure patient outcomes related to a proposed change in the nursing skills mix at a hospital?
- Finance and administration
- Human resources and the quality management department
- Clinical systems and the electronic health record
- Administration and the quality management department
*
Copyright © 2015 Wolters Kluwer • All Rights Reserved
Answer
b. Human resources and the quality management department
Rationale: The human resource department is a key source of data related to nursing skills mix and the quality management department collects data on patient outcomes through incident reports, patient satisfaction scores, and data collected for regulatory or accreditation bodies. Administration data are usually limited to patient complaints, and financial data are related to charges, such as for tests, medications, equipment, or supplies. Finally, clinical systems address lab results and pharmacy orders, while the electronic health record is based on clinical documentation, from which aggregate outcomes are hard to derive.
*
Copyright © 2015 Wolters Kluwer • All Rights Reserved
When Existing Data Sources Are Unavailable
Measurement instruments that are developed must be evaluated as to whether they are valid and reliable
- Validity: Is the instrument actually measuring what it is supposed to measure?
- Content validity: The minimum demonstration of validity needed; often reflected through a panel of experts reviewing the instrument
*
Copyright © 2015 Wolters Kluwer • All Rights Reserved
When Existing Data Sources Are Unavailable—(cont.)
Reliability: Does the instrument measure the construct consistently every time it is used?
- Cronbach’s alpha: A Cronbach’s alpha of .80 or greater usually indicates that an instrument should perform reliably each time that it is used
*
Copyright © 2015 Wolters Kluwer • All Rights Reserved
Levels of Data Measurement
| Level of Measurement | Characteristics | Example |
| Nominal | Data sorted into categories; any numbers assigned to categories used only for labeling | Gender, presence or absence of a quality (e.g., disease) |
| Ordinal | Data can be ranked in order, but the absolute difference between each level is not equal | Likert scales |
*
Copyright © 2015 Wolters Kluwer • All Rights Reserved
Levels of Data Measurement—(cont.)
| Level of Measurement | Characteristics | Example |
| Interval | Numeric data with equal and consistent mathematical values separating each discrete measurement point, however, lacks an absolute zero | Fahrenheit temperature scale |
| Ratio | Same data characteristics as interval-level data, but also has an absolute zero value | Kelvin temperature scale |
*
Copyright © 2015 Wolters Kluwer • All Rights Reserved
Reporting to Key Stakeholders
- All parties involved with the process of practice change should have an opportunity to understand the results achieved
- Two methods of presenting data in an understandable way are:
- Scorecards
- Dashboards
*
Copyright © 2015 Wolters Kluwer • All Rights Reserved
Reporting to Key Stakeholders: Scorecards
*
Copyright © 2015 Wolters Kluwer • All Rights Reserved
Reporting to Key Stakeholders: Dashboards
*
Copyright © 2015 Wolters Kluwer • All Rights Reserved
Why Would Quality Projects Need IRB Approval?
- HIPAA regulations require IRB approval of all studies involving personal health information (PHI)
- If it is possible that knowledge might be shared outside of the specific quality improvement initiative and institution (e.g., publication or presentation of strategies used and resultant outcomes), then IRB approval is required prior to initiation of the project
*
Copyright © 2015 Wolters Kluwer • All Rights Reserved
Question
A rating scale asks patients to rate their nausea by describing it as “no nausea,” “slight nausea,” “significant nausea,” or “severe nausea.” What is the highest level of data measurement that this scale provides?
- Nominal
- Ordinal
- Interval
- Ratio
*
Copyright © 2015 Wolters Kluwer • All Rights Reserved
Answer
b. Ordinal
Rationale: In Likert-type scales, data can be ranked in order, but the absolute difference between each level is not equal. It is not possible to calculate a mean or a standard deviation.
*