Conceptualization, Operationalization, and Measurement

Dr. Ekong,

What are the steps to Research Development?

1. Define a researchable question

2. Develop a conceptual model

   1. Map what is believed to cause the outcome

   2. Defines which variables and pathways are pertinent

3. Conduct a literature review

4. Operationalize the variables

   1. What will be measured and how?

5. Develop a research plan

What are the 2 models used for conceptualizing outcomes?

1. Economic, Clinical, and Humanistic Outcomes (ECHO)

2. Conceptual approach

Economic Outcomes

Assess direct and indirect cost of care

• Depends on perspective: Patient, Payer, Healthcare provider, Societal

• Can be used for decision analysis

• Cost measurement can be challenging

  • Time costs

  • Mortality or morbidity costs

Types of Economic Outcomes

• Direct healthcare costs

  • Cost of medication

  • Cost of healthcare provider services

  • Cost of hospital bills

• Direct non-healthcare costs

  • Cost of childcare to attend medical appointment

  • Cost of family member time for home care

• Indirect costs

  • Costs of lost or impaired ability to work

  • Lost economic productivity due to death

Clinical Outcomes

Assess therapeutic results, such as blood pressure control or death

• Focus on measures with documented validity and reliability whenever possible

• Measures are often distinguished by their direct relevance to a specific health condition

Types of Clinical Outcomes

• HbA1C for diabetes

• Blood pressure

• Reduction in total cholesterol

• Presence of depression (e.g., PHQ-9)

• Emergency department visits (may also add to cost)

• Hospitalization (may also add to cost)

• Reduction in symptoms

• Tumor size or cancer staging

Domains of Clinical Outcomes

• Physical Findings:

  • Blood Pressure: systolic, diastolic, ratio

  • Lung Sounds: rales, wheezes

  • Heart Sounds: S1 and S2, extra sounds S3 & S4, murmurs

  • Skin: erythema, nodules, bullae

• Laboratory Results:

  • Hemoglobin A1C

  • Cholesterol: HDL, LDL, Total Cholesterol

  • Serum creatinine

• Physiological Tests:

  • Exercise treadmill

  • Pulmonary function tests (PFTs)

• Diagnostic Procedures:

  • Ejection fraction

  • Colonoscopy

• Presence of Disease:

  • Medical diagnosis or diagnostic code

• Mortality:

  • Condition-specific mortality

  • All-cause mortality

Humanistic Outcomes

Assess patient’s quality of life and ability to function

• Reflects effects of treatment on patient’s quality of life and ability to function (emotional or physical)

  • e.g., HRQoL and patient satisfaction

• Most measures are generic

  • Comprehensive for various health concepts that apply to different health states

  • Broadly applicable across diseases, treatments, and demographic groups

HRQoL Instruments

• SF-36, SF-12, SF-2

• Child Health Questionnaire (CHQ)

• Audit of Diabetes Dependent Quality of Life (ADDQoL)

• Nottingham Health Profile

• Quality of Well-Being (QWB)

• EuroQol

• Health Utilities Index (HUI)

• Quality of Life Enjoyment and Satisfaction Questionnaire (Q-LES-Q)

Condition-specific measures

• Capture aspects unique to condition

• Audit of Diabetes Dependent Quality of Life (ADDQoL)

• Quality of life after myocardial infarction (QLM)

Functional status measures

Focuses specifically on functioning or rehabilitation

• Sickness Impact Profile (SIP)

• Functional status questionnaire

A simple example of Conceptual Model

Metformin + Adherence = Optimal glycemic control

Physical Functioning

Mobility and independence in physical abilities, self-care activities, and advanced integrated independent living activities

Psychological Well-being

Range of positive and negative emotions; anxiety and depressive symptoms

Social Functioning

Social interactions and interdependence of the individual with the social environment

Pain

Self-reported degrees of physical discomfort

Cognitive Functioning

Range of intellectual ability, specifically memory, reasoning, and orientation

Vitality

Energy, fatigue, sleep, and rest quality

Overall Well-being

Global assessment of contentment and health

Technical Considerations for Conceptualizing Outcomes

1. Conceptual and measurement model

2. Reliability (internal consistency/reproducibility)

3. Validity (content, construct, criterion)

4. Responsiveness or sensitivity to change

5. Interpretability

6. Burden (respondent/administrative)

7. Alternative forms (telephone interviews or focus groups)

8. Language and/or cultural adaptations

Variables: Dependent vs. Independent (for Conceptualizing Outcomes)

• Dependent variable

  • The variable(s) that is studied/measured/manipulated in a research study

• Independent variable

  • The variable that the researcher controls in order to assess its effects on the dependent variable

What are the 6 D’s of generic outcome measures

• Death

• Disease

• Disability

• Discomfort

• Dissatisfaction

• Destitution or dollars expended for health services

Methods to Operationalize an outcome

1. Average score on a scale (validated or unvalidated)

2. Evidence-based timeline to assess constructs or domains

3. Average change (using the scoring scale for validated measures)

4. Percent with scores ≥ XX

   1. e.g., Percent with improvement > 50%

5. Adherence ≥ 80% or ≥ 95% (based on published articles/literature)

PHQ-9 Questionnaire scores

• (5-9): Minimal symptoms → Support, educate to call if worse, return in one month

• (10-14):

  • Minor depression → Support, watchful waiting

  • Dysthymia → Antidepressant or psychotherapy

  • Major depression, mild → Antidepressant or psychotherapy

• (15-19): Major depression, moderately severe → Antidepressant or psychotherapy

• (>20): Major depression, severe → Antidepressant AND psychotherapy (especially if not improved on monotherapy)

Wong-Baker FACES Pain Rating Scale

0 - No hurt

2 - Hurts a little bit

4 - Hurt a little more

6 - Hurts even more

8 - Hurts a whole lot

10 - Hurts worst

Practical Considerations for conceptualizing outcomes

1. Format: telephone, face-to-face, secured remote data upload

2. Use of proxy respondents

3. Cost of administration (collection and entry)

4. Complexity of measurement and scoring

5. Acceptability to patients and clinicians

6. Expected format for presenting results