Diagnostic Tests, Mixed Methods, and Dissemination in Health Research

Diagnostic Tests

  • Diagnostic tests are crucial in both everyday life and rigorous research settings. They help in making informed decisions, whether in healthcare or other fields. Properly understanding and applying these tests is essential for accurate assessments and effective interventions.

  • Brief overview of diagnostic tests.

  • Mixed methods research: Combining qualitative and quantitative research approaches to provide a more comprehensive understanding of complex phenomena.

  • Dissemination of research: Sharing research findings to inform practice, policy, and further research.

Introduction to Diagnostic Tests

  • Diagnostic tests in everyday life: These tests are ubiquitous, ranging from simple inquiries to complex medical procedures.

  • Asking "How are you?" as a diagnostic test: This seemingly simple question serves as an initial assessment of a person's well-being.

Pre-test Probability

  • Pre-test probability in general practice: For instance, about 1 in 20 patients may have clinical depression before any specific testing.

  • Impact of the question "How are you?": This question can significantly alter the initial probability based on the response.

Impact of Response on Probability

  • "I feel fine": Probability of depression falls from 5% to 3%. This response suggests a lower likelihood of depression.

  • "I don't feel wonderful": Probability of depression increases from 5% to 15%. This indicates a higher likelihood of depression.

Pre-test vs. Post-test Probability

  • The question "How are you?" is the test: This question serves as an informal diagnostic test.

  • Pre-test: Probability before the question is asked.

  • Post-test: Probability after the answer is received.

Subsequent Questions

  • Post-test probability becomes the pre-test probability for the next question: The updated probability informs the next stage of assessment.

  • Example: "Have you lost interest in doing things?": A follow-up question to further refine the probability assessment.

Impact of Losing Interest

  • If they respond, actually, I have lost interest in doing most of the things that I enjoy doing, Then the probability of them being depressed increases from fifteen percent to twenty-five percent, one in four people.

  • If they still enjoy activities, probability of depression falls from 15% to 12%.

Pre-test Probability and Prevalence

  • Pre-test probability is equivalent to point prevalence (frequency at a given time): The initial probability is the same as the prevalence rate at that moment.

  • 5% of patients have major depression: The baseline prevalence of major depression in a general patient population.

  • Proportion rises to 25% for Maori women: Certain populations may have a higher prevalence of depression.

Prevalence Settings

  • Low prevalence, medium prevalence, or high prevalence: Different settings or populations can have varying levels of disease prevalence, affecting the interpretation of test results.

Mammography (Low Prevalence)

  • Less than 1% of women have breast cancer: The prevalence of breast cancer in the general population is low.

  • Problem: false positives: In low-prevalence settings, false positives are more common.

  • Less than 10% of women with a positive mammogram actually have breast cancer: Most positive mammograms in low-prevalence settings are false positives.

  • Explanation of false positives: The test indicates the presence of cancer when it is actually absent.

Breast Lump (Medium Prevalence)

  • Mammography works well: In medium-prevalence scenarios, mammography is more reliable.

  • 10% to 30% of women with a breast lump test positive: The test is more accurate when there is a palpable lump.

Disseminated Breast Cancer (High Prevalence)

  • Cancer that has spread beyond the breast.

  • Problem: false negatives (test gives reassuring response when it's not okay): In high-prevalence settings, false negatives can occur.

Case Study

  • 15-year-old Pakeha woman with a cold and ear pain.

  • Locum practitioner diagnosed a middle ear infection and prescribed antibiotics.

  • Bruce knew it was unlikely due to the patient's age, socioeconomic status, and medical history.

  • Pre-test probability of a middle ear infection was almost zero.

  • High probability of eustachian tube dysfunction (99.9%).

  • The patient had a pearly white normal eardrum, not an infected ear.

  • Pretest probabilities come from clinical experience and literature.

  • Doesn't have to be precise, a ballpark figure will do.

Risk Communication

  • Using words like "low," "high," or "very high" to communicate risk is ambiguous. Vague terms lead to misunderstandings and inconsistent interpretations of risk.

  • Study from 1980 showed varying interpretations of risk levels. This highlights the subjective nature of qualitative risk descriptors.

  • Moderate risk: varied from 20% to 75%. The perceived meaning of "moderate risk" varied widely among individuals.

  • Patho disease mnemonic (absolute certain): 55% to 100%. Even terms meant to convey certainty had a range of interpretation.

  • High probability: 55% to 95%. The term "high probability" also suffered from significant variability in understanding.

  • Lesson: Be more specific when communicating risk. Precision is essential to avoid misinterpretations and ensure informed decision-making.

  • Focus on absolute risks rather than relative risks. Absolute risks provide a clearer picture of actual impact.

  • Example: doubling the risk of thromboembolism from 1/100,000 to 2/100,000 is meaningless but can be frightening as a relative risk (100% increase). Relative risks can be misleading and cause unnecessary alarm when absolute risks are very low.

Moving from Pre-test to Post-test Probability

  • Need to know the likelihood ratio of the test: This ratio helps to quantify how much a test result changes the probability of a condition.

  • Two types of likelihood ratio:

    • Positive likelihood ratio: Likelihood of having the condition given a positive result.

    • Negative likelihood ratio: Probability of having the outcome given a negative result.

Sensitivity and Specificity

  • Need to know sensitivity and specificity to calculate likelihood ratios.

  • Example: prostate cancer

  • Matrix:

    • Disease Present, Test Positive: True Positive (A)

    • Disease Absent, Test Negative: True Negative (D)

  • Sensitivity = A / (A + C)

  • Specificity = D / (B + D)

  • Positive Likelihood Ratio = Sensitivity / (1 - Specificity)

  • Negative Likelihood Ratio can also be calculated.

  • Positive Predictive Value and Negative Predictive Value.

  • Mnemonic: WE WOE (with and without).

Likelihood Ratios Summary

  • Independent of how common the disease is.

  • Apply at the individual patient level.

  • Apply to outcomes with two or more levels.

  • Diagnosis is never certain.

  • Permit clinicians to quantify the probability of a disease for any individual patient.

  • Intuitive meaning.

  • Compiled for common tests, particularly in secondary and tertiary care.

Receiver Operating Characteristic (ROC) Curves

  • Compare diagnostic tests. ROC curves visually represent the performance of different diagnostic tests.

  • Black line bisects the graph (perfect chance). This line indicates the performance of a test that is no better than random chance.

  • Two tests represented by blue and red lines.

  • The greater the area is under the curve, the more accurate the test. A larger area under the ROC curve signifies higher accuracy.

  • Axes: sensitivity and one minus specificity.

Bias

  • Systematic error: Bias introduces consistent inaccuracies in test results.

Lead Time Bias

  • Early detection creates a longer lead time between early detection and death. Lead time bias can make a screening test appear more effective because it detects the disease earlier, but it doesn't necessarily prolong life.

Length Time Bias

  • Survival time appears to lengthen because screening detects a relative excess of cases with slowly developing disease. Length time bias occurs when screening is more likely to detect slow-progressing cases, giving the illusion of increased survival.

Overdiagnosis Bias

  • Screening increases survival by diagnosing patients with pseudo disease symptoms that mimic the real condition but don't have said condition. Overdiagnosis bias involves diagnosing conditions that would never have caused symptoms or death, leading to unnecessary treatment.

Summary

  • Diagnosis is about pre-test and post-test probabilities.

  • Taking a history and physical exam is to get to a pre-test probability that's high enough to start doing investigations or treating.

Mixed Methods Research

  • Combining qualitative and quantitative research methodologies. This approach integrates the depth of qualitative inquiry with the breadth of quantitative analysis to provide a more holistic understanding.

Qualitative Research Methodology

  • Focuses on details, settings, and context. Qualitative research emphasizes understanding the nuances and specific circumstances of a phenomenon.

  • Interested in insights of people's thoughts. It seeks to explore and interpret the perspectives and experiences of individuals.

  • Applies inductive reasoning. Qualitative research uses inductive reasoning to develop theories and insights from data.

Quantitative Research Methodology

  • Focuses on testing theories and investigating cause and effect. Quantitative research is used to test hypotheses and determine relationships between variables.

  • Allows a broad conclusion to be drawn. It aims to generalize findings to larger populations.

  • Applies deductive reasoning. Quantitative research uses deductive reasoning to test existing theories.

  • Focuses on "when", "how much", and "how many"

Need for qualitative

  • answers "why", "how", and "what"

  • Helps with developing concepts and finding the nuances that actually improve or activate the intervention.

Benefits of Mixed Methods Research

  • Brings the strengths of both methodologies to offset the limitations of each. Mixed methods research leverages the advantages of both qualitative and quantitative approaches.

  • Purposefully designed to collect, analyse, and mix both research methodology. It involves a deliberate and systematic integration of qualitative and quantitative methods.

Types of Mixed Methods Research

  • Single study focuses on both quantitative and qualitative. Some studies integrate both methodologies within a single project.

  • Mixed program has an overarching aim with different projects focusing on either quantitative or qualitative. Other programs use a combination of projects, each focusing on either qualitative or quantitative methods, to achieve a common goal.

Philosophical Assumptions and Inquiry

  • Methodology: What guides us to conduct the mixed-method research. The philosophical framework guides the research process.

  • Method: Collecting the information, analysing, and mixing. Specific techniques for data collection, analysis, and integration are employed.

  • Purpose: To help better understand a research question. The goal is to gain a deeper understanding of the research topic.

  • Purposeful.

  • Applies the rigorous of data collection analysis in both quantitative and qualitative.

Terminology

Mixed Method

  • Mixing both qualitative and quantitative.

Multi Method

  • Collecting, analysing, and mixing multiple forms of qualitative or quantitative data.

Four C's (Reasons to Use Mixed Methods)

  • Comprehensiveness: To give us a better understanding about the phenomenon that we're interested to study. Mixed methods provide a more complete understanding of the research topic.

  • Contextualize: complex health issues. They help in understanding the context surrounding complex health issues.

  • Complement: findings from qualitative and quantitative. They allow for the integration of findings from both qualitative and quantitative data.

  • Philosophical: perspectives. They offer diverse philosophical perspectives on the research question.

Challenges of Mixed Methods Research

  • Maintaining the rigor of each quality and quantitative. Ensuring the quality and validity of both qualitative and quantitative components.

  • Integrating the results, the findings out better. Effectively synthesizing the findings from different methodologies.

  • Sampling.

  • Analytical and interpretive issues.

Purpose of Mixed Methods Research

Triangulation

  • Collect both findings from qualitative and quantitative and combine or compare the findings from both. Triangulation involves comparing and combining findings from qualitative and quantitative data to validate results.

Explanatory

  • Use a connection approach where findings from one phase of the study inform the second phase. In an explanatory design, the results from one phase inform the subsequent phase.

Embedded

  • Gathers more information about how the intervention works to complement and support the findings from one phase of the study. An embedded design involves integrating one type of data within a larger study using another type of data.

Mixed Method Design Metrics

  • Emphasis: Decide if both methodologies have equal emphasis to have a pure mixed method research. Determining the relative importance of qualitative and quantitative components.

  • Time order: Decide if a one phase starts first, followed by the other. Sequencing the qualitative and quantitative phases of the study.

Sampling

Identical Sampling

  • Full sample completes both quantitative and qualitative.

Parallel Sampling

  • Sample from a different group but from the same underlying population.

Nested Sampling

  • Selecting a subset from the sample.

Multilevel Sampling

  • Draw the different sample from different level of the underlying population.

Strategy

  • First is sampling scheme to select the units/individual from sample.

  • Sampling size used determines phase.

Types of Samples

Probability Samples

  • Aligned to Quantitative Research.

Non-Probability Samples

  • Aligned with Qualitative Research.

Summary of Reports

Justify why mixed methods were chosen.

*Specify

  • Purpose.

  • Priority.

  • Methods.

  • Limitations.

  • Insights gained.

Dissemination

  • Sharing work. Dissemination involves communicating research findings to a wider audience.

  • Contributes small knowledge. It adds to the collective understanding of a topic.

  • Helps advance society. Dissemination promotes the use of research in practice and policy to improve societal outcomes.

Components of Dissemination

  • Awareness: Increase understanding of a problem among target population

  • Understanding.

  • Action.

Six T's of Disseminating Health Research

Timing

  • Think right in the beginning. Timing is crucial; dissemination strategies should be planned from the outset of the research project.

Tailor

  • Customized for a particular audience. Tailoring messages to specific audiences increases the likelihood of uptake.

Trusted Source

*To be acknowledged as credible and trustworthy.

*Theoretical Model*

Tactical Modes

  • Strategic.

    Target Audience

  • Influence how said. Identifying the intended audience influences the communication style.

    • Outlet where share. The choice of dissemination channels depends on the target audience.

Consultation

  • Consult with relevant stakeholders. Consulting with stakeholders ensures that dissemination strategies are appropriate and effective.

  • Consult to people down the line and can support you in marketing.

How to Engage End Users and Others Without Delay

  • Share interim findings (not the final findings) to your audiences who might include your key participants. Sharing interim findings keeps stakeholders engaged and informed.

  • Share findings with your key participants for them to learn what came out of the work that they've been doing with them and allow them to stay involved.

Historical Model* unidirectional researcher, talks to the user. \

Sub Models

\

  • Gradually influence seep out and hopefully influence policy.

    Active Model

  • Actively networking with other people to sell the results.

Dialogue Model

  • The researcher talks to the audience and the end users talk back. The dialogue model promotes two-way communication between researchers and end-users.

Tactical Dissemination

Publish the research:

  • reports.

  • papers.

  • journal articles.

  • books, etc.*

Popular media* online newspaper.*

Lower Level

  • newsletter or magazine.

  • leaflet.

    Communicating results.

*