Research and Evidence Flashcards
How can Cochrane Reviews help?
Outline how Cochrane systematic reviews summarise the results of relevant clinical trials
*Cochrane Reviews are Systematic with transparent processes that are published in advance as protocols
* They aim to IDENTIFY, ASSESS, SYTHESIZE and APPLY the results of Controlled Clinical Trials addressing a defined question
Forest Plots
Understand how forest plots represent the results of meta-analysis of the trials
- Forest plots visually summarise the results of multiple studies in a meta-analysis.
- Each line = one study.
- The size of the box = weight of the study (e.g. larger sample size = bigger box).
- The horizontal line = confidence interval (CI) for that study’s result.
- The vertical line = the line of no effect (e.g. 1 for ratios like RR/OR, or 0 for mean differences).
- If most studies cluster on one side of the line, the effect is consistent.
- If the diamond does not cross the line of no effect, the pooled result is statistically significant.
- Useful for understanding consistency (heterogeneity), direction, and strength of effect.
Absolute effects
Explain the importance of relative and absolute treatment effects
- If stroke risk goes from 10% to 5%,
- ARR = 5%,
- NNT = 1 ÷ 0.05 = 20 → need to treat 20 people to prevent one stroke.
- If risk goes from 1% to 0.5%, RR is still 50% reduction, but ARR = 0.05% and NNT = 200.
Why it matters: Absolute effects are more relevant for shared decision making, as they show how likely a patient is to personally benefit.
Relative effect
Explain the importance of relative and absolute treatment effects
- Compare the risk in two groups (e.g. treatment vs placebo).
- RR = 0.5 means a 50% reduction in risk relative to control.
- Often used in trials and meta-analyses because they remain constant across populations.
Cate plot
Describe how a Cates plot can be used to illustrate the absolute effects and be used in joint decision making with patients or carers
🧩 What is it?
* A Cates plot (or smiley face diagram) shows absolute benefit visually.
* Usually 100 icons (e.g. people figures), with:
* ✅ Green = would not have the outcome regardless
* 🚫 Red = would have outcome despite treatment
* 💊 Yellow = benefit from treatment (e.g. avoided outcome)
👥 How it’s used:
* Makes statistical risk more intuitive for patients/carers.
* Aids shared decision making by showing:
* “This many people will benefit”
* “Most won’t see a difference”
* “Some may still have the outcome despite treatment”
Example: For a treatment that prevents MI in 2 out of 100,
* 2 yellow (benefit)
* 98 others either unaffected or still have MIs
🗣️ Use in practice:
* Especially useful when patients ask: “What does this mean for me?”
* Helps explain trade-offs, especially in preventative or chronic conditions (e.g. statins, anticoagulants, vaccines).
PICO
Try out the process of constructing an answerable question for yourselves (using the PICO framework)
*Patient characteristics
*Intervention
*Comparison
*Outcome
*Study Design
Describe Forest plots of dichotomous and continuous outcomes
Describe Forest plots of dichotomous and continuous outcomes
Dichotomous Outcomes (e.g., alive vs. dead)
Show effect sizes such as Odds Ratio (OR), Risk Ratio (RR), or Risk Difference (RD).
The x-axis is usually on a logarithmic scale due to the nature of ratios.
Interpreted as:
OR or RR > 1 = increased risk
OR or RR < 1 = reduced risk
OR or RR = 1 = no difference
Continuous Outcomes (e.g., blood pressure, weight)
Show Mean Differences (MD) or Standardized Mean Differences (SMD).
The x-axis is usually a linear scale.
Interpreted as:
MD > 0 = higher average in intervention group
MD < 0 = lower average in intervention group
MD = 0 = no difference
Sensitivyt and specificity
Revise the interpretation of the sensitivity, specificity, positive predictive value and negative predictive values of a screening test (PBL 2)
Sensitivity – “True Positive Rate”
Definition: The proportion of people with the disease who test positive.
Mnemonic: SnNout — Sensitivity high → Negative test rules OUT disease.
* ✅ High sensitivity = few false negatives.
* 📌 Important for screening (e.g. HIV, cancer), where missing a case is risky.
Example: A test with 95% sensitivity detects 95 out of 100 actual disease cases.
Specificity – “True Negative Rate”
Definition: The proportion of people without the disease who test negative.
Mnemonic: SpPin — Specificity high → Positive test rules IN disease.
* ✅ High specificity = few false positives.
* 📌 Important for confirmatory testing, or when false positives have consequences.
Example: A test with 98% specificity correctly identifies 98 out of 100 healthy people as disease-free.
PPV and NPV
Revise the interpretation of the sensitivity, specificity, positive predictive value and negative predictive values of a screening test (PBL 2)
➕ Positive Predictive Value (PPV)
Definition: The proportion of people who test positive and actually have the disease.
* 🔄 Depends heavily on disease prevalence:
* Higher prevalence = higher PPV
* Lower prevalence = more false positives → lower PPV
Example: If PPV is 80%, 80 out of 100 positive results are true positives.
➖ Negative Predictive Value (NPV)
Definition: The proportion of people who test negative and truly do not have the disease.
* 🔄 Also depends on disease prevalence:
* Lower prevalence = higher NPV
* In rare diseases, most people are negative → higher NPV
Example: If NPV is 95%, 95 out of 100 negative results are true negatives.
sensitivity, specificity, positive predictive and negative predictive
Revise the interpretation of the sensitivity, specificity, positive predictive value and negative predictive values of a screening test (PBL 2)
Bayes Theorem
Interpret likelihood ratios for a screening / diagnostic test result and explain their application to diagnostic reasoning using Bayes Theorem
Detection of disease in asymptomatic people = screening/ symptomatic people = diagnosis
* Are a fixed characteristic of a test for given result or level of risk marker(s)
* Are thus independent of disease prevalence (baseline risk or prior risk of disease)
* Provide a clinically meaningful idea of the value of a test
* Are behind all risk scores and risk prediction algorithms
* Are key to accurate diagnostic reasoning
* Allow individualization of risk prediction – probabalistic reasoning
Testing zone
Outline how biological measures that are on a numerical (continuous) scale can be used in screening tests and in diagnostic tests
outline drug development
Outline the process of drug development from pre-clinical studies to post-marketing surveillance (session)
- Discovery
- Medicinal chemistry, biological testing, pharmacology, toxicology
- ~5,000+ compounds → 5-10 candidates
- Pre-clinical studies
- Laboratory and animal testing for toxicity, pharmacokinetics
- Clinical Trials
- Phase I: 50 healthy volunteers – Safety & pharmacokinetics
- Phase II: 200–400 patients – Safety & pharmacology in disease
- Phase III: 1,000–3,000 patients – Efficacy, long-term safety
- Phase IV (Post-marketing): < 10,000 patients – New indications, pharmacovigilance
- Regulatory Review & Licensing
- Application for Marketing Authorisation
- Approved by MHRA (UK) or EMA (EU)
- Post-Marketing Surveillance
- Ongoing safety monitoring via Yellow Card Scheme, Prescription Event Monitoring (PEM), General Practice Research Database (GPRD)
How are medicines licensed?
Explain how medicines are licensed for use in patients (session)
- General Sale Listings eg aspirin, paracetamol, ibuprofen
- Pharmacy medicine OTC eg simvastatin
- PRescription only
- Governed by:
- The Medicines Act 1968
- MHRA (Medicines and Healthcare products Regulatory Agency)
- Medicines must be proven safe, efficacious, and of acceptable quality.
- Must demonstrate superiority to placebo (or standard of care).
- After approval, they receive a Marketing Authorisation (MA), permitting them to be sold and prescribed.
How do the NHS approve new drugs?
Describe the managed entry of new drugs into the NHS including the role of NICE, drugs and therapeutics committees and local formularies and guidelines (session)
What affects prescribing choice?
Identify factors that influence individual prescribing choices (session)
Clinical Patient diagnosis, comorbidities, previous response to treatment
Evidence-based National guidelines (e.g. NICE), RCTs, Cochrane reviews
Cost Cost-effectiveness, formulary restrictions, drug availability
Patient-specific Preferences, adherence, socio-cultural factors
External Media, industry reps, peer influence, DTB, local policies
Governance Clinical audit outcomes, prescribing safety alerts, litigation risks
What are national surveys?
Describe the value of national surveys and audits for measuring the quality of care (session)
National surveys and audits provide comprehensive, large-scale data on healthcare practices, patient outcomes, and system performance.
- Identify trends
- Evaluate the effectiveness of interventions
- Pinpoint areas needing improvement
- They ensure care is:
- Evidence-based
- Patient-centered
- Efficient
Why use national surveys and audits?
Describe the value of national surveys and audits for measuring the quality of care (session)
- National surveys and audits offer benchmarks for comparing performance across regions or institutions.
- They identify best practices and highlight areas where care may fall short.
- Provide a foundation for ongoing monitoring and evaluation to support continuous quality improvement.
- Insights from these surveys help shape healthcare policies, guide resource allocation, and inform healthcare program design aimed at improving patient outcomes.
Qualititative research requires….
Describe the value of national surveys and audits for measuring the quality of care (session)
- qualitative research must use clear documentation of data collection and analysis to ensure conclusions are credible
- Identifying the research problem, designing the study, selecting data collection methods, and applying the analytical framework must all be integrated in a cohesive manner.
- Each phase of the research process should support and build upon the others, maintaining methodological consistency and cohesion throughout
Why use qualitative methods?
Discuss the contribution and value of qualitative methodology to health care research
- Focuses on how and why rather than how many or how often
- Uses non-numerical data to explore concepts, opinions or experiences
Insights from early stages of analysis can influence subsequent data collection - Allows to refocus
- Address emerging themes
- Ensure comprehensive understanding
- Understanding Social Phenomena:
- Rich and Valid Data
- Complement to Quantitative Research: Addresses gaps between population-oriented evidence
In healthcare this may help - explain why interventions may fail
- explore patient care experiences
- capture practitioner reflections
- explore health behaviours
- Identify social determinants
- explain health inequalities
What is methodological consistency?
Discuss the contribution and value of qualitative methodology to health care research
- why? ensure alignment across all research components.
- quantitative research uses standardised protocols and measures
- qualitative research must use clear documentation of data collection and analysis to ensure conclusions are credible
- Identifying the research problem, designing the study, selecting data collection methods, and applying the analytical framework must all be integrated in a cohesive manner.
- Each phase of the research process should support and build upon the others, maintaining methodological consistency and cohesion throughout
Methods of qualitative analysis
Discuss the contribution and value of qualitative methodology to health care research
Familiarisation: Immersing in the data through repeated reading or reviewing.
- Coding: Identifying themes or patterns within the data.
- Categorization: Grouping similar codes into broader categories.
- Interpretation: Developing narratives or theoretical insights from the data.
- Thematic Analysis: Focuses on identifying and analysing patterns of meaning.
- Grounded Theory: Involves generating theory from data through iterative analysis.
- Content Analysis: Systematic coding and categorizing of textual information.
- Utilizes qualitative data analysis software such as NVivo and Atlas.ti, providing practical tools to manage, code, and analyse complex data.*
Evaluation criteria for qualitative research
Communicate kaplan meier
Outline communication challenges when discussing prognosis with patients based on Kaplan-Meier (KM) survival curves
- set the scene
- “marker of microscopic disease might be somewhere else”
- empathic response
- what did we speak about last time? what do you remember?
- “to achieve that you need to put up with XYZ (risks)”
- doctor bias in the data.
- 50 to 65% is not a 15% improvement, but actually a 30% improvement.
- We can recommend an option BUT we need to have a conversation about discussing options
- We can do a shared decision making whilst recommending something
- Should build on the patient wishes.
- PICK A DENOMINATOR AND STICK WITH IT do not jump between 1 in 10 and 10 in 100.
- There is always complications with the way data is represented.
- It is so unreliable to show prognosis on an individual level, CANNOT PROMISE anything.
How to use kaplan meier
Outline how Kaplan-Meier methodology is used to interpret time-to-event data, and consider its application to clinical practice (session)
- What is time-to-event (survival) data?
- Time taken to achieve outcome (event):
- Favourable: Recovery following surgery;
- Adverse: Death following start cancer treatment.
- Time taken to achieve outcome (event):
- What is a Kaplan-Meier (KM) Survival Curve?
- Picture of group experience (time until outcome).
- Group probability (percentage) of outcome vs time.
- Picture of group experience (time until outcome).
- Why are KM Survival Curves important?
- Compare treatment options.
- Estimate prognosis e.g. ten-year survival rates.
- Favourable outcome: typically plotted going up and non favourable (death) plotted downward. Interpret differently
Kaplan meier, how long will it take for me to get back to work?
You cant give an average as there are censored observations, but you can give median (0.5)
Intervention 88 days, vs 208 days.
But how intense are those 88 days? Do people drop out is it more sustainable?
Is this classed as a side effect? Remember to communicate Risk Benefit and patient choice.
Hazard ratio ASSUMES that the return to work is proportional between groups (ignores the plateaus)
Remember probability is called risk often.
For the first 50 days it is equivalent, separates and then comes back together,
