6: Study Design I

Question 1

What are the levels of evidence-based studies from least to most reliable?

Answer 1

Least:
Editorials, expert opinion
Case series, case reports
Case-control studies
Cohort studies
Randomized controlled trials
Systematic Reviews
Most

Question 2

What must be defined for any study design?

Answer 2

• Defined population (P) for grouping
• Outcomes (O)
• Interventions (I) or exposures (E)

Question 3

What is a bias?

Answer 3

An error introduced into sampling or testing by selecting or favouring one outcome/ answer over others

Question 4

Wheat are the types of bias?

Answer 4

Information bias:
Observational, recall, classification
Selection bias:
Identification, admission rate, non-response

Question 5

Describe cohort studies

Answer 5

2 groups: group of interest and comparison group
Follow both groups over time
Compare outcomes: incidence rates, rate ratio

Question 6

Describe case-control study

Answer 6

2 groups: group of interest and comparison group
Take the histories and compare to draw conclusions

Question 7

Describe cross-sectional studies

Answer 7

Take a study population into 2 groups: controls and cases
Split 2 groups into those exposed and those not exposed
Compare.
Study occurs at a single point in time

Question 8

Describe a clinical (randomized controlled) trial

Answer 8

Randomly assign patients into treatment and control groups
Follow up and compare results (outcome, survival, recurrence)

Question 9

What are the 2 types and subtypes of basic study designs

Answer 9

Descriptive: survey (cross-sectional), qualitative
Analytic: experimental and observational analytic

Question 10

Describe the case-control study methods from the British medical journal in 1950 describing smoking and carcinoma of the lung

Answer 10

• 20 London hospitals with people admitted with clinical diagnosis of lung cancer
• control from simultaneous inpatients same sex and five year age band
• meticulous control for other variables/ confounding factors

Question 11

What are the + and - of case-control studies?

Answer 11

+ quick and cheap
+ good for rare diseases & disorders with delay between exposure & outcome
+ assess multiple exposures
- only one outcome
- prone to recall and selection bias
- no incidence/prevalnce data
- analyse with odds ratio
- not good for rare exposures

Question 12

What is an example of a famous cohort study and what did it study?

Answer 12

Mortality in relation to smoking: 50 years’ observations on male British doctors
- questionnaire for smoking habit, 2/3s response rate (high)
- follow up BMJ/NHS
- cause of death from death certificate (lung cancer, COPD, etc)

Question 13

How do you analyse cohort studies?

Answer 13

• disease incidence (new cases/person time)
• incidence ratio (incidence in exposed / incidence in unexposed)
  Rate ratio, hazard ratio

Question 14

What were the rate atio results from the 1950s biritish doctors study?

Answer 14

Mortality (/1000pyrs):
Never smoke- 0.17
Former smoker- 0.68
Current smoker- 2.49

RR compared to never smoker:
Former smoker- 4.0
Current smoker- 14.6

Question 15

Describe the nurses health cohort study

Answer 15

• 1976, 120,000 nurses aged 30-55 across US
• exploring cancer & cardiovascular disease
• focus on women’s health and diet

Question 16

Describe the validity of the nurses cohort study

Answer 16

• self-reported breast cancer diagnosis questionnaire verified by cross-referencing medical records
• deaths reported by family and cross-referenced with national death index
• exercise by questionnaire & detailed diary
• sub-cohort of survival after a diagnosis of BC dependent on exercise

Question 17

What are the + and - of cohort studies?

Answer 17

+ no recall bias
+ multiple exposures & outcomes can be assessed
+ good for rare exposures if designed appropriately
+ incidence rates
- expensive & time consuming due to continuous follow up
- loss yo follow-up results in selection ias
- not good for rare outcomes