Lecture 13: Injury epidemiology - Methodological challenges and study design

Question 1

How is the burden of injury recorded in NZ?

Answer 1

• NZHIS (new zealand health information service, part of the ministry of health)
• ACC claims database
• collected for a different purpose (compensate workers healthcare costs and rehabilitation)

Question 2

what are the pros and cons of the NZHIS?

Answer 2

pros:
- coding is internationally comparable
- subgroup by risk markers (ethnicity, gender, age group)

cons:
- exclude less serious injuries that still impact lives

Question 3

what are the pros and cons of the ACC claims database?

Answer 3

pros:
- captures non-hospitalised injuries

cons:
- major bias towards earners
- treatment claims (96%) little contextual information

Question 4

what is ACC

Answer 4

everyone in NZ is covered by ACC’s no-fault scheme if they’re injured by an accident.

helps stop people from sueing.

Question 5

name a few more data sources for recording the burden of injury?

Answer 5

• agency databases
• national surveys
• injury research publications
• research projects

Question 6

give examples of agency databases and what they record.

Answer 6

• Traffic is recorded by Crash Analysis System (CAS)
• drowning: water safety NZ
• fires: fire service commision
• suicide: ministry of health

Question 7

give an example of a national survey

Answer 7

NZ Health Survey

Question 8

give an example of an injury research publication

Answer 8

Injury research publications

Question 9

give an example of a research projects

Answer 9

using coronial files to investigate fire fatalities

Question 10

what are the things we need to consider about this data collection?

Answer 10

• the purpose for which the data was collected
• how data was collected
• what the denominator is
• potential biases - what data might be missed

Question 11

what can be used as a denominator? what could be the limitations?

Answer 11

• Can use the NZ Census population

Limitations:

• undercount in 2018
• exposure is not constant - may be transient
• total population may not be the best denominator

Question 12

what potential biases could arise when collecting data?

Answer 12

injuries that don’t result in healthcare provision

Question 13

what are the type of epidemiological study designs?

Answer 13

• systematic reviews
• descriptive studies - cross-sectional
• cohort
• case-control
• randomised controlled trial
• ecological

Question 14

what are the challenges of quantifying risk and protective factors?

Answer 14

• exposure to risk is not consistent
• risk factors may be transient or short term
• no or short lag time between exposure and outcome
• multiple contributing causes
• need to consider confounding by other causes
• outcomes are rare in the general public so need large sample sizes
• those affected are often young and mobile which leads to lower participation rates, difficult to follow-up

Question 15

what can a descriptive study do?

Answer 15

describe factors associated with injury, but not whether they are causal

Question 16

what are the pros of injury cohort studies?

Answer 16

• temporal relationship
• exposure to risk information
• can measure many different outcomes
• can use existing datasets with high-risk populations

Question 17

what are the cons of injury cohort studies?

Answer 17

• rare outcomes (most of injury) need very large sample size
• expensive and long follow up
• loss to follow up/ attrition
• difficult to measure short term/transient exposures (exposure might change)
• confounding

Question 18

what are the pros of injury case-control studies?

Answer 18

• many different exposures for single outcome
• good for rare events - efficient and needs a smaller sample
• good for short terms risk factors
• no loss to follow up
• controls representative of exposure

Question 19

what are the cons of injury case-control studies?

Answer 19

• no incidence rates; relative risk is measured by odds ratio
• confounding
• response rates very important due to selection bias (similar to loss to follow up)
• recall bias: misclassification of exposure
• control selection is very important

Question 20

how do controls need to be selected for a case control study?

Answer 20

need to be selected on the basis of transient exposure.

Question 21

how are randomised controlled trials useful for injuries?

Answer 21

• they are good for testing interventions
• but we can’t randomize potentially harmful exposures or interventions (due to equipoise)

Question 22

what are the types of randomised controlled trials?

Answer 22

• individual or group (cluster) randomisation
• community intervention trials

Question 23

what is the difference between a community intervention trial and group randomisation

Answer 23

community intervention is when you design interventions to improve in one community

if you had 3 communities with the intervention, and three communities without it, this would be a ‘cluster’ or ‘group’ randomisation

Question 24

what is an ecological study design?

Answer 24

• recognition that environmental (physical and social) determinants play an important part in the causes of injury
• looks at differences between groups rather than individuals, defined by place and time.
• exposure and outcome measures are at the level of the whole group

Question 25

what are the pros of ecological studies?

Answer 25

pros:
- they are useful for injury prevention policy evaluation

• can use existing data
• low cost
• convenient

Question 26

what are the cons of ecological studies?

Answer 26

• reliability of data
• no direct estimate of rate or injury in exposed, unexposed, unmeasured confounding

Question 27

what is important to consider in terms of cause?

Answer 27

causes of causes!
- environmental or political cause
*see notes on haddon matrix*

Question 28

how has the view of injury changed?

Answer 28

• injury is now considered to be a public health problem to be systematically studied and addressed
• research and evidence shows that most injuries and predictable and therefore preventable - they are not accidents!
• incidence of injury and prevention of injury is closely linked with policy decisions: social, mental health, alcohol. transport