POPLHLTH111 Exam

Epidemiology

The study of disease frequency in a population.

• Over a period of time (vertical arrow)

• at one point in time (horizontal arrow)

Numerator

The number of cases of disease

Denominator

The population

Age standardisation

Make the denominators look similar, eg. maintain similar age range between the two populations.

How to age standardise

1. Calculate age specific death rate (out of 1000 in each group)

2. Multiply age specific death rates to the standard population

3. Add the expected deaths in each group

PECOT

P = participants/population

E = Exposure

C = Comparison

O = Outcomes

T = Time

What are EG and CG?

Exposure group and comparison group. They are denominators for calculating dis-ease occurrence.

What are a and b (or c and d)?

Numerators, used mostly with dis-ease (a and b)

EGO

Exposure Group Occurrence/outcome = a/EG

CGO

Comparison Group Occurrence/outcomes = b/CG

Cohort Study

Measure exposures and follow people up over time to count relevant dis-ease events. Incidence and prevalence.

Incidence

Measures of occurrence/frequency over time (T).

• EGO = a/EG(/T)

• CGO = b/CG/(/T)

Prevalence

Outcomes measured at one point in time, T = 1.

• EGO = (a/EG)/1

• CGO = (b/CG)/1

Clean measure

The number of something

Dirty measure

Depending on when you measure something, factors could change.

Cross Sectional Study

The exposures and outcome is measured at the same point in time, eg. EG and CG, a and b. Associated with prevalence measures.

What does incidence involve?

Counting categorical (yes/no) dis-ease EVENTS

What does prevalence involve?

Counting categorical dis-ease EVENTS or measuring numerical dis-ease STATES

Randomised Control Trial

Like cohort studies, but participants are randomly allocated to EG or CG. Both groups start the same, meaning if something happens to one group, it is caused by the intervention.

Double Blind RCT

Neither participants nor investigators know which intervention was given to which participant.

2 reasons RCT’s can’t always be used

1. sometimes unethical

2. sometimes people won’t stay in their group - might not be reliable/practical.

Individual Participant Cohort Study

Every individual person in P gets sometime measured.

Ecological Cohort Study

A participant is a whole population; populations are allocated to EG and CG based on average levels from surveys in each population.

Risk Ratio (Relative Risk)

EGO/CGO

• RR < 1 = relative risk reduction

• RR > 1 = relative risk increase

• If EGO = CGO, RR = 1

Risk Difference

EGO - CGO

• If EGO = CGO, RD = 0

• Often called absolute risk difference

Why should all decisions be based on RD, not RR?

As RR = EGO/CGO, if the CGO is small, the risk will also be large; if we don’t know what the CGO is then there might not be any risk.

Non-random Error causes

Poor study design, poor study processes, poor study measurement

RAMBOMAN

• Recruitment

• Allocation

• Maintenance

• Blind or

• Objective

• Measurement

• ANalyses

R = Recruitment

• Are the participants a representative sample from a relevant population?

A = Allocation

• How do people get into EG and CG?

  • Measurement or random allocation

• Was allocation to EG and CG accurate?

M(1) = Maintenance

• Did participants remain in their allocated groups (EG or CG)?

• Were many participants lost to follow-up?

Why don’t cross sectional studies have maintenance error?

Everything is done at one point in time.

BOM = Blind or Objective Measurement

• Will the validity of the study results be affected by how well exposures and outcomes were measured?

• Objective - were measurements made objectively (not influenced by personal interpretation) and not subjectively?

• Blind - were participants and researchers ‘blind’?

AN = Analyses

• Were the EG and CG adjusted in analyses?

• Was there confounding?

Confounding

When the exposure is mixed with another factor that is also associated with the outcome, eg. age.

Regression to the mean

Measurements are repeated to produce less extreme outcomes.

Random sampling errors

Bigger sample, closer to the truth, smaller sample, more random error.

What is a 95% confidence interval trying to describe?

How much random error there is in a study.

• “There is about a 95% chance that the true value in a population (from which participants were recruited) lies within the 95% confidence interval)

What happens when the confidence intervals for the risk difference don’t overlap? (When EGO and CGO don’t overlap)

There is a statistically significant difference between EGO and CGO.

What happens when the confidence intervals for the risk difference don’t overlap with RD = 0? (No effect line)

There is a statistically significant difference between EGO and CGO.

What happens if the 95% confidence intervals for relative risk overlap with 1?

Probably no statistically significant difference between EGO and CGO.

• No effect line (EGO = CGO: RR = 1)

What are some similarities in different epidemiological studies?

• Same appearance/structure - PECOT

• N/D/T

• RR = EGO/CGO

• RD = EGO - CGO

• Risk of random and non-random error

What are some differences in epidemiological studies?

• Allocation - randomly/by measurement

• Time measurement; RCTs/cohort studies - over time, cross sectional studies - one point in time

• Individual vs ecological studies

• Both incidence and prevalence in cohort studies but only prevalence in cross sectional studies.

Meta-Analysis

Combining multiple good studies into one as an alternative to doing one big study.

Strengths of Meta-Analyses

• large so low random error

• relatively cheap as based on existing studies

Weaknesses of meta-analyses

Validity depends on quality of the studies and the quality of review literature/assessing if the sources are good.

3 steps in systematic reviews

1. Go through the studies and review the literature

2. Assess whether any of the sources are any good

3. COMBINE results of good studies in a META-ANALYSIS (but only if they are SIMILAR enough)

Strengths of Individual Participant studies RCTs (2)

• Measures both incidence and prevalence

• Random allocation minimises confounding

Weaknesses of individual participant studies RCTs (4)

• difficult to recruit representative populations

• unethical to randomise people to harmful exposures

• maintenance error common in long studies

• usually expensive, frequently too small increasing chance of random error.

Strengths of individual participant cohort studies (4)

• easier to recruit representative populations

• ethical to study harmful exposures (natural environment)

• less expensive than RCTs, frequently large

• measure both incidence/prevalence

Weaknesses of individual participant cohort studies (2)

• Confounding common when allocated by measurement

• Maintenance error common in long studies

Strengths of individual participant cross sectional studies (3)

• easier to recruit representative populations as little effort required by participants, ethical to study harmful exposures.

• maintenance error not an issue

• less expensive and frequently large

Weaknesses of individual participant cross sectional studies (2)

• confounding common when allocated by measurement

• reverse causality: what came first (exposure or outcomes) - measured at the same time so you don’t know.

Strengths of ecological studies (3)

• large, so low random error

• cheap

• if confounding is unlikely, they can provide very valid evidence

Weaknesses of ecological studies (1)

• confounding is common

Criteria of the Bradford Hill Framework (7)

1. Temporality

2. Strength of association

3. Reversibility

4. Biological gradient (dose-response)

5. Biological plausibility of association

6. Consistency of association

7. Specificity of association

Temporality

• First cause, then disease (outcome)

• Essential to establish a causal relation

Strength of association

• the stronger an association, more likely to be causal in absence of known biases (selection, information and confounding)

• Bigger RR indicates more a more likely causality between exposure and outcome.

Reversibility

• under controlled conditions, a change in the exposure results in a change in the outcome.

Biological gradient (dose-response)

• Incremental change in disease rates in conjunction with corresponding changes in exposure

• Changes in outcome and exposure at the same time

Biological plausibility of association

• Does the association make biological sense?

Consistency of association

• Replication of the findings by different investigators, different times, different places, different methods.

• Have multiple other studies shown similar results?

Specificity of association

• A cause leads to a single effect

• An affect has a single cause

• Weak criteria as not common in real life

Rothman’s Causal Pie Model

1. Sufficient cause (each pie)

2. Component cause (each wedge)

3. Necessary cause (a wedge)

Sufficient cause (each pie)

• Minimum set of conditions; without any one of the components, disease would not occur

• not usually a single factor

• A disease may have several sufficient causes (several pies can produce the same disease)

Component cause (each wedge)

• Each factor or slice is a component cause

• A factor that contributes towards disease causation, but is not sufficient to cause disease on its own.

• Component causes ‘interact’ to produce disease

Necessary cause (a wedge)

• For some diseases, a component cause will be a necessary cause

• A factor that must be present for a specific disease to occur.

Causes of the causes (for individuals)

Any event, characteristic or other definable entity that brings about a change for better or worse in health

Downstream determinants

Individual, easier to change. Interventions operate at the micro (proximal) level, including treatment systems and disease management.

Upstream determinants

Government, can be handled instantly, interventions operate at the macro (distal) level, such as government policies and international trade agreements.

Distal determinants

DoH that is either distant in time and/or place from the change in health status, indirectly influence health by acting on the proximal factors.

Proximal determinants

DoH that is proximate/near to the change in health status, direct association with health status

Dahlgren and Whitehead Model

Determinants operate at different scales, no arches are isolated; they can all affect each other.

Dahlgren and Whitehead: Level 1 - the individual

Age, sex, constitutional factors, individual lifestyle factors, often non-modifiable.

Dahlgren and Whitehead: Level 2 - the community

Family/friends, normalised/accepted attitudes and behaviours of people living/working in the local community, social capital

Social capital

The value of social networks that facilitates bonds between similar groups of people; provides an inclusive environment for people from diverse backgrounds.

Dahlgren and Whitehead: Level 3 - the environment

Physical, built, cultural, biological, the ecosystem, political environments.

Current living standards framework

• Captures resources and aspects of life important for wellbeing of individuals, families and communities.

• Captures role of institutions in safeguarding/building our wealth, and facilitating wellbeing of individuals and collectives.

• Captures wealth as a country, including human capability and the natural environment.

Wealth of Aotearoa: the 4 capitals

1. Natural

2. Social

3. Human

4. Financial/physical

Wealth of Aotearoa - natural

All aspects of the natural environment needed to support life and human activity.

Wealth of Aotearoa - social

Norms and values that underpin society

Wealth of Aotearoa - human

People’s skills, knowledge, physical and mental health, things which allow people to fully participate in work, study, recreation and in society more broadly.

Wealth of Aotearoa - financial/physical

Things which make up the country’s physical and financial assets which have a direct role in supporting incomes and material living conditions.

Structure in population health

Social and physical environmental conditions/patterns that influence choices and opportunities available.

Agency in population health

The capacity of an individual to act independently and make free choices.

Current health challenges

Financial pressures/affordability, ongoing recovery from covid, unhealthy lifestyles, barriers to access, quality of care, digital health

Health challenges - NZ

• Expensive to access primary care, secondary provided free

• Daily system failures - systematic injustices, inequities, lack of access

• Workforce pressures/shortages of resources

What can a minister do about health?

Investments, sets direction, monitoring system/organisational performance, setting regulation.

3 Goals - government policy statement

Access, timeliness, quality

Gaming

The way the health system response to targets may include manipulating how those targets are met rather than if the system is working better than how it was before.

Socioeconomic Position (SEP)

The social and economic factors that influence what positions individuals or groups hold within the structure of society.

Measures of SEP - individuals

Education, income, occupation, housing, assets, wealth etc.

Measures of SEP - populations

Area measures (deprivation, access), population measures (income inequality, literacy rates, GDP per capita) etc.

SEP on Dahlgren and Whitehead model - Individual lifestyle factors

You/decisions you make influence your opportunities:

• Education » knowledge

• Income » material goods, our ability to purchase necessary services/things.

• Occupation » status, power, where does your education lead you.

SEP on Dahlgren and Whitehead model - social and community influences

• Parent’s education, occupation, income

• Commonly used to measure SEP in studies of children/adolescents

• Some evidence parents’ SEP is associated with your own SEP.

SEP on Dahlgren and Whitehead model - living and working conditions

• Area based measures of SEP, eg. NZDep

• Geographic classification for health

• Social fragmentation, accessibility

Area-level deprivation

Observable and demonstrable disadvantage relative to the local community or the wider society/nation to which an individual, family or group belongs.

Dimensions of socioeconomic deprivation, SED (9)

Communication, income, employment, qualification, owned home, support, living space, living condition.

SED - communication

People with no access to internet at home

SED - income

• people aged 18-64 receiving a main means tested benefit

• people living in equivilised households with income below an income threshold.