Core Functions of Public Health

Core Functions of Public Health

• Assessment: monitor health status and understand health issues.
• Assessment: includes prevalence, incidence, mortality rates, life expectancy, QALY.
• Assessment: investigate health problems to protect people from health hazards.
• Policy Development: give people information needed to make healthy choices.
• Policy Development: engage the community to identify and solve health problems.
• Policy Development: develop public health policies and plans.
• Assurance: enforce public health laws and regulations.
• Assurance: help people receive health services.
• Assurance: maintain a competent public health workforce.
• Assurance: evaluate and improve programs.
• System Management: contribute to and apply the evidence base of public health.
• System Management: involves clinical and translational research.

Population Pyramids

• Population pyramids show the age and sex breakdown of a population at a point in time.
• A pyramidal shape is indicative of a growing population.
• Stationary pyramid: age and sex percentages are constant, birth rate equals death rate.
• Rapidly expanding pyramid: wider at younger ages, with high birth rate and low life expectancy, indicating fast population growth.
• Contracting pyramid: narrower at younger ages, with low death rate and low birth rate, indicating an older population on average.
• Contracting pyramids are typical in developed countries with high education, easy access to birth control, good healthcare, and few environmental factors.

Measures of Population Health

• Prevalence: proportion of population with a disease or condition at a single point in time which is calculated as number of cases divided by the population at risk.
• Incidence: rate at which new cases of disease occur in a population during a specified time, calculated as the number of new cases divided by the population at risk.
• The population at risk for incidence can change over time due to births, deaths, and movement.
• Incidence is often expressed in person-years, summing the time periods each individual was at risk in the study.
• Crude mortality rate: number of deaths divided by the total population.
• Crude mortality rate underestimates true population health status because death is inevitable.
• Case fatality rate: proportion of individuals with a specified disease who die as a result of the disease, and is calculated as the number of deaths due to the disease divided by the number of individuals with the disease.
• Infant mortality rate: number of infants who die before age 1 year divided by the total number of live births.
• Infant mortality rate is not related to time.
• Infant mortality rate is most reflective of population health because it reflects changes with significant alterations in community health.
• Infant mortality rate is the most sensitive indicator of overall population health.
• Life expectancy: average time a newborn would live if they experienced the exact health of every age group at that exact point in time.
• Life expectancy is not affected by age composition.
• Life expectancy is not an estimate of future experience but an estimate of current community health only.
• Life expectancy can change dramatically with wars or epidemics.
• Quality-Adjusted Life Years (QALY): summary health measure used to compare health systems/interventions.
• Unlike life expectancy, QALY captures both morbidity and mortality, as well as treatment outcomes.
• Healthcare professionals use a weighted system for each year of life, with 0 representing death and 1 representing perfect health.
• QALYs consider disability, physical function, psychological function, social function, symptoms, and impairment.
• QALYs are difficult to interpret at an individual level.

Attributable Risk Measures

• Attributable Risk: Represents the incidence of disease due to exposure and directly reflects probability of outcome.
• Attributable Risk is calculated as Incidence of disease in exposed minus Incidence of disease in unexposed.
• Attributable risk reflects the absolute number of outcomes the population may experience above the baseline (or below baseline).
• Attributable risk is also the "incidence of disease attributable to exposure".
• Proportion Attributable Risk represents the proportion of disease, in exposed individuals, that's due to exposure.
• Proportion attributable risk can be determined using attributable risk divided by the incidence of disease in exposed individuals.
• You can calculate the proportion of attributable risk as: (incidence of disease in exposed - incidence of disease in unexposed) / incidence of disease in exposed
• Proportion attributable risk indicates the "proportion of disease cases in the exposed population that is attributable to the exposure".
• Population Attributable Risk: signifies the proportion of all disease in a population that is due to the exposure.
• It is calculated by dividing attributable risk by incidence of disease in population
• Population attributable risk can be determined using: (incidence of disease in exposed – incidence of disease in unexposed) / (incidence of disease in exposed + incidence of disease in unexposed)
• Population attributable risk notes the "proportion of disease cases in the population that is attributable to the exposure".

Odds Ratio

• Odds ratio is used in studies that compare groups but cannot estimate the prevalence of disease in a population as they can't assume exposed and unexposed groups are proportional to the whole population.
• Odds are the ratio of those with an outcome to those without an outcome
• Odds of disease with exposure = # cases of disease in exposed / # healthy cases in exposed = A/B
• Odds of disease without exposure = # cases of disease in unexposed / # healthy cases in unexposed = C/D
• Odds Ratio expresses measure of outcome between exposed and unexposed groups.
• Odds ratio = odds of disease with exposure / odds of disease without exposure = (A/B) / (C / D) = AD / BC.
• You can calculate the odds of exposure given an outcome and get the same result
• Odds ratios are used for case-control studies because selecting control and experimental groups means you don't know the prevalence of disease in the general population.

Calculating Relative Risk

• Risk = incidence = rate.
• Risk of outcome in exposed group = # disease cases in exposed group / (total # of disease and healthy cases in exposed group) = A / (A + B)
• Risk of outcome in unexposed group = # disease cases in unexposed / (total # disease and healthy cases in unexposed) = C / (C + D)
• Relative risk (Risk ratio) - measures how many times more likely the exposed group is to get an outcome than the unexposed group.
• Relative risk = risk of disease in exposed / risk of disease in unexposed = A / (A + B) // C / (C + D)
• Relative risk doesn't always correlate with attributable risk when disease incidence in unexposed population begins to rise.

Relative Risk Difference

• Relative risk difference: the difference in incidence in the exposed group compared to the unexposed group.
• Often used in comparing effect of intervention (experimental group) to control group.
• Relative risk difference = (experimental event rate – control event rate) / control event rate.
• When + relative risk reduction (rate of harmful outcome decreased) and when - relative risk increase (rate of harmful outcome increased).

Absolute Risk Difference

• Absolute risk- identical to attributable risk.
• Absolute risk = incidence of disease in exposed – incidence of disease in unexposed.
• Absolute risk = experimental event rate – control event rate.
• Absolute risk= A / (A + B) – C / (C + D).
• The absolute increase/decrease in disease based on exposure indicates benefit or harm.
• Can express absolute risk as Number Needed to Treat (NNT), which is the number of patients to treat to prevent one harmful outcome.
• NNT = 1 / absolute risk.

Association vs Causation

• Association: a relationship between exposure and outcome that varies in a certain pattern.
• Association does not imply causation.
• Correlation: line of best fit on a plot of exposure and outcome variables and is the simplest measure of association.
• Correlation has no directionality and can be positive, negative, or none.
• Causation: one exposure/risk factor caused an outcome.
• There are 3 types of evidence that support causation.
• First piece of evidence to support causation is direct evidence from studies: Strength, Temporality, Reversibility
• Temporal evidence needs to have a plausible amount of time between exposure and outcome
• Second piece of evidence to support causation is clear mechanistic evidence linking exposure and outcome.
• Biologic plausibility: basic science research demonstrates pathophysiologic link between exposure and outcome.
• Dose-response relationship: outcome increases with increased exposure.
• Converse of reversibility.
• Parallel evidence: repeated studies consistently demonstrate similar association AND if same exposure indicates similar diseases.

Kaplan-Meier Survival Curve

• Kaplan-Meier survival curve: results from prognosis studies often reported as proportion of patients with certain outcome (death, cancer recurrence).
• Graphs # individuals free of event over time, and either the event is present or absent.
• Characterizes entire follow-up period regardless of # of patients at any point in time.
• The K-M survival curve is problematic because patients are usually lost to follow-up.
• As years go by, you get fewer and fewer patients in the at-risk study group the longer the cohort is followed.
• The most important part of graph = # of at-risk patients at the bottom.
• K-M probability = # of surviving patients / # of “at risk” patients.
• "# of at risk" doesn't include people who have already died or developed the event.
• A log-rank test p value <0.001 indicates significant difference between the two outcomes studied over the duration of the follow up period.

Hazard Ratio

• Hazard Ratio: K-M survival data usually referenced in a hazard ratio, where Hazard = probability estimate of an event at a given time.
• Hazard Ratio = weighted probability of the event over the entire study
• Assesses weighted odds of an event during the entire study given a specific risk factor compared to not having the risk factor.
• You can calculate Hazard ratio as hazard function with risk factor / hazard function without risk factor, or essentially the odds ratio of reaching the endpoint.
• Hazard ratio: assumes proportional hazards where constant risk over time.

Risk of Bias

• Bias distorts the results in one direction or another (above or below the true pop value) while chance creates results both above and below the true pop value.
• The risk of bias in observational studies is greater than randomized controlled trials.
• There are 3 main sources of bias: study groups have different prognosis, measurements, and complete follow-up.
• Study groups have different prognosis in ways other than measured exposure/risk factor due to selection bias or confounding.
• Selection bias: choosing a study sample that is not representative of the population at risk, and over or underestimate certain groups in the study pop vs. pop at risk.
• Eliminate selection bias with randomization.
• Confounding: unbalanced prognostic variable between groups and characteristic of the population studied.
• The ideal study design to avoid confounding is randomized controlled trials b/c you randomize both known and unknown variables.
• Confounding can be statistically corrected if confounding factors are identified.
• In cohort studies there's no way to randomize who's in control vs exposure group.
• Accurate measurement of exposures/outcomes: information bias = recall bias, interviewer bias, surveillance bias.
• Information bias: when the data obtained is better, more accurate, or more representative in one group than another.
• Recall bias: difference in accuracy or completeness of memories between study groups but is especially apparent in retrospective.
• Interviewer bias: different data collection methods results in more aggressive acquisition in one group over another, and especially apparent in face-to-face vs. written surveys.
• Surveillance bias: more frequent follow-up or more diagnostic studies being performed in one study group because of their baseline characteristics, and is apparent in prospective studies.
• Complete and adequate follow-up, and avoid loss to follow-up bias.
• Studies need to be adequately long to gather enough outcomes to make worthwhile comparison.
• Loss to follow-up bias = cases and/or controls that are lost to follow-up have different outcomes than those still in the study.
• Cohort studies: particularly prone b/c usually require long period of f/u for outcomes. * Less than 5% loss = little bias and greater than 20% loss = substantial bias.