The number of instances of a diesase in a population at a designated time.

Represents past and present events

- Lifetime record (decayed, missing, and filled teeth indicate the lifetime experience of dental caries)

-smoking status

The number of new cases of disease in a defined population within a specific period of time

Represents future events

- cleft palate

cancer

Mortality rate

Case-fatality rate

5 year survival rate

Number of deaths from a disease

divided by Total population at risk

-Reflects severity of the disease for the population

Number of deaths from a disease

divided by the number of cases of the disease

- Reflects the severity of disease for an individual

-Corollary is 5-year survival rate

Independent variables that place an individual or group at risk for a health-related outcome

Variables that may potentially

be predictive of the outcome under study

- These are adjusted/changed overtime

as they can influence comparison data

ex. age, gender, race/ethnicity, education level, poverty status, insurance status

On a 2x2 table, where is exposure

and where is outcome? What is the reason

for creating this table?

Exposure is always on the Y axis ( First row would be ex. rural, second row would be urban)

Outcome is always on the X axis (ex. has or doesn't have disease)

Outcome

     Has disease          Has no disease 

Used for statistical analysis

Exposure causes outcome

so he will draw E ---------> O

and pick what the exposure may be and what the outcome from that exposure may be

ex. exposure could be residence and you could live in a rural or urban area

Outcome could be dental caries history and you could have the disease or not have it depending

where you live

Covariates are a part of this system- factors that impact both the exposure and the outcome. For this example: access to care, diet, water supply

Categorical

-Nominal

-Ordinal

Continuous

-Interval

-Ratio

When things can't be quantified

they are described in categories

Two values

-Referred to as dichotomous

 ex. male, female

More than two categories (limited number)

- referred to as discrete (categorical type when limited)

ex. health status (good, fair, poor)

When variables have quantified

intervals on an infinite scale

ex. Height and weight

Some variables are discrete but have considerable number of possible values

- discrete (continuous type)

thought of as "counts"

ex. number of teeth

classifications (categories) without a specific order

ex. Residence (urban, rural)

blood type (A,B,O,AB)

Characteristics: Categorical, unordered

Statistics: rates, proportions, chi-square

Information: Low

What are the four levels of

measurement in healthcare research?

Classifications (categories) with a specific order

ex. Final grade (A, B, C, F)

ex. Attitudes (agree, agree, neutal, disagree)

Ordinal is discrete (categorical).

Characteristics: Categorical, Ordered

Statistics: Rank order, above plus median

Information: Medium

Nominal and Ordinal

Options for statistical analysis are limited

Ordinal is greater than nominal

Interval and Ratio

Options for statistical anaylsis are increased

Continuous is greater than categorial

Quantifiable data lacking

an absolute (meaningful) zero point

Ex. Temperature

Ex. Calendar year

Characteristics: Quantifiable, no zero

Statistics: N/A

Information: N/A

Quantifiable data with

an absolute (meaningful) zero point

Ex. Weight

Ex. Kelvin Temperature

Ratio is discrete (continuous)

Characteristics: Quantifiable, zero

Statistics: Above plus mean, t-test, ANOVA

Information: High

Ratio

From categorical to continuous to ratio the more you can do statistically. Can't do much with nominal

Ratio is better than interval

You can start with a ratio level

and create categories out of it

 (start with a more sophisticated level and go backwards - but can't do it the other way around

if you only have measured something at the categorical level you cannot convert it into a

more sophisticated variable)

(Best to worst)

Systematic reviews and meta-analyses

Randomized controlled trials (RCTs)

Cohort studies

Case-control studies

Cross-sectional studies

Case reports/case series

(not focusing on the last two)

Expert opinion

Anecdotal evidence

Correlational (ecologic)

Cross-sectional

Case report/case series

Simple and inexpensive

used for creating hypotheses of other studies

usually conducted prior to analytical studies

Analytical Research Designs

Observational

-Cohort (two types)

• retrospective
• prospective

-Case-control

complex and expensive

most useful for hypothesis testing

randomized controlled clinical studies are the best kind of study but some limitations due to ethical concerns

case report - detailed report of an outcome (disease) in a single individual

- since it has never been seen doctors want to report it so their colleagues can learn from it and discuss it

ex. rapidly progressing mass following extraction

case series - detailed report of a disease

in multiple persons

ex. Kaposi's sarcoma in "healthy" men, turned out to be HIV/aids but at the time it had never been seen before

both are least sophisticated type of study bc

the focus is on outcomes

Exposure Info: No

Outcome Info: Yes

Use: early stages of hypothesis generation

Strengths: Easy, quick, inexpensive

Weaknesses: Not amenable to statistical analysis; links between exposures and outcomes are speculative and can't be tested

Exposure and outcome data (already exisiting) used to explore possible associations

-focus on population, not individual

not possible to link exposure and outcome at the individual level

Inferring person-level associations from population data

Seen in correlational studies

Correlational studies:

Exposure and outcome info,uses, strengths and weaknesses

Exposure Info: Yes at pop level

Outcome info: Yes at pop level

Use: Hypothesis generation (early stages)

Strength: Easy, quick, inexpensive

weaknesses: can't link exposure and outcome data at individual level, subject to ecologic fallacy, dependent on availability of data

Most common descriptive study design

Simple, inexpensive, provides info at individual level, allows investigation of multiple exposure-outcome pathways simultaneously

Ex. Health Survey (NHANES)

- need to test if the association exists

limitations - inability to determine whether exposure preceded outcome (temporality)

ex. oral health and periodontitis

Exposure status not manipulated by researcher

fair support for causation

inexpensive

exposure status manipulated by researcher

complex and expensive

strongest support for causation

ethical dilemmas

-Enrollment into study dependent on outcome status

(whether participants have outcome of interest)

-good design for studying rare outcomes

-each group studied for exposure history

-start with outcome status (periodontitis - whether they have it or not) and see the exposure status of the individuals with/without diabetes for each outcome

-enrollment dependent on exposure status

(whether participants have exposure of interest)

-good design for rare exposures

-each group studied for development of outcomes

Start with exposure status and

 see developing outcomes

- following incidence as our outcome so want ppl with newly developed diseases

Case-control studies- easier, faster, and more common

can test influence of multiple exposures on one outcome, inexpensive

Cohort-require following person for a long time bc they have to develop the disease

-Recruitment of "diseased group" and "non-diseased group" may indirectly affect exposure status

-Problem when unsure if the association is due to influence of exposure or influence of recruitment

ex. diabetes and periodontitis

people with disease report more detail

-problem is when association is observed, not sure whether it is due to influence of exposure

 or differences in recall

Disease status is known at beginning of study so interviewer may lead the questioning toward preconceptions

-Problem is that when association is observed, not sure due to influence of exposure or due to way questions were asked

Solutions: use a printed questionnaire

          or use a standard script

What are the two types of cohort studies?

Describe them

prospective - subjects enrolled in present and followed into future

- expensive bc follows ppl overtime

Retrospective- limited by availability of data, everything occurred in the past, limiting factor is having all the info

-enrolled whether they have the exposure or not

*both subject to loss to follow-up bias

Subjects may not complete study

-loss of interest

-moved

-death

or loss to follow-up is associated

with exposure status bias results

ex. exposure to radiation and oral cancer

Strengths: strong support for causation, allows investigation of multiple outcomes

Weaknesses: expensive, subject to loss to follow-up bias

Strengths: inexpensive, allows investigation

of multiple exposures

Weaknesses: selection bias; observation bias; limited support of causation

This type of study is a little better than

cross-sectional studies