User:Psarka/What a mess!

This is a page to store the notes on the effort to resolve the mess of epidemiology notation and terminology.

Main references

SMM: Statistical methods in medical research (4 ed.) 20K gs citations

DoE: A Dictionary of Epidemiology (6 ed.) 6K gs citations

EBtB: Epidemiology: Beyond the Basics (4 ed.) 2K gs citations

EAI: Epidemiology: An Introduction (2 ed.) 2K gs citations

Affected pages

	shorthand	notation	Template:Medical research studies	synonyms	example	issues
Absolute risk reduction Risk difference	ARR	$I_{u}$, pA, EER	Epidemiology/methods: association: absolute	DoE, EAI, text: (negative) risk difference DoE, text: (dual) absolute risk increase text: (negative) excess risk text: risk difference text: absolute effect	ARR RRR worksheet RCT risk reduction RCT risk increase	first two references are poor and dead should be merged with Excess risk, which is a synonym formula contradicts DoE, SMM, EBtB, EAI (risk difference) Presenting results section should be clarified or removed Add image Make risk difference the main article (in progress) Add statistics
Relative risk reduction	RRR	$I_{u}$, EER	Epidemiology/methods: association: relative	EBtB, text: efficacy EAI, text: relative effect DoE, text: (dual) relative risk increase	ARR RRR worksheet RCT risk reduction RCT risk increase	formula contradicts DoE, and EBtB (efficacy). formula does not contradict EAI (relative effect) references 2 and 3 are dead reference 4 is poor relative risk increase should be mentioned in the text there should be a link to Vaccine efficacy, which is a synonym relative effect should be added as a synonym. Add image
Number needed to treat	NNT	pB, $I_{u}$, EER	Epidemiology/methods: association: absolute	-	ARR RRR worksheetRCT risk reduction	definition and all examples contradict the formula in the worksheet references could be improved Add image
Number needed to harm	NNH	pexposure, $\gamma _{+}$ $I_{u}$, EER	Epidemiology/methods association:absolute	-	custom RCT risk reduction	defined as inverse of attributable risk, while it should be an inverse of absolute risk increase worked example is way too complicated, and involves notions secondary to the topic poor references Add image
Relative risk Risk ratio	RR	pevent when not exposed, a, EER, $I_{u}$,$P(E\mid D)$	Epidemiology/methods: association: relative	text: relative risk	ARR RRR worksheet + custom RCT risk reduction	too many examples Bayesian interpretation is unclear statistical significance and signal vs noise section is too broad statistical ratios section should be removed references to relative probability are poor Add image Make risk ratio the main article
Odds ratio	OR	$N_{E}$, p1, X, n11,	Epidemiology/methods: association: relative		ARR RRR worksheet RCT risk reduction	too much for me, I give up on this one
Attributable risk	-	$I_{u}$	Epidemiology/methods: association: absolute	text: excess risk	-	this term should be depricated, and the page should indicate that, with some historical context.
Attributable fraction among the exposed	AFe	$I_{e}$	Epidemiology/methods: association: absolute	text: Attributable risk percent among exposed text: Attributable fraction among the exposed text: Attributable proportion among the exposed text: Relative attributable risk text: Attributable risk among the exposed	ARR RRR worksheet RCT risk reduction	should be renamed to attributable fraction among the exposed FAR should be clarified or removed add an interpretation section replace arr rrr by risk increase example Add image Add reference to population thing Add reference to preventable fraction
Attributable fraction for the population	AFp	$I_{p}$, $P_{e}$,	Epidemiology/methods: association: absolute	text: Attributable proportion for the population text: Population attributable proportion text: Levin's attributable risk text: Population attributable risk text: Population attributable fraction	-	needs some examples needs references from other effect sizes Add image Add reference to preventable fraction
Preventive fraction Preventable fraction among the unexposed	PFu	$I_{p}$	Epidemiology/methods: association: absolute	-	ARR RRR worksheet RCT risk reduction	preventable fraction for the population is only mentioned in the DoE. odds ratio thing is not mentioned in the main sources. Add to table
Preventable fraction for the population	PFp	$I_{p}$	Epidemiology/methods: association: absolute	-	-
Hazard ratio	HR	$\mathrm {observed\;events\;in\;interval}$, $h(t)$, $r$			-	-
Rate ratio	-	${\text{Incidence Rate 1}}$	-	text: Incidence density ratio	-	could be merged with relative risk
Excess risk	ER	$P(E\mid D)$	-	-	-	could be merged with absolute risk reduction
Experimental event rate	EER
Absolute risk	AR

Things to fix

Notation

Harmonize the notation, if possible staying close to dictionary of epidemiology (DoE):

Notation of the dictionary of the epidemiology:

$I_{e}$ - incidence rate among the exposed

$I_{u}$- incidence rate among the unexposed

$I_{p}$- incidence rate in the population

etc.

Terminology

Remove ambiguous terms and align with the dictionary of epidemiology:

• Attributable risk percent -> Attributable fraction among the exposed

• Deprecate Attributable risk. Quote:

"attributable risk (Syn: causal risk difference) The risk of the outcome had everybody

in the population been exposed minus the risk of the outcome had everybody in the

population remained unexposed. [...] Unfortunately,

this term and attributable rate have been used to denote a number of different

concepts, including the attributable fraction for the population, the attributable

fraction among the exposed, and the rate difference."

Template

Name of the category Epidemiology/methods of the Template:Medical research studies is not ideal, and further subdivisions to "occurrence", "association:absolute", "association:relative", "other" is not adding much clarity.

It needs to be sorted out.

The general terms found in pages are:

• epidemiological measure
• measure
• statistical ratio
• way to quantify

A good candidate from the dictionary of the Epidemiology is:

"effect measure A quantity that measures the effect of a factor on the frequency or risk of

a health outcome or effect. Such measures include the attributable fraction, which

measure the fraction of cases due to a factor; risk differences and rate differences,

which measure the amount a factor adds to the risk or rate of a disease; and risk

ratios, odds ratios, and rate ratios, which measure the amount by which a factor

multiplies the risk, odds, or rate of disease. The identification of these quantities with

effect measures presumes that there is no bias in the quantity."

Worked example

The "ARR RRR worksheet" focuses on control vs target group, while in epidemiology I find exposed vs unexposed more frequently.

Also, target vs control does not allow for discussion about the population.

Also, the positive effect - negative effect distinction is slightly too heavy on the understandability.

Also, some formulas do not match the literature

Random notes

What is this ?!

• Population Impact Measures

Related pages

• Incidence (epidemiology)
• Infection rate (duplicate of the above???)
• Prevalence