I have specified a negative binomial regression in `R` using `brm()`. I have three continuous explanatory variables and one interaction term based on two continuous variables. `R` has given me incidence rate ratios, which I am interpreting as "For a one-unit increase in X variable, we can expect the number of Y variable to increase by a factor of 1.07." That seems straightforward enough. Here are my two questions:
- Does the presence of an interaction term in my model change how I interpret the main effects?
- What is the best way to interpret an incidence rate ratio for an interaction term? I've not been able to find many resources online.
Hello,
I usually see incidence rate ratios (IRR) calculated by considering two incidence rates expressed as cases on person-time; like in the example below.
I was curious to know, however, if it is possible to use this same approach to compute the IRR using rates per 100,000 population (e.g. homicide per 100,000 population; not person-time). I am looking to compute the IRR by comparing the pre-intervention rate per 100,000 of a sample vs its post-intervention rate per 100,000.
Thanks!
https://preview.redd.it/rly5aop1hjz61.png?width=404&format=png&auto=webp&s=0673337ac949c96541051ebccba291be463a802a
I am comparing how deprivation (in quintiles) affects the likelihood of entering children's social care by looking at rates in quintiles, in two sub-populations.
Area 2 has higher IRR (rate in highest quintile deprivation / rate in lowest quintile deprivation) compared with area 1
...But, Area 1 has a much higher slope coefficient in linear regression compared with Area 2.
In which area is the social gradient effect strongest and why?
EDIT: made up example:
What if the two number series were both perfectly linear
Group 1
20
25
30
35
40
Group 2
30
35
40
45
50
The 5/1 RR in group 1 = 2.0
The 5/1 RR in group 2 = 1.67
Both have the same linear regression slope coefficients
Question is: if reporting on how deprivation affects social care intervention rates, is it correct (in this example) to say that the social gradients are equal, or that the social gradient is stronger in group 1 than in group 2?
UK covid incidence ratio (of positive tests) of fully vaccinated to unvaccinated by age group
Age week 32-35 week 34-37 week 38-41
Under 18 0.40 0.22 0.1
18-29 0.47 0.51 0.75
30-39 0.68 0.81 1.27
40-49 1.27 1.46 2.24
50-59 1.32 1.44 2.03
60-69 1.38 1.53 2.03
70-79 1.31 1.37 2.01
80+ 0.92 1.04 1.34
I don't remember the sales pitch being "you'll be twice as likely to test positive after a while". It's worrying the trend is up and accelerating. Could be the natural immune system is now blinded and unable to identify the mutated virus, or fighting it with wrong or only partial response.
Data:
https://www.gov.uk/government/publications/covid-19-vaccine-surveillance-report
https://www.gov.uk/government/publications/covid-19-vaccine-weekly-surveillance-reports
I've been prescribed it but haven't started yet. Reading up it seems like half the people who take it are nauseous, sometimes for days, sometimes for months.
The funny thing is I had asked my doctor for Amitriptyline because my brother takes it without problems but he said this was newer and had fewer side effects.
Some of the rarest diseases don't seem that rare at all. Menkes syndrome for example has incidence of 1 in 100,000. Given 7.9billion humans on earth, doesn't this imply approximately 79,000 people with the condition? Then you have harlequinn icthycosis at 1 in 300,000 so that implies 26,000 people. And given that you have hundreds other rare diseases, wouldn't that mean there are shit tons of people with rare diseases?
https://covid.cdc.gov/covid-data-tracker/#rates-by-vaccine-status
See the footnote in the methodology.
Asking an honest question here: isn't this methodology misleading, in that it doesn't take into account the number of tests conducted? If Number of tests on vaccinated individuals is higher by a factor of 5, and the 'positive' test rate is the same, then this methodology for incidence rates will show 5x higher rate of infection, correct?
Apologies for any typos, any feedback on where gaps need to be filled in would be much appreciated.
Related Series:
Are psychiatric drugs driving dementia rates?
Efficacy and safety of psychiatric drugs
#Fertility and Birth Rates have been falling or stagnant since the 1970's
A ‘New’ Normal? An Updated Look at Fertility Trends Across the Globe
>Basically, countries with birth rates above 1.6 or 1.7 children per woman experienced fertility declines. Countries with birth rates below that saw more stability or even increases. What we appear to be seeing is a global convergence around fertility rates of 1.6 or 1.7 children. >This is not just a rich-world phenomenon. Birth rates in Mexico are around 1.9 to 2 kids per woman, so below the rate needed to sustain Mexico’s current population levels. Brazil’s birth rate is even lower, at 1.75, similar to Colombia’s at 1.77. Costa Rica is even lower, at 1.66. El Salvador, Argentina, and Venezuela are all just barely “breaking even” demographically. Other countries like Guatemala are higher but falling fast. >Across the Pacific, Sri Lanka, Bangladesh, and India are all around 2 or 2.2 kids per woman, while Malaysia has fallen to about 1.8. Thailand is even lower, at 1.5 kids per woman. Even Muslim countries like Turkey (2), Iran (1.8), and Tunisia (2.1) have near-replacement fertility, with speedy declines still ongoing.
NCHS - Births and General Fertility Rates: United States
I don't know how to save a filter, but you can dimension by year, measure by crude birth rate.
Births: Provisional Data for 2020(PDF)
>The provisional number of births for the United States in 2020 was 3,605,201, down 4% from the number in 2019 (3,747,540) (Tables 1–3 and Figure 1). This is the sixth consecutive year that the number of births has declined after an increase in 2014, down an average of 2% per year, and the lowest number of births since 1979.
#Fertility rates are below replacement rate
[Replacement level fertility and future population growth](https://pubmed.ncbi.
... keep reading on reddit ➡
