Quick, very silly biostats question

[MD12]

---

Members don't see this ad.

Hi All,

As I continue my board prep, I realized I have a small question about incidence (new cases/those still at risk).

1) I realize that if someone contracts something incurable like HIV (something that can never really be eliminated), they will no longer be included in either the numerator or denominator of incidence, right? (B/c at another time point, they won't be a new case and they are also no longer at risk of getting it...they already have it).

2) So what about diseases where you can get it once, be cured, and get it again...like, say a cold, the flu, or pneumonia? Just because someone has had the disease before, doesn't mean they should be excluded from the denominator at a later time, right?

THANK YOU!

 

[knuckles]

Prevalence is much greater than incidence in chronic diseases.

Prevalence is similar to incidence in acute illnesses.

 

[WashMe]

Hi All,

As I continue my board prep, I realized I have a small question about incidence (new cases/those still at risk).

1) I realize that if someone contracts something incurable like HIV (something that can never really be eliminated), they will no longer be included in either the numerator or denominator of incidence, right? (B/c at another time point, they won't be a new case and they are also no longer at risk of getting it...they already have it).

2) So what about diseases where you can get it once, be cured, and get it again...like, say a cold, the flu, or pneumonia? Just because someone has had the disease before, doesn't mean they should be excluded from the denominator at a later time, right?

THANK YOU!

1) You are correct. As noted by knuckles, this is why incidence ([# of new cases over given time] / [population at risk]) can be low for diseases that have high prevalence (a cross-sectional measure: [# of people currently affected] / [population at risk]).

2) You are correct. They fall into the category of "at risk" and were not "currently affected" at time T=0, so at the end of your study they would be "new cases" if they contracted the cold, flu, etc. in your determined time period.

Prevalence is much greater than incidence in chronic diseases.

Prevalence is similar to incidence in acute illnesses.

Thank you Gunner Training haha 🙂

 

[knuckles]

first aid actually, but solid concept nonetheless

 

[WashMe]

BUMP

I actually have a few issues myself with epidemiology, particularly with regard to prevalence and odds ratio.

About prevalence:
Why is prevalence approximately equal to incidence x disease duration?
... it seems like this is only true if you assume the incidence rate is constant over the course of your study (i.e. if the incidence rate of a 14-day illness is 10 people per day and you consider a time period of 30 days, then 14*10= 140 person prevalence). In the aforementioned example, we expect 10 people at disease-day 1, 10 people at disease day-2, etc. and the "140" makes sense. What if the incidence is cyclical or very punctuated?...

About odds ratio:
On page 52 of FA, odds ratio is discussed as (1) "patients with disease X had higher odds of a history of Y behavior than those without disease X."
Now look at page 54... and at the 2x2 table. The odds ratio is (2) "odds of having disease in exposed group divided by odds of having disease in the unexposed group."

These are completely different things!

In (1), it seems we should do [a/c] / [b/d], because the reference is to the odds of previous exposure in each of the groups.

In (2), as shown on p.54, we do [a/b] / [c/d], because the reference is to the odds of disease in light of exposure!

WTF

Oh, and one more thing, why isn't the case discussed where prevalence is << incidence? This happens in my example at the top, where the point prevalence is 140 but the incidence over the course of my 30 day study is 4200 people! Likewise, if you look at the incidence of common colds from Jan. 2009 to Dec. 2009, it will be much greater than the prevalence of common colds than would have been measured by point prevalence in August 2009...

 

[WashMe]

I just realized that (1) and (2) both reduce to [ad/bc] (...duh), but I feel that FA could really do a better job of clarifying this! Also, I think my other points are valid (hopefully).

BUMP

I actually have a few issues myself with epidemiology, particularly with regard to prevalence and odds ratio.

About prevalence:
Why is prevalence approximately equal to incidence x disease duration?
... it seems like this is only true if you assume the incidence rate is constant over the course of your study (i.e. if the incidence rate of a 14-day illness is 10 people per day and you consider a time period of 30 days, then 14*10= 140 person prevalence). In the aforementioned example, we expect 10 people at disease-day 1, 10 people at disease day-2, etc. and the "140" makes sense. What if the incidence is cyclical or very punctuated?...

About odds ratio:
On page 52 of FA, odds ratio is discussed as (1) "patients with disease X had higher odds of a history of Y behavior than those without disease X."
Now look at page 54... and at the 2x2 table. The odds ratio is (2) "odds of having disease in exposed group divided by odds of having disease in the unexposed group."

These are completely different things!

In (1), it seems we should do [a/c] / [b/d], because the reference is to the odds of previous exposure in each of the groups.

In (2), as shown on p.54, we do [a/b] / [c/d], because the reference is to the odds of disease in light of exposure!

WTF

Oh, and one more thing, why isn't the case discussed where prevalence is << incidence? This happens in my example at the top, where the point prevalence is 140 but the incidence over the course of my 30 day study is 4200 people! Likewise, if you look at the incidence of common colds from Jan. 2009 to Dec. 2009, it will be much greater than the prevalence of common colds than would have been measured by point prevalence in August 2009...