need help with your account or subscription? click here to email us (or see the contact page)
join telegramNEW! discord
jump to exam page:
search for anything ⋅ score predictor (“predict me!”)

Retired NBME 21 Answers

nbme21/Block 4/Question#40 (reveal difficulty score)
A study is conducted to assess the normal ...
500 Men from a list of patients scheduled to be examined by a urologist 🔍 / 📺 / 🌳 / 📖
tags:

 Login (or register) to see more


 +19  upvote downvote
submitted by seagull(1933)
get full access to all contentpick a username

Examining patient from a urologist implies Berkson Bias which would skew the population mean of serum urea nitrogen away from the true accurate mean. Then, realize precision is dependent on statistical "Power" which is increased based on the size of the population of the study. (increased precision = increased statistical power). Therefore, an increase in population of a biased group with lead to inaccuracy with high precision.

get full access to all contentpick a username
forerofore  to add up, the urologist himself doesn't add or remove accuracy (since this is a blood test), what decreases the accuracy is the fact that in order to be sent to a urologist you probably are sick in the first place (selection bias), so your urea nitrogen is likely to be altered. +27
sharpscontainer  I thought of precision as more of a function of variance. Variance will decrease with a greater sample size. Had a hard time because I was thinking about those 4 darn targets (wouldn't 500 darts look more spread out than 10? but no, the variance will be better) that have been in my textbooks since 7th grade and for the first time I was asked a question about this concept only to discover that I didn't have it down as well as I assumed. +1
peridot  @sharpscontainer I feel you, I thought the exact same thing. Looked into it a bit and I think it has something to do with the way standard error or standard deviation or something like that is calculated, but I'm still confused and too tired to dig further. Also, wanted to mention that this NBME has a similar question but instead it's about the 95% confidence interval - maybe that'll help you understand the precision thing better since the 95% confidence interval narrows with a larger sample size? So it's kinda tied to precision? +



 +10  upvote downvote
submitted by usmleuser007(464)
get full access to all contentpick a username
1. Example of inaccurate but highly precise 
    a. 500 patients seeing a particular doctor for a particular illness
2. Example of accurate but imprecise
    a. 10 patients undergo a screening at a mall 
3. Both Accurate and precise 
    a. 500 patients (high precision) undergo a screening (high accuracy ~ no bias or systemic error)
get full access to all contentpick a username



 +1  upvote downvote
submitted by drdoom(1206)
get full access to all contentpick a username

Accuracy means the data points are dispersed, but when you take the mean of those points, that mean (“sample mean”) is nearby the population mean (“true mean”). Data points are “more precise” if the dispersion across data points is smaller than some other set of data points (notice how this is a comparison and not an “absolute” statement); precision says nothing about how close the average of the data points are to the “true mean.”

Keep in mind that accuracy and precision are relative descriptors; you can’t say “so-and-so is precise”; no, you can only say “such-and-such is more precise than so-and-so” or “so-and-so is more accurate than such-and-such.” So, in this case, we can infer that NBME considers “men at the urologist” to have BUNs that are closer to each other (more clustered; more precise; less dispersed) than the BUNs of “men at mall.”

Here’s a nice image:
https://medbullets.com/images/precision-vs-accuracy.jpg

get full access to all contentpick a username



 +0  upvote downvote
submitted by cuthbertallg0od(15)
get full access to all contentpick a username

Discussing precision only makes sense if they were to sample "X # patients" multiple times and see how close the different measurements' results were to each other. The actual size of the sample should't affect precision, but rather it should just affect accuracy (which is reduced by the biased population at the urologist). Smh

get full access to all contentpick a username



 +0  upvote downvote
submitted by ali_hassan(11)
get full access to all contentpick a username

How does 500 men with various urological conditions result in a precise estimate? Wouldn't the variety of values due to various degrees of illness reduce precision and cause a wider variety?

Maybe I overthought it

get full access to all contentpick a username



Must-See Comments from nbme21

nosancuck on Absorption atelectasis
assoplasty on Free T4
seagull on GM2
lnsetick on Apocrine
niboonsh on NMDA receptors are blocked by Mg2+ at the ...
drdoom on Deletion of a hydrophobic amino acid ...
hayayah on HCO3− transported in the plasma
mcl on Aortic
notadoctor on Usual interstitial pneumonitis
madojo on Genital herpes
hungrybox on Obstruction of the bile duct
jambo2222 on Lung
hungrybox on Mismatch repair
hungrybox on Hydrochlorothiazide

search for anything NEW!