Z-Scores of Cardiac Structures | Detroit Data

Calculate the z-scores of 21 common 2D and M-Mode echo measurements, related to body surface area. Measurement sites include the mitral valve, left ventricle, aortic valve, aortic arch, pulmonary valve, and pulmonary arteries. Data is from 782 patients evaluated at the Children's Hospital of Michigan.

Site Measured (cm) Mean Range Z-Score


Regression equations for calculation of z scores of cardiac structures in a large cohort of healthy infants, children, and adolescents: an echocardiographic study. Pettersen MD, Du W, Skeens ME, Humes RA. J Am Soc Echocardiogr. 2008 Aug;21(8):922-34.


On Saturday Apr 19 2008, Dan said: version 0.1, April 19 2008
On Tuesday Apr 22 2008, Dan said: version 0.2 constrained the calculations to subjects with BSA < 2.0. See this post.
On Thursday May 01 2008, Rick Jensen said: Very easy and helpful; feels trustworthy. I thought there were ascending aortic values?
On Thursday May 01 2008, Dan said: Curiously, the AAO did not make it into this publication, but they did include aortic measurements like the "distal aortic arch" and "aorta at diaphragm" ... Go figure.
On Sunday May 25 2008, Yael said: Excellent job!! One thing though, It is easier to have the results as you enter the numbers kike un the coronary artery section and not to hit the calculate button. Yael
On Sunday May 25 2008, Lloyd Marks said: Although I can access your site on my Treo, I can't run the application. Any way to create an application that would allow you to do this on a Treo nor other PDA. One idea - Perhaps you could get Epocrates to add this to their medmath functions. Maybe they would even pay you for it.
On Sunday May 25 2008, Dan said: @ Yael: Good point- it is quite cumbersome at this point. I only recently figured out JavaScript and it's "onChange" event... Updating this calculator is now on my "to do" list. @Dr. Marks: You are way ahead of the curve there on your Treo. This site was developed primarily as a tool for our echo lab- as an extension to our report writing software. Considerations for mobile devices never entered the picture.
On Monday May 26 2008, John Plowden said: Great job Dan. Would be nice to have the option of entering the weight in pounds and oz and the length in inches and have the program convert it automatically. Just a thought, probably would not be too hard to do. Thanks again for all of your efforts
On Tuesday May 27 2008, Giacomo Pongiglione said: Well done. Thanks.
On Tuesday May 27 2008, Dan said: Thanks for the kind words. @ Dr. Plowden: We can't afford to lose another Mars Explorer. *grin* I thought the International System was the standard for reporting height and weight units of measure, no?
On Wednesday May 28 2008, Jorge Gomez said: Seams accurate. Very Helpful!!!
On Thursday May 29 2008, Richard W Gratian, M.D. said: We started using your calculator today. It is very convenient but unfortunately, the upper limit appears to be 2.0 m2 which we found out with our third patient. Is there a calculator for patients above 2.0 m2? Thanks. Rick Gratian
On Thursday May 29 2008, Dan said: Dr. Gratian: You raise an interesting point. Shortly after I built this calculator and starting using it myself, I noticed some unbelievable results on patients with BSA > 2.0 m2. I went back to the article and sought the descriptive characteristics of their patient population. These descriptive characteristics were missing from the article. (as of right now, this article is still an "In Press Corrected Proof" so I suppose the information could yet be added). The supplied figures for the article top out at 2.0 m2, so I chose to constrain the calculations at this limit. Thanks for your comments!
On Friday May 30 2008, Dan said: modified the calculator to update the z-scores as the data is entered
On Wednesday Jun 04 2008, bleb said: entering Z-values for ascending aora (after ST junction) would be helpful. e.g. Bic Ao valve, Turner S.
On Thursday Jun 05 2008, giacomo said: Did you ceck the equations and the coefficients provided in your paper? The result of your equation 2 is -0.384 (how can a negative value be the mean?) while it should be 0.68. On the other end, if you enter 0.68 as Mean y into equation 4 the resulting z value is not - 2.48
On Thursday Jun 05 2008, Dan said: Bleb: the authors did not include data for the AAO- I can't answer as to why... Dr. Pongiglione: the results of "equation 2" in the article refers to the natural log-transformed predicted value. To get the value (y) back in the original units, it needs to be back-transformed, i.e. e raised to the power y; the JavaScript equivalent is Math.exp(y) I hope this helps.
On Tuesday Jun 17 2008, kalyani trivedi said: fantastic; now I can be very precise with ease. Thank you very much for putting this together.
On Thursday Jun 19 2008, Volker Petersen said: This ist what I was looking for a long time, thank you
On Friday Jun 20 2008, satish adwani said: Excellent! Keep it up. Very helpful. I always keep it on when I do Echo's in my clinic!
On Wednesday Jul 23 2008, Marcos said:
On Friday Jul 25 2008, Peter said: Thank you very much. Not only I have waited such a long time for an easy way to get these data calculated for each of our patients.
On Thursday Jul 31 2008, MARIO CAZZANIGA said: do you have Z score calculations of magnetic resonance: RVEDV, RVESV, LVEDV, LVESV, RVEF, LVEF ?
On Thursday Jul 31 2008, Dan said: Mario: Sorry, my primary focus has been on reference values for pediatric echocardiography.
On Monday Aug 04 2008, Jan du Plessis said: This is an extremely useful site
On Thursday Aug 07 2008, ayse sarioglu said: This an extremely useful site.Thank you very much for your excelent work.
On Thursday Aug 14 2008, Dr Rezoana said: Would u please help us to find the Z score of above mentioned psarameters
On Sunday Aug 31 2008, Keyhan Sayadpour, MD said: At Children's Medical Center, Tehran, Iran, we routinely use this website. It is great and very helpful.
On Wednesday Sep 03 2008, shanthi sivanadam said: Would appreciate if you could add Z-score/percentile .


DanDan said...

added percentiles

YaelM said...

Thank you Dan for the usefull information. I use the Detroit article. I wanted to ask how do you calculate the ranges?

DanDan said...

The range is calculated as z scores of +/- 1.65 .

This is done is by solving the "known" and "unknown" values of the equation, i.e., to solve the z score equation for the (known) z score (eg "+1.65") to determine the (unknown) measurement.



DanDan said...

Too much spaghetti code!

I have upgraded the underlying code as described here.

usha pratap said...

I have been unable to use the page from browsers in tablets for the last few days. the mobile web app is very fiddly. Please do check

DanDan said...

Usha: you are right- the site is definitely broken on my iPad. Looks like Blogger may have pushed an automatic "upgrade". I will look into it. Thanks for the heads up.

DanDan said...

Usha: can you try it again? I undid an auto-update that "enhanced the mobile experience"...

Jarosław Meyer-Szary said...

Hi! First of all thanks for a great tool and inspiration on how to practically approach "z-publications". Though I got a few concerns.
1) Why normal range is +/-1,65SD?
2) How do you manage Detroit Data above BSA 2.0 - no patients in this group in the study. Do you (a) accept "no-data" status, (b) assume that above that the values are fairly constant (they are not of course). The equations are obviously useless above BSA 2.0 -the results are outstandingly high.
3) The latter takes us to my last question - if you look into the charts in the publication you will notice that for many (most?) measurements the curves go up (sometimes steeply) just before BSA 2.0. Is there an allometric explanation for that phenomenon? I am concerned with i.e. LVDD: for 180cm and 80kg BSA is 2.0 the calculated mean is 5.81cm and upper limit 6.85cm (+1,65SD). Since this could be just an average adult I'd rather expect upper range 5.9cm according to guidelines. Having said that I am unsure where is the validity threshold for the Pettersen's otherwise great dataset.

DanDan said...

Great questions Jaroslaw.

1) I use the z-score of 1.65 as the range of normal because it represents the 95th percentile (as a cumulative representation) as opposed to z-scores of +/-2, which people often say represents 95% of the normal values... I wrote a little bit about that here.

2) The reasonable upper limit on the Detroit data is actually a bit interesting. You very correctly observe that at larger BSA's, the curves all seem to become absurd. Initially, I limited the calculations to BSA's of less than 2.0, because, as you note, that is how the authors chose to tailor their own charts... However, lots of people asked me why I limited it, and that they wanted to calculate z-scores on their population of big kids. Since the original manuscript said NOTHING about their population demographics (no mention of number of boys/girls, range of ages, heights, weights), I thought I could safely remove the guardrails I had 'arbitrarily' imposed, and we could just let the data speak for itself.

3) EXACTLY. The curves have the inflection point that you describe due to the way the authors chose to model their data-- as a 3rd order polynomial equation-- and NOT as an allometric relationship (which I think was injudicious(irresponsible?) on their part). I have also made some observations about this matter here.

Again, great questions.


Jarosław Meyer-Szary said...

Seems like I made a REdiscovery ;) Thanks for the answer and the links with clear explanations and graphs (worth 1000 words).

I'm still unsure how to approach the last problem in a rather safe and "fool-proof" way. I know that there is no ideal solution for that, but I consider to limit normal range in 2 ways:
1) if available -use adult guidelines as the maximum cap no matter BSA (e.c. 5,9cm for LVDD)
2) for the rest of the sites assume for the calculations BSA=2.0 if it is actually higher than that.

Unfortunately if "cap" is used calculation of z-score makes no sense.

This is obviously arbitrary approach with no proof to be right but "sky is the limit" doesn't work for me either. Despite no descriptive demographics were given in the publication, I assume, like you did before, that the author limited the charts to BSA=2.0 for a reason. An having adult guidelines in mind I am considering to put the limit even earlier (for selected sites). Is there a reason to do otherwise?

DanDan said...


Your struggle is also mine :-)

Trying to make sense of the available reference values is not a trivial undertaking.

I an effort to even further clarify some of these issues, I also have another incarnation of parameterz: an 'app' of sorts.

Here is a link to a measurement performed on that site by a random user on an anonymous adult-sized patient: http://app.parameterz.com/result/864001/

If you drill down on that page, you will see this chart, derived from the "Detroit" data:

I think the inflection point is clearly *before* 2.0 m^2, and that one should exercise caution on subjects with BSA > 16/1.7 m^2- this particular example is perfectly poised on that tipping point.

Your point about just falling back to the "adult reference values" on bigger kids may hold just as many mysteries though. Many of the adult norms are not size-adjusted, and if they are, are often done so crudely (i.e., by 'indexing' to BSA; more on that particular problem here)

Hope this helps,


DanDan said...

oops, the above should read " ...with BSA > 1.6/1.7 m^2"

Dr Josue Cauich Segovia said...

Greattings!!! You have a page very useful, practice and important for the better attention for congenital heart disease. But a few days ago the page is not funcion, not realized any calculate of Z score, specially "Z-Scores of Cardiac Structures / Detroit Data. Is there any reason??? Is in other pages??? Can you fix it?? O we have to pay something to continue using this useful tool???? I´m from Mérida, Yucatán and with this measurement I traid to do the best for the childrens here. I thank you for your attention, and I hope we continue in communication.

DanDan said...

Thanks for the feedback.

I made a change over the weekend (added the ability to use a 'comma' as the decimal separator). All you should have to do force a complete page refresh (CTL-F5) to clear your cached version.

Anonymous said...

CTL-F5 doesnt do anything. you broke your pageeeeeeeeeeeee!

DanDan said...

Please try again; There was a "permissions" issue with some of the files.

Mahesh MSA said...
This comment has been removed by a blog administrator.
Anonymous said...

Just wanted to bring a database error to your attention. I had a patient who was 176 cm x 108.1 kg and it said the lower limit normal was 4.68 cm for the transverse arch (ULN was 7.71 cm).



DanDan said...


There is no "database" per se for these calculations.

I cannot reproduce those numbers. Using the ht,wt of 176,108 yields a BSA of 2.23 and a mean,uln of 4.4,5.7 at the transverse arch.

I appreciate the feedback but at this point I am not sure where the issue is.

Anonymous said...

It would be nice if I could just enter the BSA. Thanks.

Dr.M.Thinakar Vel said...

Very very useful to calculate Z scores. Quicker & Easier. We started using for our daily practice. Thanks.