5

Am new to clinical research and starting off by reading some clinical papers.

In the paper, I came across a table like as shown below

enter image description here

Am I right to understand that the value of HbA1c can range from 6.2 to 9.2

Is that the meaning of (1.5)? Similarly for other measurements as well we have items in the bracket

Can help me understand this?

indigochild
  • 253
The Great
  • 3,272

2 Answers2

8

No, your table means that the mean of HbA1c is 7.7 (in %), and that the standard deviation (SD) is 1.5 (again in %). The range will usually be much wider than 6.2-9.2%.

To be honest, the table is a little confusing. If they write "mean $\pm$ SD" in the caption, they should also write "$7.7 \pm 1.5$" in the table. Or conversely, they should write "mean (SD)" in the caption.

Note that sometimes people do include the range, but from what I have seen, the mean and SD is more common.

Stephan Kolassa
  • 123,354
  • 1
    What does the standard deviation tell us for this table, if we don't know the distributions? We don't even know if the individual items all have the same (unknown) distribution. – lyinch Mar 04 '21 at 19:36
  • 6
    @lyinch By presenting data as mean+/-SD, the authors are implying that it's appropriate to approximate each variable expressed in the table with a normal distribution, otherwise they should have presented summary statistics some other way (for example, medians with IQR, or presenting the data in a figure like a histogram if it's crucial). Of course in practice this may not be followed well. – Bryan Krause Mar 04 '21 at 21:52
  • @BryanKrause: in the best of all possible worlds, you would be right. In practice, however, I see people summarizing with mean+/-SD without even looking at the distribution, simply because that's the way everyone does it. So I personally would not draw the inference that the authors considered the normal distribution as a useful approximation. More likely, they just read off mean+/-SD from their SPSS summary tables. – Stephan Kolassa Mar 05 '21 at 07:20
  • @Stephen Agreed. My last sentence was semi-deliberately understated. – Bryan Krause Mar 05 '21 at 13:54
  • Standard deviation or standard error or the mean? My first thought was that it was related to a confidence interval width. – Dave Mar 05 '21 at 16:20
  • @Dave: I usually see either "SD" or "SEM". If the authors used "SD" to refer to the SEM, all hope is lost. In cases like these, I usually interpret this as purely descriptive statistics, i.e., SD, not SEM. – Stephan Kolassa Mar 05 '21 at 16:24
4

The first line in your picture says "mean +/- SD" so it seems reasonable to assume, that the mean HbA1c was 7.7% with a standard deviation of 1.5% and the mean Cholesterol was 4.9mmol/L with a standard deviation of 1.0mmol/L

The standard deviation is the square root of the variance an thus a measure of the amount of variation or dispersion around the mean. A small value implies that most values were close to the mean, a large value means that the spread around the mean was large.

First you compute the difference of each value from the mean and then you square that. Then you compute the mean of the squared differences and take the square root from that. Voilà, that is your standard deviation or SD.

If those values were normally distributed you could draw more conclusions from that but biomarkers are usually not normally distributed. So take it as no more or less then a way to give you an idea, whether values were generally close to the mean or not.

Bernhard
  • 8,427
  • 17
  • 38