Have You Talked to Your Tables About the Dangers of Sex Bias?

The problem of bias is well documented in the biosciences. Even since the Health Revitalization Act of 1993, which laid out guidelines intended to ensure more equitable representation of women and minorities in federally funded scientific research, the problem persists. A 2010 study published in The Journal of Women’s Health found that, among 46 clinical studies enrolling both sexes, women comprised on average 37% of the participants, and among 69 studies, 87% did not conduct analyses by race or ethnicity, and 18% did not report differences in the racial makeup of the study sample at all. Examples of this sort abound and, setting aside the pernicious sociohistorical and nuanced biologic reasons for this phenomenon, the resulting reality is that medicine, as applied to women and minorities, is less evidence based because most research is extrapolated from a homogeneous population—white men.

But even as we attempt to resolve these problems—ensuring that guidelines are in place and that they are followed when conducting new research—there is another, more subtle way that these biases creep into the biomedical literature. Even if the study itself was conducted using a diverse population of participants, sometimes the reporting elides this fact.  As a manuscript editor I have encountered this problem more often than one would expect, and the culprit is usually the table.

In this table, as in many tables that I have encountered, “white” and “male” are the default. Women’s bodies and the bodies of racial and ethnic minorities are implied by the number of white male bodies present.

A good rule when presenting data in tables is to make sure that when you are reporting the sex of participants, if you choose to report only 1 sex, choose the sex that constitutes the majority of the sample. When reporting on racial and ethnic differences, be as specific as possible (even if these comprise a small percentage of participants). Who are the “others?”

The current edition of the AMA Manual of Style does not explicitly lay out these precautions, but in chapter 4, section 1, you will notice that every example shown for presenting data in tables follows these guidelines.

This is not merely a problem of “political correctness” or social equity—it is a question of accurate reporting and just plain good science.—Gabriel Dietz

 

 

 

 

Forest Plots: The Basics

When I was recently asked to give a presentation on forest plots at work, I was less than enthused. Figures are my least favorite part of a manuscript to edit because they usually require a lot of tedious work, and determining how to best visually present statistics makes my brain hurt. Forest plots in particular had become the subject of my nightmares leading up to the time of preparation of my presentation after a few experiences with editing unwieldy ones. However, thanks to being subjected to presenting on forest plots, I’ve gained some basic knowledge that I thought I would share.

There are a few types of forest plots, namely those presenting the results of meta-analyses and those presenting subgroup analyses. Here, I will focus on a forest plot for a meta-analysis. In a meta-analysis, a forest plot acts as a visual representation of the results of the individual studies and the overall result of the analysis. It also shows overall effect estimates and study heterogeneity (ie, variation in results in the individual studies). A forest plot for ratio data should include the following data:

  1. The sources included in the meta-analysis, with citations. If the source author or study name is listed more than once, query the author to ensure that the study samples are unique; overlapping samples would lead to inaccurate estimates. Also, remember to renumber the references if you have renumbered them in the body of the article.
  2. The number of events and total number of participants in each group of the study and in the combined studies.
  3. Risk ratio and 95% CI for each study and overall.
  4. Graphed relative risk and 95% CI, with top labels describing what data markers on either side of the null line mean. The squares represent the results of each study and are centered on the point estimate, with the horizontal line in the center representing the 95% CI. The diamond shows the overall meta-analysis estimate, with the center representing the pooled estimate and the horizontal tips indicating the confidence limits.
  5. Log scale for the x axis with a label indicating the measure.
  6. Percentage of weight given to the study. Weights are given when pooled results are presented. Studies with narrower confidence intervals are weighted more heavily.
  7. Heterogeneity and data on overall effect.

(Open image in a new tab to see more detail.)

The caption should indicate the test and model (fixed or random effects) used in the evaluation and may include an explanation of the meaning of the different marker sizes.

If you follow these basic rules, forest plots are a breeze. If you would like an example of a forest plot for a subgroup analysis, let us know in the Comments.—Sara M. Billings

 

 

 

Lies and Statistics

Check out this post from Skeptical Scalpel about uncool tricks with statistical graphs. Editors beware!—Brenda Gregoline, ELS

 

 

Questions From Users of the Manual

Q: How should columns with mixed units of measure indicate the unit of measure?

A: In a table with mixed units throughout, use a table footnote for the most common unit of measure, eg, “Unless otherwise indicated, data are expressed as number (percentage).” and specify in the stub or column head only those units that are different. In a table with mixed units in a single column, use the most common unit in the column head and only provide another unit in the table cell for those entries that have a different unit of measure.

Q: Because of the change from the 9th to the 10th edition in the way number and percentage are handled in running text (see page 832 in the 10th edition), should column headings in tables also be changed to read, for example, “No. of Girls (%)” rather than “No. (%) of Girls”?

A: No. The style “No. (%) of Girls” is still an acceptable table column head as here both “number” and “percentage” apply to “of girls,” whereas in the example on p 832, the percentage is given as more of an aside to the numerator and denominator and hence follows: “Death occurred in 6 of 200 patients (3%).”

Q: What recommendations do you have for the preferred typeface of a punctuation mark that follows copy set in something other than roman type?

A: Some specific recommendations are outlined below:

• If an entire sentence is set in a typeface other than roman (eg, italic, bold), any punctuation in that sentence would take the typeface of the rest of the sentence.

• If part of a sentence is set in a typeface other than roman, even if it’s the end of the sentence, the ending punctuation would be roman.

• For heads, sideheads, entries in a glossary, the punctuation would follow that of the preceding word (so, in Correct and Preferred Usage of Common Words and Phrases, the commas between the word pairs are boldface, like the words).

• For parentheses and brackets, unless the entire sentence is set in a typeface other than roman, the parentheses or brackets are roman (see the example with “[sic]” on p 358).—Cheryl Iverson, MA

 

Questions From Users of the Manual

Q: If there is a column for P values in a table and if a P value “straddles” rows (eg, provides the P value for men vs women), how should this be shown?

A: There are several options, with option 1 being preferred:

1. Center the P value between the items it compares (eg, between the values for men and women) and consider the use of a side brace.

2. If only 2 items are being compared, list the P value on the line giving the overall category (eg, Sex).

3. Use footnotes to indicate the P value for items being compared (eg, use a superscript “a” next to the value for men and the value for women and indicate the P value for this comparison in a footnote labeled “a”).

Q: If some of the confidence intervals given in a table column include negative values, how do you combine the minus sign and the hyphen that would normally be used in such a range in a table?

A: With ranges that include a minus sign, use to to express the range, rather than a hyphen. Carry this style throughout the entire table, even for those values that do not include a minus sign.—Cheryl Iverson, MA

Questions from Users of the Manual

Q: I’ve been searching the 10th edition to see where the list of footnote symbols from the previous edition is given and I cannot find it. Is that because the lowercase alphabet letters are now going to replace these symbols, as mentioned on page 91?

A: Yes, almost right. We have changed our policy on using superscript symbols for table footnotes and are now using superscript lowercase letters. There are more of them and they are not so “odd.” However, we are continuing to use the old “footnote symbols” for bottom-of-the-page footnotes (see p 43). We only show 2 here…the asterisk and the dagger…because it is not likely that more would be needed (this is the only type of bottom-of-page footnotes that we use in our journals), but if you were to require more, the “old” list would still apply.

Q: I haven’t been able to locate in the 10th edition the place where it says that the symbols “greater than” and “less than” should not be used in running text. (It’s at the top of p 256 in the ninth edition.)

A: You are correct. We neglected to include that this time, but the policy is the same. The examples on page 399 illustrate this, but having the specific statement would be good. It’s a bit like the policy we have of reserving the use of the hyphen for ranges to within parentheses and in tables (and, of course, in references, for the page ranges) and not using it in running text (P values are another exception). It all has to do with “elegance.”—Cheryl Iverson, MA

Quiz Bowl: Practice Editing Tables

During the past 15 years I have been teaching classes in medical editing. Every year I hear the same question from my students: “How can we practice our editing skills?” It’s a difficult question to answer because usually editors learn their skills on the job. But what do you do if you’re trying to break into the editorial field or have moved from, say, an editorial assistant position to a manuscript editor position? This month’s quiz, entitled Practice Editing Tables, is a first step in helping editors gain editing practice. I have focused on editing tables in this quiz because this is often one of the most challenging tasks for both novice and seasoned editors to master.

Basically, this month’s style quiz is simply to edit a table. Therefore, we have no sample question for you to try. Instead, here’s a general question about tables for you to answer.

Formal tables in scientific articles conventionally contain 5 major elements. Can you name these 5 major elements? (Use your mouse to highlight the text box.)

title, column headings, stubs (row headings), body (data field) consisting of individual cells (data points), and footnotes

Each of these elements has various style and formatting recommendations that are described in detail in the AMA Manual of Style (§4.1.3).

As we continue to post more AMA Manual of Style quizzes on the website, we will strive to provide editors with an opportunity to practice their skills. If you are interested in more practice with tables, check out the Tables Quiz and the Creating Tables and Figures Quiz at www.amamanualofstyle.com—Laura King, MA, ELS

Questions From Users of the Manual

Q: Would you hyphenate “white coat hypertension”?

A: We would follow the latest edition of Webster’s New Collegiate Dictionary. The 11th edition recommends inclusion of a hyphen: white-coat hypertension.

Q: If 2 footnote symbols appear next to each other in a table, should any punctuation be introduced between them?

A: Yes. As with the policy for citation of a reference citation and a footnote symbol side by side (see page 95 in the print), add a comma. So, you might have superscript a,b; or superscript a,c-e.

Q: I would like to know how to cite your 10th edition in the style recommended by the 10th edition.

A: Glad to oblige:

Iverson C, Christiansen S, Flanagin A, et al. AMA Manual of Style: A Guide for Authors and Editors. 10th ed. New York, NY: Oxford University Press; 2007.

Q: Section 3.10 advises beginning the subtitle of a journal article cited in a reference list with a lowercase letter. Is this true even if the title ends with a question mark?

A: Yes. Here is an example, edited to style:

Mayer AP, Files JA, Ko MG, Blair JE. Do socialized gender differences have a role in mentoring? academic advancement of women in medicine. Mayo Clin Proc. 2008;83(2):204-207.

The same policy would apply if the title were to end with an exclamation point, although those are rare in scholarly article titles!—Cheryl Iverson, MA

Quiz Bowl: Creating Tables and Figures

Have you ever been editing and suddenly experienced déjà-vu? You know, that feeling that you’ve read the material before. And I don’t mean several weeks or even days ago. I mean really recently. Don’t worry. You’re probably not suffering from posttraumatic editing syndrome. Often authors duplicate material by presenting it in both table and text forms. This is a no-no. As the AMA Manual of Style states, “The same data usually should not be duplicated in a table and a figure or in the text” (§4, Visual Presentation of Data, p 81 in print). Of course, some overlap is to be expected, but extensive duplication of data in tables and text is a waste of space and the reader’s time.

This month’s style quiz on creating tables and figures asks the user to create a figure and a table from text. There are only 2 exercises, so I’m not going to give 1 away here. Instead, here’s a bonus exercise for you to try.

Directions: Use the information in the following paragraph to create a table that can replace the text. Refer to section 4.1 of the AMA Manual of Style.

In a multivariable model of communication attributes associated with parental peace of mind (controlled for diagnostic category, time since diagnosis, child’s age, parent’s education, parent’s race/ethnicity, physician-rated prognosis, and degree of discrepancy between parent-rated and physician-rated prognosis; adjusted for clustering by physician), the odds ratios (95% confidence intervals) were as follows: 2.05 (1.14-3.70) for parent recalled receiving more extensive prognostic disclosure, 2.54 (1.11-5.79) for parent rated information received as high quality, and 6.65 (1.47-30.02) for parent had a greater sense of trust in physician.

Now, of course, there are several ways to reformat this sentence into a table, but here’s what we published (it was table 4 in this article). (Click for bigness.)

Figure

If you want more experience reformatting text into tables or figures, take this month’s quiz at www.amamanualofstyle.comLaura King, MA, ELS

Questions From Users of the Manual

Q: Are arabic numerals used for measures of time:  years, months, weeks?

A: I’m assuming you are asking about using numerals vs words.  The short answer is yes; we use arabic numerals for years, months, and weeks.  But if you should also be curious about the use of  arabic vs roman numerals, see section 19.7.5; and for specific nomenclature conventions, see chapter 15.

Q: Do you have a style for citing tweets?

A: Our blog addressed this query on August 23, 2011.  Please take a look at this archived entry.

Q: How do you handle the word continued when it’s used after a title of a table that runs over onto a second page?

A: We don’t address this specifically in the manual, but if you look at one of the longest tables in the manual (the big SI conversion table in chapter 18) you will see that we used “(cont).”  Since then, however, in our own publications, we have switched to spelling the word out (“continued”) to better serve international readers (who may not recognize cont as a “familiar” abbreviation).

Q: If there is a “compound” acronym/abbreviation defined first in a manuscript (eg, chronic myeloid leukemia in chronic phase [CML-CP]) and, later in the same manuscript, just CML is required, should CML be redefined or did the first definition cover it?

A: Good question.  AMA Manual of Style authors agree that there is no need to expand a component of an already introduced compound abbreviation.  For instance, after introducing ST-elevation myocardial infarction (STEMI), there is no need to expand MI.  In your example, there is no need to treat CML as a new abbreviation.—Cheryl Iverson, MA