HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Full Text (PDF) Submit a response Alert me when this article is cited Alert me when eLetters are posted Alert me if a correction is posted Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via Google Scholar Google Scholar Articles by Humpal, J. J. Search for Related Content PubMed PubMed Citation Articles by Humpal, J. J.

Numbers, Statistics, and Tabular Data¹

¹ From the Radiological Society of North America, 820 Jorie Blvd, Oak Brook, IL 60523. Address correspondence to the author (e-mail: humpal@rsna.org).

Index terms: Radiology (journal)

The presentation of data, including numeric data, is arguably the most important aspect of research report preparation. The Results section should provide, in a straightforward and logical manner, the answer to the research question posed by the investigators (1,2); therefore, findings must be reported unambiguously.

The RSNA publications department applies specific guidelines to the presentation of numbers, statistics, and tabular data. These guidelines, which are intended to provide a consistent "look" for data in all RSNA publications, are derived from a few well-regarded sources, prominently the American Medical Association Manual of Style (3).

The numbers zero through nine are spelled out in the body of the article, unless they are used with units of measure, and arabic numerals are always used for numbers 10 and greater; thus, "nine patients," "five of 41 cases," "the 1st year," "the third section." Because of our style for citing references, numbers should not be used by themselves inside parentheses. For example, "The results are preliminary because the number of patients was small (5)" is ambiguous: Does the author mean five patients or reference 5?

Numerals are always used with units of measure, including all time measures (eg, minutes, days, years, decades). RSNA journals, like most medical journals (3), report units according to the International System of Units (Le Système International d’Unites, or SI). RSNA journals do permit dual reporting of most units—that is, both conventional and SI units may be given—but inclusion of SI units is mandatory. Exceptions to the use of SI units are, most notably, common physiologic pressure (reported in millimeters of mercury or of water, not the SI unit newton per square meter) and temperature (reported in degrees Celsius, not the SI unit Kelvin) (1,3), and the diameter of guide wires and the thickness of film (inches, as well as metric measures, are acceptable).

Percentages and proportions are convenient for describing outcomes and population characteristics. Because these numbers are summary values, however, they can obscure important information about a study. For example, two studies of drug efficacy may report widely different outcomes: Study A showed that 80% of patients benefited from a drug, whereas study B showed that only 50% benefited. It is impossible to interpret these outcomes unless the reader knows that study A was a pilot study with only five patients, and study B was a clinical trial with 500 patients. Thus, RSNA journals require that the numbers used to calculate percentages and proportions be reported in the text or tables. RSNA journals also apply a guideline for the precision of percentages. When the denominator is less than 100, percentages are reported as whole numbers; percentages can be reported to the tenths (one decimal place) for denominators of 100–999, to the hundredths for denominators of 1,000–9,999, and so forth.

We treat some numbers in a special manner. For example, we delete the leading zero in P values because, by definition, a P value based on a sample can never equal 1.0. In contrast, many other statistical parameters, including correlation coefficients, values, and the area under a receiver operating characteristic curve, can equal 1.0, so a leading zero is not deleted. In addition, P values cannot exceed a precision of .001 (unless the Bonferroni correction has been applied), because smaller P values do not add information (3).

In most research studies, statistics are used to test hypotheses. The tests and data that were compared must be introduced in the Materials and Methods section (1,3), and the outcomes of any such tests must be reported in the Results section (3). These requirements are applicable not only for methods of hypothesis testing (eg, Student t test) but also for diagnostic discrimination statistics (eg, sensitivity).

Well-conceived and well-structured tables are an excellent way to present and organize data. Tables can be used to show the relationship between dependent and independent variables, to show the changes in measurements over time, or to detail patient or population characteristics.

The simplest table has two columns (tables consisting of one column are lists, which should be incorporated into the text or, when appropriate, recast as figures). The left-hand column is called the stub and contains the headings for the rows. Each column must have a heading, which, when applicable, states in parentheses the unit of measure for data in the column. In RSNA publications, data in parentheses in the body of a table are defined with a footnote. Numbers in a table must be consistent with those in the text. An unambiguous table contains no empty cells. Rather than dashes, ellipses, or empty cells, the use of "not applicable", "no data", or another similar phrase or zero (when appropriate) makes it easier to read and understand the table.

REFERENCES

1. International Committee of Medical Journal Editors. Uniform requirements for manuscripts submitted to biomedical journals. Ann Intern Med 1997; 126:36-47.[Free Full Text]
2. Huth EJ. How to write and publish papers in the medical sciences 2nd ed. Baltimore, Md: Williams & Wilkins, 1990.
3. Iverson C, Flanagin A, Fontanarosa PB, et al. American Medical Association manual of style 9th ed. Baltimore, Md: Williams & Wilkins, 1998.

This Article Full Text (PDF) Submit a response Alert me when this article is cited Alert me when eLetters are posted Alert me if a correction is posted Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via Google Scholar Google Scholar Articles by Humpal, J. J. Search for Related Content PubMed PubMed Citation Articles by Humpal, J. J.