Tables - Tables, Figures, and Multimedia

AMA Manual of Style - Stacy L. Christiansen, Cheryl Iverson 2020

Tables
Tables, Figures, and Multimedia

Tables and figures show relationships among data and other types of numerical, textual, or visual information. Text may be preferred if the information can be presented concisely (Box 4.1-1). For qualitative information, small amounts of data with a few simple comparisons should usually be presented in words, whereas large amounts of data with several comparisons should usually be presented in tables, graphs, or illustrations. For quantitative information, a table should be used when the display of exact values is important, whereas a figure (eg, a line or bar graph) should be used to show patterns or trends. Tables also are often preferable to graphics for small data sets and are preferred when data presentation requires many specific comparisons. However, primary outcomes should not be presented in a figure only or referred to in the text only with a general statement. For those findings, absolute numbers and measures of variance should also be included in the text.

Priorities in the creation and publication of tables and figures are to emphasize important information and to ensure that each table and figure makes a clear point. In addition to presenting study results, tables and figures can be used to explain or amplify the methods or highlight other key points in the article. Like a paragraph, each table or figure should be cohesive and focused. To be most effective, tables and figures should present ideas and information in a logical sequence. The relationship of tables and figures to the text and to each other should be considered in manuscript preparation, editorial evaluation and peer review, manuscript editing and tagging, and article layout and display for print/PDF and online.

When used properly, tables and figures add variety to article layout and are visually compelling and distinct components of scientific publications. However, authors and editors of scientific publications should avoid using tables and figures simply to break up text or to impart visual interest.

Multimedia (video, audio, interactive files), although available only online, can enhance understanding of an article’s content.

Box 4.1-1. Guidelines for Using Text, Tables, and Figures to Display Numerical Data

Uses of text

Present quantitative data that can be given concisely and clearly

Describe simple relationships among data

Uses of tables

Present more than a few precise numerical values

Present large amounts of detailed quantitative information in a smaller space than would be required in the text

Show detailed item-to-item comparisons

Display many numbers simultaneously

Display individual data values precisely

Show complex relationships among the data

Uses of figures

Highlight patterns or trends in data

Show changes or differences over time

Display complex relationships among quantitative variables

Aid decision-making

4.1 Tables.

Because of their ability to present detailed information effectively and in ways that text alone cannot, tables are an essential component of many scientific articles. Tables can summarize, organize, and condense complex or detailed data and therefore are routinely used to present study results.

The purpose of a table is to present data or information and support statements in the text. Information in the table must be accurate and consistent with that in the text in content and style. A properly designed and constructed table should be able to stand independently, without requiring undue reference to the text.

4.1.1 Types of Tables.

4.1.1.1 Table.

A table displays information arranged in columns and rows (see 4.1.4, Table Components and Table 4.1-1) and is used most commonly to present numerical data. Each table should have an object identifier (eg, Table) and a title, be numbered consecutively as referred to in the text, and be positioned as closely as possible to its first mention in the text. Formal tables usually are framed by horizontal rules, boxes, or white space. Some journals add background shading.

Table 4.1-1. Basic Table, Including Title, Columns, Rows, and footnotes

Table. Characteristics of Heat Waves During the Study Period, 1999-2010a

Heat wave definition

Counties with ≥1 heat wave event, No. (%)

Frequency of heat waves, mean No. per year (IQR)

Duration of heat waves, mean (IQR), d

Temperature, mean (IQR), °C

No. of days

Temperature percentile

During heat wave days

During non—heat wave days

≥2

>97

1943 (100)

2.6 (2.4-2.8)

3.3 (3.0-3.6)

28.3 (26.9-30.1)

23.1 (21.0-25.5)

≥2

>98

1943 (100)

1.8 (1.6-1.9)

3.1 (2.8-3.3)

28.8 (27.4-30.5)

23.1 (21.1-25.5)

≥2

>99

1943 (100)

0.9 (0.8-1.0)

2.8 (2.5-3.0)

29.5 (28.1-31.1)

23.1 (21.2-25.5)

≥4

>97

1942 (99.9)

0.8 (0.7-0.9)

5.5 (4.8-6.0)

28.5 (27.0-30.4)

23.2 (21.2-25.6)

≥4

>98

1935 (99.6)

0.5 (0.3-0.6)

5.1 (4.5-5.6)

29.0 (27.5-30.8)

23.2 (21.3-25.5)

≥4

>99

1723 (88.7)

0.2 (0.1-0.2)

4.2 (4.0-5.0)

29.6 (28.1-31.4)

23.2 (21.1-25.8)

Abbreviation: IQR, interquartile range.

a Heat waves are defined as at least 2 or at least 4 consecutive days with temperatures exceeding the 97th, 98th, or 99th percentile of a county’s daily temperatures, 1999-2010. The heat wave definition used in the main analysis was at least 2 days’ duration with temperatures exceeding the 99th percentile of the county’s daily temperatures. Non—heat wave days are not in the heat wave period but are matched by county and week.

4.1.1.2 Matrix.

A matrix is a tabular structure that uses numbers, short words (eg, no, yes), symbols (eg, bullets, check marks), or shading to depict relationships among items in columns and rows and to allow comparisons among entries. A matrix may also be processed as a figure, depending on the complexity of the construction and the need for multiple colors or shading (Table 4.1-2).

Table 4.1-2. Matrix That Is Presented as a Figure

Figure. Paired Grading Class Comparison for Paravalvular Regurgitation (PVR)

Image

4.1.1.3 Boxes, Sidebars, and Other Nontabular Material.

Information that is complementary to the text (eg, lists) can be set off in a box or sidebar within the article (see 4.3, Nontabular Material).

4.1.2 Organizing Information in Tables.

For a table to have maximum effectiveness, the information it contains must be arranged logically and clearly so that the reader can quickly understand the key point and find the specific data and comparisons of interest.

During the planning and creation of a table, the author should consider the primary purpose of the table: what data need to be included, compared, or emphasized. Because the English language is read first horizontally (from left to right) and then vertically (from top to bottom), the primary comparisons should be shown horizontally across the table. Data that depict cause-and-effect correlations, before-and-after relationships, or trends (eg, change over time) should be arranged from left to right if space allows or, alternatively, from top to bottom. Information being compared (such as numerical data) should be juxtaposed within adjacent columns to facilitate comparisons among items of interest.1 The tables in Table 4.1-3 present the same information. In the first table, the values for the dependent variable (years of study) are placed in the row headings, and the data for each independent variable must be compared vertically, up and down rows. Reversing the axes of the table so that the values for the dependent variable are placed in the column headings allows the data to be compared horizontally, between columns.

Table 4.1-3. Two Tables With Identical Information Present the Use of a Vertical Format for the Primary Outcome vs a Horizontal Format

Table. Relative Risk for Death After Onset of Heart Failure Defined by the Framingham Criteriaa

Year

Men by age, y

Women by age, y

60

70

80

60

70

80

1996-2000

1 [Reference]

1 [Reference]

1 [Reference]

1 [Reference]

1 [Reference]

1 [Reference]

2001-2005

0.84 (0.69-1.02)

0.84 (0.73-0.97)

0.85 (0.72-1.00)

0.80 (0.63-1.03)

0.91 (0.77-1.06)

1.02 (0.90-1.15)

2006-2010

0.63 (0.50-0.80)

0.74 (0.63-0.88)

0.88 (0.75-1.04)

0.95 (0.73-1.24)

0.99 (0.83-1.18)

1.03 (0.90-1.17)

2011-2015

0.48 (0.36-0.64)

0.59 (0.49-0.71)

0.72 (0.61-0.87)

0.67 (0.48-0.92)

0.79 (0.64-0.98)

0.94 (0.82-1.09)

a Data are presented as relative risk (95% CI).

Table. Relative Risk for Death After Onset of Heart Failure Defined by the Framingham Criteria

Age, y

Relative risk (95% CI)

1996-2000

2001-2005

2006-2010

2011-2015

Men

60

1 [Reference]

0.84 (0.69-1.02)

0.63 (0.50-0.80)

0.48 (0.36-0.64)

70

1 [Reference]

0.84 (0.73-0.97)

0.74 (0.63-0.88)

0.59 (0.49-0.71)

80

1 [Reference]

0.85 (0.72-1.00)

0.88 (0.75-1.04)

0.72 (0.61-0.87)

Women

60

1 [Reference]

0.80 (0.63-1.03)

0.95 (0.73-1.24)

0.67 (0.48-0.92)

70

1 [Reference]

0.91 (0.77-1.06)

0.99 (0.83-1.18)

0.79 (0.64-0.98)

80

1 [Reference]

1.02 (0.90-1.15)

1.03 (0.90-1.17)

0.94 (0.82-1.09)

Although tables frequently are used to present many quantitative values, authors should remember that tabulating all collected study data is unnecessary and actually may distract and overwhelm the reader. Data presented in a table should be pertinent and meaningful. For example, when reporting the sex of participants, choose the sex that constitutes most of the sample if you report data for only 1 sex.2 (Note that JAMA prefers complete reporting of all participants in clinical trials.) When reporting on racial and ethnic differences, be as specific as possible and define who is included in an “other” category (see 11.12, Inclusive Language).

The length of the table should also be considered. For ease of reading print journals and PDFs as well as for practical reasons, a table that would span more than a printed page horizontally or run vertically onto a second page should, if possible, be recast into 2 or more smaller tables. If this is not possible, the table may be set in smaller type. Another option is to publish the table as supplementary material in electronic form only, with a note in the print publication, but the same difficulty in following column headings down the table or losing one’s place while scrolling wide tables occurs online as well. Yet another option is to run a table in landscape format, requiring the print reader to turn the article sideways to view (these are usually rotated online).

4.1.3 Tables in Online Journals.

For the JAMA Network journals online, tables appear as downloadable images, whereas other journals may use HTML tables for the online version of an article. The advantage of using tables that are images is that the table’s appearance, such as the design and data alignment, can be controlled and allows for consistent presentation of the data, which is an important consideration for larger, more complex tables. The use of HTML tables has its advantages as well, the first being that a graphic file need not be downloaded to view the table, making the data more immediately accessible on any electronic device (smartphone, tablet, or computer). In addition, the table’s text is searchable, and hyperlinks can be included within the table, such as reference citations that link to the reference in the list of all references at the end of the file or in a bibliography.

4.1.4 Table Components.

Formal tables in scientific articles conventionally contain 6 major elements: object identifier, title, column headings, stubs (row headings), body (data field) consisting of individual cells (data points), and footnotes (Table 4.1-1). Details pertaining to elements of style for table construction vary among publications; what follows is based on the general style of the JAMA Network journals.

4.1.4.1 Object Identifier.

Each table should be formally identified as such, and if there is more than 1 in an article, the tables should be numbered consecutively according to the order in which they are mentioned in the text (Table 1, Table 2, etc). If the article contains only 1 table, it is referred to in the text as “Table.”

4.1.4.2 Title.

Each table should have a brief, specific, descriptive title, usually written as a phrase rather than as a sentence, that distinguishes the table from other data displays in the article. The title should convey the topic of the table succinctly but should not provide detailed background information or summarize or interpret the results.

The word “Table” and the table number are part of the title. The capitalization style used in article titles should be followed for table titles (see 10.2, Titles and Headings). The following are examples of table titles:

Table 4. Ten-Year Prevalence of Fractures, Falls, or Musculoskeletal Injuries Among Beneficiaries With Disorders of Binocular Vision

Table 3. Adjusted Relative Risks for Asthma and Asthma Duration Predicting 4-Year Incidence of Obstructive Sleep Apnea

Certain article types have required tables, such as a baseline characteristics table in a randomized clinical trial or a listing of studies included in a meta-analysis; the table titles should reflect this content:

Table 1. Baseline Characteristics of Patients in the HOME Trial

Table. Characteristics of Studies Included in the Meta-analysis

4.1.4.3 Column Headings.

The main categories of information in the table should be contained in separate columns. In tables for studies that have independent and dependent variables, the independent variables conventionally are displayed in the left-hand column (row heading) and the dependent variables in the columns to the right. Each column should have a brief heading that identifies and applies to all items listed in that column. Terms used in the text should be consistently used in table headings as well. It is important to label groups consistently to avoid confusing readers.1 For example, “placebo,” “control,” and “untreated” should not be used to describe the same group of individuals; instead, use the same term in tables as can be found in the abstract, text, and figures. When possible, use highly descriptive terms (eg, “Metronidazole group” is preferable to “Treatment group”).

It is acceptable to omit the heading for the first column if the entries are different or the entries do not need further explanation (Table 4.1-4).3

Table 4.1-4. Omitted Heading for First Column

Table. Baseline Characteristics of Study Participants by Ferritin Level


Low ferritin (≤26 ng/mL)

Higher ferritin (>26 ng/mL)

Iron (n = 51)

No iron (n = 50)

Iron (n = 60)

No iron (n = 54)

Women, No. (%)

33 (64.7)

31 (62.0)

38 (63.3)

34 (63.0)

Age ≥60 y, No. (%)

12 (23.5)

11 (22.0)

17 (28.3)

12 (22.2)

Age, mean (SD), y

47.5 (15.5)

45.9 (15.7)

49.3 (14.6)

48.1 (14.6)

Weight, mean (SD), kg

75.9 (16.4)

76.8 (15.8)

81.5 (16.4)

77.9 (16.2)

Hemoglobin, mean (SD), g/dL

13.2 (1.0)

13.7 (1.3)

14.1 (1.0)

14.3 (1.2)

Ferritin, mean (SD), ng/mL

14.9 (5.8)

15.2 (6.0)

54.0 (24.3)

58.9 (32.9)

sTfR, mean (SD), mg/L

4.0 (1.33)

3.9 (1.19)

3.1 (0.65)

3.1 (0.62)

Estimated blood volume, mean (SD), L

4.59 (0.8)

4.66 (0.91)

4.79 (0.84)

4.64 (0.84)

Abbreviation: sTfR, soluble transferrin receptor.SI conversion factor: To convert ferritin to pmol/L, multiply by 2.247.

Units of measure should be indicated in the column heading, unless provided in the row heading, and are preceded by a comma (Table 4.1-1). Column headings are set in boldface type. If necessary, column subheadings may be used. For more complex headings, braces (or spanning rules) may be used or additional explanatory information may be provided in the footnotes (Table 4.1-1).

If all elements in a column are identical (eg, if all studies in a review used the same assay in the study methods), this information could be provided in a footnote or in the table title and the column deleted (Table 4.1-5).3

Table 4.1-5. Use of a Footnote to Convey Information That Would Have Been Identical for Each Row (See Footnote a)

Table. Cases of MOG-IgG—Associated Seizures Identified on Literature Reviewa

Source

No. of patients

Age, y/sex

Ethnicity

Other antibodies

Type of seizure

Clinical syndrome

Hino-Fukuyo et al,9 2015

3

12/M; 14/M; 5/M

Japanese

None

NA

ADEM

Tsuburaya et al,10 2015

1

7/M

Japanese

None

Partial (eye deviation and L arm clonic seizures)

ADEM, ON

Ramberger et al,11 2015

22b

NA

NA

None

NA

ADEM

Titulaer et al,12 2014

1

4/F

Hispanic

NMDAR

NA

Seizures, hemiparesis; later: mutism, chorea, and orofacial dyskinesia

Ogawa et al,13 2017

4

39/M; 36/M; 23/M; 38/M

Japanese?

None

GTC; GTC; GTC + focal; GTC

Encephalopathy, ON; seizure, ON; encephalopathy; seizure, aphasia, and R hemiparesis

Fujimori et al,14 2017

1

46/M

Japanese

None

Focal progressed to secondary generalized

Encephalopathy, paraplegia

Abbreviations: ADEM, acute disseminated encephalomyelitis; FLAIR, fluid-attenuated inversion recovery; GTC, generalized tonic-clonic; L, left; MOG, myelin oligodendrocyte glycoprotein; MRI, magnetic resonance imaging; NA, not available; NMDAR, N-methyl-D-aspartate receptor; R, right; ON, optic neuritis.

a All studies used a cell-based assay for MOG antibody testing.

b Represents patients who presented with 1 or a combination of cognitive impairments or seizures. It is unclear how many of these patients had seizures.

In column headings, style guidelines regarding numbers (eg, use of ordinals) and abbreviations may be relaxed somewhat to save space, with abbreviations expanded in a footnote. However, when space allows spelled-out headings, expansions are preferable to abbreviations. Column and row headings are set in sentence case (only an initial cap), similar to axis labels in figures (see 4.2.6.2, Axis Labels).

4.1.4.4 Row Headings.

The left-most column of a table contains the row headings, which apply to all items in that row. If a unit of measure is not included in the column heading, it should be included here. Row headings are capitalized according to the style for sentences, not titles. Therefore, if a symbol (such as %), an arabic numeral, or a lowercase Greek letter (such as β) begins the entry, the first major word to follow should be capitalized. Row headings are left-justified, and indents are typically used to depict hierarchical relationships (Table 4.1-6). However, some publications use bold or shading instead.

Table 4.1-6. Hierarchy of Stubs (Row Headings)

Table. Characteristics of Men With Androgen Deprivation Therapy Use

Characteristic

Data value

No. of men

35 487

Age at index date, y


 Mean (SD)

75.67 (5.93)

 Median (IQR)

75 (71-80)

Income quintile, No. (%)

 1 (Lowest)

5093 (14.4)

 2

5959 (16.8)

 3

5866 (16.5)

 4

5954 (16.8)

 5 (Highest)

6455 (18.2)

Living in rural area, No. (%)

6160 (17.4)

Prior diagnosis of osteoporosis, No. (%)

1387 (3.9)

Prior fragility fracture, No. (%)

774 (2.2)

Charlson Comorbidity Index score, median (IQR)

0 (0-0)

Abbreviation: IQR, interquartile range.

For a table that may be readily divided into parts to enhance clarity or for 2 closely related tables that would be better combined, cut-in headings may be used. A cut-in heading may be set in boldface type to draw the reader’s attention. It is placed above the table columns but below the column heads and applies to all the tabular material in the portion of the table immediately below it (Table 4.1-7). The cells with “NA” (under “Chronic pancreatitis”) indicate “not applicable” because the training cohort had no individuals and hence contributed no data.

In some publications, cut-in headings are centered. However, centered cut-in headings may interfere with downward scanning and may not be as readable and thus should be used with care.3,4

Both column headings and row headings should be consistent in style and presentation between tables in the same article.

Table 4.1-7. Cut-in Headings Divide the Table Into Related Sections

Table. Demographics of Patients and Healthy Participants in the Discovery, Training, and Validation Cohortsa


No. (%) of patients and healthy participantsb

Discovery cohort (n = 230)

Training cohort (n = 379)

Validation cohort (n = 137)

Pancreatic cancer

143 (62)

180 (47)

86 (63)

Sex




 Men

88 (61.5)

102 (56.7)

49 (57)

 Women

55 (38.5)

78 (43.3)

37 (43)

Age, median (range), y

65 (38-75)

68 (37-89)

67 (46-86)

Resection of tumors




 Yes

23 (16.0)

10 (5.6)

11 (12.8)

 No

120 (84.0)

170 (94.4)

75 (87.2)

Cancer stage




 IA

3 (2.1)

2 (1.1)

1 (1.2)

 IB

2 (1.4)

2 (1.1)

0

 IIA

11 (7.7)

7 (3.9)

1 (1.2)

 IIB

14 (9.8)

15 (8.3)

12 (14.0)

 III

26 (18.2)

44 (24.4)

22 (25.6)

 IV

87 (60.8)

107 (59.4)

50 (58.1)

 Unknown

0

3 (1.7)

0

Serum CA19-9, median (range), kU/L

791 (3-608 500)

508 (1-467 000)

435 (2-182 300)

Chronic pancreatitis

18 (8)

0

7 (5)

Sex




 Men

13 (72.2)

NA

3 (42.9)

 Women

5 (27.8)

NA

4 (57.1)

Age, median (range), y

54 (33-67)

NA

56 (47-85)

Serum CA19-9, median (range), kU/L

27 (3-1134)

NA

41 (3-159)

Healthy participants

69 (30)

199 (53)

44 (32)

Sex




 Men

31 (44.9)

95 (47.7)

22 (50.0)

 Women

38 (55.1)

104 (52.3)

22 (50.0)

Age, median (range), y

53 (33-66)

50 (18-66)

55 (41-65)

Serum CA19-9, median (range), kU/L

NA

NA

4 (3-40)

Abbreviation: CA19-9, cancer antigen 19-9; NA, not applicable.

a More information is provided in the Supplement (see Patients).

b Unless otherwise indicated.

4.1.4.5 Field.

The field or body of the table presents the data. Each data entry point is contained in a cell, which is the intersection of a column and a row. Table cells may contain numerals, text, symbols, or a combination of these. However, all data must be consistent with their row and column headings. Data in the field should be arranged logically, either by sensibly ordering the rows and column headings, which helps readers find the value of interest, or by ordering the values, which can help reveal patterns in the data. For instance, time order should be used for data collected in sequence (Table 4.1-1). Similar types of data should be grouped. Numbers that are summed or averaged should be placed in the same column. Text in the field cells should be capitalized in sentence style (ie, the first word is capitalized and all words that follow in the cell are lowercased except proper nouns). Justified margins are not recommended because the spacing between words can be uneven, impairing readability.1 For example, the JAMA Network journals set columns with text and numbers primarily flush left.

Missing data and blank space in the table field (ie, an empty cell) may create ambiguity and should be avoided, unless an entry in a cell does not apply (eg, a column head does not apply to one of the stub items).4 Use of ellipses in cells is also ambiguous and should be avoided. The numeral 0 should be used only to indicate that the value of the data in the cell is zero. Designations such as NA (for “not available,” “not analyzed,” or “not applicable”) may be used, provided their meaning is explained in a footnote.

Blank cells may be acceptable when an entire section of the table does not contain data, or the use of “NA” is acceptable.

4.1.4.6 Merged Cells.

In some tables, values may apply across multiple rows of the table, either because they belong to various subgroups or because the analysis was performed across categories. One of the ways this can be illustrated is by merging the cells (usually in the right-most column) to depict this relationship (see the last 4 columns in Table 4.1-8).

Table 4.1-8. Rows Merged Vertically to Show That the Adjusted Mean Differences, Accompanying 95% CIs, and P Values Were Computed Across Subcategories

Table. Unadjusted and Adjusted Difference Scores of BREAST-Q Patient-Reported Outcomes (PROs)

BREAST-Q survey

Cohort

Unadjusted scores, mean (SD)

Adjusted mean difference (95% CI)a

Baseline

1 y Postoperative

2 y Postoperative

1 y Postoperative

P value

2 y Postoperative

P value

Satisfaction with breast

Fat grafted

58.7 (21.5)

60.1 (16.7)

65.6 (17.1)

−4.74 (−8.21 to −1.28)

.008

−0.68 (−4.42 to 3.06)

.72

No fat graft

59.2 (22.5)

66.1 (17.2)

66.0 (18.3)

Psychosocial well-being

Fat grafted

68.4 (18.7)

67.2 (19.3)

73.2 (19.2)

−3.87 (−7.33 to −0.40)

.03

−0.59 (−3.92 to 2.74)

.73

No fat graft

68.8 (18.5)

73.5 (19.2)

75.3 (19.1)

Physical well-being

Fat grafted

77.2 (16.0)

72.5 (13.5)

74.8 (15.2)

−1.23 (−3.71 to 1.25)

.33

−0.50 (−3.36 to 2.36)

.73

No fat graft

78.4 (14.7)

76.2 (14.9)

76.8 (14.9)

Sexual well-being

Fat grafted

55.7 (20.3)

48.0 (20.5)

52.8 (20.9)

−5.59 (−9.70 to −1.47)

.008

−2.94 (−7.01 to 1.12)

.15

No fat graft

54.4 (20.9)

54.7 (21.0)

55.4 (21.9)

a Fat-grafted and non—fat-grafted differences based on mixed-effects regression models with each PRO measured at 1 or 2 years postoperatively as the dependent variable. Each model included an indicator for fat grafting between years 1 and 2 as the primary predictor, and included as covariates baseline PRO, age, body mass index, procedure type, laterality, indication for mastectomy, timing of reconstruction, radiotherapy, smoking history, race, ethnicity, fat grafting before year 1 PRO measures, concurrent revision procedure, cancer recurrence, and prior complication. Also included are random intercepts for study sites (hospitals) and an interaction variable between fat grafting and concurrent revision procedures. Analyses were performed and combined using 10 imputed data sets.

4.1.4.7 Totals.

Totals and percentages in tables should correspond to values presented in the text and abstract and should be verified for accuracy. Any discrepancies (eg, because of rounding) should be explained in a footnote.

Boldface type for true totals (ie, those that represent sums of values in the table) should be used with discretion, although the JAMA Network journals typically do not do this. Boldface should not be used to overemphasize data in the table (eg, significant odds ratios or P values).

4.1.4.8 Alignment of Data.

Horizontal alignment (across rows) must be considered in setting tables. If the table row heading column contains lines of text that exceed the width of the column (runover lines) and the cell entries in that row do not, the field entries should be aligned across the first or top line of the entry (Table 4.1-9). This top-line alignment of data applies to tables that have numbers, words, or both as cell entries. If some entries within the table field contain information that cannot be contained on a single line in the cell (runover lines in the table field), the table entries in that row also should be aligned across on the first line of the row heading entry (Table 4.1-5).

Table 4.1-9. Alignment of Data With the First Line in the Stub Entry

Table. Response of IOP at Final Follow-up at 6 Months

Outcome

SLT (n = 50)

PGA (n = 50)

Mean difference (95% CI)

P value

BCVA





 Mean (SD), logMAR

0.12 (0.12)

0.09 (0.12)

−0.02 (−0.07 to 0.03)

.57

 Mean Snellen equivalent

20/32

20/25



IOP, mean (SD), mm Hg

19.5 (3.3)

18.1 (2.4)

−1.2 (−2.5 to −0.15)

.05

Mean change in IOP from baseline (95% CI), mm Hg

4.0 (3.2 to 4.8)

4.2 (3.5 to 4.9)

0.2 (−0.8 to 1.3)

.78

Adjusted mean change in IOP from baseline (95% CI), mm Hga

3.7 (3.7 to 4.5)

4.4 (3.8 to 5.2)

0.7 (−0.3 to 1.7)

.17

Change in IOP from baseline, mean (SD), %

16.9 (12.1)

18.5 (10.6)

1.5 (−2.9 to 6.0)

.52

Abbreviations: BCVA, best-corrected visual acuity; IOP, intraocular pressure; PGA, prostaglandin analogue; SLT, selective laser trabeculoplasty.

a Adjusted for baseline differences in IOP.

Vertical alignment within each column of a table is important for the visual presentation of data. In some publications, data are aligned on common elements, such as decimal points, plus or minus signs, hyphens (used in ranges), virgules, or parentheses.

For an explanation of the use of “to” vs a hyphen in ranges with negative numbers (Table 4.1-9), see 8.3.1.3, Expressing Ranges and Dimensions.

However, in some journals, including the JAMA Network journals, all columns are set flush left whether the cell contains text or data.

4.1.4.9 Rules and Shading.

Some journals add rules and shading during the production process. For example, the JAMA Network journals use horizontal rules to separate rows of data. Other journals use shading for the same purpose (Table 4.1-10). The JAMA Network journals request that tables be submitted without rules drawn in (as opposed to table borders, which are appropriate) or shading. If these elements are included they will be manually removed during the editing process (see 4.1.10, Guidelines for Preparing and Submitting Tables).

Table 4.1-10. Shading Used to Separate Rows of Data

Table. Baseline Characteristics of Study Participants by Ferritin Level


Ferritin ≤26 ng/mL

Ferritin >26 ng/mL

Iron (n = 51)

No iron (n = 50)

Iron (n = 60)

No iron (n = 54)

Number (%) women

33 (64.7)

31 (62.0)

38 (63.3)

34 (63.0)

Number (%) ≥60 y old

12 (23.5)

11 (22.0)

17 (28.3)

12 (22.2)

Age, mean (SD), years

47.5 (15.5)

45.9 (15.7)

49.3 (14.6)

48.1 (14.6)

Weight, mean (SD), kg

75.9 (16.4)

76.8 (15.8)

81.5 (16.4)

77.9 (16.2)

Hemoglobin, mean (SD), g/dL

13.2 (1.0)

13.7 (1.3)

14.1 (1.0)

14.3 (1.2)

Ferritin, mean (SD), ng/mL

14.9 (5.8)

15.2 (6.0)

54.0 (24.3)

58.9 (32.9)

Soluble transferrin receptor, mean (SD), mg/L

4.0 (1.33)

3.9 (1.19)

3.1 (0.65)

3.1 (0.62)

Estimated blood volume, mean (SD), L

4.59 (0.8)

4.66 (0.91)

4.79 (0.84)

4.64 (0.84)

SI conversion factor: To convert ferritin to pmol/L, multiply by 2.247.

4.1.4.10 Footnotes.

Footnotes may contain information about the entire table, portions of the table (eg, a column), or a discrete table entry. The order of the footnotes is determined by the placement in the table of the item to which the footnote refers. The letter for a footnote that applies to the entire table (eg, one that explains the method used to gather the data or format of data presentation) should be placed after the table title. A footnote that applies to 1 or 2 columns or rows should be placed after the heading(s) to which it refers. A footnote that applies to a single entry in the table or to several individual entries should be placed at the end of each entry to which it applies.

For both tables and figures, footnotes are indicated with superscript lowercase letters in alphabetical order (a-z). The font size of the footnote letters should be large enough to see clearly without appearing to be part of the actual data. For tables in which superscript numbers and/or letters are used to display data, care should be taken to ensure that superscript footnote letters are distinguished clearly from superscripts used for data elements. Although some publications use symbols (*, †, etc) to indicate footnotes in tables, such symbols are ordered arbitrarily and are limited in number. Use of superscript letters ensures a logical order to the entries and a much larger supply of notations (26 characters). In addition, use of letters reduces the likelihood that the symbol could be misinterpreted as an exponent or reference citation.4

Footnotes are listed at the bottom of the table, each on its own line. However, to save space, tables with more than a few footnotes can use 2 columns for the footnotes (Table 4.1-11).

Table 4.1-11. When Tables Span the Width of the Page or Have Many Footnotes, They Can Be Presented in 2 Columns Instead of With a Single Footnote on Each Line

Table. Univariable and Multivariable Analyses of the Association Between at Least Moderate PVR and 1-Year Outcomes

End point

Univariable analysis

Multivariable analysisa

HR (95% CI)b

P value

HR (95% CI)

P value

All-cause mortality

2.40 (1.30-4.43)

.005

2.59 (1.39-4.85)

.003

Cardiovascular mortality

2.68 (1.24-5.81)

.01

2.87 (1.30-6.30)

.009

Rehospitalization

2.27 (1.34-3.83)

.002

2.27 (1.31-3.94)

.003

Composite of mortality and rehospitalization

2.35 (1.52-3.62)

.001

2.36 (1.50-3.69)

<.001

Aortic valve reintervention

13.14 (3.39-50.85)

<.001

NA

NA

Abbreviations: HR, hazard ratio; NA, not applicable; PVR, paravalvular regurgitation; STS, Society of Thoracic Surgeons.

a Adjusted for age, sex, body mass index, STS score, diabetes, at least moderate baseline aortic regurgitation, and at least moderate baseline mitral regurgitation.

b Hazard ratio is for at least moderate PVR vs less than moderate PVR. Multivariable analysis was not performed for aortic valve reintervention because there were only 10 events.

Journals often have underlying grids to permit different sizes of tables to run in a standard 2- or 3-column print format. For example, tables can be sized to fill 1.5 columns of content and footnotes run in the space beside the table (Table 4.1-12).

Table 4.1-12. A Partial-Width Table (1.5 Columns in This Publication) With the Footnotes Beside the Table, Aligned With the Bottom of the Table

Table. Use of the Intervention Components Among Individuals in the Intervention Group and Smoking Cessation Rates by Component

Component use

Individuals in the intervention group who quit smoking, No./Total (%) (n = 399)

P value

Used intervention component

Did not use intervention component

Spoke to the tobacco treatment specialist

58/274 (21.2)

13/125 (10.4)

.01

Received nicotine replacement patchesa

50/218 (22.9)

8/56 (14.3)

.16

Received a HelpSteps.com referrala

30/128 (23.4)

28/146 (19.2)

.39

Reported using a HelpSteps.com referralb

24/55 (43.6)

34/219 (15.3)

<.001

a Among intervention participants who spoke to the tobacco treatment specialist.

b Among intervention participants who spoke to the tobacco treatment specialist and received a HelpSteps.com referral.

Footnotes may be phrases or complete sentences and should end with a period. Any operational signs, such as <, >, or =, imply a verb. For example, P = .01 is considered a complete sentence (“P is equal to .01.”) when used as a table footnote. Footnote letters should appear before the footnote text (preferably in superscript to distinguish them from the text) and are followed by a space for clarity. In addition, in the JAMA Network journals, the abbreviations and units of measure conversion footnotes appear first and are set off with an introductory word or phrase instead of a letter. In such footnotes, abbreviations are expanded in alphabetical order, and units of measure are listed by consecutive mention in the table (Table 4.1-13).

If several tables share a detailed or long footnote that explains several abbreviations or methods, this footnote may appear in the first table for which it is applicable, and a footnote in each succeeding table for which the footnote also is applicable may refer the reader to the first appearance of the detailed information:

Study acronyms are explained in the first footnote to Table 1.

The reader also may be referred to a relevant discussion in the text by a footnote:

See the Statistical Analysis section for a description of this procedure.

Several of the most common uses of footnotes include the following:

To expand abbreviations:

Abbreviations: OR, odds ratio; RR, relative risk.

To designate reporting of numerical values:

aScores are based on a scale of 1 to 10, with 1 indicating least severe and 10, most severe.

To provide information on statistical analyses or experimental methods:

bAdjusted for age, smoking status, and body mass index.

To explain a discrepancy in numerical data:

aBecause of rounding, percentages may not total 100.

To cite references for information used in the table:

cInternational Classification of Health Problems in Primary Care.45

To acknowledge that data in the table are taken from or based on data from another source:

aData from the US Census Bureau.5

To acknowledge credit for reproduction of a table (if the table has been reprinted or modified with permission from another source, credit should be given in a footnote):

aReproduced with permission of the AMA Manual of Style. American Medical Association, 2020.

aAdapted with permission from Vega and Avner.41

References for information in a table or figure should be numbered and listed as if this information were part of the text. For instance, if the source from which the material referred to in the table or figure is one of the references used in the text, that reference number should be used in the table or figure. If the reference pertains only to the table or figure (ie, the source is not cited elsewhere in the text), the reference should be listed and numbered according to the first mention of the table or figure in the text (see 3.6, Citation). All references in an article should appear in the reference list.

Note that references cited at the end of table titles are ambiguous. Instead, a footnote should be added to explain the source of the data.

Adapted from. . .

Reproduced with permission from. . .

Data were derived from. . .

When both a footnote letter and reference number follow data in a table, set the reference number first, followed by a comma and the footnote letter (see 3.6, Citation):

427 Patients5,b

4.1.5 Units of Measure.

The JAMA Network journals report laboratory values in conventional units (see 13.12, Units of Measure, and 17.0, Units of Measure). In tables, units of measure, including the variability of the measurement, if reported, should follow a comma in the column or row heading:

Age, mean (SD), y

Systolic blood pressure, mean (SD), mm Hg

Body mass index, median (IQR)

Duration of hypertension, mean (SD) [range], y

Change in rate, % (95% CI)

The JAMA Network journals use a conversion footnote to indicate how to convert values to the SI or another system (Table 4.1-13). See 17.5, Conventional Units and SI Units in JAMA Network Journals.

Table 4.1-13. The Factor for Converting Total Serum Bilirubin Values to SI Units Is Provided in a Separate Footnote

Table. Risk for CP Associated With Varying Elevated TSB Levels


No. of infants

Absolute CP risk, %

Relative risk (95% CI)

Absolute risk difference (95% CI), %

CP

Exposed

Elevation of TSB level >ETT, mg/dL

 <0

86

104 716

0.1

1 [Reference]

[Reference]

 0-4.9

4

1705

0.2

2.9 (1.0 to 7.8)

0.2 (0 to 0.5)

 5.0-9.9

1

102

1.0

11.9 (1.7 to 84.9)

0.9 (0.1 to 5.3)

 ≥10.0

2

26

7.7

93.7 (24 to 361)

7.6 (2.1 to 24.1)

Peak TSB level, mg/dL

 <20.0

87

103 271

0.1

1 [Reference]

[Reference]

 20.0-24.9

2

2772

0.1

0.9 (0.2 to 3.5)

0 (−0.1 to 0.2)

 25.0-29.9

2

459

0.4

5.2 (1.3 to 20.9)

0.4 (0 to 1.5)

 30.0-34.9

0

32

0.0

NA

−0.1 (−0.1 to 10.6)

 ≥35.0

2

15

13.3

158.0 (43.0 to 585.0)

13.2 (3.7 to 37.8)

Abbreviations: CP, cerebral palsy; ETT, exchange transfusion threshold; NA, not applicable; TSB, total serum bilirubin.

SI conversion factor: To convert TSB to µmol/L, multiply by 17.104.

4.1.6 Punctuation.

As with numbers and abbreviations, rules for punctuation may be less restrictive in tables to save space (see 8.0, Punctuation). For example, virgules may be used to present dates (eg, 4/02/17 for April 2, 2017) and hyphens may be used to present ranges (eg, 60-90 for 60 to 90) (see 18.0, Numbers and Percentages). However, when virgules are used to present dates, consider that some international readers place the day before the month in such constructions, causing 4/02/17 to be interpreted as February 4, 2017, instead of April 2, 2017. Also, if any range in a table includes a negative number, the word to should replace the hyphen in all ranges in that row or column to avoid a hyphen followed by a minus sign as well as for consistent presentation within the table (eg, −13.7 to −4.3) (Table 4.1-8 and Table 4.1-9). Phrases and sentences in tables may use end punctuation if required for readability (eg, if cells contain multisentence entries).

4.1.7 Abbreviations.

Within the body of the table and in column and row headings, units of measure and numbers normally spelled out may be abbreviated for space considerations (see 13.12, Abbreviations, Units of Measure; 17.0, Units of Measure; and 18.0, Numbers and Percentages). However, spelled-out words should not be combined with abbreviations for units of measure. For example, “First Week” or “1st wk” or “Week 1” may be used as a column heading, but not “First wk.” Abbreviations or acronyms (but not abbreviations used to indicate units of measure) should be explained in a footnote (see 4.1.4.10, Table Components, Footnotes).

4.1.8 Numbers.

Additional digits (including zeros) should not be added (eg, after the decimal point) to provide all data entries with the same number of digits. Doing so may indicate more precise results than actually were calculated or measured. A percentage or decimal quotient should contain no more than the number of digits in the denominator. For example, the percentage for the proportion 9 of 28 should be reported as 32% (or decimal quotient 0.32), not 32.1% (or 0.321) (see 19.4, Significant Digits and Rounding Numbers). Values reporting laboratory data should be provided and rounded, if appropriate, according to the number of digits that reflects the precision of the reported results to eliminate reporting results beyond the sensitivity of the procedure performed (see 17.4.1, Expressing Quantities).

Values for reporting statistical data, such as P values and CIs, also should be presented and rounded appropriately (see 19.4, Significant Digits and Rounding Numbers). Although some publications3(p512) suggest use of specific designations for levels of significance (eg, a single asterisk in the table to denote values for entries with P< .05, 2 asterisks for P < .01), exact P values are preferred, regardless of statistical significance. In most cases, P values should be expressed to 2 digits to the right of the decimal point unless the first 2 digits are zeros, in which case 3 digits to the right of the decimal place should be provided (eg, instead of P < .01, report as P = .002). However, values close to .05 may be reported to 3 decimal places because the .05 is an arbitrary cut point for statistical significance (eg, P = .053, P = .047). P values less than .001 should be designated as “P < .001” rather than exact values (eg, P = .00006). For some studies, it is important to express P values to more significant digits, such as genome-wide association studies, studies that involve Bonferroni correction, and other types of studies with adjustments for multiple comparisons, and when the level of significance is defined as substantially less than P < .05. For very small numbers, scientific notation may be used to express the value (eg, a genome-wide significant association of P = 1.1 × 10−8) (see 19.5, Glossary of Statistical Terms, P value).

For study outcomes, individual statistically significant values should not be expressed as “P < .05” either in the table or in the table footnote, and nonsignificant P values should not be expressed as “NS” (not significant). For CIs, the number of digits should correspond to the number of digits in the point estimate. For instance, for an odds ratio reported as 2.45, the 95% CI should be reported as 1.32 to 4.78, not as 1.322 to 4.784.

4.1.9 Tables That Contain Online-Only Supplementary Information.

Tables that contain important supplementary information that is too extensive to be published in the journal article may be made available from other sources, including the journal’s website, or other means (eg, online database, institutional website). Online-only supplementary tables posted with an article on a journal’s website should undergo review because they are considered part of the article’s content.5,6

4.1.10 Guidelines for Preparing and Submitting Tables.

Authors submitting tables in a scientific article should consult the publication’s instructions for authors for specific requirements and preferences regarding table format. Although details about preferred table construction vary among journals, several general guidelines apply. Each table should be created using the table functionality in the word processing software or spreadsheet program and inserted in the electronic manuscript file. Reduced type should not be used. For most journals, if a table is too large to be contained on 1 manuscript page, the table should be continued on another page with a “continued” line after the title on the subsequent page. Alternatively, if the table is large or exceedingly complex, the author should consider separating the data into 2 or more simpler tables. Tables should not be submitted on oversized paper, as a graphic image, or as photographic prints.

An example of table creation instructions are provided for authors submitting manuscripts to the JAMA Network journals (Box 4.1-2).

Box 4.1-2. Author Instructions for Table Creation

Creating the table

Use the table editor of the word processing software to build a table. Regardless of which program is used, each piece of data needs to be contained in its own cell in the table.

Avoid creating tables using spaces or tabs. Such tables must be retyped during the editing process, creating delays and opportunities for error. Do not try to align cells with hard returns or extra spaces. Similarly, no cell should contain a hard return or tab. Although individual empty cells are acceptable in a table, be sure there are no empty columns.

Each row of data must be in a separate row of cells:

Table 1. Title

Treatment

Group A

Group B

Medical

500

510

Surgical

500

490

Note that percentages are presented in the same cell as numbers and measures of variability are in the same cell as their corresponding statistic:

Table 2. Title

Characteristic

Group A (n = 50)

Group B (n = 50)

Relative risk (95% CI)

Women, No. (%)

25 (50)

20 (40)

1.25 (1.11-1.57)

Age, mean (SD), y

35 (8)

37 (7)

0.98 (0.92-1.05)

To indicate data that span more than 1 row, merge the cells vertically and use centered vertical alignment in the merged cell as shown in the example below.

Table 3. Title

Age, y

Blood pressure, mm Hg

P value

18-34

120/75

.08

35-50

110/80

51-80

125/82