Format

Send to

Choose Destination
Obstet Gynecol. 1997 Mar;89(3):351-4.

Treatment of continuous data as categoric variables in Obstetrics and Gynecology.

Author information

1
New York University School of Medicine, New York, USA.

Abstract

OBJECTIVE:

To assess the treatment of continuous data in a sample of obstetric and gynecologic literature.

METHODS:

We reviewed articles in Obstetrics and Gynecology published in the first 4 months of 1985 and 1995. Data were tabulated on a data form created specifically for this purpose and reviewed for accuracy.

RESULTS:

The sample set included 170 variables in 102 original articles from Obstetrics and Gynecology published from January to April 1995, inclusive (group A, contemporary articles), and 117 variables in 89 articles published between January and April 1985, inclusive (group B, historic articles). Fifty-three variables (31% of total variables) in group A and 27 variables (23% of total variables) in group B (chi 2, P = .05) were continuous predictor variables. The historic-period articles (63%) were significantly more likely to represent continuous data only as categoric variables than were articles in the contemporary period (38%) (Fisher exact test, P = .037). Correlation coefficients, r values, were provided where possible in ten articles in the contemporary period (83%) and four articles in the historic period (31%) (Fisher exact test, P = .008). In articles in which continuous predictors were treated only as categoric variables, an emphasis was placed on the findings based only on categories in four of 12 (33%) of these articles in 1995 and nine of 13 (69%) in 1985 (Fisher exact test, P = .05).

CONCLUSIONS:

The treatment of continuous data has improved over the time period reviewed. However, clinicians should be aware that continuous data may be mischaracterized as categoric variables in some journal articles. We hope that in the future, editors will consider requesting r values for all continuous data relations. The quality-of-care implications of using discrete cutoffs of continuous data for patient care should be investigated.

PMID:
9052583
DOI:
10.1016/S0029-7844(96)00504-2
[Indexed for MEDLINE]

Supplemental Content

Loading ...
Support Center