Access to data and the information explosion: oral contraceptives and risk of cancer

Abstract

This report describes a technically feasible method to deal with the enormous volume of literature published regarding oral contraceptives. This subject was discussed in 3735 publications during the period from November 1977 through October 1980. Our findings showed that those papers reporting original, numeric relationships involving risk of cancer and use of oral contraceptives were one percent of the total 3735 publications. However, only seven of the 27 articles involved were authored by individuals from departments of obstetrics and gynecology. Further, only four of the 27 were published in journals devoted to obstetrics and gynecology. The analyses suggest a form of censorship taking place, in that the obstetrics/gynecology specialists do not have ready access to the primary data.

PIP: Effective decision making necessitates access to pertinent data. Because of information explosion, however, the mechanisms used in the literature review process can overwhelm a researcher or a practitioner. This paper describes a technically feasible method for dealing with a current large volume situation using the subjects oral contraceptives and risk of cancer as examples. The National Library of Medicine's MEDLARS system identified 3735 citations covering the period November 1977 through October 1980 at the time of the survey. It was found that papers reporting original numeric relationships involving risk of cancer and use of oral contraceptives were 1% of the 3735 publications and only 7 of 27 articles involved were authored by individuals from depts. of obstetrics and gynecology. Only 4 of the 27 were published in the obstetric/gynecology journals, suggesting that obstetricians/gynecologists are not having ready access to primary data. The authors therefore recommended an information processing technique which separates the analysis into 2 parts: 1) the mechanical, performed by the technician; and 2) the cognitive, performed by the professional. The technician examines the cited article and classifies it relative to the presence of numeric displays, originality of work reported, and inclusion of data of interest to the professional. The cognitive functions are left to the professionals. This process provides relevant articles and identifies verifiable reasons for excluding others. It is also effective in terms of cost and time. In this survey, the retrieval; analysis; and photocopying took 14 days or 112 hours. In contrast, the professional time traditionally spent in reviewing citations before the retrieval process begins is about 7 hours (given 741 citations containing 250,000 words and reading time of 600 words per/minute. Also, the accuracy of selections using the title; abstract; and indexing terms is questionable given the variable nature of matching author's and user's purposes and the presence of specific indexing terms and the relationships of interest to the user. The literature search and analysis described in this paper is particularly effective in isolating articles with relevant data directed to different medical groups.