Skip to main content
NCBI home page
Journal of the Medical Library Association : JMLA logoLink to Journal of the Medical Library Association : JMLA
. 2002 Oct;90(4):380–392.

Cohort studies in health sciences librarianship

Jonathan Eldredge 1
PMCID: PMC128954  PMID: 12398244

Abstract

Question: What are the key characteristics of the cohort study design and its varied applications, and how can this research design be utilized in health sciences librarianship?

Data Sources: The health, social, behavioral, biological, library, earth, and management sciences literatures were used as sources.

Study Selection: All fields except for health sciences librarianship were scanned topically for either well-known or diverse applications of the cohort design. The health sciences library literature available to the author principally for the years 1990 to 2000, supplemented by papers or posters presented at annual meetings of the Medical Library Association.

Data Extraction: A narrative review for the health, social, behavioral, biological, earth, and management sciences literatures and a systematic review for health sciences librarianship literature for the years 1990 to 2000, with three exceptions, were conducted. The author conducted principally a manual search of the health sciences librarianship literature for the years 1990 to 2000 as part of this systematic review.

Main Results: The cohort design has been applied to answer a wide array of theoretical or practical research questions in the health, social, behavioral, biological, and management sciences. Health sciences librarianship also offers several major applications of the cohort design.

Conclusion: The cohort design has great potential for answering research questions in the field of health sciences librarianship, particularly evidence-based librarianship (EBL), although that potential has not been fully explored.

A young obstetrician took a position at the equivalent of a county hospital a number of years ago. Not long after beginning his new job in one of the two maternity wards at the hospital, the physician noticed that patients in his particular ward were contracting a mysterious infection at a rate of 11.4%. Yet, patients in the other maternity ward only contracted the infection at a rate of 2.7%. Four years prior to his arrival, the annual infection rate in his ward actually had been nearly 16%. The young obstetrician could find no obvious reason for this one difference between the wards, because conditions in both wards were essentially the same. Except for one possible factor: in his ward, only physicians and medical students delivered babies, whereas only midwives and nurse midwifery students delivered babies in the other ward. The obstetrician attempted to equalize conditions between the two wards in every conceivable way. He even instructed the physicians and medical students in his ward to imitate the seemingly gentler delivery methods used in the midwifery ward. Physicians both at his hospital and at other hospitals where outbreaks of the same infection had occurred had described their experiences in case reports, but none were able to understand what caused these infections.

While still grappling with this seemingly unsolvable problem a year later, one of the obstetrician's physician friends contracted the identical infection. This physician friend had accidentally cut himself with a scalpel while working on a cadaver during an autopsy, which led to his developing the same infection. The obstetrician soon recognized that the physicians and medical students from his maternity ward, who had been working on cadavers elsewhere in the hospital during slow periods in the maternity ward, had inadequately washed their hands before delivering babies. These physicians and medical students consequently were unintentionally infecting the postpartum women. In contrast, the midwives and nurse midwifery students rarely came into contact with cadaveric material.

The obstetrician, Ignaz Semmelweis, began work at the Vienna Lying-in Hospital in 1846. The clinical research method that he employed, now known as the cohort study design, enabled him to detect the cause of the frequently fatal puerpural (childbed) fever in his maternity patients. This brief biographical sketch [1–3] illustrates the effectiveness of the cohort study in observing possible causal relationships among variables. All cohort studies contain a population (e.g., pregnant women delivering at the Vienna Lying-in Hospital), an exposure (e.g., cadaveric material), and an outcome (e.g., risk of childbed fever). The cohort design will be defined and described in detail later in this article.

The British physician James Lind initially pioneered the use of the cohort design in 1747, predating Semmelweis by ninety-nine years, while determining the effective treatment of scurvy with citrus fruits in sailors [4]. PCA Louis of France in 1835 published his famous cohort study Recherches sur les effets de la saignée based upon a cohort design that disproved the effectiveness of bleeding patients for diseases such as pneumonia, which had a subsequent influence on both British and U.S. medicine [5]. In the United States, the cohort design first appeared in a study involving a cohort of tuberculosis patients, who had been treated during the years 1885 to 1901 at the Adirondack Cottage Sanitarium in upstate New York [6]. Another early cohort study linked increased dietary animal protein and a cure for pellagra [7]. Epidemiologists later employed the cohort design to detect patterns of familial contagion of tuberculosis during the 1930s in Kingsport, Tennesssee [8]. Beginning in 1949, citizens of Framingham, Massachusetts, were enrolled in a now famous prospective cohort study intended to detect causal relationships between lifestyle and heart disease [9]. Samet and Munoz provide interested readers with an excellent history on the development of cohort studies in epidemiology [10].

The cohort study design continues to be a popular research method in medicine, public health, and other scientific disciplines [11]. One commonly encounters research results from cohort designs reported in the contemporary clinical and public health literatures. Table 1 provides brief descriptions of the defined populations, exposures, and outcomes of health sciences cohort studies. These descriptions illustrate the diversity of research questions in the health sciences that can be addressed by utilizing the cohort study design.

Table 1 Examples of cohort study designs in the health sciences

graphic file with name i0025-7338-090-04-0380-t01.jpg

Open in a new tab

COMPONENTS OF THE COHORT STUDY DESIGN

Figure 1 offers a generic cohort study design with its three major components: (1) a defined population, (2) exposure status, and (3) outcomes. Figure 1 pertains primarily to applying health sciences uses of the cohort design as a vehicle for practicing evidence-based librarianship. The author will explain below how the same basic cohort design has been adapted to the social, behavioral, biological, and earth sciences. Most importantly, this article will demonstrate that the cohort design also has wide potential applicability to evidence-based librarianship.

Figure 1.

Figure 1

Open in a new tab

Generic cohort study

Defined population

All cohort studies begin with a defined cohort. A cohort may be most generally defined as a “band” or a “group,” harking back to the earliest known use of the term meant to designate one out of ten Roman legions [22]. Rothman and Greenland observe that “In epidemiology, the word cohort is often used to designate a group of people who share a common experience or condition” [23]. This group may comprise all of the defined population or simply a sample from that defined population. Regardless of data collection method, no member of the cohort should have the outcome of interest at the chronological beginning of the cohort study. This factor aids later inferences of causality as researchers observe changes over time. The first column in Table 1 lists a wide variety of defined populations studied in the health sciences. Cohorts can be composed of people who belong to any conceivable categorical group including, but not limited to, religious sects, vegetarians, smokers, occupational groups, lower socioeconomic status groups, victims of traumatic injury, adults having experienced adverse childhood events, former cancer or surgery patients, or those living in defined geographic regions [24]. Cohort studies even have been deployed for humorous purposes in mock studies of cohorts consisting of soap opera characters, jazz musicians, and teething infants [25–27].

Outside of the health sciences, cohorts generally are defined more broadly. Glenn, for example, defines a cohort as “Those people within a geographically or otherwise delineated population who experienced the same significant life event within a given period of time” [28]. Ryder defines a cohort even more generically: “A cohort may be defined as the aggregate of individuals (within some population definition) who experienced the same event within the same interval” [29]. Social and behavioral scientists have defined cohorts of every conceivable grouping of people: students, graduates, professionals, married couples, parents, divorced people, twins, blended families, children raised in dysfunctional families, adulterers, prostitutes, substance abusers, and agents of social change to name a few familiar variations. Table 2 lists diverse defined populations who have been studied with the cohort design in the social, behavioral, biological, and earth sciences. A recent dictionary by Bégaud defines a cohort as a “Group of subjects selected according to one or more common characteristic(s) and followed over time in order to identify, describe or quantify an event” [47].

Table 2 Diverse examples of cohort studies from outside the health sciences

graphic file with name i0025-7338-090-04-0380-t02.jpg

Open in a new tab

Cohorts need not be comprised of humans either. Antarctic seals, Atlantic salmon, shellfish, dogs, rabbits, and evergreen trees also can be cohorts. Biological cohort studies tend to focus upon a group of individuals born at the same time, particularly in species with shorter life spans than humans [48–52]. “There are many different sources of cohorts and the most appropriate cohort often depends on the study questions, disease frequency, and study financial resources” observes one author [53]. For this article, it must be noted that a defined population in a cohort need neither be alive nor present for its changes to be measured through surrogates or artifacts, such as fossils [54].

Exposure or non-exposure

Figure 1 utilizes the triangle, as did the ancient Greeks, to symbolize change. In this diagram, the triangle can represent the existence of an exposure or non-exposure. Health sciences cohort designs normally focus upon a single exposure or intervention (i.e., change) that researchers suspect causes an outcome. In the opening medical example of Semmelweis's discovery of the cause of childbed fever, the exposure would be the manual contact of cadaveric tissue with postpartum women by physicians and medical students in one of the two maternity wards. Lind noted the exposure of sailors to citrus fruit as a cure for scurvy [55]. In the health sciences, these exposures or non-exposures can span a range of conditions as diverse as the ionizing radiation from an atomic bomb blast to a prescribed drug treatment to the avoidance of certain unhealthy lifestyles in the case of vegetarians. An exposure can include personal characteristics such as the presence of a psychiatric condition [56]. An exposure can be either an uncontrolled event or a conscious decision to make a medical intervention [57]. Or, an exposure actually can include people's behavior [58]. Cohort studies can measure exposures in different ways: intensity, duration, regularity, or even variability [59]. Different members of a cohort might be exposed over time in varying degrees to the potential cause of the outcomes. For example, not everyone who survived the bombings of Nagasaki or Hiroshima was exposed to identical amounts of ionizing radiation due to either their distance from ground zero or their shielding from the blast by structures. As another example, vegetarians vary in their consumption choices: some might be vegans, while others might engage in potentially risky behaviors such as excessive alcohol consumption or tobacco use.

The social and behavioral sciences, perhaps reflecting the complexity of human activity, tend to focus upon multiple rather than just one identified set of exposures or interventions hypothesized to cause an outcome. Table 2 indicates that the concept of exposure or non-exposure outside the health sciences may be best described in terms of a variable that causes change.

Outcomes

Figure 1 clarifies that not all hypothesized outcomes actually occur in a cohort study. Sometimes the hypotheses, null hypotheses, or alternative hypotheses fail to predict what turns out to be a completely unexpected outcome. A cohort study of soft-shelled clams by Belding, for instance, revealed the surprising discovery that clam beds in polluted waters were “extremely productive” [60]. One cohort study produced the unexpected finding that moderate amounts of alcohol might have beneficial health effects [61]. Another cohort study on the risk of developing Parkinson disease noted the possible preventive effect of caffeine [62]. In other instances, cohort studies yield no observed outcomes.

Semmelweis could have calculated that the relative risk for patients to contract childbed fever in his ward was 4.2 times greater than in the other ward. The relative risk ratio represents a probabilistic statement that helps decision makers in the health sciences, and other disciplines, recognize the chances that a certain outcome might occur [63]. Gordis indicates that in epidemiology, “The essential characteristic in the design of cohort studies is the comparison of outcome(s) in an exposed group and a non-exposed group (or a group with a certain characteristic and a group without that characteristic)” [64]. Samet and Munoz observe that “The dynamic nature of many risk factors and their relation in time to disease occurrence can only be captured in the cohort design” [65]. Semmelweis was able to compare the incidence rate between his ward and the other ward at the Vienna Lying-in Hospital. When cohort studies make comparisons, these comparisons can be within the cohort of interest (intracohort) or against other cohorts (intercohort) [66]. Some suggest that intracohort arrangements are more desirable [67]. Cohort studies sometimes can be combined through either systematic reviews or possibly in a more rigorous meta-analysis to make even greater generalizations [68].

Outside the health sciences, these comparisons between exposed and non-exposed groups and the group or groups with an outcome of interest simply do not seem as important. At least these aspects are not emphasized to the same extent. An outcome needs only to be descriptive to be of interest to the social, behavioral, or biological scientists [69]. As already noted, researchers in the health sciences have been keenly interested in making comparisons between outcomes and then linking these varied outcomes to differences in exposures. Outside the health sciences, interest in these dimensions of cohort studies span a continuum from simply describing differences in outcomes all the way to the aforementioned more rigorous statistical analyses found in the health sciences.

The use of cohort studies apparently arose to meet different needs within various disciplines. In the social sciences, Karl Mannheim has been credited for first recognizing the utility of the cohort study design as early as 1928. The social sciences then began to harness this study design in the 1930s through the 1950s, with further theoretical investigations by Talcott Parsons in 1959. Political scientists found new applications of the cohort design with voting studies in 1954. The popular media also used the cohort design in more casual ways during the 1950s to describe social phenomena [70–74]. The cohort design has wide applicability in the biological sciences due to the interest in studying the “natural history” of species. The author serendipitously, rather than systematically, identified a 1903 publication describing basic growth experiments at Cold Spring Harbor on Long Island as possibly the first known use of the cohort design in biology [75].

Data collection

Cohort studies collect data with either prospective or retrospective approaches, depending upon the sequence in which researchers begin to study the cohort. Prospective cohort studies, sometimes called “concurrent cohort studies,” such as the Framingham Study normally begin to measure relevant indicators of variables prior to an exposure of interest. These measurements continue throughout the study until a certain endpoint. Retrospective cohort studies, sometimes called “historic cohort studies,” identify the cohort, their exposure, and outcomes after an exposure, as follow-up studies.

Longitudinal studies are a specific subtype of cohort study involving multiple measurements of individuals in a cohort, sometimes continuously [76, 77]. Some researchers from the biological and social sciences tend to use the terms “longitudinal studies” and “cohort studies” interchangeably, however. The biological sciences tend to describe cohort studies as longitudinal studies, even when only one baseline and one follow-up measurement of each individual in a cohort occur [78]. This semantic issue has implications for any literature searches across different disciplines, although the Dictionary of Epidemiology [79] and Medical Subject Headings (MeSH) both subsume longitudinal studies categorically under cohort studies.

The cohort design specific to epidemiology, and more generally applicable to other fields, thus may be readily summarized. A cohort design includes members of a defined population who are exposed to a factor of interest over a period of time. These members are compared to members of the same population (or a similar population) who have not been exposed to the factor of interest. The data analysis then compares the outcomes over two or more periods of time between the two or more groups [80].

COHORT DESIGNS IN LIBRARIANSHIP

Librarianship both borrows from and influences other fields [81]. Librarianship employs cohort studies more closely resembling the social sciences models, which tend to describe phenomena rather than make inferences about possible causality. By presenting the diverse examples in the preceding text and in Tables 1 and 2, the author hopes to encourage librarians to explore new applications of the cohort study design to apply this study design to other research questions in librarianship. The author further hopes that some of the examples from the health sciences are particularly inspiring. References in the text and the tables refer readers to the actual reported studies as well as other resources for calculating the statistics appropriate to this design.

Methods for identifying cohort studies

The examples of cohort studies in Tables 1 and 2 represent well-known historic cohort studies or more recent diverse applications of the cohort study design outside librarianship. The author identified many of the health sciences examples by searching PubMed during March 2000, utilizing MeSH terms such as “Cohort Studies,” “Followup studies,” and “prospective studies” plus keywords such as “cohort” or “cohorts.” The author identified most of the cohort studies outside the health sciences by searching FirstSearch during April and May 2000 using keywords like “cohort” or “cohorts.” FirstSearch databases used included Agricola, BIOSIS, EconLit, Education Abstracts, Geobase, Humanities Abstracts, Legal Periodicals, Library Literature, PAIS, Social Sciences Abstracts, and Wilson Business. Some articles listed references to historically significant cohort studies in various fields. This approach represented a less scientific, narrative review literature review methodology [82].

The search strategy for identifying possible cohort studies in librarianship, particularly those from health sciences librarianship, began with searching the FirstSearch Library Literature database and PubMed during April 2000. The author used the aforementioned MeSH terms and keywords in PubMed and the keywords “cohort” and “cohorts” in Library Literature. These approaches were almost completely fruitless. The aforementioned controlled subject headings or keyword approaches used for other subject areas yielded hardly any possible studies in librarianship. The author ultimately had to engage in the kind of handsearching found in systematic reviews to supplement the database searches. The author made no special effort to extend his search beyond the library literature for library and information science articles, even though he already was aware of at least a few relevant articles published in other fields. The author also drew upon papers and posters presented at the 1998 through 2001 Medical Library Association (MLA) annual meetings, the MLA South Central Chapter meetings during 1999 to 2000, and the 2000 Joint Meeting of the Medical Library Group of Southern California and Arizona (MLGSCA) and the Northern California and Nevada Medical Library Group (NCNMLG), for examples. In this respect, the search for diverse illustrative examples of cohort studies in health sciences librarianship followed a systematic review [83] rather than a narrative review approach, although not all identified studies ultimately appeared in this article, because they featured either redundant adaptations or had been described already in a previous article by the author [84].

Researchers reporting in the library literature hardly ever categorize these studies by their formal name of cohort studies [85]. This lack of categorization thereby thwarts efforts to identify studies using standard controlled vocabulary or keyword search strategies. Consequently, the author manually searched the following journals specifically for the noted years: Bulletin of the Medical Library Association (BMLA), 1990 to 2000; Bibliotheca Medica Canadiana (BMC), 1995 to 2000; Health Libraries Review, 1995 to 2000; and Medical Reference Services Quarterly, 1994 to 2000. The restricted local availability of the latter three titles explains the limited coverage for a manual search. Following this systematic review, the author identified more recently published reports through his professional reading in these journals, attendance at MLA 2001, and reading of back and current issues of journals such as Hypothesis.

Collection resources cohort studies

Table 3 summarizes some applications of the cohort study design in health sciences librarianship concerning collection resources. The most common and enduring application of the collection resources cohort design traditionally has been the book or journal usage study. These studies can be found throughout the professional literature and even in library newsletters. Beginning in 1939, Postell pioneered adapting the cohort study design for collection resources use studies [98]. Health sciences librarians at Yale University, the Philadelphia College of Physicians, the Mayo Clinic, the National Library of Medicine, and Columbia University [99–105] also adapted the basic cohort design to answer important collection resources questions during the 1950s and early 1960s.

Table 3 Collection resources cohort studies

graphic file with name i0025-7338-090-04-0380-t03.jpg

Open in a new tab

This usage studies genre of cohort studies endures, because it answers many practical—and sometimes a few theoretical—research questions. Table 3 includes most collection resources cohort studies identified through the author's systematic review. In Table 3, nearly all of the populations under study consist of users of various libraries or information centers. The exposures mainly consist of access to these collection resources. The observable outcome usually consists of the aggregate usage of the resources. In this adaptation, the use measurement becomes the locus for observing actual changes in the user population. Thus, while populations of humans are involved in collection resources cohort studies, their activity becomes recorded in the form of collection resources usage in ways similar to the ways that laboratory results, medical images, and other data become surrogates in health sciences cohort studies. Usage becomes the surrogate measure for changes in the user population without linking individuals to their specific usage data, however. Privacy policies and professional librarians' codes of ethics usually do not allow linking materials used with specific users, so humans are not studied per se, just the artifact of their use, in most of these cohort studies. The collection resources described in Table 3 offer the practical advantage over other types described in Tables 4 and 5 of not normally requiring approval from human subjects research committees, because individuals are not linked to their confidential usage data. All of these examples are meant to convey the wide applicability of the cohort design rather than to offer a complete inventory or to endorse the quality of the presented examples.

Table 4 User population cohort studies: user education studies

graphic file with name i0025-7338-090-04-0380-t04.jpg

Open in a new tab

Table 5 User population cohort studies: information-seeking behavior cohort studies

graphic file with name i0025-7338-090-04-0380-t05.jpg

Open in a new tab

User population cohort studies

Tables 4 and 5 summarize various types of user population cohort studies utilized by health sciences librarianship to answer research questions. Within this major categorization, Tables 4 and 5 distinguish between two subcategories: user education cohort studies and information-seeking behavior cohort studies. User education cohort studies typically involve a population of students or professionals acquiring additional library or informatics skills. The exposure becomes training of the population by librarians or other experts. The outcome or outcomes describes the post-exposure changes in trainees' library or informatics skills. The information-seeking behavior cohort studies outlined in Table 5 have diverse user populations. The exposures in Table 5 range from the proximity of users to a health sciences library to the running of mediated MEDLINE searches to personality profiles of potential users. The outcomes also vary due to the diverse exposures suspected of causing the changes. The cohort studies in Tables 4 and 5 either link user changes directly at the individual level of analysis or employ surrogates for these changes as would be typically found in the cohort studies found in Table 3.

Other cohort studies in librarianship

Tables 3, 4, and 5 exclude at least one additional health sciences librarianship cohort study, which simply does not fit into the two major typologies. This study further suggests the potential diversity of applications of the cohort study design. This retrospective cohort study by Newcomer and Piscotti examines the career paths of academic health sciences directors [132]. Such career path cohort studies also exist in other fields [133].

STRENGTHS AND WEAKNESSES OF COHORT STUDIES

By now, readers can recognize that cohort studies are adaptable to studying a great variety of circumstances involving observable change. The basic orientation of a cohort study can range from simply descriptive to actually aiding predictions about the future. This article has shown that cohort studies moreover can be combined with other methodologies such as surveys, which may be a more familiar research method for many librarians. Finally, cohort designs are capable of addressing important questions involving probabilities of observed outcomes, such as projected use of collection resources or use of acquired library or informatics skills, that face the profession [134] as librarians enter an era of evidence-based librarianship (EBL). In the EBL hierarchy of evidence for decision making, cohort studies occupy one of the highest levels, just below randomized controlled trials (RCTs). Under most circumstances, a well-constructed cohort study simply has greater validity and probably greater reliability than even a series of well-done case studies.

Cohort studies, as observational studies, can be used in situations in which it would be unethical to subject participants either to harmful exposures or to deny beneficial exposures. RCTs, while offering the advantages of confirmatory or experimental research approaches, simply may be unethical under certain circumstances. Librarianship normally does not involve high-stakes situations in which harm could be dramatically caused by an exposure such as may be found in medicine. Yet, ethical issues do arise in the profession about denying a potentially beneficial exposure (to a library resource or a service) that may best be resolved with a cohort study instead of an RCT. Cohort studies also observe changes in naturally occurring circumstances, which lends a practical as well as ethical advantage over RCTs. Of course, the EBL question should drive the choice of best research design.

Cohort studies provide great versatility across different exposures, which might produce either dramatic or subtle outcomes [135]. The historic cohort study examples of both Semmelweiss and Lind yielded dramatic results. Semmelweiss's obstetrics patients stopped dying at a terrifying rate from childbed fever, and Lind's sailors nearly all made dramatic recoveries from scurvy. Cohort studies also can measure far more subtle and long-term effects without too much trouble from confounding such as in the Framingham Study or the other varied examples listed in Table 1. Established statistical methods are available for analyzing results in either the dramatic or the more subtle instances [136, 137].

The most controversial argument for the strength of cohort studies resulted from an actual comparison study reported in the New England Journal of Medicine during 2000. This study by Benson and Hartz suggested that, from their review, cohort and RCTs produced similar results in seventeen of the nineteen health conditions. Their suggestion was vigorously disputed by Pocock and Elbourne in the same issue, although both Moses and Feinstein previously had pointed to the realistic use of cohort studies [138–141] in place of RCTs. Additionally, psychological evidence indicated that “Experimenter Expectancy Effects” might lead scientific observers conducting RCTs as well as cohort studies to mistakenly perceive anticipated outcomes in study situations in which actual events moved in radically different directions. Two authors reviewing this impressive body of evidence have written that “The result of hundreds of studies demonstrate that experimenters find what they expect to find.” This body of evidence contributed to the implementation of double blinding [142].

The principal problems with the cohort design revolve around the possible introduction of bias in these kinds of studies. Pocock and Elbourne make this point when they note that cohort studies do not prove causality; they only suggest probabilities of causality [143]. Many other researchers have made similar points. Cohort studies do not randomly assign populations into exposure status so that some members of the population receive no exposure, while others will receive some variation of exposure as done in RCTs. This leads to the possibility of population members either self-selecting exposures on their own or haphazardly becoming exposed (or non-exposed) in ways that bias study outcomes. RCTs obviously are not immune from various forms of bias, either. In general, cohort studies seem to be more susceptible to either the aforementioned selection bias or confounding bias. Confounding bias occurs when variables other than the exposure identified in the cohort design really cause the outcome. Library science researchers can avoid confounding bias, in part, by articulating many diverse alternative hypotheses at the outset when designing their research to anticipate the existence of other explanatory variables. The many forms of bias need not be summarized here, but readers need to be keenly aware that no study design, whether cohort or RCT, can be immune to the many forms of bias. Interested readers are encouraged to refer to two chapters in the Encyclopedia of Epidemiologic Methods for concise discussions of bias relevant to cohort studies [144, 145].

CONCLUSION

Most library science examples described in this article do not match the methodological rigor of cohort studies utilized in contemporary clinical medicine. Cohort studies in health sciences librarianship more closely resemble studies in the biological, behavioral, or social sciences disciplines in most instances. The presence of a study here furthermore does not signify an endorsement of its validity. Yet, librarians have successfully adapted cohort study designs to solve practical problems, frequently with limited resources, to improve their collection resources or services. This article provides numerous references to guide interested readers to specific examples of cohort studies or to methodological discussions concerning cohort studies. This article and the cohort studies it references should inspire health sciences librarians to apply the cohort design to both familiar and novel adaptations. The cohort design offers librarians a powerful tool to improve their libraries.

Acknowledgments

The author wishes to thank Thomas Becker, M.D., Ph.D., and Kristine Tollestrup, Ph.D., for their explanations of the subtle intricacies of health sciences cohort designs. The author also appreciates David Bennahum, M.D., for alerting him to the first known cohort study carried out by Dr. James Lind.

REFERENCES

  1. Speert H. Obstetric and gynecologic milestones, illustrated. New York, NY: Parthenon Publishing Group, 1996:377–87. [Google Scholar]
  2. Carter KC, Carter BR. Childbed fever: a scientific biography of Ignaz Semmelweis. Westport, CT: Greenwood Press, 1994:23–56. [Google Scholar]
  3. Thoms H. Classical contributions to obstetrics and gynecology. Springfield, IL: Charles C. Thomas, 1935:184–90. [Google Scholar]
  4. Lind J. A treatise on the scurvy. in three parts. containing an inquiry into the nature, causes, and cure, of that disease. 3d ed. enlarged and improved. London, U.K.: S Crowder, D Wilson, G Nicholls, T Cadell, T Becket and Co, 1777:149–67. (Reprinted 1980 Birmingham, AL: The Classics of Medicine Library.). [Google Scholar]
  5. Bull JP. The historical development of clinical therapeutic trials. J Chron Dis. 1959 Sep; 10(3):218–48. [DOI] [PubMed] [Google Scholar]
  6. Brown L, Pope EG. The postdischarge mortality among the patients of the Adirondack Cottage Sanitarium. Am Med. 1904 Nov 19; 8(21):879–82. [Google Scholar]
  7. Goldberger J, Waring CH, and Tanner WF. Pellagra prevention by diet among institutional inmates. Public Health Rep. 1923 Oct 12; 38(41):2361–8. [Google Scholar]
  8. Frost WH. Risk of persons in familial contact with pulmonary tuberculosis. Am J Public Health. 1933 May; 23(5):426–32. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Dawber TR. The Framingham Study: the epidemiology of atherosclerotic disease. Cambridge, MA: Harvard University Press, 1980. [Google Scholar]
  10. Samet JM, Munoz A.. Evolution of the cohort study. Epidemiol Rev. 1998;20(1):1–14. doi: 10.1093/oxfordjournals.epirev.a017964. [DOI] [PubMed] [Google Scholar]
  11. Benson K, Hartz AJ. A comparison of observational studies and randomized, controlled trials. N Engl J Med. 2000 Jun 22; 342(25):1878–86. [DOI] [PubMed] [Google Scholar]
  12. Speert H. Obstetric and gynecologic milestones, illustrated. New York, NY: Parthenon Publishing Group, 1996:377–87. [Google Scholar]
  13. Schull WJ. Effects of atomic radiation: a half-century of studies from Hiroshima and Nagasaki. New York, NY: Wiley-Liss, 1995. [Google Scholar]
  14. Dawber TR. The Framingham Study: the epidemiology of atherosclerotic disease. Cambridge, MA: Harvard University Press, 1980. [Google Scholar]
  15. Gennaro S, Medoff-Cooper B, and Lotas M. Perinatal factors and infant temperament: a collaborative approach. Nurs Res. 1992 Nov/Dec; 41(6):375–7. [PubMed] [Google Scholar]
  16. Brown AS, Van Os J, Driessens C, Hoek HW, and Susser ES. Prenatal famine and the spectrum of psychosis. Psychiatr Ann. 1999 Mar; 29(3):145–50. [Google Scholar]
  17. Notzer N, Soffer S. Determination in rejected medical school applicants: a 10-year follow-up. Med Educ. 1987 Sep; 21(5):405–9. [DOI] [PubMed] [Google Scholar]
  18. Kant AK, Schatzkin A, Graubard BI, and Schairer C. A prospective study of diet quality and mortality in women. JAMA. 2000 Apr 26; 283(16):2109–15. [DOI] [PubMed] [Google Scholar]
  19. Colditz GA, Willett WC, Hunter DJ, Stampfer MJ, Manson JE, Hennekens CH, and Rosner BA. Family history, age, and risk of breast cancer: prospective data from the Nurses' Health Study. JAMA. 1993 Jul 21; 270(13):338–43. [PubMed] [Google Scholar]
  20. Niedhammer I, Goldberg M, Leclerc A, David S, Bugel I, and Landre MF. Psychosocial work environment and cardiovascular risk factors in an occupational cohort in France. J Epidemiol Community Health. 1998 Feb; 52(2):93–100. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Lantz PM, House JS, Lepkowski JM, Williams DR, Mero RP, and Chen J. Socioeconomic factors, health behaviors, and mortality: results from a nationally representative prospective study of US adults. JAMA. 1998 Jun 3; 279(21):1703–8. [DOI] [PubMed] [Google Scholar]
  22. Gove PB. ed. Webster's third new international dictionary. Springfield, MA: Merriam-Webster, 1993:441. [Google Scholar]
  23. Rothman KJ, Greenland S. Cohort studies. In: Rothman KJ, Greenland S, eds. Modern epidemiology. 2d ed. Philadelphia, PA: Lippincott-Raven, 1998:79–91. [Google Scholar]
  24. Lynch JW, Everson SA, Kaplan GA, Salonen R, and Salonen JT. Does low socioeconomic status potentiate the effects of heightened cardiovascular responses to stress on the progression of carotid atherosclerosis? Amer J Public Health. 1998 Mar; 88(3):389–94. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Crayford T, Hooper R, and Evans S. Death rates of characters in soap operas on British television: is a government health warning required? BMJ. 1997 Dec 20–27; 315(7123):1649–52. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Kinra S, Okasha M. Unsafe sax: cohort study of the impact of too much sax on the mortality of famous jazz musicians. BMJ. 1999 Dec 18–25; 319(7225):1612–3. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Bennett HJ, Brudno DS. The teething virus. Pediatr Infect Dis. 1986 Jul–Aug; 5(4):399–401. [DOI] [PubMed] [Google Scholar]
  28. Glenn ND. Cohort analysis. Beverly Hills, CA: Sage Publications, 1977:8. [Google Scholar]
  29. Ryder NB. The cohort as a concept in the study of social change. Am Sociol Rev. 1965 Dec; 30(6):843–61. [PubMed] [Google Scholar]
  30. Dempsey D. First novelists, last words. Saturday Review. 1963 Oct 12; 46(41):34. [Google Scholar]
  31. Enzer HA. The American first novelist: a study of commitment and the literary career. Diss Abstr. 1964 Apr; 24(10):4304. [Google Scholar]
  32. Clark DL. Extinction of conodonts. J Paleontol. 1983 Jul; 57(4):652–61. [Google Scholar]
  33. Nye MA. Party support in the House of Representatives: generational replacement, seniority, or member conversion? Am Polit Q. 1994 Apr; 22(2):175–89. [Google Scholar]
  34. Glenn ND. Television watching, newspaper reading, and cohort differences in verbal ability. Sociology of Education. 1994 Jul; 67(3):216–30. [Google Scholar]
  35. Brewer DJ, Eide E, and Ehrenberg RG. Does it pay to attend an elite private college? cross cohort evidence on the effects of college quality on earnings. Cambridge, MA: National Bureau of Economic Research, 1996. [Google Scholar]
  36. Morgan LA. Glass-ceiling effect or cohort effect? a longitudinal study of the gender earnings gap for engineers, 1982 to 1989. Am Sociol Rev. 1998 Aug; 63(4):479–83. [Google Scholar]
  37. Bonzi S, Day DL.. Faculty productivity as a function of cohort group, discipline, and academic age. ASIS '91: Proceedings of the 54th ASIS annual meeting. 1991;28:267–75. [Google Scholar]
  38. Bonzi S.. Trends in research productivity among senior faculty. Information Processing & Management. 1992;28(1):111–20. [Google Scholar]
  39. Otta E, Queiroz RS, Campos LS, Dasilva MWD, and Silveira MT. Age differences between spouses in a Brazilian marriage sample. Evol Hum Behav. 1999 Mar; 20(2):99–103. [Google Scholar]
  40. Little MA. The impact of the custody plan on the family: a five-year follow-up: executive summary. Family & Conciliation Courts Review. 1992 Apr; 30(2):243–51. [Google Scholar]
  41. Belding DL.. The cause of the high mortality in the Atlantic salmon after spawning. Trans Amer Fish Soc. 1934;64(2):219–22. [Google Scholar]
  42. Gerrish G, Mueller-Dombois D. Measuring stem growth rates for determining age and cohort analysis of a tropical evergreen tree. Pacific Science. 1999 Oct; 53(4):418–29. [Google Scholar]
  43. Alwin DF, Cohen RL, and Newcomb TM. Political attitudes over the life span: the Bennington women after fifty years. Madison, WI: University of Wisconsin Press, 1991. [Google Scholar]
  44. Hackman EM, Moe GL. Evaluation of newspaper reports of nutrition-related research. J Am Diet Assoc. 1999 Dec; 99(12):1564–6. [DOI] [PubMed] [Google Scholar]
  45. Brewer DD, Potterat JJ, Garrett SB, Muth SQ, Roberts JM, Kasprzyk D, Montano DE, and Darrow WW. Prostitution and the sex discrepancy in reported number of sex partners. Proc Natl Acad Sci. 2000 Oct 24; 97(22):12385–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Werner EE. Overcoming the odds. Developmental and Behavioral Pediatrics. 1994 Apr; 15(2):131–6. [PubMed] [Google Scholar]
  47. Bégaud B. Dictionary of pharmacoepidemiology. New York, NY: John Wiley & Sons, 2000:36. [Google Scholar]
  48. Belding DL, Pender MJ, and Rodd JA. The early growth of salmon parr in Canadian hatcheries. Trans Amer Fish Soc. 1932 Sep; 62(2):211–23. [Google Scholar]
  49. Massachusetts Department of Fisheries and Game. A report upon the quahaug and oyster fisheries of Massachusetts. Boston, MA: State of Massachusetts, 1912. [Google Scholar]
  50. Tanner JT. Guide to the study of animal populations. Knoxville, TN: University of Tennessee, 1978:81. [Google Scholar]
  51. Elseth GD, Baumgardner KD. Population biology. New York, NY: Van Nostrand, 1981:271. [Google Scholar]
  52. Hedrick PW. Population biology: the evolution and ecology of populations. Boston, MA: Jones and Bartlett, 1984:201. [Google Scholar]
  53. Cudkowicz ME. Epidemiologic methods in neurology. In: Batchelor T, Cudkowicz ME, eds. Principles of neuroepidemiology. Woburn, MA: Butterworth Heinemann, 2001:3. [Google Scholar]
  54. Clark DL. Extinction of conodonts. J Paleontol. 1983 Jul; 57(4):652–61. [Google Scholar]
  55. Lind J. A treatise on the scurvy. in three parts. containing an inquiry into the nature, causes, and cure, of that disease. 3d ed. enlarged and improved. London, U.K.: S Crowder, D Wilson, G Nicholls, T Cadell, T Becket and Co, 1777:149–67. (Reprinted 1980 Birmingham, AL: The Classics of Medicine Library.). [Google Scholar]
  56. Daly LE, Bourke GJ. Interpretation and uses of medical statistics. 5th ed. Oxford, U.K.: Blackwell Science, 2000:146. [Google Scholar]
  57. Herbst AL, Poskanzer DC, Robboy SJ, Friedlander L, and Scully RE. Prenatal exposure to stilbestrol: a prospective comparison of exposed female offspring with unexposed controls. N Eng J Med. 1975 Feb 13; 292(7):334–9. [DOI] [PubMed] [Google Scholar]
  58. Daly LE, Bourke GJ. Interpretation and uses of medical statistics. 5th ed. Oxford, U.K.: Blackwell Science, 2000:146. [Google Scholar]
  59. Greenberg RS, Daniels SR, Flanders WD, Eley JW, and Boring JR. Medical epidemiology. 3d ed. New York, NY: Lange Medical Books, 2001:119. [Google Scholar]
  60. Belding DL. The soft-shelled clam fishery of Massachusetts: including the natural history of the soft-shelled clam and a discussion of sewage pollution and shellfish. Boston, MA: Commonwealth of Massachusetts Division of Fisheries and Game, 1930:47. [Google Scholar]
  61. Doll R, Peto R, Hall E, Wheatley K, and Gray R. Mortality in relation to smoking: 40 years' observations on male British doctors. BMJ. 1994 Oct 8; 309(6959):901–11. [DOI] [PMC free article] [PubMed] [Google Scholar]
  62. Ross GW, Abbott RD, Petrovitch H, Morens DM, Grandinetti A, Tung KH, Tanner CM, Masaki KH, Blanchette PL, Curb JD, Popper JS, and White LR. Association of coffee and caffeine intake with the risk of Parkinson Disease. JAMA. 2000 May 24/31; 283(20):2674–9. [DOI] [PubMed] [Google Scholar]
  63. Zhang J, Yu KF. What's the relative risk? a method of correcting the odds ratio in cohort studies of common outcomes. JAMA. 1998 Nov 18; 280(19):1690–1. [DOI] [PubMed] [Google Scholar]
  64. Gordis L. Epidemiology. 2d ed. Philadelphia, PA: WB Saunders, 2000:131–9. [Google Scholar]
  65. Samet JM, Munoz A.. Evolution of the cohort study. Epidemiol Rev. 1998;20(1):1–14. doi: 10.1093/oxfordjournals.epirev.a017964. [DOI] [PubMed] [Google Scholar]
  66. Ryder NB. Cohort analysis. In: Sills DL, ed. International encyclopedia of the social sciences. v. 2. New York, NY: The Macmillan Company & The Free Press, 1968:546–50. [Google Scholar]
  67. Callas PW, Pastides H, and Hosmer DW. Survey of methods and statistical models used in the analysis of occupational cohort studies. Occup Environ Med. 1994 Oct; 51(10):649–55. [DOI] [PMC free article] [PubMed] [Google Scholar]
  68. Key TJ, Fraser GE, Thorogood M, Appleby PN, Beral V, Reeves G, Burr ML, Chang-Claude J, Frentzel-Beyme R, Kuzma JW, Mann J, and McPherson K. Mortality of vegetarians and nonvegetarians: detailed findings from a collaborative analysis of 5 prospective studies. Am J Clin Nutr. 1999 Sep; 70(3):516S–524S. [DOI] [PubMed] [Google Scholar]
  69. Armstrong LE, Deluca JP, Christensen EL, and Hubbard RW. Mass-to-surface area index in a large cohort. Am J Phys Anthropol. 1990 Nov; 83:321–9. [DOI] [PubMed] [Google Scholar]
  70. Mannheim K. The problem of generations. In: Mannheim K, ed. Essays on the sociology of knowledge. New York, NY: Oxford University Press, 1952:276–322. [Google Scholar]
  71. Alwin DF, Cohen RL, and Newcomb TM. Political attitudes over the life span: the Bennington women after fifty years. Madison, WI: University of Wisconsin Press, 1991. [Google Scholar]
  72. Parsons T. The school class as a social system: some of its functions in American society. Harvard Educational Review. 1959 Fall; 29(4):297–318. [Google Scholar]
  73. Lipset SM, Lazarsfeld PF, Barton AH, and Linz J. The psychology of voting: an analysis of political behavior. In: Lindzey G, ed. Handbook of social psychology. v. 2. Cambridge, MA: Addison-Wesley, 1954:1124–75. [Google Scholar]
  74. Bracker M. There's no class like the class of '29. New York Times Magazine. 1954 May 23; 103(35,183):14– 59–60. [Google Scholar]
  75. Kellogg JL.. Feeding habits and growth of venus mercenaria. Mus Bull—NY State Mus. 1903;10(71):3–32. [Google Scholar]
  76. Halloran ME. Overview of study design. In: Thomas JC, Weber DJ, eds. Epidemiologic methods for the study of infectious diseases. Oxford, U.K.: Oxford University Press, 2001:86–115. [Google Scholar]
  77. Coggon D, Rose G, and Barker DJP. Epidemiology for the uninitiated. 4th ed. London, U.K.: BMJ Publishing Group, 1997:39–42. [Google Scholar]
  78. Rosner B. Fundamentals of biostatistics. 2d ed. Boston, MA: Duxbury Press, 1986:240–1. [Google Scholar]
  79. Last JM. ed. Dictionary of epidemiology. 3d ed. New York, NY: Oxford University Press, 1995:32. [Google Scholar]
  80. Becker TM. Email to: Jonathan Eldredge. 21 Apr 2000. [Google Scholar]
  81. Yerkey N, Glogowski M. Bibliographic scatter of library and information science literature. J Educ Libr Inf Sci. 1989 Fall; 30(2):90–101. [Google Scholar]
  82. Mulrow C, Cook D. Systematic reviews: synthesis of best evidence for health care decisions. Philadelphia, PA: American College of Physicians, 1998:6–9. [Google Scholar]
  83. Mulrow C, Cook D. Systematic reviews: synthesis of best evidence for health care decisions. Philadelphia, PA: American College of Physicians, 1998:6–9. [Google Scholar]
  84. Eldredge JD. Evidence-based librarianship: an overview. Bull Med Libr Assoc. 2000 Oct; 88(4):289–302. [PMC free article] [PubMed] [Google Scholar]
  85. Eldredge JD. Evidence-based librarianship: searching for the needed EBL evidence. Med Ref Serv Q. 2000 Fall; 19(3):1–18. [DOI] [PubMed] [Google Scholar]
  86. Morse DH, Clintworth WA. Comparing patterns of print and electronic journal use in an academic health science library. Presented at: 2000 Joint Meeting of the Medical Library Group of Southern California and Arizona (MLGSCA) and Northern California and Nevada Medical Library Group (NCNMLG); 2000 Feb 11. [Google Scholar]
  87. Dee CR, Rankin JA, and Burns CA. Using scientific evidence to improve hospital library services: Southern Chapter/Medical Library Association journal usage study. Bull Med Libr Assoc. 1998 Jul; 86(3):301–6. [PMC free article] [PubMed] [Google Scholar]
  88. Eldredge JD. The vital few meet the trivial many: unexpected use patterns in a monographs collection. Bull Med Libr Assoc. 1998 Oct; 86(4):496–503. [PMC free article] [PubMed] [Google Scholar]
  89. Blecic DD. Monograph use at an academic health sciences library: the first three years of shelf life. Bull Med Libr Assoc. 2000 Apr; 88(2):145–51. [PMC free article] [PubMed] [Google Scholar]
  90. Dyer H, Buckle P. Who's been using my CD-ROM? Results of a study on the value of CD-ROM searching to users in a teaching hospital library. Health Libr Rev. 1995 Mar; 12(1):39–52. [DOI] [PubMed] [Google Scholar]
  91. Schneider E, Mankin CJ, and Bastille JD. Practical library research: a tool for effective library management. Bull Med Libr Assoc. 1995 Jan; 83(1):22–6. [PMC free article] [PubMed] [Google Scholar]
  92. Byrd GD. Medical faculty use of the journal literature, publishing productivity and the size of the health sciences library journal collections. Bull Med Libr Assoc. 1999 Jul; 87(3):312–21. [PMC free article] [PubMed] [Google Scholar]
  93. Farrar S, Bang D. Info Scripts: gateways to consumer health information. Bibl Medica Can. 1998 Summer; 19(4):135–8. [Google Scholar]
  94. Lacroix EM. Interlibrary loan in U.S. health sciences libraries: journal article use. Bull Med Libr Assoc. 1994 Oct; 82(4):363–8. [PMC free article] [PubMed] [Google Scholar]
  95. Ball AL. Is free online really free? Hypothesis. 2000 Fall; 14(3):1–2. [Google Scholar]
  96. Colborne D, Summers R, and Desjardins J. Client satisfaction and utilization of electronic and traditional library services. Bibl Medica Can. 1999 Summer; 20(4):173–7. [Google Scholar]
  97. D'Alessandro MP, D'Alessandro DM, Galvin JR, and Erkonen WE. Evaluating overall usage of a digital health sciences library. Bull Med Libr Assoc. 1998 Oct; 86(4):602–9. [PMC free article] [PubMed] [Google Scholar]
  98. Postell WD. Further comments on the mathematical analysis of evaluating scientific journals. Bull Med Libr Assoc. 1946 Apr; 34(2):107–9. [PMC free article] [PubMed] [Google Scholar]
  99. Kilgour FG. Annual report of the Yale Medical Library. New Haven, CT: Yale University, 1953:3–8;28–35. [Google Scholar]
  100. Kilgour FG. Use of medical and biological journals in the Yale Medical Library. part I. frequently used journals. Bull Med Libr Assoc. 1962 Jul; 50(3):429–43. [PMC free article] [PubMed] [Google Scholar]
  101. Kilgour FG. Use of medical and biological journals in the Yale Medical Library. part II. moderately used journals. Bull Med Libr Assoc. 1962 Jul; 50(3):444–9. [PMC free article] [PubMed] [Google Scholar]
  102. Morse EH, Beatty WK, and Hodge HM. Annual report on the library. Transactions & Studies of the College of Physicians of Philadelphia. 1954 Apr; 21(4):138–150. [PubMed] [Google Scholar]
  103. Keys TE. Applied medical library practice. Springfield, IL: Charles C. Thomas, 1958:389–93. [Google Scholar]
  104. Kurth WH. Survey of the interlibrary loan operation of the National Library of Medicine. Bethesda, MD: U.S. Department of Health, Education, and Welfare, 1962. [Google Scholar]
  105. Fleming TP, Kilgour FG. Moderately and heavily used biomedical journals. Bull Med Libr Assoc. 1964 Jan; 52(1):234–41. [PMC free article] [PubMed] [Google Scholar]
  106. Kolner SJ, Dalrymple PW, and Christiansen R. Teaching skills in medical information retrieval to medical students. J Med Educ. 1986 Nov; 61(11):906–10. [DOI] [PubMed] [Google Scholar]
  107. Rankin JA. Problem-based medical education: effect on library use. Bull Med Libr Assoc. 1992 Jan; 80(1):36–43. [PMC free article] [PubMed] [Google Scholar]
  108. Marshall JG, Fitzgerald D, Busby L, and Heaton G. A study of library use in problem-based and traditional medical curricula. Bull Med Libr Assoc. 1993 Jul; 81(3):299–305. [PMC free article] [PubMed] [Google Scholar]
  109. Frasca MA, Dorsch JL, Aldag JC, and Christiansen RG. A multidisciplinary approach to information management and critical appraisal instruction: a controlled study. Bull Med Libr Assoc. 1992 Jan; 80(1):23–8. [PMC free article] [PubMed] [Google Scholar]
  110. Ikeda NR, Schwartz DG. Impact of end-user search training on pharmacy students: a four-year follow-up study. Bull Med Libr Assoc. 1992 Apr; 80(2):124–30. [PMC free article] [PubMed] [Google Scholar]
  111. Fox LM, Richter JM, and White NE. A multidimensional evaluation of a nursing information-literacy program. Bull Med Libr Assoc. 1996 Apr; 84(2):182–90. [PMC free article] [PubMed] [Google Scholar]
  112. Noesgaard C, Bayley L, Blythe J, Dzugan R, Lunyk-Child O, Royle JA, and Valaitis R. Nursing students' use and comfort with computer technology. Bibl Medica Can. 1999 Fall; 21(1):231–8. [Google Scholar]
  113. Burrows SC, Tylman V. Evaluating medical student searches of MEDLINE for evidence-based information: process and application of results. Bull Med Libr Assoc. 1999 Oct; 87(4):471–6. [PMC free article] [PubMed] [Google Scholar]
  114. Nesbitt KW, Glover J. Can we prove that medical students can be taught to search MEDLINE effectively? Paper presented at: MLA 2001, 101st Annual Meeting of the Medical Library Association; Orlando, FL; 29 May 2001. [Google Scholar]
  115. Shelstad KR, Clevenger FW. On-line search strategies of third year medical students: perception vs fact. J Surg Res. 1994 Apr; 56(4):338–44. [DOI] [PubMed] [Google Scholar]
  116. Wildemuth BM, Moore ME. End-user behaviors and their relationship to search effectiveness. Bull Med Libr Assoc. 1995 Jul; 83(3):294–304. [PMC free article] [PubMed] [Google Scholar]
  117. McKibbon KA, Haynes RB, Walker-Dilks CJ, Ramsden MF, Ryan NC, Baker L, Flemming T, and Fitzgerald D. How good are clinical MEDLINE searches? a comparative study of clinical end-user and librarian searches. Comput Biomed Res. 1990 Dec; 23(6):583–93. [DOI] [PubMed] [Google Scholar]
  118. Eldredge J. BobCats, infestation, and a new library. Coll Res Libr News. 1984 Jul/Aug; 45(7):336–40. [Google Scholar]
  119. Dorsch JL. Equalizing rural health professionals' information access: lessons from a follow-up outreach project. Bull Med Libr Assoc. 1997 Jan; 85(1):39–47. [PMC free article] [PubMed] [Google Scholar]
  120. Tabur S, Devakos R, Newman S, Bayne J, Bartlett J, Reid EA, Devore L, Russell ML, and Uleryk E. Peer informatics tutoring for medical students: the Toronto experience. Bibl Medica Can. 2001 Spring; 22(3):121–7. [Google Scholar]
  121. Marshall JG. End-user training: does it make a difference? Med Ref Serv Q. 1989 Fall; 8(3):15–26. [DOI] [PubMed] [Google Scholar]
  122. Marshall JG. The impact of the hospital library on clinical decision making: the Rochester Study. Bull Med Libr Assoc. 1992 Apr; 80(2):169–78. [PMC free article] [PubMed] [Google Scholar]
  123. Bowden VM, Kromer ME, and Tobia RC. Assessment of physicians' information needs in five Texas counties. Bull Med Libr Assoc. 1994 Apr; 82(2):189–96. [PMC free article] [PubMed] [Google Scholar]
  124. Gorman PN, Ash J, and Wykoff L. Can primary care physicians' questions be answered using the medical journal literature? Bull Med Libr Assoc. 1994 Apr; 82(2):140–6. [PMC free article] [PubMed] [Google Scholar]
  125. Urquhart C, Hepworth J. The value of information supplied to clinicians by health libraries: devising an outcomes-based assessment of the contribution of libraries to clinical decision-making. Health Libr Rev. 1995 Sep; 12(3):201–13. [DOI] [PubMed] [Google Scholar]
  126. Gorman PN, Helfand M. Information seeking in primary care: how physicians choose which clinical questions to pursue and which to leave unanswered. Med Decis Making. 1995 Apr–Jun; 15(2):113–9. [DOI] [PubMed] [Google Scholar]
  127. Curtis KL, Weller AC, and Hurd JM. Information-seeking behavior of health sciences faculty: the impact of new information technologies. Bull Med Libr Assoc. 1997 Oct; 85(4):402–10. [PMC free article] [PubMed] [Google Scholar]
  128. Feldman JD, Bowden VM. Library use survey of Texas Health Science Center–San Antonio faculty: comparison of 1996 and 2000. Poster presented at: MLA 2001, 101st Annual Meeting of the Medical Library Association; Orlando, FL; 25–30 May 2001. [Google Scholar]
  129. Klein MS, Ross FV, Adams DL, and Gilbert CM. Effect of online literature searching on length of stay and patient care costs. Acad Med. 1994 Jun; 69(6):489–95. [DOI] [PubMed] [Google Scholar]
  130. Kars M, Olson S.. A retrospective study of PubMed training and Loansome Doc registration improving access to the medical literature. J Hosp Libr. 2001;1(2):67–72. [Google Scholar]
  131. Baker LM. Monitors and blunters: patient health information seeking from a different perspective. Bibl Medica Can. 1994 Winter; 16(2):60–3. [Google Scholar]
  132. Newcomer AP, Piscotti RA. Career progression of academic medical library directors. Bull Med Libr Assoc. 1989 Apr; 77(2):185–95. [PMC free article] [PubMed] [Google Scholar]
  133. Wienker CW, Bennett KA.. Trends and developments in physical anthropology, 1990–91. Am J Phys Anthropol. 1992;87:383–93. doi: 10.1002/ajpa.1330510311. [DOI] [PubMed] [Google Scholar]
  134. Eldredge JD. Evidence-based librarianship: an overview. Bull Med Libr Assoc. 2000 Oct; 88(4):289–302. [PMC free article] [PubMed] [Google Scholar]
  135. Prentice RL. Design issues in cohort studies. Stat Methods Med Res. 1995 Dec; 4(4):273–92. [DOI] [PubMed] [Google Scholar]
  136. Daly LE, Bourke GJ. Interpretation and uses of medical statistics. 5th ed. Oxford, U.K.: Blackwell Science, 2000:146. [Google Scholar]
  137. Gordis L. Epidemiology. 2d ed. Philadelphia, PA: WB Saunders, 2000:131–9. [Google Scholar]
  138. Benson K, Hartz AJ. A comparison of observational studies and randomized, controlled trials. N Engl J Med. 2000 Jun 22; 342(25):1878–86. [DOI] [PubMed] [Google Scholar]
  139. Pocock SJ, Elbourne DR. Randomized trials or observational tribulations? N Eng J Med. 2000 Jun 22; 342(25):1907–9. [DOI] [PubMed] [Google Scholar]
  140. Moses LE. Measuring effects without randomized trials? Options, problems, challenges. Med Care. 1995 Apr; 33(4 Suppl):AS8–14. [PubMed] [Google Scholar]
  141. Feinstein AR.. Epidemiologic analyses of causation: the unlearned scientific lessons of randomized trials. J Clin Epidemiol. 1989;42(6):481–9. doi: 10.1016/0895-4356(89)90142-x. [DOI] [PubMed] [Google Scholar]
  142. Weinberger J, Eig A. Expectancies: the ignored common factor in psychotherapy. In: Kirsch I, ed. How expectancies shape experience. Washington, DC: American Psychological Association, 1999;357–82. [Google Scholar]
  143. Pocock SJ, Elbourne DR. Randomized trials or observational tribulations? N Eng J Med. 2000 Jun 22; 342(25):1907–9. [DOI] [PubMed] [Google Scholar]
  144. Hill HA, Kleinbaum DG. Bias in cohort studies. In: Gail MH, Benichou J, eds. Encyclopedia of epidemiologic methods. Chichester, U.K.: John Wiley & Sons, 2000:92–4. [Google Scholar]
  145. Hill HA, Kleinbaum DG. Bias in observational studies. In: Gail MH, Benichou J, eds. Encyclopedia of epidemiologic methods. Chichester, U.K.: John Wiley & Sons, 2000:94–100. [Google Scholar]

Articles from Journal of the Medical Library Association are provided here courtesy of Medical Library Association

RESOURCES