To address the perception that ethical issues are underrepresented in the rheumatology literature, a systematic review was conducted using multiple databases to identify articles that addressed ethical and rheumatologic issues.
To address the perception that ethical issues are underrepresented in the rheumatology literature, a systematic review was conducted using multiple databases to identify articles that addressed ethical and rheumatologic issues.
A rheumatologist, research librarian, and clinician-ethicist designed queries for 4 electronic and ethics databases, searching for articles with content that was relevant to rheumatology/rheumatic diseases and that primarily focused on ethics. Based upon the Beauchamp and Childress framework, the retrieved articles were classified according to their ethical content, and the proportions addressing each Beauchamp and Childress ethical principle were analyzed using Cochran's Q statistic. Correlations between the appearance of discussions involving each of the 4 ethical principles were determined via chi-square analysis.
The total number of manuscripts in the rheumatologic literature with an ethical focus was 104 out of an estimated library of >400,000 rheumatologically oriented manuscripts (0.026%). Very few manuscripts consisted of original research studies. Nonmaleficence (66%) was the most common theme, whereas justice represented the least frequently addressed ethical issue (12%). The differences in the proportions of each ethical principle reached statistical significance (Q = 73.8, P < 0.0001). Only 8 articles addressed >2 ethical principles. Discussion touching on autonomy and nonmaleficence frequently appeared in the same article (Pearson's χ2 = 14.9, P < 0.001).
Despite the frequency of ethical issues while caring for patients, few reports within the rheumatic disease literature have focused on ethical issues. Further work should ascertain the degree to which the literature addresses the ethical questions in rheumatology.
Ethical dilemmas are ubiquitous for health care providers, with >99% of primary care physicians reporting the occurrence of ethical difficulties in their practice within the last year (1). Ethical issues arise frequently for nurse practitioners (2), pharmacists (3), and allied health professionals (4), as well as for physicians in both outpatient and inpatient settings (1, 5).
Current controversial ethical issues include the appearance of conflicts of interest associated with pharmaceutical companies (6), the need to conduct health-related research while protecting the privacy of data/subjects (7), and the emergence of ethically sensitive biotechnology, such as stem cell research (8). In the practice of rheumatology, there are many ethical issues that arise, such as the inaccessibility of biologic therapies for underinsured patients.
Despite the frequency of ethical quandaries in medicine and their potentially profound impact upon patients, only 20% of practitioners describe the available ethical literature as useful in addressing moral dilemmas (1). Some authors describe this inadequacy as one of insufficient quantity; indeed, it has been suggested that there is limited ethical discourse in rheumatology journals (9). Unfortunately, there are very few data regarding the quantity and breadth of the bioethics literature (10). Without even such basic estimates, it is impossible to assess whether a gap exists between the available ethics literature and the needs of providers and patients.
On a more fundamental level, we would hypothesize that the number and content of manuscripts devoted toward any subject within a body of literature would reflect upon the priority assigned to that subject. In the case of bioethics, the perceived frequency of moral questions inherent to the practice of health care in general and rheumatology in particular would be predicted to be reflected in a substantial volume of relevant manuscripts.
Thus, to address the perception that ethical issues are underrepresented in the rheumatology literature, we applied search strategies to electronic databases in order to conduct a systematic review of articles that addressed ethical as well as rheumatologic issues. The identified resources were then classified according to their ethical content to evaluate the completeness of the bioethics rheumatologic literature.
A rheumatologist (LC), professional research librarian (LH), and clinician-ethicist (AVP) formulated the search strategy. Inclusion criteria for the identified literature consisted of 1) a retrievable full-length quantitative analysis manuscript, qualitative research manuscript, editorial/commentary, letter to the editor, case report, or highlighted news story within a scientific journal with content that was relevant to rheumatology/rheumatic diseases, and 2) a primary focus within each identified article on an ethical principle, based upon the framework developed by Beauchamp and Childress (11). The Beauchamp and Childress framework represents one of the most widely adopted systems for understanding medical ethics.
Briefly, the Beauchamp and Childress framework has identified 4 clusters of ethical principles (respect for autonomy, beneficence, nonmaleficence, and distributive justice) that function as guidelines for moral behavior with regard to biomedical issues. Each of these ethical principles is described in Table 1. This table also provides practical examples of potential challenges to each ethical principle, and includes a citation in which the ethical principle was operative, for illustration purposes. Respect for autonomy, for example, is the ethical principle that encompasses an individual's freedom of will and decision-making capacity. It is thus inexorably intertwined with issues of informed consent, which depend upon freedom from coercion and the ability to make choices. Distributive justice embodies the concept of fair and equitable treatment among individuals. Racial and ethnic disparities in health status and access to appropriate treatments are examples of specific issues relevant to the principle of distributive justice.
|Principle||Description of principle||Example of a potential challenge to principle||Relevant reference|
|Autonomy||Respect for an agent's self-rule and independence from controlling influences; enabling individuals to make reasoned informed choices; respect for confidentiality||Inadequacy of subjects' understanding of the informed consent for a randomized trial may call into question whether subjects' independent wishes are being respected||27|
|Beneficence||Contributing to the welfare of persons; actions done to benefit others; the obligation to help others||Investigators using true placebos, rather than using active therapies, for the control arm of randomized trials||28|
|Nonmaleficence||An obligation to avoid the causation of harm||Use of sham operations in randomized controlled studies of a surgical procedure||29|
|Distributive justice||Distributing benefits, risks, and costs fairly, equitably, and appropriately; treating patients in a similar position in a similar manner||Preferential recruitment of subjects from economically disadvantaged backgrounds for therapeutic trials, although the drugs being studied are unavailable to study subjects after termination of the study (due to the considerable cost of the therapy)||30|
Relatively recent attempts to compile and organize the existing ethics literature have capitalized on electronic data (12) and standardized search strategies (13, 14). These efforts provide a means for identifying and characterizing the available ethics literature. In general, interrogation of databases for this systematic review utilized both general conceptual terms (e.g., rheumatology) and, when necessary, specific terms (e.g., “Takayasu's arteritis,” which is not subindexed under “rheumatology” in Medline). Databases searched and their interfaces are shown in Table 2. The specific search strategies are shown in Appendix A (available at the Arthritis Care & Research Web site at http://www.interscience.wiley.com/jpages/0004-3591:1/suppmat/index.html) or by request from the authors.
|Database||Interface||Years of access||Reference|
|Medline||Ovid||1950 to present||31|
|Cochrane Central Register of Controlled Trials, CENTRAL||Ovid||About 1948 to present||32|
|NRCBL database||ETHXWeb||1988 to present||17|
|Science Citation Index Expanded||Web of Science||1970 to present||18|
|Social Science Citation Index||Web of Science||1974 to present||19|
The Ovid interface was used to search the Medline database (15) and the Cochrane Central Register of Controlled Trials (16). The National Reference Center for Bioethics Literature (NRCBL) database was accessed via ETHXWeb (17). The Web of Science interface was used to search the Science Citation Index Expanded (18) and the Social Science Citation Index (19). EndNote, a computerized bibliographic management application, was used to organize citations (20). The authors' library of articles and the bibliographies of identified manuscripts were reviewed to locate additional primary sources (group of articles noted as “author identified” in Figure 1).
When possible, the searches utilized exploded medical subject headings (MeSH) with “multipurpose” truncated searches for text words that the MeSH terms did not capture (e.g., rheumatolog$.mp). MeSH terms were also focused to minimize the number of irrelevant hits. To maximize retrieval, the search in CENTRAL involved explosion of MeSH combined with individual text word searches for EMBase subject headings. A Boolean search function with truncated wildcards was used for the NRCBL, an archive of bioethics and professional ethics sponsored by Georgetown University's Kennedy Institute of Ethics and the National Library of Medicine (17). Finally, the Science Citation and Social Science Citation Indices were searched utilizing text words in the advanced search feature of Web of Science.
The citations identified through the database queries were reviewed for eligibility criteria, using the 2 criteria stipulated in the first paragraph of the Materials and Methods section. Manuscripts not conforming to the specified criteria were eliminated from consideration. Every source identified through the database queries was assessed for the presence of material related to each of the 4 specific ethical principles (derived from Table 1). To satisfy the eligibility criteria, the ethical content of the source had to represent a primary focus of the article, in the opinion of the investigators. Equivocal classifications were reconciled by consensus of the clinicians (LC and AVP). Articles addressing multiple ethical principles were classified accordingly.
From the final cohort of citations that met the rheumatologic and ethical criteria, the proportion addressing each Beauchamp and Childress ethical principle was reported separately. These 4 proportions were analyzed using Cochran's Q statistic to determine if the rheumatologic/ethical literature favored the treatment of certain ethical principles. The comprehensiveness of the ethical discussions was assessed by reporting the number of articles touching on 1, 2, 3, or all 4 of the Beauchamp and Childress principles.
To examine whether the ethical principles that were addressed in the articles demonstrated a pattern, correlations between the appearance of discussions involving each of the 4 ethical principles were determined. This was achieved by testing for an association between all possible pairs of Beauchamp and Childress principles through chi-square analysis (Pearson's chi-square) that examined the 6 possible permutations (for example, “Do articles addressing beneficence tend to also discuss justice?” etc.). For those correlations found to be statistically significant, effect size was assessed using phi coefficient for nominal variables (same as Cohen's w). We assumed a priori values <0.10 as indicating no effect, 0.10–0.29 as indicating a small effect size, and values >0.30 consistent with moderate effect size (21).
To understand the format through which ethical discourse occurs in the rheumatology literature, eligible articles were classified according to 6 possible categories: those that contained primarily original quantitative analyses, original qualitative research, editorials/commentaries, letters to the editor, case reports, or highlighted news stories in scientific journals. Categories were mutually exclusive. The proportion of articles in each format was determined as a percentage of the total number of eligible articles.
Search results are shown in Figure 1. Redundant citations refers to citations identified in multiple databases. The preponderance of articles was initially identified through ETHXWeb, followed by Ovid Medline (Figure 1). The majority of articles that were excluded (n = 336) did not address issues relevant to rheumatology; most of the excluded manuscripts dealt with type 1 diabetes mellitus (48% of nonrheumatologic manuscripts) or multiple sclerosis (25% of nonrheumatologic manuscripts), and were likely identified through the use of the search term “autoimmunity.” Less than 20 articles initially captured through database inquiries failed to meet both the rheumatology and ethics eligibility criteria.
The total number of manuscripts in the rheumatologic literature with an ethical focus was 104 out of an estimated eligible library of >400,000 rheumatologically oriented published manuscripts (∼0.026%). Of these 104 citations, 15 represented original quantitative or qualitative research (14%, <0.004% of the source library). An accurate count of the total number of rheumatologic manuscripts is not possible, because some of the initial search strategies for the Web of Science reported retrievals of >100,000. This result, therefore, likely represents an overestimate for the percentage of ethics articles within the body of rheumatology literature.
In terms of the article format (Table 3), 14% of the articles were original research (quantitative research: 13%, and qualitative research: 1%), whereas 86% were not research in nature. Sixty-four percent of the articles were editorials or commentaries, 13% were letters to the editor, 4% were case studies, and 4% were highlighted news stories within a scientific journal with content that was relevant to rheumatology or the rheumatic diseases.
|All articles||104 (100)||44 (42)||19 (18)||69 (66)||12 (12)|
|Original research||15 (14)||11 (73)||0 (0)||4 (27)||2 (13)|
|Quantitative study||14 (13)||10 (71)||0 (0)||4 (29)||2 (14)|
|Qualitative study||1 (1)||1 (100)||0 (0)||0 (0)||0 (0)|
|Nonresearch||89 (86)||33 (37)||19 (21)||65 (73)||10 (11)|
|Editorial/commentary||67 (64)||25 (37)||17 (25)||49 (73)||8 (12)|
|Letter to the editor||14 (13)||5 (35)||0 (0)||11 (79)||0 (0)|
|Case study/series||4 (4)||3 (75)||2 (50)||2 (50)||1 (25)|
|News stories||4 (4)||0 (0)||0 (0)||3 (75)||1 (25)|
Many of the initially retrieved rheumatology citations touched on ethical concepts; however, for the majority of these articles, the primary focus was not ethics oriented. For those with an ethics focus, the relevant citations addressed a broad spectrum of ethical concepts, broaching each of the 4 principles within the Beauchamp and Childress scheme (Table 3). Nonmaleficence (66%) and respect for autonomy (42%) were the most common themes within the rheumatology literature. Beneficence was an uncommon ethical subject, although theoretically, any randomized controlled trial at its core is concerned with the discovery of beneficial therapies or optimizing care. However, these controlled trials are rarely explicitly framed in ethical terms. Justice represented the least frequently addressed ethical issue in rheumatology literature, present in only 12% of those articles with an ethical focus. Testing using Cochran's Q statistic suggested that the differences in the proportions of each ethical principle reached statistical significance (Q = 73.8, 3 df, P < 0.0001).
Restricting the analysis to original research articles (n = 15) revealed a greater proportion of articles concerned with issues of autonomy (73%) and fewer with nonmaleficence (27%). Testing using Cochran's Q statistic again suggested that the differences in the proportions of ethical principles were significant (Q = 17.5, 3 df, P < 0.001). Nonresearch articles (n = 89) produced a distribution similar to that of the overall citation cohort, with significant differences in the proportions of ethical principles (Q = 74.4, 3 df, P < 0.0001).
The number of ethical principles broached in each article is displayed in Figure 2. The majority of articles focused only on a single ethical principle. Only 8 articles addressed >2 of the ethical principles. Among the 15 original research manuscripts, none addressed >2 ethical principles.
The correlation between the appearance of the 4 ethical principles was determined through chi-square analysis. Of the 6 possible permutations, the relationships found to be statistically significant were as follows: respect for autonomy and nonmaleficence (Pearson's χ2 = 14.9, 1 df, P < 0.001), beneficence and justice (Pearson's χ2 = 5.0, 1 df, P < 0.03), and nonmaleficence and justice (Pearson's χ2 = 6.6, 1 df, P = 0.01). The effect size for these correlations was moderate for articles addressing both respect for autonomy and nonmaleficence (phi = 0.38) and small for the 2 remaining relationships (phi = 0.22 and 0.25, respectively).
This study represents the emerging field of empirical ethics research, whereby objective data (an assessment of the number of articles and ethical principles within a body of literature) enlighten our understanding of ethics (22). This is in contrast to traditional ethics (i.e., normative ethics), which relies on philosophy, theology, and reason—rather than data—to arrive at conclusions or stimulate questions. In this case, the relatively sparse collection of ethical articles within the literature of a medical subspecialty evokes questions as to the reasons behind this paucity.
Despite an exhaustive and systematic search, our results suggest that a relatively small number of reports within the rheumatic disease literature have focused on ethical issues. These findings are consistent with prior work on the subject; preliminary reports have proposed that when compared with the general medical literature, the surgical (23) and subspecialty (24) literature appears to be relatively deficient in material focusing on ethics. The reason for this scarcity is unclear but may reflect priorities within rheumatology, a general unfamiliarity with formal ethical concepts and scholastic trends within bioethics, discomfort with nonquantitative disciplines, or alternative explanations. In any case, it is concerning that ethical discourse represents such a small fraction of the literature.
A substantial percentage of articles were identified through queries of the ETHXWeb database, indicating the importance of relying upon nontraditional sources for systematic reviews in the field of bioethics. While articles may often mention issues with ethical overtones in the results section (e.g., sociodemographic risk factors), the associated discussion often does not delve into the ethical implications. Because such ethics-related dialogue often does not take place, these articles would not have met our eligibility criteria.
The retrieved literature included sources addressing all 4 main ethical concepts that comprise the original Beauchamp and Childress framework; however, the frequency of the appearance of each of the concepts varied. Nonmaleficence represented the most frequent theme, whereas justice was least frequently encountered. This finding may provide empiric evidence for endorsement of the maxim primum non nocere (“above all [or first], do no harm”). The scarcity of articles addressing the subject of justice within the rheumatologic literature is provocative and warrants further inquiry.
The assembled library also suggests that bioethics articles within the corpus of rheumatologic literature tend to restrict their focus to a single ethical principle, rather than comprehensively considering a topic. Indeed, it was uncommon that more than 1 ethical principle was discussed per article. Letters to the editor, editorials, and commentaries may be composed in response to original research articles or specific current events. As such, they may focus on only the most salient points, rather than thoroughly addressing an issue. Limitation on length imposed by journals and editors may further contribute to their brevity. However, the tendency to address only 1 or 2 ethical principles also characterized original research manuscripts (Figure 2), which do not share these restrictions. For the authors of original research, this may reflect a desire to investigate issues in a focused manner. A less optimistic interpretation points to potentially incomplete or nonsystematic treatments of ethical issues within the literature.
When more than 1 principle was broached, chi-square analysis revealed that respect for autonomy and nonmaleficence commonly appeared together, underscoring conceptual similarities between these 2 constructs. To illustrate, the need to obtain adequate patient consent (a concept closely linked to autonomy) reflects a desire to avoid subjecting the patient to risks of potential harm against their will (maleficence).
Last, the majority of ethical discussions were conveyed through editorials and commentaries, with quantitative analyses and letters to the editor contributing equal, but small, proportions to the overall literature. The relatively frequent use of editorials and letters to the editor to convey ethical themes suggests that in addition to allowing for traditional communication and scientific contextualization of quantitative manuscripts, these formats are an extremely important vehicle for ethical discourse in the medical literature.
This descriptive report contains a number of limitations. Articles were retained only if their primary emphasis was both rheumatologic and ethics related. In point of fact, the majority of articles incorporate some nominal element of ethics (e.g., choice of informed consent process and use of institutional review boards in randomized controlled trials, sociodemographics risk factors in observational studies, etc.). However, we were concerned only with those articles that were primarily focused on ethics. In addition, we did not perform a comprehensive search of the entire ethics literature looking for rheumatologic-oriented issues. For logistical reasons, we instead chose to rely upon a search that incorporated the ETHXWeb database, as well as a search for ethical issues within 4 scientific databases. We hypothesized that this strategy was sufficient to characterize the relevant body of available sources. Finally, the ethical framework adopted in this report (Beauchamp and Childress ) was not originally designed as an ethical classification scheme. Consequently, it may not have been the optimal method of classifying the manuscripts identified. However, in the absence of any obvious alternatives, we selected it because it is the most commonly used and accessible conceptual scheme within the bioethics literature.
We have demonstrated that articles with an ethical focus comprise an extremely small percentage of the rheumatologic literature, and that commonly cited bioethical principles are not addressed with equal frequency. If medicine represents a “moral enterprise” (25) that is “inextricably bound to an ethical lattice” (26), these findings raise questions about why ethics discussions represent such a small contribution to the literature and what ethical lattice physicians base their decisions upon.
The practice of rheumatology, as a discipline with unique patient populations and clinical circumstances, no doubt engenders many ethical issues. The degree to which the available literature addresses ethically oriented issues specific to rheumatology is a potentially fertile ground for inquiry.
Dr. Caplan had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study design. Caplan, Prochazka.
Acquisition of data. Caplan, Hoffecker, Prochazka.
Analysis and interpretation of data. Caplan, Prochazka.
Manuscript preparation. Caplan, Hoffecker, Prochazka.
Statistical analysis. Caplan.
The authors wish to thank Diane Fairclough, PhD, for statistical support and Eliana Caplan, MA, for reviewing the manuscript.