End-of-Life Care Content in 50 Textbooks From Multiple Specialties FREE

Many patients in the United States currently receive suboptimal care at the end of life.^1- 3 Inadequate physician training likely contributes to both deficient care for dying patients and increased anxiety for caring physicians.^4- 11 Medical education typically provides little training in care of the dying.^12- 15 Most medical schools, residencies, and fellowships offer almost no formal training in palliative care, the information taught is not well integrated into the curricula, and the few courses available are generally elective.¹⁶ National medical licensing examinations have relatively few questions assessing students' end-of-life care competence.¹⁷ In general, students and physicians feel ill prepared to provide end-of-life care.¹⁸

Textbooks serve as a cornerstone in the training of medical students and residents, as authoritative references and reviews for more experienced clinicians, and as an important feature of professional orthodoxy and culture. Textbooks are central to clinical medicine, not only describing the expected best practices but also codifying the principles and standards of clinical care. Previous research suggests that the end-of-life content in internal medicine, pediatrics, and nursing textbooks is limited.¹⁹ Carron et al²⁰ reviewed 4 of the classic internal medicine texts and found little helpful information on providing care to patients at the end of life. Hill et al²¹ analyzed the end-of-life care content in 5 major pediatric textbooks and found similar deficiencies in the areas of symptom management, advance care planning and communication. Ferrell et al²² have recently documented limited end-of-life content in 50 nursing textbooks.

These results suggest that the environment for learning necessary skills in end-of-life care is inadequate. Although generalist physicians and those from nearly every specialty provide care for patients at the end of life, a broad survey of a large number of general and specialty textbooks in the disciplines of medicine, surgery, pediatrics, and psychiatry has not been performed. To assess the end-of-life care content in a wider, more representative range of textbooks, we undertook a detailed review of 50 leading medical textbooks, analyzing the presence and utility of the end-of-life care information presented.

To assess the current status of end-of-life care content, 2 readers reviewed in 1998 the most recent editions available for 50 best-selling medical textbooks. We identified best-selling textbooks based on a widely used publishing industry sales report, the Login Brothers Report for 1997.²³ To ensure review of a wide spectrum of textbooks, we selected a variety of general and specialty textbooks from a variety of publishers (Table 1 ). Once we selected the 50 textbooks, we requested review copies from their publishers or purchased them.

The format for review included 13 domains of end-of-life content that one might reasonably expect to find in medical textbooks. The domains chosen were based on position statements from important national and international medical organizations.^3,24- 28 Two national experts who had previously analyzed textbooks reviewed the domains chosen.^20,22 Furthermore, the domains selected address all elements in the biopsychosocial model of health.²⁹ The domains were epidemiology (vital statistics), natural history (prognosis, time course, mode of death, and symptoms), pain management, nonpain symptom management (dyspnea, nausea and vomiting, delirium, and fatigue, among others), psychological issues (depression, anxiety, fear, loneliness, and grief), social and demographic issues (interpersonal relationships with spouses or partners, family, and friends; gender; race; and cultural and economic issues), spiritual issues (abandonment, completion of tasks, acceptance, religious tasks, and choices), family issues (communication of patient and family member wishes, grief and bereavement, informal caregiver role and support, education, and economic issues), definition of end-of-life care (definition of death and goals of care), ethics, law, and policies (individual vs organizational ethics, patients' self-determination, double effect, withdrawal, and withholding of life support), physician after-death responsibilities (including pronouncement, autopsy, and organ donation), physician roles (communication with patient and family, personal grief, and bereavement), and context of care (advance directives, options for end-of-life care, referral to hospice, and funeral arrangements).

In each book, we reviewed textbook chapters that pertained to the most common causes of death in the United States.³⁰ Using the tables of contents of textbooks from a particular specialty, we identified chapters for review that concerned disease entities or topics relevant to end-of-life care in that specialty. For example, we reviewed chapters on coronary artery disease in cardiology texts but not in neurology texts. Some chapters selected for review overlapped multiple specialties (for example, lung cancer was reviewed in pulmonary, oncology, family and primary care medicine, and internal medicine texts), allowing comparisons across and between textbooks in various specialties. The number of chapters reviewed in each textbook ranged from a minimum of 4 chapters (psychiatry) to a maximum of 13 chapters (internal medicine).

Reviewers evaluated all textbooks within a specialty concurrently; for example, all neurology textbooks were reviewed sequentially before moving to the next specialty. Each textbook was rated by at least 1 reviewer. The review process entailed copying the selected chapters, reading the entire chapter, and highlighting content appropriate to each of the 13 domains. Following a complete reading of each chapter, readers rated the marked content in each domain using the following ratings: 0 for absent content; 1 for minimal content; and 2 for helpful content and topic well discussed. Rating was done as liberally as possible, with domains receiving a 1 for any mention of the topic or for a cross-reference to content elsewhere in the text. Domains received a 2 for providing any helpful information. For example, a chapter noting simply that most oncologists believe that patients with cancer should be told about their diagnosis would receive a rating of 1 for the communication domain. A rating of 2 would be assigned if the chapter gave any further information about communication, such as how to discuss a new diagnosis with patients, common barriers to communication, or patient expectations for communication. This rating process was derived from an earlier study of end-of-life care content in 4 internal medicine textbooks.¹⁹ The 2, 1, and 0 domain ratings were then entered into a computerized database.

We also reviewed the table of contents of each text to identify chapters specifically devoted to topics related to care for patients at the end of life. Finally, we examined the index of each textbook for 18 key words concerning end-of-life care issues (advance directives, autopsy, bereavement, brain death, death, death certificate, do not resuscitate, double effect, dying, end of life, family, grief, hospice, mortality, palliation, pronouncement, withdrawal, and withholding support). We selected these particular key words because they directly reflected or were synonyms for the domains under review. For each keyword, we entered into the database the total number of pages with citations to that word and the total number of pages in the textbook.

To determine interrater reliability for the 13 domains of end-of-life care content, the 2 reviewers both rated 10% of the chapters. The chapters read by both reviewers included the first textbook reviewed in each specialty to ensure that each reader was scoring that specialty's texts similarly. If discrepancies in scores were present, resolution was achieved by a consensus of all 4 of the investigators.³¹ This process standardized the reading of chapters for subsequent textbooks. The total observed agreement of 97% between the 2 raters indicates a high degree of interrater reliability. Adjudication was required for only 3% of the rater scores for which there were disagreements.

We entered all ratings into a database prior to any data analysis. Given the descriptive nature of this study, we limited our analyses to descriptive statistics (averages) and frequency distributions. We used Microsoft Excel 1998 (mfr) for all data management and analysis.

For each chapter reviewed, each of the 13 end-of-life content domains received a rating of 2, 1, or 0. We used these raw domain ratings to create a content score for each chapter, textbook, and specialty. For each chapter, we calculated the helpful, minimal, or absent content score as the percentage of 2, 1, or 0 domain ratings across all 13 domains in a particular chapter. For example, a chapter on stroke in a neurology textbook that received a rating of 2 in 1 domain, 1 in 4 domains, and 0 in the other 8 domains would be described as having a helpful content score of 7.7%, a minimal content score of 30.8%, and an absent content score of 61.5%. We calculated the content score for each textbook as the average of the content scores for all chapters reviewed in that particular book. We calculated the content score for each specialty as the average of the content scores for all textbooks reviewed in that particular specialty. In addition, we calculated the overall averages of the helpful, minimal, and absent content scores for all 50 textbooks reviewed. To evaluate how various specialties compared in their coverage of a common chapter, we calculated the average content scores for a particular chapter topic across all texts in a given specialty.

Given the relatively small numbers of elements in each category of analysis, content scores and their averages were not adjusted or weighted to account for the amount of content (number of pages). However, when a chapter topic was completely absent from a particular textbook, the analysis was adjusted for this absence. For example, Rakel's Textbook of Family Practice contained only 9 of 10 chapter topics selected for review in the family and primary care medicine textbook category. In the analysis, the chapter not present (adult acute respiratory distress syndrome) was ignored and the content score for this textbook was calculated based on the average of content scores for 9 rather than 10 chapters.

To compare coverage of various end-of-life domains, we conducted an in-depth analysis of the domains in textbooks from a single specialty—internal medicine. There were 14 chapter topics reviewed in each of the 6 internal medicine textbooks. We identified the domains that received a consistent 0 or 2 rating for a particular chapter topic in all 6 internal medicine texts. For example, the domain of epidemiology consistently (in all 6 internal medicine texts) received a helpful content rating in 7 of the 14 chapter topics reviewed. We evaluated the 13 content domains to identify consistently well-covered or consistently poorly covered domains. We considered a domain to be well covered if it had consistent helpful content ratings in all 6 texts in at least 7 of the 14 chapter topics reviewed. We considered a domain to be poorly covered if it had consistent absent content ratings in at least 7 of the 14 chapter topics reviewed.

The content scores for individual textbooks varied considerably (Table 1a). The highest helpful content score was 44.2% in a psychiatry text, indicating that, on average, the chapters reviewed in that text contained helpful content for 44.2% of end-of-life content domains. The highest absent content score was 77.9% in an infectious diseases and AIDS text, indicating that, on average, the chapters reviewed in that text contained no information for 77.9% of end-of-life content domains. For the 50 textbooks reviewed, the average helpful content score was 24.1% (range, 8.7%-44.2%), the average minimal content score was 19.1% (range, 6.2%-38.5%), and the average absent content score was 56.9% (range, 23.1%-77.9%); thus, overall, the textbooks contained helpful information in about one quarter of end-of-life content domains, minimal information in about one fifth of end-of-life content domains, and no information at all for more than one half of end-of-life content areas.

Figure 1 displays the content scores for each specialty. As a group, surgical textbooks had the least end-of-life content with an absent content score of 71.8%. The quartile of specialties with the least end-of-life content also included infectious diseases and acquired immunodeficiency syndrome (AIDS) (absent content score, 70.0%) and oncology and hematology (absent content score, 61.9%). The quartile of specialties with the greatest amount of helpful end-of-life content included family and primary care medicine and psychiatry (helpful content score, 34.4% each) and geriatrics (helpful content score, 29.6%).

Some diseases or chapter topics that we reviewed were covered in the textbooks of several specialties. As one might expect, the greatest amount of helpful end-of-life information for coronary artery disease was found in cardiology texts (average helpful content score, 34.6%). However, for many other topics, end-of-life information was not necessarily where one might expect to find it. For example, the highest average helpful content score for Pneumocystis carinii pneumonia was found in internal medicine texts (19.3%), whereas the lowest (7.7%) was found in the infectious diseases and AIDS texts. For Alzheimer disease, the greatest amount of helpful end-of-life information was present in family and primary care medicine texts (average helpful content score, 50.8%), while neurology texts had the least amount of helpful information (15.4%). Family and primary care medicine texts had the greatest amount of useful end-of-life content for chronic obstructive pulmonary disease (average helpful content score, 43.1%), not pulmonary texts (32.7%) or internal medicine texts (20.5%). The most helpful information about end-of-life care in lung cancer was found in family and primary care medicine texts (average helpful content score, 28.2%), not in oncology texts (11.6%). Specific content scores by chapters are provided in Table 2.

We compared the end-of-life domains for all 14 chapter topics reviewed in the 6 internal medicine texts. The domains of epidemiology and natural history were generally well covered in internal medicine, having consistently received a domain rating of 2 (helpful content) in at least one half of all chapters reviewed. In general, the domains of social issues, spiritual issues, family issues, ethics, and physician after-death responsibilities were poorly covered. For these domains, a 0 domain rating (absent content) was consistently recorded in at least one half of all chapters reviewed. Two domains (spiritual issues and physician responsibilities after death) were completely absent from chapters covering particular diseases and were discussed only in a few chapters devoted specifically to care at the end-of-life. Coverage of specific domains by chapter are provided in Figure 2.

Review of the tables of contents of the 50 texts revealed that 24 (48%) had chapters on topics related to end-of-life care. Some of these chapters were comprehensive, with titles such as "Care at the End of Life." However, many of these chapters were focused on a specialized aspect of end-of-life care, with titles such as "Ethical Issues in Clinical Medicine." Each of the texts in emergency medicine (n = 4), family and primary care medicine (n = 5), and internal medicine (n = 6) contained a chapter on a topic related to end-of-life care. However, one half or less of the textbooks in oncology and hematology (n = 6), pediatrics (n = 4), and psychiatry (n = 3) had a chapter on end-of-life care. None of the cardiology (n = 4), infectious diseases and AIDS (n = 3), neurology (n = 3), pulmonary (n = 4), or surgery (n = 3) texts included such a chapter.

We evaluated the indexes of the 50 textbooks for citations to 18 end-of-life content key words. Out of 34,845 textbook pages, 697 pages (2%) contained citations to at least 1 of the end-of-life care key words. For individual textbooks, the percentage of pages with end-of-life care key words compared with the total number of textbook pages ranged from 0% to 12%. Specialties with the highest average index percentages were geriatrics (4.74%) and family and primary care medicine (2.81%). Specialties with the lowest average percentages were infectious diseases and AIDS (0.07%), pulmonary (0.09%), neurology (0.23%), and surgery (0.36%). No end-of-life key words were found in the index of 7 of the 50 textbooks.

In analyzing how common causes of death are discussed in 50 best-selling general and specialty textbooks, we found helpful information for less than one fourth of end-of-life content domains and minimal coverage for one fifth of those domains. We found no information for more than one half of the end-of-life content domains one might reasonably expect to be covered in a medical textbook. Consistently, there was a paucity of attention to the domains of social, spiritual, family, and ethical issues at the end-of-life, and to the domain of physician responsibilities after death.

Almost one half of the textbooks had chapters focused on at least some aspect of care at the end of life, but many of these chapters were not comprehensive. In addition, chapters focused on end-of-life care were uncommon in some specialties (such as oncology, pediatrics, and psychiatry) and were completely absent in the cardiology, infectious diseases and AIDS, neurology, pulmonary, and surgery texts reviewed. Textbooks with the greatest amount of helpful end-of-life content were in the specialties of family and primary care medicine, psychiatry, and geriatrics. Textbooks with the least end-of-life content were in the specialties of infectious diseases and AIDS, oncology and hematology, and surgery. Unfortunately, specialty textbooks that physicians might read for information about a particular disease often did not contain helpful information about caring for patients dying from that disease. In several important instances, general rather than specialty textbooks had better end-of-life coverage for a particular disease. Finally, our analysis also confirmed major deficiencies in the indexing of end-of-life topics.

This textbook analysis was limited in that although 50 texts were reviewed overall, the number of texts in each specialty was small and the content score averages were not weighted. Although a top-selling book, only a single AIDS text was reviewed. Some texts may have been updated subsequent to our review. Also, some influential textbooks may have been excluded because they are not top sellers.

In reviewing each chapter, it is possible that some domain ratings assigned were inaccurate. However, the chapters were read, reviewed, and rated carefully by experienced clinical researchers with evidence of high interrater reliability. The rating was performed generously and liberally. It is possible that some end-of-life content was present elsewhere in the textbook but not in the chapters we reviewed. However, in the scoring, textbooks were not penalized for chapters that were completely absent (despite their presence in competing textbooks). And while some relevant content may have been missed if it was not identified in the table of contents, not listed in the index, or not included in the chapters reviewed, this hidden end-of-life content is likely not to be found by busy students or clinicians.

It is possible that textbooks may have located all end-of-life content in a single chapter devoted to end-of-life care or ethics, consequently receiving a lower content score than texts that included their end-of-life content in each individual disease chapter. However, in our analysis, we gave credit for any mention of a topic, including cross-references to content elsewhere in the text. Moreover, our analysis revealed that texts that include a chapter devoted to end-of-life care generally scored well in specific disease chapters throughout the book as well. Ideally, all texts would be cross-referenced, even if general information (such as attending to spiritual issues) is presented in a general chapter and disease-specific end-of-life content (such as management of dyspnea) is presented in the relevant disease chapter.

The index analysis may be incomplete since the list of end-of-life care key words reviewed included only 18 words. Other key words could have been chosen. For example, although pain was one of the 13 domains, it was not included in the index analysis because the word pain has great lack of specificity for end-of-life care. Most index references for "pain" are for acute pain. Futility was not included because it is primarily an ethical issue whose clinical implications are captured by key words included in the analysis, such as withdrawal or withholding support.

The utility of an index analysis is only as good as the index itself. Some textbooks had good end-of-life content that was not appropriately indexed and that was therefore underrepresented in the index analysis. For underrepresentative indexes, the lack of accessible information is a potentially major critique of textbook quality and utility. Some textbooks received credit for end-of-life content in the index analysis but when reviewed, the cited pages were actually unrelated to end-of-life care (overrepresentative index). The finding of overrepresentative indexes only strengthens our conclusion that commonly used textbooks from multiple specialties provide limited end-of-life content.

Given the failing grade for medical textbooks, efforts should be undertaken to improve them. We and others have suggested a number of steps that can be taken by authors, editors, and publishers.^20,32- 33 Change has already begun. Partly in response to textbook reviews,^20- 22 publishers have already commissioned updates of end-of-life care content in a major textbook in each of the disciplines of nursing, pediatrics, and psychiatry and in 2 medical textbooks, including the creation of new chapters devoted to end-of-life care. Increasingly, our national professional accreditation and licensing bodies (including the Joint Commission on Accreditation of Healthcare Organizations, the National Board of Medical Examiners, and the US Medical Licensing Examination) are requiring a higher standard of physician competence in end-of-life care.¹⁷ This may prove to be a powerful motivation for improvement in best-selling textbooks.