Characteristics and impact of the most-cited palliative oncology studies from 1995 to 2016.

BACKGROUND
Palliative care, as a relatively young field within medicine, has increasingly used original research to validate and standardize its practice. In particular, palliative care has been incorporated into oncology to better address end-of-life decisions and care. The goal of this study is to identify seminal studies in the field of palliative oncology while more broadly characterizing the trends across the literature.


METHODS
The publication databases Scopus and Web of Science were queried using predefined search terms to identify palliative oncology studies published between 1995 and 2016. The 100 most-cited articles from the time periods 1995-2005 and 2006-2016 were selected and analysed for publication data and study content.


RESULTS
Palliative oncology studies were found to primarily examine patients with multiple rather than single cancer types and rarely were randomized controlled trials. Early research topics of pain, symptoms, and survival studies have been replaced by the issues of access to care, healthcare utilization, and religion and spirituality.


CONCLUSIONS
By identifying and analyzing notable studies in palliative oncology, we found areas of research that are commonly investigated or overlooked and identified model studies that highlight the need for additional disease-specific randomized control trials to provide high quality clinical evidence in the field.


Background
The concept of palliative care was notably and quite recently articulated by Dr. Cicely Saunders in her 1960 publication on the subject of caring for dying patients. The resulting chapter "The Management of Patients in the Terminal Stage" was a unique entry in an expansive six-volume series on cancer in the UK. At the time, care for dying patients was limited to a small number of religious hospice organizations, one of which, St. Joseph's Hospice in Hackney, London, employed Saunders as a research fellow. [1] The term "palliative care" was not adopted until 1974, when Dr. Balfour Mount from The Royal Victoria Hospital of McGill University in Montreal, Canada, popularized it in order to avoid the perceived connotations of the word hospice. This burgeoning field gained heightened attention 10 years later when the Institute of Medicine published the report "Approaching Death: improving care at the end of life" (M.I. Field and C.K. Cassel, editors), which documented deficiencies in end-of-life care in the United States. Despite this relatively short history as a distinct and officially recognized field within medicine, palliative care is rapidly and broadly being incorporated into the practice models of medical institutions worldwide. Only in 2006 was Hospice and Palliative Medicine recognized as an official subspecialty by the American Board of Medical Specialties. [2] Today, palliative care has increasingly become a focus of investigation as reflected by the growth in medical literature published in the field. [3] Characterizing the direction of research and advances in palliative care presents a uniquely challenging task. [4] Palliative care, as defined by the World Health Organization, "is an approach that improves the quality of life of patients and their families facing the problem associated with life-threatening illness, through the prevention and relief of suffering by means of early identification and impeccable assessment and treatment of pain and other problems, physical, psychosocial and spiritual." [5] Given its broad implications for all fields of medicine, palliative care is often wide-ranging in its scope and publication. In order to identify the most salient information, certain specialties have analysed how palliative care is specifically being employed or investigated within their field. [6] Oncology, in particular, is often associated with end-of-life decisions and care, and therefore, has great utility for palliative care in the management of the symptoms and stressors of life-limiting illness. A central tenant of conventional oncology care is focused on improving quality of life and survival by modifying the natural course of cancer through cancer directed treatments: systemic therapy, radiation, and surgery. The established framework of medical, surgical, and radiation oncology in cancer care may have led to a delay in integration of distinct palliative care services into the field of oncology.
Previous reports on palliative care in oncology have not comprehensively evaluated studies, identified seminal works, and trended their impact in the field. [7] In this paper we define and map the evolving trends within the palliative oncology literature. In order to achieve this goal, we have identified and characterized the 100 most-cited palliative oncology studies published between 1995-2005 and 2006-2016.

Methods
The search terms (palliat$ OR hospice$ OR "terminal care") developed by Sladek et al. were used to filter publications specific to palliative care from the general medical literature. [8] This expression was combined via "AND" with the terms (adenocarcinoma* OR adenosarcoma* OR antineoplastic OR cancer* OR carcinoid* OR carcinoma* OR glioblastoma* OR *cytoma OR glioma* OR hodgkin* OR leukaemi* OR leukemi* OR lymphoma* OR malignan* OR melanoma* OR metasta* OR myelodysplas* OR myeloma* OR neoplas* OR non-hodgkin* OR nonhodgkin* OR nonseminoma* OR oncolog* OR osteosarcoma* OR sarcoma* OR seminoma* OR teratoma* OR tumor* OR tumour* OR waldenstrom* OR mesothelioma* OR blastoma* OR wilms*) to select for palliative medicine literature specifically applied to the field of oncology.
On October 9, 2017, this Boolean expression was used to search the Scopus (http://www.scopus.com) database with the Article title, Abstract, Keywords input. The results were filtered by publication date in the timeframe 1995-2005 and subsequently sorted by Cited by (highest). Each result was reviewed in descending order until 100 articles had been identified that met the study's inclusion criteria. An article had to present original research in palliative medicine as it relates to oncology. Systematic reviews, case studies, review articles, and opinion pieces were excluded from the study. Studies addressing treatment (e.g. palliative chemotherapy) or other measures specifically intended to be life-prolonging were excluded as well. Only English-language studies were included. The same process was repeated for the years 2006-2016.
The aforementioned Boolean expression was also used to perform a Topic search in the database Web of Science (http://www.webofscience.com). Results were again divided into the two timeframes, sorted by Times Citedhighest to lowest, and selected based on the study inclusion criteria. The Web of Science results were cross-referenced with the Scopus output. Duplicates were excluded, and after inclusion criteria was met, articles were selected based on number of citations until a list of the 100 most-cited palliative oncology articles for each timeframe was compiled.
For each article, the citation count, author names, journal of publication, affiliated institution, country, and year were extracted using the information available from Scopus, Web of Science, and PubMed. Population data for each country of publication was retrieved from the United Nations World Population Prospects. [9,10] In order to further characterize each study, the cancer type, study topic, study population, study design, and data collection were categorized based on a review of the article's abstract. If further clarification was required, the study manuscript was reviewed in its entirety. Data was analyzed using Microsoft Excel 2017.

Results
The median number of citations from the 100 articles published between 1995 and 2005 was 142 with a range of 91 to 687. Of those, 18 articles were cited more than 250 times, and 6 were cited more than 500 times. From 2006 to 2016, the median number of citations was 78 with a range of 53 to 2453. Only 6 articles were cited more than 250 times, 2 of which were cited over 500 times. The vast majority (n = 66) were cited less than 100 times in this time frame (Additional files 1 and 2). The 10 most-cited studies in 1995-2005 and 2005-2016 were identified as potential seminal works in the field of palliative oncology (Tables 1 and 2). The sample studies from 1995 to 2016 were compared as a whole and ranked based on number of citations per year (Table 3).  (Table 4), studies predominantly used mixed cancer types (3 or more) to answer research questions, while lung cancer was most likely to be addressed independently. The subjects of pain and symptoms decreased between the two time frames, but the topic of survival decreased the most both in absolute and relative terms. There has been an increase in studies addressing access to palliative care, end-of-life discussions, religion and spirituality, and healthcare utilization. The psychosocial experience of cancer and palliative care remained a popular subject across both time frames. New study populations of hospitals and countries were examined in 2006-2016, but geriatric patients were less represented as compared to 1995-2005. In regard to study design, there was a decrease in prospective cohorts nearly equivalent to an increase in randomized control trials. Clinical means of data collection, such as physician assessment or laboratory and exam data, were used less frequently, while pre-existing datasets obtained via chart review or national and local databases were increasingly utilized.
The majority of palliative oncology research is concentrated in the United States (Tables 5 and 6), particularly in the Harvard-affiliated Boston medical institutions ( Table 7).  (Table 8). Over both periods, the Journal of Clinical Oncology (n = 40) and the Journal of Pain and Symptom Management (n = 29) were the most frequent sources of publication (Table 9).

Discussion
In our analysis of palliative oncology literature, the studies identified in our search were subdivided based on publication date, either from 1995-2005 or 2006-2016. The 100 most-cited articles from each time frame were selected in order to limit the scope to the most impactful studies in the field. Each study was systemically examined and categorized with the intent of characterizing palliative oncology literature collectively, charting trends within the field, and elucidating new avenues for subsequent research. In such a nascent field, early studies have primarily sought to substantiate the benefits of palliative oncology care. As a result, this foundational work has contributed to an increased awareness of palliative oncology research and its application to patient care. Across all publications included in our study, 8 articles were cited over 500 times. As expected, these predominantly occur in the 1995-2005 timeframe, which would allow each study to accumulate more citations in a time dependent manner. All but 1 of the articles were published after 1999, and the 2 overall most- In both time frames, 1995-2005 and 2006-2016, the analysed studies primarily used mixed cancer types in their inclusion criteria. Not selecting for specific cancer types facilitates a large enough patient population to generate a sufficiently powered study. This pragmatic approach to patient recruitment may reflect the general increase in clinical oncology research. Research protocols which limit participation in additional studies or trial fatigue on the part of patients may impact the type and quantity of patients available for study.
Nonetheless, the underlying premise appears to be that type of cancer would have limited influence on the administration of palliative care as opposed to having cancer in general. Lung cancer, in contrast, may have been the most frequently used specific cancer type because its experiences and symptoms present a unique scenario for receiving palliative care. Although a more intuitive explanation for studying palliative care in lung cancer is simply because more patients are available for study due to overall cancer epidemiology and prevalence. [11] If principles of palliative care are held to be largely universal in their application to oncology, the mixed cancer approach is appropriate. Nonetheless, cancer, more so than many other medical problems, represents a very heterogeneous set of diseases, each with a very distinct set of palliative needs. This stems from differences in comorbidity, life style, cancer symptoms, and treatment complications, which vary widely from one cancer to the next. Therefore, future investigation of palliative care in oncology should focus on specific cancer types in order to better delineate the impact of treatment.
After classifying the topics of each study, several trends were suggested from the outcome of this A particularly interesting topic is the example of end-of-life discussions. This exact phrase did not appear in any of the studies from 1995-2005 but was referred to in more general terms such as "communication." This heightened emphasis and validation of end-of-life discussions contributed to the inclusion of advance care planning for the first time on the Medicare physician fee schedule in 2016. [12] The increasing acceptance of the proven benefits of palliative oncology likely contributed to the increase in studies assessing the effectiveness or limitations in providing access to its services. The topic of health care utilization was increasingly studied more recently. Emergency department visits, hospitalization, admission to the intensive care unit, and other metrics  of care were often used a proxy for cost and aggressive medical intervention for patients at the end of life. Lastly, the psychosocial aspects of palliative oncology remained the consistent and most frequent study topic. These studies often reflected on the holistic experience of cancer and the ability of palliative care to alleviate the dying process. Overall, the distribution of study topics increased between 1995-2005 and 2006-2016, indicating a diversification of the research being undertaken in this field. In order to address the aforementioned topics, investigators did not limit their studies exclusively to patients (although this represented the majority), but included studies examining caregivers and healthcare providers as well. Studies examining palliative oncology in geriatric populations are underrepresented considering the incidence and mortality rates due to cancer are highest in patients 65 years and older. [13] In 2006-2016, hospitals and countries became study populations, representing a shift in scope from patient-level to a more macroscopic, population based analysis. Very few randomized controlled trial (RCT) were used until 2006-2016. This increase may be a response to the paucity of high-quality clinical evidence due to the ethical and practical challenges of conducting RCTs in palliative care. [14,15] Analysis of the literature reveals that palliative oncology studies consistently rely on interviews and surveys, which have the potential of introducing greater subjectivity into the data.
An ongoing challenge of this relatively young field is the concentration of highly-cited literature coming from a select number of countries. Within these countries, such as the United States, studies are originating most often from handful of institutions. This is in contrast to research in other fields of oncology, particularly in drug development or therapeutic cancer clinical trials, which have robust involvement from Europe and East Asia.   The inherent bias of this analysis is that it favours older articles and may neglect recently published studies that, while significant, may have less citations. [16] In an attempt to control for lead-time publication date, the studies have been ranked based on number of citations per year, which will limit the effect of the absolute number of citations. However, more recently published articles may still not have had adequate time for inclusion in the top 100. Recently published articles with sufficient citations may have added importance, particularly those in Table 3. Although citation analysis is a frequently employed tool in scientific literature evaluation, number of citations may not necessarily correlate with a study's impact and is an imperfect proxy for determining significance. [17] This analysis is unable to control for disproportionate citation due to institutional, language, or self-citation biases. Of note, several studies included subjects with non-malignant disease, resulting in a mixed study population. This may limit the ability to directly infer all findings to an exclusive cancer population.
By examining the 100 most-cited palliative oncology studies in 1995-2005 and 2006-2016, we were able to identify and analyze research trends across the literature. Existing palliative care studies in oncology largely focused on multiple-as opposed to singlecancer types, addressed the impact of palliative care on communication and end-of-life decisions, and were rarely RCTs. Early questions of pain, symptoms, and survival in palliative oncology have been replaced with studies addressing access to care and healthcare utilization. The relation of religion and spirituality to the field also has recently become a frequent area of investigation. Such information may prove useful in directing future study and highlighting the ongoing need for high-quality clinically-based research.

Conclusions
Palliative oncology care is increasingly being delivered as a distinct service, ushering in a new era of research questions. A unique billing code for palliative care will allow investigators to study questions related to adoption of palliative care and its effect on healthcare utilization/cost of care using electronic medical record data and large datasets of medical claims. Oncology practices are now integrating patient reported outcomes into routine clinical care. This will provide a rich source of relevant, patient centric data where palliative oncology-specific quality metrics can be developed and tailored to individual cancer type and clinical setting (e.g. urban vs. rural, academic vs. community). Best illustrated by the seminal paper by J.S. Temel et al. (2010), more insight about the application of palliative oncology can be obtained from disease-specific randomized control trials. Ideally such rigorous research will propagate beyond a select number of well represented institutions and countries to incorporate the global medical community.

Additional files
Additional file 1: Table S1.

Funding
This study received no funding.

Availability of data and materials
The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.
Authors' contributions MSB originated the study. CE compiled and analyzed the study data. MSB and CE wrote the manuscript. NB aided in dataset design. EP and RR reviewed and edited the manuscript. All authors have read and approved the final manuscript.