Methods used to identify the prevalence of palliative care needs: An integrative review

Objectives. Early identification of palliative care (PC) needs is crucial to provide appropriate holistic care to patients. The objective of this integrative review is to synthesize the methods used to identify the prevalence of PC needs. Methods. An integrative review search of the Cumulative Index to Nursing and Allied Health Literature (CINAHL) Plus with full text, ProQuest, Wiley InterScience, ScienceDirect, Scopus, PubMed, and Web of Science with publications from 2010 to 2020 was carried out in English. Empirical studies examining the methods used to determine the prevalence of PC needs were included. The methods of data extraction of the included articles were categorized by data source, study setting, and data collector. Quality appraisal was performed using QualSyst. Results. Of the 5,410 articles screened, 29 were included in this review. Two articles identified the prevalence of PC needs in a community that was supported by a network of volunteers, while 27 studies considered this at a continent, country, hospital, and/or primary care facility level as represented by physicians, nurses, and researchers. Significance of results. Various methods have been used to determine the prevalence of PC needs, and the outcomes are valuable for policymakers in developing PC services when allocating resources at the national and community levels. Future research to identify PC needs across health settings, especially primary care facilities, should consider providing PC across a spectrum of care settings.


Introduction
In 2014, the World Health Assembly agreed to integrate palliative care (PC) as a component of the comprehensive care provisions of a health system needed by all citizens. This decision necessitates PC access within the definition of universal health coverage (The Worldwide Hospice Palliative Care Alliance 2014). In addition, the International Association for Hospice and Palliative Care has, through consensus, revised its definition of PC to "the active holistic care of individuals across all ages with serious health-related suffering due to severe illness and especially of those near the end of life" (Radbruch et al. 2020). When PC is considered from an epidemiological perspective, the focus moves away from addressing cancers in response to severe illnesses and chronic conditions with distinct illness trajectories (Gó mez- Batiste et al. 2013). Therefore, the concept of PC shifts from being previously an emphasis on the older adult, provided at the end of life and primarily to individuals with cancer, to now a "palliative approach to care, " which starts earlier in the illness trajectory and is not only limited to cancer but providable for people of all ages. A person can begin to receive PC services early on their life-threatening illness to manage their symptoms and improve quality of life.
PC must, therefore, be provided within all health-care settings via the integration of systems, organizations, administrative means, and clinical provisions if patients' and their families' quality of life are to be appropriately maintained (Brazil 2018;Radbruch et al. 2020). Zheng et al. (2013) indicated that 25% of patients with cancer, 81% of patients with organ failure, and 80% of patients with frailty/dementia were not identified for PC. This might have led to the limited palliative provisions.
Early identification of people with PC needs is essential if appropriate care is to be offered, which can serve complex needs covering the remaining life of patients, as well as the bereavement of their families (Teike Lüthi et al. 2020). Various methods exist to identify PC needs.
2 Nithra Kitreerawutiwong et al. Swarbrick et al. 2019). PC needs must be reported in the form of prevalence based on the concept of public health-oriented PC.
In a given population, prevalence can be calculated using several methods. The estimation, in this context, demonstrates the routine use of mortality statistics to estimate population-based PC needs (Higginson 1997;Kane et al. 2015;Sleeman et al. 2019). Some studies used the International Statistical Classification of Diseases and Related Health Problems-10th Revision (ICD-10) (Etkind et al. 2017;Fumaneeshoat 2018;Khalid and Chong 2019;Rosenwax et al. 2005). In one study, patients with chronic progressive diseases and a limited life prognosis were considered, and for people with all causes of death, it was estimated that 75% of all deaths lacked PC (Etkind et al. 2017). Moreover, the study of Yang et al. (Lan Yang SL et al. 2020) applied the methods from Murtagh et al. (2014) and Gó mez-Batiste et al. (2014) to project national PC needs.
In addition, some studies have conducted cross-sectional surveys on the prevalence of PC needs (Amblàs-Novellas et al. 2016;Daya et al. 2017;Gó mez-Batiste et al. 2013. However, these studies have utilized different tools, facility approaches, and data collectors. Two systematic reviews and one literature review reported on the screening tools that were employed and the accuracy gained (ElMokhallalati et al. 2020;Healthcare Improvement Scotland 2020;Teike Lu¨Thi et al. 2021). However, evidence regarding the methods used to identify PC at the community level is limited. These methods are crucial in different settings and are valuable in designing integrated PC services that enable the achievement of quality outcomes for patients in need of PC. The objective of this integrative review is to synthesize the methods used to identify the prevalence of PC needs.

Study design
Integrative reviews were conducted to summarize past empirical literature and provide a more comprehensive understanding of the selected topics. This study determined the topics used to identify the prevalence of PC needs. This study was conducted in accordance with the steps of problem formulation and literature search, data evaluation, and data analysis, (Whittemore and Knafl 2005) with presentation of the data in the results section.

Problem formulation and literature search
This step aims to clearly state the topic of interest by listing the variables of interest. The areas of the methods are available through which the prevalence of PC needs is determined as the core concept. An integrative search was performed in the following databases: Cumulative Index to Nursing and Allied Health Literature (CINAHL) Plus with full text, ProQuest, Wiley InterScience, ScienceDirect, Scopus, PubMed, and Web of Science. The search term keywords included "palliative*, " "hospice*, " "terminal care, " "supportive care, " "end of life care" alongside "prevalence, " "epidemiology, " "how many, " "need*, " "demand, " "method, " "approach, " "procedure, " and "instrument. " Keywords were identified through the titles and abstracts of the articles found via a preliminary search. The indexed terms from the selected databases were identified and included in the search terms for the utilized databases (Table 1). Articles published between 2010 and 2020 were reviewed. This study began in October 2020. We searched the previous 10 years of published articles as 10 years of research was determined to have value cocreation in dealing with this still emerging concept and clarifying insights into its definition, foundations, drivers, related processes, expected consequences, and measurement (Leclercq et al. 2016). The search results were imported into Endnote 7 to enable management. Duplicate references were addressed in the Endnote Library by K.K. Four investigators (N.K., K.K., O.K., and S.M.) independently screened the titles and abstracts of the searched articles. The inclusion criteria were that each article must have been peerreviewed, written in English, published between 2010 and 2020, have a primary outcome pertaining to the prevalence of PC needs, and include prospective and/or retrospective cohort studies, casecontrol studies, and cross-sectional studies. Review articles and studies published as chapters in a book along with opinion/editorial/conference publications were excluded. Articles whose full text could not be accessed were excluded. Finally, studies that included non-PC patients or that focused on populations with specific symptoms or syndromes (e.g., patients with pain, depression, or delirium) were excluded. Disagreements in the screening process were resolved through discussions and consultations with a third researcher (R.T.). The excluded studies and the main reasons for exclusion were recorded in a separate database.

Data evaluation
The identified research articles were assessed using QualSyst software. The checklist for assessing the quality of quantitative studies consists of 14 items, with scoring being measured in relation to the degree to which specific criteria are met ("yes" = 2, "partial" = 1, and "no" = 0). Items not applicable to a particular study design are marked "n/a" and excluded from the calculated summary score. A summary score is calculated for each paper by summing the total score obtained across all relevant items and then by dividing the total possible score [i.e., 28 -(number of "n/a" × 2)] (Leanne et al.   2004). The assessment was completed independently by 4 investigators (N.K., K.K., O.K., and S.M.). The scores obtained were compared and discussed to reach a consensus.
A standardized system, with scores ranging from 0 to 2, was used to assess all aspects of an article's research quality. The total possible scores (28 for quantitative studies and 20 for qualitative studies) were then converted to a standardized score of 1. A QualSyst cutoff score of 0.55 was chosen to capture 75% of the articles considered, thereby ensuring that valuable data were included. Following this quality scoring, the final selection of articles was reviewed by the authors, thereby guaranteeing that each article's inclusion was based on its originality and relevance. The interrater agreement for the overall scale range of each article ranged 0.86-1, with the discrepancies in the overall scores having ranged 0.02-0.10.

Data analysis
Data were simplified using the 2 methods of primary and secondary data collection to identify PC needs. The data were displayed in the format of a table to show the connection with the topic. The data of this study were extracted to accomplish the aim of determining the methods available to identify the prevalence of PC needs. The table designed was populated to denote each article's author, year of publication, country of focus, objective, data collection methods, subgroup of the setting, approach (e.g. facility, hospital, primary care, and/or community level), and data collector. The authors (N.K., K.K., O.K., and S.M.) extracted the data and R.T. reviewed the data. A narrative synthesis was conducted. Subsequently, the findings of this study are presented. The data were verified in the results section, including methods, setting, and tools and data sources of data collection regarding PC prevalence.

Results
The search yielded 5,410 articles, of which 3,441 were removed following the elimination of duplicate titles using Endnote (K.K.). Furthermore, 45 articles were removed after discovering that they were published more than 10 years ago. According to the inclusion and exclusion criteria, 1,895 articles were excluded and 29 studies were eligible, and they were included for quality appraisal (Fig. 1). In terms of the quality of the paper was reported inter-rater agreement for overall scale 0.86−1.

Methods of data collection as to PC prevalence
The methods used to gather information fall between the primary and secondary data collection approaches. Thirteen studies were undertaken via the collection of primary data, whereby various tools and data collection methods utilized were supported by a network of volunteers in 2 studies and by trained health-care professionals in 11 studies. The remaining 16 studies were undertaken via the collection of secondary data, as analyzed and interpreted primary data utilized recorded hospital data and national statistics. Therefore, PC needs were calculated by the researcher(s). One study employing secondary data reported a global estimate (Connor et al. 2017), while one study reported an estimate for 12 European and non-European countries (Morin et al. 2017).

Setting of data collection as to PC prevalence
Regarding the data collected on PC needs prevalence, 13 studies undertook primary data collection, while 16 studies produced an estimation. Two of the 13 primary data collection studies emphasized community-collected data for all populations of the considered households, while 11 pertained to data derived from hospitals and primary care facilities. Of these 11 studies, 7 collected data in the context of a hospital and 3 collected data in the context of a general practitioner clinic (as a primary care facility) in Japan (Hamano et al. 2018), primary care centers, a district general hospital, social health centers, nursing homes in Spain (Gó mez- Batiste et al. 2014Batiste et al. , 2012, and a nursing home in Belgium (Hermans et al. 2017). One study was conducted at a hospital and primary care facility (as offered primary care, social health, and nursing home services) (Gó mez-Batiste et al. 2012). For the secondary data collection studies, 2 studies pertained to a hospital (Hua et al. 2014;Wirasorn et al. 2016) and 14 studies reported on demographics at the population and country levels (Aristodemou and Speck 2017;Bowers et al. 2020;Calvache et al. 2020;Connor et al. 2017;Engeser et al. 2020;Etkind et al. 2017;Fraser et al. 2020

Tools and data sources of the data collection as to PC prevalence
In primary data collection studies to determine the prevalence of PC needs at a community level, 2 examples were undertaken in India by a trained team of medical social workers, medical interns, nursing students, postgraduates, and faculty from the department with the support of a network of volunteers in the villages (Daya et al. 2017;Elayaperumal et al. 2018). The study of NECPAL (from Necesidades Paliativas in Spanish [Palliative Needs]) can identify patients in need of PC in both primary and secondary care facilities, as recorded by trained doctors and nurses (Daya et al. 2017;Gó mez-Batiste et al. 2014, 2012. The other instruments available are the Supportive and Palliative Care Indicators Tool (SPICT TM ), the Palliative Outcome Scale (POS), surprise questions, admission characteristics, and examined outcomes, the Sheffield Profile for Assessment and Referral to Care (SPARC), Gold Standards Framework (GSF), and PALLIA-10 (Table 2) (Creutzfeldt et al. 2015;Gardiner et al. 2013aGardiner et al. , 2013aHamano et al. 2018;Hamdi et al. 2018;Hermans et al. 2017;Molin et al. 2019;Niekerk and Raubenheimer 2014;Ryan et al. 2013).
In contrast, estimations given as to the prevalence of PC needs via the use of secondary data sources employ data that is already collected by investigator and organizations, wherein emphasis is given to the population and/or country level. The data sources used here depend on the health data of each country that may comprise administrative health data by calendar year, the ICD-10 code relating to life-limiting conditions eligible for PC, World Health Organization (WHO) mortality data from the Global Health Estimates: Causes of Death 2000-2011, UNAIDS data on HIV prevalence, United Nations population data by country (for ages 0-19) and year, World Bank country population estimates for 2010, national inpatient hospital data, hospital episode statistics as to the admitted patient care, Office for National Statistics mortality data, and death certificates. Calculations through populationbased methods as to estimating PC needs can be derived from (1) age-and sex-specific proportions of deaths from defined chronic progressive illnesses, (2) multiplying the number of people dying from a particular condition by the percentage of pain prevalence for the condition, (3) projected mortality data, combined with projected population data and observed individual-level data from a prospective longitudinal study on aging, (4) Higginson, (5) Rosenwax, (6) Gó mez-Batiste, and (7) Wirasorn et al. 2016). The studies were grouped according to their author/year/country, objective, and data collection methods (Table 3).

Discussion
Prevalence is an epidemiological measure that represents the proportion of a given population affected by a certain condition (Fletcher et al. 2014). The present research focuses on the prevalence of PC needs and knowledge, which are needed to reflect the importance of such services being provided under Human Right to Health. Prevalence estimates are also important for healthrelated decision-making and such outcomes must correspond with national policies, programs, resources, and PC-related training for health professionals, volunteers, and families (World Health Organization 2021). This conclusion is consistent with previous research, which has indicated that the identification of PC patients can help health-care providers assess the need to develop a care plan (Amblàs-Novellas et al. 2016).
The methods of collecting data on the prevalence of PC needs, with the use of primary and secondary data sources, depend on the objective of each study. Primary data are collected to address a specific problem at hand using the fit that best responds to that problem. By contrast, secondary data are collected for purposes other than the problem at hand (Hox and Boeije 2005). The present study was supported by primary data collected regarding the prevalence of PC needs, produced in community, hospital, and primary care facility contexts, in an effort to provide appropriate PC services that respond to the needs faced. For example, at the community    level, such information will contribute to the training of volunteers, families, local organizations, and health-care professionals in accordance with their respective roles and responsibilities. The prevalence of PC needs identified at each level can be used to design PC delivery, such as community-based (CBPC), hospital-based, or primary care facility-based PC (such as that offered in hospices or nursing homes). The model produced for each PC level corresponds to numerous interdisciplinary and discrete tasks and activities associated with the delivery of PC. This is reflected in the work of Bhavsar et al. (2017), who focused on CBPC. This study reported that physicians spent 94% of their time on tasks related to patient care and 1% on administrative tasks. In contrast, nurse practitioners and registered nurses spent 82% and 53% of their time on patient-related tasks and 2% and 37% on administrative tasks, respectively. For the primary data collected at the hospital level, this study reported 7 studies relevant to hospital-based PC services proposed by the WHO. This concept covers services provided via an outpatient PC clinic, PC consultation service for hospital inpatients, PC day-care service, inpatient PC unit, and PC outreach/home care provision. Several types of staff are required to provide these services including doctors, nurses, auxiliary/palliative nurse aides, and pharmacists. In addition, psychologists or counselors, social workers, and volunteers should be involved. Hospital-based PC also facilitates discussion on the values, diagnosis, prognosis, and agreements of patients alongside the goals of the care offered (World Health Organization 2016). Moreover, Paramanandam et al. (2020) found that collaboration between a hospital's PC team and CBPC results in high-quality transitions across care settings and a reduction in acute care utilization.
Through primary data collection at the primary care facility level, this study reported on 3 studies undertaken in (1) general practitioner clinics in Japan (Hamano et al. 2018), (2) primary care centers and 1 district general hospital, 1 social health center, and 4 nursing homes in Spain (Gó mez-Batiste et al. 2012), and (3) a nursing home in Belgium (Hermans et al. 2017). These primary care facilities serve the needs of older adults in countries having the highest proportion (28.2% of the total population aged over 65 years): 28.0% in Japan, 19.1% in Spain, and 18.7% in Belgium (Population Reference Bureau 2021). The context of the increasing older adult populations invokes the need for residential care homes and/or assisted living facilities. The identification of PC needs for these older adults living in these facilities and accessing other provisions is important in the early phase (Galiana and Haseltine 2019). However, the need for PC is not restricted to older adults; yet, this demographic requires greater access to PC than other groups (Daya et al. 2017;Elayaperumal et al. 2018). This is consistent with the general regime and practice of geriatric PC (Goldhirsch et al. 2014). In line with the concept of home-based PC, CBPC and aging within the community encourage the provision of PC (Galiana and Haseltine 2019) in places other than hospitals.
For the primary data collection undertaken at the community level, this study reported on 2 studies relevant to the CBPC as proposed by the WHO. This concept holds that CBPC should be offered at primary care facilities (such as community health centers, sub-district health promoting hospitals, and nursing homes) and should involve people in the community as witnesses of the facilitation of PC assessments, planning, implementation, resource mobilization, and collaboration with local health authorities. This approach shall allow community organizations to play a role in establishing PC services in their community and for these efforts to be integrated into the wider health system (World Health Organization 2016). The involvement of volunteers in assessing PC needs within a community must arise alongside a system designed to enable assessment tools to be used, and the roles and responsibilities of PC teams should be clarified based on the concept of intersectoral collaboration and the appropriate training of relevant teams. Additionally, one study developed a tool to identify PC needs in both primary care facilities and hospital settings (Gó mez-Batiste et al. 2012).
When considering the data collectors of the primary data studies considered here, it was found that 11 studies were undertaken in relation to primary and secondary care facilities by trained healthcare professionals (e.g., doctors and nurses). This reflects how PC, in this setting, is provided by trained staff at primary care facilities and hospitals. However, collaboration among health-care professionals working in community hospitals at the district level occurs through the production of discharge plans by health-care professionals operating at the primary care level. Therefore, volunteers working in communities are required. Khongsrichay et al. (2020) indicated that an obstacle faced by nurses working at the primary care level is the lack of knowledge development in PC. Thus, greater training of health-care professionals in PC is needed if PC provisions are to be comprehensively integrated into health-care systems. Furthermore, it was found that 2 studies utilizing primary data were carried out by trained volunteers with this activity being in concordance with CBPC, whereby people in a community are involved in such assessment processes (World Health Organization 2016). Data collected by trained volunteers will be the input of primary care facilities, whereas the latter will design PC services. In other words, through primary health care enacted alongside community participation, appropriate technology that is compatible with local, cultural, and economic conditions can emerge. Simultaneously, responsive tools and processes can be maintained and operationally controlled by the local population, while intersectoral collaboration in the community should be strengthened. The information derived from and contributed by each community shall therein be culturally and demographically responsive and will acknowledge the resources available therein, thereby providing solutions that are community-centered and advocating the continuity of care across all levels (Calvache et al. 2020) The secondary data approach used to estimate the prevalence of PC needs has mostly been undertaken at the country or continent level (Aristodemou and Speck 2017;Bowers et al. 2020;Calvache et al. 2020;Connor et al. 2017;Engeser et al. 2020;Etkind et al. 2017;Fraser et al. 2020;Kane et al. 2015;Maetens et al. 2019;May et al. 2019;Morin et al. 2017;Murtagh et al. 2014;Scholten et al. 2016;Swarbrick et al. 2019). Information on the overall prevalence of people with PC needs can, therefore, be positioned to determine a country's PC-related policy, resource/funding allocation, and workforce training (among both professionals and lay people). Additionally, secondary data sources derived from hospitals can pinpoint the prevalence of PC needs (Hua et al. 2014;Wirasorn et al. 2016). To understand the trends of PC needs, from which processes can manifest and appropriate methods are developed through which PC needs within hospitals are determined, the accessibility of PC must be comprehended. Furthermore, secondary data (or routine data), including those derived from death registries, hospital activity records, primary care data, and specialist PC registers, are widely underutilized. The benefits of secondary data pertain to this existing information and are often formatted on a population basis. The combination of existing data pertaining to given areas, services, and demographics may be vital in contributing to a comprehensive understanding of the current context of such health care and where improvements may be made or are required. Patient-centered outcome measures are increasingly seen as the gold standard for measuring quality of care; therefore, person-level data must be collected uniformly over time. On the other hand, the use of secondary data encounters a number of challenges, including varying quality of the data, associated safety and ethical issues, lack of linkage with other information, and non-specificity of PC and end-of-life care needs (Davies et al. 2016).
Data collected on PC needs prevalence via different means allows population-based estimations at national or continent levels to emerge through which determination can be given as to implementing responsive policies and resource allocation. In this sense, primary data sources from primary, secondary, and tertiary care facilities can be used to improve PC services and determine policy and resource management at the facility level. Primary data from community-level sources can be facilitated by primary care facilities to improve their services by using a network of community PC volunteers and working together with health-care providers in the community, such as providing information on population illness in the community (Daya et al. 2017;Elayaperumal et al. 2018) and caring and spiritual supporters of PC patients. The diversity of methods used to identify PC needs is crucial for the comprehensive assessment and treatment of pain and other physical, psychosocial, and/or spiritual problems (World Health Organization 2016). The PC needs must be identified to manage resources such as those of health-care professionals (in terms of the extent and quality of their training), volunteers, medical equipment, information, and pertinent networks. The methods of determining PC needs in a given context will vary depending on whether the given PC program is designed to be community-, facility-(primary care facility and hospital), or country-based. This will impact PC assessment, planning, implementation, and evaluation processes designed and operated.

Strengths and limitations of the present study
The strengths of this study pertain to its focus on addressing the methods used to identify the prevalence of PC needs. The integrative review method that was employed has set out stringent inclusion and exclusion criteria enacted by 2 researchers and was then independently reviewed. When disagreement arose regarding the inclusion or exclusion of an article, justification and discussion were raised in collaboration until a consensus was reached. This process was designed to minimize potential bias. Furthermore, a comprehensive evaluation has been given to the methods used to identify the prevalence of PC needs via various levels of coverage (e.g. continental, country, facility, and community). However, the present integrative review had several limitations. First, in addressing the prevalence of PC needs presented in each context and the economic disparities across populations, generalization needs to be considered. Second, language bias must be acknowledged as only papers published in English were included in this integrative review.

Implications for further research
This study suggests that most of the research undertaken with secondary data sources (12 studies) pertains to the country level, whereas research undertaken with primary and secondary data (12 studies) pertains to hospital-level data. The determination of PC needs at the community level was found in only 2 studies. Thus, there is a need to enhance the provision of PC within communities, integrate this effort with national health systems, and strengthen collaboration between health sectors, local organizations, and the people sector. Moreover, the prevalence of PC needs should be reported across all levels of care, including community, primary, secondary, and tertiary care, in order to design integrated PC.

Conclusions
This integrative review revealed that there are several methods through which data can be collected regarding the prevalence of PC needs, including primary and secondary data collection. The benefits of each method correspond to their respective purposes of utilization (e.g., policy development or practice improvement). Here, data on the prevalence of PC needs can be collected at continental, country, facility (primary care facility and/or hospital), and community levels. The tools and data sources accessible in this context demonstrate the involvement of trained health-care providers, researchers, and support from trained volunteers. The prevalence of PC needs was recorded as the population proportion (i.e. per 100 and 1,000). A bottom-up approach is used to understand the prevalence of PC needs spanning both the community level and primary care facilities. Thus, there is a need to develop CBPC services designed to be integrated with national health systems.