Step 1. Locating existing theories	An initial programme theory was devised, drawing on the knowledge of the research team. To identify the focus of the review, an exploratory search was conducted using keywords in MEDLINE/PubMed and Google Scholar. Following initial searches, the limited evidence on the management of complicated grief in primary care settings in the UK and other settings was apparent, and thescope of the searches broadened to include general bereavement care.
Step 2. Searching for evidence	The search strategy was conducted with the support of an information specialist (IK). The main search strategy used combinations of search terms for the concepts 'bereavement’, ‘complicated grief’, ‘primary care’, and ‘United Kingdom’ to run four searches (see Supplementary Box S2). In August 2019 the following databases were searched: MEDLINE, Embase, CINAHL (the Cumulative Index to Nursing and Allied Health Literature), PsycINFO, Web of Science, Scopus, ASSIA (Applied Social Sciences Index and Abstracts), Sociological abstracts and SCIE (Social Care Institute for Excellence) Social Care Online.Citation chasing and manual searches of the reference lists of articles and reports were also completed.Following initial screening, the limited evidence on the management of complicated grief in UK primary care was apparent. Therefore, an additional search using a refined search filter for the United Kingdom was undertaken in January 2020 to ensure all relevant UK literature was retrieved.
Screening	Titles and abstracts were screened by CP against the inclusion and exclusion criteria. Inclusion criteria were studies of: (1) adults (aged >18 years) who have lost a significant other and whose grief and/or bereavement experience is considered complicated or prolonged in accordance with a variety of criteria (formal diagnosis of ICD/DSM [International Classification of Diseases/the Diagnostic and Statistical Manual of Mental Disorders] or other not required); and (2) where care is sought and received in primary care settings, focusing on the UK but also including studies from other high and middle income countries where psychosocial conditions can, routinely, be managed in primary care. All study designs were included. Exclusion criteria were studies of neo- and peri-natal bereavement, bereaved children and adolescents, military bereavement, non-death bereavement, grief of healthcare professionals.Judgments on the relevance of the data within these articles for programme theory development were reviewed by the whole project team.
Step 3. Selecting articles	Full-text documents were selected for inclusion based on their ability to provide relevant data to the review. This included articles that directly investigated, reviewed, and discussed how bereavement and complicated grief is managed in primary care settings, in the UK and elsewhere. CP read the full texts and classified them into categories of high and low relevance, based on her judgments on the relevance of the data within these articles for programme theory development. These decisions were then reviewed by SB and GW.At the point of data inclusion based on relevance, when needed, the trustworthiness and rigour of the methods used in each study to gather relevant data was also assessed. One means of assessing rigour was examining whether the study methods had been clearly explained and justified. For example, if using questionnaires or diagnostic tools, then the trustworthiness of data would be considered greater if the questionnaire or tool had been tested and validated.A range of articles were included in the synthesis, including empirical studies, grey literature, and non-empirical articles such as editorials and opinion pieces. Articles were not excluded on the basis of rigour alone if the article contained data relevant to the development of the programme theory.Realist reviews are a type of configurational review, where data are interpreted and used to develop theoretical explanations of phenomena, in this case bereavement and complicated grief in primary care. To build these explanations, even less rigorous data can be useful. Assessments of rigour of included data tend to be undertaken when there is a paucity of data to inform context–mechanism–outcome configurations (CMOCs). It is also equally, if not more, important to judge the explanatory powers of any theories developed, and this was done in the review using the criteria of consilience, simplicity, and analogy at the level of the CMOC and programme theory. A more in-depth explanation and justification of the approach may be found in Wong 2018.85
Step 4. Extracting and organising data	The articles were coded in NVivo (version 12), by CP, and initial codes were discussed and refined through iterative discussions with GW and SB. As analysis continued, a refined version of the programme theory was developed. Codes were developed by referring back to the refined programme theory and codes were applied deductively with new codes created as needed (inductively).A realist logic of analysis was applied to the data and sections of texts found within each broad conceptual theme.
Step 5. Synthesising the evidence and drawing conclusions	Working across and within coded-data extracts, CMOCs were developed as part of an iterative development of causal explanations. This included comparing and contrasting data presented in different articles, and making judgments about whether similarities between findings presented in different sources were adequate to form patterns in the development of CMOCs and programme theory.The final product of this stage was a refined realist programme theory that explained how bereaved patients are identified and managed in primary care. The programme theory formed the basis of the implications for practice.