Study population and data collection
Phase I – Theoretical Phase: The theoretical phase comprises four sub-projects.
First, an
organizational ethnographic field approach will be applied to gain insights into the practices involved in delivering primary PC and to analyze barriers and facilitators. Data will be collected by means of document analysis (such as care plans, case notes, registers); by work shadowing (observation of daily routines, team meetings, staff interactions, home visits); and by interaction with staff (informal interviews, cognitive interviewing / think-aloud technique) [
28]. Field notes will be taken according to a pre-defined scheme [
29] and will be transformed into accounts including information about procedures, observations of actions and interactions, quotations of conversations, interpretations (e.g. of underlying attitudes), and the researcher’s intuitive thoughts and memos. These accounts will be imported into data files using software for qualitative text analysis (MaxQDA®).
Second, a secondary analysis of qualitative data [
30] will be undertaken with a particular focus on barriers and facilitators of PC provision by GPs. Data are available from a recently concluded project on end of life care for frail older patients in family practice (ELFOP), a longitudinal study on needs, appropriateness and utilization of services (funding number: BMBF-FK 01 GY 1120) funded by the Federal Ministry of Education and Research. Participants were selected through theoretical sampling to include GPs with different backgrounds, working conditions (single or joint practice), and working in rural and urban areas characterized by diverse care infrastructures. Both the interviews and the group discussion include relevant information on the GPs’ perception of tasks, problems, challenges in the end-of-life care for frail older people, as well as on their perceived roles, responsibilities, and professional self-perception within this context of care.
Third, based on the findings from ethnographic research and secondary interview analysis focused interviews will be conducted with GPs, GP assistants or nurses from home care teams or inpatient service provider, patients, and their relatives (Table
1) using the
Critical Incident Technique (CIT) [
31,
32] to identify significant mechanisms that result in excellent practice versus in undesirable or even harmful events.
Table 1
Target/ study population detailed for each work package of data collection
Ethnographic research | GPs; GPs assistants, professional caregivers | n = 30 | The Institute of General Practice’s education and research network of n = 250 general practices; theoretical sampling contrasting: rural vs. urban; single vs. joint practice; specific palliative education vs. nonspecific palliative education |
Secondary analysis | GPs, GPs assistants, professional caregivers | interviews: n = 52 group discussions: n = 3 | Existing data set of interviews / one focus group with GPs conducted within the ELFOP project; all interviews and the focus group are included |
CIT | GPs, GPs assistants, professional caregivers, patients and relatives | n = 16 | Theoretical sampling contrasting: rural vs. urban; diagnosis; single vs. joint practice; men vs. women |
Analysis of Health insurance claims data | GPs’ patients identified as being in need of PC | not applicable | Health insurance claims data: all AOKN and BARMAR insured persons in Lower Saxony with palliative billing codes |
PAR | GPs, community nursing teams, patients and informal carers, other relevant PC providers | not applicable | Theory-based purposive sampling [ 55] |
Delphi study | GPs’ professional representatives and other relevant experts for health policy decision making and healthcare planning | n = 30 | DEGAM; DGP; GPs’ chairmen / quality circles; the Institute of General Practice’s close links with GPs’ professional boards and associations (Hausärzteverband) |
Consensus workshops | GPs’ professional representatives; health insurers; experts for medical education; and other relevant experts for health policy decision making and healthcare planning | n = 20 | Two groups with approximately 10 participants each; theory-based purposive sampling [ 55] |
Implementation of the intervention package | GPs’, community nursing teams, health insurers; and other relevant experts for health policy decision making and healthcare planning | not applicable | Exemplary test region in Germany (Lower Saxony) involving n = 8 general practices, as well as all relevant service providers and stakeholders in the respective region |
Questionnaire survey (Summative evaluation) | GPs’, community nursing teams, health insurers; and other relevant experts for health policy decision making and healthcare planning | n = 130 | Service providers, insurers, and other local stakeholders involved in the implementation of the tailored intervention package (Phase II) and from two exemplary other regions in Germany |
Patient-relevant outcomes (Summative evaluation) | Patients and informal carers | standardized scales n = 32 interviews n = 16 | Patients and informal carers attended by general practices who are involved in the implementation of the tailored intervention package (Phase II) and from two exemplary other regions in Germany |
Fourth,
secondary analysis of health insurance claims data of the AOK – Die Gesundheitskasse für Niedersachsen (AOKN, statutory health insurance) and BARMER (statutory health insurance), and from the GPs’ office software will serve to describe the morbidity and service usage of patients, assess patient-relevant outcomes such as hospital admissions, estimate the time investment and costs for the provision of primary PC by GPs, and to draw inferences on the relevant laws, professional regulations, and reimbursement systems (see Table
2).
Table 2
Outcomes, comparators, and data sources
Patient-related outcomes |
Quality of life | Physical and psychosocial wellbeing, social support, financial strain | Standardized quality of health / care related measures such as POSa; QODDb; QUELCc |
Quality of care | Unplanned hospital admissions | Health insurance claims data |
Service usage and time investment |
Service usage | GPs’ consultations and home visits; prescriptions (medication, medical aids, interventions); referral to specialists and other service providers; out-of-hours contacts; emergency interventions | Health insurance claims data GP’s office software |
GPs’ time investment | Time dedicated to assessments, home visits, telephone consultations, physical care, and psychosocial care | Time registration (ethnographic field research) |
Impact on GPs’ and practice staff’s ability and capacity to deliver primary palliative care |
Quality of palliative care delivery | GPs’ and practice staff’s appraisal of the quality of PC provided to their patients | Organizational ethnographic field research including individual and group interviews |
Quality of care other than palliative care | GPs’ and practice staff’s appraisal of the quality of care provided to other patients | Organizational ethnographic field research including individual and group interviews |
Work satisfaction | GPs’ and practice staff’s work-related satisfaction and sense of meaningfulness | Organizational ethnographic field research including individual and group interviews |
Economic evaluation |
Direct costs for the healthcare system | Service usage (consultations and home visits, prescriptions of medicines and medical aids, specialist referrals, social services, out-of-hours contacts, emergency interventions) | Health insurance claims data GP’s office software |
Direct costs for patients and relatives | Additional costs for medications and therapeutic interventions not covered by the health insurance funds | Organizational ethnographic field research including individual and group interviews |
Indirect costs for patients and relatives | Social isolation; (temporary) work loss; abandon of activities or relationships | Organizational ethnographic field research including individual and group interviews |
Direct costs for GPs and other healthcare providers | Costs invested for PC not remunerated within the medical compensation system | Organizational ethnographic field research including individual and group interviews |
Indirect costs for GPs and other healthcare providers | Personal strain; impact on healthcare professionals’ private lives; abandon of activities or relationships | Organizational ethnographic field research including individual and group interviews |
Feasibility of the intervention package |
Process monitoring and formative evaluation of the implementation phase | Normalization Process Theory core constructs (coherence, cognitive participation, collective action, and reflexive monitoring) | NoMADd assessment instrument for the evaluation of the implementation of complex interventions |
Phase II – Modelling Phase: The modelling phase consists of three sub-projects building up on each other.
First,
Participatory Action Research (PAR) [
33] will be undertaken with general practices, community nursing teams, patients and informal carers, as well as other relevant PC service providers to define adequate operationalization of PC processes and quality outcomes, and to condense a set of service development statements [
34,
35].
Second, a nation-wide consensus Delphi study will be undertaken with GPs and their professional representatives such as the GPs’ professional federation, the Medical Chamber, and the German association of general practice (DEGAM) to seek expert consensus on these service development statements.
Third,
consensus workshops will be conducted in the follow-up of the
Delphi study, using the
Nominal Group Technique [
36] to elaborate a strategic action plan on how to put the service development statements into practice. Workshop participants will be key representatives from the practical, education, economic, and policy sphere, as well as representatives of patients and relatives. Data collected during this phase will be integrated during analysis (see [
2,
37]) to design the tailored intervention package.
Finally, an intervention package will be developed for enhancing the integration of PC provision into general practice, drawing on the example of the Gold Standards Framework for Palliative Care [
12], but adapting it to the German context. A participatory framework and consensus building techniques will be applied to promote ownership of the developed elements and to enhance commitment among the relevant stakeholders.
Phase III – Implementation and evaluation: The intervention package developed in phase II will be implemented and evaluated in an exemplary test region in Lower Saxony.
First, formative evaluation will use an organizational ethnographic approach including individual and group interviews with relevant stakeholders to (a) test the feasibility and acceptance of the intervention in practice, and (b) to compare the intervention to two exemplary other regions in Germany with different approaches to generalist and specialist PC provision in terms of contracts with health insurance funds, different roles of GPs in the local delivery of PC (e.g. Westphalia-Lippe).
Second,
summative evaluation will be accomplished to assess the effects of the intervention on patient care in terms of physical, practical, and psychological outcomes. Data sources will be a questionnaire survey among the relevant stakeholders involved in PC provision; analysis of clinical and health insurance funds routine data to assess the effects on patient care, service usage, and costs of care; and perceptions of patients and relatives regarding the outcomes of care using established instruments such as the Palliative Outcome Scale [
38] (Table
2).
Sample size calculation
Phase I: For the ethnographic field research the participation of
N = 8 GPs, GP assistants and other medical professionals was planned considering methodological arguments of theoretical sampling when contrasting practices characteristics such as rural vs. urban, single vs. joint practice, men vs. women. At least
N = 16 interviews following the Critical Incident Technique were proposed which relates to the methodological experience that data saturation in qualitative analysis via Grounded Theory. Data collection will end once data saturation is obtained which will be the case when no more new or valuable information can be attained, further coding is no longer feasible, and once the study can be replicated [
39]. This usually takes place after 12–15 interviews. The sample size of secondary analysis of qualitative data from ELFOP is pre-determined due to the retrospective design of the sub project. The sample size of the secondary analysis of health insurance claims data is not applicable and will be defined by the amount of specific data sets of the current insured persons of the AOKN and BARMER.
Phase II: Reflecting the current literature on successful Delphi studies,
N = 30 participants who complete each Delphi round are planned. Due to the expected drop-out of 60% during the course of the Delphi assessment, probably three times as many participants will be initially invited to this sub project [
18]. The final consensus workshops aims to include
N = 20 stakeholders following a theory-based purposive sampling [
40].
Phase III: For the questionnaire survey in terms of the summative evaluation a participation of N = 130 key stakeholders of local (primary) palliative care delivery from the practical, education, economic, and policy sphere, as well as representatives of patients and relatives is proposed based on content-related reflections taking the accumulated sample size of the prior subprojects into account.
Methods of data analysis
Phase I + III: Analysis of qualitative data (ethnographic field research, CIT, secondary analysis of interviews and focus group), as well as for the formative evaluation during the
implementation phase (organizational ethnographic approach with individual and group interviews) will be guided by the methodology of Grounded Theory. For each qualitative data-subset (field notes, interviews, focus group), Strauss and Corbin’s single coding paradigm [
41] will be used to identify salient determinants and mechanisms of relevance to primary PC delivery in general practice. To ensure inter-subjective validity and reliability, qualitative data will be analyzed independently by the members of the junior research group and discussed within the team until consensus on core categories will be reached. Analysis of health insurance claims data and GPs’ office software will follow the guidelines for best practice of secondary data analysis [
42]. For quantitative data analysis univariate and bivariate descriptive methods as well multivariate methods (e.g. logistic regression or survival analysis [
43,
44]) of empirical social research will be used. The findings from the different data sources within the
theoretical phase will be triangulated to draw inferences on the structural, economic, and interactional mechanisms that either hamper or facilitate PC delivery in general practice.
Phase II: A characteristic feature of
Participatory Action Research is community and practice partners’ participation also in the phase of data analysis [
45]. Mutual reflection, shared learning and co-construction of meaningful outcomes are key elements of this process; for this aim, pragmatic and comprehensible methods are recommendable such as thematic or content analysis, or the single coding paradigm. A stepwise approach to data analysis will be conducted in working groups composed of academic staff and representatives from the field [
45]. Outcomes of the
Nominal Group Technique employed during the consensus workshops will be analyzed using quantitative (descriptive comparison between ranks and weights of ideas and statements) and qualitative methods (combination of structured-thematic and formal qualitative content analysis) [
46].
The Delphi study and the questionnaire survey will be analyzed using descriptive statistics (percentages, measure of central tendencies such as mean and median values, and measures of dispersion such as variance, interquartile ranges, and standard deviation). For the Delphi study, results of concluded survey rounds will be processed to provide feedback to the participants in the subsequent round [
47]. If necessary, service development statements will be modified based on the experts’ responses in order to inform the questionnaire for the next survey round and hereby improve the degree of consensus within the expert panel. For the development of the tailored intervention package, data from the different sources in the modelling phase will be integrated in accordance with the framework for the design of complex interventions [
26] and the corresponding guidance for synthesizing and triangulating mixed methods data (48). Following the recommended phases of the mixed methods analysis process [
48], a package of expedient interventions will be proposed and elaborated based on the data collected during the participatory process, will be translated into service development statements, and consented during the subsequent Delphi survey and consensus workshops.
Statistical analysis will be compared and contrasted with qualitative data on direct and indirect costs resulting from organizational ethnographic research including individual and group interviews using triangulation. Standardized patient-related outcomes such as Palliative Outcome Scale [
38] or Quality of Dying and Death questionnaire [
49] (Table
2) will be analyzed according to the instructions for scoring, analysis, and interpretation of responses indicated by the authors in the instruments’ manual, on the respective website, or in key publications.
All qualitative analysis will be conducted using software for qualitative text analysis (MAXQDA®). Quantitative data will be analyzed using IBM SPSS Statistics (Version 24) and R (Version 3.3.0).