Decentralizing referral prioritization to general practitioners at the primary care level: A qualitative case study based on the Grounded Theory

Abstract

BACKGROUND:

Patient referral prioritizations is an essential process in coordinating healthcare delivery, since it organizes the waiting lists according to priorities and availability of resources.

OBJECTIVE:

This study aims to highlight the consequences of decentralizing ambulatory patient referrals to general practitioners that work as family physicians in primary care clinics.

METHODS:

A qualitative case study was carried out in the municipality of Rio de Janeiro. The ten health regions of Rio de Janeiro were visited during fieldwork, totalizing 35 hours of semi-structured interviews and approximately 70 hours of analysis based on the Grounded Theory.

RESULTS:

The findings of this study show that the obstacles to adequate referrals are beyond the management of vacancies, ranging from the standardization of prioritization criteria to ensuring the proper employment of referral protocols in diverse locations assisted by overloaded health workers with different backgrounds and perceptions. Efforts in decentralizing patient referral to primary care still face the growing dilemmas and challenges of expanding the coverage of health services while putting pressure on risk assessment, as well as sustaining the autonomy of physicians’ work while respecting the eligibility when ordering waiting lists.

CONCLUSION:

A major strength of this work is on the method to organize and aggregate qualitative data using visual representations. Limitations concerning the reach of fieldwork in vulnerable and hardly accessible areas were overcame using snowball sampling techniques, making more participants accessible.

Keywords

Waiting lists health systems resilience health care reform ambulatory care risk assessment qualitative research

1 Introduction

Primary healthcare systems are critical to providing adequate care to local populations [1, 2]. Following this approach, primary care clinics are responsible for assessing patients’ demands, solving them if possible, and referring more complicated demands to specialized procedures, exams, and ambulatories. Thus, in strong public health systems, primary care is responsible for coordinating assistance, prioritizing the use of resources, and referring patients to other levels of the system [3, 4].

Extensive debate has been carried out on the role of primary care in coordinating assistance, especially concerning delivery of specialized services. While some advocate that centralizing patient referral to specialized procedures in primary care overloads primary care clinics and workers, others emphasize the benefits of holding primary care accountable for organizing patient referral, since primary care is the closest focal point for patients, holds strong familiarity with the health conditions of the assisted populations, and is able to strengthen bonds with the communities [5].

Differently from the original conceptualization of centralized referral, a recent reform in Rio de Janeiro’s health system decentralized patient referral to general practitioners (GPs) who work as family physicians in primary care clinics. Therefore, among many other tasks and clinical responsibilities, the GPs also became responsible for patient referral [6 –8]. Decentralizing referral means that family physicians will organize their patients’ priorities, rather than sending requests to a central referral unit. In a decentralized referral process, primary care clinics employ their own referral processes, coordinate, and organize their waiting lists, and, therefore, refer their patients according to their priorities and cope with the availability of services. When centralized, the requests are handled by a referral center that organizes the queues based on the information that clinics provide on the conditions of each patient.

These new work activities brought some degrees of variability and flexibility. Therefore, adjustments are required to keep the system functioning accordingly. In fact, adaptations in performance are paramount assets for proper functioning. However, feedback and communication are necessary with those involved in implementing this system — managers and physicians — to ensure optimal performance, as performance adaptations and variability may result in both beneficial and adverse outcomes [9]. Understanding and reflecting on such adaptations made at the sharp end of the system prevent a common idea of seeking adequate functioning by constraining performance variability and the use of non-technical skills [10 –12].

The referral prioritization reform carried out in Rio de Janeiro expanded the access to health services significantly. The access to primary care services, which was approximately 8% in the 2000 s, rose to almost 70% in the second half of the 2010 s. However, despite the increase of access towards a more universal system, the corresponding workload constrained the system, jeopardizing the quality of assistance. Therefore, discussing the effects of decentralization on workers is necessary.

This paper explores the challenges in decentralizing patient referral to GPs through a case study in Rio de Janeiro, Brazil, to provide reflections on how decentralized referral affects the outcomes and the resilience of the overall health system. Describing worker performance and highlighting how adaptations change the system from the way it was designed on the blunt end (e.g., policymakers, managers, service providers) bridges the gaps that hinder new processes, technologies, and support devices aimed at coping with variability, unpredictability, and adaptation. This study relies on theoretical models of everyday practices, including variations inherent to complex systems. The Grounded Theory framework [13] was chosen as the methodological background for this study to enable the emergence of new theories along the data collection cycles. This background proved to be suitable to the research problem given the complexity and variability involved in the process of referral prioritization, in which health workers are active players and the context promotes different strategies and solutions. Thus, different workers perform patient referrals based on different ideas and representations, and a methodology capable of representing such a scenario is needed.

2 Methods

This is a qualitative case study that follows the Grounded Theory [13], taking as major sources of information a literature review (grey and scientific), observations in 10 primary care clinics, and semi-structured interviews with GPs based on the participant clinics. The literature review included scientific and official technical documentation of different typologies, including policies, resolutions, manuals, protocols, and processes to understand the recommended referral activities. The contents of semi-structured face-to-face interviews with the GPs were analyzed using qualitative analytical tools and coded as concept maps. This methodology enabled knowledge representation using graphical structures. For explicit and comprehensive description of the study’s findings, this article was elaborated according to the consolidated criteria for reporting qualitative research (COREQ) checklist [14].

2.1 Research settings

The municipality of Rio de Janeiro has the second largest population in Brazil in an area of 1,224.56 km² organized in 10 health regions. The population distribution is not homogeneous among health regions. Densities vary from 2,246 inhabitants per km² in the smallest region to1 21,731 inhabitants per km² in the largest region.

According to the Brazilian Policy for Primary Health Care, each team covers from 3,000 to 4,000 patients. Since most primary care clinics are in poorly developed areas, like slums, most patients are impoverished, and the vulnerability of families covered by each team determines the team size. Approximately 950 teams are responsible for covering an estimated 3,305,100 registered patients, and health teams are responsible for geographically delimited target populations [15, 16]. Each health region hosts a Regional Referral Coordination.

2.2 Participant recruitment and data collection procedures

We conducted observations and semi-structured interviews with GPs in 10 primary care clinics. This work studied at least one clinic per health region and at least one GP per clinic. We also performed two pre-testing interviews with two physicians in management positions.

Once interviewed, participants could recommend new participants. This process, known as snowball sampling [17], is useful for studying groups that are difficult to access. It allows for permanent information gathering, which benefits from social networks of the participants that increasingly extend the potential contacts until saturation.

Through snowball sampling, 16 GPs were invited to participate during four months of field work in the 10 clinics. Among participants, 14 worked exclusively as GPs, eight worked as both GPs and medical residency supervisors, and one worked as a GP and as a responsible technician. Five participants hold positions as GPs, residency supervisors, and technicians. In total, 35 hours of interviews were recorded. The analysis required approximately 70 hours. All participants signed informed consent forms that included authorizations for recording their interviews. At least two researchers conducted each interview. Besides recording (when authorized), researchers also took field notes during each interview and subsequently summarized each interview’s main findings.

2.3 Semi-structured interview script

The interviews were organized in two topics: first, the participants were asked to talk about the context of patient referral, including the scenario pre-reform (when familiar with it) and the changes with the decentralization reform; secondly, the participants were asked to describe their work processes, practices, and strategies.

The research team designed subtopics to which the conversation during the interviews could possibly branch, to cover the essentials for adequate understanding of work in patient referral, as described below:

The context of patient referral

Did the decentralization contribute to the ordering of the system and to the coordination of care? In what way?

How could the decentralization be improved?

What difficulties are still observed, even after decentralization?

The practice of patient referral

How does referral occur? (risk assessment > authorize requests > schedule procedures)

What is crucial for referral to occur effectively?

How does the risk assessment occur? What are the criteria? Who participates in it? Is there a relationship with patient screening?

How does the clinic and/or team organize itself to account for the workflows?

Are the criteria described in the protocol aligned with the practice? To what extent do they aid or hinder your decision-making?

Are decisions made collaboratively? When is this factor most present?

What strategies do you and the other members of your team employ to overcome challenges?

Do these strategies have obstacles or risks? What could be improved to support them?

How do information systems help or hinder referral? Suggestions?

How much time is left in the week for you to refer patients? What would be the ideal timeframe to exclusively perform referrals?

Which singular cases are most impactful?

2.4 Data analysis and coding procedures

Concept maps were developed to depict the contents of interviews using Novak’s methodology [18]. The maps were written on two different levels. The first level aimed at creating a general map of expected functioning according to the literature review.

The second level, carried out to code the contents of the interviews, focused on two questions: “How does patient referral occur in your clinic?” and “What is your opinion on the decentralized patient referral in Rio de Janeiro?” A direction map (Fig. 1) guided the basic structure of data coding.

Fig. 1

Partial view of the direction map.

The contents of the interviews were organized into six sub-topics: Actual functioning; Obstacles; Support mechanisms; Opportunities for improvement; Workload; and Information technology. “Actual functioning” focuses on analyzing variability between clinics, strategies employed, and the organization of the referral system. “Obstacles” highlights constraints in referral as described by the participant GPs. “Support mechanisms” presents factors that assist the referral activity. “Opportunities for improvement” describes possibilities for future solutions. The “Workload” concept displays physical and cognitive harms to GPs. Finally, the “Information technology” concept highlights software and applications used.

3 Results

3.1 Expected functioning

The general map of expected functioning (Fig. 2) presents the ideal functioning of referral systems according to the literature on referral standards.

Fig. 2

The concept map of expected functioning.

Family physicians based on primary healthcare clinics request a referral for a specialized outpatient procedure for their patients using specific software. Then, the assigned GP decides the referral status (options include “deny,” “return,” “schedule,” or “toggle pending”). When denying a request, GPs must highlight eventual mismatches between the request and the requirements for the appointed vacancy. When returning, GPs indicate that there is something unclear about the request.

Scheduling an appointment is dependent on availability. GPs assess the request’s urgency and determine its priority according to specific criteria. In some clinics, the prioritization criteria are reviewed daily. When scheduling, GPs choose the closer location and sooner appointment. When no suitable vacancies exist, the request is toggled pending. The referral protocol recommends reassessing pending cases on hold for more than six months. Once scheduled, patients receive text messages to confirm dates. GPs send schedule reports to health teams, and then community health workers start communication and guidance procedures and deliver the referral documents to patients.

3.2 Actual functioning

The actual functioning, or “work-as-done,” encompasses adaptations in work activities done by GPs and other team members to operate the system. Regarding the complexity of the referral process in primary care clinics, the findings of this study highlight the importance of dividing tasks among doctors, management, and other teams, as shown in Fig. 3.

Fig. 3

Partial view of the actual functioning.

When a GP is also the responsible technician, more autonomy for making decisions and managing internal workflows is entailed. GPs ground their actions on information, protocols, and directions from existing policies. However, these protocols and guidelines entail interpretation, constant adaptations, and improvisation. Some adaptations in the work processes are described in Table 1.

Table 1

Adaptations and recommendations due changing in work processes

Roles and work processes changes due decentralization
Description of adaptations	Recommendations
GPs and residency supervisors organize the workflow during the week. Responsible Technicians are entitled more responsibility, e.g., checking availability for all pending requests. Therefore, the Responsible Technician handles the most challenging cases and audits the quality of requests	Redesign GPs’ and RTs work processes to cope with decentralization issues
Each team is responsible for referring their own patients, which minimizes the discussion between the GPs and the family physician, preventing eventual denials. Groups discuss specific cases, e.g., alternatives for low availability, information from counter-referrals, etc. Scheduling usually occurs during regular appointments	More integration between GPs, family physicians and specialists
Residency supervisors and GPs typically refer their own patients only, while Responsible Technicians refer any patients from their clinics. However, in some clinics, all GPs refer the entire clinic	Discussions and training enabling a more consistent referral practice
In some clinics, staff members perform referrals in the referral software. A clinic with seven family health teams and four residency supervisors (one is a Responsible Technician) performs referral together with GPs in the same room at the same time (usually 8 am)	Reflect upon different practices among clinics and improve in information and communication technologies

Residency supervisors spend time analyzing and verifying the software during their shifts, especially when they are aware of new vacancies. During appointments, family physicians can ask GPs to schedule their patients by the end of the appointment as well. Sometimes, GPs also aid assistance (either covering the absence of a resident or fulfilling the care schedule) and referrals during appointments. Scheduling usually occurs 15 days to a month prior to a consultation to minimize absences. Table 2 depicts strategies to reduce cancellations and vacancies.

Table 2

Bottlenecks in schedules minimizing cancels and vacancies

Bottlenecks in the supply and distribution of vacancies
Description of adaptations	Recommendations
A note is shared with the patient informing them of cancelation protocols	More efficient communication to inform patients
Scheduling urgent cases account for patients’ proximity to the primary care clinic	Improvements in the priority criteria considering vulnerability and localization
Scheduling to the closest know vacancy	Improvement in the software referral system

During referrals, GPs make decisions for scheduling patients based on the location of the service provider and the available dates, combined with prioritization criteria, the risk of absence, and the need for a community health worker to accompany the patient during the procedure. In some clinics, the requesting doctor is informed of scheduling, and, in others, there is little monitoring. Table 3 shows adaptations on how confirmed appointments have been monitored.

Table 3

Monitoring confirmed appointments

Monitoring confirmed appointments
Description of adaptations	Recommendations
Staff members staple the appointment receipts for confidentiality and distribute it to the team	Include as software feature
Patients with a registered telephone number receive a text message confirming the appointment	Use social networks e.g. WhatsApp
Staff members deliver the appointment receipt daily to the teams, and community health workers deliver it to patients	Integration of CHWs in referral processes in all clinics
Each team has autonomy to deliver and store receipts	Discussion on variations among clinics, include in the software systema
Requests and print screens of the referral software tabs are stored in the team’s folder. As appointments are scheduled, staff members order them by date. Then, CHWs deliver them to patients within a month of the procedure	Improve software system functionalities to monitor appointments
A staff member prints the receipts and attaches them to referral forms and cancellation alert forms. Managers store this paperwork in a folder associated with the patients’ respective community health workers. Every morning, community health workers check their folders and convey schedules to patients	Improve software system functionalities avoiding paperwork

Table 4 shows the strategies that are employed to improve communications that the GPs and the health teams have with patients.

Table 4

Improving dialog among GPs, health teams and patients

Communications between GPs, Health teams and communities
Description of adaptations	Recommendations
Community health workers are responsible for delivering information regarding the request to the patient. If patients cannot be found at their residences, community health workers reach out via telephone asking them to come to the clinic to get the forms. This occurs when the time between scheduling and performing the procedure is little	Improve ways for communications between clinics and patients, e.g., using social networks
Paper forms delivered to patients for appointments and rescheduling	Improve access in vulnerable communities
Community health workers typically deliver the forms at patients’ residences, but when residences are unreachable for some reason, teams can choose alternative methods to deliver the forms	Improve software and communication system to use social networks and digital forms

Many procedures were often unavailable due to a lack of providers and physicians. When vacancies open in such situations, the few offered procedures available in the system close quickly, and scheduling is often circumstantial. Some strategies GPs use to mitigate these bottlenecks are shown in Table 5.

Table 5

The strategies used to overcome schedule issues in some procedures

Strategies for schedule issues in low available procedures
Description of adaptations	Recommendations
Discussing directly with specialists for urgent cases. Some of these specialists are outside the system, but part of the requesting physicians’ or the clinics’ GPs’ personal contact network	Reflect upon ethics referral criteria within the set of clinics
GPs train family physicians to increase their abilities to solve low-complexity cases at the clinic, aiming to reduce referrals and to save resources for high-complexity cases	Improve the resolutive capacity in family clinics
Negotiations to refer patients to private providers based on personal contacts	Develop procedures to integrate private providers
Consultation of the Regional Referral Coordination to help solve complex cases.	Improve communication between clinics and Regional Referral

3.3 Obstacles

Figure 4 summarizes the obstacles that hamper coordination of care under which the adaptations emerge to keep the decentralized system functioning.

The map of Fig. 4 shows many new work activities that GPs needs to add to their normal clinical practices to cope with decentralization processes, which, even with all these obstacles, have shown significant improvement in referral numbers in Rio de Janeiro. Beside the adaptation issues depicted in previous tables, Fig. 4 also points out that GPs without background in family and community medicine appear to have a lower capacity for solving prevalent health problems, leading to unnecessary referrals and more returns, which delays processing times. It is the responsibility of primary care to bridge communication and coordinate a multidisciplinary effort for each patient. However, communication gaps are frequent.

Fig. 4

A partial map view highlighting the “obstacles” set of concepts in a concept map.

Fig. 5

A partial map view about support mechanisms.

Fig. 6

A partial view of the opportunities for improvement.

Other major obstacles include:

Nonexistence of a formal period to perform referrals.

Difficulty of referring patients during or shortly after their medical appointment (due to the tight schedule).

Unpredictability of vacancies (may occur between 8 h and 20 h).

Accumulation of tasks (especially for responsible technicians).

3.4 Support mechanisms

Findings highlight three significant major support mechanisms: the Regional Referral Coordination, WhatsApp groups, and continuing education. The results indicate that support from the Regional Referral Coordination is essential to the organization of internal flows.

Although informal, WhatsApp groups circulate information on availability, and GPs receive warnings when a sparse vacancy becomes available after a cancellation. Concerning continuing education, training proved to be useful to improve the quality and reduce refusals. Types of useful training include the residency on family and community medicine.

3.5 Opportunities for improvement

Suggested changes to the referral system are organized proposals as follows:

GPs’ work processes.

Improvements in information and communication technologies.

Integration between GPs and specialists.

Consistent referral practice.

Resolution of bottlenecks in the supply and distribution of vacancies.

Informing the patient.

Training.

Improvements in the priority criteria.

Regarding the work process, GPs lack specific time for referring during working days, increasing both the GPs’ workload and the risks of inequity. It is noteworthy, however, that these modifications could delays GPs if there is no synchrony between the work shift dedicated to referrals and the vacancies’ opening hours. To mitigate information gaps and enable tracking of patients’ flows throughout the care network, referrals and respective counter-referrals should be computerized. Moreover, formal practices to increase dialogue between family physicians, GPs, and service providers must be adopted, such as discussions of clinical cases and the definition of new protocols and flows, enhancing transparency and trust in the system.

3.6 Workload

The large number of assignments prevents discussions of cases with specialists, making it difficult to resolve cases related to some bottlenecks. Several clinics indicated that the need for working after hours was more evident when vacancies opened at midnight, a situation that no longer occurs, as shown in Fig. 7. Although the current window is narrower for vacancies (most vacancies open at 8 a.m.), there is still some uncertainty on the opening hours of the remaining vacancies, which might occur from 8 a.m. to 10 p.m. This increases workload since the opening of vacancies must be monitored throughout shifts.

Fig. 7

A partial map view about workload.

There are still other activities that increase workload, such as the need to reinsert requests in the software, and the return of requests to applicants after long waiting times. It is worth mentioning that this practice, common in many clinics, occurs because the software does not allow GPs to update pending requests. It is also worth remembering that the software does not queue up the waiting requests. Finally, patients’ misunderstandings about vacancies are also a factor in raising the workload of the GP. Participants report that many times, even after clarifications, the difficulty of scheduling specific procedures creates obstacles in the relationship between the patient and the clinic’s health teams.

3.7 Information technology

The low interchangeability of the software with other systems, especially with electronic medical records, is a problem. The referral software issues the form with appointment data, while the electronic medical record issues the referral receipt, and both documents are necessary for referring the patient to other levels of care, as shown in Fig. 8.

Fig. 8

A partial map view about information technology.

There were no significant reports of problems such as crashes and bugs. However, there were criticisms regarding usability. It is also worth highlighting the difficulties in tracking absences and non-attendance. Also, regarding the traceability of requests, it is not clear how to handle returned requests in the systems. Regarding the view of available vacancies, participants state that the ordering of vacancies is confusing due to the large amount of data.

4 Discussion

As part of a broad health reform, the decentralization of referrals explored in this case study was an important factor in increasing the coverage of primary care. Between 2008 and 2013, the coverage of primary care in Rio de Janeiro went from 3.5% to approximately 40%. In 2017, coverage was up to 70% [19, 20]. Moreover, with the referral decentralized, family physicians get closer to GPs who perform referrals.

4.1 Adaptation, situated standardization, and resilience

Our study showed that clinics manage the referral process differently, and decentralization hampered the employment of basic referral standards for criteria, processes, and accountability. Some clinics concentrate referral responsibilities on the responsible technician. Others included GPs in the referral and reserve a specific time for referring on different days of the week. Some organize referring teams. Such diversity shapes the GPs’ work as much as the context elements, resulting in a kind of situated standardization, where referral criteria and work processes differ among clinics. Our results indicated that to keep the system functioning and to improve their own referral processes, workers developed ad hoc adaptations, in many ways, in several work situations. Such adaptions are needed even to cope with limitations and drawbacks of available technology to support decentralized referral. Such adaptations enable a certain kind of resilient performance, which was not fully reconciled with prescribed or imagined work and, sometimes, not even perceived by managers. To achieve more resilient performance, section 3.5 lists some recommendations.

4.2 Training and selection of family doctors

A better training and even selection of family doctors with background in family and community medicine is needed because the physicians’ formation and training also shape how physicians understand patient referral. This specific expertise affects physicians’ mindset and their perceptions on ethical principles on primary care activities. Formation on community medicine enables more consistent referral practice among different clinics, as already indicated in other studies.In the same way, gaps in GPs’ backgrounds in community medicine bring the challenge for continuing education of new doctors within the clinics, not only about the relationship between clinical cases and the priority criteria within communities, but also about primary care and its importance in the coordination of care, as studies in the recent literature show [23, 24].

4.3 Referral decision-making

Adaptations in decision-making are natural and even required for resilient performance in complex systems like healthcare, especially when covering large and vulnerable populations. However, reflections on how such ad hoc adaptations impact the overall results of referral process are needed. Such results are not only the number of procedures realized; aspects like equity, communities’ vulnerability, workload of practitioners, and many other issues influence referral decision-making. One of the main findings of this study concerning the disparities in prioritization refers to the struggle of management levels in ensuring equity while expanding coverage, as well as sustaining at the same time that the queues are organized according to severity and eligibility [23]. Moreover, it is necessary to understand and reflect upon ad hoc practices to maintain adequate integration between different levels of care, and centralized referrals aim to mitigate fragmentation, enforce coordination, and implement strategies to control the provision of health services [25].

4.4 Information and communication technologies for referrals

Our results also pointed out the need for improvements in information and communication technologies, and efforts to have more informed patients are needed. The difficulties are proportional to the complexity of cases and availability of service providers. The more complex the cases, the more they rely on strong communication between the patients’ physicians, GPs, and providers. This is particularly challenging because dialogue between the family physicians and the provider is limited, and GPs are usually overloaded. Patients themselves are usually responsible to communicate procedure outcomes and the indications for diagnosis and treatment to their family physicians. This gap in information between the GPs and service providers hampers the closure of the patient’s diagnosis and treatment, increasing waiting times, especially in more complex cases [26, 27]. Moreover, gaps of information vacancies for bottlenecked procedures appear unexpectedly, contributing to non-optimal prioritization [28, 29].

4.5 More than clinical priority criteria for referrals?

Another issue of our study involves improvements in priority criteria. The foundations of the referral lie on clinical priority criteria. In developing countries, issues related to the social vulnerability of communities and patients are essential in assessing patients’ conditions and shape priorities as well [30, 31]. Although the clinical protocols for assigning priority criteria are relevant and practical in most cases, they are not suitable to every situation, as recent literature describes [32 –35]. Long waiting times, far from what’s recommended by known protocols, concentrate referrals on specific kinds of requests, usually simpler ones, putting pressure on the upper levels of care and undermining the resilience of the system [36, 37]. Our results indicated that decentralization entitled GPs based on primary care clinics to schedule specialized procedures directly, streamlining referrals and increasing the coverage, but also generated competition among them, jeopardizing risk assessments. This type of adaptations must be managed to avoid compromising the overall functioning of the system, and procedures to mitigate unwanted results should be discussed and encompassed into decentralized referral protocols.

Furthermore, ethical issues are involved in assigning protocol-coherent priorities, although flexibility is useful in cases of higher complexity and social vulnerability [31 , 39]. This study shows that decentralization of referral enforces local ethics, but also creates conflict and mistrust between different locations, in addition to putting pressure on GPs. However, managing the adaptations accordingly may lead to significant improvements combined with the necessary expansion of health coverage, especially in low-income locations.

4.6 Limitations and validity of the qualitative approach

The Grounded Theory is a systematic methodology that has been largely applied to qualitative research, as it involves the construction of hypotheses and inductive reasoning incrementally throughout data collection and analysis of data. Like usual studies based on the Grounded Theory, this study began with a question and subsequent collection of qualitative data. As the research team reviewed the data collected, new concepts emerged. The researchers highlighted such concepts with different codes and grouped them into diverse levels and categories that enabled new hypotheses and theories.

Thus, the qualitative approach of the present study allowed the identification of processes that were outside the previously planned attributions, pointing out innovations and possible adjustments to be made in health policies. Likewise, it allowed the indication of experiences to be shared, encouraging the formulation of adaptations that generate more efficiency in each territory. In addition, the grounded approach allowed to identify system adjustments that may promote changes in health policies for widespread implementation and reflections on the experiences of different clinics and practitioners, validating adaptations that generate more efficiency and reflecting on the ones that jeopardize the referral system.

In order to mitigate the eventual tendency to commit the first conceptualizations, two rounds of data analysis were carried out. The first deductive round of the analysis ensured that only relevant cases are included for a more inductive analysis in the second round. The second round of inductive analysis allowed codes to emerge directly from the data, basing the inferences on the empirical findings. This is necessary to prevent misleading interpretations. This two-step analytical process allowed the resulting analysis to be aligned with existing concepts and theories and ensured the validity of the methodological background.

The limitations of the study did not hinder the development and outcomes of the research. However, the difficulties of exploring the field and understanding the activity and compiling the data of this complex system were highlighted. Naturally, the validity of results from a descriptive study is a function of the representativeness of the sample. This study presents results from a diversity of primary healthcare clinics studied throughout all health regions of Rio de Janeiro. Thus, although the sample size limits the ability to draw conclusions on the prevalence of certain perceptions in the entire population of GPs, most key elements of the system’s operation that showed relevance to a broader picture were mapped.

5 Conclusions

Understanding performance adaptations is useful to improve health systems. In this study, we aimed to understand how the patient referral system operates and described how practitioners adjust their work to overcome shortcomings in a decentralized scenario.

A major strength of this work is that it describes a method to organize qualitative data in concept maps that focus on a specific topic and aggregate pertinent data. This unique visual representation facilitated analysis and can be employed in further studies. The major limitation concerned fieldwork since the areas explored were vulnerable and hardly accessible. Moreover, family physicians at work are difficult to access due to the exhausting nature of their work. To overcome these limitations and expand the reach of fieldwork, the snowball sampling proved to be useful in making more participants accessible and available.

Despite the positive opinions of all participants regarding the decentralization of referrals, there is a need for more studies that broaden and deepen the findings of this study. The topic of patient referral, as a strategic element in coordination of care, would make good use of further research due to its magnitude and its impact on the health system and the lives of low-income populations.

Ethical approval

This study was carried out according to the Resolution from the Brazilian Health Council #510/2016 (https://conselho.saude.gov.br/resolucoes/2016/Reso510.pdf) that regulates ethics procedures in scientific research with human subjects in Brazil. Moreover, this article is part of a research project that has been approved by the Institutional Review Board (IRB) of the Oswaldo Cruz Institute (opinion #.994.410; https://plataformabrasil.saude.gov.br).

Informed consent

All participants declared their agreements in participating in this study on signed consent forms.

Conflict of interest

The authors have no competing interests to declare.

Footnotes

Acknowledgments

None to report.

Funding

Alessandro Jatobá is partially funded by the National Council for Scientific and Technological Development - CNPq (grants #307029/2021-2 and #402670/2021-3) and the Carlos Chagas Filho Foundation for Supporting Research in the State of Rio de Janeiro - FAPERJ (grant #E-26/201.252/2022). Paulo V. R. de Carvalho is partially funded by the National Council for Scientific and Technological Development - CNPq (grant #304770/2020) and the Carlos Chagas Filho Foundation for Supporting Research in the State of Rio de Janeiro - FAPERJ (grant #260003/001186/2020). Paula de Castro Nunes is partially funded by the Inova Fiocruz Program (grant #310815559697153).

References

Starfield

. Primary Care: Concept, Evaluation, and Policy. USA: Oxford University Press; 1992.

Starfield

, Shi

. Policy relevant determinants of health: An international perspective, Health Policy. 2002;60:201–18. doi: 10.1016/S0168-8510(01)00208-1.

Starfield

. Reinventing primary care: Lessons from Canada for the United States, Health Affairs 2010;29:1030–6.

Bousquat

, Giovanella

, Campos

EMS

, et al. Primary health care and the coordination of care in health regions: Managers’ and users’ perspective, Ciência & Saúde Coletiva.-54 2017;22:1141–54. doi: 10.1590/1413-81232017224.28632016.

Costa do

N R

. Community health workers’ attitudes, practices and perceptions towards the COVID-19 pandemic in brazilian low-income communities. IOS Press. https://content.iospress.com/articles/work/wor205000 [Accessed 3 Feb 2021].

Pinto

, Soranz

, Scardua

, et al. Ambulatory municipal regulation of the Unified Health System services in Rio de Janeiro: Advances, limitations and challenges, Ciênc saúde Coletiva 2017;22:1257–67. doi: 10.1590/1413-81232017224.26422016.

Silva Junior

, Guabiraba de

KPL

, Gomes

, et al. Outpatient regulation in Primary Care in the municipality of Rio de Janeiro, Brazil, based on the local regulatory doctors, Ciênc Saúde Coletiva 2022;27:2481–93.

Soranz

, Pisco

LAC

. Primary health care reform in the cities of lisbon and rio de Janeiro: Context, strategies, results, learning and challenges, Ciênc Saúde Coletiva 2017;22:679–86. doi: 10.1590/1413-81232017223.33722016.

Bulhões

, Bellas

, Arcuri

, et al. Shifting the management model of Brazilian health services: Perceptions of major stakeholders on the participation of the private sector in public hospital administration, Dialogues in Health 2022;1:100011. 10.1016/j.dialog.2022.100011.

10.

Arcuri

, Bellas

, Ferreira de

, et al. On the brink of disruption: Applying Resilience Engineering to anticipate system performance under crisis, Applied Ergonomics 2022;99:103632. 10.1016/j.apergo.2021.103632.

11.

Hollnagel

. Safety–I and Safety–II: The Past and Future of Safety Management. 1st ed. CRC Press; 2018. doi: 10.1201/9781315607511

12.

Bellas

, Bulhões

, Arcuri

, et al. Community health workers–non-technical skills for delivering primary health care in low-income areas. Work. published online first: 19 May 2022. doi: 10.3233/WOR-205089.

13.

Glaser

, Strauss

The Discovery of Grounded Theory: Strategies for Qualitative Research. 1st ed. Routledge; 2017. doi: 10.4324/9780203793206

14.

The EQUATOR Network. Consolidated criteria for reporting qualitative research (COREQ): A 32-item checklist for interviews and focus groups; 2015. https://www.equator-network.org/reporting-guidelines/coreq/ [Accessed 19 Oct 2022].

15.

Acosta Ramírez

, Giovanella

, Vega Romero

, et al. Mapping primary health care renewal in South America, Family Practice 2016;33:261–7. doi: 10.1093/fampra/cmw013.

16.

Pessoa

, Rigotto

, Carneiro

,et al. Meanings and methods of territorialization in primary health care, Ciênc Saúde Coletiva. 2013;18:2253–62. doi: 10.1590/S1413-81232013000800009.

17.

Goodman

. Snowball sampling, The Annals of Mathematical Statistics 1961;32:148–70.

18.

Novak

. Learning, Creating, and Using Knowledge: Concept Maps as Facilitative Tools in Schools and Corporations. 2nd ed. Routledge; 2010. doi: 10.4324/9780203862001

19.

Soranz

, Pinto

, Penna

. Themes and reform of primary health care (RCAPS) in the city of Rio de Janeiro, Brazil, Ciênc Saúde Coletiva 2016;21:1327–38. doi: 10.1590/1413-81232015215.01022016.

20.

Melo

, Mendonca de

MHM

, Teixeira

. The economic crisis and primary health care in the SUS of Rio de Janeiro, Brazil, Ciênc Saúde Coletiva. 2019;24:4593–8. doi: 10.1590/1413-812320182412.25432019.

21.

Marchand

, Peckham

. Addressing the crisis of GP recruitment and retention: A systematic review, Br J Gen Pract 2017;67:e227–37. doi: 10.3399/bjgp17X689929.

22.

Finn

, Brown

MEL

, Laughey

. Holding a mirror up to nature: The role of medical humanities in postgraduate primary care training, Education for Primary Care 2021;32:73–7. doi: 10.1080/14739879.2020.1816860.

23.

Deslauriers

, Raymond

M-H

, Laliberté

, et al. Prioritization of referrals in outpatient physiotherapy departments in Québec and implications for equity in access, Bioethics 2018;1:49–60. doi: 10.7202/1058251ar.

24.

Déry

, Ruiz

, Routhier

, et al. Patient prioritization tools and their effectiveness in non-emergency healthcare services: A systematic review protocol, Systematic Reviews 2019;8:78. 10.1186/s13643-019-0992-x.

25.

Batista

, Vilarins

GCM

, Lima de

, et al. The Regulatory Complex for Health Care in the Federal District, Brazil and the challenge for integrating levels of health care, Ciênc Saúde Coletiva 2019;24:2043–52. doi: 10.1590/1413-81232018246.08132019.

26.

Oliveira

CCRB

, Silva

EAL

, Souza de

MKB

. Referral and counter-referral for the integrality of care in the Health Care Network. Physis. 2021;31. doi:10.1590/S0103-73312021310105.

27.

López-Vázquez

, Pérez-Martínez

, Vargas

, et al. Barriers and factors associated with the use of coordination mechanisms between levels of care in Mexico. Cad Saude Publica. 2021;37. doi: 10.1590/0102-311X00045620.

28.

Schulz

, Messikh

, Reboul

, et al. Characteristics of outpatients referred for a first consultation with a nephrologist: Impact of different guidelines, J Nephrol 2022;35:1375–85. doi: 10.1007/s40620-021-01204-w.

29.

Chang

, Allen

, McClung

, et al. Criteria for referral of patients with advanced heart failure for specialized palliative care, Journal of the American College of Cardiology 2022;80:332–44. doi: 10.1016/j.jacc.2022.04.057.

30.

Bellas

, Jatobá

, Bulhões

, et al. Effects of urban violence on primary healthcare: The challenges of community health workers in performing house calls in dangerous areas, J Community Health 2019;44:569–76. doi: 10.1007/s10900-019-00657-2.

31.

Jatobá

, Bellas

, Koster

, et al. Patient visits in poorly developed territories: A case study with community health workers, Cognition, Technology & Work 2018;20:125–52.

32.

Zwi

, Joshua

, Moran

, et al. Prioritizing vulnerable children: Strategies to address inequity, Child: Care, Health and Development 2015;41:827–35. doi: 10.1111/cch.12265.

33.

Adamson

, Barrass

, McConville

, et al. Implementing theaccess and waiting time standard for earlyintervention in psychosis in the United Kingdom: An evaluation ofreferrals and post-assessment outcomes over the first year ofoperation, Early Intervention in Psychiatry 2018;12:979–86. doi: 10.1111/ei12548.

34.

Marzouk

, Kelley

, Fadhil

, et al. “If I have a cancer, it is not my fault I am a refugee”: A qualitative study with expert stakeholders on cancer care management for Syrian refugees in Jordan, PLOS ONE 2019;14:e0222496. doi: 10.1371/journal.pone.0222496.

35.

Martinez

, Zhang

, Bastias

, et al. Prolonged wait time is associated with increased mortality for Chilean waiting list patients with non-prioritized conditions, BMC Public Health 2019;19: 233. 10.1186/s12889-019-6526-6.

36.

Neimanis

, Gaebel

, Dickson

, et al. Referral processes and wait times in primary care, Canadian Family Physician 2017;63:619–24.

37.

Lewis

, Harding

, Snowdon

, et al. Reducing wait time from referral to first visit for community outpatient services may contribute to better health outcomes: A systematic review, BMC Health Serv Res 2018;18:869. 10.1186/s12913-018-3669-6.

38.

Bellas

, Arcuri

, Ferreira de

D S

, et al. Complex systems design based on actual system functioning: Coping with variability in a national water ambulances service, Work 2022;73:S265–77. doi: 10.3233/WOR-211211.

39.

Arcuri

, Bulhões

, Jatobá

, et al. Gatekeeper family doctors operating a decentralized referral prioritization system: Uncovering improvements in system resilience through a grounded-based approach, Safety Science 2020;121:177–90. doi: 10.1016/j.ssci.2019.08.023.