An overview of systematic reviews on the collaboration between physicians and nurses and the impact on patient outcomes: what can we learn in primary care?

Background Primary care needs to be strengthened in order to address the many societal challenges. Group practices in primary care foster collaboration with other health care providers, which encourages care co-ordination and leads to a higher quality of primary care. Nursing roles and responsibilities expanded over time and nurses have been found to often provide equal high-quality chronic patient care compared to physicians, even with higher patient satisfaction. Inter-professional collaboration between primary care physicians and nurses is a possible strategy to achieve the desired quality outcomes in a strengthened primary care system. The objective of this research is to synthesize the evidence presented in literature on the impact of collaboration between physicians and nurses on patient outcomes in primary care or in comparable care settings. Methods A systematic review of peer-reviewed reviews was performed in four databases: COCHRANE, MEDLINE, EMBASE and CINAHL. All studies from 1970 until May 22 2016 were included in the search strategy. Titles, abstracts and full texts were respectively reviewed. At least two of the three authors independently reviewed each of the 277 abstracts and 58 full texts retrieved in the searches to identify those which contained all the inclusion criteria. Two authors independently appraised the methodological quality of the reviews, using the AMSTAR quality appraisal tool. Results A total of eleven systematic reviews met all the inclusion criteria and almost fifty different patient outcomes were described. In most reviews, it was concluded that nurses do have added value. Blood pressure, patient satisfaction and hospitalization are patient outcomes where three or more systematic reviews concluded better results when physicians and nurses collaborated, compared to usual care. Colorectal screening, hospital length of stay and health-related quality of life are outcomes where collaboration appeared not to be effective. Conclusions Collaboration between physicians and nurses may have a positive impact on a number of patient outcomes and on a variety of pathologies. To address future challenges of primary care, there is a need for more integrated inter-professional collaboration care models with sufficiently educated nurses. Electronic supplementary material The online version of this article (10.1186/s12875-017-0698-x) contains supplementary material, which is available to authorized users.


Background
Populations around the world are rapidly ageing. It is estimated that between 2015 and 2050, the world's population of over 60 year olds' will nearly double from 12 to 22% [1]. As people age, they are more likely to experience several health conditions at the same time. The demand for health care is evolving rapidly in the context of an ageing population and the growing number of people living with one or more chronic conditions [2]. In Europe, patients are more demanding and expect health care to be accessible and high qualitative at the same time [2,3]. Professional caregivers, on the other hand, experience a high workload and demand a better work-life balance [4,5]. At the same time, financial resources in health care are decreasing, while the demand for financial support is increasing [6][7][8]. In an attempt to address these challenges, the following four aims have the potential to guide innovations in health care delivery: improving the health of populations, improving the experience of care, reducing per capita costs of health care, and diminishing the workload for professional caregivers so they can rediscover meaning and joy in their work [7,9,10].
Reforms are shifting care from hospitals to community, partly due to a growing prevalence of chronic diseases [11,12]. In addition, countries in the European Union show many potentially avoidable hospital admissions for several chronic conditions including diabetes mellitus, chronic heart failure, chronic obstructive pulmonary disease and asthma. Potentially avoidable hospitalizations for these conditions are commonly used to measure access and quality of primary care systems [13,14]. In order to address the needs of ageing populations and to reduce the unnecessary use of hospital care, primary care systems should be strengthened [2].
It was suggested that group practices in primary care foster collaboration with other health care providers, which encourages care co-ordination and leads to a higher quality of primary care [8]. Primarily, nurses were introduced in primary care practices to meet a perceived shortage of primary care physicians [15]. Over time, nursing roles and responsibilities expanded. Practice nurses were able to provide holistic care for patients that was not limited to traditional nursing boundaries [16]. Nurses have been found to often provide cost effective patient care and equal high-quality chronic patient care compared to primary care physicians, even with higher patient satisfaction [2,12,16,17]. By expanding the roles and responsibilities of nurses, primary care systems can be strengthened.
Improved inter-professional collaboration is important and diversity of disciplines is needed in a time when the provision of primary health care becomes more complex and one health professional can no longer meet all patient needs [18,19]. As the largest health care workforce group, and because of their specific skills and competencies, nurses are in an ideal position to collaborate with other team members in the delivery of more accessible and effective chronic disease management in primary care. Inter-professional collaboration between primary care physicians and nurses is a possible strategy to achieve the desired quality outcomes in an effective and efficient manner in an integrated health system. Therefore, there's a need to explore to what extent an integration of physician and nurse competencies impacts patient outcome.
The objective of this research is to synthesize the evidence presented in literature on the impact of collaboration between physicians and nurses on patient outcomes in primary care or in comparable care settings.

Data sources
We searched for reviews of the literature containing synthesized evidence relating to collaboration between physicians and nurses, and the impact of their collaboration on patient outcomes.
Searches were performed in four literature databases: COCHRANE, MEDLINE, EMBASE and CINAHL. All databases were searched from 1970 (or from their inception if this was later than 1970) until May 2016. In addition, reference lists of the selected reviews were reviewed to identify other eligible reviews, but no additional review articles were identified.
All detailed search strategies can be found in Additional file 1.
The retrieved references were entered into Endnote© and duplicates were removed.

Study selection
The included studies had to fulfil a number of criteria in order to be included. First, the manuscript had to be a systematic review of the literature. A review was considered a systematic review if two of the following criteria were met: a search strategy was reported, a search was performed in Medline(PubMed) at least, and the included studies were subjected to a methodological assessment. There were no inclusion criteria based upon the research design of the primary research articles included in the systematic reviews.
Second, the manuscript needed to concern 'collaboration between physicians and nurses' in a primary care setting or in a hospital setting. Since there is no generally accepted definition of what inter-professional collaboration means, the intervention was defined as collaboration by the researchers if at least one physician provided care along with at least one nurse.
Third, the outcomes in the reviews needed to concern clinical patient outcomes and/or patient satisfaction outcomes. The review also needed sufficient methodological quality according to the AMSTAR quality appraisal tool (studies with a score ≥ 11 were included) [15][16][17]. And finally, none of the exclusion criteria listed below were met.
Research publications were excluded when they were primary research studies, when they were written in a language other than English or Dutch, or when the setting was considered 'inappropriate'. Settings were defined as inappropriate when the presented patient population was dissimilar or incomparable to the primary care population. Inappropriate settings were determined as; an intensive care unit (ICU), radiology, neonatology intensive care unit (NICU), obstetrics and gynecology. Studies were also excluded when the outcomes merely concerned nurse/physician outcomes.
A four-stage inclusion process was applied. Initially, titles and abstracts of research articles identified from the search strategies were screened, in order to determine their relevance and whether they met the inclusion criteria. No further analysis was done on the subsequent criteria as soon as one criterion was not met. In the first stage, one reviewer screened all references. When the title provided insufficient information to determine inclusion or exclusion, the research article proceeded to the second stage.
In the second stage, two reviewers independently examined all abstracts of the articles selected in the first stage, in order to determine whether they met the inclusion criteria. Any disagreements were resolved by discussion between the two reviewers.
In the third stage, two reviewers independently examined all full texts of the articles selected in the second stage. Any disagreements were resolved by discussion between the two reviewers. If no agreement could be reached, a third reviewer decided.
The final stage of inclusion related to the methodological assessment of the reviews. All reviews remaining after the third stage, were assessed with the AMSTAR quality appraisal tool [15,17]. This assessment tool was formed by combining the enhanced Overview Quality Assessment Questionnaire (OQAQ), a checklist created by Sacks, and three additional items judged to be of methodological importance. 11 different components were identified [15]. The eleven criteria were scored as followed: 2 points were given when the criterion was fully met, 1 point when it was partly met and zero points when it was not met. Therefore, a maximum of 22 points on methodological quality could be achieved (see Table 2). Two reviewers independently examined the methodological quality of the reviews, using the AMSTAR quality appraisal tool [18]. The mean of the scores of the two reviewers was computed and classified as the final quality score [17]. In case the scores of the reviewers differed more than two points, reviewers reached consensus by discussion. Only moderate and high quality reviews (mean scores ≥11) were used for data extraction.

Data-analysis and synthesis
Data were extracted about the search strategies, time frame of the searches, studied interventions, selected outcomes, selected patient populations, selected study setting, the collaboration between physician(s) and nurse(s) and the different nursing roles within the collaboration.
Data-analysis was done primarily by description of the characteristics, interventions and outcomes. Metaanalyses and quantitative assessments from the included reviews were described. No quantitative pooling was performed across the reviews.

Search and inclusion results
After duplicates were removed, the searches in the different databases resulted in one unique database, encompassing 4004 studies. Titles, abstracts and full texts were respectively reviewed and subsequently 277 studies and 58 studies were identified as potentially meeting the inclusion criteria (See Fig. 1). A total of 36 systematic reviews met all the inclusion criteria. Two reviewers independently assessed the remaining 36 reviews on their methodological quality, using the AMSTAR quality appraisal tool. A mean of the two scores was computed and classified as the final quality judgement. Eleven systematic reviews had a mean quality score higher than 11 and were included for data-extraction and analysis.
The flow diagram of the inclusion process is shown in Fig. 1.
Characteristics of the 11 included systematic reviews. Search periods for each systematic review are shown in Table 1.
A narrative overview of the included review articles is described in Table 2. The eleven reviews only included quantitative studies. Four reviews [19][20][21][22] were limited to randomized controlled trials only, while the other seven reviews also included other comparative designs such as controlled before and after studies, interrupted time series and intervention studies. Three reviews included observational studies [23][24][25]. One review author additionally included other systematic reviews [26].
Four systematic reviews performed a meta-analysis [21,[26][27][28]. The methodological quality of the included review articles varies from moderate [20,24] to high [21,28,29]. Nine review articles included studies that were conducted in both a primary care setting and a hospital setting [19-24, 26, 28, 29]. Two review articles included studies that were exclusively conducted in a hospital setting [25,27].
The eleven systematic reviews included a total of 285 different primary studies, the number of primary studies included in the review articles varies from 6 to 69. Most of the primary studies were included only once in a review, with the exception of 12 papers that were included in two reviews. Additional file 2 presents a list of all primary studies included in at least one of the reviews. Table 3 presents the main findings of the metaanalyses. Four different review articles are presented. The table includes: intervention, control group and the different outcomes. The number of studies within the systematic review and the total number of patients are presented, followed by the (weighted median) effect size, a measure of heterogeneity and an appraisal of the quality of evidence/risk of bias (if available). The included systematic reviews provided no information on the performance of a statistical process for small-study effects. The table shows that interdisciplinary teams targeting either informational or management continuity had a positive impact, with a weighted median effect size (95% confidence interval) of respectively 2.0% (−0.03, 3.20) and 2.0% (−1.90, 3.20), on the quality of life of patients diagnosed with cancer. A measure of heterogeneity was not available. The quality of evidence of the included research articles, according to GRADE, was rated very low. Team based models of end-of-life care (home and comprehensive) caused a decrease in the number of people admitted to hospital and an increase of the number of people dying at home. Nurse-coordinated care as well as nurse-managed protocols had a positive effect on patients' blood pressure and caused a decrease in patients' low-density lipoprotein cholesterol levels.   -Significant reduction in HbA1c among diabetes patients.
-Significant increase in the proportion of CAD patients with controlled BP and total cholesterol.
-Significant reduction in hospitalization after 1 year for CAD patients receiving secondary prevention.
-More patient satisfaction with care provided by the nurse plus physician intervention.
-Inconsistency regarding outcomes related to HRQOL.
No outcomes indicated specialized nursing interventions to be more harmful than physicians alone. Unclear role of the specialized nurse, making it difficult to determine the impact on efficiency. More research is needed to better understand the impact of specialized nurses on primary health care efficiency.   -High level of evidence to support equivalent levels of patient satisfaction, self-reported perceived health, functional status outcomes, glucose control and BP control.
-High level of evidence to support equivalent rates of emergency department visits, hospitalization and mortality.
-High level of evidence to support better serum lipid levels.
-Moderate level of evidence to support equivalent length of stay.
-High level of evidence to support equivalent patient satisfaction scores.
-High level of evidence to support equivalent or lower length of stay for patients cared for in the CNS group.
-High level of evidence to support the CNS group has a lower cost of care        Table 3 presents the results of the meta-analyses of four of the included systematic reviews. The different 'collaboration interventions' are presented, followed by the control group, patient outcomes, number of studies, number of patients, effect size, a measure of heterogeneity (if available) and a measure of quality of evidence/risk of bias (if available). The improved patient outcomes are written in bold NA 'not applicable' , To handle the diverse set of outcomes within each individual study, the median value was computed of all the measured effects across all the outcomes of the same class. To pool the results from multiple studies, the median effect size was calculated for each class of outcome, by computing the median from all the median effects in outcomes obtained from individual studies. The researchers chose this pooling strategy to be consistent with the median approach used in other reviews [45][46][47] b non-parametric bootstrap confidence intervals Table 4 presents an overview of the systematic reviews that did not provide a meta-analysis. Seven different review articles are presented. The table includes: intervention, control group and the different outcomes. The number of studies within the systematic review and the total number of patients are presented, followed by a statement on heterogeneity (if available) and an appraisal of the quality of evidence/risk of bias (if available).
All eleven articles describe the impact of collaboration between physicians and nurses on patient outcomes. Table 5 presents an overview of the different outcomes described in the review articles. Table 5 provides an overview of the improved patient outcomes (collaboration between physicians and nurses led to better results for these outcomes), Table 6 shows an overview of the equivalent patient outcomes (collaboration between physicians and nurses led to equal results for these outcomes) and Table 7 presents an overview of the mixed patient outcomes (collaboration between physicians and nurses led to better and/or equal and/or worse results for these outcomes). Blood pressure, patient satisfaction and hospitalization are the outcomes where three or more systematic reviews concluded better results when physicians and nurses collaborated, compared to usual care. Systematic reviews often described a combination of improved and equivalent patient outcomes when the included articles showed mixed results. Table 8 describes the collaboration between physicians and nurses in the different review articles. Collaboration was described as a 'multidisciplinary' , 'inter-disciplinary' or 'inter-professional'. Other health care providers are often part of the team [20,21,26,27]. Figure 2 presents the nursing roles/tasks in the collaboration with physicians in the included systematic reviews. The most frequently represented tasks are: specific nursing tasks (e.g. blood pressure control), communication/consultation tasks (e.g. communication with the multidisciplinary team), patient education tasks (e.g. lifestyle counseling) and coordination/organization/referral tasks (e.g. coordination of care, conducting a discharge planning). Two review articles did not clearly describe the tasks performed by the nurses.

Discussion
Eleven systematic reviews describing the impact of collaboration between physicians and nurses on patient outcome were included in this overview of systematic reviews. Collaboration between physicians and nurses may have a positive impact on a number of patient outcomes and on a variety of pathologies.
Almost fifty different patient outcomes were described (Table 3). In most reviews, it was concluded that nurses do have added value. Maybe we observe some publication bias here since most of the author groups included nurses [30]. We also obtained mixed results in the other reviews. Blood pressure was the only patient outcome exclusively reported as improved in three different systematic reviews [19,21,28]. Two of them even performed a meta-analysis [21,28]. Patient satisfaction is an improved patient outcome as well. No less than five different systematic reviews confirmed this [19,20,22,26,27]. However, two systematic reviews reported an equivalent patient satisfaction when physicians and nurses collaborated [22,23]. Number of hospitalization is another improved patient outcome, confirmed by four different systematic reviews [19,20,22,26]. However, three systematic reviews [20,23,26] also reported an equivalent number of hospitalizations and one [22] even reported an increase of hospitalizations when physicians and nurses collaborated. These mixed results make it difficult to make an accurate interpretation and conclusion towards the different patient outcomes.
Colorectal screening, hospital length of stay and health-related quality of life are three patient outcomes that also improved when physicians and nurses did not collaborate. However, only colorectal screening and health-related quality of life were merely categorized as negative outcomes. Allen et al. reported the length of hospital stay as a negative outcome. But the same review article also reported improvement in length of hospital stay, as well as two other review articles [20,23]. Quality of life in general was reported as an improved outcome when physicians and nurses collaborated in two different review articles [26,27].
The included systematic reviews often combined different interventions such as patient education [22,29], medication adjustment [28], discharge planning protocol and shared decision making [21,22] while measuring patient outcomes. Adding one or more interventions, besides collaboration between physicians and nurses, also makes it more difficult to determine which effect can be attributed to which intervention.
The evidence of collaboration between physicians and nurses on patient outcome can be applied to the primary care setting for almost all the measured patient outcomes. Only two systematic reviews included articles conducted in a hospital setting [25,27]. Therefore, the improvement of global quality of life, and the decline of patient falls and pressure ulcers cannot be allotted to collaboration between physicians and nurses in the primary care setting.

Collaboration
The different systematic reviews used a variety of terms describing the collaboration between health care providers including inter-professional collaboration, multidisciplinary collaboration, coordination, communication, teamwork and shared care. A clear definition and subsequent   Table 4 presents the 'collaboration intervention' , control, patient outcome, number of studies, number of patients (if available), a statement on heterogeneity (if available) and a measure of quality of evidence/risk of bias (if available) of seven included systematic reviews that did not conduct a meta-analysis elaboration of the nature of the collaboration was lacking in most of the reviews. This is consistent with findings in the existing literature, where there seems to be no agreement on the use of terms to describe collaboration between health professionals [31]. This also makes it difficult to know how the collaboration translates itself in daily practice: were the studied collaborations between physicians and nurses merely focused on nurses performing dedicated tasks, based on physicians orders (a rather more instrumental collaboration)? Or were the studied collaborations focused on nurses' competences and tasks with autonomous decision-making capacity, based on structured agreements between nurses and physicians (a rather more integrated collaboration)? A total of 173 RCTs were finally included in this overview of systematic reviews. Although RCTs are the gold standard in establishing a firm evidence base in quantitative research, complex practice settings like health service settings, often require a more diverse methodology [32]. The relationship between teamwork and patient outcomes seems to be difficult to investigate with RCTs. A Cochrane review on interventions to promote collaboration between nurses and physicians concluded that rigorous evaluations are difficult to conduct. This is because the interventions are complex and the intermediate processes are difficult to assess [33]. Researchers in the United Kingdom increasingly use qualitative research methods alongside RCTs to gain a more comprehensive understanding of the impact of health service delivery [32]. Direct observation of collaborative practice in primary care settings holds promise as a method to better understand and articulate the complex phenomena of inter-professional collaboration. Despite methodological challenges, observation data may contribute in a unique way to the teamwork discourse by identifying elements of inter-professional collaboration that are not so obvious to caregivers when asked to self-report [34].
Open communication between physicians and nurses is an important element of collaboration that appeared to be appreciated [25]. More often, researchers reported that deficiencies in collaboration and communication between healthcare professionals have a negative impact on the provision of health care and patient outcomes [35][36][37]. In addition to open communication, trust, respect, shared leadership, recognition of unique contribution and collegiality are mentioned by researchers as enabling factors for good inter-professional relationships [38]. On the other hand, barriers to good inter-professional collaboration reported by researchers are time pressure, lack of explicit descriptions of each other's roles and tasks (and therefore unawareness of one another's roles and competencies), poor organizational support, absence of clear leadership, different standards and professional values, different aims and priorities, and    Table 7 presents the patient outcomes that were found to be improved and/ or equivalent and/or declined vertical management structures with discriminatory power structures [39][40][41].

Nursing roles
Although the review articles often lacked a comprehensive description of the nursing roles in collaboration with physicians, we identified seven different categories of nursing roles in the systematic reviews in our review. 'Nursing tasks' and 'drug prescription' may be more distinct instrumental roles or nursing tasks, and are probably based on physicians' orders. 'Communication/consultation' and 'coordination/organization/referral' may be rather more related to integrated nursing roles with nurses' autonomous decision-making capacity based on structured agreements between nurses and physicians. Existing literature confirms the nursing skill mix, and the shift from task delegation to team care with shared responsibilities [42,43]. Two systematic Collaboration between nurses and general practitioners or cardiologists.
The following strategies were used: -Risk factor management.

Stalpers et al.
Collaborative nurse-physician relationships in the care for hospitalized patients. Table 8 presents an overview of the interpretation of collaboration between physicians and nurses (and other health care providers) within the eleven included systematic reviews reviews failed to describe the nature of the nursing roles and their responsibilities [23,27]. Furthermore, nursing titles differed across the included systematic review articles and ranged from 'advanced practice (registered) nurses' , 'nurse practitioners' , 'registered nurses' , 'primary care nurses' , 'clinical nurses specialists' and 'practice assistants'. The literature confirms the increasing diversity of the primary care workforce. A wider range of health professionals is included, such as those mentioned above [43]. The difference in professional titles might be attributed to a difference in education, which points out the importance of (postgraduate) education of nurses, especially in collaboration with physicians [29]. Expanding the role of primary care nurses is possible with appropriate training and on-going support from primary care physicians [44]. Improving care quality requires investing in a distinct primary care workforce that has followed a defined program of post-graduate training in primary care [2].

Strengths and limitations
A comprehensive research was performed and the methodological quality of the included review articles was carefully assessed. Overall, the quality of available systematic reviews on this research topic appeared to be limited. 15 potentially useful systematic reviews were excluded based upon an inadequate methodological quality. The included systematic review articles were heterogeneous in terms of patient populations, setting, type of nurse and geographic region. Limited descriptions of the collaboration, and the different nursing and physician roles in the included systematic reviews are limitations of this overview of systematic reviews. The included systematic reviews often lacked a detailed description of the evidence of the different primary studies, therefore it is difficult to make conclusions about the strength of the evidence of the results. For future systematic reviews concerning this research topic we suggest to define more precisely the nature of the collaboration between the two professions and to provide a clear description of the concept of inter-professional collaboration. Primary research articles concerning this research topic within the primary care setting are often limited to one pathology or diagnosis. However, the patient population in primary care presents itself with a wide range of pathologies. This overview of systematic reviews provides a more comprehensive view on the impact of collaboration between physicians and nurses in primary care on a wide variety of patient outcomes, for a wide range of patients.
Future research is necessary to define 'integrated interprofessional collaboration' in primary care more clearly, and to explore the impact of this collaboration on relevant patient and health care provider outcomes. These include hospital (re)admissions of patients with chronic conditions, patient satisfaction and primary health care provider satisfaction. We suggest using complementary methods to find a more robust evidence base for the collaboration of nurses and general practitioners in primary care.

Implications for practice
This overview of systematic reviews provides a firm evidence base to engage practice nurses in general practices. Moreover, current and future challenges in primary care require a more integrated inter-professional collaboration instead of a task shift between general practitioners and nurses. Therefore, we recommend that collaboration between health care providers should be well described Overview of the nursing roles in the collaboration with physicians Fig. 2 Overview of the nursing roles in the collaboration with physicians. Presents an overview of the 7 different nursing roles in collaboration with physicians within the eleven included systematic reviews. 2 systematic reviews failed to give a clear description of the nursing roles. The numbers within the graph represent the eleven included systematic reviews and discussed concerning roles, tasks and responsibilities of individual caretakers. A clear description is important in order to address the needs of the patient populations, and in order to address the individual patient needs.