A mixed‐methods systematic review of nurse‐led interventions for people with multimorbidity

Abstract Aims To identify types of nurse‐led interventions for multimorbidity and which outcomes are positively affected by them. Design Mixed‐methods systematic review following the Joanna Briggs Institute (JBI) methods for convergent‐integrated reviews. PROSPERO ID: CRD42020197956. Data Sources Cochrane CENTRAL, CINAHL, Embase and MEDLINE were searched in October 2020. Grey literature sources included OpenGrey, the Journal of Multimorbidity and Comorbidity and reference mining. Review Methods English‐language reports of nurse‐led interventions for people with multimorbidity were included based on author consensus. Two reviewers performed independent quality appraisal using JBI tools. Data were extracted and synthesized using a pre‐existing taxonomy of interventions and core outcome set. Results Twenty studies were included, with a median summary quality score of 77.5%. Interventions were mostly case‐management or transitional care interventions, with nurses in advanced practice, support to self‐manage conditions, and an emphasis on continuity of care featuring frequently. Patient‐centred outcomes such as quality of healthcare and health‐related quality of life were mostly improved, with mixed effects on healthcare utilization, costs, mortality and other outcomes. Conclusion Interventions such as case management are agreeable to patients and transitional care interventions may have a small positive impact on healthcare utilization. Interventions include long‐term patient management or short‐term interventions targeted at high‐risk junctures. These interventions feature nurses in advanced practice developing care plans in partnership with patients, to simplify and improve the quality of care both in the long and short‐term. Impact This is the first mixed‐methods review which includes all types of nurse‐led interventions for multimorbidity and does not focus on specific comorbidities or elderly/frail populations. Using adapted consensus‐developed frameworks for interventions and outcomes, we have identified the common features of interventions and their overall typology. We suggest these interventions are of value to patients and healthcare systems but require localization and granular evaluation of their components to maximize potential benefits.


| INTRODUC TI ON
People are living longer now than ever before and spending an increasing number of years in worsening health. The proportion of people living in the community with multimorbidity (two or more chronic conditions) is estimated to be around 33% (Nguyen et al., 2019), growing significantly with advanced age and increasing deprivation (Head et al., 2021). The cumulative burden of disease alongside socioeconomic drivers of poor health mean that those with multimorbidity are at increased risk of mortality, disability, functional decline, increased healthcare use and reduced quality of life (Xu et al., 2017).
Isolation compounds this risk, as older people with multimorbidity who live alone are more likely to require emergency care than those who live with others (Barrenetxea et al., 2021). The structure of healthcare services also places people with multimorbidity at risk of care fragmentation and treatment burden (Morris et al., 2021).
Nurse-led care delivered by experienced nurses is comparable in quality to physician-led primary care (Laurant et al., 2018), but it should not simply be considered a replacement for doctor-led care.
Nursing care is traditionally holistic and if provided in the context of a supportive and adequately resourced organization, builds on the foundation of a therapeutic relationship where patient's goals and wishes are central to their care (Bridges et al., 2013). Nursing interventions are ideally suited to supporting people with multimorbidity, for whom priorities may change with time and are not necessarily tied to specific conditions. To do this effectively, nurses require a robust evidence base to inform their practice (O'Connor et al., 2018).

| Multimorbidity
Multimorbidity refers to the coexistence of two or more chronic conditions (van den Akker et al., 1996). Comorbidity also refers to concurrent chronic conditions, yet it differs from multimorbidity by focussing on a specific index condition to which others are considered comorbid (Feinstein, 1970).To adopt a multimorbidity framework in research or clinical practice requires one to accept that no condition holds precedence, and that clinical decisions are taken within the overall context of the person (Boyd & Fortin, 2010).
This distinction matters in the way interventions are developed and evaluated. Interventions for index conditions and comorbidities can be targeted and use disease-specific outcomes but may not be generalizable to people with multimorbidity. Multimorbidity interventions will be more generic in their design, and the selection of outcomes to measure their effectiveness is more difficult (Harrison et al., 2021).
We must also consider what constitutes a chronic condition.
Diagnosed diseases such as asthma or diabetes meet this requirement. But in the context of the person, symptoms (such as breathlessness or pain) or risk factors (such as obesity or poverty) are also important (Willadsen et al., 2016). The number of chronic conditions which are considered is also a concern (Fortin et al., 2012), as identifying multimorbidity from a small number of conditions may limit the generalizability of findings.
The strongest evidence to support interventions for concurrent conditions comes from studies which focus on specific disease clusters, or improving outcomes of common comorbidities, such as depression . Most multimorbidity interventions are evaluated using disease-specific outcomes (Xu et al., 2017), theoretically limiting the generalizability of these effects to groups with different combinations of disease. Approaches which are patientoriented and support self-management are increasingly viewed as essential (Poitras et al., 2018;Smith et al., 2021).

| Nurse-ledcare
The concept of person-centred (or person-oriented) care underpins much of contemporary health policy but is particularly associated with nursing, providing a theoretical framework on which nursing interventions can be planned, implemented and evaluated. Key processes in person-centred nursing include engaging with patients, shared decision-making, creating a sympathetic presence, providing for physical needs and working with the patient's beliefs and values.
Evaluating the effectiveness of such interventions should incorporate patient satisfaction, the level of involvement they have with care, their feelings of well-being, and the presence of a therapeutic culture (McCormack & McCance, 2006).
Community-based nurse-led interventions cover a diverse range of areas including walk-in clinics, primary care clinics, smoking cessation, women's health and healthcare for homeless persons. These types of interventions have been shown to improve access to healthcare, symptom management, and a range of disease-specific measures (Randall et al., 2017). Conversely, there is strong evidence to suggest that peri-discharge interventions of varying complexity are not effective in reducing hospital reattendance when compared with usual care (Wong et al., 2021). We cannot, however, generalize these findings to people with multimorbidity.

| Definitions
Based on the work of Boyd and Fortin (2010), Le Reste et al. (2013) and Willadsen et al. (2016), we define multimorbidity as the K E Y W O R D S comorbidity, literature review, mixed-methods, multimorbidity, nurse-led care, nurses, nurse's practice patterns, nursing, systematic review coexistence of two or more chronic conditions, where no condition holds precedence. Conditions are not restricted to diagnosed chronic diseases and can include biopsychosocial and somatic risk factors.
Nurse-led interventions were required to satisfy at least one of the following requirements: (1) the service is led by a senior (possibly consultant) nurse, (2) nurses manage and are accountable for a caseload of patients or (3) nurses practice with a discernible degree of autonomy compared with medically led care.

| Rationaleforreview
Previous reviews established that interventions which are personoriented and incorporate self-management have the potential to improve outcomes in persons with multimorbidity (Crowe et al., 2016;Poitras et al., 2018;Smith et al., 2021). However, the evidence-base for certain types of nursing interventions for this group has not provided robust evidence of their impact on a wide range of outcomes . Given the important role person-centred care plays in nursing practice (McCormack & McCance, 2006), we believe a review which seeks to categorize and evaluate nurse-led interventions for people with multimorbidity is warranted.

| Aims
This review aims to identify what types of nurse-led interventions are effective in improving outcomes for people with multimorbidity.
3.1.1 | Specific objectives 1. To identify and categorize the different types of nurse-led interventions for people with multimorbidity, and 2. To identify which outcomes are improved by nurse-led interventions for people with multimorbidity.

| Design
This review follows the Joanna Briggs Institute (JBI) guidance for mixed-methods systematic reviews using a convergent-integrated approach (Lizarondo et al., 2020). This involves simultaneous analysis of qualitative and quantitative findings. Evidence grading is not recommended due to heterogeneity of sources. We have instead sought to classify whether different interventions have an effect which is best described as 'mostly improved', 'mostly unaffected', or 'mixed'. The review was registered on PROSPERO in October 2020 (ID: CRD42020197956), and is reported following the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) 2021 guidance (Page et al., 2021).

| Databases
Cochrane CENTRAL, CINAHL, Embase and MEDLINE were searched from inception using the terms in Table 1, which were based on a comprehensive list of synonyms from a recent systematic review . All searches were conducted in October 2020. File S1 details search strategies.

| Inclusion/exclusion criteria
Peer-reviewed research using qualitative, quantitative or mixedmethods to evaluate nurse-led interventions for people with multimorbidity were eligible. We applied the definitions of nurse-led care and multimorbidity as outlined earlier. Only English-language sources were included. Articles which dealt with specific clusters of conditions, only mental-health conditions or those which assumed multimorbidity purely on the basis of advanced age or frailty were excluded.

| Screening for eligibility
The first author applied inclusion/exclusion criteria to titles/abstracts and borderline decisions were agreed among the remaining authors. The first author read all full-text articles, with the role of second reviewer being shared between the second/third authors.
Decisions were agreed between the group. Endnote (The EndNote Team, 2013) and Rayyan (Ouzzani et al., 2016) were used to manage and screen citations.

| Searchoutcome
The search outcome is detailed in Figure 1. Three thousand four hundred and twenty citations were retrieved, deduplicated and screened. Ninety-seven full-text articles were read and 88 were excluded, mostly for not being related to multimorbidity (n = 34), for not detailing a nurse-led intervention (n = 11), or a combination of both (n = 8). Thirty-six articles were retrieved through other means as detailed in Figure 1, of which 19 were eligible for inclusion. The final review includes 28 reports detailing 20 studies.

| Qualityappraisal
Studies were independently appraised by two authors using JBI tools, and discrepancies were resolved through discussion between all three authors. A summary quality score (the percentage of checklist outcomes satisfied) was calculated for each study. Studies were not excluded based on quality. Full details are given in File S2.

| Dataextractionandabstraction
Data were extracted using JBI tools. Citation, methodology, participant details, intervention details, setting, outcomes affected, and authors' conclusions were extracted. Intervention details were extracted qualitatively and background papers were consulted as necessary. All quantitative findings related to the review questions were extracted. Meta-analysis was not attempted, primarily due to heterogeneity in intervention types, but also in study designs, populations, multimorbidity definitions and outcome measures. Qualitative data were extracted at the level of themes, subthemes or other distinct concepts outlined in the results sections of included papers.

| Synthesis
A convergent-integrated approach requires that quantitative data be synthesized narratively so it can be integrated with the qualitative findings. Data extraction forms were imported into NVIVO 12 (QSR International Pty Ltd, 2018), and synthesis was theoretically guided by the use of existing frameworks alongside inductive coding.

Multimorbidity terms
Nurse-ledcareterms  and Canada (n = 2), followed by the United Kingdom (n = 3), Spain (n = 2) and Sweden (n = 2). Hong Kong, Israel and Portugal each had one study. Quantitative studies were mostly quasi-experimental (n = 10), and all cohort studies (n = 4) were retrospective. Only three randomized controlled trials (RCTs) were found, and these included cluster and pragmatic designs. Qualitative studies (n = 4) and the qualitative component of mixed-methods studies (n = 2) were mostly descriptive or exploratory.

Mortality-mixed
Deaths were lower in intervention group at 1 year, no difference by 2 years.

Healthcareuse-mostlyunaffected
No significant improvement in hospitalization and ED use were detected. Improvements were noted across outcomes in a diabetes subgroup Home visits generated strong reduction in 30-day readmission and reduced 6-month inpatient admission in 4 of 6 subgroups stratified by risk.

Costs-mixed
6-month inpatient costs were reduced in the highest risk strata but were unaffected in the remaining five subgroups Nurse-led interdisciplinary intervention to support self-management.

Mentalhealth-mixed
Difference in depression [CES-D-10] favouring the intervention group in Ontario. No difference for depression in Alberta. Anxiety  unaffected at both sites.

Health-relatedqualityoflife-mixed
The mental component score of  was higher in the Ontario intervention group. There was no difference in mental component score in Alberta, and there was no difference in physical component score at either site.

Costs-mixed
At both sites, small reductions in costs across healthcare services were offset by an increase in diabetes-related costs, indicating cost neutrality.

Self-managementbehaviour-mixed
There was a small significant improvement in diabetes self-management in Ontario. No difference was detected in Alberta [Summary of Diabetes Self-Care Activities]

Qualityofhealthcare-mixed
There were many illustrations where patients and staff felt patient care had been improved by the intervention, but also gaps identified particularly in communicating with secondary care.

Communication-mostlyimproved
Communicating with the community matrons was seen as easy and preferable to going through GP surgery

Self-managementbehaviour-mostlyimproved
There were illustrations of participants engaged in good self-management behaviours, which were credited to the community matrons' intervention

Health-relatedqualityoflife-mostlyimproved
There was a significant increase in the number of people reporting good quality of life at 12 months [Revised Dementia Quality of Life Self-Esteem subscale]

Healthcareuse-mixed
There was a significant reduction in ED use over the preceding 12 months but no difference in hospitalization.

Prioritization-mostlyimproved
There was qualitative evidence of nurses engaging in patient-centred goal setting, and that patients appreciated this approach Self-managementbehaviour-mostlyimproved There was qualitative evidence of patients adopting positive health behaviours as a result of intervention

Nutrition-mostlyimproved
There was a significant reduction in proportion of those at 'high nutritional risk' at 12 months [DETERMINE checklist]

Fallsrisk-mostlyunaffected
The proportion of those at 'high-risk' of falls was marginally higher post-intervention [STEADI assessment]. Authors attribute this to likely disease progression or improved detection of high-risk individuals.

Pain-mostlyimproved
There was a significant reduction in the number of people reporting poorly controlled pain at 12 months [Modified Universal Pain Assessment Tool]

Proactivecase-finding-mixed
There was qualitative-only evidence to indicate the intervention improved early detection of high-risk patients and appropriate referral

Diseasemanagement-mostlyimproved
There was qualitative evidence to indicate nurses provided effective disease management and early detection of deterioration in health status

Trustandadvocacy-mostlyimproved
There was qualitative evidence to indicate nurses provided an advocacy role for the patients with other health and social care providers

Caregiversupport-mostlyimproved
There was qualitative evidence to suggest nurses were involved in proactively identifying and supporting caregivers

Mortality-mostlyunaffected
No differences detected in mortality between groups at 12 months

Mentalhealth-mostlyimproved
The telephone group saw a reduction in cognitive impairment [Pfeiffer Score] over 12 months, but not the telehealth group Health-relatedqualityoflife-mostlyimproved Both intervention arms were associated with improved [EQ-5D] scores, and the telehealth group also had improved [EQ-5D VAS] at 12 months Activitiesofdailyliving-mostlyimproved The telephone group had improved functional ability/ADLs at 12 months compared with control [Barthel Index]

Healthcareuse-mostlyunaffected
There was no reduction between groups for any aspect of healthcare utilization.
a Study title decided based on either the principal investigator (if known), or the first author of the most recent or significant publication. This does not signify authorship as determined by the publication authors.
b Summary quality scores calculated as a percentage of the domains in the scoring tool which were scored 'yes'.
c Outcome measures in parentheses where identified by study authors.

TA B L E 2 (Continued)
quasi-experimental studies scored less than 50% but did not add unique findings. The most common issues in RCTs related to blinding of participants or clinicians (n = 3). Absence of control groups (n = 5) and single measurements of outcomes (n = 6) were common in quasi-experimental studies. Qualitative studies lacked statements locating the researcher culturally or theoretically (n = 5) and most did not consider the impact of the researcher on the research (n = 3). Failure to discuss theory and reflexivity in qualitative research suggests that studies have been conducted in a value-neutral or atheoretical manner, which is at odds with the assumptions of most qualitative approaches (Braun & Clarke, 2021;Sandelowski, 1993) and limits the transferability of findings. We have highlighted which results are based purely on qualitative findings. File S2 provides further detail.

| Participantcharacteristics
The average age of participants was reported in 13 studies and ranged from 38.0 to 83.1 (median: 75.6, IQR: 73.9-76.5), reflecting the fact that most (n = 11) studies targeted older adults. Nine studies reported an average number of chronic conditions, ranging from 2.85 to 11.5 (median: 4.45, IQR: 4.02-8.00). All studies recruited participants from both sexes and were broadly evenly distributed.

| Interventioncomponents
File S3 details the adapted EPOC taxonomy. Interventions were complex, with the average intervention comprising 5.75 components (range 2-12). Figure 2 outlines components by frequency.
Common components included nurses in advanced practice (n = 16), case-management (n = 13) supporting self-management (n = 10), discharge planning/transitional care (n = 8) and interventions to improve continuity of care (n = 8). A common theme was involvement of an experienced nurse managing complex care to reduce healthcare utilization or improve the quality of patient care.
Interventions were delivered either as a continuous intervention (with the nurse serving as primary caregiver) or targeted at high-risk transitions, such as hospital discharges.

Effect of case-management interventions
Case-management interventions had positive effects on healthrelated quality of life, self-management behaviour, pain and disease management, nutrition, activities of daily living, communication with healthcare providers, prioritization of needs, fostering trust and advocacy and the overall quality of care from the patient perspective (Table 4).
Effects on health services were mixed. While some studies noted reductions in bed-days and emergency care use for communitybased interventions (Lupari, 2011;Sadarangani et al., 2019), two large studies of primary care interventions did not detect a reduction across the majority of interactions with healthcare services (Boult et al., 2011;Dorr et al., 2008). Two hospital-based interventions also did not detect reductions in healthcare use (García-Fernández et al., 2014;Valdivieso et al., 2018). Cost reduction and mortality were similarly mixed. A community intervention in Northern Ireland demonstrated a significant reduction in healthcare costs (Lupari, 2011), while a primary care-based intervention from the United States failed to generate significant reductions (Leff et al., 2009;Sylvia et al., 2008). Two studies involving hospital case-managers did not detect any reduction in mortality at 90 days (García-Fernández et al., 2014) and 12 months (Valdivieso et al., 2018), while a large study of a community-based nurse casemanager intervention (3432 participants) found significant differences in the proportion of deaths favouring the intervention group at 1 year, although despite a crude reduction at 2 years, significance was not sustained (Dorr et al., 2008).  Randall et al., 2014).
Qualitative findings suggested case-finding for referral to other services may be improved by a community-based intervention, and quantitative findings indicated that there was no reduction in falls risk as a result of the same intervention (Sadarangani et al., 2019).

| Transitional care interventions
Transitional care interventions are relatively short-term and focus on the coordination of patient care and changes to the care process (EPOC, 2015); the interventions included in this review all targeted the transition from acute hospital to home (Chow & Wong, 2014;Hanson et al., 2018;Jackson et al., 2016;Karlsson & Karlsson, 2019). Home visits were common to all interventions however, one study compared a combined telephone and home visit service with a telephone-only service (and usual care) (Chow & Wong, 2014). One study examined the effect of a home visit by a nurse nested within an existing transitional care intervention (Jackson et al., 2016).

Effect of transitional care interventions
Transitional care interventions demonstrated positive effects on health-related quality of life, self-rated health, self-efficacy and healthcare use ( Table 4). One study evidenced improvements in physical functioning, self-efficacy, self-rated health and reduced readmission within 84 days of discharge, although not within 28 days (Chow & Wong, 2014). Conversely, a propensity-matched cohort analysis of another intervention demonstrated a reduction in readmission and ED attendance at 30 days which was not sustained at 90 or 180 days (Takahashi et al., 2016). In a large US study (35,174 participants), the addition of a nurse home visit to an existing transitional care intervention also generated positive reductions in 30-day readmission, as well as reduced 6-month readmission in four of six risk strata (Jackson et al., 2016).
Mortality results were mixed, with evidence of un-sustained reductions in one study (Takahashi et al., 2016). Cost reductions were only evidenced for those with the highest risk of healthcare utilization, as determined by predictive modelling (Hanson et al., 2018;Jackson et al., 2016), suggesting that targeting such groups in this way may be cost-effective. Limited qualitative findings suggested a Swedish intervention may have improved communication and quality of healthcare from the patient perspective (Karlsson & Karlsson, 2019).

| Supported self-management interventions
Two studies from the same programme of research were predominantly focused on supporting self-management. The six-month intervention included home visits, group education sessions, case conferences and nurse-led care coordination. There were variations in delivery between study phases and sites, however, the above components were consistent throughout (Markle-Reid et al., 2016;Markle-Reid et al., 2018;Miklavcic et al., 2020).

Effect of supported self-management interventions
Effects varied between reports. It did not generate an overall reduction in costs, yet savings in some areas were explained to offset increases in others, achieving cost-neutrality. Mixed results were reported across studies for health-related quality of life and selfmanagement. No improvement was noted in mental health and selfefficacy (Markle-Reid et al., 2016Miklavcic et al., 2020).

| Nurse-led interdisciplinary team interventions
Interdisciplinary team working featured in six interventions, however, only one study specifically focussed on evaluating the effect of a nurse practitioner-led intensive management patient-aligned care team, comprised of a nurse practitioner (team leader), social worker, recreation therapist, administration coordinator and a parttime physician champion. This was compared with a less intensive physician-led team. The key features of the intervention involved 24/7 contact to either the nurse practitioner or physician, comprehensive interdisciplinary assessment and frequent follow-up, health education and coaching, hospital 'in-reach', co-attendance at appointments and connecting patients with community resources (Hummel et al., 2017).

Effect of nurse-led interdisciplinary team interventions
The intervention reported a crude improvement in identification and referral of people with palliative and end-of-life care needs, however, when the imbalanced distribution of people with cancer and dementia was removed in a sensitivity analysis, the difference was no longer significant (Hummel et al., 2017).

| Information and communication technology (ICT) interventions
The use of ICT and telehealth was present in many interventions, however only one intervention evaluated the effect of a nurse practitioner-led smart technology intervention to a cohort of adults with multimorbidity. The 12-week intervention made use of a patientfacing web application provided to participants on a tablet computer alongside a range of devices for physiological monitoring, such as Bluetooth-enabled scales, glucometer and sphygmomanometer.
Participants were expected to use the application in place of attending the clinic and the nurse acted as primary clinician (Mallow et al., 2018).

Effect of ICT interventions
The small (30 participants) study reported improvements in some physiological measures (blood pressure and glycaemic control) but not others (weight control; Mallow et al., 2018).

| DISCUSS ION
We have found that nurse-led interventions for multimorbidity are mostly either case-management or transitional care interventions, often employ a nurse in advanced practice and are focussed on simplifying care and supporting self-management. Effects on outcomes were mixed; case-management had positive effects on patient-centred outcomes, yet the impact on health service outcomes was inconsistent. Transitional care interventions were more consistent in reducing healthcare usage and reattendance, albeit with variations in the longevity of the effect. The overall aim of this review was to identify the types of nurse-led interventions which improve outcomes for people with multimorbidity; we conclude that interventions which focus on coordination and management of care in both the long-and short-term have the potential to improve patient-centred and patient-reported outcomes but are less consistent in improving health-service oriented outcomes such as usage and costs.
To better understand the types of nurse-led interventions which are most effective, we should locate these interventions in relation to established theory. First, interventions were generally grounded in the Chronic Care Model (Wagner et al., 2002). Importance was placed on the organization of healthcare services, and on connecting nurses and patients with resources available in the wider community.
Productive interactions between nurse and patient were seen as essential to the shared development of care plans and to setting realistic goals and priorities about the individuals' care. Both case management and transitional care interventions tend to operate within this framework, albeit over different durations and in different settings.
Nested within the Chronic Care Model is the concept of supported self-management, which also underpinned many interventions in this review. Yet, studies which measured the effect interventions had on treatment burden were conspicuously absent, despite this being an important outcome to consider in multimorbidity research (Smith et al., 2018) and one which can negatively impact an individual's ability to self-manage multimorbid conditions (Boyd et al., 2014). Burden of Treatment Theory describes the complex interplay between the individual, the health system and their social network, and how these relationships mediate the individual's capacity to self-manage effectively (May et al., 2014). Validated tools exist to measure treatment burden in people with multimorbidity (Tran et al., 2014), and these should be deployed in future studies of similar interventions.
While interventions were spread between primary, secondary and community care, their orientation is closest to the generalist and person-centred approach of primary care. A primary care approach to multimorbidity care should include comprehensive assessment which considers interactions between the many drugs and conditions a person has, prioritization of care based on patient preferences, and individualized management and follow-up. All this should be centred around realistic and shared treatment goals (Muth et al., 2014).
We should also consider where this approach aligns with nursing models. At their most fundamental level, we suggest these interventions are grounded in primary nursing theory, where an individual nurse is responsible for an individual patient and for providing continuous care to them. This is in contrast with alternative and more common models such as team-based nursing (where nurses work together to care for a group of patients) and patientallocation (where nurses care for a different allocation of patients each shift) (Fernandez et al., 2012). A primary nursing approach requires experienced nurses who are well-prepared to undertake this level of practice, which we note in the large number of interventions which required a nurse undertaking various levels of advanced practice.
Despite these elements of nurse-led interventions being present across case-management and transitional are interventions, the effect they had on outcomes was varied. Small improvements in patient-centred outcomes coupled with lesser impacts on healthcare services and health/disease outcomes are common features of most evaluations of multimorbidity interventions (Crowe et al., 2016;Smith et al., 2021) and this review finds that nurse-led interventions are no different. Interventions for this patient group are complex and multi-faceted, deployed within complex healthcare systems and target patients who are, by definition, also complex. These layers of complexity reflect the real-world challenges of developing and testing interventions for this group. It is challenging and resourceintensive to isolate the effects of the components or system factors which are responsible for observed effects (Poitras et al., 2018), but such approaches may be necessary in the future development of nurse-led interventions for people with multimorbidity. Adopting a primary care and primary nursing-based approach may provide a foundation on which to build such interventions.

| Limitations
Identifying studies concerning multimorbidity is challenging due to inconsistent terminology (Nicholson et al., 2019). We designed our definition of multimorbidity to account for this, however, it is possible some studies may have been missed if not identified by their authors as being concerned with multimorbidity. A similar problem exists in defining nurse-led interventions, although we sought to ensure the studies included met our agreed definition. Finally, only articles published in English were included, therefore some non-English studies may have been missed.
We adopted a convergent-integrated design primarily because we identified significant heterogeneity in study design and intervention type through scoping and because we wanted to include qualitative evidence in our analysis. Measuring certainty (through systems such as GRADE) is not recommended in convergent-integrated reviews, therefore we have been cautious in our interpretation of intervention effects. A similar issue exists regarding quality appraisal; we have applied a summary quality score to provide a summary of the extent to which quality outcomes were satisfied, but it is important to note that this does not allow for comparisons to be made between studies with different designs. Full details of quality assessment can be found in File S2.

| Futureresearch
For those developing and evaluating complex interventions for people with multimorbidity, future research should focus on evaluating the components within these interventions, to better distil which elements are effective. RCTs can be conducted within existing interventions, and qualitative methods may be able to guide the selection of intervention components to be analysed. The way we interact with healthcare services has also been impacted by the COVID-19 pandemic, and while some countries are moving towards a degree of 'normality' in their practices, the effect of remote consultations, selftesting, shielding and such may have altered perspectives on what is desirable in an intervention. Further research should explore this in the context of multimorbidity.
Treatment burden is an important factor in multimorbidity research, and future evaluations should make use of validated tools to measure the effect of interventions on this outcome. Finally, researchers should strive to ensure a uniform terminology is employed to define multimorbidity in studies of interventions.

| Implicationsforpractice
Nurses are well-placed to manage people with multimorbidity, as the transdiagnostic approach required compliments the person-centred approach of nursing care. Nurses working with people with multimorbidity should encourage involvement in decisions about their care and endeavour to create and regularly review care plans which align with the person and their needs.

| CON CLUS ION
Nurse-led interventions for multimorbidity are characterized by the development and review of individualized patient-centred care plans by a primary nurse responsible for continuous care, working in partnership with the patient. They may constitute long-term interventions such as case management, or they may be brief and targeted at high-risk periods (such as transitional care). The evidence indicates that both approaches are acceptable to patients and may increase satisfaction, but the evidence for health service improvements is less clear. As researchers and practitioners, we need to consider the complexity of such interventions and the patients for whom they are designed, and we must plan our evaluations and care accordingly.

AUTH O R S 'CO NTR I B UTI O N S
All authors have agreed on the final version and meet at least one of the following criteria (recommended by the ICMJE*): 1. substantial contributions to conception and design, acquisition of data, or analysis and interpretation of data; 2. drafting the article or revising it critically for important intellectual content.

ACK N OWLED G EM ENTS
The authors thank Paul Cannon, University of Glasgow librarian who peer-reviewed the search strategies, PPI adviser Annabelle who reviewed this manuscript and the corresponding authors who kindly responded to requests for further information.

FU N D I N GI N FO R M ATI O N
This study was funded by NHS Greater Glasgow and Clyde as part of the first author's PhD fellowship.

CO N FLI C TO FI NTE R E S T
No conflict of interest has been declared by the authors.

PEER R E V I E W
The peer review history for this article is available at https://publo ns.com/publo n/10.1111/jan.15427.

DATAAVA I L A B I L I T YS TAT E M E N T
Data sharing is not applicable to this article as no new data were created or analysed in this study.

PATI E NTA N DPU B LI CI N VO LV E M E NT
A patient advisor has reviewed, approved and provided comments on this manuscript.