Article Text
PDF
Abstract
Introduction The increase in the number of patients with uncontrolled type 2 diabetes mellitus (T2DM) is in need of effective management interventions. However, research to date has been limited to the evaluation of the outcomes of community pharmacists alone. Therefore, the aim of the study protocol is to compare the effects of clinical pharmacist-led intervention strategies for the management of T2DM in the outpatient settings.
Method and analysis The study will collect and analyse data applying standard Cochrane methodological procedures. A search for eligible studies and ongoing trials will be conducted using PubMed, Embase, Medline (via Ovid), EBSCO (via Ovid), Lippincott Williams & Wilkins (LWW) Journals (via Ovid), ProQuest Health and Medical Complete, and ClinicalTrials.gov (clinicaltrials.gov) from database inception to December 2023. Clinical and health outcomes will be measured using both glycaemic control related indicators (eg, glycated haemoglobin, fasting blood glucose, postprandial glucose) and general indicators (eg, adherence, disease management and health-related quality of life). The meta-analysis will conduct pairwise meta-analysis using random effects models and network meta-analysis (NMA) employing the Bayesian hierarchical model. The visualisation and statistical analysis will be carried out using RevMan, R Studio and ADDIS. Additionally, we will evaluate the certainty of the evidence by using Grading of Recommendations Assessment, Development and Evaluation system.
Ethics and dissemination There will be no primary data collection from NMA participants, and there is no requirement for formal ethical review. Our aim is to present the results of this NMA in a peer-reviewed scientific journal, at conferences, and in the mainstream media.
PROSPERO registration number CRD42022355368.
- protocols & guidelines
- diabetes & endocrinology
- quality in health care
- clinical pharmacology
This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.
Statistics from Altmetric.com
STRENGTHS AND LIMITATIONS OF THIS STUDY
Network meta-analysis (NMA) extends traditional pairwise meta-analysis to allow simultaneous comparison of multiple interventions using direct or indirect evidence.
NMA is useful to rank the interventions according to their effectiveness and to assess the implications of observed effects within the evidence network.
The variety of interventions in the studies might lead to a high heterogeneity, which makes NMA infeasible or difficult to interpret.
The validity of results of NMA may be affected by the lack of intransitivity of indirect comparisons.
Introduction
Diabetes is a progressive health problem that associated with increased morbidity and mortality rates1–4 and a significant burden of healthcare costs.5 6 Globally, an estimated 537 million people with diabetes were reported in 2021; even more is predicted to rise to 643 million by 2030.7 Type 2 diabetes mellitus (T2DM) is the major contributor to the global diabetes burden, accounting for 90% of diabetes mellitus cases worldwide.8 Patients with T2DM are at risk of organ failure, dysfunction and long-term damage.1 Poorly controlled glycaemic variability can lead to microvascular and macrovascular complications.9–11 To date, evidence from the literature shows that despite the benefits of antihyperglycaemic medications and tighter glycaemic control, patients with T2DM fail to achieve the desired clinical outcomes,12–14 with 42.9%–66.7% failing to achieve adequate glycaemic control.15–18 These unsatisfactory outcomes may result from patient-related issues such as poor compliance19 20 and suboptimal ability to manage diabetes.21 22 Thus, the diabetes management approach implemented based on professional and proactive healthcare in which patients play an active role is critical to improving patients’ therapeutic outcomes.23–26
As non-physician clinicians, pharmacists play a vital role in monitoring the health and progress of patients to ensure the safe and effective use of medication.27 Pharmacists have been involved in chronic disease management, taking direct responsibility for patients’ disease states, medications and overall management to improve outcomes.24 25 Extensive global studies have demonstrated the effectiveness of pharmacist-based interventions on patients with diabetes, resulting in improved quality of care and clinical outcomes.23 26 28–30 In addition, several systematic reviews31–34 have identified various non-pharmacological pharmacist-led interventions, such as diabetes education,35–38 medication review,39 drug counselling/advice,40 41 lifestyle modification, self-care, peer support and behavioural intervention.42–44
However, current reviews focus only on the effectiveness of pharmacist-led interventions without considering the differences between different pharmacist roles. Notably, in contrast to community pharmacists who work primarily in community settings,45 46 clinical pharmacists partner with hospitals, care homes, physicians and other professionals.47 48 Given the different working places and responsibilities among multiple pharmacist positions, the existing relevant systematic reviews were limited to the generalisability of implementations. Thus, a meta-analysis measuring the effectiveness of specific pharmacist roles is needed to determine the feasibility of intervention and to identify which components are more likely to work in their respective context.
In fact, comprehensive evaluations of community pharmacist-led interventions have been carried out,23 49 50 but meta-analyses of clinical pharmacist-led interventions in the outpatient setting are limited.51 52 Although evidence exists the efficacy of clinical pharmacist-led interventions in controlling T2DM glycaemia in the outpatient setting,53 implementing a programmatic randomised controlled trial (RCT) is challenging due to the different intervention categories available.54 Therefore, our study goal is to compare previously published clinical pharmacist-led interventions in the outpatient setting to provide policy makers with practical implementation strategies for developing guidelines.
Methods and design
The study will systematically review and compare the effectiveness of clinical pharmacist-led interventions on patients with T2DM in outpatient settings. Effectiveness comparisons will be performed quantitatively using direct or indirect evidence through network meta-analysis (NMA).55 56 The relative effectiveness of two interventions delivered in an outpatient setting can be estimated even in the absence of direct comparisons.57
This protocol has been registered in the International Prospective Register of Systematic Reviews (PROSPERO) database. The development of this protocol followed the Preferred Reporting Items for Systematic Review and Meta-Analysis Protocols (online supplemental appendix A).58
Supplemental material
Patient and public involvement
The study will apply existing studies; no patient-level or public data will be involved directly in this study.
Eligibility criteria
Our study is designed to include all types of RCTs (eg, cluster RCT and programmatic RCT). The setting of inclusion and exclusion criteria followed the framework of ‘PICOTS’ (population, intervention, comparison group, outcomes, time frame, setting) (table 1).59
Inclusion and exclusion criteria
Type of participants (P)
The inclusion criteria include (1) adult patients (at least 18 years old) and (2) diagnosed with T2DM. The excluded trials include (1) patients with other diabetes types (eg, type 1 diabetes mellitus, gestational diabetes) or with diabetes secondary to medication or with other chronic diseases; (2) involved participants who are pregnant women.
Types of interventions (I)
All non-pharmacological but clinical pharmacist-led strategies in outpatient settings to improve clinical outcomes and diabetes management ability of patients with T2DM will be considered interventions. Studies will be excluded if they involve testing of clinical pharmacy interventions in community pharmacies, community settings and home-based care.
As there is no clear classification and definition of interventions of clinical pharmacists and considering the assumption of transitivity among those particular strategies, we will predesign the intervention types from clinical pharmacists as (1) diabetes education, (2) medication review, (3) drug consultation/counselling, (4) lifestyle adjustment, (5) self-care, peer support and (6) behavioural intervention according to the previous reviews.29 60 61 These types of intervention will be used as the preliminary classification hypothesis of clinical pharmacists’ interventions in our NMA, and the specific classification will be adjusted according to the results of the actual literature review.
This review will include studies that involved clinical pharmacist-only interventions. The multidisciplinary collaborative or interdisciplinary interventions with pharmacists’ involvement will be excluded. The interventions can be implemented as a standalone or a combination of categories. Trials that differed from our study’s settings of interventions (eg, comparing the effectiveness of pharmacological interventions, treatment/clinical operations) will be excluded.
Types of comparisons (C)
The included studies compared clinical pharmacist-led interventions against usual care of T2DM or the previous reviewed multiple intervention strategies58 and above will be considered as a comparison. Direct and indirect comparisons of all strategies will be performed. Studies that differed from our study’s settings of comparisons (eg, compare the effectiveness of pharmacological treatments) will be excluded.
Types of outcomes (O)
The predetermined primary outcomes indicator was glycated haemoglobin (HbA1c) (%), the most used parameter to assess glycaemic control.62 Secondary outcome indicators included (1) other blood glucose indicators, including fasting blood glucose and postprandial glucose; (2) medicine adherence; (3) disease management and (4) health-related quality of life (HRQoL).
Studies with unclear diagnosis and without our interested efficacy outcomes indicators (eg, economic outcomes) will be excluded.
Time-frame of outcome evaluation (T)
The minimum duration of the intervention is no restrictions. If the included studies have more than one follow-up point, the final follow-up values of the outcome will be used for analysis.
Type of studies and setting (S)
We designed the meta-analysis to involve both RCT studies (including cluster RCT and programmatic RCT) and ongoing trials concerning clinical pharmacist-led non-pharmaceutical interventions in clinical settings for patients with T2DM. The study types of reviews, case reports, editorials, commentaries, clinical practice guidelines, conference abstracts, literature not in peer-reviewed journals and qualitative studies will be excluded.
Data sources and search strategy
The search strategies have developed to identify related studies and ongoing trials in PubMed (1966–2023), Embase (1947–2023), Medline (via Ovid, 1946–2023), EBSCO (via Ovid, 1994–2023), Lippincott Williams & Wilkins (LWW) Journals (via Ovid), ProQuest Health and Medical Complete (1986–2023) and ClinicalTrials.gov (clinicaltrials.gov). The search terms used in this review include medical subject headings (MeSH) and entry terms.
Key search terms included: “Intervention”; “Pharmaceutical Services”; “Pharmaceutical Services, Online”; “Pharmacy Service, Hospital”; “Diabetes Mellitus, Type 2”; “glycemic control”; “Blood Glucose”; “Patient Compliance”; “Treatment Adherence and Compliance”; “Medication Adherence”; “Self-Management”; “Disease Management”; “therapy”; “Self Medication”; “Quality of Life”. According to the search order, synonyms with the same search results will be removed. The MeSH terms and synonyms will be combined with Boolean operators.
This set of search terms was slightly modified when searching in different databases due to a different system and technical limitations. The detailed search strategies for each database were presented in online supplemental appendix B. There will be no language or year restriction.
Supplemental material
Study selection
The study selection will follow three steps: (1) two reviewers will independently screen all titles and abstracts and retrieve the full article from the electronic databases based on defined eligibility criteria; (2) the full text of each potentially eligible article will be obtained and screened independently by two reviewers to assess further its suitability for inclusion in this review; (3) the reference lists of the selected publications will be screened for additional studies of interest. Any reviewer disagreements will be resolved by discussion or consultation with a third party.
Data extraction
Two reviewers will abstract data independently, and the third party will match and double-check the data. Data will be extracted in standard format using Microsoft Excel. The collected data includes general study characteristics (eg, author name, publication title, year of publication), description of the study population, follow-up time, number and duration of contact moments during the intervention, description and components of the intervention, clinical outcomes (HbA1c, blood glucose and other) and patient-reported outcomes (adherence, quality of life, disease management and other). The data extraction process will use the free web application Rayyan QCRI.63
Risk of bias assessment
Based on our inclusion criteria, a risk of bias assessment accompanying each included randomised trial will use the Cochrane tool (RoB2 tool),64 which is structured into the following five bias domains for each outcome evaluated: (1) bias arising from the randomization process, (2) bias due to deviations from intended interventions, (3) bias due to missing outcome data, (4) bias in the measurement of the outcome and (5) bias in the selection of the reported result. Reviewers assessed each study item as either ‘high risk’, ‘low risk’ and ‘some concerns’ (RoB 2) of bias. Two reviewers will assess the bias of included studies independently and the third party will match and double-check the results.
Data synthesis
Dealing with missing data
If necessary, we will contact the authors of the original studies to access the missing data of each included study.
Assessment of heterogeneity
Heterogeneity between the studies in effect measures will be assessed using both the χ2 test (test level: α=0.05) and I² statistic that categorises heterogeneity as low (0%–40%), moderate (30%–60%), substantial (50%–90%) and considerable (75%–100%).65
Assessment of transitivity
As a unique assumption of NMA, the transitivity combines direct evidence to create indirect evidence about a related comparison (eg, from comparison A-B and C-B to create comparison A-C).66 However, unequally distributed covariates or other effect modifiers (such as the age group of the participants or the treatment intensity) between studies might threaten the plausibility of the transitivity assumption and, thus, the validity of indirect comparison.67
Transitivity, as such, cannot be tested statistically. However, the risk of violating this assumption can be attenuated by comparing the distribution of effect modifiers between different comparisons and only including studies for which the population, methodology and target condition are as similar as possible.68 Subgroup analysis and meta-regression are commonly used to assess or improve trial similarity for adjusted indirect comparison.67
However, the transitivity might be challenging to estimate due to these studies’ unknown, unobserved or unreported effect modifiers. Thus, we will consider downgrading the certainty of the indirect evidence.
Assessment of coherence
The statistical manifestation of transitivity is typically called coherence, which differs from the notion of inconsistency (or heterogeneity) within standard meta-analyses. The coherence assumption is that a specific comparison’s direct and indirect intervention effects are identical.69
The coherence assumption might be violated when different sources of information for a particular relative effect are in disagreement.70–72 Incoherence between the direct and indirect summary estimates for pair-wise comparisons will be measured by incoherence factor (IF).
We will evaluate the local and global incoherence in NMA. The Separating Indirect from Direct Evidence (SIDE) will evaluate local incoherence. SIDE can detect possible ‘incoherence spots’ by evaluating the IF.73 The global incoherence will be evaluated by models that add additional terms to allow intervention effects to vary when estimated directly and indirectly.71 The global approaches can evaluate coherence in the entire network.
Due to the low power of the incoherence tests, we will conclude by consideration of statistical significance and by interpreting the range of values included in CIs of the IFs.74 75 If the statistical significance of incoherence occurs, we will seek the possible explanations (eg, errors in data collection, broad eligibility criteria and imbalanced distributions of effect modifiers) and possible analytical strategies.72 76
Pairwise and NMA
Meta-analysis will be undertaken with RevMan software to analyse the direct comparison results of the included studies. The NMA will be performed for each outcome to compare multiple interventions simultaneously by the R programming language. The R Studio and ADDIS software will be used for data merging, statistical analysis, and NMA while drawing a network relationship diagram and anecdotal sequence diagram of various intervention measures.
Considering the expected between-study heterogeneity, we will use a Bayesian random-effects model to build a data pool for each outcome. For dichotomous data, effect estimates will be calculated using an OR with a 95% credible interval (CrIs). The continuous data will be expressed as means and SDs for each study. The (standardised) mean difference (if different metrics are used across studies) will be calculated with their respective 95% CIs. All the p values will be set to be 0.05 with 95% CIs.
The relative efficacy of different interventions for the primary outcome (HbA1c) will be evaluated using the surface under the cumulative ranking curve (SUCRA), with its 95% CrI and mean ranks. The best intervention is expected to have high SUCRA values, while the worst will have low values.
Subgroup and sensitivity analysis
If there is heterogeneity and enough studies are available, we will use subgroup analysis to find the reasons for heterogeneity and compare each group’s effects. We propose six hypotheses to explain the variability between studies as possible effect modifiers: (1) patient characteristics, (2) intervention types of T2DM, (3) time of T2DM diagnosis, (4) trial with low risk of bias compared with a high risk of bias (considering concealment allocation and intention to treat analysis), (5) presence of diabetes chronic complications and (6) duration of intervention and follow-up time. The assessment of usual care of T2DM will be their similarity across intervention comparisons.
If the included studies are sufficient, we will perform a sensitivity analysis of the statistical results to assess the robustness of the results.77 78 For each excluded study, the meta-analysis will be performed again, and the results will be compared with the results before exclusion. If there is no substantial change in the comparison, the result is stable; otherwise, the result is unstable.
Assessment of publication biases
According to the Cochrane Handbook,79 if more than ten studies are included,80 we will construct a funnel plot to analyse the potential publication bias. Egger’s test is commonly used to assess potential publication bias via the degree of funnel plot asymmetry.81 Asymmetry in the funnel plot indicates bias; the more pronounced the asymmetry, the greater the bias.
Quality of evidence
We will use the Grading of Recommendations Assessment, Development, and Evaluation system (GRADE) approach for NMA to evaluate the certainty of the direct, indirect and network estimates for evidence with available data. The evaluation process will be conducted using GRADE’s official software package: (GRADEpro GDT; www.gradepro.org). The process will be performed in pairs and independently.
The following four steps determine GRADE assessments of the studies’ certainty: (1) present each direct and indirect comparison in NMA; (2) rate the certainty of direct and indirect estimates; (3) present the network estimate for each pairwise comparison and (4) rate the certainty of the network estimate.82
The classifications of estimated quality rating of the treatment effect include four levels (‘high’, ‘moderate’, ‘low’ or ‘very low’). Given that our involved study types are RCT, the initial quality rating level is ‘high’.68
The direct and indirect estimates rating will consider to be rated down due to the following reasons: risk of bias, inconsistency, indirectness, imprecision and publication bias. The insensitivity can be an additional reason to rate down for the indirect evidence.83 For each component, the quality of the evidence can be maintained or downgraded by up to two level—1 (‘serious concern’) or 2 (‘very serious concern’), subject to a maximum downgrade by three levels, to the quality rating of ‘very low’, across the assessment components.
The network estimates rating will be based on the direct evidence alone or mixed evidence and the assessment of incoherence at pairwise comparisons. If we encounter incoherence between direct and indirect evidence, we will focus on whichever of the direct or indirect estimates with higher certainty. The network estimates rating may be rated up when the certainty of the direct or indirect evidence is rated down for both imprecision, and combining the direct and indirect narrows the CI sufficiently.84 The judgements will be performed in the NMA results and be included in the ‘Summary of findings’ table.85
Ethics and dissemination
There is no primary data collection from participants in NMA, and no formal ethical assessment is necessary. We plan to present the results of this NMA in a peer-reviewed scientific journal, conferences and the popular press.
Discussion
With the increasing prevalence of T2DM86 and the unsatisfactory glycaemic control by some individuals,87–91 it is necessary to find out the proven and best effective actions in multiple strategies to prevent and control this metabolic disease. The NMA studies will be helpful to the policy-makers to find out the best interventions led by pharmacists to support patients with T2DM in achieving glycaemic control targets and improving their adherence, the ability of disease management, and HRQoL.
Ranking the interventions based on SUCRA may have a controversy on the uncertainty of rankings.92 For example, even if there are no clinically or statistically relevant differences between the efficacy of treatments and are observed between treatments near each other’s SUCRA value, the differences in their ranks, which might be because of random error, will imply there is one.93 Previous studies have noticed several reasons that can be responsible for the substantial uncertainty of rankings: (1) the limited amount of evidence for each comparison; (2) the substantial heterogeneity between studies and (3) the similar interventions have frequent overlapping uncertainty intervals.94 With a quick increase in pairwise comparisons, summarising and interpreting the relative effects of multiple interventions can be challenging.
In short, the conclusions on the effectiveness rank of multiple interventions regarding specific health outcomes need to consider the estimates of effects comparing those interventions and the certainty of the evidence (confidence in evidence, quality of evidence).95 To ensure the robustness of results, we will set rigorous eligibility criteria following the ‘PICOTS’ and use the Cochrane Handbook69 to undertake NMA step-by-step. Notably, the analysis will develop into three phases to control the quality: (1) group the interventions into categories based on the magnitude of the effect; (2) conduct a critical appraisal of the evidence-based on GRADE guidance to assess whether the evidence supports causal assumptions and its quality and (3) rank the interventions’ effectiveness based on the classified categories and assessed evidence quality. Further, we will perform a sensitive analysis to test the robustness of the results. Suppose a single study can change the pooled evidence (eg, relative effects) to influence SUCRA-based intervention ranks.93 In that case, we will investigate whether the change is significant enough to impact the results.
The potential limitation can be the heterogeneity due to different study implementation strategies. If high heterogeneity fails to be synthesised, we propose six hypotheses to explain the variability between studies as possible effect modifiers; subgroup analysis will be used to explore the reasons for heterogeneity and compare each group’s effects. More statistically significant incoherence must be needed to ensure transitivity in the network. Thus, intransitivity in indirect comparisons may be another potential limitation. Thus, we will always examine the effect modifiers before undertaking the analysis and try to find out and report the reasons for the intransitivity.
In this study, we will explore the most effective non-pharmacological intervention led by pharmacists in outpatient settings to help patients, clinical pharmacists, and policy-makers in decision-making. Based on the results of this study, we will conduct an RCT study to verify the feasibility, effectiveness, and hindrances of the interventions for patients with T2DM in a factual environment.
Ethics statements
Patient consent for publication
References
Supplementary materials
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Footnotes
Contributors YW and HW planned and designed this network meta-analysis project. MZ and BW collected and identified data resources. YW conceived and drafted the protocol manuscript. DH and HW revised the manuscript and approved the final version. In practice, YW and BM will monitor each procedure of the review and be responsible for quality control. All authors have read and approved the publication of the protocol. HW and QZ will be the guarantor of the review.
Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.
Competing interests None declared.
Patient and public involvement Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.
Provenance and peer review Not commissioned; externally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.