This protocol was registered with the International Prospective Register of Systematic Reviews (PROSPERO) on 29 August 2022 with registration number CRD42022353977 and has been written under the Preferred Reporting Items for Systematic Review and Meta-analyses Protocols (PRISMA-P) guidelines (see checklist in additional file) (34). We will perform the review according to the Cochrane handbook for systematic reviews of interventions (35).
Type Of Studies
We will include randomised controlled trials (RCT), cluster-randomised controlled trials, non-randomised controlled trials, cohort studies with a concurrent comparison group and controlled-before-after studies and interrupted-time series studies (36). The latter will be included if three measures before and after the intervention were conducted (37)
Type Of Participants
Patients with COPD diagnosed according to international standards (2). We will include all studies, in which at least 80% are COPD patients or that report the results for COPD patients and other patients (e.g. asthmatic patients) separately.
Type Of Interventions
We will include studies that analyse an intervention which could have a direct or indirect positive impact on patient adherence. We are particularly interested in studies that aim to improve the management, intake or administration of the entire COPD pharmacological and oxygen COPD therapy (i.e. therapy management programs) because we assume that adherence problems, at least in part, arise from the complexity of the whole COPD therapy. Moreover, the optimal treatment outcome can only be reached if all different types of therapies are correctly used and are geared to each other. Nevertheless, we will include studies that examine adherence interventions only targeting certain types of therapies (e.g. inhalers) as some patients might only get a single type of therapy. However, we will include inhalation technique training interventions only if the were part of an adherence intervention. In this way, we will receive additional evidence on the effectiveness of individual intervention components (see section data synthesis).
All types of adherence measures are eligible. This includes education (e.g. information material), behavioural counselling (e.g. motivational interviewing), managing support, reminder and incentives.
Type Of Comparators
The comparator study arm must be either no adherence-enhancing intervention (i.e. usual care) or another adherence-enhancing intervention.
Type Of Outcome Measures And Prioritisation
We plan to perform two patient interviews (one before and one after the systematic review) and follow a sequential approach to integrating qualitative and quantitative information. The first interview aims to understand patients’ needs and prioritise the selection of the outcomes. The second interview will be conducted after the evidence synthesis to present the results to patients.
Overall, outcomes may be as follow:
Primary outcomes
-
Adherence: it is categorised into three stages: initiation, implementation and discontinuation. Sometimes, the term ‘persistence’ is added. We will analyse each component as a part following what is available in the literature
-
COPD Exacerbations
-
Functional exercise capacity
-
Health-Related Quality of Life (HRQoL)
Secondary outcomes
-
Hospital admission: if possible, we will analyse hospitalisation beyond the emergency department (e.g. pneumology department, intensive care unit) separately.
-
Mortality
-
Inhaler technique
-
Respiratory function:
-
Forced expiratory volume at 1 second (FEV1)
-
Tiffneau coefficient: FEV1/CV (as reported by authors)
-
Adverse events
Search Methods For Identification Of Studies
Electronic searches
We will search for all published and unpublished studies regarding adherence-enhancing interventions for COPD. We will develop a comprehensive literature search strategy in collaboration with an experienced librarian and without restrictions on language and publication status (e.g. published, unpublished, ongoing). The search strategy will follow the Peer Review of Electronic Search Strategies (PRESS) guideline.
We will search the following databases to identify relevant studies:
-
MEDLINE and MEDLINE in process (via PubMed): inception to present;
-
EMBASE (via EMBASE): inception to present;
-
CENTRAL (via Cochrane Library): inception to present;
-
CINHAL (via EBSCO): inception to present;
We will search manually for additional studies by:
We will search the following trial registries:
The search strategy will be combined with a highly sensitive filter for RCTs and a study filter for comparative non-randomised studies (38,39). Our pubmed search strategy is detailed in Table 1.
Table 1
Pubmed (16.09.2022; 2609 Hits)
|
("chronic obstructive lung disease"[tiab] OR "chronic obstructive lung diseases"[tiab] OR "chronic airflow obstruction"[tiab] OR "chronic airflow obstructions"[tiab] OR "chronic airway obstruction"[tiab] OR "chronic airway obstructions"[tiab] OR "chronic obstructive lung disorder"[tiab] OR "chronic obstructive lung disorders"[tiab] OR "chronic obstructive pulmonary disease"[tiab] OR "chronic obstructive pulmonary diseases"[tiab] OR "chronic obstructive pulmonary disorder"[tiab] OR "chronic obstructive pulmonary disorders"[tiab] OR copd[tiab] OR coad[tiab] OR cobd[tiab] OR cold[tiab] OR emphysema[tiab] OR "Pulmonary Disease, Chronic Obstructive"[mh])
|
AND
|
(adherence[tiab] OR adherent[tiab] OR adhere[tiab] OR nonadherence[tiab] OR non-adherence[tiab] OR nonadherent[tiab] OR non-adherent[tiab] OR compliance[tiab] OR “patient compliance”[mh] OR medication adherence[mh] OR compliant[tiab] OR comply[tiab] OR noncompliance[tiab] OR non-compliance[tiab] OR noncompliant[tiab] OR non-compliant[tiab] OR “patient compliance”[mh])
|
AND
|
(((cohort[all] OR (control[all] AND study[all]) OR (control[tw] AND group*[tw]) OR epidemiologic studies[mh] OR program[tw] OR clinical trial[pt] OR comparative stud*[all] OR evaluation studies[all] OR statistics as topic[mh] OR survey*[tw] OR follow-up*[all] OR time factors[all] OR ci[tw]) NOT ((animals[mh:noexp] NOT humans[mh:noexp]) OR comment[pt] OR editorial[pt] OR review[pt] OR meta analysis[pt] OR case report[tw] OR consensus[mh] OR guideline[pt] OR history[sh]))
|
OR
|
(“Randomized controlled trial”[pt] OR “controlled clinical trial”[pt] OR random* [tiab] OR "clinical trials as topic"[mh:noexp] OR trial[ti] NOT ((animals[mh:noexp] NOT humans[mh:noexp]) OR comment[pt] OR editorial[pt] OR review[pt] OR meta analysis[pt] OR case report[tw] OR consensus[mh] OR guideline[pt] OR history[sh])))
|
Searching Other Resources
We will check the reference lists of all included primary studies and systematic reviews on the same topic for additional references. We will search EPISTEMONIKOS to identify relevant systematic reviews. When appropriate, we will contact experts in the field to ask for any ongoing trials or newly published papers.
Data Collection And Analysis
We will follow the recommendations of the Cochrane handbook when conducting the screening process, data extraction and management (35).
Selection Of Studies
Two review authors, one with clinical expertise and one with methodological expertise will independently perform study selection. They will screen the titles and abstracts of the search results using Rayyan and code them as 'retrieve' (eligible or potentially eligible/unclear) or 'do not retrieve'. Subsequently, the same reviewers will retrieve the full text of all potentially relevant titles/abstracts and screen them for inclusion while recording the reasons for excluding ineligible studies. In case of discrepancies, a discussion will determine eligibility until consensus. If necessary, a third person will be involved. We will record the selection process in sufficient detail to complete a PRISMA flow diagram (34).
Data Extraction And Management
Two review authors will use a data collection form piloted on at least one study in the review to extract characteristics from included studies. One of the reviewers will be a statistician or epidemiologist. We will extract data using an Excel spreadsheet. Any missing information will be recorded as unclear or not described. Descriptive data (e.g. study characteristics) will be extracted by one reviewer and verified by a second. Two review authors will independently extract outcome data from included studies (40). We will report if outcome data were not reported in a usable way. If multiple reports from the same study are identified, we will directly extract data from all reports into a single data collection form.
We will extract the following study characteristics from the included studies:
-
General information: study ID, author contact details, study centres, locations, and setting
-
Methods: study design, total study duration.
-
Participants: inclusion criteria, exclusion criteria, total number randomised, number randomised per group, age, gender, COPD stage, smoking history, clusters (if applicable)
-
Intervention/comparison groups: We will follow the template for intervention description and replication (TIDieR) (41). For further details, please refer to table 2.
-
Outcomes: primary and secondary outcomes, baseline measures, measurement instrument, and time points.
-
Notes: funding for studies and notable conflicts of interest of trial authors.
Table 2: The TIDieR (Template for Intervention Description and Replication) Checklist (1)
Item
|
Item number
|
Extraction
|
Brief name
|
1
|
Extract the name or a phrase that describes the intervention.
|
Why
|
2
|
Extract any rationale, theory, or goal of the elements essential to the intervention.
|
What
|
3
|
Materials: extract any physical or informational materials used in the intervention, including those provided to participants or used in intervention delivery or in training of intervention providers. Provide information on where the materials can be accessed (e.g. online appendix, URL).
|
4
|
Procedures: extract each of the procedures, activities, and/or processes used in the intervention, including any enabling or support activities.
|
Who provided
|
5
|
For each category of intervention provider (e.g. psychologist, nursing assistant), extract their expertise, background and any specific training given.
|
How
|
6
|
Extract the modes of delivery (e.g. face-to-face or by some other mechanism, such as internet or telephone) of the intervention and whether it was provided individually or in a group.
|
Where
|
7
|
Extract the type(s) of location(s) where the intervention occurred, including any necessary infrastructure or relevant features.
|
When and how much
|
8
|
Extract the number of times the intervention was delivered and over what period of time including the number of sessions, their schedule, and their duration, intensity or dose.
|
Tailoring
|
9
|
If the intervention was planned to be personalised, titrated or adapted, then extract what, why, when, and how.
|
Modification
|
10
|
If the intervention was modified during the course of the study, extract the changes (what, why, when, and how).
|
How well
|
11
|
Planned: if intervention adherence or fidelity was assessed, extract how and by whom, and if any strategies were used to maintain or improve fidelity, extract them.
|
12
|
Actual: if intervention adherence or fidelity was assessed, extract the extent to which the intervention was delivered as planned.
|
- Hoffmann TC, Glasziou PP, Boutron I, Milne R, Perera R, Moher D, et al. Better reporting of interventions: template for intervention description and replication (TIDieR) checklist and guide. BMJ. 2014 Mar 7;348:g1687.
Risk Of Bias Assessment
Two review authors will independently assess the risk of bias (RoB) of each outcome following criteria outlined in the Cochrane Handbook for Systematic Reviews of Interventions (35). Discrepancies will be resolved in a discussion until a consensus is reached.
The risk of bias of RCTs will be assessed with the Cochrane risk of bias 2.0 tool (42). We will use the RoB 2 Excel tool to complete RoB 2 assessment, and the robvis tool to generate “traffic light” plots of the domain-level judgements for each outcome and weighted bar plots of the distribution of 'Risk of bias' judgments within each bias domain (43). We will assess the risk of bias, which can be low, some concern or high, according to the following domains:
-
Bias arising from the randomisation process
-
Bias due to deviations from intended interventions
-
Bias due to missing outcome data
-
Bias in measurement of the outcome
-
Bias in selection of the reported result
Our effect of interest is starting intervention. We will judge each outcome as being at low risk, some concerns, or high risk according to the RoB2 algorithm. We will provide a quote from the study report and a justification for our judgment in the 'Risk of bias' table. We will report information on the risk of bias relates to unpublished data or correspondence with a trialist.
For cluster-RCTs, we will use the test version of RoB 2.0 for this study design (10 November 2020, revised 18 March 2021) (44).
The risk of bias of non-randomised studies will be assessed with ROBINS-I (35,45), considering it low, moderate, serious or critical. We will assess the following domains:
-
Bias due to confounding
-
Bias in selection of participants into the study
-
Bias in classification of interventions
-
Bias due to deviations from intended interventions
-
Bias due to missing data
-
Bias in measurement of the outcome
-
Bias in selection of the reported results
We will report the risk of bias assessment in the results section. It will also be part of the GRADE assessment of the certainty of evidence (along with precision, directness, consistency, and publication bias). When considering treatment effects, we will take into account the risk of bias for the studies that contribute to that outcome. Our primary analysis will include all studies without considering the risk of bias assessment.
In addition to risk of bias, we will assess the quality of recruitment strategy (46).
Measures Of Treatment Effect
The choice of the summary effect depends on the type of the outcome and how they were reported. We will use risk ratio (RR) for dichotomous outcomes, and we will analyse continuous outcomes as a mean difference (MD) or, when needed, as a standardised mean difference (SMD) (i.e. combine different scales).
For time-to-event data (as reported by the authors), our treatment effect will be the hazard ratio (HR).
Data Synthesis
To determine the degree of complexity, we will use the iCAT_SR checklist (47). Based on this assessment and the logical model, the findings will be systematically tabulated and graphically displayed (48).
We will group the interventions according to the following criteria to explore which factors might affect the effectiveness of adherence-enhancing interventions:
-
Intervention target: inhalers, oral medications, oxygenation, or multiple types of adherence intervention: education, behavioural counselling, managing support, reminder and incentives
-
Number of effective components (e.g. education [1] vs. education plus reminder [2], as determined with iCAT_SR))
-
Behaviours or actions of intervention recipients or participants to which the intervention is directed (as determined with iCAT_SR)
-
The duration of the intervention (short, medium and long-term intervention)
-
Degree of tailoring (as determined with iCAT_SR) to the individual patients
-
Targeted adherence type: unintentional vs. intentional non-adherence (49)
-
Optionally, further if suggested by the patient interviews or logical model.
We will prepare harvest plots and forest plots (with or without pooled estimates) (50) and perform meta-analyses for studies with sufficient clinical and methodological homogeneity. Then, statistical heterogeneity will be explored using prediction intervals.
We will perform two types of meta-analyses to assess the effectiveness of adherence-enhancing interventions (components).
We will run pair-wise random-effects meta-analyses using the Paule-Mandel heterogeneity variance estimator and (modified) Hartung-Knapp confidence intervals (CIs) to determine the overall effectiveness of the adherence interventions (51). The variance correction factor for the Hartung-Knapp confidence intervals will only be applied if the 95%-CIs of the conventional Hartung-Knapp CIs are narrower than Wald-type CIs. We will use beta-binomial models for meta-analyses of less than five studies and zero event studies (52,53), Bayesian random-effects meta-analyses with weakly informative priors for tau-square for sensitivity analysis, and Bayesian random-effects meta-analyses with weakly informative priors for the treatment effect for zero events studies (54,55).
In addition, we will conduct random-effects component network meta-analyses to identify the most effective components and combinations of components (50,56). We will check if the transitivity assumption is met. Network meta-analyses has recently been shown promising to offer additional insights when analysing adherence interventions (57). In these models, each adherence intervention component (e.g. education or reminder) will be treated as a separate component (separate node in the network). We will build two types of models. First, an additive model assuming that each component has a fixed effect. This model will answer the question of the most effective adherence intervention/component. The second is an interaction model in which different types of adherence interventions can interact. This model will answer the question of which adherence-enhancing components have the strongest synergies.
All analyses will be performed using R (package meta, bayesmeta, and netmeta), and SAS.
In addition to the meta-analyses, we will carry on a structured narrative synthesis to understand which patients benefit from which interventions and contexts. For this analysis, the results of the structured tabulations and graphical displays will be contrasted with the patient characteristics and study characteristics (e.g. setting). Furthermore, the narrative synthesis will incorporate information from the second round of patient interviews.
Subgroup Analysis
Furthermore, we plan to perform the following subgroup analysis:
Sensitivity Analysis
We plan to conduct a sensitivity analysis by removing RCTs at high risk of bias and some concern, and non-RCTs judged at critical and serious risk of bias.
Confidence In Cumulative Evidence (Grade Assessment)
We will assess the certainty of the body of evidence for each prioritised outcome with GRADE and prepare a summary of findings tables (58). We will use the methods and recommendations described in the Cochrane handbook (35) using GRADEpro GDT software (59). We will justify all decisions to downgrade the quality of studies using footnotes and make comments to explain the summary of the evidence.
We will follow the GRADE guidance for using ROBINS-I to facilitate the integration of non-randomised and randomised studies in the body of evidence (60).
Meta-bias
To detect reporting bias, we will compare the study protocol with the published report, if possible. We will use the ROB-ME tool to assess the risk of bias due to missing evidence in the synthesises (61). We will try to contact the study authors to identify missing or partially reported data. If more than ten studies are included in the meta-analysis, we will create a funnel plot to explore publication bias. We will use the Copas selection model-based tests suggested by Duan et al. to test for publication bias (62).