Introduction

Statin medications reduce low-density lipoprotein cholesterol (LDL-C) blood concentrations and cardiovascular events in patients with hypercholesterolemia, and guidelines recommend statin therapy to lower LDL-C in patients who are at risk for developing or have known atherosclerotic cardiovascular disease [1]. Despite evidence for the benefits of statins, the medications are widely underutilized [2,3,4,5,6]. Previous studies highlight both patient- and prescriber-barriers to statin use including side effects, competing medical conditions, busy clinics, and patient reluctance affecting adherence to prescribed medications [7,8,9]. Lack of adherence is associated with increased mortality in a dose dependent relationship [10].

Implementation strategies can be used to promote the uptake of interventions, such as statin therapy, and are defined as “methods or techniques used to enhance the adoption, implementation, and sustainability of a clinical program or practice” [11]. Numerous implementation strategies have been attempted to improve utilization of statins, all with varying degrees of success. These studies have targeted a variety of actors (e.g., patients, clinicians, or systems) and employed a variety of implementation strategies (e.g., education, reminders, or financial incentives). A computer-based clinical decision support system to aid in prescribing of evidence-based treatment for hyperlipidemia, which targeted clinicians, was found to significantly reduce blood LDL-C concentrations [12]. However, when providing financial incentives to providers, patients, or both, a study found that only the combination incentive was successful in reducing LDL-C levels to target [13]. The absolute and comparative effectiveness of these strategies, however, is unclear. Knowing which strategies are most effective can facilitate the uptake of statins and lead to reduce mortality.

To address this issue, we aimed to address the following key questions:

  1. 1.

    What implementation strategies have been used to promote the uptake of statins?

  2. 2.

    How completely are the implementation strategies utilized reported in studies designed to promote statin uptake?

  3. 3.

    Which implementation strategy, or combination of strategies, is (are) the most effective at promoting the uptake of statins?

We conducted a systematic review of studies aimed at improving statin use and categorized implementation strategies by the Expert Recommendations for Implementing Change (ERIC) compilation [14]. Our primary objective was to better understand the impact of specific implementation strategies on the utilization of statins in patients with hypercholesterolemia. Our secondary objective was to evaluate statin adherence, statin prescribing, and lowering of LDL-C after intervention.

Methods

This registered (PROSPERO CRD42018114952) systematic review adhered to the reporting guidelines of the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) statement [15].

Search strategy

A medical librarian (L.H.Y.) searched the literature for records including the concepts of hypercholesterolemia, hyperlipidemia, and statins. The search strategies used a combination of keywords and controlled vocabulary and searched the following databases from inception to October 2018: MEDLINE, Embase, Scopus, Cochrane Database of Systematic Reviews, Cochrane Central Register of Controlled Trials, and Clinicaltrials.gov. References were imported into Endnote™ and duplicates were identified and removed. An example of the search string can be found in Table 1 and the fully reproducible search strategies for each database can be found in Additional file 1: Appendix 1.

Table 1 Example search string
Full size table

Study selection

We included studies reported in English, regardless of the country where the study was conducted, that had at least one strategy promoting statin uptake that could be categorized using the ERIC compilation [14, 16]. Seven manuscripts were excluded for this reason. The ERIC compilation was created so that researchers have a standardized way to name, define, and categorize implementation strategies. The ERIC compilation was selected for use in this review because the implementations strategies in the included articles most closely matched the ERIC taxonomy compared to other available choices [17]. For key questions 1 and 2, we did not limit inclusion based on study design or outcome. For key question 3, we limited inclusion to randomized controlled trials (RCTs). Studies were excluded for key questions if full text was not available.

Search results were uploaded into systematic review software (DistillerSR, Ottawa, Canada). In the first round of screening, abstracts and titles were evaluated for inclusion. Following abstract screening, eligibility was assessed through full-text screening. Prior to both abstract and full text screening, reviewers underwent training to ensure a basic understanding of the background of the field and purpose of the review as well as comprehension of the inclusion and exclusion criteria. The initial 20 abstracts were reviewed independently and then discussed as a group. Eligibility at both levels (abstract and full-text) was assessed independently and in duplicate (L.K.J., S.T., L.R.F., and C.G.). Disagreements at the level of abstract and full text screening were resolved by consensus. If consensus could not be achieved between the two reviewers, a third reviewer arbitrated (M.R.G., T.W., or T.S.).

Data collection

The following characteristics were extracted from included studies: first author, year of publication, location, age of patient population (adult vs. child), study design, implementation strategies, inclusion and exclusion criteria, and any of the following outcomes: statin prescribing or use, statin adherence, or LDL-C measurements.

Key question 1: what implementation strategies have been used to promote the uptake of statins?

We first summarized and described the populations, interventions, comparisons, and outcomes presented for all studies that reported at least one implementation strategy that could be mapped to the ERIC compilation. The ERIC compilation of nine implementation strategies categories (73 total strategies) was applied to each of the interventions to (1) count the total number of strategies and (2) describe how complete each implemented strategy was defined. One study team member, who was an author on the original ERIC compilation, ensured validity of the categories selected (T.W.) [14].

Key question 2: how completely are the implementation strategies utilized reported in studies designed to promote statin uptake?

Based on guidance from proctor and colleagues, we assessed the degree to which each strategy was completely reported including actor, action, action target, temporality, dose, implementation outcome affected, and justification (Table 2) [11].

Table 2 Summary of the implementation strategies’ defining characteristics
Full size table

Key question 3: which implementation strategy, or combination of strategies, is (are) the most effective at promoting the uptake of statins?

When present, we extracted data related to statin prescribing, statin adherence, and LDL-C reported from included RCTs. All outcomes were collected at intervention completion. Statin prescribing or use included all orders for statin medications. Statin adherence included only objective measures of adherence by either medication possession ratio (MPR) or proportion of days covered (PDC) [18]. MPR or PDC were captured as a percentage or attainment of greater than 80% adherence. LDL-C levels were recorded as LDL-C measured or achievement of an LDL-C target.

Risk of bias assessment

The Cochrane Collaboration’s risk of bias tool version 2 to evaluate methodological quality of studies included in the meta-analysis for key question 3 [19]. The risk of bias in included studies was assessed in duplicate by two reviewers (L.K.J. and L.R.F.) working independently. Any disagreements were resolved by consensus; if consensus was unable to be achieved, a third reviewer arbitrated (M.R.G.).

Statistical analysis

Standardized mean differences (SMDs) with corresponding 95% CIs were estimated for continuous outcomes, and odds ratios (ORs) and 95% confidence intervals (CIs) were calculated for binary outcomes from included studies. Publication bias was evaluated by Egger’s test [20]. Variability between included studies was assessed by heterogeneity tests using I2 statistic [21]. If overall results showed significant heterogeneity, potential sources of heterogeneity were explored by subgroup analysis. All analyses were conducted using RStudio (Version 1.0.136) using the “Meta” and “Metafor” package.

Results

Description of study selection

We initially identified 65,118 studies. After removing duplicates, we identified 38,585 unique citations (Fig. 1). Through abstract and title screening, 208 reports were identified for full-text review. During full-text review, 86 were selected for inclusion [12, 13, 22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76,77,78,79,80,81,82,83,84,85,86,87,88,89,90,91,92,93,94,95,96,97,98,99,100,101,102,103,104,105]. A complete list of excluded full-text studies with rationale for exclusion is available in Additional file 1: Appendix 2.

Fig. 1
figure 1

PRISMA flow diagram

Full size image

Description of studies

Table 3 describes the included studies (more details are included in Additional file 1: Appendix 3). Almost all the implementation strategies targeted adults (two studies included pediatric patients), half were implemented in the USA, and almost all were conducted in individuals with hypercholesterolemia (two studies were conducted in individuals with familial hypercholesterolemia).

Table 3 Study demographics
Full size table

Implementation strategies

All implementation strategies except “provide interactive assistance” were used (Table 4). A total of 258 uses of strategies were identified across 86 studies. On average, each study utilized three strategies (SD 2.2, range 1–13). The most utilized strategies were “train and educate the stakeholders” (studies utilized strategies in this grouping 79 times), “support clinicians” (68), and “engage consumers” (47). The most utilized individual strategies were “intervene with patients and consumers to enhance uptake and adherence” (41), and “distribute educational materials” (41) (Additional file 1: Appendix 4). Implementation strategies often did not include key defining characteristics: temporality was reported 59% of the time, dose 52%, affected outcome 9%, and justification 6% (Table 2 provides a summary and Additional file 1: Appendix 5 provides a more detailed version).

Table 4 Summary of implementation strategies by strategy category
Full size table

Meta-analysis

Due to the large heterogeneity between studies, effectiveness outcomes (statin prescribing, statin adherence, and LDL-C) were only extracted from RCTs. Thirty-one trials reported at least one of the three outcomes of interest. The implementation strategies examined demonstrated: significantly reduced LDL-C (LDL-C reduction: SMD − 0.17, 95% CI − 0.27 to − 0.07, p = 0.0006; met LDL-C target: OR 1.33, 95% CI 1.13 to 1.58, p = 0.0008) (Fig. 2), increased rates of statin prescribing (OR 2.21, 95% CI 1.60 to 3.06, p < 0.0001) (Fig. 3), and improved statin adherence (PDC/MPR: SMD 0.13, 95% CI 0.06 to 0.19; p = 0.0002; ≥ 80% PDC/MPR: OR 1.30, 95% CI 1.04 to 1.63, p = 0.023) (Fig. 4). There was inconsistency across trials based on the outcome measured; statin prescribing (I2 = 73%), statin adherence (I2 = 0%), and LDL-C (I2 = 79% (LDL-C reduction) and 76% (met LDL-C targets)). Publication bias using the Egger’s test indicated no publication bias for statin prescribing (p = 0.63), statin adherence (p = 0.83 for SMD, p = 0.22 for OR), and potential publication bias for LDL-C (p = 0.08 for SMD, p = 0.01 for OR).

Fig. 2
figure 2

Forest plot of implementation strategies’ impact on LDL-C compared to control. a Achievement of target LDL-C. b Standardized mean difference in LDL-C

Full size image
Fig. 3
figure 3

Forest plot of implementation strategies’ impact on statin prescribing compared to control

Full size image
Fig. 4
figure 4

Forest plot of implementation strategies’ impact on statin adherence compared to control. a Medication possession ratio or portion of days covered > 80%. b Standardized mean difference in medication possession ratio or portion of days covered

Full size image

Although subgroup analyses were conducted for statin prescribing and LDL-C, there were not enough studies to conduct a subgroup analysis for statin adherence (Table 5). We identified a significant difference among studies published in 2013 or later for LDL-C measured as a binary outcome (OR 1.62, 95% CI1.19–2.19, p = 0.05). We also found a significant effect on LDL-C measured as a continuous variable when more than 2 implementation strategies were utilized (SMD − 0.38 95% CI − 0.67; − 0.09, p = 0.05). There was no significant effect in the between country analysis.

Table 5 Subgroup analyses
Full size table

Most studies were found to be at a low risk of bias (Fig. 5 and Additional file 1: Appendix 6); therefore, we did not conduct subgroup analyses based on the risk of bias.

Fig. 5
figure 5

Risk of bias of RCTs included in the meta-analyses

Full size image

Discussion

Our findings

In this review of implementation strategies regarding uptake of statins in hypercholesterolemia, we found that 38 different strategies were utilized to lower LDL-C, improve statin prescribing, and promote adherence. However, strategy components were not well defined and there was not a single strategy or group of strategies that demonstrated superior impact compared to others. Consistent with management of other diseases and conditions and literature from implementation science [106], we found evidence to support the use of multiple concurrent strategies; the use of three or more implementation strategies was associated with a greater reduction in LDL-C. We also found that studies published after 2012 had, on average, greater reductions in LDL-C through the use of the reported implementation strategies. While it cannot be definitely attributed, this could result from a better understanding of which strategies work best or could reflect a switch toward the utilization of high dose statin therapy. There was no difference in outcomes based on country where the study was conducted.

An important limitation of the many strategies described was incomplete definitions, limiting generalizability to other settings. Often, we were able to discern the actor, action, and action target but were unable to determine temporality, dose, implementation outcome affected, or justification. Without clear reporting of these factors, we are unable to interpret when these strategies should be used (temporality), how often (dosage), how the success of a specific strategy is measured (implementation outcomes affected), or when to justify the choice of a particular strategy (justification) to influence clinical practice. While the interventions appeared to be effective at increasing the utilization of statins and reducing LDL-C overall, the variable nature of the interventions studied and outcomes examined, the effectiveness of any specific strategy or set of strategies was unclear.

In addition, one category of strategies, “provide interactive assistance,” was not utilized in any of the studies included in the analysis. Among the strategies that were used, many were used in combination, but specific combinations were not used frequently enough to permit reliable subgroup analysis.

Comparison with other studies

In the field of implementation science, there has recently been a desire to improve specification of implementation strategies utilized in practice and to develop standard language and definitions for reporting these implementation strategies [11, 14, 107]. This trend has led to the development of two implementation strategy taxonomies: the ERIC compilation [14], used in this study, and the Effective Practise and Organization of Care (EPOC) taxonomy [17]. Use of these taxonomies has allowed for consistent language in reporting implementation strategies and development of tailored compilations of strategies specific to certain disease states [108, 109]. Other systematic reviews of implementation strategies in other fields (i.e., intensive care setting and oral health) have found improved outcomes when multiple implementation strategies are used but have not been able to identify the groups of strategies most likely to produce the most favorable outcomes [110,111,112].

An investigation of enablers and barriers to treatment adherence in familial hypercholesterolemia found seven enablers for patients that could be used to develop new interventions and matched to implementation strategies we identified in our study [113]. These enablers were “other family members following treatment regime,” “commencement of treatment from a young age,” “parental responsibility to care for children,” “confidence in ability to successfully self-manage their condition,” “receiving formal diagnosis of familial hypercholesterolemia,” “practical resources and support for following lifestyle treatment,” and a “positive relationship with healthcare professionals” [113]. By linking the two most frequently used strategies identified in our systematic review “intervene with patients and consumers to enhance uptake and adherence” and “distribute educational materials,” with the enablers identified above, effective implementations strategies for statin utilization can be designed.

The sustainability of interventions to promote the uptake of guidelines when intervening at the clinician level has been limited in a variety of settings [114,115,116]. Specifically, in cardiovascular disease, a systematic review of interventions to improve uptake of heart failure medications saw an increase in guideline uptake but not improvement in clinical outcomes [117]. Similar findings have been found in hypertension [118]. However, the success of these interventions have been limited.

Limitations and strengths

Our review is the first to comprehensively map the strategies used to increase utilization of statins among persons with hypercholesterolemia to the ERIC compilation. We chose to use ERIC due to a perceived better fit over alternatives (i.e. EPOC); however, we identified 7 studies (out of 208 identified) which could not be mapped to ERIC, exclusion of which could lead to missing important strategies. Other strengths include utilization of a medical librarian to conduct the search, searching of multiple databases which covered parts of the gray literature, and utilizing trained reviewers. Finally, we limited our search to studies in English with full-texts available. Thus, we may have missed studies not published in English or published in the gray literature (e.g., only conference abstract available in published literature) and be at risk for language bias [119] or publication bias [120]. While the Egger’s test suggested possible publication bias, we think that the risk of this is low due to our comprehensive search strategy. Further, while language bias is a possibility [119], few studies were excluded based on language so any potential impact is likely to be small.

Suggestions for future research

Consistent strategies for reporting LDL-C would significantly improve the ability to assess efficacy of an intervention. Some studies used arbitrary cut-offs for LDL-C, some used absolute values, and others used thresholds published in cholesterol guidelines [121]. This led to difficulty in aggregating data across studies. Future studies should report absolute values of LDL-C to facilitate meta-analyses directed at change of LDL-C with intervention. Generating a core outcome set for trials in hypercholesterolemia would facilitate meta-analyses and ensure all relevant outcomes are consistently measured [122]. Ideally, these studies should be registered and included in a meta-analysis in a prospective manner [123].

Clarity in the terminology, definition, and description of implementation strategies by researchers would help translation and replication of efforts. Completely reporting implementation strategies facilitates interpretation of results as well as facilitating reproducibility and scalability [11]. The field of implementation science offers guidance on how to name and report these strategies [11]. Even though this study was unable to identify a single or gold standard approach to improving statin therapy for hypercholesterolemia disorders, it provides examples of many different approaches that have some impact on outcomes relevant to care. In this way, this study provides a roadmap for future implementation to better define implementation strategies and to rigorously define and test the outcomes associated with those strategies. More guidance will be needed on the impact of strategies in different healthcare settings, because different strategies may work better in different healthcare settings so these idiosyncrasies need to be understood.

Conclusion

Implementation strategies to improve the uptake of statins among patients with hypercholesterolemia exist but they are poorly reported and generalizability is limited. While these strategies lowered LDL-C and improved adherence, significant heterogeneity made assessment of the comparative effectiveness of strategies difficult. Future studies for increasing the utilization of statins among patients with hypercholesterolemia should more clearly define strategies used, prospectively test comparative effectiveness of different strategies, and use standardized efficacy endpoints.