Effect of drug reminder packaging on medication adherence: a systematic review revealing research gaps

Adherence is defined as the extent to which a patient’s behavior matches the agreed recommendations from the prescriber [1]. Reported rates vary from 4.6% to 100% of patients of all age classes with different medical conditions and on long- or short-term treatments [1, 2]. Mean adherence rates for specific diseases are 88.3% for HIV infection, 76.6% for cardiovascular disease, 67.5% for diabetes mellitus and 58% for psychosis patients [2, 3]. Adherence depends on patients’ capability (e.g., physical, cognitive and economic) and willingness to initiate and execute their treatment plan: if either is insufficient, unintentional or intentional non-adherence will be the consequence [4, 5]. Non-adherence is known to impair clinical, economic and humanistic outcomes [6–12]. In a study across five European countries, increasing the percentage of patients adhering to antihypertensive treatment to 70% was estimated to lead to a reduction of cardiovascular related health-care costs by €332 million ($461 million) [13]. Reasons for non-adherence are highly individual and complex. Therefore, individual needs and necessities have to be assessed to find the optimal aid for each patient.

Drug reminder packaging, such as weekly pillboxes or multidrug punch cards, is widely used in everyday practice. It usually consists of a certain number of compartments containing solid oral medication for specific dosing times. Compared to other adherence-enhancing programs, such as patient counseling, education or motivation [14], drug reminder packaging is a simple technical option and requires little resources on the patient’s as well as on the provider’s side. The provision of drug reminder packaging aims at enhancing adherence by facilitating medication organization and intake, by decreasing medication errors and by (self-) monitoring medication intake. Various authors suggest that drug reminder packaging supports mainly unintentionally non-adherent patients, e.g., geriatric patients and patients with complex drug regimens [4, 15–17]. Previous reviews with restrictive inclusion criteria investigated the effect of reminder packaging on adherence and were inconclusive [18–20]. This review uses evidence mapping [21, 22] to analyze data from a different perspective, highlighting methodological strength and completeness of information as well as research gaps, to identify areas for future research.

A systematic review was conducted, complying with the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) statement. We proceeded by following the evidence mapping methodology in four steps: question development, question prioritization, evidence search and selection, and data extraction [21].

Of the total 855 identified references, 30 fulfilled the inclusion criteria. The PRISMA flow diagram of study inclusion and the PRISMA checklist are provided in Additional files 1 and 2, respectively. According to the EPHPP assessment tool for study design, 10 studies were RCTs, 19 controlled clinical trials and 1 was a cohort study (one group with a pre- and post-intervention comparison). Compared to the previously published reviews [18–20], a total of 13 studies were additionally included, from which 7 were controlled clinical trials, 5 RCTs and 1 was a cohort study.

Methodological quality and completeness of information

Methodological quality was rated as strong for 5 studies, moderate for 12 and weak for 13. Overall, weaknesses were in the methods used for data collection (mostly not valid and not reliable) and the report of confounders and their comparison between groups (insufficient or missing). The most accurate standard in statistical analysis, the intention-to-treat analysis, was applied by seven studies. The number of studies with strong and moderate methodological quality doubled after 1996, the year of the first publication of the CONSORT statements [59], while the number of weak methodological quality studies diminished by a factor of 3.

Completeness of information ranged from 0 to 1.0 with a mean score of 0.3. Two studies [40, 55] gave complete information for all required details. Reported criteria for the completeness of information are depicted in Figure 1. Criteria 7 and 8 were not applicable for 5 studies [40, 53–56] and practically non-existent in all 25 remaining studies (criterion 7: 0; criterion 8: 1). Information on the person in charge and place of intervention were often missing from the description of the intervention and control conditions. Figure 2 shows the included studies according to their methodological quality, completeness of information and outcome measures.

Although more than half of the studies included in this review reported significant effects, only three studies were graded as methodologically strong. Drug reminder packaging had a significant effect on adherence in a geriatric population [30], for chronic mental illness [31] and for cardiovascular disease [40]. The overall effect of drug reminder packaging on adherence parameters remains inconclusive, as reported by previous reviews with more restrictive selection criteria [18–20]. Three studies reported a significant effect on adherence but not on clinical outcomes [31, 50, 56]. Thus, the question of how much adherence is necessary for altering treatment success is raised and there is a requirement to present the clinical benefits for the patients [60]. We observed that drug reminder packaging offers a broad field of application and is mostly used for polypharmacy. As a consequence, disease-unspecific, generalizable clinical outcomes like morbidity or re-hospitalization rates would provide viable and comparable results rather than measures of disease-specific clinical parameters. Only two trials investigated such outcomes [44, 58], with one showing that drug reminder packaging significantly reduced the mean hospitalization rate.

We included five RCTs in the evidence map that were excluded by three previous reviews [18–20] because of their multiple intervention design. In a direct comparison (factorial trials), the effect was higher with multiple interventions, which is consistent with previous findings [14, 18]. Yet, the evidence is limited, for these trials were graded as weak in methodological quality.

The overall methodological quality of the studies included is poor and thus evidence for the effect of drug reminder packaging on adherence is low. We used a quality assessment tool that is applicable to a variety of study designs and was specifically developed to provide research evidence for studies on public health services with a focus on behavior change education [24]. In comparison to previous reviews, we were able to include four additional studies of strong methodological quality [30, 38, 40, 43]. However, information on intervention and control settings was incomplete in three of these additional studies (completeness scores: 0.13, n = 2; 0.25, n = 1). As a consequence, being graded as strong and complying with all the criteria for completeness of information was observed in one out of the 30 studies included [40]. It therefore represents a thin basis for informed clinical decision support.

The increasing number of methodologically strong trials after 1996, the year when the CONSORT statements were released, is intriguing and probably follows from under-reporting in studies published before 1996. Various authors indeed stated that complete reporting of methodological quality according to the CONSORT criteria was inadequate, but that poor reporting did not necessarily correlate with the quality of how the trial was conducted [61–64]. The CONSORT statements of non-pharmacological treatment require ‘precise details of both, the experimental treatment and the comparator’ [27] and omission of trial details has been shown to lead to decreased uptake of trial results into clinical practice [65, 66]. Thus, to obtain valuable and reliable study results, high methodological quality and detailed information are crucial.

Most studies were designed as RCTs, which provide the most reliable results through the minimization of confounding. However, RCTs might not be the appropriate design for all research questions and settings, especially in the field of behavior research. Alternative designs might be worth considering. Firstly, randomized allocation of study participants to a predefined intervention may not be practicable since tailored interventions, in respect of patients’ needs and abilities, are expected to be the most effective [14]. Secondly, in studies on survival outcomes for HIV patients, investigating adherence-enhancing strategies in a randomized controlled fashion has been declared to be ethically difficult [67, 68]. The reason for this declaration was the assumption that allocation to the control condition equaled withholding a tool, which could possibly lead to higher survival rates through an optimal clinical response due to increased adherence [67, 68]. Thirdly, behavioral interventions are often complex and can only be controlled poorly under real-life conditions and therefore randomization might not be practical in a primary care setting [69]. Consequently, confounding could even persist despite randomization. Alternatives to conventional randomization designs, i.e., randomization at the patient level, include pre- and post-cohort studies, historical control studies, pre-randomized designs and cluster randomization [70].

More studies could be included and research gaps identified using our approach of evidence mapping. Patient-relevant disease-unspecific long-term clinical outcomes, e.g., (re-)hospitalization, admission to a nursing home, etc., were neglected. Economic outcomes as defined by Kozma et al. [29] were not reported in any study on drug reminder packaging. This may be due to the fact that drug reminder packaging is generally supposed to be inexpensive, and thus cost-effective. Humanistic outcomes were measured in two studies [52, 56], which is insufficient for judging whether a condition optimally treated through drug reminder packaging leads to increased quality of life. Improved adherence could lead to increased adverse events as well. However, safety issues were reported by two studies only [49, 56]. Patient satisfaction and other aspects of safety, such as opening medication packaging, confusion with new packaging and decreased ability to identify one’s own medication [71–74], were hardly mentioned by the studies.

Our study has strengths. First, evidence mapping allows the inclusion of more studies and gives an overall view of the subject. Second, the tool used to assess methodological quality is independent of study design (EPHPP) and was developed specifically to assess studies within the scope of public health. Third, with completeness of information, a further element for judging quality is added. Fourth, the consolidation of adherence outcomes and economic, clinical and humanistic parameters allows an overall presentation and highlights research gaps. Our study has limitations also, such as the language restriction, which led to the exclusion of articles considered relevant. Information may also have been missed due to the exclusion of studies performed in developing countries.

A suggestion for future research is to develop methodologically strong studies reporting complete information to clarify the effect of drug reminder packaging on medication adherence.

New information was extracted from the 30 studies included and several studies had statistically significant and relevant results for adherence and clinical outcomes with drug reminder packaging. However, firm conclusions cannot be given for the effect of drug reminder packaging on adherence, mainly because the studies lack methodological quality and the information was incomplete. The main research gaps concerned economic, disease-unspecific clinical outcomes and humanistic outcomes. Safety issues and satisfaction with the intervention were marginally reported. Researchers of behavioral interventions might consider alternative study designs for similar research questions, without neglecting methodological issues and reporting important details. Future research should aim at filling the observed gaps with a focus on patient safety and the benefit to patients as well as on implementable and valuable interventions. Drug reminder packaging should be distributed with respect to patient needs, requests and abilities.