https://orcid.org/0000-0003-1559-7885
Figures
Abstract
Background
Depression, osteoporosis, and cardiovascular disease impose a heavy economic burden on society. Understanding economic impacts of suboptimal use of medication due to nonadherence and non-persistence (non-MAP) for these conditions is important for clinical practice and health policy-making.
Objective
This systematic literature review aims to assess the impact of non-MAP to antidepressants, bisphosphonates and statins on healthcare resource utilisation and healthcare cost (HRUHC), and to assess how these impacts differ across medication classes.
Methods
A systematic literature review and an aggregate meta-analysis were performed. Using the search protocol developed, PubMed, Cochrane Library, ClinicalTrials.gov, JSTOR and EconLit were searched for articles that explored the relationship between non-MAP and HRUHC (i.e., use of hospital, visit to healthcare service providers other than hospital, and healthcare cost components including medical cost and pharmacy cost) published from November 2004 to April 2021. Inverse-variance meta-analysis was used to assess the relationship between non-MAP and HRUHC when reported for at least two different populations.
Results
Screening 1,123 articles left 10, seven and 13 articles on antidepressants, bisphosphonates, and statins, respectively. Of those, 27 were rated of good quality, three fair and none poor using the Quality Assessment Tool for Observational Cohort and Cross-Sectional Studies. In general, non-MAP was positively associated with HRUHC for all three medication classes and most prominently for bisphosphonates, although the relationships differed across HRUHC components and medication classes. The meta-analysis found that non-MAP was associated with increased hospital cost (26%, p = 0.02), outpatient cost (10%, p = 0.01), and total medical cost excluding pharmacy cost (12%, p<0.00001) for antidepressants, and increased total healthcare cost (3%, p = 0.07) for bisphosphonates.
Conclusions
This systematic literature review is the first to compare the impact of non-MAP on HRUHC across medications for three prevalent conditions, depression, osteoporosis and cardiovascular disease. Positive relationships between non-MAP and HRUHC highlight inefficiencies within the healthcare system related to non-MAP, suggesting a need to reduce non-MAP in a cost-effective way.
Citation: Park KH, Tickle L, Cutler H (2022) A systematic review and meta-analysis on impact of suboptimal use of antidepressants, bisphosphonates, and statins on healthcare resource utilisation and healthcare cost. PLoS ONE 17(6): e0269836. https://doi.org/10.1371/journal.pone.0269836
Editor: Daoud Al-Badriyeh, Qatar University, QATAR
Received: August 11, 2021; Accepted: May 28, 2022; Published: June 29, 2022
Copyright: © 2022 Park et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: All data used in this study are publicly available in the form of published journal articles. The summary data included in the tables are extracted directly from the articles.
Funding: This research is conducted as part of Kyu Hyung Park's PhD study funded through a program jointly funded by Macquarie University, Australia (https://www.mq.edu.au/) and five pharmaceutical companies - Amgen Australia Pty Ltd (https://www.amgen.com.au), Janssen-Cilag Pty Ltd (https://www.janssen.com/australia/), Merck Sharp & Dohme Australia Pty Limited (https://www.msd-australia.com.au/), Pfizer Australia Pty Ltd (https://www.pfizer.com.au/), and ROCHE Products Pty Limited (https://www.roche-australia.com/). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. The funders provided support in the form of scholarship for an author [KP] through Macquarie University Centre for the Health Economy, the research organisation directed by another author [HC], but did not have any additional role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript. The specific roles of these authors are articulated in the ‘author contributions’ section.
Competing interests: The authors have declared that no competing interests exist.
Introduction
Poor medication adherence or persistence (MAP) is related to increased morbidity and mortality [1–4] and greater healthcare resource utilisation and healthcare cost (HRUHC) [5–7]. Interventions to improve MAP are reported to give positive effects on morbidity, HRUHC and patient satisfaction [8, 9].
Suboptimal use of medication due to nonadherence and non-persistence (non-MAP) is prevalent in chronic conditions [10–12] because a long-term therapy is often interrupted by undesirable medication use, including erratic use, under-use and premature discontinuation of therapy. The prevalence of non-MAP across all chronic conditions has been estimated at approximately 50% by the World Health Organisation [10]. The annual cost of non-MAP to the US healthcare system has been estimated at between USD 100 billion and USD 289 billion [11, 13, 14].
Depression, osteoporosis, and cardiovascular disease (CVD) are among the most prevalent conditions in developed countries [15–17] and impose a large health and economic burden on society [18–21]. For example, in the US, the total annual economic burden of major depressive disorder was estimated at $210 billion [20]; total annual healthcare spending associated with osteoporosis fractures among Medicare beneficiaries at $57 billion [22]; and total annual healthcare system cost for heart disease or stroke at $214 billion [23]. In 2015 in Australia, the three disease groups, CVD, musculoskeletal conditions and mental and substance use disorders, accounted for around 39% of the total burden of disease as measured using disability-adjusted life year (DALY) [21]. Although MAP for these conditions is important in achieving clinical goals [16, 24, 25], reported MAP is relatively low [16, 26, 27].
Antidepressants, bisphosphonates, and statins are medications used for the chronic conditions depression, osteoporosis, and CVD, respectively. Antidepressants aim to correct chemical imbalances of neurotransmitters in the brain responsible for changes in mood and behaviour, bisphosphonates are used to prevent loss of bone density, and statins are used to lower cholesterol. While the minimum recommended length of antidepressant and bisphosphonate therapy is six months [24, 28] and three to five years [29], respectively, discontinuation of statins is generally not recommended [30].
Understanding how and to what extent non-MAP impacts health outcomes, leads to premature death, and exhausts valuable healthcare resources is important for improving clinical practice, developing health policies and prioritising research. Awareness of MAP patterns can help clinicians improve clinical practice and better manage health outcomes by regularly checking MAP, identifying reasons for non-MAP, and implementing interventions to improve MAP including better patient and clinician communication and shared decision making, better support from other health system stakeholders such as community care nurses, better medication packaging, and patient education. More accurate and informative MAP measures enable healthcare policymakers to better evaluate costs and benefits of MAP policies and interventions. In addition, better understanding of the link between MAP and HRUHC provides insights into prioritisation of future research.
This systematic literature review and meta-analysis aims to provide a comprehensive summary of the impact of non-MAP to three medication classes, antidepressants, bisphosphonates and statins, on HRUHC as measured by hospitalisation, emergency department (ED) presentation, visit to other healthcare service providers, healthcare cost and pharmacy cost. Evaluation of the three different medication classes under one systematic literature review permits understanding of whether different medication classes used with different patterns have different impacts on HRUHC using the same evaluation criteria.
Methods
This review was conducted in accordance with a protocol (see dx.doi.org/10.17504/protocols.io.b4m4qu8w) developed using the process recommended by the Centre for Reviews and Dissemination [31] and written in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guideline [32]. The abstraction and analysis of data were conducted by the first author based on the methods developed by all authors, and reviewed by all authors.
Selection criteria
The scope of the systematic review was studies assessing the impact of non-MAP on HRUHC in antidepressants, bisphosphonates or statins. There was no restriction on the definition of MAP while HRUHC for the review included use of hospital, visit to healthcare service providers other than hospital, and healthcare cost components including medical cost and pharmacy cost. Studies were required to address the distinct impact of MAP to one of the three classes of medications on HRUHC, to be peer-reviewed, to be available as full articles, to be written in English, and to include quantitative analysis of the impact. The eligibility criteria are summarised in Table 1.
Search strategy
We used five search engines or registries: PubMed, Cochrane Library, ClinicalTrials.gov, JSTOR and EconLit. The period over which the search was conducted was 1 November 2004 to 30 April 2021. The period was set to achieve a balance between the number of studies included and focusing on a more recent period to ensure relevance of the findings. The search strategy was developed based on preliminary reviews of literature aiming at comprehensively including internationally used terminologies. Subsequent reviews of reference lists were conducted using the snowballing technique [33] to identify additional publications. The search strategy used with PubMed and Cochrane Library is displayed by item 38 in Table 2. For ClinicalTrials.gov, JSTOR and EconLit, due to the restriction on the length or form of search terms, we broadened our search specification to find studies having keywords (in their abstract for JSTOR and EconLit) showing the names of medications including statin, antidepressant, bisphosphonate or disphosphonate, and MAP or non-MAP using several words including adherence, compliance, nonadherence, noncompliance or persistence.
After removing duplicates, abstracts were screened to arrive at the eligible or possibly eligible studies for a full-text review. The full-text review was conducted to exclude articles not meeting the eligibility criteria, and to extract data.
Data extraction
Extracted information for the review includes authors, country, year of publication, study type (e.g., retrospective cohort study), medications studied, data period, characteristics of cohort, statistical method of analysis, measure of MAP, reported MAP of cohort, and summary of impact of non-MAP on HRUHC. In cases where both adjusted estimates (incorporating covariates) and unadjusted estimates (not incorporating covariates) for the relationship between MAP and HRUHC were reported, we summarise the adjusted estimates only because the inclusion of covariates facilitates meaningful interpretation and comparison. We report statistically significant or insignificant coefficients showing the relationships between MAP and HRUHC rather than absolute amounts of change in HRUHC where possible. This is to account for different baseline levels of HRUHC and different contexts (e.g., time frame, country). Other supporting information (e.g., covariates, reported conflicts of interest) was also collected to support quality assessments.
Quality criteria
Quality of study and risk of bias were evaluated using the Quality Assessment Tool for Observational Cohort and Cross-Sectional Studies [34]. This tool is used to assess the quality of observational cohort or cross-sectional studies [e.g., 35, 36] and was considered appropriate for the review because all included studies were observational cohort studies. The tool rates a study quality as good, fair or poor based on 14 questions about study objectives, sample selection, definition and use of exposure and outcome, analysis method and risk of bias.
Aggregate data meta-analysis
A meta-analysis was conducted for a certain relationship between MAP and HRUHC when reported for at least two different cohorts of population. Note that “different cohorts” here refers to both cohorts in different studies and different cohorts within a single study. For multiple results to be synthesised, the measure of MAP (e.g., adherence defined by amount of medication used relative to total amount recommended greater than 80% over six months) was required to be comparable in terms of the length of measuring period and which aspect of MAP is measured (e.g., adherence, adherence rate, persistence). In addition, medication class (e.g., antidepressants) and type of HRUHC (e.g., hospitalisation cost) were required to be the same. When the requirements were met, the combined result was estimated using the inverse-variance method which allocates each study a weight equal to the inverse of the variance of the effect estimate [37], i.e.
where Yi is the effect estimated in the ith study, SEi is the standard error of the estimate, and the summation is across all studies. When the standard error was not reported, it was approximated from the confidence interval or p-value [38]. Analysis was implemented using RevMan v5.4 [39].
Results
Study selection
As shown by the PRISMA flow diagram in Fig 1, the initial search, abstract screening and full-text review have left 30 articles for the review. The characteristics of individual studies on antidepressants, bisphosphonates and statins are summarised in Tables 3–5, respectively. Study findings of the impacts of MAP to antidepressants, bisphosphonates and statins on several different types of HRUHC are summarised in Tables 6–8, respectively. A summary of the directions of the impact of MAP on HRUHC based on the study findings is presented in Table 9.
The PRISMA diagram details the search and selection process applied during the overview.
Quality assessment
We used the Quality Assessment Tool for Observational Cohort and Cross-Sectional Studies [34] to assign each study a grade of either good, fair or poor. Of the 30 studies, 27 achieved a good rating while three studies achieved a fair rating primarily due to not reporting detailed statistical results for the total healthcare cost including pharmacy costs [40], not reporting detailed statistical results for hospitalisations and outpatient visits [41], and not reporting clear definitions of MAP and HRUHC measures [42]. The assessments are summarised in Table 10.
Descriptive results
Study characteristics.
Antidepressants. Ten studies reviewed considered the impact of MAP to antidepressants on HRUHC. These were all retrospective cohort studies using health research or claims database. Eight studies were conducted in the US. The studies were conducted with the data covering 3.6 years on average with standard deviation (SD) of 1.3 years, between 1999 and 2014. Three studies [43–45] were on all classes of antidepressants, two studies [46, 47] were on all selective serotonin reuptake inhibitors (SSRIs) medications, and other studies were on selected classes of antidepressants.
Seven studies broke down the examination period into three periods, commonly used in MAP research. The three periods are the baseline period in which patient characteristics are measured, the MAP period in which MAP is measured, and the follow-up period in which outcomes are measured. While lengths of each period are the same for each individual within a study, the date at which each period starts or ends depended on an individual date for the start of therapy, or index, which is screened in a pre-specified index period.
Cohort sizes ranged from 1,361 to 79,642 with an average of 31,659 (SD 25,382). All studies included both genders. Most studies targeted adult patients aged at least 18 years while one study was without age restriction [45], one study was for seniors aged 66 years or greater [48], and one study was for working-age patients aged between 16 and 65 [47]. Seven studies targeted patients with depression, and three other studies targeted patients with specified diseases other than depression but taking antidepressants: type-2 diabetes [44]; coronary artery disease, dyslipidaemia or diabetes [49]; and chronic obstructive pulmonary disease [45].
Bisphosphonates. Seven studies reviewed considered the impact of MAP to bisphosphonates on HRUHC: all were retrospective cohort studies using health research or claims database. Four studies were conducted in the US. The studies were conducted with the data covering 8.0 years on average (SD 1.8 years), between 1999 and 2013. Five studies broke down the examination period into baseline, MAP, and follow-up period. All studies considered alendronate and risedronate either exclusively [41, 50, 51] or as part of a wider medication range. Cohort sizes ranged between 17,770 and 38,234 with an average of 29,544 (SD 7,974). Most studies targeted female patients except two studies that considered both male and female [50, 52]. All studies set a minimum age to be included in the study cohort, the most common being 55 years [51–53]. No studies limited the cohort to those with specific underlying diseases other than osteoporosis.
Statins. There were 13 studies reviewed that considered the impact of MAP to statins on HRUHC. These studies were retrospective cohort studies using health research or claims data except for a prospective observational study [54], a longitudinal study using a survey [63], and a secondary analysis using a randomised clinical trial [55]. Ten studies were conducted in the US. The studies were conducted with data covering 4.3 years on average (SD 2.6 years) between 1997 and 2015. Four studies divided the examination period into baseline, MAP and follow-up period.
Five studies were on all classes of statins [56–60], three studies were on selected classes of statins [54, 61, 62], and five studies were on several medications including statins [42, 55, 63–65]. All studies were conducted with relatively large cohort sizes of over 500 individuals (between 682 and 381,422), except Cheng, Chan [54], which included 83 individuals. The average sample size was 44,744 (SD 101,804). Five studies [54, 56–59] targeted all adult patients and four did not specify an age range [42, 63–65], while other studies were on different age ranges, as seen in Table 5. Six studies were for patients with baseline diseases including diabetes [42, 58, 63], coronary heart disease [54], and acute myocardial infarction [55, 65]; the remaining studies were for any patients newly prescribed statins.
Medication adherence or persistence.
Antidepressants. While various methods were used to measure MAP to antidepressants, the most common measure was persistence, defined as the duration of time from initiation to discontinuation of therapy. Five studies used it as a single method [48, 66], as one of several methods [40, 44], or in combination with another method (for an overall MAP measure) [49]. Discontinuation was defined by having a treatment gap greater than 15 days [40, 44, 49], or greater than 30 days [48, 66].
Medication possession ratio (MPR), calculated by adding the days’ supply for all medications and then dividing over a set period [67], was used by four studies, either as a single measure [68], as one of several measures [40, 44] or in combination with a persistence measure [49]. Proportion of days covered (PDC), the proportion of days a patient has a drug administered in a study interval [67], was used by one study [45]. For MPR and PDC measures, the measured adherence was categorised to form an independent variable in the studies, the most common method being categorising MPR or PDC of at least 80% as adherent [40, 44, 49, 68]. Other methods used include nonadherence for the first prescription [47] and customised rules [43, 46].
Nine studies measured MAP for a fixed duration of six months or 180 days [40, 44, 48, 49, 66], of one year [46, 68], of one month [47] and of 214 days [43]. One study measured MAP during the available follow-up period for each patient [45].
Reported average values of MAP ranged between 19% and 85%. However, heterogeneity in the length of MAP period and type of MAP measure used did not allow an estimation of an aggregate average.
Bisphosphonates. For bisphosphonate studies, MPR was used to measure MAP by six studies, as a single measure [50, 52, 53], in combination with a persistence measure [51, 69], or in addition to a persistence measure [41].
Of those, four studies defined adherence as MPR at least either 70% [51–53] or 80% [41]. One study categorised MPR into several groups by threshold [50] and one study did not form categories but reported a proportion of patients who achieved at least 80% of MPR [69]. The studies using persistence measures defined discontinuation as a treatment gap greater than 30 days [41], 60 days [51], or three months [69]. One study that did not use MPR categorised longitudinal quarterly MAP into four categories: non-switching, switching, discontinuing, and reinitiating [70].
Three studies measured MAP for a one-year period [51–53]. Other studies measured MAP during the available period for each patient. The wide range of reported average values of MAP between 20% and 85% could not be summarised further because of the heterogeneity in the length of MAP period and type of MAP measure used.
Statins. Seven statin studies used MPR as a single measure of MAP by defining MAP as an MPR of at least 80% [56–58, 62], by having multiple categories of MAP using the MPR [60, 61], or by using the MPR as a numerical variable [63]. Three studies used PDC by using the threshold at 80% to define MAP for multiple medications [64], as a numerical variable [65], and as a numerical variable along with GlowCap adherence measure, the number of days the electronic pill bottle was opened divided by the total number of days followed [55]. Other methods include assessment based on percentage of doses taken within the suggested time interval and without time constraint [54], statin supply for at least 90 days in the year prior to hospitalisation [59], and annual number of prescription fills [42].
Ten studies measured MAP for fixed duration of one year [55, 58, 60, 61, 65], of six months [54, 59, 64], and of 18 months to two years [56, 57]. Other studies measured MAP for the available period for each patient. The wide range of reported average values of MAP between 17% and 96% could not be summarised further because of the heterogeneity in the length of MAP period and type of MAP measure used.
Impact of MAP on HRUHC.
Antidepressants. Among the ten studies on antidepressants, six reported significantly increased HRUHC by non-MAP, including total healthcare cost [44, 66], medical cost excluding pharmacy cost [40, 44, 49], hospitalisation cost [49], outpatient cost [49], cost of non-psychiatric hospitalisation [48], psychiatric and other specialty visits [48] and general practice (GP) visits [47].
Three studies reported reduced pharmacy costs [44, 48, 68]. Some studies also found reduced total and mental health specific healthcare costs [43], reduced cost of GP services [48], mixed results for total healthcare cost and medical cost excluding pharmacy cost [46], and mixed results for medical cost excluding pharmacy cost [68] over several categories of MAP. Insignificant impacts were reported on total healthcare cost [68], mental health specific hospitalisation cost [48], hospitalisations [45], ED visits [45] and specialty visits [47]. In addition, insignificant impacts of MAP on hospitalisation cost, ED cost, outpatient cost, GP service cost, and antidepressant and other pharmacy costs were reported [46].
Bisphosphonates. All seven studies on bisphosphonates reported significantly increased utilisation or cost of at least one type of health resource following non-MAP, including total healthcare cost [41, 50, 51, 53, 70], osteoporosis-related healthcare cost [52, 53], all-cause combined and osteoporosis-related medical cost excluding pharmacy cost [53], hospitalisation cost [50], outpatient cost [50], outpatient visits and use of ED services [53], hospitalisations [41, 53], osteoporosis-related hospitalisations and outpatient services [53], and combined HRUHC [69].
Four studies reported reduced utilisation or cost of several types of resource, including pharmacy cost [50, 53], cost of bisphosphonates [53, 70], osteoporosis-related healthcare cost [70], and outpatient visits [41].
Statins. Nine of 13 statin studies reported that at least one type of HRUHC significantly increased following non-MAP. The increased HRUHCs include total healthcare cost [57, 61, 63], CVD-related healthcare cost [58], medical cost excluding pharmacy cost [42, 57, 58, 61], CVD-related medical cost excluding pharmacy cost [61], pharmacy cost other than statins [61], hospitalisation cost [62], ED visits [58, 60], hospitalisations [42, 57–59, 62], hospital days [42], and CVD hospitalisations [56, 61]. In contrast, Zhao, Zabriski [60] found decreased total healthcare costs for nonadherent patients. Pittman, Chen [61] found lower statin prescription cost for nonadherent patients.
Several studies found insignificant impacts of MAP to statins on total healthcare cost [54, 56, 65], hospitalisation cost [65], outpatient cost [65], remedy and aid cost [65], hospitalisations [60, 64], CVD hospitalisations [62], and ED visits [62, 64]. Gibson, Mark [56] divided patients into two groups–new users and continuing users–and found mixed results for pharmacy cost, GP visits, ED visits, and hospitalisations. Kirsch, Becker [65] found non-linear impact on pharmacy cost only for female patients and positive impact on rehabilitation cost only for male patients. Mehta, Asch [55] used multiple measures of MAP and found mixed results on hospitalisation.
Aggregate data meta-analysis.
We conducted a meta-analysis to estimate the impact of MAP on HRUHC when reported for at least two different population cohorts using comparable measures of MAP to the same medication class and the same type of HRUHC. Of 30 studies, only eight were used in the meta-analysis to obtain five synthesised results. Further findings were not possible due to the heterogenous types of MAP and HRUHC examined by the reviewed studies.
Table 11 summarises the averaged impacts of MAP on HRUHC from the meta-analysis. Forest plots and more detailed figures are found in S2 File. For antidepressants, having greater than 80% adherence by either medication possession ratio (MPR) or proportion of days covered (PDC) during a six-month or 180-day period was found to reduce the total medical cost not including pharmacy cost by 12% [-16%, -8%], the hospitalisation cost by 26% [-48%, 4%] and outpatient cost by 10% [-17%, -2%]. The impact of persistence in using antidepressants for a 180-day period without 15-day gap on total healthcare cost was found insignificant. For bisphosphonates, having greater than 70% or 80% adherence by one-year MPR was found to reduce the total healthcare cost by 3% [-6%, 0%]. For statins, no meta-analysis could be performed mainly due to inconsistency in the measures of MAP within studies.
Discussion
This review provides a comprehensive view on the impact of MAP to antidepressants, bisphosphonates and statins on HRUHC. It is the first to provide an integrated understanding of the impact of MAP to three medication classes on HRUHC using a broad array of HRUHC measures including total healthcare cost, disease-specific healthcare cost, medical cost excluding pharmacy cost, pharmacy cost, cost of medication, hospitalisation, outpatient service use, ED visits, GP visits and specialty visits. Previous reviews on the impact of MAP to these three medication classes on HRUHC did not address diverse measures of MAP or HRUHC, or did not make comparisons with other medication classes used for prevalent conditions [e.g., 6, 71, 72].
MAP varies by medication class, MAP measure, length of follow-up period and study design, limiting scope for comparison. Nevertheless, the review found broad consistency in reported MAP to antidepressants, with the percentage of patients with MPR or PDC at least 80% during a 180 day or six month period ranging between 36% and 43% [40, 44, 49, 68] and percentages of persistent patients as measured by continued use in 180 days without a 30-day gap of 19% [66] and 35% to 47% [48]. These findings are comparable with a previous review (on studies not limited to those on the impact of MAP on HRUHC) indicating that 35% to 55% of patients remain adherent or persistent to antidepressant therapy at six months [73].
The proportion of patients adherent to bisphosphonates were between 20% and 70% in the studies that measured MAP as a MPR or PDC of at least 70% or 80% during a one year period [50–53]. The mean persistence of oral bisphosphonates for one year similarly ranged between 18% and 75% in a systematic review by Fatoye, Smith [74]. The wide range of MAP could be partly due to variance between studies in factors related to non-MAP including younger age [74] and more frequent dosing [74, 75]. Of the four studies compared above [50–53], the lower bound of the range (i.e., 20%) was found from Briesacher, Andrade [50] on patients aged at least 40, compared with the other three studies which were all on patients aged at least 55.
Among statin users, 17% to 68% were found to have a MPR at least 80% during a one year period [58, 60, 61]. These findings are comparable with a previous review showing that patients with a MPR for statins at least 80% ranged between 18% and 92% for different lengths of MAP period [76]. Consistent with previous studies that found lower MAP for new statin users compared to continuous users [56, 76], Wu, Seiber [58] and Zhao, Zabriski [60] reported a lower percentage of adherence for new statin users than Pittman, Chen [61] for continuous users.
The MPR and PDC were the most frequently used methods to summarise MAP. While they both measure the percentage of the time that a patient has medication available, the PDC was introduced to mitigate the overestimation problem of the MPR in which early refill (i.e. refill when the medication is still available) is included in the amount of medication available in the measuring period [77]. We found the tendency that the PDC measure was used for studies published later.
Several studies [40, 41, 44, 45, 54, 55] reported results that allow comparison among multiple methods of measuring MAP. For example, Mehta, Asch [55] compared MAP as measured by PDC based on pharmacy claims as well as GlowCap adherence using electronic pill bottles. They found that the significantly lower risk of all-cause readmission of patients previously discharged with a diagnosis of acute myocardial infarction was found only when the GlowCap adherence was used, highlighting the importance of measurement method for MAP.
In general, non-MAP was found to be associated with increased HRUHC. Of the 30 papers included, 25 found a significant positive association between non-MAP and one or more measures of HRUHC, although in some cases negative or mixed associations were found for other HRUHC measures. This generally positive association between non-MAP and HRUHC has also been found in previous reviews [e.g. 6, 78].
This review revealed that the association between non-MAP and increased healthcare costs is most definitive for medical costs excluding pharmacy costs. Of 10 papers that assessed total healthcare costs net of pharmacy costs, eight reported a positive association with non-MAP, the remaining two reporting mixed results. Non-MAP was found to be negatively associated with total pharmacy costs for five of the seven papers that assessed this, reflecting higher pharmacy costs for patients that are adhering to, and therefore consuming more of, their medication. Of the 18 papers that assessed total healthcare costs, 10 reported a positive association between total healthcare costs and non-MAP, and a further six found no significant association or mixed results across multiple MAP categories. This result is a combination of generally lower pharmacy costs and generally higher medical costs excluding pharmacy costs for non-MAP patients. These findings related to healthcare and pharmacy costs apply across all three medications considered.
Increase in hospitalisations and ED visits associated with non-MAP was frequently reported. Of 12 studies that measured the impact of non-MAP on hospitalisation, seven found a positive impact while the rest found either mixed or insignificant results. Of seven studies that assessed the impact of non-MAP on ED visits, three found a positive impact while four found insignificant results or mixed results across multiple patient categories. Several HRUHC measures were reported by too few papers to enable comparison; for example, the impact on specialty visits was reported by only one study [47].
While general patterns of resource use were similar across the three medication classes, some differences exist. First, the pattern of increased healthcare cost following non-MAP was the most apparent in bisphosphonate studies. All seven studies on bisphosphonates found an increase in at least one type of HRUHC. Of six studies that assessed the total healthcare cost, five found a positive association with non-MAP, while one found no significant association. The clearer pattern found could be related to the tendency that studies on bisphosphonates used longer data periods (average 8.0 years compared to average 3.6 and 4.3 years for the studies on antidepressants and statins, respectively) and less heterogenous cohorts (typically females aged at least 55 compared to all adult patients in the studies on antidepressants and statins).
Second, the impacts of non-MAP on hospitalisation and ED visits were more frequently studied on statins and were found to be generally positively associated. Third, the associations between non-MAP and total pharmacy costs were negative for all studies on antidepressants and bisphosphonates that assessed this, but were heterogenous for the studies on statins; negative for continuing users [56], non-linear for female patients [65] and insignificant for others. This shows that non-MAP to statins does not necessarily imply non-MAP to other medications. Future studies to examine such selective non-MAPs will be useful to further understand the reasons for non-MAP to statins.
Last, in the studies on antidepressants and statins compared to bisphosphonates, notwithstanding the general finding of positive associations, there were greater numbers of reported insignificant impacts of non-MAP on HRUHC. The impacts of non-MAP may not be sufficiently captured with a short follow-up period given that nine [45–48, 54, 60, 64, 65, 68] of 11 studies that reported insignificant associations measured MAP and HRUHC for the same (i.e., overlapping) six-month to one-year period. There could also be idiosyncratic healthcare system factors. Li and Huang [62] found an insignificant impact on ED visits and discussed that the finding may not accurately show the impact on the occurrence of emergency situations that are potentially costly due to frequent non-emergency use of emergency care services in Taiwan.
Comparing the directions of impact on different types of HRUHC within an individual study can suggest underlying mechanisms linking MAP and HRUHC. For example, Tournier, Moride [48] found that persistence to antidepressants increases GP service costs and pharmacy costs but decreases specialty visits and non-psychiatric hospitalisation costs. This shows that spending on primary care to maintain mental health can reduce the costs associated with adverse health outcomes. Another similar pattern was found by Sunyecz, Mucha [41] reporting that persistence to bisphosphonates reduces total healthcare cost and hospitalisations but increases outpatient visits.
The meta-analysis found positive associations between non-MAP and HRUHC for total medical cost excluding pharmacy cost, hospital cost and outpatient cost for antidepressants all at 5% significance level, and total healthcare cost for bisphosphonates at 10% significance. The average magnitude of the impact of non-MAP to antidepressants on hospitalisation cost was estimated at 26%.
Heterogeneities in study location, data type, cost calculation method, HRUHC measure, analysis method and other characteristics limited the extent to which further meta-analysis could be conducted and overall conclusions drawn. This meant we could not conduct a meta-analysis on statins and a meta-analysis on other HRUHC components, and it limited the number of studies included in the meta-analyses that were conducted. Study characteristics will also influence the meta-analysis results. For example, more costly healthcare in the US than the UK [79] along with the preponderance of US studies included in the meta-analysis is expected to result in higher estimates of impact of non-MAP on HRUHC which may not be generalisable to non-US locations.
Comparisons across studies having different characteristics were not possible primarily due a limited number of studies having certain characteristics different to the majority. For example, a comparison by location was limited because the review included at most one study conducted in a non-US country for each medication class; a comparison by analysis method was limited because most studies used generalized linear models; and a comparison by data type was limited because most studies used an administrative dataset with limited use of other (e.g., survey data) types.
The quality assessments in Table 10 show that all included studies meet the majority of assessment criteria related to study objectives, selection criteria for study populations, justification of sample selection and size, measurement of MAP, study timeframes and quality of analysis. While 28 of 30 studies reviewed did not assess MAP more than once, a single MAP figure comprises multiple observations on medication use over time and hence any inaccuracy that may arise from assessing an exposure once only would be small. There was no evident difference in quality across the studies on the three medication classes.
Of the 30 reviewed studies, 27 studies used a large administrative dataset allowing for a sample size greater than 1,000 and measured MAP using administrative records of filling prescriptions (e.g., pharmacy claims). This approach to measuring MAP is standard and has the advantages that MAP is passively measured and easy to track for large populations [80] and does not influence participant behaviour. However, limitations include that filled medications are not necessarily taken; a diagnosis for which medications are prescribed is mostly unavailable; the reason for non-MAP is not known; a case of discontinuation recommended by health service provider is not identified (e.g., side effects); and data does not capture all medications used by a patient (e.g., data extracted from an insurance plan does not capture medications funded from other means). In addition, the measure of MAP can be sensitive to modelling decisions in preparing pharmacy data sets [81], and most reviewed studies did not provide details of these decisions.
Not measuring MAP prior to HRUHC was a common problem, most prominent in the studies on antidepressants. Although 22 studies set separate measurement periods for MAP and HRUHC, only eight clearly showed that MAP was measured strictly prior to HRUHC. As MAP could be affected directly by HRUHC (e.g., GP visits to get prescriptions) or indirectly by health conditions suggested by HRUHC (e.g., reassuring the need for medication at ED visit), measuring MAP prior to HRUHC would avoid potential reverse causality problems. The findings within these eight studies were generally consistent with the overall findings of this review.
Limited generalisability was found for several studies. For example, results within Gibson, Mark [56] may be only applicable for an insured population as it was based on patients covered by employer-sponsored health insurance. The patient population of Ferguson, Feudjo Tepie [69] is atypical in that a significant proportion had a high level of glucocorticoid use.
Many studies did not clearly specify the payer and recipient for the measured healthcare costs although such specification will be useful when study findings are used for developing health policy. Only a few studies specifically stated from whose perspective the costs are measured. For example, Cheng, Chan [54] reported that the costs were calculated for each patient from the perspective of a public health provider. Considering the type of data used by the majority (i.e., administrative claim data), most studies that assessed healthcare costs are likely to have measured the cost paid by an insurance company or the public healthcare system to healthcare service providers.
There are several limitations of this systematic review. First, only a limited meta-analysis could be undertaken due to heterogeneous types of MAP and HRUHC within the studies. There is a need for agreement on consistent methods to be applied to measure MAP and more studies on each type of HRUHC to improve comparability. Second, the review does not attempt to directly evaluate individual patient factors that may influence the impact of MAP on HRUHC (e.g., age, sex, comorbidity, severity of disease): such factors were highly heterogenous across different studies and could not be meaningfully incorporated. However, 28 of 30 studies did measure these and other potentially confounding variables and made statistical adjustments accordingly in measuring the impact of MAP on HRUHC, enabling valid comparison without a need for direct evaluation of these factors in this review.
Third, 22 of the studies were conducted in the US and therefore results may fail to reflect experience in other countries; this is due partly to the exclusion of non-English articles as well as to the preponderance of US studies. Several studies have found that exclusion of non-English articles is unlikely to result in bias [82, 83]. Among English articles, the strict search protocol limits bias in study selection and ensures that the US dominance is due to dominance of research. Additional research conducted outside the US will permit greater understanding of the impact of MAP on HRUHC dependent on healthcare systems.
Last, the review focuses only on HRUHC from a healthcare system perspective and does not address other non-healthcare burdens such as loss of productivity, absence from work, loss of quality of life and costs of home care or informal care. Several previous studies examined such burdens following non-MAP [84–87]. These aspects are outside the scope of our review however it should be acknowledged that total economic impact of non-MAP will be greater than that indicated by HRUHC within this review.
Conclusions
This systematic literature review is the first to compare the impact of non-MAP to medications for three prevalent conditions—depression, osteoporosis and cardiovascular disease—on healthcare resource utilisation and cost. While previous reviews generally focused on finding the impact of MAP on particular healthcare costs or clinical outcomes, this review considered a wide range of measures and three different medication classes. From 30 included studies assessed to be of good or fair quality, we found generally positive associations between non-MAP and healthcare resource utilisation and cost for all three medication classes but most prominently for bisphosphonates. Notwithstanding this general finding, the significance and direction of associations was heterogenous across alternative HRUHC measures and medication classes. In some cases, non-MAP reduced healthcare resource utilisation or cost, particularly for pharmacy. The ability to quantitatively summarise the impact of non-MAP on healthcare resource utilisation and cost was challenged by a small number of studies reporting comparable results; the development of more consistent measures would enable more meaningful analysis. The study highlights the need to understand how and to what extent poor MAP exhausts healthcare resources to inform clinical practice, health policy and research.
Supporting information
S1 Fig. Forest plots for aggregate meta-analysis.
https://doi.org/10.1371/journal.pone.0269836.s002
(DOCX)
References
- 1. DiMatteo MR, Giordani PJ, Lepper HS, Croghan TW. Patient adherence and medical treatment outcomes: a meta-analysis. Medical care. 2002;40(9):794–811. Epub 2002/09/10. pmid:12218770.
- 2. Blackburn DF, Dobson RT, Blackburn JL, Wilson TW. Cardiovascular morbidity associated with nonadherence to statin therapy. Pharmacotherapy. 2005;25(8):1035–43. Epub 2005/10/07. pmid:16207093.
- 3. Cramer JA, Benedict A, Muszbek N, Keskinaslan A, Khan ZM. The significance of compliance and persistence in the treatment of diabetes, hypertension and dyslipidaemia: a review. Int J Clin Pract. 2008;62(1):76–87. Epub 2007/11/05. pmid:17983433.
- 4. Korhonen MJ, Ruokoniemi P, Ilomäki J, Meretoja A, Helin-Salmivaara A, Huupponen R. Adherence to statin therapy and the incidence of ischemic stroke in patients with diabetes. Pharmacoepidemiology and drug safety. 2016;25(2):161–9. Epub 2015/12/22. pmid:26687512.
- 5. Ling Y-L. Adherence to antidepressants and healthcare resource utilization and costs among medicare advantage beneficiaries with Parkinson’s disease and depression 2016.
- 6. Cutler RL, Fernandez-Llimos F, Frommer M, Benrimoj C, Garcia-Cardenas V. Economic impact of medication non-adherence by disease groups: a systematic review. BMJ open. 2018;8(1):e016982. Epub 2018/01/24. pmid:29358417; PubMed Central PMCID: PMC5780689.
- 7. Walsh CA, Cahir C, Tecklenborg S, Byrne C, Culbertson MA, Bennett KE. The association between medication non‐adherence and adverse health outcomes in ageing populations: A systematic review and meta‐analysis. British journal of clinical pharmacology. 2019;85(11):2464–78.
- 8. Wilhelmsen NC, Eriksson T. Medication adherence interventions and outcomes: an overview of systematic reviews. European Journal of Hospital Pharmacy. 2019;26(4):187–92. pmid:31338165
- 9. Bassett SM, Schuette SA, O’Dwyer LC, Moskowitz JT. Positive affect and medication adherence in chronic conditions: A systematic review. Health Psychology. 2019;38(11):960. pmid:31368717
- 10.
Sabaté E. Adherence to long-term therapies: evidence for action: World Health Organization; 2003.
- 11. Osterberg L, Blaschke T. Adherence to medication. The New England journal of medicine. 2005;353(5):487–97. Epub 2005/08/05. pmid:16079372.
- 12. Brown MT, Bussell JK. Medication adherence: WHO cares? Mayo Clinic proceedings. 2011;86(4):304–14. Epub 2011/03/11. pmid:21389250; PubMed Central PMCID: PMC3068890.
- 13.
New England Healthcare Institute. Thinking outside the pillbox: a system-wide approach to improving patient medication adherence for chronic disease: New England Health Care Institute; 2009.
- 14. Peterson AM, Takiya L, Finley R. Meta-analysis of trials of interventions to improve medication adherence. American Journal of Health-System Pharmacy. 2003;60(7):657–65. pmid:12701547
- 15. Kessler RC, Chiu WT, Demler O, Walters EE. Prevalence, Severity, and Comorbidity of 12-Month DSM-IV Disorders in the National Comorbidity Survey Replication. Archives of General Psychiatry. 2005;62(6):617–27. pmid:15939839
- 16. Hadji P, Papaioannou N, Gielen E, Tepie MF, Zhang E, Frieling I, et al. Persistence, adherence, and medication-taking behavior in women with postmenopausal osteoporosis receiving denosumab in routine practice in Germany, Austria, Greece, and Belgium: 12-month results from a European non-interventional study. Osteoporosis International. 2015;26(10):2479–89. pmid:26018090
- 17. König H, König HH, Konnopka A. The excess costs of depression: a systematic review and meta-analysis. Epidemiology and psychiatric sciences. 2019;29:e30. Epub 2019/04/06. pmid:30947759.
- 18. Wells KB, Stewart A, Hays RD, Burnam MA, Rogers W, Daniels M, et al. The Functioning and Well-being of Depressed Patients: Results From the Medical Outcomes Study. Jama. 1989;262(7):914–9. pmid:2754791
- 19. Budhia S, Mikyas Y, Tang M, Badamgarav E. Osteoporotic fractures: a systematic review of U.S. healthcare costs and resource utilization. PharmacoEconomics. 2012;30(2):147–70. Epub 2011/12/23. pmid:22187933.
- 20. Greenberg PE, Fournier A-A, Sisitsky T, Pike CT, Kessler RC. The economic burden of adults with major depressive disorder in the United States (2005 and 2010). The Journal of clinical psychiatry. 2015;76(2):155–62. pmid:25742202
- 21.
Australian Institute of Health and Welfare. Australian Burden of Disease Study: impact and causes of illness and death in Australia 2015. Canberra: AIHW, 2019.
- 22.
Gagel R. Preventing Costly Bone Fractures—Why Don’t We Do What Works?: AJMC.com; 2019 [cited 2021 25 May]. Available from: https://www.ajmc.com/view/for-national-women-s-health-week-a-look-ahead-at-the-coming-pandemic-induced-challenges.
- 23. Benjamin EJ, Virani SS, Callaway CW, Chamberlain AM, Chang AR, Cheng S, et al. Heart Disease and Stroke Statistics-2018 Update: A Report From the American Heart Association. Circulation. 2018;137(12):e67–e492. Epub 2018/02/02. pmid:29386200.
- 24. Schulberg HC, Katon W, Simon GE, Rush AJ. Treating major depression in primary care practice: an update of the Agency for Health Care Policy and Research Practice Guidelines. Archives of general psychiatry. 1998;55(12):1121–7. pmid:9862556
- 25. Corrao G, Conti V, Merlino L, Catapano AL, Mancia G. Results of a retrospective database analysis of adherence to statin therapy and risk of nonfatal ischemic heart disease in daily clinical practice in Italy. Clinical therapeutics. 2010;32(2):300–10. pmid:20206788
- 26. Jackevicius CA, Mamdani M, Tu JV. Adherence With Statin Therapy in Elderly Patients With and Without Acute Coronary Syndromes. Jama. 2002;288(4):462–7. pmid:12132976
- 27. Hunot VM, Horne R, Leese MN, Churchill RC. A cohort study of adherence to antidepressants in primary care: the influence of antidepressant concerns and treatment preferences. Primary care companion to the Journal of clinical psychiatry. 2007;9(2):91. pmid:17607330
- 28. Gelenberg A, Freeman M, Markowitz J, Rosenbaum J, Thase M, Trivedi M, et al. American Psychiatric Association practice guideline for the treatment of patients with major depressive disorder. Am J Psychiatry. 2010;167(suppl).
- 29. Compston J, Cooper A, Cooper C, Gittoes N, Gregson C, Harvey N, et al. UK clinical guideline for the prevention and treatment of osteoporosis. Archives of osteoporosis. 2017;12(1):43.
- 30. Jasińska-Stroschein M, Owczarek J, Wejman I, Orszulak-Michalak D. Novel mechanistic and clinical implications concerning the safety of statin discontinuation. Pharmacological reports: PR. 2011;63(4):867–79. Epub 2011/10/18. pmid:22001974.
- 31.
Centre for Reviews and Dissemination. Systematic Reviews: CRD’s Guidance for Undertaking Reviews in Health Care. University of York, Centre for Reviews and Dissemination; 2009.
- 32. Moher D, Liberati A, Tetzlaff J, Altman DG. Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. Annals of internal medicine. 2009;151(4):264–9. pmid:19622511
- 33. Sayers A. Tips and tricks in performing a systematic review. Br J Gen Pract. 2007;57(542):759–. PMC2151802.
- 34.
National Heart L, Institute B. Quality assessment tool for observational cohort and cross-sectional studies. Bethesda: National Institutes of Health, Department of Health and Human Services. 2014.
- 35. Buchholz I, Janssen MF, Kohlmann T, Feng Y-S. A Systematic Review of Studies Comparing the Measurement Properties of the Three-Level and Five-Level Versions of the EQ-5D. PharmacoEconomics. 2018;36(6):645–61. pmid:29572719
- 36. van Koppen B, Zandwijk P, de Vries J, van Mameren H, de Bie R. Adherence to home-based exercises and/or activity advice in low back pain patients: a systematic review. European Journal of Physiotherapy. 2020:1–16.
- 37.
Higgins JP, Thomas J, Chandler J, Cumpston M, Li T, Page MJ, et al. A generic inverse-variance approach to meta-analysis. Cochrane handbook for systematic reviews of interventions: John Wiley & Sons; 2019. p. 245.
- 38. Higgins J, Green S. Cochrane Handbook for systematic reviews of interventions version 5.1.0 [updated march 2011]. The Cochrane Collaboration. 2011.
- 39.
The Cochrane Collaboration. Review Manager (RevMan). 5.4 ed2020.
- 40. Cantrell CR, Eaddy MT, Shah MB, Regan TS, Sokol MC. Methods for evaluating patient adherence to antidepressant therapy: a real-world comparison of adherence and economic outcomes. Medical care. 2006;44(4):300–3. Epub 2006/03/28. pmid:16565629.
- 41. Sunyecz JA, Mucha L, Baser O, Barr CE, Amonkar MM. Impact of compliance and persistence with bisphosphonate therapy on health care costs and utilization. Osteoporosis international: a journal established as result of cooperation between the European Foundation for Osteoporosis and the National Osteoporosis Foundation of the USA. 2008;19(10):1421–9. Epub 2008/03/21. pmid:18351427.
- 42. Stuart BC, Simoni-Wastila L, Zhao L, Lloyd JT, Doshi JA. Increased persistency in medication use by U.S. Medicare beneficiaries with diabetes is associated with lower hospitalization rates and cost savings. Diabetes care. 2009;32(4):647–9. pmid:19171724.
- 43. Robinson RL, Long SR, Chang S, Able S, Baser O, Obenchain RL, et al. Higher costs and therapeutic factors associated with adherence to NCQA HEDIS antidepressant medication management measures: analysis of administrative claims. Journal of managed care pharmacy: JMCP. 2006;12(1):43–54. Epub 2006/01/20. pmid:16420107.
- 44. Vega C, Becker RV, Mucha L, Lorenz BH, Eaddy MT, Ogbonnaya AO. Impact of adherence to antidepressants on healthcare outcomes and costs among patients with type 2 diabetes and comorbid major depressive disorder. Curr Med Res Opin. 2017;33(10):1879–89. Epub 2017/06/24. pmid:28644095.
- 45. Albrecht JS, Khokhar B, Huang TY, Wei YJ, Harris I, Moyo P, et al. Adherence and healthcare utilization among older adults with COPD and depression. Respiratory medicine. 2017;129:53–8. Epub 2017/07/25. pmid:28732836; PubMed Central PMCID: PMC5607946.
- 46. Eaddy MT, Druss BG, Sarnes MW, Regan TS, Frankum LE. Relationship of total health care charges to selective serotonin reuptake inhibitor utilization patterns including the length of antidepressant therapy—results from a managed care administrative claims database. Journal of managed care pharmacy: JMCP. 2005;11(2):145–50. Epub 2005/03/16. pmid:15766321.
- 47. Aznar-Lou I, Iglesias-Gonzalez M, Gil-Girbau M, Serrano-Blanco A, Fernandez A, Penarrubia-Maria MT, et al. Impact of initial medication non-adherence to SSRIs on medical visits and sick leaves. J Affect Disord. 2018;226:282–6. Epub 2017/10/13. pmid:29024901.
- 48. Tournier M, Moride Y, Crott R, du Fort GG, Ducruet T. Economic impact of non-persistence to antidepressant therapy in the Quebec community-dwelling elderly population. J Affect Disord. 2009;115(1–2):160–6. Epub 2008/08/13. pmid:18694602.
- 49. Katon W, Cantrell CR, Sokol MC, Chiao E, Gdovin JM. Impact of antidepressant drug adherence on comorbid medication use and resource utilization. Archives of internal medicine. 2005;165(21):2497–503. Epub 2005/11/30. pmid:16314547.
- 50. Briesacher BA, Andrade SE, Yood RA, Kahler KH. Consequences of poor compliance with bisphosphonates. Bone. 2007;41(5):882–7. Epub 2007/08/21. pmid:17707710.
- 51. Sharman Moser S, Yu J, Goldshtein I, Ish-Shalom S, Rouach V, Shalev V, et al. Cost and Consequences of Nonadherence With Oral Bisphosphonate Therapy: Findings From a Real-World Data Analysis. The Annals of pharmacotherapy. 2016;50(4):262–9. Epub 2016/01/20. pmid:26783359.
- 52. Eisenberg DF, Placzek H, Gu T, Krishna A, Tulsi BB. Cost and consequences of noncompliance to oral bisphosphonate treatment. Journal of managed care & specialty pharmacy. 2015;21(1):56–65. Epub 2015/01/07. pmid:25562773.
- 53. Kjellberg J, Jorgensen AD, Vestergaard P, Ibsen R, Gerstoft F, Modi A. Cost and health care resource use associated with noncompliance with oral bisphosphonate therapy: an analysis using Danish health registries. Osteoporosis international: a journal established as result of cooperation between the European Foundation for Osteoporosis and the National Osteoporosis Foundation of the USA. 2016;27(12):3535–41. Epub 2016/07/11. pmid:27394414.
- 54. Cheng CW, Chan JC, Tomlinson B, Woo KS, You JH. Factors associated with healthcare utilization costs for statin therapy—a pilot study in Hong Kong. International journal of clinical pharmacology and therapeutics. 2006;44(10):484–8. Epub 2006/10/27. pmid:17063979.
- 55. Mehta SJ, Asch DA, Troxel AB, Lim R, Lewey J, Wang W, et al. Comparison of Pharmacy Claims and Electronic Pill Bottles for Measurement of Medication Adherence Among Myocardial Infarction Patients. Medical care. 2019;57(2):e9–e14. Epub 2018/07/26. pmid:30045159.
- 56. Gibson TB, Mark TL, Axelsen K, Baser O, Rublee DA, McGuigan KA. Impact of statin copayments on adherence and medical care utilization and expenditures. The American journal of managed care. 2006;12 Spec no.:Sp11-9. Epub 2006/12/19. pmid:17173486.
- 57. Aubert RE, Yao J, Xia F, Garavaglia SB. Is there a relationship between early statin compliance and a reduction in healthcare utilization? The American journal of managed care. 2010;16(6):459–66. Epub 2010/06/22. pmid:20560689.
- 58. Wu J, Seiber E, Lacombe VA, Nahata MC, Balkrishnan R. Medical utilization and costs associated with statin adherence in Medicaid enrollees with type 2 diabetes. The Annals of pharmacotherapy. 2011;45(3):342–9. Epub 2011/02/18. pmid:21325098.
- 59. Chen JY, Ma Q, Chen H, Yermilov I. New bundled world: quality of care and readmission in diabetes patients. Journal of diabetes science and technology. 2012;6(3):563–71. Epub 2012/07/10. pmid:22768887; PubMed Central PMCID: PMC3440042.
- 60. Zhao Y, Zabriski S, Bertram C. Associations between statin adherence level, health care costs, and utilization. Journal of managed care & specialty pharmacy. 2014;20(7):703–13. Epub 2014/06/27. pmid:24967523.
- 61. Pittman DG, Chen W, Bowlin SJ, Foody JM. Adherence to statins, subsequent healthcare costs, and cardiovascular hospitalizations. The American journal of cardiology. 2011;107(11):1662–6. Epub 2011/03/29. pmid:21439533.
- 62. Li YC, Huang WL. Effects of Adherence to Statin Therapy on Health Care Outcomes and Utilizations in Taiwan: A Population-Based Study. Biomed Res Int. 2015;2015:149573. Epub 2015/11/06. pmid:26539464; PubMed Central PMCID: PMC4619755.
- 63. Stuart B, Davidoff A, Lopert R, Shaffer T, Samantha Shoemaker J, Lloyd J. Does medication adherence lower Medicare spending among beneficiaries with diabetes? Health services research. 2011;46(4):1180–99. Epub 2011/03/19. pmid:21413981; PubMed Central PMCID: PMC3130847.
- 64. Roberts AW, Crisp GD, Esserman DA, Roth MT, Weinberger M, Farley JF. Patterns of medication adherence and health care utilization among patients with chronic disease who were enrolled in a pharmacy assistance program. North Carolina medical journal. 2014;75(5):310–8. Epub 2014/09/23. pmid:25237868.
- 65. Kirsch F, Becker C, Kurz C, Schwettmann L, Schramm A. Effects of adherence to pharmacological secondary prevention after acute myocardial infarction on health care costs—an analysis of real-world data. BMC health services research. 2020;20(1):1145. Epub 2020/12/22. pmid:33342431; PubMed Central PMCID: PMC7751107.
- 66. Ereshefsky L, Saragoussi D, Despiegel N, Hansen K, Francois C, Maman K. The 6-month persistence on SSRIs and associated economic burden. Journal of medical economics. 2010;13(3):527–36. Epub 2010/08/13. pmid:20701432.
- 67. Martin BC, Wiley-Exley EK, Richards S, Domino ME, Carey TS, Sleath BL. Contrasting Measures of Adherence with Simple Drug Use, Medication Switching, and Therapeutic Duplication. Annals of Pharmacotherapy. 2009;43(1):36–44. pmid:19126828
- 68. Stein MB, Cantrell CR, Sokol MC, Eaddy MT, Shah MB. Antidepressant adherence and medical resource use among managed care patients with anxiety disorders. Psychiatric services (Washington, DC). 2006;57(5):673–80. Epub 2006/05/06. pmid:16675762.
- 69. Ferguson S, Feudjo Tepie M, Taylor A, Roddam A, Critchlow C, Iqbal M, et al. The impact of persistence with bisphosphonates on health resource utilization and fracture risk in the UK: a study of patient records from the UK Clinical Practice Research Datalink. Journal of evaluation in clinical practice. 2016;22(1):31–9. Epub 2015/08/19. pmid:26279458; PubMed Central PMCID: PMC5042106.
- 70. LaFleur J, DuVall SL, Willson T, Ginter T, Patterson O, Cheng Y, et al. Analysis of osteoporosis treatment patterns with bisphosphonates and outcomes among postmenopausal veterans. Bone. 2015;78:174–85. Epub 2015/04/22. pmid:25896952.
- 71. Ho SC, Chong HY, Chaiyakunapruk N, Tangiisuran B, Jacob SA. Clinical and economic impact of non-adherence to antidepressants in major depressive disorder: A systematic review. J Affect Disord. 2016;193:1–10. Epub 2016/01/11. pmid:26748881.
- 72. Nguyen E, Weeda ER, Sobieraj DM, Bookhart BK, Piech CT, Coleman CI. Impact of non-medical switching on clinical and economic outcomes, resource utilization and medication-taking behavior: a systematic literature review. Current Medical Research and Opinion. 2016;32(7):1281–90. pmid:27033747
- 73. Ta JT, Sullivan SD, Tung A, Oliveri D, Gillard P, Devine B. Health care resource utilization and costs associated with nonadherence and nonpersistence to antidepressants in major depressive disorder. Journal of managed care & specialty pharmacy. 2021;27(2):223–39. pmid:33506730
- 74. Fatoye F, Smith P, Gebrye T, Yeowell G. Real-world persistence and adherence with oral bisphosphonates for osteoporosis: a systematic review. BMJ open. 2019;9(4):e027049. pmid:30987990
- 75. Cramer JA, Gold DT, Silverman SL, Lewiecki EM. A systematic review of persistence and compliance with bisphosphonates for osteoporosis. Osteoporosis international: a journal established as result of cooperation between the European Foundation for Osteoporosis and the National Osteoporosis Foundation of the USA. 2007;18(8):1023–31. Epub 2007/02/20. pmid:17308956.
- 76. Deshpande S, Quek RGW, Forbes CA, de Kock S, Kleijnen J, Gandra SR, et al. A systematic review to assess adherence and persistence with statins. Current Medical Research and Opinion. 2017;33(4):769–78. pmid:28076703
- 77.
Nau DP. Proportion of days covered (PDC) as a preferred method of measuring medication adherence. Springfield, VA: Pharmacy Quality Alliance. 2012.
- 78. Bitton A, Choudhry NK, Matlin OS, Swanton K, Shrank WH. The impact of medication adherence on coronary artery disease costs and outcomes: a systematic review. The American journal of medicine. 2013;126(4):357. e7-. e27. pmid:23507208
- 79. Hargraves J, Bloschichak A. International comparisons of health care prices from the 2017 iFHP survey 2019 [cited 2021 13 May]. Available from: https://healthcostinstitute.org/hcci-research/international-comparisons-of-health-care-prices-2017-ifhp-survey.
- 80. Sattler EL, Lee JS, Perri M 3rd. Medication (re)fill adherence measures derived from pharmacy claims data in older Americans: a review of the literature. Drugs & aging. 2013;30(6):383–99. Epub 2013/04/05. pmid:23553512.
- 81. Bjarnadottir MV, Czerwinski D, Onukwugha E. Sensitivity of the Medication Possession Ratio to Modelling Decisions in Large Claims Databases. PharmacoEconomics. 2018;36(3):369–80. Epub 2017/12/13. pmid:29230712.
- 82. Morrison A, Polisena J, Husereau D, Moulton K, Clark M, Fiander M, et al. The effect of English-language restriction on systematic review-based meta-analyses: a systematic review of empirical studies. International journal of technology assessment in health care. 2012;28(2):138–44. Epub 2012/05/09. pmid:22559755.
- 83. Moher D, Pham B, Klassen TP, Schulz KF, Berlin JA, Jadad AR, et al. What contributions do languages other than English make on the results of meta-analyses? J Clin Epidemiol. 2000;53(9):964–72. Epub 2000/09/27. pmid:11004423.
- 84. Burton WN, Chen CY, Conti DJ, Schultz AB, Edington DW. The association of antidepressant medication adherence with employee disability absences. The American journal of managed care. 2007;13(2):105–12. Epub 2007/02/09. pmid:17286530.
- 85. Wade SW, Satram-Hoang S, Nadkar A, Macarios D, Tosteson AN. Impact of medication adherence on health care utilization and productivity: self-reported data from a cohort of postmenopausal women on osteoporosis therapy. Clin Ther. 2011;33(12):2006–15. Epub 2011/11/22. pmid:22099387.
- 86. Loeppke R, Haufle V, Jinnett K, Parry T, Zhu J, Hymel P, et al. Medication Adherence, Comorbidities, and Health Risk Impacts on Workforce Absence and Job Performance. Journal of occupational and environmental medicine. 2011;53(6).
- 87. Carls GS, Roebuck MC, Brennan TA, Slezak JA, Matlin OS, Gibson TB. Impact of Medication Adherence on Absenteeism and Short-Term Disability for Five Chronic Diseases. Journal of occupational and environmental medicine. 2012;54(7). pmid:22796923