Authors: Sakunika Ranasinghe, Abarna Nadeshkumar, Savini Senadheera, Nithushi Samaranayake
Categories: Systematic Review, Cost-effectiveness, Health policy, Medical error, measurement/epidemiology, Medication safety, 1506
Source: BMJ Open Quality
Medication errors are an unnecessary cost to a healthcare system and patients of a country. This review aimed to systematically identify published cost variables used to calculate the cost of medication errors and to explore any updates on findings already known on calculating the cost of medication errors during the past 10 years.
A systematic review was conducted according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines. Electronic databases, PubMed, Scopus, Emerald and JSTOR were searched, using keywords “medication error” AND “cost” and predetermined inclusion criteria. Duplicate articles were removed. Quality check was done using 10 criteria. Cost variables used in calculating the cost of medication errors were extracted from each article.
Among 3088 articles, 33 articles were selected for review. Most studies were conducted in Western countries. Cost variables used (types and number) by different studies varied widely. Most studies (N=29) had used direct costs only. A few studies (N=4) had used both direct and indirect costs for the purpose. Perspectives considered when calculating cost of medication errors also varied widely. A total of 35 variables used to calculate medication error costs were extracted from selected articles.
Variables used to calculate the cost of medication errors were not uniform across studies. Almost a decade after systematic reviews previously reporting on this area, a validated methodology to calculate the cost of medication errors has still not been reported to date and highlights the still pending necessity of a standard method to be established.
Keywords: Health policy; Medication safety; Medical error, measurement/epidemiology; Cost-effectiveness
Medication error can be defined as a ‘failure in the treatment process that leads to or has the potential to lead to, harm to patient’. Medication errors can happen at any stage of the treatment process^1^ be it prescribing, compounding, dispensing, medication administration or monitoring.^2^ They are medication related, always preventable and would, therefore, include preventable adverse drug reactions as well.^2^ In addition to the unacceptable harm caused to patients, medication errors are an unnecessary cost to a healthcare system and patients of a country.^3^ This study focuses on the cost of medication errors.
Cost of healthcare can differ according to the perspective that is being considered; provider, patient or third-party payer (insurance companies). The cost to the provider would be, the expenses of delivering healthcare services to patients. For the patient, it is the cost that they have to pay out of pocket for healthcare. Cost for third-party payer would be the amount that they pay to the providers for the services rendered for their client (patient).^4 5^
There are three types of healthcare costs. They are direct costs, indirect costs and intangible costs. Direct costs are the monetary costs directly related to prevention, treatment and diagnosis of the disease and include fees for services such as professional, medication, surgery, hospital stays, diagnostic tests like X-rays, ambulances and food. Indirect costs are also monetary costs, but they are not directly related to treating the disease. They include losses due to an inability to engage in normal daily activities, work, domestic responsibilities and loss of income. Intangible costs are social, emotional and human costs (damage or loss to people). They are not related to money and are not measurable. Costs of pain, worry and other suffering that a patient or his family might endure are examples of intangible costs.^6^ The cost of a medication error is ideally the sum up of all direct, indirect and intangible costs spent due to that error. Evaluation of direct and indirect costs is quite objective while evaluation of intangible cost is subjective.^7^
Appropriate and feasible cost variables (or elements, or parameters) of direct and indirect costs should be established first in order to calculate the cost of medication errors, and for these costs to be comparable across settings and countries.^8^ These variables could vary by country because healthcare services and resources differ from country to country depending on their economy.
Among related studies, there were two systematic reviews that collated research on cost of medication errors. Patel et al ^8^ and Walsh et al,^9^ reporting in 2016 and 2017 respectively, assessed the cost variables used in calculating the cost of medication errors by different research groups. Different perspectives of calculating costs, using a multitude of variables were reported, but neither reviews concluded on a universal/standardised set of variables that could be considered to compute the cost of medication errors, nor a formula or a model that could be used globally for this purpose.^8^
Patel et al ^8^ conducted a systematic review to identify approaches for calculating medication error costs across healthcare settings and included studies from 1993 to 2015. The review concluded that, a standard approach for exploring the costs of medication errors was lacking. He reported inconsistencies in the terminology used, and in the methods used to calculate cost of medication errors. The review found different methodologies used to derive the cost while in some cases, the same methodology was applied in different ways. Cost inputs used varied across studies, were not explicitly defined and did not describe how the cost inputs were relevant to medication errors. The number of cost inputs used to calculate medication error cost varied across studies based on subjective judgement of researchers. Therefore, Patel et al ^8^ recommended that future research is required to determine the most appropriate context-specific method for calculating costs.
Walsh et al ^9^ conducted a systematic review to quantify the economic burden of medication errors and also to identify methods and parameters used when calculating the cost of medication errors. This review included studies from 2004 to 2016 and reported similar findings to that of Patel’s review. Further, Walsh et al ^9^ observed that the difference between calculated costs of medication errors was as large as €100 000 between studies which confirmed the lack of a standardised methodology for this purpose. His findings were mostly related to the hospital setting and lacked the social perspective of economic burden. Importantly, Walsh et al too noted the variability in financial information sources used to determine costs.
The burden of medication errors needs to be costed or otherwise its gravity cannot be justified against the investment made to avoid them. It is beneficial to have a universal formula which can be adjusted according to the country and healthcare setting, and enable objective comparisons across countries. According to Patel et al ^8^ and Walsh et al,^9^ a standard methodology for calculating the economic burden of medication errors was lacking as at 2016. The current review aimed to explore any updates to this information based on recent research, and to systematically identify published cost variables used to calculate the cost of medication errors in local and international literature during the past 10 years (2011–2021).
The systematic review was conducted based on the methodology specified in Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines.^10^ The search was carried out in electronic databases, PubMed, Scopus, Emerald and JSTOR using keywords “medication error” AND “cost”. Related articles published from 1 January 2011 to 13 November 2021 in English language were searched in databases using predetermined inclusion and exclusion criteria specified in table 1.
Title and abstract screening were done simultaneously, relevant articles were selected by two reviewers, and discrepancies were resolved through discussion. Duplicate articles were removed by EndNote V.X9 software. Then, full texts were read by first reviewer to assess if articles were compatible with inclusion criteria. Articles including the rejected ones were reviewed by a second reviewer and any discrepancies were resolved until 100% agreement was reached. Final articles to be included in the systematic review were decided. Quality of each selected article was checked through a checklist of 10 criteria as done by Elliott et al.^11^ Cost variables used in the calculation of medication error cost were extracted from each article.
A total of 3088 articles resulted from the initial keyword search in stated databases. Among them, 1485 articles were removed after limiting for language and year of publication, and 50 were removed due to duplicating of articles. After the title and abstract screening, 188 articles were selected for full-text reading. After removing articles which did not comply with inclusion criteria, 52 articles were selected for detailed analysis. Discrepancies were resolved through discussion among the two reviewers until 100% agreement was reached to include 33 articles for the systematic review. PRISMA flow diagram of articles selected for systematic review is shown in figure 1.
Figure 1 PRISMA flow diagram of articles selected for systematic review. PRISMA, Preferred Reporting Items for Systematic Reviews and Meta-Analyses.
When quality of each article selected for the systematic review was assessed, only 6 articles matched 9 out of 10 criteria (18.18%) considered. Criteria such as, clearly mentioning the objective/s of the study, recruiting/selecting all subjects from the same or similar populations and reliability of the method of collecting data (medication errors), were met by all articles (100%).
Online supplemental table gives the general details of studies selected. We reviewed 33 articles which resulted from the systematic procedure explained in the methodology. They were from different countries (USA=7, UK=3, Brazil=3, France=3, Spain=2, Netherland=2, South Korea=1, Taiwan=1, Singapore=1, Romania=1, Ireland=1, Sri Lanka=1, Arabia=1, Germany=1, Iran=1, Malaysia=1, Switzerland=1, Mexico=1 and Canada=1). Most of the studies in the systematic review were conducted in western countries (78.78%).
Perspectives considered when calculating medication error cost differed considerably. Seven approaches of perspectives were considered in studies; provider perspective (N=10), patient perspective (N=3), insurance company perspective (N=1), patient and insurance company perspective (N=2), provider and patient perspective (N=2), provider, patient, and insurance company perspective (N=1), and no perspective defined (N=14). Most studies had not defined the perspective.
The type of medication errors considered for calculating costs varied among studies. Some studies had calculated the cost of any medication error,^3 12–14^ while some had considered prescribing errors only.^15 16^ Some studies had limited the scope to more specific types of errors like, DePuy et al ^17^ had calculated cost for antiretroviral-related errors.^17^ Forster et al ^18^ had calculated cost for inaler handling errors, Al-lela et al ^19^ had calculated cost for imunisation dose errors, and Ranchon et al ^20^ had calculated the cost for hospital readmissions due to drug-related problems^21 22^ and cost due to adverse drug reactions.^23–25^
Cost variables used for calculating cost of medication errors identified by each article were extracted and are shown in table 2.
As in previous systematic reviews,^8 9^ our systematic review also showed that different studies calculated cost of medication errors in different ways. There was a wide variation in the methodologies used to calculate the cost of medication error. Some studies had used economic models to calculate medication error cost,^3 14 18^ some had calculated cost avoidance due to clinical pharmacists’ interventions where avoided costs were mainly medication cost in most studies.17 22 26–30 Some studies had calculated incremental cost-effective ratios.^21 24 31^ There was still no reported research on a validated methodology to calculate the cost of medication errors.
Cost variables used (types and number) by different studies also had a wide variation, similar to findings by Patel et al ^8^ and Walsh et al.^9^ Most studies (N=29) had used direct costs only. Even then, some studies (N=21) had specified the type of direct cost variables such as, cost of hospitalisation, medication costs, nursing care costs, diagnostic tests costs and emergency department visit cost,^23 32^ while some (N=8) had just only mentioned direct costs without disaggregating the variable.^17 33 34^ Some studies (N=4) had used both direct and indirect costs to calculate medication error costs.^18 21^ Studies such as by Karapinar-Çarkıt et al ^21^ had described the cost variables used in their study very clearly. Where indirect costs variables were used, the cost variables often used were cost for absenteeism from paid and unpaid work, and cost of permanent harm to patient.18 21 35 36 Litigation cost variable was rarely considered in studies. However, McCullagh and Slattery,^37^ in a 6-year review about medication related litigation in Ireland, stated that ‘the median total cost, in purely financial terms, of a medication‐related claim that closed with a payment to the plaintiff was €60 991, including median damages of €33 858’.^37^ This suggests that litigation costs are also an impactful cost variable to consider.
The two systematic reviews by Patel et al and Walsh et al which were on medication error cost calculation at the initial literature survey of this study were mostly outside the study period considered in this review. In fact, we deliberately selected the study period to avoid these two systematic reviews so as to avoid duplication of findings and to explore if any changes had taken place since. Patel et al ^8^ and Walsh et al ^9^ included studies up to 2015 but the current search was from 2011 January to 2021 November. We found 19 related articles which were published after 2017 which contributed to our systematic review, and 6 articles which were published before 2017 but not included by Patel et al and Walsh et al. The current systematic review had eight articles that overlapped with Patel et al’s (24.2% overlap), and six articles that overlapped with Walsh et al’s (18.18% overlap). There were only seven articles that overlapped between Walsh et al and Patel et al (21.21% overlap). Considering that there was minimum overlap between the current systematic review and past literature, it is noteworthy that the findings of research conducted after 2015 had not changed. Still, the most appropriate context-specific method for calculating the cost of medication errors has not been established as recommended by Patel et al.^8^ Clear description of cost sources and explicit cost calculations were not available as recommended by Walsh et al.^9^ We were able to extract the various cost variables that were used for the purpose of calculating the cost of medication errors, but as found in previous reviews, they were not systematically used by all. The appropriateness, adequacy, relevance and feasibility of using these cost variables were not assessed and appeared to be subjective.
There are some limitations to this systematic review to be acknowledged. We only included studies related to the past decade, mainly to avoid duplications in findings with previous reviews. However, by including articles from 2011 onwards we may have missed some important articles which are related. We used only “medication error” AND “cost” as keywords which may have led to miss some important articles. Electronic databases search was limited to four databases, and some articles which we felt were relevant could not be downloaded (N=21).
This systematic review revealed that different studies had used different cost variables to calculate the cost of medication error. Most studies used one or two variables only, and a very few had considered different possible cost variables. That too, the appropriateness, feasibility and relevance of the variables used were not established nor validated and appeared to be subjective. Almost a decade after systematic reviews reporting on this area, a validated methodology to calculate the cost of medication error has still not been reported to date and highlights the still pending necessity of a universal formula or standard method to be established.
Absence of a universal formula or at least a standard list of cost variables to calculate the cost of medication errors has led to inconsistencies and generation of non-comparable medication errors costs across countries. This systematic review is evidence for these irregularities, and thus, we recommend that a standard methodology that is universally acceptable should be devised on calculating the cost of medication errors which could be adjustable according to a healthcare setting of interest. Also, cost variables identified through this systematic review provide an evidence base for policy-makers in the world on developing a standard methodology/guideline for calculating cost of medication errors. The cost variables identified through this review could be a draft to finalise a standard list of cost variables with the help of experts in healthcare such as doctors, nurses, pharmacists, cost accountants and statisticians. Once a universally acceptable standard list of cost variables has been established, the appropriateness, accessibility and measurability of each of these cost variables would have to be explored for a particular country before implementation. Deviations from the standard methodology (or standard list of cost variables) could be disclosed when publishing costs of medication errors in order to overcome limitations in comparability.
All data relevant to the study are included in the article or uploaded as online supplemental information.
Not applicable.
All data relevant to the study are included in the article or uploaded as online supplemental information.