Pons‑Mesquida et al. BMC Medical Informatics and Decision Making (2022) 22:70 https://doi.org/10.1186/s12911-022-01809-6 RESEARCH Open Access Impact of a system to assist in clinical decision‑making in primary healthcare in Catalonia: prescription Self Audit M. Àngels Pons‑Mesquida1,2*, Míriam Oms‑Arias1, Albert Figueras2 and Eduard Diogène‑Fadini2,3 Abstract Background: In 2008, in the context of a complete computerisation of medical records, the Institut Català de la Salut (ICS, Catalan Health Institute) implemented a system in its electronic clinical workstation (ECW) to assist decision-mak‑ ing at the prescription level. This system is known as Self Audit, and it supports physicians in reviewing the medication of their patients. Self Audit provides lists of patients presenting medication-related problems (MRPs) that have poten‑ tial for improvement, and provides therapeutic recommendations that are easy to apply from the system itself. The aim of this study was to analyse the main results derived from the use of Self Audit in primary care (PC) in Catalonia, and the effect of an incentive-based safety indicator on the results obtained. Methods: A descriptive cross-sectional study was carried out to analyse variations in the MRPs detected by Self Audit during 2016, 2017, and 2018 in PC in Catalonia. The effect of a safety indicator on the results obtained was also studied. This safety indicator includes the most clinically relevant MRPs (i.e., therapeutic duplications, safety alerts from the Spanish Medicines Agency, and incidences of polymedication in patients over 65 years of age). Variation in the MRPs was measured using the differences between two evaluation points (initial and final). An MRP was considered resolved if the recommendation specified in the alert was followed. The prescriptions of 6411 PC doctors of the ICS who use the ECW and provide their services to 5.8 million Catalans through 288 PC teams were analysed. Results: Analysis of the total safety-based MRPs detected by Self Audit gave overall resolutions from April to Decem‑ ber of 9% (21,547) in 2016, 7% (15,924) in 2017, and 1% (2392) in 2018 out of the total number of MRPs recorded in April each year. Examination of the 3 types of MRPs with the highest clinical relevance that were linked to the safety indicator gave overall resolutions of 41% in 2016 (17,358), 20% in 2017 (7655), and 21% in 2018 (8135). Conclusions: The ICS Self Audit tool assists in reducing the number of safety-based MRPs in a systematic manner, and yields superior results for the MRPs linked to a safety indicator included in the incentives of PC physicians. Keywords: Decision support system, Primary care, Clinical safety, Electronic prescription Background the cost-effectiveness and quality of patient care. In this In the past few decades, the development of new infor- context, a range of technical reports from the American mation and communication technologies in the field National Institute of Medicine have confirmed that an of healthcare has potentially contributed to improving electronic record of healthcare activity, such as an elec- tronic health record (EHR), together with the integra- tion of clinical decision support systems (CDSSs) in such *Correspondence: aponsmesquida@gencat.cat 1 Unitat de Coordinació i Estratègia del Medicament (UCEM), Institut EHRs, constitute a guarantee of quality for the health sys- Català de la Salut, Barcelona, Spain tem [1, 2]. According to several authors, CDSSs aimed at Full list of author information is available at the end of the article © The Author(s) 2022. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://​creat​iveco​mmons.​org/​licen​ses/​by/4.​0/. The Creative Commons Public Domain Dedication waiver (http://​creat​iveco​ mmons.​org/​publi​cdoma​in/​zero/1.​0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data. Pons‑Mesquida et al. BMC Medical Informatics and Decision Making (2022) 22:70 Page 2 of 12 the prescription of medications have the greatest impact practitioner can also check the schedule of visits for the on improving patient safety [3]. Although a variety of day, which will indicate any patients who have an MRP, different designs and functionalities exist, these systems and allow the doctor to take advantage of the visit to have a common role in intelligently combining clini- review the medication. This system therefore does not cal knowledge and patient information, with the aims of alter the workflow during the consultation [11, 12], and ultimately improving the overall prescribing process. The allows the professional to decide when is the most appro- possibility of integrating a CDSS into the EHR system has priate time to review the MRP. made it possible to provide medical histories with inter- From 2008 to 2016, Self Audit evolved both at the tech- active signals that alert professionals to situations of risk nological level and at the level of its clinical content. Ini- for their patients [4], thereby helping to improve the pre- tially, the tool only allowed the detection of patients with scription process and the overall clinical safety of patients certain therapeutic duplications and/or cases of poly- [3, 5, 6]. Indeed, several Spanish studies have indicated medication (i.e., > 10 drugs). During this period, a num- that 50% of adverse events related to medication errors ber of new MRP detections were incorporated into the are avoidable [7, 8], and that the implementation of such tool, and the detection specificity was improved overall. technologies can help to reduce them. Thus, it was not until approximately 2014 that this tool With this background in mind, in 2008, the Institut was completed in its current Self Audit configuration. In Català de la Salut (ICS, Catalan Health Institute) inte- addition, the process of obtaining data to monitor the use grated a combination of CDSSs into its electronic clinical of the tool was expensive, taking a long time to validate workstation (ECW), namely PREFASEG, which generates and debug the data until the level of quality and detail online notifications when starting a treatment to prevent required for analysis was obtained. medication errors [9], and the Self Audit tool, which gen- Linked to the Self Audit prescription system, an incen- erates lists of patients presenting with active medication- tive-based safety indicator was designed in 2008, which related problems (MRPs). This study focuses on the Self selected some of the most clinically relevant MRPs, and Audit tool. was included in the “payment for objectives” program The Self Audit tool is a computerised system that is for ICS PC physicians (N.B. according to this program, integrated into the EHR, and based on the combination objectives are linked to annual economic incentives up of clinical with therapeutic data, it simplifies the search to approx. 6000 €). The aim of this indicator was to pro- for patients with an MRP related to an ongoing medica- mote a culture of safety in the use of medicines, and also tion, thereby facilitating changes and/or suspensions to encourage the use of the Self Audit tool. of treatment. In an agile and visual manner, it provides The aims of this study are therefore to determine the the professional with a list of patients with an MRP, main results derived from the use of Self Audit in the such as a therapeutic duplication or a drug contraindi- Catalan PC system, and to evaluate the effect of the cated by a previous or current pathology, thereby allow- safety indicator on the results obtained. A further aim ing the review and assessment of any possible change in of this article is to provide the international audience treatment. The tool itself provides a therapeutic recom- with details regarding a computer tool aimed at improv- mendation in each case and facilitates the management ing clinical safety, which has been widely managed in the of changes and/or suspensions of treatment, without Catalan PC system by 6411 users with more than 10 years the need to leave the program. Thus, in this system, of experience. This tool is of particular importance since a number of aspects related to the review of a patient’s it helps doctors to detect patients with potential MRPs, medication are systematised based on an optional and and as a result, any ongoing treatments related to these individual self-evaluation exercise. Any changes carried MRPs can be reviewed and modified to benefit the health out are recorded in the EHR. of the patient. Self Audit is a versatile and dynamic tool The MRPs are defined by a group of expert profession- that can be updated with new or modified warnings as als from the ICS. Each year, the clinical content of the desired. Due to the considered importance of this tool to MRPs is reviewed according to the scientific information provide improvements in healthcare practice, its imple- available, and, for this reason, they may vary from year mentation was essentially immediate for all PC teams and to year. Each MRP is classified as high or low clinical rel- professionals. Physicians refer to Self Audit as a useful evance, as recommended in the literature [10]. tool whose usability needs, nevertheless, to be assessed. The Self Audit tool is activated voluntarily, wherein the practitioner can use the ECW to consult the lists of Methods patients with an MRP in his assigned population. All pri- A descriptive, cross-sectional study was designed that mary care (PC) practitioners (i.e., 100% of practitioners) began in April 2016 and continued until December 2018. use this tool at some point over the course of a year. The This study was developed within the scope of the PC Pons‑Mesquida et al. BMC Medical Informatics and Decision Making (2022) 22:70 Page 3 of 12 system of the ICS, which is the main provider of health more prescribed medications (in 2016 or 2017) and with services in Catalonia, a region in the northeast of Spain, some specific MRPs, such as double antiplatelet therapy and covers a population of 5.8 million inhabitants over for more than 12 months, a combination of anticholin- the different Catalan territories. Overall, it serves the ergic drugs, or other avoidable medications. In 2018, the population through a network of 288 PC teams and 8 denominator changed, and polymedication was defined hospitals. The ICS is a public company that has a total as a patient with 8 or more prescribed medications. of 42,374 professionals, who provide services to 80% of the population of Catalonia. Since all PC doctors of the Data collection and analysis ICS employed the Self Audit tool during routine practice, The data were collected from the ECWs, where the no control group was available to establish a comparison. active prescriptions of the patients are stored. The study We therefore analysed the evolution of the results over was restricted to drugs prescribed and financed by the time. National Health System and employed in PC centres. The extraction of active prescription data was carried Study sample out automatically on a monthly basis, and identified the The sample studied consisted of all the prescriptions of MRPs out of the prescritions of each physician detected the 6411 ICS PC physicians (i.e., 100% of the physician by Self Audit. staff ) who used the EHRs during the study period. Throughout the three years analysed (2016, 2017, and 2018), 6 points or cross-sections of information were Variables and indicators studied. Within each year, the variations in the number of The main variable was the number of resolved MRPs. MRPs between the considered baseline data and the final An MRP was considered “resolved” when: (1) the drug data were calculated and thus the percentage variation or drugs causing the MRP had been dropped from the was established. Data could not be compared between patient’s active prescription, or (2) the diagnosis was reg- different years because the criteria that defined the detec- istered as resolved. tion of an MRP were different from year to year, and so The main types of safety MRPs detected by the Self such a comparison would not have been appropriate. For Audit tool were analysed, wherein clinically relevant example, if a new pharmacological group was to be added MRPs that had been linked to the incentive-based safety to the “duplicate therapy” MRP in a particular year, or indicator were emphasised. This safety indicator included additional drugs were to be included in an existing dupli- 3 MRPs: (1) Therapeutic duplications; (2) safety alerts cation group, the number of MRPs related to duplicate from the Spanish Agency for Medicines and Health therapies would increase. Products (AEMPS, Agencia Española de Medicamentos y The data obtained from the extractions carried out for Productos Sanitarios), and (3) polymedication in patients the month of April were considered as the baseline data over 65 years of age with some specific MRPs, wherein because this was the point at which the MRP detection polymedication is defined as the case where more than criteria were defined and updated, and the incentive- 10 medicines were prescribed in 2016 and 2017, and based goals were proposed. The data obtained for the more than 8 medications were prescribed in 2018. month of December were considered to be the final data The MRP related to therapeutic duplication detected since they correlated to the final month of the calendar patients with a non-beneficial prescription of two or year, and they coincided with the last evaluation point of more drugs that exhibit the same active principle (alone the safety indicator. The difference between the baseline or in combination) and/or the same pharmacological and the final data points reflected the number of resolved action. In addition, “clinically relevant duplications” and MRPs and the number of generated PRMs. The MRPs of “duplications of dose adjustments” (i.e., combinations individual patients were not followed over time. sought with a therapeutic objective) were clearly differen- The safety indicator averages the variation in the selec- tiated, and only those considered relevant were linked to tion of the MRPs mentioned above. The effect of the the safety indicator. incentive-based care indicator was therefore evaluated by During the study period, the AEMPS safety alerts the reduction in the number of MRPs at the PC level over included the following contraindications: The “Triple a year, which was the time that the indicator remained Whammy”; coxibs, diclofenac, and aceclofenac; cilosta- unchanged, and which coincided with the validity of the zol; ivabradine; escitalopram and citalopram; trimetazi- management contract signed by the PC doctors. dine; raloxifene and bazedoxifene; strontium ranelate; To evaluate the indicator, the ability of the PC doc- aliskiren; and canagliflozin (see Table 1). tors to reach the goal established at the beginning of the The MRP related to incidences of polymedication year was measured for specific months, and was calcu- detected patients older than 65 years of age with 10 or lated from the baseline data. More specifically, the goal Pons‑Mesquida et al. BMC Medical Informatics and Decision Making (2022) 22:70 Page 4 of 12 Table 1 Summary of the various AEMPS safety alert criteria Drug Alert criteria Citalopram High doses: Above 40 mg/day Above 20 mg/day in patients > 65 years of age Above 20 mg/day in patients suffering from liver dysfunction Administered in combination with other drugs that also prolong the QT interval of the electrocardiogram Escitalopram High doses (> 10 mg/day in patients > 65 years of age) Administered in combination with other drugs that also prolong the QT interval of the electrocardiogram Aliskiren In patients with a diagnosis of diabetes mellitus II or undergoing treatment with antidiabetic drugs Jointly administered with ACE inhibitors Cilostazol In patients suffering from a health problem where its use is contraindicated, i.e., cerebral haemorrhage, severe ventricular arrhythmias, or heart failure Or, in concomitant treatment with: 2 Antiplatelet agents Antiplatelet + oral anticoagulant Trimetazidine In patients with a diagnosis of extrapyramidal and movement disorders Raloxifene or bazedoxifene In patients suffering from any health problem where it is contraindicated, e.g., venous thromboembolism, uterine sac, endometrial cancer, or liver failure of any degree COXIBS In patients suffering from any health problem where it is contraindicated, e.g., ischemic heart disease, peripheral arterial disease, cerebrovascular disease, heart failure, or inflammatory bowel disease Diclofenac or Aceclofenac In patients suffering from any health problem where its use is contraindicated, e.g., ischemic heart disease, peripheral arterial disease, cerebrovascular disease, or heart failure Agomelatine In patients ≥ 75 years of age Ivabradine Co-administration with verapamil “Triple Whammy” In patients ≥ 75 years of age or undergoing treatment for diabetes (NSAIDs + RAS inhibi‑ tors + diuretics) Canagliflozin In patients suffering from a health problem in which it is necessary to be more careful due to an increased risk of amputa‑ tion for each physician corresponded to a specific number of Upon analysis of the total clinical safety MRPs detected MRPs less than that existing at the beginning of the year. by Self Audit, an overall resolution of 9% (21,547) was In the years studied, there were 2 or 3 months of the year observed in 2016, while resolutions of 7% (15,924) and in which the extraction of information from the active 1% (2392) were found in 2017 and 2018, respectively (see prescription MRPs was evaluated, and the ability of the Table 2). doctors to reach the goal was measured. In 2016, the Upon the analysis of all clinical safety MRPs detected evaluation was carried out in September and December, by Self Audit, overall resolutions of 9% (21,547), 7% while in 2017 and 2018 the evaluations were carried out (15,924), and 1% (2392) were observed in 2016, 2017, and in June, September, and December. 2018, respectively. Subsequent analysis of the resolutions of the different MRPs throughout the whole study period Results (i.e., 2016–2018) showed an overall trend towards reso- General analysis of the MRPs of the Self Audit tool lution, especially in the cases where AEMPS safety alerts The data extractions corresponding to the months of were implemented, since this resulted in 49% resolution December 2016, December 2017, and December 2018 of the cases in 2016, 11% in 2017, and 23% in 2018. showed that the ECW had registered 9.5, 9.6, and 9.7 mil- The behaviours of the specific MRPs were then exam- lion active prescriptions, respectively. In these months, ined in further detail. More specifically, in April 2016, 210,916 MRPs (December 2016), 227,856 MRPs (Decem- a total of 46,242 duplications were detected, while in ber 2017), and 230,959 MRPs (December 2018) were December of the same year, such duplications had been identified. Based on these data, it was observed that the reduced by 10% (i.e., to a total of 41,589). However, percentage of MRPs studied with respect to the total there was a significant increase in the absolute number number of active prescriptions represented 2.2% in 2016 of duplications detected in 2018 (65,377 in April and and 2.4% in 2017 and 2018. 64,650 in December), with a reduction of only 1% being Pons‑Mesquida et al. BMC Medical Informatics and Decision Making (2022) 22:70 Page 5 of 12 Table 2 Problems related to medications detected by Self Audit: April 2016–December 2018 Problem detected by Self Audit Year 2016* Apr 2016 Dec 2016 Variation Percentage (%) Duplicate therapies 46,242 41,589 − 4653 − 10 AEMPS safety alerts 13,521 6849 − 6672 − 49 Contraindications due to medical devices and/or clini‑ 37,359 37,421 62 0 cal variables Treatment duration Bisphosphonates ≥ 5 years 8246 7434 − 812 − 10% Double anti-aggregation ≥ 12 months 4805 4332 − 473 − 10% Drugs advised against in geriatrics 88,393 83,638 − 4755 −5 Combination of anticholinergic drugs 2913 2320 − 593 − 20 Avoidable medication 30,984 27,333 − 3651 − 12 Total number of problems detected 232,463 210,916 − 21,547 −9 Problem detected by Self Audit Year 2017 Apr 2017 Dec 2017 Variation Percentage (%) Duplicate therapies 65,679 59,536 − 6143 −9 AEMPS safety alerts 11,212 9935 − 1277 − 11 Contraindications due to medical devices and/or clini‑ 44,687 44,000 − 687 −2 cal variables Treatment duration Bisphosphonates ≥ 5 years 6386 4964 − 1422 − 22% Double anti-aggregation ≥ 12 months 4394 4508 114 3% Drugs advised against in geriatrics 82,252 79,187 − 3065 −4 Combination of anticholinergic drugs 2184 1819 − 365 − 17 Avoidable medication 26,986 23,907 − 3079 − 11 Total number of problems detected 243,780 227,856 − 15,924 −7 Problem detected by Self Audit Year 2018 Apr 2018 Dec 2018 Variation Percentage (%) Duplicate therapies 65,377 64,650 − 727 −1 AEMPS safety alerts 7046 5441 − 1605 − 23 Contraindications due to medical devices and/or clini‑ 45,175 46,469 1294 3 cal variables Treatment duration Bisphosphonates ≥ 5 years 4552 4123 − 429 − 9% Double anti-aggregation ≥ 12 months 4631 4630 −1 0% Drugs advised against in geriatrics 79,225 79,384 159 0 Combination of anticholinergic drugs 1854 1736 − 118 −6 Avoidable medication 25,491 24,526 − 965 −4 Total number of problems detected 233,351 230,959 − 2392 −1 *In the three years studied, April was taken as the baseline data because it is the time at which the definitions of the MRPs were updated according to the consensus of a group of experts, and it also is the month in which the incentive-based goals were proposed achieved throughout the year. In addition, it was found in 2016 (− 12%), 3079 cases in 2017 (− 11%), and 965 that the MRPs related to drugs not recommended for cases in 2018 (− 4%). use in geriatric patients exhibited reductions of 5% (4755 Taking the last analysis point, namely that of Decem- cases) in 2016 and 4% in 2017 (3065 cases), although no ber 2018, 34% of the 230,959 MRPs detected were due reduction was found in 2018. Furthermore, the MRP to the use of drugs not recommended for use in geri- related to avoidable medications (including chondropro- atric patients, while 28% were attributed to therapeu- tectors and citicoline) showed a reduction of 3651 cases tic duplications, and 20% were due to pathological Pons‑Mesquida et al. BMC Medical Informatics and Decision Making (2022) 22:70 Page 6 of 12 contraindications. As a result, these three MRPs Furthermore, for the MRP related to polymedication, the accounted for more than three-quarters of the overall corresponding reductions were 5130 cases in 2016, 603 MRPs detected at this point (i.e., 190,503 cases, 82.5%). cases in 2017, and 2915 cases in 2018. The annual data for the safety indicator showed that at Analysis of clinically relevant MRPs linked to the PA safety the time of the evaluation, there was a greater decrease indicator in the number of MRPs, while after each evaluation In the period studied, 41,492 MRP cases were resolved point there was a rebound in the number of cases, as in the Self Audit tool, of which 80% (33,148) were linked can be seen in Fig. 2. The most pronounced rebound was to the safety indicator. Upon examination of the 3 types observed after the December evaluation point, as will be of MRPs linked to this indicator, joint resolutions of 41% discussed later. in 2016 (17,358), 20% in 2017 (7655), and 21% in 2018 Upon analysis of the detail relating to the AEMPS (8135) were observed, as detailed detail in Fig. 1. How- safety alerts (Table 3), it was observed that the Triple ever, despite these promising percentages of resolution, Whammy represented 89% (9767/11,035) of the total the total number of these three MRPs increased from alerts included in 2016, 80% (4123/5153) of those in 24,720 in December 2016 to 31,501 in December 2018. 2017, and 73% (3260/4460) of those in 2018. Its reduction With reference to the specific MRPs, in the case of ther- percentage ranged from 66% in 2016 to 43% in 2018. apeutic duplications, the resolution of 5413 cases in 2016, As indicated in Table 3, the reduction in the number 4892 cases in 2017, and 3485 cases in 2018 was achieved. of cases related to the AEMPS safety alert for diclofenac In addition, when considering the MRPs related to the was 56% in 2016, while the reduction for acecloflenac was AEMPS alert, 6815 cases were resolved in 2016, which 27%. In 2017, the corresponding reductions for diclofenac dropped to 2160 cases in 2017, and 1735 cases in 2018. and acecloflenac were 42 and 38%, respectively, and in MRPs base: 42,078 MRPs base: 37,533 MRPs base: 39,636 MRPs resolved: 17,358 MRPs resolved: 7,655 MRPs resolved: 8,135 MRPs % MRPs resolved: 41% % MRPs resolved: 20% % MRPs resolved: 21% 3,485 4,892 5,413 (18%) (26%) (31%) 5,130 603 2,915 (39%) (5%) (18%) 6,815 2,160 1,735 (61%) (41%) (38%) Base Final Base Final Final Base Base Final Final Base Base Base Final Final Final Base Base Final 2016 2017 2018 Fig. 1 Variation in the MRPs linked to the 2016–2018 incentive-based safety indicator Pons‑Mesquida et al. BMC Medical Informatics and Decision Making (2022) 22:70 Page 7 of 12 2nd Evaluation 2nd Evaluation 2nd Evaluation 3rd Evaluation 3rd Evaluation 1st Evaluation 1st Evaluation 1st Evaluation 2016 2017 2018 Fig. 2 Annual performance of the 2016–2018 incentive-based safety indicator 2018, they were 49 and 45%. Furthermore, the reductions Through analysis of the 20 groups of therapeutic dupli- in cases related to alerts for the coxibs were 31, 31, and cations with the highest number of cases, it was observed 26% in 2016, 2017, and 2018, respectively, while the cor- that in 2017, the groups that presented the greatest responding reductions for the cilostazol alert were 9% in degrees of reduction in cases were the gastric protectors 2016 but 23% in 2017 and 2018. The remainder of alerts (− 18%), the thiazide diuretics (− 15%), and the metam- represented few cases in absolute numbers. izole-type analgesics (− 14%, data not shown). In 2018, Following analysis of the therapeutic duplications the groups exhibiting the greatest degrees of reduction (Table 4), it was found that 70% fell into 10 pharmaco- were the sulfonamide diuretics (− 75%), the thiazide diu- logical groups out of a total of 63. Of these 10 groups, retics (− 24%), and the beta-blockers (− 12%). the renin-angiotensin system inhibitors stood out par- ticularly (2828 detected in December 2016, and 1939 Discussion detected in December 2018), along with non-steroidal The prescription Self Audit system is a clinical manage- anti-inflammatory drugs (2084 and 1969), long-acting ment computer tool, aimed at increasing the quality of benzodiazepines (1796 and 1854), and inhaled gluco- care by giving support to health professionals in the move corticoids (1741 and 1771). These four groups of drugs towards the safe and effective prescription of drugs. represented 68% of the duplicities detected in December The main finding of this study was that Self Audit is 2016, and 48.5% of those detected in December 2018. positioned as a CDSS, which is widely used among the The groups that experienced an increase in detected doctors of the Catalan PC system to help identify and duplications over the same period were the antidepres- resolve safety PRMs in a systematic manner, and leads sants (1417 and 1701 in December 2016 and December to superior results for the MRPs linked to the incen- 2018, respectively), the urinary antispasmodics (1219 and tive-based safety indicator developed for PC physicians. 2020), and the thiazide diuretics (943 and 1595). Thus, Therefore, we think that results could be improved these three groups accounted for 29.1% of the dupli- by implementing awareness strategies and providing cations detected in December 2016, and 34% of those feedback to physicians. This should could be followed detected in December 2018. by more specific recommendations, which should be Table 3 Evolution and reduction percentages by type of AEMPS safety alert Year 2016 2017 2018 AEMPS ALERT Apr Dec Variation Percentage (%) Apr Dec Variation Percentage (%) Apr Dec Variation Percentage (%) Triple whammy 9767 3356 − 6411 − 66 4123 2242 − 1881 − 46 3260 1850 − 1410 − 76 Coxibs 407 282 − 125 − 31 349 240 − 109 − 31 384 283 − 101 − 36 Diclofenac 325 144 − 181 − 56 189 109 − 80 − 42 189 96 − 93 − 97 Cilostazol 115 105 107 82 101 78 Pons‑Mesquida et al. BMC Medical Informatics and Decision Making − 10 −9 − 25 − 23 − 23 − 29 Ivabradine 102 88 − 14 − 14 87 73 − 14 − 16 82 65 − 17 − 26 Aceclofenac 82 60 − 22 − 27 69 43 − 26 − 38 51 28 − 23 − 82 Agomelatine 74 62 − 12 − 16 78 69 −9 − 12 81 63 − 18 − 29 Escitalopram 55 51 −4 −7 54 61 7 13 60 45 − 15 − 33 (2022) 22:70 Citalopram 41 34 −7 − 17 48 47 −1 −2 59 46 − 13 − 28 Trimetazidine 40 27 − 13 − 33 34 24 − 10 − 29 34 26 −8 − 31 Raloxifene and Bazedoxifene 14 9 −5 − 36 10 7 −3 − 30 9 8 −1 − 13 Strontium ranelate 8 9 1 13 3 0 −3 − 100 NA NA NA NA Aliskiren 5 4 −1 − 20 2 3 1 50 3 2 −1 − 50 Canagliflozin NA NA NA NA NA NA NA NA 147 126 − 21 − 17 TOTAL ALERTS 11 4 −7 − 62 5 3 −2 − 42 4460 2716 − 1744 − 64 NA: Not applicable Page 8 of 12 Pons‑Mesquida et al. BMC Medical Informatics and Decision Making (2022) 22:70 Page 9 of 12 Table 4 Top 10 prescribed duplicate groups in 2017 and 2018, including June–December variations Duplicate group* June 2017 Dec 2017 Variation Percentage (%) June 2018 Dec 2018 Variation Percentage (%) Renin-angiotensin system inhibitors 2458 2141 –317 –13 2107 1939 –168 –9 Anti-inflammatories 2154 1880 –274 –13 2203 1969 –234 –12 Long-acting benzodiazepines 2032 1911 –121 –6 2064 1854 –210 –11 Inhaled glucocorticoids 1551 1616 65 4 1583 1771 188 11 Alpha adrenergic antagonists 1273 1299 26 2 1337 1346 9 1 Gastric protectors 1425 1162 –263 –18 1364 1276 –88 –7 Other anti-depressants I 1467 1389 –78 –5 1562 1702 140 8 Urinary antispasmodic agents 1655 1547 –108 –7 1876 2020 144 7 Thiazide diuretics 964 823 –141 –15 2085 1595 –490 –31 Paracetamol (analgesic) 545 668 123 23 523 570 47 8 For technical reasons, the April data were not recorded at the level of detail required for the duplication group and so they have not been included in the table. The data were analysed in June and December, at which points they met the level of quality and detail required for analysis *For each duplicate group, all available active prescription data are shown, which coincide with the evaluation points repeated and regularly inspected These hypothesis, of the AEMPS alerts. This was perhaps due to the fact that course, need to be verified. there is greater response from professionals when a safety In the period studied, 41,492 cases of potential safety alert is issued by a regulatory body [13]. Although the problems that could affect patient health were resolved, use of AEMPS alerts resulted in a significant reduction of which 80% (33,148) were linked to the safety indica- in the number of cases in 2018, it should also be pointed tor. In general terms, the percentage of MRPs detected out that part of this reduction was due to a change in the by Self Audit ranged between 2.2 and 2.4% of the active clinical content, wherein the warnings for citalopram and prescriptions in the ECW (> 9 million) during the years escitalopram became more specific (i.e., alerts were only studied. It should be noted there that the detection of an issued if these drugs were prescribed together with other MRP depended on the defined clinical content, and in medications that prolong the QT interval), and so these the case of Self Audit, this content was updated annually. medications generated fewer detections. As a result, data could not be compared between differ- Several Self Audit MRPs also showed increases in the ent years. number of MRP in the period of study, and this could be Although the number of MRPs resolved was signifi- attributed to various factors, that need to be validated. cant, the percentage of MRPs resolved each year with For example, the increase in therapeutic duplications respect to the number detected by Self Audit was low, over the years could be explained by the fact that new namely less than 10%. In addition, the numbers of some groups of drug duplications or new active ingredients MRPs increased over time, as in the case of therapeutic marketed in different groups had been included. duplications (> 60,000 cases pending resolution in 2018) Another MRP that attracted attention due to its neg- and contraindications due to pathologies (> 46,000 cases ligible decrease, or even a certain increase, was the con- pending resolution in 2018). This result indicates that traindication group. The results related to this MRP can significant numbers of MRPs must still be solved, and so be explained by considering that the content of this MRP supports the need to design interventions that contribute changed substantially during the study period. More spe- to improving the prescribing attitude. Moreover, it will cifically, in 2017, the contents were expanded to include be necessary to analyse the reasons for these increases contraindications due to the altered values of some clini- and/or low resolution levels, in addition to assessing the cal variables (e.g., potassium and glomerular filtration), requirement to make the detections of some MRPs more while in 2018, a global update of the contraindications specific, and/or to more clearly detail the therapeutic rec- took place, thereby resulting in increased detections of ommendations that are offered. this MRP. However, the reasons behind their low resolu- As pointed out in a previous study [9], Self Audit is tion percentages require further investigation. It is pos- common among PC physicians. However, during the sible that this could be attributed to a lack of specificity study period examined herein, resolution of the differ- of the warning and/or recommendation, since different ent Self Audit MRPs was found to be heterogeneous authors [14, 15] have supported the fact that giving a and irregular. More specifically, some MRPs presented clear and precise recommendation constitutes one of the high percentages of resolution, such as those related to success criteria of the CDSS. It is also a possibility that Pons‑Mesquida et al. BMC Medical Informatics and Decision Making (2022) 22:70 Page 10 of 12 the recommendations provided in some cases suggest [3], in addition to increasing the average cost of care [5], that a clinical follow-up should be carried out, and there- increasing the number of visits to primary healthcare fore do not result in the withdrawal of any medication. centres, and increasing hospital admissions [25]. Under such circumstances it would be assumed that the Upon analysis of the MRPs linked to the safety indica- MRP is not resolved, despite the fact that the recommen- tor, it was observed that the resolution of these MRPs dation is actually being followed. was significantly higher than that of the general Self In contrast, the MRP related to the drugs not recom- Audit data. This could be explained by considering that mended for use in geriatric patients exhibited a particu- the included MRPs are of greater clinical relevance, or larly low or no reduction during the years of study. In this that it is an economically incentivised indicator. Another case, there was no change in the clinical content; how- key point is that the reduction in cases decreased year on ever, the low resolution percentage was attributed to this year, both in terms of the absolute number and the per- being an MRP of low clinical relevance, and the fact that centage. One explanation for this could be that the com- the literature [16–18] does not consider that the use of position of the indicator varied each year, and therefore these drugs are fully contraindicated in older patients, the target population for intervention was different, and but instead it is simply recommended that they not be could have been smaller. Another hypothesis that was used. The same argument would serve to justify the low considered was that the baseline starting point improved resolution of the MRP related to avoidable medications over time, until it reached a point where further improve- (i.e., chondroprotectors and citicoline). Thus, when doc- ments were difficult to achieve. tors are faced with different MRPs, they prioritise those The annual plots obtained for the evolution of the that are clinically more relevant, or that can be solved MRPs linked to the indicator clearly showed a decrease more rapidly or with less effort [19]. Another explanation in cases at the time of evaluation. The highest degree to consider for the low resolution percentages associated of MRP resolution occurred at the end of December, with these two MRPs is that they are not included in the and this was accounted for by considering that histori- safety indicator. cally, this indicator had always been evaluated in a sin- It is also known that healthcare practice generates gle evaluation point at the end of the year. Every January, multiple incidences of medication, which suggests that a relevant increase in cases was observed, although the the total resolution of MRPs through the Self Audit tool baseline point was not reached, and so it was assumed was considerably higher than that indicated in the results that the professionals were indeed acquiring a certain of the study. This could be attributed to the resolution culture of safety, and that the MRPs generated during the of some MRPs at the same time as new ones being cre- daily healthcare practice were being solved. The results of ated; this behavior is not reflected in the current study. our study therefore appear to be in line with a previous It should also be noted that the patients with MRPs were study, wherein the authors suggest that incentive-based not followed over time, but instead, the existing MRPs systems could influence physicians, and ultimately lead to under active prescription were compared at two differ- an improvement in healthcare provision [26, 27]. How- ent times within a year. In addition, it must be taken into ever, in a Cochrane review by Scott et al. [28] regarding account that the world population is continually aging, this point, it was concluded that there was insufficient and this is accompanied by a greater incidence of pathol- evidence to indicate whether financial incentives had a ogies, and an increase in the use of medications [20–22]. positive impact on the quality of care in PC systems. Indeed, it has recently been reported that if recent health On the other hand, it is known that intervention strat- trends continue, Spain is on its way to becoming the egies based on improving the prescription of drugs leading country in terms of the highest life expectancy through audits and feedback to physicians have improved in 2040 (i.e., 85.8 years) [23]. As a result of such aging, the quality of care, wherein such feedback includes infor- the greater incidence of multiple associated pathologies mation corresponding to their own patients, in addition results in an increased consumption of drugs, which to specific improvement recommendations; these strate- favours the appearance of increasingly complex thera- gies are repeated and supervised by other colleagues [11, peutic regimens. This in turn is associated with a higher 12, 29, 30]. In this context, the ICS can highlight that frequency of adverse effects, interactions, and hospital this individualised feedback is standard practice for its admissions, in addition to a poorer quality of life and a pharmacists and PC pharmacologists [31, 32]. However, lack of treatment compliance [24]. the collected data show that there is significant room for In terms of the health impact, a reduction in the num- improvement, as the number of MRPs that are pending ber of MRPs can be translated into the avoidance of resolution is considerable. It is therefore evident that it adverse drug effects in patients, which are known to have will be necessary to design specific intervention strate- a considerable impact on patient morbidity and mortality gies to attain a change in the prescribing attitude. Such Pons‑Mesquida et al. BMC Medical Informatics and Decision Making (2022) 22:70 Page 11 of 12 strategies could include the close monitoring of data at an information required to conduct this study. We would also like to thank the Self Audit working group for helping to maintain a useful and up-to-date tool. individual level, training support, and continuous review of the clinical contents to ensure that they are specific Authors’ contributions and that they are accompanied by concrete therapeutic MAP extracted, analysed, and interpreted the data on the use of the Self Audit tool. She also wrote the majority of the manuscript. MO helped to interpret recommendations [33]. the data on the use of the Self Audit tool, in addition to giving support and Furthermore, it does not go unnoticed that it is nec- making contributions to the writing of the manuscript. AF and ED reviewed essary to evolve and improve the Self Audit tool at a the data analysis process, reviewed the various parts of the manuscript, and made the relevant contributions to give clarity and understanding to the technological level to make it more user-friendly and study. All authors read and approved the final manuscript. intuitive, and to impart a greater degree of integration with the patient’s medical records. Moreover, this tool Funding The primary care component of this study was funded by the Institut Uni‑ should be provided with artificial intelligence elements versitari d’Investigació en Atenció Primària (IDIAP Jordi Gol). Scholarship to that possess more agile algorithms for information inter- complete a PhD thesis in primary care. pretation, and to facilitate decision making. Availability of data and materials Finally, it should be noted that one of the main limi- The datasets used and/or analysed during the current study are available from tations of this study is that there is no follow-up over the corresponding author on reasonable request. time for patients with certain MRPs, thereby prevent- ing us from knowing how many MRPs persist over time, Declarations how many are new MRPs, and how many MRPs return Ethics approval and consent to participate or reappear after a while. Indeed, such follow-ups would Not applicable. This manuscript does not report studies involving human be beneficial to allow the consequences on the patient’s participants, human data or human tissue. This manuscript does not report health to be evaluated. Likewise, continuous changes in studies involving animals. the clinical contents also made it difficult to analyse the Consent for publication temporal evolution of each type of MRP. Not applicable. Competing interests The authors declare that they have no competing interests. Conclusions The Self Audit clinical decision support system developed Author details 1 Unitat de Coordinació i Estratègia del Medicament (UCEM), Institut Català by the Institut Català de la Salut helps to systematically de la Salut, Barcelona, Spain. 2 Departament de Farmacologia, Terapèutica i identify and resolve safety medication-related problems Toxicologia, Universitat Autònoma de Barcelona, Barcelona, Spain. 3 Servei de (MRPs) in a systematic manner, wherein superior results Farmacologia Clínica, Hospital Universitari Vall d’Hebron, Institut Català de la Salut, Barcelona, Spain. were obtained for the MRPs linked to a safety indicator that is included in the incentives of primary care phy- Received: 11 May 2021 Accepted: 25 February 2022 sicians. However, it is noted that significant room for improvement exists in the prescribing attitude, and as a result, additional medical awareness strategies will be necessary, as well as improvements to the tool itself. References 1. Institute of Medicine (US) Committee on Quality of Health Care in Such improvements should be based at a technical level America. Crossing the quality chasm: a new health system for the 21th and should be aimed at increasing specificity in MRP century. Washington, DC: National Academies Press; 2001. detection and subsequent recommendations. Finally, we 2. Institute of Medicine (US) Committee on Identifying and Preventing Medication Errors. Quality chasm series. Preventing medication errors. believe that in the context of clinical safety, the imple- Washington, DC: National Academies Press; 2007. mentation HER tools similar to Self Audit could be a use- 3. Bates DW, Cullen DJ, Laird N, Petersen LA, Small SD, Servi D, et al. ful and beneficial healthcare strategy that could benefit Incidence of adverse drug events and potential adverse drug events: implications for prevention. JAMA. 1995;274(1):29–34. patients from other healthcare systems worldwide. 4. McCoy AB, Waitman LR, Lewis JB, Wright JA, Choma DP, Miller RA, et al. A framework for evaluating the appropriateness of clinical decision support alerts and responses. J Am Med Inform Assoc. 2012;19(3):346–52. Abbreviations 5. Bates DW, Spell N, Cullen DJ, Burdick E, Laird N, Petersen LA, et al. The AEMPS: Spanish Agency for Medicines and Health Products (Agencia Española costs of adverse drug events in hospitalized patients. Adverse Drug del Medicamentos y Productos Sanitarios); CDSS: Clinical decision support Events Prevention Study Group. JAMA. 1997;277(4):307–11. system; ECW: Electronic clinical workstation; EHR: Electronic health record; ICS: 6. Bobb A, Gleason K, Husch M, Feinglass J, Yarnold PR, Noskin GA. The Institut Català de la Salut (Catalan Health Institute); MRP: Medication-related epidemiology of prescribing errors: the potential impact of computerized problem; PC: Primary care. prescriber order entry. Arch Intern Med. 2004;164(7):785–92. 7. Aranaz-Andrés JM, Aibar C, Limón R, Mira JJ, Vitaller J, Agra Y, et al. A study Acknowledgements of the prevalence of adverse events in primary healthcare in Spain. Eur J This study has been prepared within the framework of the Doctorate in Public Health. 2012;22(6):921–5. Pharmacology at the Universitat Autònoma de Barcelona (UAB). We would like 8. Aranaz-Andrés JM, Aibar-Remón C, Vitaller-Murillo J, Ruiz-López P, Limón- to thank ICS Information Systems for supporting us during aquisition of the Ramírez R, Terol-García E. Incidence of adverse events related to health Pons‑Mesquida et al. BMC Medical Informatics and Decision Making (2022) 22:70 Page 12 of 12 care in Spain: results of the spanish national study of adverse events. J 32. Madridejos R, Amado E. Formación continuada y feedback informativo. Epidemiol Community Health. 2008;62(12):1022–9. En: Manual de gestión de la prescripción farmacéutica en Atención 9. Catalán A, Borrell F, Pons À, Amado E, Baena JM, Morales V. Seguridad del Primaria. Nin E. Madrid: SEDAP; 2001. p. 125–3. paciente en atención primaria: Proyecto PREFASEG (PREscripción FArma‑ 33. Roshanov PS, Fernandes N, Wilczynski JM, Hemens BJ, You JJ, Handler SM, cológica SEGura). Med Clin (Barc). 2014;143(Suppl. 1):32–5. et al. Features of effective computerised clinical decision support sys‑ 10. Paterno MD, Maviglia SM, Gorman PN, Seger DL, Yoshida E, Seger AC, tems: meta-regression of 162 randomised trials. BMJ. 2013;346(7899):f657. et al. Tiering drug-drug interaction alerts by severity increases compli‑ ance rates. J Am Med Inform Assoc. 2009;16(1):40–6. 11. Saleem JJ, Patterson ES, Militello L, Render ML, Orshansky G, Asch SM. Publisher’s Note Exploring barriers and facilitators to the use of computerized clinical Springer Nature remains neutral with regard to jurisdictional claims in pub‑ reminders. J Am Med Inform Assoc. 2005;12(4):438–47. lished maps and institutional affiliations. 12. Hysong SJ. Meta-analysis: audit and feedback features impact effective‑ ness on care quality. Med Care. 2009;47(3):356–63. 13. Agencia Española de Medicamentos y Productos Sanitarios. Notas informativas de la AEMPS. Madrid: Agencia Española de Medicamentos y Productos Sanitarios (AEMPS). https://​www.​aemps.​gob.​es/​accio​nes-​infor​ mativ​as/​notas-​infor​mativ​as-​de-​la-​aemps/. Accessed 19 Feb 2021. 14. Osheroff JA. Improving medication use and outcomes with clinical deci‑ sion support: a step by step guide. Chicago: HIMSS; 2009. 15. Kawamoto K, Houlihan CA, Balas EA, Lobach DF. Improving clinical prac‑ tice using clinical decision support systems: a systematic review of trials to identify features critical to success. BMJ. 2005;330(7494):765–8. 16. O’Mahony D, O’Sullivan D, Byrne S, O’Connor MN, Ryan C, Gallagher P. STOPP/START criteria for potentially inappropriate prescribing in older people: version 2. Age Ageing. 2015;44(2):213–8. American Geriatrics Society 2019 Updated AGS Beers Criteria® for poten‑ 17. Fick DM, Semla TP, Steinman M, Beizer J, Brandt N, Dombrowski R, et al. tially inappropriate medication use in older adults. J Am Geriatr Soc. 2019;67(4):674–94. 18. Renom-Guiteras A, Meyer G, Thürmann PA. The EU(7)-PIM list: a list of potentially inappropriate medications for older people consented by experts from seven European countries. Eur J Clin Pharmacol. 2015;71(7):861–75. 19. Loo TS, Davis RB, Lipsitz LA, Irish J, Bates CK, Agarwal K, et al. Electronic medical record reminders and panel management to improve primary care of elderly patients. Arch Intern Med. 2011;171(17):1552–8. 20. Turgeon J, Michaud V, Steffen L. The dangers of polypharmacy in elderly patients. JAMA Intern Med. 2017;177(10):1544. 21. Onder G, Marengoni A. Polypharmacy. JAMA. 2017;318(17):1728. 22. Payne RA. The epidemiology of polypharmacy. Clin Med J R Coll Phys Lond. 2016;16(5):465–9. 23. Foreman KJ, Marquez N, Dolgert A, Fukutaki K, Fullman N, McGaughey M, et al. Forecasting life expectancy, years of life lost, and all-cause and cause-specific mortality for 250 causes of death: reference and alterna‑ tive scenarios for 2016–40 for 195 countries and territories. Lancet. 2018;392(10159):2052–90. 24. Wimmer BC, Cross AJ, Jokanovic N, Wiese MD, George J, Johnell K, et al. Clinical outcomes associated with medication regimen complexity in older people: a systematic review. J Am Geriatr Soc. 2017;65(4):747–53. 25. Otero López MJ, Alonso Hernández P, Maderuelo Fernández JA, Ceruelo Bermejo J, Domínguez-Gil Hurlé A, Sánchez RA. Prevalencia y factores asociados a los acontecimientos adversos prevenibles por medicamentos que causan el ingreso hospitalario. Farm Hosp. 2006;30(3):161–70. 26. Bell CM, Levinson W. Pay for performance: learning about quality. CMAJ. 2007;176(12):1717–9. 27. Petersen LA, Woodard LCD, Urech T, Daw C, Sookanan S. Does pay- for-performance improve the quality of health care? Ann Intern Med. Ready to submit your research ? Choose BMC and benefit from: 2006;145(4):265–72. 28. Scott A, Sivey P, Ait Ouakrim D, Willenberg L, Naccarella L, Furler J, et al. • fast, convenient online submission The effect of financial incentives on the quality of health care provided by primary care physicians. Cochrane Database Syst Rev. 2011;9:CD008451. • thorough peer review by experienced researchers in your field 29. Ivers N, Jamtvedt G, Flottorp S, Young JM, Odgaard-Jensen J, French SD, • rapid publication on acceptance et al. Audit and feedback: effects on professional practice and healthcare • support for research data, including large and complex data types outcomes. Cochrane Database Syst Rev. 2012;6:CD000259. 30. Van Wyk JT, Van Wijk MAM, Sturkenboom MCJM, Mosseveld M, Moorman • gold Open Access which fosters wider collaboration and increased citations PW, Van Der Lei J. Electronic alerts versus on-demand decision support • maximum visibility for your research: over 100M website views per year to improve dyslipidemia treatment: a cluster randomized controlled trial. Circulation. 2008;117(3):371–8. At BMC, research is always in progress. 31. Amado E, Pellicer A. Otras herramienta útiles en la gestión de la farmacia. En: Manual de farmacia de Atención Primaria. Madrid: SEFAP; 2006. p. Learn more biomedcentral.com/submissions 693–705.