Pharmacist-Led Interventions to Reduce Drug-Related Problems in Prescribing for Pediatric Outpatients in a Developing Country: A Randomized Controlled Trial

“Drug-related problems (DRPs) are drug treatment-related events or situations that interfere with the intended health outcome after treatment.”¹  DRPs can occur anytime during prescribing, dispensing, or administrating, and anyone taking a drug can experience DRPs. One study showed common DRPs included poor therapy control, non-optimal drug, intentional non-adherence, drug-drug interactions, inappropriate drug, and inappropriate dosages.^2,3  In our study, we surveyed DRPs related to prescribing through examining physician prescriptions, including unintentional prescribing errors and non-compliance with prescribing guidelines made by physicians. DRPs are usually caused by errors in prescribing by physicians and pharmacists, errors in dispensing medicine by nurses, or a lack of drug knowledge and compliance on the part of patients. DRPs can lead to treatment failure and increase the need for follow-up visits, rehospitalization, or additional medications to manage negative consequences, thus significantly increasing the cost of treatment. A study by Ayalew et al⁴  indicated that DRPs accounted for between 1.3% and 41.3% of hospital admissions, depending on the studies and the populations studied. Moreover, between 0% and 5.7% of patients died because of DRPs. Annually in the United States, DRPs are estimated to lead to 17 million emergency department visits and 8.7 million hospital admissions. In developing countries (including Vietnam), 38.46% of patients experienced DRPs during treatment.⁵  Many studies worldwide have proven the role of pharmacists in detecting and amending DRPs.^5–7  Pharmacist-led interventions have been shown to reduce the rate of DRPs, health care costs, and mortality and improve patients' quality of life by providing information on the condition of disease for patients, or side effects of the drug, and proposing appropriate drug guidelines.^8–10  A study of a French pharmacist-led intervention on prescribing antibiotics in pediatric outpatient clinics reported that physicians agreed to adjust 93.7% of the DRPs identified.¹¹  The involvement of clinical pharmacists has influenced patient-oriented clinical outcomes in Canada, contributing to a decrease in the incidence or slow course of disease and complications, and increased survival rate and control-related risk factors.¹²  Moreover, clinical pharmacist assistance reduced the number of patients exposed to DRPs related to medications.¹³  Despite the important role of pharmacists-led interventions, many medical facilities in Vietnam have incorrect prescriptions owing to a lack of clinical pharmacists and management software, resulting in DRPs from prescription errors, such as overuse of vitamins and the prescribing of too many drugs or inadequate instructions for drug use. Therefore, it leads to non-guaranteed treatment, affecting the process of drug adherence and increasing DRPs in Vietnam.¹⁴ 

DRPs can occur among any patient group and at any age, particularly in children. Compared with adults, children have a continuously changing metabolism and incompletely developed physical structure and biological functions; they thus have a greater risk of developing DRPs. Pediatric patients also face a greater risk of DRPs because of limited data on pediatric drug efficacy and safety, unsuitable dosage form, and differences in pharmacokinetics or pharmacodynamics of drugs for different age groups.¹⁵  It is estimated that half of all children have at least 1 chronic disease. Still, only about 60% received prescriptions over the past 12 months.¹⁶  Even though children need good health care for comprehensive mental and physical development, many studies on reducing DRPs have been conducted among patients aged 18 years and older,^17–19  but studies among children are still limited. We therefore conducted a study to assess the effectiveness of a pharmacist-led intervention to reduce DRPs in prescription, thus ensuring effective treatment outcomes for pediatric outpatients at Can Tho Children's Hospital.

Study Design and Population. To study prescription errors, we conducted a randomized controlled trial involving all physicians at outpatient clinics for patients with health insurance at a children's hospital in Can Tho City.

Study Setting and Recruitment. Vietnam has enhanced the role of clinical pharmacist interventions in outpatient prescribing-related DRPs in recent years. Pharmacists usually analyze prescriptions and report their findings at hospital progress meetings to reduce prescribing issues. In hospitals with complete pharmacist human resources, pharmacists check prescriptions manually or with software before they are given to patients to immediately recommend interventions for physicians. The development of a prescribing checking system on software has supported prescribing pharmacist-driven interventions with remarkable results in front-line hospitals with many outpatient prescriptions (with a volume of 1000 to 2000 per day). In the case of the studied hospital, no software is available for pharmacists to review prescriptions. The hospital prescriptions are retrospectively reviewed, then reported during the progress meeting or medical review to limit outpatient prescribing-related DRPs. Specifically, clinical pharmacists record DRPs identified and develop information about drugs related to the identified DRPs. Such information, together with recommended information for prescribing (produced from such information), is presented during hospital meetings by a clinical pharmacist. Two other clinical pharmacists are in charge of answering physicians' questions related to the presented content.

Baseline recruitment took place in January 2020. Data were collected after the intervention in August 2020. We recruited all clinical pharmacists in the hospital, namely 3 people who received postgraduate training or short-term training in clinical pharmacy. All of the fixed or rotating physicians at 14 internal clinics under the health insurance department were recruited for the study, excluding those absent during the research period (owing to school attendance or maternity leave). The examining outpatient schedule consists of 4 fixed physicians in the outpatient department (scheduled weekly, with 2 being on duty every other week), and 2 to 5 daily arranged physicians from the internal medicine departments. Therefore, each physician prescribed medications at least once every 2 weeks. Thus, we collected prescriptions for at least 2 weeks for each period to collect sufficient prescriptions of all physicians recruited. We screened prescriptions of recruited physicians for eligibility. The exclusion criteria were prescriptions at follow-up visits or lacking patient information (age, sex, and weight). In our study, the minimum number of physician prescriptions at each stage was 25, so we randomly selected 25 prescriptions per physician at each period.

Randomization and Intervention. We used an online randomization number generator (randomization.com) to divide physicians randomly into 2 groups and select prescriptions by each physician. We sorted physicians alphabetically by name and randomly allocated them to 2 equal parallel groups (control and intervention), using a blocking technique with random permuted blocks of 2, 4, or 6 physicians. We sorted prescriptions according to the physicians who had prescribed them and numbered them in reverse chronologic order. To arrange prescriptions with the same date, we used patient codes. Numbered prescriptions were randomly permuted and selected in order of their appearance. Allocation concealment was not implemented in our study owing to the short intervention period because it was unlikely uninfluenced on the results.

In 3 intervention weeks, physicians from the control group received only the usual interventions (via reports and meetings). In contrast, their counterparts in the intervention group were provided with recommended prescribing information, edited by pharmacists, apart from the usual interventions. Pharmacists intervened in DRPs in the following ways: 1) in the first week, preparing content and reporting DRPs during a hospital progress meeting; (2) in the second week, providing physicians with further information on medicine related to each DRP group; and (3) in the third week, repeating the information (Supplemental Table S2). Further, during the intervention period, clinical pharmacists oversaw answering questions and providing additional information required by physicians. During the intervention period, the information provided to physicians was from the same references as the DRP assessments.

Data Collection. Clinical pharmacists identified DRPs related to baseline prescriptions, based on references commonly used at the hospital: instruction manual; Pediatric Treatment Regimen (outpatient section) of the Children's Hospital; guidelines for diagnosis and treatment of several children's diseases by the Ministry of Health; and the Vietnamese National Drug Formulary. Additional references were the British National Formulary for Children²⁰ ; Micromedex²¹ ; UpToDate²² ; AHFS (American Hospital Formulary Service) Drug Information²³ ; and eMC (electronic Medicines Compendium)²⁴ . Any non-conformity to the above-indicated DRPs is classified according to PCNE (Pharmaceutical Care Network Europe) 9.1 (Supplemental Table S3). We examined prescriptions collected at each stage to evaluate pharmacist-led intervention efficacy in decreasing the prevalence of prescription-related DRPs. The proportion of prescriptions with DRPs was the primary outcome. Secondary outcomes were the proportions of prescriptions with specific types of DRPs (drug choice, dose selection, dosage form, and dosing timing relative to meals). Patient characteristics might influence the effect of DRP occurrence in addition to pharmacist-led interventions, especially when these features differ at baseline and post-intervention. We therefore performed multivariate logistic regression to evaluate the effectiveness of the pharmacist-led intervention and adjust the effects of the survey factors (age group, sex, primary disease, total drugs in prescription).

Analysis. We analyzed data with Microsoft Excel 2019 and IBM SPSS statistics 26.0 software. Qualitative variables (patient characteristics, DRPs) were described by frequency and percentage. We compared the differences between patient characteristics and DRPs of prescriptions and differences between characteristics of the control and intervention groups at baseline and post-intervention, with 95% confidence. We used a chi-square test and Fisher test (when more than 20% of the comparison table cells had an expected value of <5) to compare the 2 ratios. The independent samples t tests (for quantitative variables with normal distribution) and Mann-Whitney U test (for quantitative variables without normal distribution) were used to compare the 2 mean values (mean number of DRPs per prescription) of 2 independent samples (control and intervention group at baseline and post-intervention). A multivariable logistic regression model and variable Enter method were used to determine factors affecting DRP occurrence.

Sample Characteristics. The Figure summarizes collecting data on baseline characteristics and outcomes after the pharmacist-led intervention. Of 40 eligible physicians, 31 (77.5%) were included; 9 (22.5%) physicians were excluded on the basis of our exclusion criteria. Of the 31 included physicians, 15 (48.4%) were randomly assigned to the control group and 16 (51.6%) to the intervention group. We randomly collected 25 prescriptions per physician at each stage, with 775 prescriptions (375 from the control group and 400 from the intervention group). Most physicians had a university degree (5 times higher than a master's degree) and less than 5 years seniority. The difference between the control and the intervention groups' physician characteristics (age, sex, education level, seniority) was not statistically significant, as presented in Table 1.

At the start of the study, most patients receiving prescriptions were older than 2 years but not older than 6 years (47.2%); 55.5% were males, and 77.4% were diagnosed with respiratory disease. Most patients had no comorbidity and fewer than 5 drugs in their prescriptions. Patient characteristics between the control and intervention groups showed no significant difference (p > 0.05). The other patient characteristics are shown in Table 2. Prescriptions with DRPs in the control and intervention groups accounted for 231 of 375 (61.6%) and 257 of 400 (64.3%), respectively; prescriptions involving 1 DRP were twice as many as those involving 2 to 5 DRPs. Overall the most common DRPs were dose selection and administration timing: 34.5%, and 31.5%, respectively (Table 3). The difference between the proportion of DRPs and specific types of DRPs (except for DRPs related to dosage and overdose) in the control and intervention groups was not statistically significant (p > 0.05). This serves as the foundation for comparing these variables throughout the post-intervention (Table 3).

At the endpoint, age, sex, comorbidities of the control group, the intervention group, and the 2 combined were not significantly different from baseline (p > 0.05). Still, differences were found in the main disease and the total number of drugs in prescriptions (p < 0.001) (Table 2).

Intervention Outcomes. After the pharmacist-led intervention, the overall proportion of prescriptions with DRPs decreased from 63% to 45%. The figure for the intervention group was 41%, lower than that for the control group (49.3%). The difference was statistically significant (p = 0.02). The proportion of prescriptions with 1 DRP in the control and intervention groups decreased to 39.7% and 32.8%, respectively, with a statistically significant difference (p = 0.043). Overall, the proportion of specific DRP types (except for those related to administration times associated with meals) had decreased significantly by endpoint, with no significant difference between the control group and the intervention group (p > 0.05). Overall, the proportion of DRPs related to administration time associated with meals had not decreased significantly at the endpoint (p > 0.05). Specifically, the figure for the control group increased from 31.7% to 34.9%. In contrast, it decreased from 31.3% to 25.3% for the intervention group, with a statistically significant difference between the control and intervention groups at the endpoint (p = 0.003) (Table 3).

Multivariate logistic regression analysis showed that physicians' prescriptions in the intervention group were less likely to lead to DRPs than prescriptions in the control group (OR, 0.672; 95% CI, 0.496–0.910); differences were statistically significant (p = 0.01). Patients aged >2 years to ≤6 years were more likely to have DRPs than those aged ≤2 years (OR, 1.871; 95% CI, 1.340–2.613); differences were statistically significant (p < 0.001). Prescriptions with 5 or more drugs were more likely to lead to DRPs than those with fewer drugs (OR, 5.037; 95% CI, 2.472–10.261); differences were statistically significant (p < 0.001) (Table 4).

Principal Findings. Our pharmacist-led intervention involved identifying and reporting DRPs, repeatedly providing drug information, and answering physicians' drug-related questions for 3 weeks. The main contents of the interventions were DRP-related information on drug selection, dosage form, dose selection, and timing of drug administration. The effect of the intervention in reducing DRP occurrence was proven by the lower proportion of prescriptions with DRPs in the intervention group (41%) than in the control (49.3%) at the endpoint (p = 0.02). Noticeably, the proportion of DRPs related to the timing of administration relative to meals increased in the control condition and decreased in the intervention, with a statistical difference between the 2 conditions at the endpoint. When prescribing, health care providers should pay particular attention to drug administration time relative to meals and the number of drugs in prescriptions (≥5 drugs in prescriptions) for patients aged >2 to ≤6 years, because these were significantly related to DRP occurrence.

Explanations and Comparison with Other Studies. Our results corroborated previous studies that demonstrated pharmacist-initiated interventions' important role in reducing prescription-related DRPs. A systematic review showed that clinical pharmacists contributed to a decrease in DRPs by providing prompt and accurate drug-related information to prescribers.²⁵  Many pharmacist-led interventions were applied, including additional prescription review, diagnosis assessment, drug information, pharmacokinetic consultation, identification and management of DRPs, and adverse event prevention. An intervention study by Nguyen Anh Nhut¹⁹  on prescriptions for outpatients ≥18 years old, using the same intervention methods as our study, also showed a positive outcome, reducing DRP prevalence from 89% to 74.9%.

Throughout the years, DRPs related to dose were consistently reported to be the most common, with prevalence ranging from 34.2% to 78%.^15,26–28  This tendency was seen at baseline in our study (34.5%). After the intervention, the proportion in the intervention group decreased from 38.8% to 16.5% (p < 0.001), and the figure for the control group decreased from 29.9% to 16.5% (p < 0.001), with no statistical difference between the 2 groups at the end (p = 0.990). Overdoses and underdoses in both groups also decreased (p < 0.001). In pediatric prescribing, dose selection and calculation are often complicated and time-consuming. Overdoses increase the risk of adverse drug reactions, while underdoses decrease the effectiveness of treatment and prolong its duration, thereby affecting treatment costs. Most drugs (such as paracetamol, antibiotics, corticosteroids, and proton pump inhibitors) require weight-based dose calculations. In addition, the content of the solution, for example, syrup for pediatrics, is usually indicated in “mg,”; however, to facilitate pediatric administration, “mL, teaspoon, tablespoon” are often used when prescribing, often leading to confusion and errors in dose selection. As reported in our findings, the reduction in incorrect doses showed that the hospital was paying special attention to reducing dose-related DRPs. This may be because inappropriate dosage causes significant negative effects for patients and increases treatment costs and time. Our findings also showed that physicians now regularly update drug use instructions and more carefully calculate doses.

The proportion of incorrect timing of administration relative to meals decreased in the intervention condition. It increased in control, resulting in the most significant difference between both at the endpoint (31.3% to 25.3% vs 31.7% to 34.9%, respectively). With specific drugs, such problems could reduce or even prevent effective treatment outcomes, and the lack of studies that evaluate this DRP type made it difficult to compare results. Furthermore, the timing of administration related to meals is often ignored in prescribing by physicians because they consider it unimportant for treatment. This led to a minor post-intervention decrease in the proportion of DRPs reported concerning administration time. We detected that DRPs and their negative effects should be carefully identified and reported to increase the awareness of physicians and other health care providers.

The proportion of DRPs related to drug choice and dosage form was significantly reduced at the endpoint, but without statistical difference between the 2 groups (p > 0.05). DRPs related to drug selection in the intervention group decreased from 4.5% to 2.8% (p = 0.001), similar to the control group but without statistical significance (p = 0.185). Nevertheless, intervention on drug selection is favorable, as pharmacists explained that DRPs related to drug choice, prescribing unnecessary drugs, and inappropriate diagnosis could lead to adverse reactions, increase in treatment costs, or refusal of payment by health insurance. The proportion of DRPs related to dosage form in the intervention and control groups decreased (from 10% to 4.5%, p < 0.001 vs from 11.2% to 4.5%, p = 0.003, respectively). Thus, both groups had almost equal reduction rates, but that did not mean that our intervention was ineffective. This may have been due to inadequate information from the pharmacy department regarding dosage forms. This issue occasionally raises conflicting opinions from physicians who question whether the dosage form significantly affects outcomes. Nevertheless, choosing the adequate form makes it easier for family members to measure doses, ensuring the best treatment effects.

Multivariate logistic regression analysis showed that prescriptions by physicians in the intervention group were less likely to lead to DRPs than in the control group (OR, 0.672; 95% CI, 0.496–0.910); results were statistically significant (p = 0.010) and confirmed the effectiveness of the pharmacist-led intervention in reducing DRPs. Patients aged >2 to ≤6 years were more likely to experience DRPs than patients aged ≤2 years (OR, 1.871; p < 0.001); these results seem to contradict those of other studies.^22,24,25  The difference may be because in our study, patients aged >2 to ≤6 years were prescribed more drugs than younger patients, which led to a higher risk of DRPs. Polypharmacy (≥5 drugs) raised the probability of developing DRPs when compared with <5 drugs (OR, 5.037; p < 0.001). These results were consistent with previous studies, which reported that the higher the number of drugs, the more likely the occurrence of DRPs ^26,31,32  because each prescribed drug could cause 1 or more types of DRPs.

Like other pediatric studies, we did not find sex related to DRPs (p = 0.692).^26,27,29,30  Moreover, our study did not find comorbidities related to DRPs; these results differed from findings in the inpatient study.²⁷  In the inpatient study, the number of prescribed drugs sometimes did not increase with the number of diseases diagnosed.

In our study, physician-related factors (education and seniority) did not affect DRPs. This suggested that DRP occurrence could be due to work overload, leading to physicians' inadequate updating of drug information. Therefore, pharmacist-led interventions to provide drug information can help physicians remember and keep up with drug information, reducing prescription DRPs.

Strengths and Weaknesses of the Study. For our study, we performed an in-depth assessment of a pharmacist-led intervention that helped to minimize prescription-related DRPs in pediatrics in Vietnam. Providing prescribing information repeatedly over 3 weeks was shown to reduce DRPs significantly. The feasibility of the study is worth mentioning, because it is inexpensive and can be widely replicated. The research was conducted when the clinical pharmacist's role in Vietnam was gaining attention, and hospitals were focused on limiting inappropriate prescribing involving errors in indications and contraindications. The intervention study provided concrete evidence of its value, encouraging closer cooperation between physicians and clinical pharmacists to help patients achieve optimal treatment results.

However, several limitations in our study should be considered. Firstly, we only identified DRPs in prescriptions without evaluating their effects on treatment outcomes. Further studies could consider assessing the in-depth effect of DRPs on patients, based on our results to propose precise and effective interventions. Further, our intervention focused on hospital reports and only provided drug information on identified DRPs. Different physicians often had different types of prescription-related DRPs. Still, pharmacists in the study did not join with physicians to solve specific cases of DRP occurrence, resulting in the relatively high post-intervention proportion of prescriptions with DRPs in both groups. Moreover, the number of prescriptions with more than 5 drugs decreased after intervention both for control and intervention groups, which is a bias in our study. The reason is that there are also hospital interventions in addition to the study's intervention. Additionally, owing to drug supply at the hospital, the number of available drug items decreases from time to time, leading doctors to prescribe fewer medications in the post-intervention period. We recommend conducting further studies to provide drug information for specific physicians, based on our study intervention programs. Such “tailored” information would be more concise and easily understandable, leading to a higher effect. Another bias might be the absence of allocation concealment and controlling physicians in the control group learning about the intervention. However, the intervention period in our study was short, without a follow-up period; thus, this was less likely to affect physicians' prescribing behaviors and the study results. Further interventions should be implemented on the basis of our study with an extended period to understand the pharmacist-led intervention efficacy with allocation concealment to reduce errors.

In conclusion, a pharmacist-led intervention improved DRP occurrence related to physicians' prescribing for pediatric outpatients with health insurance. In further research, pharmacists could cooperate with physicians in the prescribing process to provide tailored interventions.

AHFS

American Hospital Formulary Service;

DRPs

drug-related problems;

eMC

electronic Medicines Compendium;

PCNE

Pharmaceutical Care Network Europe

Phuong Minh Nguyen, PhD, and Kien Trung Nguyen, PhD, contributed equally to the work and are co-first authors. We sincerely thank you for the support from University of Medicine and Pharmacy at Ho Chi Minh City, Can Tho University of Medicine and Pharmacy, and Can Tho Children's Hospital.