Non-adherence to the WHO Medicine Prescribing Practice Indicators and Contributing Factors in Ethiopia. A Systematic Review and Meta-analysis

Background: The availability and appropriate utilization of medicine are important to reduce morbidity and mortality from various medical conditions. However, more than 50% of them were used inappropriately. This study aimed to evaluate non-adherence to the WHO prescribing practice indicator and contributing factors in Ethiopia.

Methods: PubMed, EMBASE, and CINAHL databases, and supplementary sources were searched. To be included, studies are required to assess non-adherence to the WHO’s medicine-prescribing practice indicators (i.e., mean number of medicines per encounter, the proportion of generic, antibiotics, and injection prescribing, and the proportion of medicines prescribed from local formulary) at least in one of the five domains. Comprehensive Meta-Analysis (CMA) Software (version 4) was used to estimate the pooled prevalence of the outcomes. I2-test and Egger’s test were used to assess statistical heterogeneity and bias, respectively.

Results: Forty-five studies with total encounters of 80,782 were included. The mean number of medicines per prescription exceeds the WHO reference value by 0.23 (CI: 0.60 – 1.06). The mean prevalence of medicines prescribed by non-generic or brand names and outside the list of essential medicine/local formulary were 10.0% [Pooled Proportion – (PP) = 0.10 CI: 0.08 – 0.12] and 8.0% [PP = 0.08 CI: 0.07 – 0.09], respectively. The mean prevalence of inappropriate prescribing of antibiotics and injections was 24.0% [PP = 0.24 CI: 0.19-0.28] and 21.0% [PP=0.21 CI: 0.13 – 0.29], respectively.

Conclusion: All of the WHO core prescribing indicators are not within the reference value. Therefore, appropriate measures including a multidisciplinary approach to prescribing and involving clinical pharmacists as medical expertise should be considered to lower potentially inappropriate prescribing of medications.

Keywords: Prescribing of medicines; mean number of medicines per encounter; non-generic prescribing; prescribing outside local formulary; inappropriate prescribing of antibiotics; inappropriate prescribing of injections; systematic review

Medications are important in the provision of health services and the management of various medical conditions [1]. Hence, their availability and appropriate utilization are important to reduce global morbidity and mortality from various illnesses [2]. Evidence showed that nearly half of the total healthcare expenditures are spent on medicines and medical products [3,4]. Therefore, appropriate utilization of medicines is important to optimize the use of medicines and the quality of healthcare services; and to maintain the health of individual patients and the community [2,6].

Medicines were used appropriately when patients get the right medicine, in the correct doses, for an adequate period, and at an affordable cost [5]. However, this could be influenced by the health of patients, the behaviors of prescribers, the healthcare environment, the supply chain, and the legislation and available information about the medicine [5]. Therefore, to tackle the inappropriate utilization of medicines and their adverse consequences, appropriate monitoring of medicine utilization in healthcare facilities and the patient is critical.

The extent of medicine utilization was estimated based on the World Health Organization (WHO) medicine utilization indicators [2]. These include prescribing practice, patient care, and healthcare facility indicators [2]. The prescribing practice indicators (i.e., mean number of medicines per encounter, the proportion of generic, antibiotics, and injection prescribing, and the proportion of medicines prescribed from local formulary) reflect how medicines were utilized by healthcare professionals [2]. According to the WHO’s criteria, the mean number of medicines per prescription should be less than 2, antibiotic prescribing should constitute less than 30% of the prescription, injection prescribing should constitute less than 20% of the prescription, and prescribing generic and from local formulary should constitute 100% [2].

According to the WHO criteria, prescribing of medicines should be in line with the above recommendations. For example, the average number of medicines per prescription reflects the degree of polypharmacy [2]. Polypharmacy in turn could predispose patients to experience drug-related problems, adverse medicine events, unnecessary hospitalization and health care costs, and mortality [6]. Likewise, excess utilization of antibiotics could contribute to adverse events, resistance, and unnecessary costs [7]. Moreover, inappropriate use of injection medicines could predispose patients to infections including hepatitis and human immunodeficiency virus (HIV) [8].

The proportion of generic prescribing and prescribing of medicine from local formulary demonstrate the degree of generic prescribing and the level of adherence to the country’s drug utilization policy for the facility surveyed. Therefore, assessing the extent of non-adherence to the WHO’s Core prescribing practice indicators has significant importance in evaluating how medicines are prescribed and utilized in the country.

A study protocol and reporting

The protocol was registered and available in Prospero (R.NO CRD42021257969), and the study was reported in line with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guideline.

Literature Search

PubMed, EMBASE, and CINAHL databases, and supplementary sources such as Google Scholar, and Research gate were searched for pieces of literature. Additionally, the references of the selected studies were searched. The keywords used in the search were; drug use evaluation, drug utilization, drug prescribing, rational use of drugs, and inappropriate prescribing. The literature search was limited to English language articles and had no time restraints.

Eligibility Criteria

Observational studies reporting non-adherence to the WHO’s prescribing practice indicator in any patient group and healthcare facilities in Ethiopia and published in English were considered for inclusion. Studies that reported non-adherence at least to one of the five WHO’s prescribing indicators were included. Review articles, case reports, conference abstracts, qualitative studies, letters, and articles without raw data available/accessible for retrieval were excluded.

Screening

The titles and abstracts of the retrieved studies were screened by five reviewers (AG, DA, AJ, NB and JB) to assess eligibility. Further, the four review authors (FA, AJ, NB and JB) were assessed full-text articles for the pre-defined inclusion criteria, and any disagreement was reached to a consensus through a discussion with the other co-authors (JA, HM and TG). A detailed flowchart outlining the screening process is illustrated in Figure 1.

Figure 1: Database search

Outcome variables

The primary outcomes of the review were adherence to any of the five domains of the WHO’s core prescribing indicators.

Quality assessment and data extraction

Joanna Brigg’s Institutes (JBI’s) Checklist was utilized to evaluate the quality of the included studies. Two review authors (FA and AG) independently extracted the details of articles included in the study such as author, year, design, setup, type of setup, number of setups, total encounter (sample size), and mean number of medicines per encounter, the proportion of generic, antibiotics, and injection prescribing, and the proportion of medicines prescribed from the local formulary. Any disagreement was resolved through the involvement of other review authors (TG and JA).

Definition of terms

An average number of medicines per encounter: the total number of medicines prescribed divided by the total number of encounters (Reference value < 2).

Percentage of medicines prescribed by generic name: the total number of generic medicines prescribed divided by the total number of medicines prescribed, and multiplied by 100 (Reference value = 100%)

Percentage of encounters with antibiotics prescribed: the total number of patients who received one or more antibiotics divided by the total number of encounters multiplied by 100 (Reference value < 30%).

Percentage of encounters with an injection prescribed: the total number of patients who received one or more injections divided by the total number of encounters multiplied by 100 (Reference value < 20%).

Percentage of medicines prescribed from essential drugs list (EDL) or formulary: the total number of EDL medicines prescribed divided by the total number of prescribed medicines multiplied by 100 (Reference value = 100%).

Statistical analysis

Data were analyzed by using Comprehensive Meta-Analysis (CMA) Software (version 4) using the random effect (Der Simonian and Laird method) model. I² statistics were used to look for the presence of heterogeneity whereas the eggers test was used to identify publication bias. Sensitivity analysis was done to check the effect of individual studies on the outcomes of interest and sub-group analysis was performed by total encounter/samples (> 600 vs < 600) and types of healthcare facilities (public vs private). A p-value of < 0.05 was considered significant in all cases.

The systematic search yielded 894 original studies. Out of these, 91 studies were duplicates and excluded and the remaining 803 titles and abstracts were reviewed. Out of these, 717 studies were excluded due to not meeting the pre-defined eligibility criteria. Eighty-six full texts were reviewed for eligibility and 41 studies were excluded for reasons. Therefore, 45 articles that met the inclusion criteria were included in the review (Figure 1).

Characteristics of the included studies

The included original articles assessed the prevalence of nonadherence to one or more of the five WHO’s core prescribing indicators, and all of them were cross-sectional in design. Forty-one studies [9-46, 48-51] were conducted in public health care facilities and 33 were confined to single centers [9-16, 18, 19, 21-25, 27, 28, 30, 32-40, 42, 45, 46, 48-51]. The total number of encounters/samples ranged from 342 [9] to 3058 [17] and it was < 600 in 24 studies [9, 10, 12, 14, 16, 18, 19, 21, 23-25, 27-29, 32-35, 39, 40, 42, 43, 45, 51]. Twenty-eight [9, 11-13, 16, 17, 21, 22, 26, 27, 29-31, 33, 34, 36, 38, 40-46, 50-53] were studies reported the average number of medicines per encounter that exceeds the WHO’s reference value. Prescribing of medicines by their non-generic names was reported in forty-four studies [9-21, 23-28, 30-43, 45, 46, 48-53] while inappropriate prescribing outside of essential medicine list/local formulary was reported in thirty-four studies [10-13-19, 21-28, 32, 35-43, 45, 46, 50, 51, 53]. Inappropriate prescribing of antibiotics was reported in forty-two studies [9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 28, 30, 31,32, 33, 34, 35, 36, 37,  40, 41, 42, 43, 45, 46, 48-53] while inappropriate prescribing of injection drugs was reported in eighteen studies [9, 10, 14, 17,  25, 27, 30, 33, 35, 37, 41, 42, 45, 46, 50, 52] (Table 1).

Prevalence of potentially inappropriate prescribing

The average number of medicines per encounter/samples exceeds the WHO’s reference value by 0.23 (95 % CI: 0.60 – 1.06) (fig 2).

Fig 2: Inappropriate average number of drugs per-encounter.

The pooled prevalence of medicines prescribed by their non-generic/ brand names was 10.0% (Pooled Proportion = 0.10 [95% CI: 0.08 – 0.12]) (fig 3) and the estimated prevalence of medicines prescribed outside the list of local formularies was 8.0% (PP=0.08 95% CI: 0.07 – 0.09) (fig 4). The estimated prevalence of potentially inappropriate prescribing of antibiotics and injections was 24.0% (PP=0.24 95% CI: 0.19-0.28) (fig 5) and 21.0% (PP=0.21 95% CI: 0.13 – 0.29) (fig 6), respectively.

Fig 3: pooled prevalence of potentially inappropriate prescribing of generics in Ethiopia.

Fig 4: Pooled prevalence of potentially inappropriate prescribing outside of the essential medicine list.

Fig 5: The prevalence of potentially inappropriate prescribing of Antibiotics in Ethiopia.

Fig 6: The pooled prevalence of potentially inappropriate prescribing of injection in Ethiopia.

Subgroup Analysis

The inappropriate mean number of medicines was higher in private healthcare facilities as compared to public health sectors (0.71 vs 0.21) (fig 7). But, the inappropriate average number of medicines was similar between larger (> 600) (0.23 95% CI: 0.91 – 1.37) and smaller (< 600) (0.23 95% CI: 0.99 – 1.45) (fig 8) encounters/samples. The estimated prevalence of medicines prescribed by their non-generic names/branded forms was slightly higher at larger (> 600) encounters/samples (PP=0.12 95% CI: 0.09 – 0.15) as compared to smaller (< 600) encounters/samples (PP=0.08 95% CI: 0.07- 0.10) (fig 9) and in public health care facilities (PP=0.10 95% CI: 0.08 – 0.12 (fig 10).  Likewise, the prevalence of potentially inappropriate prescribing of antibiotics was higher at larger (> 600) encounters/samples (PP=0.31 95% CI: 0.25 – 0.38) as compared to the prevalence at smaller (< 600) encounters/samples (PP=0.18 95% CI: 0.13 – 0.23) (fig 11) and in public health care facilities (PP=0.23 95% CI: 0. 19 – 0.28) (fig 12). However, the prevalence of potentially inappropriate prescribing of injections at larger (> 600) (PP=0.21 95% CI: 0.07 – 0.34) and smaller (< 600) encounters/samples (0.21 95% CI: 0.12 – 0.30) (fig 13) and in public health care facilities were similar (PP=0.21 95% CI: 0.11 – 0.30) (fig 14). On the other hand, the prevalence of medicines prescribed outside the list of essential medicine/local formulary at larger (> 600) (0.08 95% CI: 0.07-0.10) and smaller encounters/samples (0.08 95% CI: 0.07-0.09) were similar (fig 15) but slightly lower in public health care facilities 0.07 95% CI: 0.06-0.09) (fig 16).

Fig 7: Inappropriate average number of drugs per-encounter by a study setup in Ethiopia.

Fig 8: Inappropriate average number of drugs per-encounter by number of total encounters in Ethiopia.

Fig 9: Pooled prevalence of potentially inappropriate prescribing of generics by sample size in Ethiopia.

Fig 10: Pooled prevalence of potentially inappropriate prescribing of generics by a study setup in Ethiopia.

Fig 11: The prevalence of potentially inappropriate prescribing of Antibiotics by sample size in Ethiopia.

Fig 12: The prevalence of potentially inappropriate prescribing of Antibiotics by types of healthcare facilities in Ethiopia.

Fig 13: The prevalence of potentially inappropriate prescribing of injection by sample size in Ethiopia.

Fig 14: The prevalence of potentially inappropriate prescribing of injection by the types of healthcare facilities in Ethiopia.

Fig 15: The prevalence of potentially inappropriate prescribing of drugs outside of the essential medicine list by sample size in Ethiopia.

Fig 16: The prevalence of potentially inappropriate prescribing of drugs outside of the essential medicine list by the types of healthcare facilities in Ethiopia.

Publication Bias

The funnel plot symmetry and egger’s test were used to check for the presence of publication bias. The funnel plot symmetry revealed the presence of publication bias for medicines prescribed by non-generic names/branded forms (Fig 17) and outside of the essential medicine list (fig 18). But it did not reveal publication bias for an average number of medicines (fig 19), inappropriate prescribing of antibiotics (fig 20), and injections (fig 21). The eggers regression test showed no publication bias for the mean number of medicines per encounter/samples (p=0.290), non-generic prescribing (p=0.000) (fig 14), prescribing outside the list of essential medicine (p=0.000) (fig 14), inappropriate prescribing of antibiotic (p=0.008) (fig 16) and injections (p=0.096) (fig 17).

Fig 17: Funnel plot revealing Publication bias for potentially inappropriate prescribing of generics.

Fig 18: Funnel plot revealing Publication bias for potentially inappropriate prescribing outside of essential medicine list.

Fig 19: Funnel plot revealing Publication bias for average number of drugs per encounter.

Fig 20: Funnel plot revealing Publication bias for potentially inappropriate prescribing of antibiotics.

Fig 21: Funnel plot revealing Publication bias for potentially inappropriate prescribing of injection.

Measures of prescribing appropriateness are used to flag problematic medicines in a specific population group and areas requiring attention when prescribing medicines Inappropriate prescribing of medicines can lead to adverse health outcomes, particularly in vulnerable populations such as elderly patients with multi-morbidity.

This review found that the non-adherence to the mean number of medicines per encounter was greater in privately owned healthcare sectors as compared to public health sectors. However, this was similar between larger (> 600) and smaller (< 600) encounters/sample. The magnitude of non-adherence to the mean number of medicines per encounter/samples identified in this study was lower when compared with reports of study conducted by Ofori-Asenso et al [54]. Likewise, this finding was lower when compared with WHO reports for African, European, North American, and South East Asian regions [55]. Prescribing of the generic form of medicines could be influenced by various factors such as the supply chain system, prescribers, and patient health conditions. The results of this study revealed that the overall pooled prevalence of medicines prescribed by their non-generic names was 10.0%. This was higher among studies with larger (> 600) encounter/samples as compared to studies with lower encounter/samples (< 600) (12.0% vs 8.0%). The overall estimated rate of non-generic prescribing identified in this review reflects better results when compared with the non-generic prescribing rate in Africa (40%), the pacific region (22.0%), East Mediterranean (72.3%), and Asian region (51.1%) [55]. Similarly, our finding was better than the findings of study done by Ofori-Asenso et al (32%) [54]. Prescribing of the innovator (brand) name medicines may be related to the clients’ request or prescribers’ intention to satisfy their clients who believe that brand names are better in quality, pleasant in taste or color, and has fewer side effects.  It is known that there is no difference in terms of patient outcomes between innovator (brand) and generic drug products except for the cost [56]. The variation in costs between the two forms is estimated to be 20 – 27.5% [57].

In our review, we found that the magnitude of medicines prescribed outside the list of essential medicine/local formulary was 8.0% (Range: 7.0 – 9.0) [i.e.  8.0% lower than the WHO recommendation (100%)]. This was not different by the number of encounters/samples and comparable at private and public health care facilities. The magnitude of medicines prescribed outside the local formulary estimated in this study was lower when compared to findings identified in Europe (44.9%), the USA (28.6%), and South East Asian Region (19.0%) [55]. Likewise, our finding was also lower when compared with the report in a review by Ofori-Asenso et al where 12% of medicines were prescribed outside the list of essential medicines [54]. Good adherence to prescribed medications from the essential medicine list in our review may be due to wider availability and optimal utilization of the essential medicine list /local formulary. Non-optimal use of essential medicine list is often due to supply chain-related factors such as inadequate distribution of medications, characteristics of the medicine formulary [2], unavailability of essential medicine list [58], over-representation of the efficacy of innovator medicine by companies [59] and absence of an obligation to prescribe from the essential medicine list [60].

The prevalence of potentially inappropriate antibiotics prescribing identified in our study was 24.0%. This rate was 31.0% at larger (> 600) and 18.0% smaller (< 600) encounters/samples, and 23.0% in public healthcare facilities. Prescribing antibiotics could be influenced by multiple factors such as cultural beliefs and patient expectations and preference of receiving some types of antibiotics, the extent of marketing and distribution of antibiotics, the effectiveness of drug quality assurance system, and the availability of laboratory facilities to make definitive and differential diagnosis [2]. The overall estimate (24.0%) of non-adherence to the antibiotic prescribing rate identified in this review was comparable with the magnitude reported in Eastern Mediterranean region (23.2%) but higher than the USA (19.3%) and Europe (3.5%) [55]. Our result was also comparable with the inappropriate antibiotics prescribing rate reported in research done in Western Africa (28.0%) [61], and findings of a review conducted in African countries (26.8%) [54, 62]. The inappropriate use of antibiotics could be due to the high magnitude of infection in African region, antibiotic utilization without clear indications, inappropriate adherence to national antibiotic utilization guidelines, demand by clients [63], and financial incentives for overprescribing [64].

The estimated nonadherence to the injection utilization identified in our study was 21.0% (i.e., exceeding WHO recommendation by 21.0%) [2], and did not differ by the number of total encounters/samples and healthcare setting. The estimated nonadherence in the injection prescribing rate found in our review was higher than the WHO report for the eastern Pacific region (3.2%) [55] and findings of previous systematic reviews (5%) of studies conducted in African countries (54). The patient perception about the efficacy of injection form of medicine as compared to oral, and other health professionals (for example, number of previous clinical encounters) and healthcare facility-related factors (ex. availability of intravenous formulation) could contribute to the inappropriate overuse of injections [2]. The inappropriate use or overuse of injections could predispose individuals to infections such as hepatitis and HIV [8] and unnecessary healthcare costs.

Heterogeneity among the studies such as study area/setting, study design, quality, and the types of prescribing indicators reported are possible limitations of our review. Another possible limitation was that we were unable to evaluate factors contributing to the non-adherence to prescribing practice indicators.

All of the WHO core prescribing indicators are not within the reference value. Therefore, appropriate measures including a multidisciplinary approach to prescribing and involving clinical pharmacists as medical expertise should be considered to lower potentially inappropriate prescribing of medications.

Authors’ Contribution

All authors contributed to data analysis, drafting or revising the article, and gave final approval of the version to be published.

Funding

No funding was received to conduct the study

Competing interests

None

Data Sharing

All data relevant to the study are uploaded as supplementary

Ethical approval

Not applicable

Acknowledgment

The authors acknowledged the staffs of the school of pharmacy, College of health and medical sciences, Haramaya University, Ethiopia.