OPERATIONAL DEFINITIONS

Prescribing: laying down, in writing or otherwise, as a rule or a course of action to be followed by patients.

Prescribing error: a prescribing error is ‘a failure in the treatment writing process that results in an incorrect order about one or more of the normal structure of a prescription’.

Prescription structure: the ‘normal prescription structure’ includes the identity of the recipient, the identity of the drug, the formulation, dose, route, timing, frequency, and duration of administration.

ILLEGAL ACCESSS

ABSTRACT

Introduction: Prescribing errors are too often in general practice in health facilities. They can occur during the process of writing a prescription due to many factors, it could be slips, lapses or mistakes such as unintended omissions in the transcription of the medicine yet they have the potential to result into harming the patient.

Objective of the study: To determine the predicators of prescribing errors in medical practice in public and private health facilities in Kampala City, Uganda.

Methodology: A total of 158 prescribers from public, private not for profit and private for profit health facilities in Kampala city were involved in a cross sectional study. A total of 9 health facilities were selected using simple random sampling. The prescribers were selected using purposive sampling and were required to fill in a structured self-administered questionnaire. Key informants included physicians and pharmacy technicians. Data was collected and analyzed using STATA, Poisson regression model was carried out both at bivariate and multivariate analysis and variables were significant at p-value < 0.05.

Results: Prescribers in the department of obstetrics & gynecology were 0.1 times less likely to make prescribing errors compared to the prescribers in OPD (IRR=0.1; 95%CI=0.0-0.2; p=0.003). Prescribers who always consulted other prescribers when not sure about a certain prescription were 0.2 times less likely to make a prescribing error than prescribers that never consulted (IRR=0.2; 95%CI=0.1-0.9; P=0.35). Prescribers that received on job training on recognizing and avoiding prescribing errors were 0.2 times less likely to make prescribing errors than prescribers that never received any on job training ( IRR=0.2; 95% CI=0.1-0.5; P=0.001). Prescribers who noted that the health facilities where they practiced always retained prescriptions for review were 0.5 times less likely to make prescribing errors than the prescriber’s whose health facilities never retained prescriptions for review. Prescribers who reported working less than 8 hours per day were 0.4 times less likely to make prescribing errors than prescribers who worked more than 8 hours a day (IRR=0.4; 95% CI=0.2-0.9; P=0.033).

Conclusion and Recommendations: The proportion of prescribing errors among prescribers in both public and private health facilities in Kampala is relatively high (47%). The identified predictors for prescribing errors include; type of department, never consulting while prescribing, lack of on job training, not reviewing prescriptions and more than 8 working hours per day. Therefore, prescribers should take prescribing very seriously and endeavor to consult where necessary, busy departments like the outpatient department should have adequate staff, health facilities should introduce several interventions like regular reviews and double checks. The health facilities should encourage working in shifts so that prescribers don’t work for many hours.

CHAPTER ONE

INTRODUCTION

1.0 Introduction

Prescribing is the practice where all registered professionals such as; doctors authorize the use of treatments or medications to patients and also provide instructions on how and when these medications and treatments can be used. These errors are too often in general practice and in health facilities, they can occur during the process of writing a prescription due to many factors and they have the potential to result into harming the patient. During the course of prescribing, an error can occur; it could be slips, lapses or mistakes such as unintended omissions in the transcription of the medicine. Mistakes in selecting an appropriate dose, overlooking transcriptions and unreadable handwritings translate to prescribing errors(Dekker and Nyce, 2013).

The introduction chapter describes the background to the study, statement of the problem, objectives and research questions, significance of conducting the study and the conceptual frame work which is the diagrammatic representation of the study.

1.1 Background to the study

Prescribing errors are common incidences worldwide, with systematic reviews suggesting that as many as 50% of hospital admissions and 7% of medication orders are being affected(Lewis et al., 2009). They are broadly secondary to several factors that contribute to a favorable environment for medication errors to happen. In order to prevent these errors, it is very important to have adequate knowledge regarding the prescribing process and developing strategies to prevent errors at each step in the process. The strategies may include; analysis of medication errors, computerized order entry system, dispensing cabinets that are automated, reconciliation of medication, standardizing the whole medication process and continuous education to the prescribers. Whatever strategy implemented, prescribing errors can only end by developing a culture that encourages reporting of prescribing errors and using a non-punitive approach to manage the errors (BRADY et al., 2009).

Errors in prescribing include irrational, inappropriate, and ineffective prescribing, under prescribing and overprescribing (collectively called prescribing faults) and errors in writing the prescription (including illegibility). Avoiding medication errors is important in balanced prescribing, which is the use of a medicine that is appropriate to the patient's condition and, within the limits created by the uncertainty that attends to therapeutic decisions (Aronson, 2009).

Before the Institute of Medicine reported on prescribing errors in 1999, the American Academy of Pediatrics and its members had been committed to improving the health care system to provide the best and safest health care for infants, children, adolescents, and young adults. This commitment includes designing health care systems to prevent errors and emphasizing the pediatrician's role in this system. Human and device errors can lead to preventable morbidity and mortality (Stucky, 2003).

Canada has seriously tried to correct prescribing errors that were made in the 1990’s when they had inadequate trained numbers of physicians. In the past few years, the government of Canada tried desperately to remedy errors made in the 1990s by training physicians and increasing medical school enrolments. Unfortunately this took several years to take effect, leaving millions of Canadians without timely access to services. To address this challenge some provinces decided that medical regulatory and licensing authorities must offer faster tracks to licensure for international medical graduates (IMGs). They are also legislating expanded scopes of practice for other health professionals, hoping these individuals will provide services traditionally offered by physicians. These high-risk decisions could compromise patients in ways that would make past government errors seem trivial (Rosser et al., 2005).

Insecure working environments, complicated or unclear processes and poor communication among healthcare workers, mainly between the nurses and the doctors have been recognized as serious fundamental predictors for prescribing errors. Several interventions are seriously recommended in order to prevent prescribing errors and these include; focusing on educating and training prescribers and use of on- line aids. Others interventions include using an automated system, use of similar prescribing charts, immediate review of transcriptions by a pharmacist and periodic audits (Basey et al., 2014).

In Malaysia, the need to transform community pharmacy practice has been discussed by all interested parties. However, the transition has been slow due in part to the nonexistence of a dispensing separation policy between pharmacists and medical doctors in private community practices. For decades, medical doctors in private community practices have had the right to prescribe and dispense, thus diluting the role of community pharmacists because of overlapping roles(Shafie et al., 2012).

In South Africa, statistics that have been referred to in the media include reports that Gauteng province alone faces R3.7billion in legal claims for clinical negligence whilst in Kwazulu-Natal the Department of Health has legal claims against it in excess of R2billion. With regards to common errors, reports from the Eastern Cape tell of how medical professionals left surgical sponges or instruments inside patients after surgery. The patients are now suing for damages and compensation. Avoidable medical errors include the 22 HIV-positive women who were involuntarily sterilized in South Africa without their informed consent. The patients later sued the government and were awarded for damages in compensation by a South African Court of Law, the prevalence of the prescribing errors has never been established (Makholw et al.,2014).

The incidence rate of prescribing errors in Nigeria is high (40.5%) and majority of identified errors were related to prescription of incorrect ART regimens and potential drug- interactions; the prescribers are always contacted and the errors resolved in majority of cases. Preventing prescribing errors is feasible in resource-limited settings following a capacity building intervention(Agu et al., 2014).

In Ethiopia prescribing errors in the intensive care unit (ICU) are frequent and lead to patient morbidity and mortality, increased length of stay, and substantial extra costs. The prevalence of medication prescribing errors in other units has not previously been studied(Agalu et al., 2011).

Uganda’s experience in strengthening routine health data reporting through the roll-out of the District Health Management Information Software System version 2 (DHIS2) has resulted in improved timeliness and completeness in reporting of routine outpatient, inpatient and health service usage data from the district to the national level. Continued onsite support supervision and mentorship and additional system/infrastructure enhancements, including internet connectivity, are needed to further enhance the performance of DHIS2. However, little is mentioned on how this will reduce prescribing errors (Kiberu et al., 2014).

The general health sector of Uganda is relatively poor compared to other sectors. According to Dukes and colleagues, one of the challenges Uganda faces is drug prescription errors & this problem lacks enough media coverage. The HMIS which the Uganda health system relies on for data does not capture prescription errors and adverse events. This leaves the patients prone to prescription errors(Dukes et al., 2014).

A study conducted by Govule and colleagues in public health facilities in Kampala revealed that 58% of all medical errors were prescribing errors; however predictors for the errors were not established (Govule et al., 2015). It’s upon this background that a study on the predictors of prescribing errors among prescribers in both public and private health facilities in Kampala city was conducted.

1.2 Statement of the problem

In a bid to save the many numbers of patients in health facilities in Kampala City, prescribers need to be fast and sometimes provide healthcare under pressure which subjects them to committing prescribing errors as noted by Dr Joaquim Saweka (WHO representative for Uganda). Ideally patient safety is an essential discipline complete with an integrated body of knowledge and expertise to improve health care, so no one would be expecting to observe prescribing errors at a health facility level (WHO, 2014). Prescribing errors can be prevented but surprisingly they keep happening from time to time, unfortunately most of them pass unnoticed (Sandars and Langlois, 2005).

Though data gathered clearly shows that too often patient management is far from ideal (diseases are misdiagnosed or missed, incorrect or incomplete treatment regimens are prescribed and the patients are inadequately counselled on how to adhere to correct treatment) the government of Uganda provides clinical guidelines with standard prescribing protocols to minimize errors in medical practice (MOH- Uganda, 2010).

Despite all the above advances in medicine & healthcare, health facilities have remained places where patients are getting harmed due to prescribing errors. If the problem is not resolved patient safety will be compromised leading to complications, drug resistance and premature death.

1.3 Objectives of the study

1.3.1 Broad Objective

To determine the predicators of prescribing errors in medical practice in public and private health facilities in Kampala City, Uganda in the period of July- September 2016.

1.3.2 Specific objectives

i. To determine the prevalence of prescribing errors in medical practice in public and private health facilities in Kampala City in the period of July- September 2016.

ii. To identify the nature of prescribing errors as perceived by prescribers in medical practice in public and private health facilities in Kampala City in the period of July- September 2016.

iii. To assess the prescriber related factors influencing prescribing errors in medical practice in public and private health facilities in Kampala City in the period of July- September 2016.

iv. To establish the institutional factors influencing prescribing errors in medical practice in public and private health facilities in Kampala City in the period of July- September 2016.

1.4 Research questions

i. What is the prevalence of prescribing errors in medical practice in public and private health facilities in Kampala City in the period of July- September 2016?

ii. What is the nature of prescribing errors as perceived by prescribers in medical practice in public and private health facilities in Kampala City in the period of July- September 2016?

iii. What are prescriber-related factors influencing prescribing errors in medical practice in public and private health facilities in Kampala City in the period of July- September 2016?

iv. What are the institutional factors influencing prescribing errors in medical practice in public and private health facilities in Kampala City in the period of July- September 2016?

1.5 Significance of the study

The findings from this study may serve as a useful remainder to all prescribers in Kampala public health facilities and to all prescribers in other regions in the country to be keen on the possibility of error occurrence during prescribing so that they avoid the reoccurrence of the most common prescription errors.

In addition to that, the results from this study may be very helpful to the general public if well utilized by responsible departments in the public health facilities to devise means of preventing similar occurrences of prescription errors, thereby protecting the patients from adverse events resulting from prescription errors such as poor health of the population, drug resistance, complications, increased morbidity and mortality and low productivity.

More to that, the findings from this study may benefit the entire nation, if utilized by policy makers at M.o.H, Uganda to formulate policies that will protect citizens from adverse events resulting from prescription errors.

Last but not least, different scholars may be able to compare and contrast their findings to those of this study and generate new knowledge for future researchers in the same field.

1.6 Conceptual framework

PRESCRIBING ERRORS

Independent variables Dependent variable

Prescriber perceived errors

· Unclear dose

· Drug name unclear

· Wrong dose

· Unreadable prescription

· Unclear abbreviation

· Others

Prescriber related factors

· Period of practice

· Nature of department of a prescriber

· No. of working hours per day

· No. of patients seen per day

Institutional factors

• Computerized prescriptions

• Double checks

• Signing on the prescription

• Standard prescribing procedures

• Review of prescription

• Training

• Reporting system

Outcome

• Poor health of the population

• Drug resistance

• Complications

• Increased morbidity and mortality

Adopted from Donabedian concept of medical errors and individual errors (Donabedian, 1980)

Narrative of the framework

Prescription errors may include; unclear dose, drug name unclear, wrong dose, unreadable prescription, unclear abbreviation and many others. There are several predictors for prescription errors in general practice in health facilities and these include; prescriber related factors such as; nature of department, period of practice, number of working hours per day, number of patients seen per day and consulting during the prescribing process.

Institutional factors influencing prescribing errors in health facilities could be in the form of prescriptions, double checks, and training, signing on the prescriptions, reviewing prescriptions and standard prescribing procedures and reporting systems.

CHAPTER TWO

LITERATURE REVIEW

2.0 Introduction

This chapter presents reviewed literature of similar studies according to the objectives; the prevalence of the prescribing errors, the nature of prescribing errors, the prescriber related factors influencing prescribing errors and the institutional factors influencing prescribing errors. A summary of the reviewed literature is also provided in this chapter.

2.1 The prevalence of prescribing errors in medical practice in health facilities

To describe the prevalence, type, and factors associated with prescribing errors in a pediatric emergency department, protocols and clinical guides were used and the results were: In 377 of 1906 checked reports, some treatments were prescribed. A total of 92 errors (15%) were detected and all of them were prescription errors: 50 (8%) for inappropriate indication and 42 (7 %) for inadequate dose. Also, 87 were considered insignificant errors, 5 were moderate and none were severe. In the weekends and holidays, more errors were committed compared in weekdays (28% vs 18 %, P = 0.02). Between 24 and 8 hours, more errors were registered than between 8 and 16 and between 16 and 24 hours (32.3% vs 17.9% vs 21.2%; P = 0.03). The high assistance pressure during weekends and holidays and the tiredness during the night were risk factors of prescribing errors (Vilà-de-Muga et al., 2011).

A total of 194 error reports related to prescriptions were abstracted and analyzed using a medication error coding tool, the main outcome measures were type, severity and preventability of medication errors and associated adverse drug events. The results were as follows; 126 (70%) of the medication errors were prescribing errors, 17 (10%) were medication administration errors, 17 (10%) documentation errors, 13 (7%) dispensing errors. The severity of harm from reported errors were: prevented and did not reach patients, (72, 41%), reached patients but did not require monitoring (63, 35%), reached patients and required monitoring (15, 8%), reached patients and required intervention (23, 13%) and reached patients and resulted in hospitalization(Kuo et al., 2008).

A pre-intervention and post-intervention cross-sectional study was conducted of a sample of prescriptions that were ordered by physicians and medications that were administered by nurses to patients at the Hospital Italiano de Buenos Aires Department of Pediatrics in 2002 and2004. The prevalence of prescribing error rate in the second phase was 7.3% (199 of 2732) and 11.4% (201 of 1764) in the first phase. The risk difference was _4.1%. the researcher concluded that, the development of a program mainly centered on the promotion of a cultural change in the approach to prescribing errors would effectively diminish medication errors in neonates and children(Otero et al., 2008).

A systematic literature search for studies that examined the incidence or cause of medication error in one or more stage(s) of the medication-management process in the setting of a community or hospital-based mental healthcare service was undertaken. The results revealed that all studies examined medication management processes, as opposed to outcomes. The reported rate of error was highest in studies that retrospectively examined drug charts, intermediate in those that relied on reporting by pharmacists to identify error and lowest in those that relied on organizational incident reporting systems. Only a few of the errors identified by the studies caused actual harm, mostly because they were detected and remedial action was taken before the patient received the drug (Maidment et al., 2006).

In Dean Delphi’s study in a single health facility involving 41 doctors in UK, 538 errors were detected out of 36,168 prescriptions made giving a prevalence of prescribing errors of 1.5%. A total of 98% of the prescribing errors were made by the junior staff (Dean et al., 2000).

A prospective observational study involving 21,464 adults and children admitted in a US tertiary care teaching health facility and 840 physicians of which 378 were residents was conducted to quantify prescribing errors made by physicians. All prescriptions received were examined by centralized pharmacists before being dispensed and all possible prescribing errors were noted and discussed with all prescribers. The researcher defined a prescribing error as; wrong patient, wrong medicine, in accurate dosage, in correct frequency or prescriptions with vital missing information. Results indicated that out of 289,411 prescriptions, 905 had errors i.e. 3.1/1000 prescriptions. A total of 95.7% of the errors were committed by the junior staff (Lesar, 2002).

Frey and colleagues carried out a retrospective observational study in one children’s ICU in Australia involving total of 202 patients and all prescribers. The aim of the study was to evaluate the incidence, nature and implication of prescribing errors. The pharmacist at the ICU ward studied all prescriptions and recorded all the errors that had been established and characterized by the consultant physician. The results indicated that a total of 4.48% of the prescriptions were identified as having prescribing errors of which 87% were made by junior staff (Frey et al., 2002).

A study was conducted in a 580 bed Canadian tertiary care teaching hospital, where patients and doctors were the subjects for the study. Data was collected for a period of 25 weeks on the number, times, origin, and effect of the prescribing error and any prescription that had an omission or any inaccuracy was captured as a prescribing error. For the errors that had earlier on been detected by the staff in the pharmacy were not included in the study. The findings indicated that from a total of 237,798 prescriptions, 0.5% of them had prescribing errors by doctors (Tam et al., 2008).

In a university affiliated teaching hospital, a study involving adults on admission on surgical and medical wards and 57 doctors working on the respective wards was conducted to determine the prevalence of prescribing errors. Any prescribing error identified by the pharmacist was recorded and at the end of the study and 177 prescribing errors were recorded out of 8195 prescriptions giving a prevalence of prescribing errors of 2.2% (Hendey et al., 2005).

Haw and colleagues conducted a retrospective observational study in eight acute mental health trust health centers and one independent psychiatric center. Both consultants and non-consultants were involved in the study and the results revealed that a total of 63% (329/523) of the prescribing error were made by the non-consult (Haw et al., 2007).

In an Urban teaching health facility, 352 patients on admission were involved in a study to determine predictors for prescribing errors which were defined as errors as result of ordering, transcribing, dispensing, administering and monitoring. The results indicated that from a total of 6916 prescriptions, 104 were found prospective observational study in an outpatient family medicine health center, 20 family practice residents containing errors (Sard et al., 2008).

In a pediatric hospital emergency department, a study involving patients 1532 patients attending the department and 80 doctors was carried out and the nature of prescribing errors was; Wrong dose, wrong route of administration, wrong timing or wrong units. The junior doctors were 1.5 times more likely to commit an error than senior doctors, this would calculate as 162 errors of the total 271, written in 154 charts (2.1 errors made by junior staff per 1000 charts written by all doctors(Kozer et al., 2002).

A systematic evaluation involving all physician and all staff pharmacists involved in routine review of medication orders was conducted and results revealed that the most common specific factors associated with errors were decline in renal or hepatic function requiring alteration of drug therapy (97 errors, 13.9%), the most common groups of factors associated with errors were those related to knowledge and the application of knowledge regarding drug therapy (209 errors, 30%). The researcher concluded that; several easily identified factors are associated with a large proportion of medication prescribing errors. By improving the focus of organisational, technological, and risk management educational and training efforts using the factors commonly associated with prescribing errors, risk to patients from adverse drug events should be reduce(Kozer et al., 2006).

2.2 The nature of prescribing errors as perceived by prescribers in medical practice

A total of thirty four physicians, surgeons, pharmacists, nurses and risk managers were involved in a study to form prescriber description of a prescribing error to be used in quantitative studies of the incidence of prescribing errors. The results revealed that 88% of the participants came to a consensus that prescribing errors involve transcription errors, not communicating useful information, use of inappropriate drug or dose for individual patients. They all agreed that deviating from policies or guidelines were not prescribing errors. The researchers concluded that those agreed classification of errors can be used to compare prescribing error rates among different hospitals I both research and clinical governance initiatives (Dean et al., 2000).

A retrospective cross sectional study involving senior doctors (clinicians) and junior doctors (clinician assistants) was conducted in 2 teaching hospitals in Netherlands with the aim of exploring an epidemiological framework to assess the causes of prescribing errors. Prescribing errors considered as errors in the dosage, therapeutic, unreadable prescriptions, incomplete data, and unclear name of medicine wrong medicine, incorrect route and unrecognized abbreviations. Results showed that 23.5% of the prescribing errors were written by the doctors where junior doctors were 1.57 times more likely to write a prescribing error than the senior doctors (Fijn et al., 2002).

A prospective observational study involving 21,464 adults and children admitted in a US tertiary care teaching health facility and 840 physicians of which 378 were residents was conducted to quantify prescribing errors made by physicians and to determine the risk factors associated with such errors. All prescriptions received were examined by centralized pharmacists before being dispensed and all possible prescribing errors were noted and discussed with all prescribers. The researcher defined a prescribing error as; wrong patient, wrong medicine, in accurate dosage, in correct frequency or prescriptions with vital missing information. Results indicated that out of 289,411 prescriptions, 905 had errors i.e. 3.1/1000 prescriptions. A total of 95.7% of the errors were committed by the junior staff (Weingart et al., 2000).

A retrospective observational study in one children’s ICU in Australia involving total of 202 patients and all prescribers was carried out. The aim of the study was to evaluate the incidence, nature and implication of prescribing errors. The pharmacist at the ICU ward studied all prescriptions and recorded all the errors that had been established and characterized by the consultant physician. The results indicated that a total of 4.48% of the prescriptions were identified as having prescribing errors of which 87% were made by junior staff. This study defined prescribing errors as wrong patient, wrong medicine, wrong dosage, and wrong frequency and incompatibility or not recognizing a drug to drug interaction (Wong et al., 2004).

2.3 The prescriber related factors influencing prescribing errors in medical practice

A Prospective observational study in a 800-bed teaching hospital with catchment population of 320 000 was conducted on four wards admitting general medical emergencies the junior doctors involved were 13and the identified prescription errors were; failure to take into account pharmaceutical issues (intravenous drug incompatibilities, drug interactions, contraindications, lack of monitoring of drug or patient parameters); failure to communicate essential information (such as omissions in medication history taking); use of drugs or doses inappropriate for the patient. The researcher proposed a clinical teaching pharmacist programme to improve prescribing skills among newly qualified prescribers (Webbe et al., 2007).

A study was conducted in an acute medical unit of a teaching hospital to investigate the way prescribers obtained details from patients before admitting them and prescribing to assess the likelihood of any potential medical errors. Data was collected for a period of four months where the process of admitting a patient was carefully observed. The observed findings indication that only 45.5% used one source, 58.3% of the prescribers confirmed the medication with patients yet the observations showed that only 16% of the prescribers that confirmed medications with their patients. 97% of the cases were able to discuss medication with the prescribers. A total of 14 patients were not asked any question relating to medication. From a total of 688 reviewed chats, 46.2% had prescription errors with 86.6% involving omission of a medicine. The researcher concluded that although prescribers know the importance of obtaining correct information and the importance checking prescriptions with patients, they however, fail to practice hence resulting into prescribing errors (Wanzel et al., 2000).

To examine the nature, frequency and potential severity of prescribing errors in UK mental health units, a survey of errors detected by pharmacy staff in nine NHS trusts was conducted. In total, 523 errors meeting the study definition were detected (2.4% of prescription items checked). Prescription writing errors (77.4%) were most common, while decision-making errors accounted for 22.6% of errors. In 280 (53.5%) cases the prescribed drug had been administered before the error was detected. Most errors were of doubtful or minor importance but 22 (4.3%) were deemed likely to result in serious adverse effects or death. The researcher concluded that pharmacy staff had an important role to play in their management(Nath and Marcus, 2006).

A review was done to outline the methods used, highlight their strengths and limitations, and summarise the incidence of prescribing errors reported. Methods used were retrospective or prospective and based on process or on outcome. Reported prescribing error rates varied widely, ranging from 0.3% to 39.1% of medication orders written and from 1% to 100% of hospital admissions. Unfortunately, there was no standard denominator for use when expressing prescribing error rates. It could be argued that the most meaningful is the number of medication orders written; however, it is also helpful to consider the number of medication orders written per patient stay in order to understand the risk that a given prescribing error rate poses to an individual patient. Because of wide variation in the definitions and methods used, it is difficult to make comparisons between different studies(Evans et al., 2007).

Each method for identifying prescribing errors has advantages and disadvantages. Process-based studies potentially allow all errors to be identified, giving more scope for the identification of trends and learning opportunities, and it may be easier to collect sufficient data to show statistically significant changes in prescribing error rates following interventions to reduce them. However, studies based on process may be criticised for focusing on many minor errors that are very unlikely to have resulted in patient harm. Focusing instead on harm, as in outcome-based studies, allows efforts to reduce errors to be targeted on those areas that are likely to result in the highest impact. Therefore, the most appropriate method depends on the study’s aims. However, using a combination of methods is likely to be the most useful approach if comprehensive data are required (Franklin et al., 2005).

An examination of questionnaire studies, the literature on reports of sleep loss, studies of the reduction of work hours on performance as well as observational and a few interventional studies have yielded contradictory and often equivocal results. The resident doctors generally find they feel better working fewer hours but improvements in patient care are often not reported or do not occur. A change in the attitude of the resident toward his role and his patient has not been salutary. Decreasing sleep loss should have had a positive effect on patient care in reducing medical error, but this remains to be unequivocally demonstrated(Kramer, 2010).

2.4 The institutional factors influencing prescribing errors

In 2002, the UK Department of Health and the Design Council jointly commissioned a scoping study to deliver ideas and practical recommendations for a design approach to reduce the risk of medical error and improve patient safety across the National Health Service (NHS). Despite the multiplicity of activities and methodologies employed, what emerged from the research was a very consistent picture. This convergence pointed to the need to better understand the healthcare system, including the users of that system, as the context into which specific design solutions must be delivered. Without that broader understanding there can be no certainty that any single design will contribute to reducing medical error and the consequential cost thereof (Clarkson et al., 2004).

Shaughnessy and colleagues carried out a study among 20 family practice residents in their first, second and third year of training. Main or slight oversights of serious information, dosage or route error, failure to meet legal requirements, nonprescription medicine, uncertain quantity, or unfinished charts were all considered as errors. A total 21% of prescriptions contained errors before the intervention and after the intervention prescriptions with errors reduced to 17% (O’Shaughnessy et al., 2008).

Two years later, a similar study was conducted at the same health center but this time the subjects were 12 family practice residents in their first and second year of training. The educational program consisted of an evaluation and feedback on prescription writing, before the program 14.4% of the prescriptions contained errors and after the educational program, the prescribing errors reduced to 6%. Indicating that educational programs on prescribing can prevent errors from occurring(O'Shaughnessy et al., 2010).

Anton and colleagues conducted a prospective observational study in a renal unit of a teaching hospital in UK. The study included; 257 renal patients, and 42 doctors of which 9 were consultants, 13 were registrars, 6 were SHOs and 14 were all prescribing in the renal unit. Warnings produced via an electronic prescribing system (computers) were utilized as a proxy measure for possible errors. Results revealed that there were 5462 warnings of which 2642 reflected true errors, giving a prevalence of 51.4% of prescribing errors. This clearly illustrated that electronic warning can be effective intervention measures in preventing prescribing errors (Anton et al., 2004).

A study was conducted in a teaching hospital in US to assess whether the introduction of computer- aided prescribing would reduce prescribing errors in an emergency department. Any prescription that needed a pharmacist to clarify, and all those with missing information, in accurate dose, not readable, wrong and those with non-formulary drugs was considered as prescribing errors. Prescribers were provided with computers and they were expected to make computer aided prescription writing. Before computers were provided the prevalence of prescribing errors was 2.5% and after the intervention the prevalence of prescribing errors decreased to 0.54% (Bizovi et al., 2002).

To evaluate the proportions, nature and severity of prescribing errors among outpatients, a prospective cohort study involving 1202 adult outpatients was carried out in four primary care health facilities in Boston. The results revealed that prescribing errors were detected in 7.6% of the patients, 62% of the prescription errors had a potential to cause injury, 2% of the prescription errors had a life threatening potential. Medication rates and use of computers were not statistically associated to prescribing errors. It was also noted that 95% of the prescribing errors would have been prevented by doing advanced checks. The researcher concluded that 7.6% of the prescription errors that harmed patients would have been prevented by having more advanced systems with dosage and frequent checks (Forster et al., 2005).

At a US family practice health facility, the effect of an educational intervention for all first year residents was studied. Residents were provided in service training in writing prescription and reviewing common prescribing errors. The study considered misplaced prescription information, incorrect dose or frequency, uncertain quantity or instructions, and failure to comply with regulation as prescribing errors. The results revealed that 17.9% of the prescribing errors were detected before the intervention and 19.7% after the intervention. This implies that training is very essential (Howell et al., 2003).

A study was conducted to determine whether the use of a computerized bar code–based blood identification system resulted in a reduction in transfusion errors or near-miss transfusion episodes. The results indicated that a total of 388,837 U were transfused during the 2002-2005 period. There were 6 misidentification episodes of a blood product being transfused to the wrong patient during that period (incidence of 1 in 64,806 U or 1.5 per 100,000 transfusions; 95% CI, 0.6-3.3 per 100,000 transfusions). The researcher concluded that the institution of a computerized bar code–based blood identification system was associated with a large increase in discovered near-miss events(Nuttall et al., 2013).

Bates, (2003) reveals that highly unstable critically ill patients are more vulnerable to medication errors, and the risk of errors is increased in these patients because of the number of drugs they receive. Therefore, reducing errors is crucial to improving patients' outcomes. Information technology and automated systems have been introduced to improve the medication process. For example interactive knowledge based prescription systems which are computer-controlled dispensing units providing secure prescription in care units as seen in this study. These systems have improved medication use in medical units, with an impact on administration time errors, omissions, and work activities (Bates and Gawande, 2003).

To determine whether indication-based computer order entry alerts intercept wrong-patient medication errors at an academic medical center serving inpatients and outpatients, Galanter and colleagues developed and implemented a clinical decision support system to prompt clinicians for indications when certain medications were ordered without an appropriately coded indication on the problem list. The results indicated that over a 6-year period, 127 320 alerts fired, which resulted in 32 intercepted wrong-patient errors, an interception rate of 0.25 per 1000 alerts. Neither the location of the prescriber nor the type of prescriber affected the interception rate. No intercepted errors were for patients with the same last name, but in 59% of the intercepted errors the prescriber had both patients' charts open when the first order was initiated(Galanter et al., 2013).

To investigate the impact of different measures, implemented by clinical pharmacists, on prescribing error rates in a pediatric intensive care unit in Cairo, Egypt, a pre-post study of prescribing errors in a 12 bed hospital was conducted. Results indicated that; of pre-intervention orders, 1107 (78.1%) had at least one prescribing error. The intervention resulted in significant reduction in prescribing error rate to 35.2% post-intervention (p < 0.001). The intervention resulted also in a significant reduction in the rate of potentially severe errors from 29.7% pre-intervention to 7% post intervention (p < 0.001) and the rate of potentially moderate errors from 39.8% pre-intervention to 24.2% post-intervention (p < 0.001). Besides, rates of all types of prescribing errors were declined to different degrees as a result of the intervention (Alagha et al., 2011).

Summary of the literature review

Although prescribing errors is a worldwide issue, the majority of the studies on this topic have been carried out in developed nations such as North American and European countries, while the issue has been relatively neglected in Asia and Africa. Research about prescribing errors in Africa is limited. In particular, very little is known about the incidence of error in hospital settings or about the harm caused by it. In Africa people get care from non-hospital settings and the extent of harm is not well described. Evidence is available from other sources that a substantial number of adverse drug events are caused by mainly prescribing errors. Some of these are preventable and might probably, therefore, be due to processes in the respective health establishment and features of the organization. If much on prescribing errors is not established and reported, the healthcare given might predispose clients to several harmful complications, therefore priorities for future research are suggested especially in Africa and Uganda in particular.

CHAPTER THREE

METHODOLOGY

3.0 Introduction

This chapter describes the methods of how the study was carried out. It includes the study area, study scope, study design, sources of data, study population, inclusion and exclusion criteria, sample, sample size determination, sampling unit, sampling procedure, study variables, data tools, how the quality of data was maintained, plan for analysis, ethical considerations and limitations of the study.

3.1 Study Area

Kampala city is the capital of Uganda where most of the economic activities take place. It is surrounded by the fast growing Wakiso district with a population of close to 2 million and population density of 159 per sq.km. Kampala city is near the banks of Lake Victoria, before the British, it was called Buganda kingdom. The main economic activity is trade and industry. The city is comprised of several health facilities, both private and public. The national referral hospital (Mulago) is situated in Kampala city. The city was made an authority five years ago, now known as Kampala Capital City Authority (KCCA) and it manages the city on behalf of the central government.

3.2 Study Scope

The study was carried out at only nine health facilities in Kampala city, each sector (private for profit, private not for profit and public) three health facilities were considered; Kampala Hospital, AAR Healthcare and Kololo Hospital (private for profit), Mengo Hospital, Nsambya Hospital and Kibuli Hospital (private not for profit), Kisenyi health center IV, Kisugu health center III and Kiswa health center III (public).

3.3 Study Design

This was a cross sectional study employing both quantitative and qualitative data collection techniques. This kind of design was best suited to determine the predictors for prescribing errors in Kampala city since predictors and prescribing errors were determined at a point in time. The advantage with this study design is that; it is very inexpensive, it provides results in the shortest time possible and it is easy to conduct making it best suited for the study.

3.4 Sources of Data

Primary data

Data was collected primarily from prescribers (consultants, medical officers and clinical officers) at the selected health facilities for a period of three months. The key informants (pharmacists and pharmacy technicians) also provided primary data.

Secondary data

Secondary data was utilized in the discussion section and this secondary data was gotten from journals published articles and internet sources.

3.5 Target population

The target population comprised of prescribers (consultants, medical officers and clinical officers) in Kampala city. These were both males and females employed in public, private for profit and private not for profit health facilities situated in Kampala city.

3.6 Study Population

The study population comprised of consultants, physicians, medical officers and clinical officers from Mengo Hospital, Nsambya Hospital, Kibuli Hospital, Kisenyi health center IV, Kisugu health center III, Kiswa health center III, AAR, Kololo Hospital and Kampala Hospital that were present during the period of July to September, 2016.

3.7 Inclusion and Exclusion Criteria

3.7.1 Inclusion criteria

A respondent was included in the study only if he or she were prescribers at the selected health facilities and had been prescribing for at least more than six months.

3.7.2 Exclusion criteria

All prescribers at the selected health facilities that had joined the medical profession less than 6 months ago and those that were very busy with patients during the study period were excluded from the study.

3.8 Sample Size Determination

The total number of participants to be included in the study was determined using Yamane simplified formula. This formula was chosen because it is ideal for finite population (since the number of prescribers in the selected health facilities is known).

The formula is;

n = N/1 +N (e²)

Where n = desired sample size

N = total population of all prescribers (consultants, medical officers and clinical officers) at the selected health facilities.

e = margin of error at 95% level of significance, which is 0.05

N= 260 prescribers (established from the respective health facility data base).

Substituting in the formula above

n = 260/1+ 260*0.05²

n = 158 prescribers in Kampala city

Therefore, the desired sample size is158 prescribers in Kampala city

3.9 Study Unit

The study unit was a prescriber, (a consultant, medical officer or clinical officer) at any of the selected 9 health facilities in Kampala city.

3.10 Sampling Procedure

3.10.1 Selection of health facilities in Kampala

Private not for private health facilities (Mengo, Kibuli and Nsambya) Hospitals were purposively selected based on the fact that they were the only private not for profit health facilities in Kampala city.

The other health facilities were selected using a probability sampling method (simple random sampling) this sampling method was used to give an equal chance for all the health facilities to be selected. A list of all public health facilities in Kampala city was obtained and each health facility was noted on a single sheet of paper and folded, the folded papers were all put in one box and shuffled several times, then one research assistant was asked to pick only three sheets of folded papers from the box. The selected public health facilities were Kiswa health center III, Kisenyi health center IV and Kisugu health center III. The same procedure was used to select the private for profit health facilities and the three health facilities selected were; AAR Healthcare Ntinda, Kololo and Kampala Hospitals.

3.10.2 Selection of prescribers for quantitative data

A total of 158 prescribers (consultants, medical officers and clinical officers) were selected using a stratified method of sampling to calculate a proportional number of prescribers to represent a particular health facility as illustrated below;

Table 1: Proportional number of prescribers to represent a particular health facility

S/N	Health facilities	Number of prescribers (a)	Desired sample size (c) *c =( a/b) n**
1	Mengo Hospital	54	32
2	Nsambya Hospital	86	52
3	Kibuli Hospital	24	15
4	Kiswa Health center iii	05	3
5	Kisugu Health center iii	05	3
6	Kisenyi Health center iv	10	6
7	Kololo Hospital	16	10
8	AAR Healthcare	23	14
9	Kampala Hospital	37	23
Total others	9 Jumbo Clinic Ssekungu Rubaga Hospital MakerereUniversity Hospital Sidina Hospital Fieca Pharmacy Abi Clinic Kimera Clinic Mukama Clinic Kadic Clinic	260 (b)	158 (n)

In each of the health facility, a list of prescribers was obtained and used to systematically select the K^th number and the prescribers possessing that number were selected to participate in the study. If a respondent selected was not present to answer the questionnaire on two different occasions or not willing to participate, the same procedure was repeated to select another respondent.

3.10.3 Selection of participants for qualitative data collection

A total of 9 pharmacists/ pharmacy technicians each from the selected health facility were purposively selected as key informants. They were chosen based on the fact that they receive most of all the medical prescriptions so they are familiar with the nature of prescribing errors in the health facilities.

3.11 Study Variables

3.11.1 Dependent variable

Prescribing errors measured as a binary outcome (made prescribing errors and did not make prescribing errors).

3.11.2 Independent variables

The following independent variables were measured to determine the predictors for prescribing errors in the health facilities in Kampala city.

1. Prescriber perceived errors

· Unclear dose

· Drug name unclear

· Wrong dose

· Unreadable prescription

· Unclear abbreviation

2. Prescriber related factors

· Nature of department

· Period of practice

· Number of patients seen per day

· Number of working hours

· Consulting during the prescribing process

3. Institutional factors

· Computerized prescriptions

· Double checks

· Signing on prescriptions

· Standard operating procedures

· Reporting system

3.12 Data collection tools

Quantitative tools

A semi- structured self-administered questionnaire with both open and closed ended questions was used for quantitative data of the sampled prescribers.

Qualitative tools

A key informant guide was used to collect qualitative data from purposively selected pharmacists and pharmacy technicians based on the fact that receive medical prescriptions on a daily basis.

3.13 Description of the tools

A semi- structured researcher administered questionnaire consisted of 22 questions that sampled prescribers answered and where necessary they filled the answer in the space provided. The questionnaire had four parts; part one consisted the health facility information, section two consisted the prescribing errors, part three consisted questions relating to prescriber factors and part four consisted questions relating to institutional factors.

A key informant guide consisted of 3 questions for the selected key informants to answer. The questions were basically designed to capture the nature of errors encountered, the prescriber related factors and the institutional factors influencing prescribing errors among prescribers.

3.14 Quality Control

The quality of data was ensured by taking into account the following measures;

Reliability of the instrument

To ensure that the research instrument was able to provide the same results repeatedly each time it was used, the data collection instrument was pretested among 20 respondents from international hospital Kampala (IHK) and the results were subjected to alpha reliability which is internal test reliability. IHK was chosen because it is not the study population but it possesses similar characteristics as the study population.

The researcher used Respondent debriefing technique of pretesting in order to establish whether there were any misunderstandings of terms or phrases used in the data collection tool, to ascertain the extent to which respondents’ understandings of questions and concepts were consistent with official definitions, to evaluate whether some questions in the main survey were superfluous, to examine whether alternate versions of a question did a better job of identifying or measuring specific activities and to construct comparable subsets of respondents from different questionnaire versions to allow comparative analyses. According to Esposito et al., 1993, respondent debriefing is the best pretesting technique for self-administered questionnaire

Pre- visit to the study area

A pre- visit was conducted to the study area (9health facilities) to identify the number of prescribers in each health facility.

Training of research assistants

Nine research assistants with adequate knowledge on prescribing (dispensers) were recruited and trained on how to collect data.

Editing of data

Editing of data and corrections were done immediately at the end of each data collection day to rule out any missing data. Double data entry was done using Epi data version 3.1 and cleaning was done to reduce chances of errors made during the entry.

3.15 Data Analysis

Exploratory analysis was done on all variables to detect any missing and any data inconsistencies. Descriptive analysis was also done on all variables to determine the proportions and for all continuous data variables, the means and their standard deviations (SD) were calculated.

The outcome of the study was prescribing errors which were measured as a binary outcome. Poisson regression model was used to estimate the incidence risk ratios (IRR) and their 95% confidence interval for the prescribing errors comparing them with the independent variables and robust standard errors were estimated. The Poisson regression model was used because it is best suited for determining the relationship of a dependent variable with multiple independent variables when the outcome of interest (the prescribing errors) is more than 10% (47%).

All variables in the bivariate analysis with p<0.15 or potential confounders were included in the multivariate analysis. All statistical analyses used Stata version 12. Data was presented in tables and figures as shown in the results section.

Data management and analysis of qualitative data

Data collected from key informant interviews was transcribed from the audio recordings. Further analysis was done using coded word processed text organized and analyzed using content and factor analysis with Atlas/ti software.

Data was divided into meaningful analytical units and marked with descriptive words. The codes were merged into lager categories and themes. Content from each coded groups were summarized and illustrated with direct quotes from the discussion. A 10% back translation was done for quality control.

3.16 Ethical Consideration

Permission and approval to conduct the study was obtained from International Health Sciences University and from the in charges and medical directors of the respective health facilities.

Informed consent was obtained from respondents after explaining adequately the aim, procedures and anticipated benefits of the study. It was also explained to the study participants that their participation was voluntary with no payment involved and they were free to withdraw consent at any time during the study.

Respect for prescribers was observed, those who were not ready to be interviewed at a particular time were scheduled for another day. Confidentiality was also maintained throughout the study period by using serial numbers as opposed to prescribers’ names and no participant was harmed during the study.

3.17 Plan for dissemination of a report

A report of findings was submitted to International Health Sciences University (IHSU) in partial fulfillment of a master’s degree of Science in public health. A copy was submitted to the respect health facilities.

3.18 Limitations of the study

Data was retrospectively obtained from respondents, so recall bias could have influenced the results. Because of sampling, the population sampled may not have represented all the views of prescribers in Kampala city.

CHAPTER FOUR

RESULTS

4.0 Introduction

This chapter presents the quantitative and qualitative results according to the study objectives; the prevalence of prescribing errors, the nature of prescribing errors as perceived by prescribers, the prescriber related factors influencing prescribing errors and the institutional factors influencing prescribing errors in medical practice in public and private health facilities in Kampala City in the period of July- September 2016.

4.1 Descriptive Analysis

This section provides descriptive statistics (frequencies and percentages) of a sample of 158 prescribers practicing in the 9 selected public and private health facilities in Kampala city.

4.1.1Respondents per health facility level and type

The vast majority 83.5% (132/158) of the prescribers were from hospitals and the highest proportion 62.7% (99/158) of the prescribers were from private for profit health facilities as shown in table 2 below;

Table 2: Respondents per health facility 1evel and health type

Variable	Frequency (N=158)	Percentage (%)
Health facility level
Hospital	132	83.5
Health center iii	6	3.8
Health center iv	6	3.8
Clinic	14	8.9
Health facility type
Public	12	7.6
PNFP	99	62.7
PFP	47	29.7

Source: primary data

4.1.2 Characteristics of respondents

4.1.2.1 Age distribution

The age of prescribers was normally distributed with mean age being 32.7 years and standard deviation of 6.7 years as illustrated in figure 1 below;

Figure 1: Histogram showing the age distribution of respondents

4.1.2.2 Other characteristics of respondents

More than half (91/158) of the respondents were male and the highest proportion of respondents 46.2% (73/158) was medical officers.

Table 3: other characteristics of respondents

Variable	Frequency (N=158)	Percentage (%)
Sex
Male	91	57.6
Female	67	42.4
Professional cadre
Consultants	30	19
Physicians	32	20.3
Medical officers	73	46.2
Clinical officers	23	14.6

Source: primary data from prescribers

4.1.3 Prevalence of prescribing errors

From figure 2 below; 47.5% (75/158) of the prescribers reported to having made a prescribing error at one point in time.

Figure 2: Pie chart showing the prevalence of prescribing errors

Source: primary data from respondents

4.1.3.1 Prescribing errors per health facility level and type

There were no prescribing errors reported in health center iii whereas more than half (73/131) of the prescribing errors were made in hospitals.

Table 4: Prevalence of prescribing errors per health facility level and type

Variable	Made prescribing errors(N=158)
	Yes Frequency (%)	No Frequency (%)
Health facility level
Hospital	73 (55.7)	59 (44.3)
Health center iv	2 (13.3)	6 (86.7)
Health center iii	0 (0.0)	4 (100.0)
Clinic	0 (0.0)	14 (100.0)
Health facility type
Public	2 (16.7)	10 (83.3)
PNFP	53 (53.5)	46 (46.5)
PFP	20 (42.6)	27(57.4)

Source: primary data

4.1.4 The nature of prescribing errors as perceived by prescribers

Prescribers were asked to report any of the types of prescribing errors they have encountered during their period practice of practice, they were asked to report as many types they have encountered and a total of 93 prescribers reported drug name being unclear as the most common encountered prescribing error, the other prescribing errors identified were; wrong dose (identified by 72 prescribers), unreadable prescriptions (identified by 65 prescribers), unclear dose (identified by 64 prescribers) and finally unclear abbreviations (identified by 55 prescribers) as indicated in figure 3 below;

Figure 3: Bar graph showing the nature of prescribing errors

Source: primary data from respondents

Key informants (KI) were asked to describe the nature of prescribing errors that they had ever encountered. All the key informants (pharmacists and pharmacy technicians) identified unreadable prescription as the most common prescribing error ever encountered followed by unclear drug name as illustrated in their exact words below;

It is really very difficult to read some of these prescriptions, most prescribers just write carelessly like they are not sure of what they are writing, as if they want us to figure it out. As noted by a key informant.

Bad handwriting, sometimes I just look at the diagnosis and give an appropriate drug since it is hard to read what some prescribers write, then at times you look at the drug name and it has never existed. As reported by a key informant.

Prescribing errors are very often especially unreadable prescriptions and drug names that are not recognizable, As noted by a key informant.

4.1.5 The prescriber related factors influencing prescribing errors

Several prescriber- related factors were analyzed and results are presented below;

4.1.5.1 The nature of department where selected prescribers practiced

The highest proportion (34.8%) of the prescribers was from the outpatient department (OPD), followed by a proportion of 29.7 % of the prescribers in the department of medicine. Other specialist departments were the least (3.2%) as indicated in figure 4 below:

Figure 4: Graph showing the different departments where respondents were selected

Source: primary data

4.1.5.2 Respondents that made prescribing errors per department

Majority (81.3%) of the of the prescribers from the obstetrics and gynecology department made prescribing errors while OPD registered the least number (30.9%) of prescribers that had made prescribing errors as indicated in figure 5 below.

Figure 5: Respondents that made prescribing errors per department

Source: primary data

4.1.5.3 Other prescriber related factors

Table 5: below shows that more than half (57%) of the prescribers that had been in practice for 1- 5 years made prescribing errors, the highest proportion (54.2%) of the prescribers that reported seeing more than 30 patients a day, did not make prescribing errors. A total of 61.5% of the prescribers working for more than 8 hours a day did not make prescribing errors. The highest proportion (69.7%) of prescribers that was always certain when prescribing did not make any prescribing errors.

Table 5: Showing the prescriber-related factors influencing prescribing errors

Variable	Made prescribing error	Didn’t make prescribing error
	N= 158
	Frequency (%) n=75	Frequency (%) n=83
Period of practice
< 1 year	12 (46.2)	26 (53.8)
1-5 years	45 (57.0)	34 (43.0)
>5 years	18 (49.9)	23 (56.1)
No. of patients seen per day
< 30 patients	31 (50.0)	31 (50.0)
>30 patients	44 (45.8)	52 (54.2)
Hours of work per day
< 8	25 (62.5)	15 (37.5)
8	20 (50.0)	20 (50.0)
>8	30 (38.5)	48 (61.5)
Consult while prescribing
Consult	48 (50.5)	47 (49.5)
Never consulted	17 (56.7)	13 (43.3)
Always certain	10 (30.3)	23 (69.7)

Source: primary data from prescribers

Key informants were also asked about factors they thought were influencing prescribing errors and most of the key informants noted that; work load was the major predictor of prescribing errors. The responses are summarized just as stated by the KI below;

The prescribers can get tired; sometimes they work for over 12 hours, so they are liable to committing prescribing errors, noted by a key informant (KI).

Another KI reported that; Patients are very many yet prescribers are few and to err is human so since prescribers are human, the chances of committing a prescribing error are very high.

Some prescribers work in different health facilities, you know how much the government pays them, so that is the only way to make extra money, so they can be exhausted, and since they sit in their examination rooms alone, they take decisions solely, it is not easy to consult other prescribers especially when the patient is in front of you. As noted by a key informant.

Prescribers at this health facility can work without taking any break, as soon as they enter the consultation rooms, they remain there till all patients are seen, yet here we can get even 200 patients a day and prescribers are just three. I think prescribers get tired and make errors sometimes; even when they leave the consultation rooms they really look very exhausted.

4.1.6 Institutional factors influencing prescribing errors

Table 6 below shows that the highest proportion 68.3% (41/60) of prescribers that didn’t make prescribing errors reported not having a system to recognize prescribing errors, more than two thirds; 67.1% (49/73) of prescribers that received on job training regarding prescribing errors did not make prescribing errors, the highest proportion 52.5%% (73/139) of the participants that signed on the prescriptions before handing them to the patients did not make prescribing errors. Majority 72.3% (17/22) of prescribers that were not sure whether the health facility retained a copy of the signed prescriptions did not make prescribing errors. Most 72.5% (37/51) of the prescribers that were not sure whether retained copies of prescriptions were later on reviewed at the health facility did not make prescribing errors. Most 63.3% (14/22) of the respondents that noted that their health facilities had no prescribing standards/ alerts made prescribing errors. The highest proportion (54.5%) of prescribers that made computerized prescriptions did not make prescribing errors

Table 6: Showing the institutional factors influencing prescribing errors

Variable	Made prescribing error	Didn’t make prescribing error
	N= 158
	Frequency (%)n=75	Frequency (%) n=83
System to recognize errors
Yes	42 (57.5)	31 (42.5)
No	19 (31.7)	41(68.3)
Not sure	14 (56.0)	11(44.0)
On job training
Yes	24 (32.9)	49 (67.1)
No	51 (60)	34 (40)
Signing on prescriptions
Yes	66 (47.5)	73 (52.5)
No	9 (47.4)	10 (52.6)
Facility retaining signed copies	N = 139
	n=66	n=73
Yes	48 (49.5)	49 (50.5)
No	13 (65)	7 (35)
Not sure	5 (22.7)	17 (72.3)
Review of retained signed copies	N = 124
	n=55	n=69
Yes	31 (57.4)	23 (42.6)
No	10 (52.6)	9 (47.4)
Not sure	14 (27.5)	37 (72.5)
Prescribing standards/ alerts
Yes	49 (43.3)	64 (56.6)
No	14 (63.6)	8 (36.4)
Not sure	12 (52.2)	11 (47.8)
Form of prescription
Computerized	50 (45.5)	60 (54.5)
Hand written	25 (52.1)	23 (47.9)

Source: primary data from prescribers

Reporting system as an institutional factor influencing prescribing errors

Figure 6 below, shows that more than half 53.2% (84/158) of the prescribers reported not having a reporting system in the health facility and out of the 84; a total of 58.3% (49/84) of the prescribers did not make errors.

Figure 6: Showing responses of prescribers in regards to presence of a reporting system

Source: primary data

The key informants also pointed out certain interventions that can be put in place to prevent prescribing errors in the health care industry.

It is very expensive for every prescriber to use an electronic system but health facilities that are able to provide their prescribers with computers, it would be one intervention to prevent prescribing errors, as noted by one key informant.

Prescribers should also be monitored; all prescriptions should be reviewed to identify errors that can be that could be potentially very harmful. As suggested by a key informant.

Another KI reported that; there are many new drug formularies introduced, so institutions have to keep updating their prescribers through continuous medical education. Prescribers should also make effort to do double checks before handing over prescriptions to patients.

4.2 Bivariate Analysis

To determine the predictors for prescribing errors, all variables under institutional processes and interventions were run in a Poisson regression model (since the outcome variable was more than 10%) to estimate the incidence risk ratios (IRR) and their 95% confidence interval for the prescription errors comparing them with the independent variables. Variables were significant at p-values less than 0.05.

4.2.1 Bivariate analysis of the prescriber related factors influencing prescribing errors

Table 7, below shows that prescribers in the department of medicine were 0.3 times less likely to commit a prescription error than the prescribers in the OPD (95% CI= 0.1-0.9, p-value= 0.022).Prescribers in the surgery department were 0.2 times less likely to make prescribing errors than prescribers in the OPD (IRR= 0.2; 95% CI= 0.1-0.7; P= 0.009). Prescribers in the department of pediatrics were 0.2 times less likely to make prescribing errors than prescribers in the OPD (IRR= 0.2; 95% CI= 0.0-0.9; P= 0.049). Prescribers in the department of obstetrics &gynecology were 0.1 less likely to make prescribing errors than the prescribers in the OPD (IRR=0.1; 95% CI=0.0-0.4; P= 0.003).

Prescribers that always consulted when they were not sure during the prescribing process were o.2 times less likely to make prescribing errors than the prescribers that thought were always certain during the prescribing process (95%CI= 0.1-0.7; p= 0.006).

Period of practice, number of patients seen per day and hours of work per day did not significantly influence prescribing errors.

Table 7: The Bivariate analysis of the prescriber related factors influencing prescribing errors

Variable	Prescribing errors	P-value
Variable	IRR (Confidence Interval)	P-value
Departments
OPD	1
Medicine	0.3 (0.1-0.9)	0.022**
Surgery	0.2 (0.1-0.7)	0.009**
Pediatrics	0.2 (0.0-0.9)	0.049**
Obstetrics& gynecology	0.1 (0.0-0.4)	0.003**
Other specialties	0.5 (0.1-4.1)	0.538
Period of practice
>5 years	1
< 1 year	1.7(0.6-4.9)	0.343
1-5 years	0.6 (0.3-2.0)	0.617
No. of patients seen per day
> 30 patients	1
<30 patients	0.9 (0.3-1.9)	0.688
Hours of work per day
>8	1
<8	0.5 (0.2-1.4)	0.218
8	0.9 (0.4-2.4)	0.936
Consult while prescribing
Always certain	1
Consult	0.2 (0.1-0.7)	0.006**
Never consulted	0.3 (0.1-1.0)	0.053

** Statistically significant at 0.05 level of significance

4.2.2 Bivariate analysis of the institutional factors influencing prescribing errors

Table 8; below shows that prescribers that had received on job training on avoiding prescribing errors were 0.3 times less likely to make prescribing errors than the prescribers that did not receive on job training (IRR=0.3;95%CI=0.1-0.8; P=0.019).Prescribers who reported that retained prescriptions were reviewed by the health facility personnel were 0.2 times less likely to make prescribing errors than health facilities that did not review prescriptions. The form of prescription was not significant.

Table 8: Bivariate analysis of the institutional factors influencing prescribing errors

Variable	Prescribing errors	P-value
Variable	IRR (Confidence Interval)	P-value
System to recognize errors
Not sure	1
Yes	1.2 (0.3-5.2)	0.812
No	2.2 (0.5-9.9)	0.304
On job training
No	1
Yes	0.3 (0.1-0.8)	0.019**
Signing on prescriptions
No	1
Yes	0.9(0.9-9.3)	0.895
Review of retained signed copies
Not sure	1
Yes	0.2(0.0-0.7)	0.008**
No	0.4 (0.1-1.2)	0.103
Prescribing standards/ alerts
Not sure	1
Yes	3.6 (0.6-20.4)	0.149
No	1.2 (0.1-9.8)	0.844
Reporting system
Not sure	1
Yes	0.7 (0.1-3.7)	0.713
No	0.6 (0.1-3.4)	0.579
Form of prescription
Hand written	1
Computerized	0.9 (0.4-2.4)	0.959

** Statistically significant at 0.05 level of significance

4.3 Multivariate Analysis

All variables in the bivariate analysis with p<0.35 or potential confounders were included in the multivariate analysis to determine the strength of the risk of making prescribing errors.

4.3.1 Multivariate analysis of predictors of prescribing errors

Table 9; below shows that prescribers from the department of obstetrics & gynecology were 0.1 times less likely to make prescribing errors compared to the prescribers in OPD (IRR=0.1;95%CI=0.0-0.2; p=0.003). Prescribers who always consulted other prescribers when not sure about a certain prescription were 0.2 times less likely to make a prescribing error than prescribers that never consulted (IRR=0.2; 95%CI=0.1-0.9; P=0.35). Prescribers that received on job training on recognizing and avoiding prescribing errors were 0.2 times less likely to make prescribing errors than prescribers that never received any on job training on prescribing errors ( IRR=0.2; 95% CI=0.1-0.5; P=0.001).

Prescribers who noted that the health facilities where they practiced always retained signed prescriptions for review were 0.5 times less likely to make prescribing errors than the prescriber’s whose health facilities never retained any signed prescription for review. Prescribers who reported working less than 8 hours per day were 0.4 times less likely to make prescribing errors than prescribers who worked more than 8 hours a day (IRR=0.4; 95% CI=0.2-0.9; P=0.033).

Having standard prescribing procedures and alerts, error detecting systems and the period of practice were not significant predictors for prescribing errors.

Table 9: Multivariate analysis of the predictors for prescribing errors

Variable	Prescribing errors	P-value
Variable	IRR (Confidence Interval)	P-value
Departments
OPD	1
Medicine	0.7 (0.2-2.7)	0.649
Surgery	0.5 (0.1-2.3)	3.373
Pediatrics	0.1 (0.0-1.6)	0.108
Obstetrics & gynecology	0.1 (0.0-0.2)	0.001**
Other specialties	3.2 (0.1-3.7)	0.360
Period of practice
>5 years	1
< 1 year	2.2(0.8-6.0)	0.134
1-5 years	0.7 0.3-1.6)	0.388
Hours of work per day
>8	1
<8	0.4 (0.2-0.9)	0.033**
8	0.8 (0.3-2.0)	0.840
On job training
No	1
Yes	0.2 (0.1-0.5)	0.001**
Review of retained signed copies
Not	1
Yes	0.5(0.2-1.5)	0.225**
Not sure	0.3 (0.9-1.1)	0.067
System to recognize errors
Not sure	1
Yes	0.8 (0.3-2.3)	0.671
No	3.0 (1.0-9.0)	0.503
Prescribing standards/ alerts
Not sure	1
Yes	2.1 (0.4-10.6)	0.365
No	0.6 (0.1-4.4)	0.691

** Statistically significant at 0.05 level of significance

CHAPTER FIVE

DISCUSSION

5.0 Introduction

This chapter discusses the study findings, compares and contrasts them with the findings of related studies done before according to the study objectives; the prevalence of prescribing errors, the nature of prescription errors as perceived by prescribers, institutional processes influencing prescribing errors in medical practice in public and private health facilities in Kampala City and institutional interventions in place to prevent prescription errors.

5.1 Prevalence of prescribing errors in public and private health facilities

The prevalence of prescribing errors among prescribers in public and private health facilities in Kampala city (47.5%) is relatively higher compared to other reviewed studies (Frey et al., , et al., 2002, Hendey et al., 2005, Otero et al., 2008) in USA where the prevalence was found to be 4.48%,1.5% and 2.2%, and 7.3% respectively. The discrepancy could be attributed to the difference in the study settings. In USA, prescribers are familiar with indication-based prescribing which intercepts errors because of the alerts in the computer, this is evident in a study (William et al., 2013) conducted in USA yet in Uganda there are very few health facilities with electronic prescribing yet they still make prescribing errors because they are not designed to intercept or detect errors, what is recorded.

However, in a single health facility in South California (Haw et al., 2003) the prevalence of prescribing errors was way too high (83%) compared to the current study, this could probably be attributed to the well-established reporting system in California as opposed to what we have in Uganda where data is cooked (forged). This was evident in a study (Kiberu et al., 2014) to strengthen district health reporting through Health Management Information software system after noting that the data recorded did not match what was on ground. In patient safety, the more errors reported the better for system because it does not allow a reoccurrence; hence improvement in the quality of services (Donabedian, 1980). The government and administrators of the respective health facilities have to design robust systems that are able detect and prevent prescribing errors so as to reduce the prevalence of prescribing errors and these errors should also be reported so that the system is designed better so that the errors don’t reoccur.

5.2 The nature of prescribing errors as perceived by the prescribers

Prescribers in both public and private health facilities identified prescribing errors ever encountered as unreadable prescriptions, unclear drug name, drug dose, drug name unclear, and unclear dose. Most prescribers identified unclear drug name as the most common prescribing error. The key informants (pharmacists and pharmacy technicians) were in line with the prescribers as regards to unclear drug name being one of the most common prescribing errors. However, they also noted that unreadable prescriptions were also common and a challenge to them, unlike the report from the National Health Survey (NHS report, 2012) where dosage and timing were the serious errors in 1/20 of all reviewed prescriptions.

The key informants also noted that it was very hard to read certain prescriptions;

It is really very difficult to read some of these prescriptions, most prescribers just write carelessly like they are not sure of what they are writing, as if they want us to figure it out, bad handwriting; sometimes I just look at the diagnosis and give an appropriate drug since it is hard to read what some prescribers write, then at times you look at the drug name and it has never existed. “Prescribing errors are very often especially unreadable prescriptions and drug names that are not recognizable.

As noted by several key informants. Prescribers have to take prescribing seriously; any errors committed could potentially harm patients. Drug names should not be confusing since the Uganda medicines Policy emphasizes prescribing generic names as opposed to trade names, this should be uniform in all health facilities. The U.S food and drug administration agency also advices prescribers to reduce the confusion of drug names by including the generic (established) name of medication in addition to the brand name (FDA, 2015). Since the policy is already in existence in Uganda, now health facilities in Kampala city have to implement and monitor to ensure that all prescribers use generic names, if they prescribe a brand it should have a corresponding generic name so that the dispensers are not confused with the drug name.

Another study (Lesar et al., 2002) identified several types of prescribing errors that were not identified by prescribers in Kampala city, for instance; wrong patient, incorrect frequency and missing of vital information. Then in another study (Fijn et al., 2002), the nature of prescribing errors as perceived by the prescribers were similar to those perceived by prescribers in Kampala city with an exception of incorrect route. No prescriber in the selected health facilities in Kampala city had ever encountered a prescription with an incorrect wrong route.

5.3The prescriber- related factors influencing prescribing errors

Due to the nature of work per department, prescribers in particular departments were more liable to making errors; prescribers in the obstetrics and gynecological department were 0.1 times less likely to make prescribing errors than prescribers in outpatient department (OPD). This was consistent with what Sard and colleagues found in their study in an urban teaching hospital where prescribers in OPD made more prescribing errors than any other department (Sard et al., 2008). The consistence could be attributed to the nature of work in the outpatient department, since most prescribing is done at OPD.

The number of working hours per day was found to be significantly influencing the likelihood to make prescribing errors. Prescribers that worked for less than 8 hours a day were 0.4 times less likely to make prescribing errors than those who worked for more than 8 hours a day. This is consistent with a study conducted by Wanzel and colleagues (Wanzel et al., 2010) where an experiment was done on reduction of working hours for resident medical workers and results revealed that a reduction in the number of working hours increased performance and reduced prescribing errors among resident prescribers. Prescribers rarely take breaks like other professionals especially in the public sectors and this predisposes them to committing errors. This argument was raised by one of the key informants;

prescribers at this health facility can work without taking any break, as soon as they enter the consultation rooms, they remain there till all patients are seen, yet here we can get even 200 patients a day and prescribers are just three. I think prescribers get tired and make errors sometimes; even when they leave the consultation rooms they really look very exhausted.

As noted by one of the key informants from a public health facility. The government has to properly plan for the delivery of health services, the staffing norm per facility level may not help much, a health center iii in Kampala city may not serve the same population as a health center iii in a rural setting, so since most of Uganda’s population lives and works in Kampala city, a considerable number of prescribers should be placed to these health centers. That might enable prescribers work in shifts and reduce on the working hours for increased productivity and a reduction in prescribing errors.

However, Wanzel and colleagues found that prescribing errors were entirely dependent on the prescriber’s attitude (Wanzel et al., 2010). They argued that even when processes are clearly designed to suit the prescriber, he or she will still make errors, actually the number of working hours did not have any significant association with making prescribing errors which was contrary to the current study findings. This implies that prescribers should actually take their roles seriously and abide to the professional and ethical requirements especially in regards to Nonmaleficence “Do no harm”.

5.4 The institutional factors influencing prescribing errors

The institutional factors that were found to be significantly associated with preventing errors were on job training and review of prescriptions. Prescribers that received on job training on recognizing and preventing prescribing errors were 0.2 times less likely to make errors than the prescribers that did not receive on job training. This is consistent with what several studies (Shaughnessy et al., 2000 and Howell et al., 2004) where in service training on prescribing errors reduced the frequency of errors among prescribers. This indicates that on job/ in service training is very paramount when it comes to quality in the healthcare industry. This kind of training can be offered through continuous medical education programs.

The reemergence of diseases has forced the invention of different drug formularies with different potent and efficacy levels, so it is crucial that prescribers get continuous education programs on these new inventions in form of in-service training. Howell and colleagues realized that in-service training on how to prevent prescribing errors reduced the prevalence of prescribing errors by 17.9%. The government and Health facility managers in Kampala city should endeavor to have all prescribers frequently trained on how best to avoid making prescribing errors, this will definitely promote safe prescribing.

Another study (Webbe et al., 2007) illustrated how in service training was the main intervention in preventing prescribing; a group of prescribers that got in serving training on prescribing errors did not frequently make errors like the prescribers that did not receive in service training. This evidently shows that training is very effective in reducing prescribing errors among prescribers.

In a similar study (Kozer et al., 2006) prescribing errors were attributed to lack of adequate knowledge regarding prescribing certain medications. The current study did not assess the level of knowledge related to prescribing among prescribers, instead the study focused on the training offered in regards to prescribing.

Reviewing prescriptions was also found to significantly influence prescribing errors. Prescribers who reported that health facilities normally retained prescriptions which were later on reviewed were 0.5 times less likely to make prescribing errors than prescribers in health facilities that did not review prescriptions. This was in agreement with a study in India (Ghandi et al., 2005) where frequent checks were responsible for a 9.6% reduction in the prevalence of prescribing errors. Evans’s findings were also consistent with the current study findings (Evans et al., 2007); they discovered that prescriptions that were reviewed, similar prescribing errors never reoccurred than errors on none reviewed prescriptions which often reoccurred. Prescribers in Kampala city tend to work in more than one health facility;

As noted by a key informant, therefore, it is very necessary for health facilities to retain copies of these prescriptions which can later be reviewed and discussed to prevent reoccurrence of similar prescribing errors.

The Front Health Services Management (Stefl, 2001) and the Institute of Medicine in America also acknowledge that to err is Human (Linda et al., 2000), so prescribers are human beings who are liable to making errors; “Patients are very many yet prescribers are few and to err is human so since prescribers are human, the chances of committing a prescribing error are very high”, “the prescribers can get tired; sometimes they work for over 12 hours, so they are liable to committing prescribing errors”, noted by a key informants (KI). A call for a redesign of the system to ensure that frequent checks are done before a patient receives a prescription is urgently needed in health facilities in Kampala city. However, prescribers should also be conscious as they prescribe because some errors could be very serious and cause harm to the patients. If Prescribers take prescribing seriously, the errors will be minimized.

Computer controlled prescribing was found not to significantly prevent prescribing errors which was contrary to studies Bates et al., 2003 and Anton et al., 2004) in USA and UK respectively where electronic prescribing with computers with warnings and alerts that appeared every time one made an error. In agreement with the previous studies was a study (Bizovi et al., 2002) in USA where computer controlled prescribing reduced the prevalence of prescribing errors from 2.5% to 0.54%. The discrepancy of the findings from the current study findings could be probably attributed to the study setting. Electronic prescribing has not been fully embraced in Uganda. There are a few private health facilities that do electronic prescribing and the few that have computers are not designed to have warnings/ alerts pop up when a prescribing error is made. Therefore, having them would only reduce one form of prescribing errors (unreadable prescriptions as a result of bad hand writing) unless they get designed to showcase warnings and alerts every time an error is made.

CHAPTER SIX

CONCLUSION AND RECOMMENDATION

6.0 Introduction

This chapter includes the conclusion as per the posed research questions (what is the prevalence of prescribing errors? What is the nature/ type of prescribing errors encountered? What are the prescriber related factors influencing prescribing errors? And what are the institutional factors influencing prescribing errors?) Several recommendations to the respective bodies are also included in this chapter.

6.1 Conclusion

The prevalence of prescribing errors among prescribers in both public and private health facilities in Kampala city is relatively high (47%). The identified prescribing errors as perceived by the prescribers were unreadable prescriptions, wrong dose, drug name and dose unclear. However, the most encountered prescribing error was drug name unclear.

The identified prescriber-related factors influencing prescribing errors include; nature of the prescriber’s department, never consulting other prescribers while prescribing when not sure of the medication and many hours of work (more than 8 working hours per day).

The institutional factors that were significantly associated with a likelihood of committing a prescribing error were; lack of on job training regarding prescribing and not reviewing prescriptions retained at the facility for any inconsistencies.

6.2 Recommendations

i. Prescribers should take prescribing very seriously and endeavor to consult where necessary since these errors have the potential to harm patients.

ii. The government and private health facility managers should endeavor having busy departments like the outpatient department adequately staffed. Busy departments can potentially predispose one to making an error, or it is very crucial that these departments are considered during staffing decisions.

iii. Health facilities should introduce several interventions like regular reviews and double checks to prevent prescribing errors. Regular reviewing of health-facility-retained copies of prescriptions can be very helpful in terms of recognizing the errors and deriving solutions on how to prevent a recurrence of such errors.

iv. All prescribers should sign on every prescription so that the pharmacy department can trace the prescriber in case any clarification is required especially for prescribers with very poor hand writing.

v. The health facilities should encourage working in shifts so that prescribers don’t work for many hours and become liable to committing errors. Prescribers who work in more than one facility should schedule some time off to rest in order to avoid fatigue that would result in making prescribing errors.