Perspectives of healthcare professionals in Qatar on causes of medication errors: A mixed methods study of safety culture

Background There is a lack of robust, rigorous mixed methods studies of patient safety culture generally and notably those which incorporate behavioural theories of change. The study aimed to quantify and explain key aspects of patient safety culture which were of most concern to healthcare professionals in Qatar. Methods A sequential explanatory mixed methods design of a cross-sectional survey followed by focus groups in Hamad Medical Corporation, Qatar. All doctors, nurses and pharmacists were invited to complete the Hospital Survey on Patient Safety Culture (HSOPS). Respondents expressing interest in focus group participation were sampled purposively, and discussions based on survey findings using the Theoretical Domains Framework (TDF) to explain behavioural determinants. Results One thousand, six hundred and four questionnaires were received (67.9% nurses, 13.3% doctors, 12.9% pharmacists). HSOPS composites with the lowest levels of positive responses were non-punitive response to errors (24.0% positive) and staffing (36.2%). Specific TDF determinants potentially associated with these composites were social/professional role and identity, emotions, and environmental context and resources. Thematic analysis identified issues of doctors relying on pharmacists to correct their errors and being reluctant to alter the prescribing of fellow doctors. There was a lack of recognition of nurses’ roles and frequent policy non-adherence. Stress, workload and lack of staff at key times were perceived to be major contributors to errors. Conclusions This study has quantified areas of concern relating to patient safety culture in Qatar and suggested important behavioural determinants. Rather than focusing on changing behaviour at the individual practitioner level, action may be required at the organisational strategic level to review policies, structures (including resource allocation and distribution) and processes which aim to promote patient safety culture.


Introduction
Promoting patient safety in healthcare settings is a global challenge, with an estimated one in ten patients being harmed whilst receiving care [1]. In an effort to raise awareness of key concepts and strategies in patient safety, the World Health Organization (WHO) published 'Medication Without Harm, WHO Global Patient Safety Challenge' in March 2017 [2,3]. The challenge calls for action to reduce patient harm which occurs as a result of unsafe medication practices and medication errors [2,3]. The goal is to 'gain worldwide commitment and action to reduce severe, avoidable medication-related harm by 50% in the next five years, specifically by addressing harm resulting from errors or unsafe practices due to weaknesses in health systems'. Accumulation of evidence confirms that healthcare professionals often prescribe, dispense and administer medication in ways and circumstances that may increase the risk of patient harm [4][5][6][7][8].
Whilst it is noted that the magnitude and nature of medication harm may differ between countries, globally the cost associated with medication errors has been estimated at US$ 42 billion annually [2,3]. The most commonly cited and accepted definition of the term 'medication error' is that of the National Coordinating Council for Medication Error Reporting and Prevention (NCCMERP) in the United States (US), 'any preventable event that may cause or lead to inappropriate medication use or patient harm, while the medication is in the control of the health care professional, patient, or consumer' [9]. Most research literature focuses on errors relating to prescribing, administration and dispensing, with evidence that causation is often complex and multifactorial. Systematic reviews have focused on causes of medication errors in different patient populations and settings [10][11][12][13]. Common to all systematic reviews is the relatively poor research methodologies reported in most of the primary literature, a lack of behavioural theory and organisational culture in study design. Furthermore, very few studies employed a mixed methods approach to allow quantification and in-depth description and explanation of contributory factors.
Behavioural theories may be used to provide explanation hence providing a robust and rigorous foundation for development of behaviour change interventions. The United Kingdom (UK) Medical Research Council (MRC) framework, 'Developing and implementing complex interventions' highlights the importance of considering theory, noting that interventions grounded in theory are more likely to be effective than those developed empirically or pragmatically' [14]. The Theoretical Domains Framework (TDF) is being used increasingly within health-related research to provide insight into influences on behaviour. TDF derives from 33 psychological theories and 128 theoretical constructs organised into 14 domains of behavioural determinants, as described in Table 1 [15].
It is apparent that there is also a need to focus attention on organisational safety culture. The 'Study Group on Human Factors' defines organisational safety culture as, 'the product of individual and group values, attitudes, perceptions, competencies, and patterns of behaviour that determine the commitment to, and the style and proficiency of, an organization's health and safety management [16].' While two systematic reviews have explored interventions to promote safety culture in hospitals in general and acute hospitals specifically, medication safety was not a feature of any primary research [17,18].
In an attempt to promote and standardise the measurement of organisational safety culture, the US Agency for Healthcare Research and Quality (AHRQ) and Medical Errors Workgroup of the Quality Interagency Coordination Task Force (QuIC) sponsored the development of the Hospital Survey on Patient Safety Culture (HSOPS) [19]. Items are clustered within 12 composites as presented in Table 2.
Research on medication errors within the Middle East has historically been reported to be of poor quality [20]. Recently, Elmontsri et al. conducted a systematic review to explore the status of patient safety culture in Arab countries based on the findings of the HSOPS [21]. Data from 18 studies across seven countries (excluding Qatar) were included, identifying that composites relating to non-punitive response to error to be infrequently practised in their organisation, that staffing levels were often inadequate and that communication needed to be more open. The authors concluded that further research is warranted to provide explanation of these findings and to identify potential interventions to enhance culture and patient safety.
The aim of the present study was to quantify and explain key aspects of patient safety culture which were of most concern to health professionals in Qatar.

Study design
A sequential explanatory mixed methods design was employed, with a cross-sectional survey followed by focus groups in samples of questionnaire respondents to provide further depth and explanation of survey findings [22,23].

Setting
The research was conducted within Hamad Medical Corporation (HMC), the main provider of secondary and tertiary healthcare in Qatar.

Cross-sectional survey
The first phase of the research was a cross-sectional survey.

Questionnaire development
The questionnaire was adapted from AHRQ HSOPS with items presented as 5-point Likert type scales; personal and practice demographic items were added. The common language of care delivery at HMC is English thus translation into other languages (e.g. Arabic) was not required. The questionnaire was piloted in a convenience sample of 100 healthcare professionals. Test-retest reliability was assessed in pilot respondents by requesting that the questionnaire be completed on a second occasion within an interval of two weeks. A high level of test-retest reliability was achieved (p<0.001, Cohen's kappa) for all Likert statements. The final questionnaire was formatted in Snap 10 Professional (software for web and email questionnaire design, publication, data entry and analysis). On completion of the questionnaire, respondents were invited to participate in focus groups to discuss responses in more detail.

Recruitment
All doctors, nurses and pharmacists working within HMC were eligible to participate, with no exclusions. Three hundred and sixty responses were required to give a margin of error of 5% with 95% confidence intervals [24]. Online participation was encouraged via HMC web alerts and promotional posters; paper-based questionnaires were distributed to all doctors, nurses and pharmacists. Data were collected from mid-January 2016 to mid-April 2016.

Data analysis
Anonymised online submissions were imported into Snap before direct export to SPSS version 21.0 and data cleaned prior to analysis. Paper-based questionnaire data were entered manually using the survey link. Descriptive statistics were used to describe respondents' demographic and professional characteristics as well as their responses to individual HSOPS safety culture items. For each composite, a score was determined in line with the recommendations of the AHRQ task force [19]. This involved firstly calculating the percentage of positive responses to each item. A positive response may reflect agreement with a positively phrased statement, or may reflect disagreement with a negatively phrased statement. Therefore, for positively worded items the percentage of positive responses is the proportion of 'agree' and 'strongly agree' responses to the item, and for negatively worded items the percentage of positive responses is the proportion of disagree and strongly disagree responses to the item. Higher percentages (%) therefore reflect more positive responses to HSOPS safety culture items as defined in Table 2.
The composite score was then expressed as the mean of the positive responses. For those composites with overall positive responses of <50%, Chi-square was used to determine any statistically significant associations with demographic variables.

Focus groups
To clarify, explore and explain issues identified in the survey phase, a qualitative approach was employed.

Sampling and recruitment
Respondents of the survey who expressed interest in participating in the focus groups were sampled purposively to represent a range of professions, hospitals and number of years of experience.

Data generation
The focus group topic guide was developed following analysis of the questionnaire data, with reference to the TDF [15]. Initial discussions were based around views and experiences on safety culture in relation to causes of medication errors. This was followed by targeted discussions around patient cases of errors illustrating errors in prescribing, dispensing and administration in diverse patient groups which led to significant harm. TDF domains were used as prompts in relation to potential causes of errors. Focus groups were moderated by two experienced qualitative researchers, with informed consent obtained from each participant at the outset. Discussions were audio-recorded (with permission), transcribed verbatim and checked for transcribing reliability. A clear audit trail was maintained which documented details of data gathering to promote dependability [25]. Sampling and recruitment continued to the point of data saturation, when no new themes emerged from data analysis [26]. Focus groups were conducted between mid-May 2016 and mid-June 2016.

Data analysis
Data analysis followed the Framework Approach, using TDF domains deductively for to generate a coding framework [27]. Two researchers independently read each focus group transcript repeatedly to ensure familiarity, then coded text to one or more TDF domains. Any disagreements were resolved by discussion which involved a third researcher if needed.

Ethics
The study received ethical approval from Hamad Medical Corporation, Medical Research Center Qatar, Qatar University Institutional Review Board and Robert Gordon University Research Ethics Sub-Committee. Return of the questionnaire was taken as an indication of informed consent; written informed consent was obtained from all focus group participants.
Patient safety culture items. Positive responses to the HSOPS composites and items are given in Table 4. Composites with the lowest levels of mean positive responses were: non-punitive response to errors (24.0%); staffing (36.2%); communication openness (50.5%); handoffs and transitions (53.1%); and supervisor/manager expectations and actions promoting patient safety (56.5%). Composites with the highest levels of positive responses were: organisational learning-continuous improvement (85.5%); team working within unit (82.1%); and management support for patient safety (75.4%). For the two composites with mean positive responses of <50%, Chi-square was used to determine the associations between percentage positive responses and demographics/professional characteristics.
Non-punitive response to errors-all individual items contributing to this HSOPS composite attracted a low level of positive response, this was particularly the case for items relating to staff concerns over errors being kept in their personnel files (26.2%), and the perception that errors counted against them (14.6%). There were highly statistically significant associations with mean composite agreement and gender (females most positive, Χ 2 (1, N = 1547) = 8.23, p<0.005), age (older most positive, Χ 2 (4, N = 1555) = 11.62, p<0.05) and experience as a healthcare professional (the most experienced being most positive, Χ 2 (5, N = 1536) = 18.42, p<0.005).
Staffing-while all responses attracted a low level of positive response, this was particularly the case for work pressures and speed of work (23.5%). There were highly statistically significant associations with mean composite agreement and healthcare professions (doctors most positive and pharmacists least, Χ 2 (2, N = 1494) = 42.06, p<0.001), age (youngest least and For those ten composites with higher mean agreement, several items had less than half responding positively. There were issues around: supervisors/ managers overlooking recurring patient safety problems (31.9% positive); that it was due to chance that serious errors did not occur (36.0%); problems occurring when exchanging information across hospital units (42.9%); staff being able to ask questions if things did not seem right (44.0%); that at particular pressure points supervisors/ managers wanted staff to work faster, even if this required shortcuts to be taken (46.1%); and staff feeling able to question those in positions of authority (46.6%).
More detailed data on the responses to individual items within each composite are given in S1 File.

Focus groups
Demographics of participants. Two hundred and ninety-five survey respondents (18.4%) expressed interest in participating in focus groups. Nine focus groups were conducted (duration 45-60 minutes), at which point data saturation was deemed to have been achieved. Fifty-four individuals from different disciplines participated, with just under half (n = 26, 48.1%) being nurses, followed by 18 (33.3%) pharmacists and 10 (18.5%) doctors. Most were highly experienced with only 11 (20.4%) having <5 years of experience. During the focus groups, there was wide-ranging discussion across the spectrum of medication errors of prescribing, administration and dispensing.
Behavioural determinants associated with errors. Themes and subthemes relating to safety culture identified during focus group discussions are mapped to TDF behavioural determinants, with illustrative quotes provided for each.
A. Social/professional role and identity (a coherent set of behaviours and displayed personal qualities of an individual in a social or work setting)   When an error is made, but is noticed and corrected before affecting the patient, how often is this reported?

(agreed)
When an error is made, but has no potential to harm the patient, how often is this reported?

(agreed)
When an error is made that could potentially harm the patient but does not, how often is this reported?

(agreed)
Overall perceptions of patient safety, overall positive response = 59.1% Patient safety is never sacrificed to get more work done 70.6 (agreed)

Doctors reliance on pharmacists to correct errors.
During discussion, it emerged that there were instances where doctors would rely on pharmacists to correct their prescribing errors and this led to complacency around prescribing,

Doctors reluctance to alter other doctors' prescribing.
During one focus group, there was concern that doctors were unwilling to alter prescriptions written by other doctors, particularly for doctors from other specialities. The doctors considered this to be the responsibility of the original prescriber, even if a prescribing error had been made and initial prescriber was unavailable, 'This will happen when you're in the Ob-Gyn [obstetrics and gynaecology] setup. If one physician came from Hamad from other. . . from cardiac or other site, if they write any prescription, if you call the Ob-Gyn doctor here, the on duty doctor, she will never agree to change because B. Emotions (a complex reaction pattern, involving experiential, behavioural, and physiological elements, by which the individual attempts to deal with a personally significant matter or event) 1. Stress leading to medication errors. Stress and high pressure situations were described in all focus groups as influences on medication errors. While workload was a common factor leading to stress, patients themselves could also put undue pressure and hence stress of health professionals,

' And I think that probably the stresses of the work [lead to errors].' (FG1 Doctor 2)
' And parents are too tense than they are. . . even the parents they are too much angry. Yeah, they will scold the staff then like that time they will get pressure.' (FG7 Nurse 3)

C. Environmental Context and Resources
(any circumstance of a person's situation or environment that discourages or encourages the development of skills and abilities, independence, social competence, and adaptive behaviour) Much of the discussion centred on aspects of environmental context and resources as key influences leading to medication errors. These were discussed by all participants in all focus groups. There were several key themes within this domain.
1. Workload issues leading to medication errors. Workload issues were discussed by doctors, nurses and pharmacists. Doctors believed one of the reasons for errors to happen was the heavy workload that they had. One pharmacist noted that the excessive workload for the doctor can lead to errors occurring and that this workload also put pressure on other health professionals which could compound errors.
'There are two problems here, a load on the physician that can lead to many mistakes and a load on the pharmacist because he needs to dispense medication for this patient and at the same time answer the questions of physician, nurses.' (FG5 Pharmacist 4) One of the nurses also explained that the main cause of errors committed by junior medical staff was workload rather than lack of knowledge.
' And this is why the medication errors are also increasing, so it's not always related to the knowledge of the resident. And if the resident is overloaded because he has to document for all the patients and see all the patients and he is receiving calls from other units as well' (FG3 Nurse 3)

Lack of staff at key times.
Closely related to workload issues was a critical lack of staff at key times such as weekends and evening which could compromise patient safety.
'On the whole days of the week, there is complete staff, complete number of physicians. In weekend, well, only one physician is doing the whole work.' (FG4 Doctor 2) 'Especially the areas like emergency, less staff. They will be get. . . too tense by the patients and they just want to do the things for faster. so it will make errors. (FG2 Nurse 1) 3. System-related issues. Discussion also centred on key issues related to the systems in operation in various wards and departments. There was particular concern over the implementation of Cerner (electronic health record system for hospitals, health care providers, clinics) from doctors, nurses and pharmacists. 'The electronic system is not robust, and I mean, the hardware is not good enough.' (FG2 Doctor 1) 'We have now to concentrate on the mistakes or medication errors happening by the prescribing system.' (FG5 Pharmacist 2) One senior doctor commented that following implementation of Cerner, fewer checks were being performed compared to the previous paper-based system. Themes and subthemes for those behavioural determinants less related to safety culture are summarised in Table 5. communication openness; handoffs and transitions; and supervisor/manager expectations and actions promoting patient safety. During focus group discussions, specific TDF determinants suggested as being potentially associated with these composites were: social/professional role and identity; emotions; and environmental context and resources. Thematic analysis identified issues of doctors relying on pharmacists to correct their errors and being reluctant to alter the prescribing of fellow doctors. There was a lack of recognition of nurses' roles and frequent policy non-adherence. Stress was perceived to be a major contributor to errors, as was excessive workload and lack of staff at key times.

Strengths and weaknesses
The mixed methods design is a key study strength providing quantification of results followed by in-depth explanation. Further strengths are the use of the validated HSOPS tool and embedding psychological behaviour change theory (TDF) within qualitative data generation and analysis [15,19]. There are, however, several limitations hence findings should be interpreted with caution. Self-reported questionnaire responses could not be validated and may have been impacted by response and social desirability biases [22]. While responses were received from healthcare professionals in all HMC hospitals, these may have been skewed towards females and nurses hence there are potential issues of lack of generalisability within Qatar and beyond. Similarly, qualitative findings may not be transferable to other healthcare professionals, settings and countries.

Interpretation
This mixed methods study has contributed to the expressed need for robustness and rigour in patient safety research within the Middle East [20]. Furthermore, it aligns to the WHO 'Global Patient Safety Challenge' calling for action to reduce severe, avoidable medication-related harm by 50% in the next five years [2,3]. Whilst the HSOPS questionnaire has been used within the Middle East [21], this is the first study to publish Qatari data. There are, however, similarities between the Qatari data and those reported by Elmontsri et al. [21], with the lowest agreement scores (and hence of most concern) relating to the composites of non-punitive response to errors; staffing; communication openness; handoffs and transitions; and supervisor/ manager expectations and actions promoting patient safety. Within the two composites of lowest scores (non-punitive response to errors and staffing) there were issues with staff perceiving that errors counted against them and that details of errors committed were kept in their personnel files. This appeared to be an issue for male, younger and less experienced healthcare professionals. Staffing was the other key composite with very low agreement scores, particularly in relation to work pressures and speed of work, with similar statistically significant associations as for the non-punitive response to errors. There may be some merit in initially prioritising any intervention towards these specific groupings. One limitation of the published studies using the HSOPS is the lack of qualitative research to provide in-depth explanation of the results [21]. The use of behavioural theory within the focus groups in this study identified key determinants which could facilitate intervention development. TDF has been incorporated within intervention developments for smoking cessation, physical activity, hand hygiene, acute low back pain and schizophrenia [28]. To date only one other published study has applied TDF to explore potential causes of medication errors, focusing on prescribing errors in a sample of junior doctors in Scotland [29]. There are some similarities with the findings of this study, most notably within the domains of knowledge and skills, particularly the general lack of medication-related knowledge. While pharmacists can provide support, and indeed doctors were found to rely on pharmacists to correct errors, the HSOPS data and the focus groups identified issues around staff complement and workload, particularly at key times.
TDF domains of social/ professional role and identify, emotions and environmental context and resources are related to organisational safety culture, as defined by 'Study Group on Human Factors' [16]. Concerns were expressed around nurses perceiving that their professional role was not recognised leading to poor communication compromising patient safety. This is also reflected in the HSOPS score of~50% agreement for communication openness. There were instances of doctors relying on pharmacists to correct their prescribing errors and, at times, would not alter the prescribing of others, even when errors could potentially lead to patient harm. Themes of environmental context and resources also emerged in the discussions around workload as a leading cause of errors, with lack of staff at key pressure times of evening and weekends. Furthermore, the electronic prescribing and records system was considered to have introduced potential for error. While such systems have been shown to enhance patient safety, others have also highlighted the risky human factors and user-centred design issues that have been encountered [13].
Stress was the main theme which emerged in the TDF emotions domain as a determinant of error, arising due to workload, work pressures and the influence of patients. Issues of workload were also identified in the HSOPS data around staff numbers to handle the workload, working under pressure to do too much, too quickly.
These TDF l determinants which were highlighted as potential contributors to medication errors can be used during the development of behaviour change interventions, defined as 'coordinated sets of activities designed to change specified behaviour patterns'. These are often complex, consisting of interacting components known as 'behaviour change techniques' (BCTs), 'observable and replicable components designed to change behaviour' [30]. Michie et al. developed a cross-disciplinary taxonomy of evidence based BCTs [31], mapped to specific TDF domains [32]. Whilst knowledge and skills can be impacted through education and training [31,32], altering aspects of social/ professional role and identity and environmental context and resources are more complex. Indeed, the work of Michie et al. [31,32] did not identify any evidence-based BCTs which mapped reliably to social/professional role and identity. Those for environmental context and resources relating mainly to restructuring the physical environment and providing prompts and cues for safer practice, which in this case would focus on the electronic medication systems [31,32]. Rather than focusing on changing behaviour at the individual practitioner level, action may be required at the organisational strategic level to review policies, structures (including resource allocation and distribution) and processes which aim to promote patient safety culture and minimise harm. Qualitative research focusing on understanding the perspectives of key strategic decision-makers in relation to promoting all aspects of medication safety is warranted.

Conclusion
This mixed methods study has provided further confirmation of key areas of concern relating to patient safety culture in Qatar. Non-punitive response to errors and staffing had the lowest levels of agreement, followed by communication openness, handoffs and transitions, and supervisor/manager expectations and actions. The qualitative component provided further detail of specific TDF determinants highlighting issues of social/professional role and identity, emotions, and environmental context and resources. Further attention on these issues at strategic and policy levels is required.