Research in Social and Administrative Pharmacy 15 (2019) 546–557

Contents lists available at ScienceDirect

Research in Social and Administrative Pharmacy

journal homepage: www.elsevier.com/locate/rsap

Understanding the causes of prescribing errors from a behavioural T

perspective
Douha F. Bannana,b,∗, Mohammed A. Aseeric, Aeshah AlAzmid, Mary P. Tullyb
a
Faculty of Pharmacy, King Abdulaziz University, Jeddah, Saudi Arabia
b
School of Health Science, Division of Pharmacy and Optometry, University of Manchester, Manchester, UK
c
Pharmacy Department, King Saud bin Abdul Aziz University for Health Sciences, King Abdul Aziz Medical City/ National Guard Health Affairs, Jeddah, Saudi Arabia
d
Pharmacy Department, King Abdul Aziz Medical City/ National Guard Health Affairs, Jeddah, Saudi Arabia

A R T I C LE I N FO A B S T R A C T

Keywords: Introduction: While many attempts have been made to reduce prescribing errors (PEs), they persist. PE is not in
Causes itself a behaviour, but a consequence of a prescribing behaviour. Interventions aimed at prescribers should focus
Prescribing errors on understanding prescribers' behaviours.
Behavious Objectives: The aim of this study was to use the capability, opportunity, motivation - behaviour (COM-B) model
COM-B
to explore the behaviours that could have caused PEs made by senior doctors in a speciality paediatric inpatient
ward.
Methods: A qualitative approach was used to investigate prescribers' behaviours in a 26-bed paediatric oncology
ward. Error data were collected over a two-month period and were presented during focus groups with pre-
scribers, which were audio-recorded and transcribed verbatim. Thematic analysis was used to identify con-
tributory factors to errors, which was used to identify sources of behaviours using the COM-B model.
Results: Behaviours related to prescribers' capabilities were: prescribers' improper use of the software because of
insufficient skills, and prescribers' inability to prescribe correctly because of lack of knowledge. Behaviours
related to opportunities in the environment were: prescribers' inability to make an informed decision because of
poor access to patient information, inability to properly complete a task because of heavy workload and in-
terruption, and having to re-check doses frequently because of frequent change in patients' weight and surface
area. Those related to motivation were: prescribers unquestioningly following recommendations and not com-
municating with other specialists because they over-trusted them or feared a negative reaction, and prescribers
inability to complete a task because of other competing and preferable tasks at the same time.
Conclusion: Employing COM-B helped in identifying causes of PEs from a new perspective. Future work could
focus on mapping identified sources of behaviour and errors against appropriate intervention functions and
policies in order to design more successful interventions.

1. Introduction As with many clinical activities, the behaviour involved in the

Behaviour can be defined as ‘anything a person does in response to
internal or external events.’1 Many factors contribute to individual
decisions and thus the behaviour that individuals express. Some of
these factors relate to the individuals themselves, such as knowledge
and beliefs (i.e. some of the ‘internal’ events mentioned above). Other
factors could relate to the environment, such as hectic or unfriendly
surroundings (i.e some of the ‘external’ events). Of course, individual
behaviour in the presence of such factors varies, and while some of this
behaviour can be predicted based on patterns or experience, other be-
haviours are more difficult to predict.
prescribing process is complex, as the process involves multiple linked
steps that are interconnected and interact in a nonlinear way, which
could produce unpredictable results.2 For many reasons, prescribing for
children is more complex than prescribing for adults. For example,
pharmacokinetic parameters differ between children of different ages,
the weight and height of young children can change dramatically over
short periods, and the ability of children to take different forms of
medication can vary by age.3,4 In addition, compared to prescribing for
children in general, prescribing for children with cancer is even more
complex. Here, additional factors exist related to the nature of the ill-
ness, the additional vulnerability of the patient group, and the use of a

∗
Corresponding author. Faculty of Pharmacy, King Abdulaziz University, P.O. Box 80260, Jeddah, 21589, Saudi Arabia.
E-mail address: dbannan@kau.edu.sa (D.F. Bannan).

https://doi.org/10.1016/j.sapharm.2018.07.007
Received 28 December 2017; Received in revised form 2 July 2018; Accepted 8 July 2018
1551-7411/ © 2018 Published by Elsevier Inc.
D.F. Bannan et al.
combination of high-risk and toxic medications (such as chemotherapy
drugs) over long periods.5,6 The complexity of these conditions and
their treatment plans make hospitalised children with cancer particu-
larly vulnerable to errors.
A prescribing error (PE) is the outcome that occurs when ‘as a result
of a prescribing decision or prescription writing process, there is an
unintentional significant reduction in the probability of treatment being
timely and effective or increase in the risk of harm when compared with
generally accepted practice.’7 PEs are relatively common in hospitals,8
and the error rate for hospitalised children ranges from 16.8 to 86.5
errors per 100 medication orders.9,10 In Saudi Arabia (SA), studies on
PEs for children are limited, but the few published studies suggest that
errors occurred with 41.5%–56% of total written medication or-
ders.11,12
As the definition suggests, PE occurs “as a result of a prescribing
decision or prescribing writing process”. Thus, an error is not in itself a
behaviour that could be targeted for change, but rather it can be con-
sidered as a consequence of a subset of prescribing behaviours, some of
which may have been conducted in an incorrect fashion. To minimise
errors, it is important to understand why errors have occurred.
The causes of errors have been commonly investigated from a
human error perspective using models such as Reason's model of active
causation (i.e. active failure, error provoking conditions and latent
factors).13–16 However, such models typically focus on classifying
causes of errors and offer little information on the behaviours asso-
ciated with the cause of these errors. For example, many studies have
identified performance deficits, knowledge deficits, and lack of com-
munication as causes of PEs in hospitalised patients.17,18 However, how
prescribers behave when they lack knowledge or how their behaviours
are affected when they lack communication with others are not always
investigated.
Within behavioural psychology, there are multiple behaviour
models and theories that could have been appropriate for under-
standing prescribing behaviours. These include the health belief
model,19–21 social cognitive theory,22–24 and the theory of reasoned
action.25,26 However, they are largely stand-alone approaches that do
not, in themselves, provide guidance on how to link the actual beha-
viour change the theories would suggest with the theory of developing
interventions. The theoretical domains framework has been used in
healthcare studies,27–29 but there is no formal guidance on how to apply
the framework for designing interventions. The Capacity, Opportunity,
Motivation – Behaviour (COM-B) model30 was chosen for this study as it
is a component of a step-by-step intervention design framework called
the Behaviour Change Wheel, which focuses on both understanding and
changing behaviour.1 The model states that behaviour is a result of
interactions between the three sources of behaviour: capability, op-
portunity, and motivation. The COM-B model has been employed in
healthcare settings both to understand behaviour and to design inter-
ventions. For example, the COM-B model was used to explore facil-
itators of and barriers to medication adherence,31 to identify inter-
ventions that facilitate the transfer of information on medication
discharge summaries,32 to understand behaviours when using a stop
smoking services,33 to change behaviour of women with gestational
diabetes,34 and to increase hearing aid use.35 The aim of this study was
to use the COM-B model to explore the behaviours that could have
caused PEs made by senior doctors in a speciality paediatric inpatient
ward. This work constitutes a starting point for designing a behaviour
intervention to reduce PEs.
2. Methods
2.1. Study design and population
A qualitative approach was used to investigate prescribers' beha-
viour, to ensure that an in-depth understanding could be achieved.
Interviews and focus groups are two commonly used approaches that
547
Research in Social and Administrative Pharmacy 15 (2019) 546–557
are considered.36–38 Focus groups were selected as the primary data
collection method for two reasons. First, the research ethics committee
argued that, for cultural reasons, doctors in SA would not talk about PEs
in one-to-one interviews due to concerns about sigma, privacy and
negative impact on relationships between the doctors and pharmacists.
Focus groups about non-attributable PEs were acceptable. Second, focus
groups would allow researchers to use discussion dynamics to capture
participants' perceptions and opinions about concepts and topics as to
why prescribers behave in such a way that resulted in errors.39,40
All senior doctors (n = 18) working in the study ward, and who
were authorised to prescribe medication (i.e. consultants, assistant
consultants, staff physicians, and clinical fellows), were eligible to
participate in this study. Unlike in some nations or even other settings
in SA, junior doctors in their first foundation year of postgraduate
training (i.e. interns and junior residents) were not authorised to pre-
scribe in this hospital, and thus were not included in the study.
All of these 18 potential participants were informed about the study
both via email from the department secretary and through a presenta-
tion of the project by researchers at a staff meeting. The participants
were then given participant information sheets and consent forms.
Based on individual preference, the consent forms were collected either
during a subsequent staff meeting or just before the beginning of focus
group sessions. Participation was voluntary. Ethical approval was
granted both by the Institutional Review Board (IRB) in SA and by the
University of Manchester Research Ethics Committee (UREC) in the UK.
2.2. Setting
The study was conducted in a 26-bed paediatric oncology ward in a
tertiary hospital in SA. Two types of prescribing software systems were
used to order medications in this ward, which were given the pseudo-
nyms of O-software and C-software. The O software was the standard
software used by all healthcare professionals working throughout the
hospital. This software was used to order all medications except che-
motherapy protocols. These protocols had not been incorporated within
the O-software and therefore they had to be prescribed using the on-
cology-specific C-software. The major differences between the two
software systems, such as who had access and the proportion of avail-
able patient data captured by each system, are shown in Table 1. The
two systems were not integrated or connected with each other, and
patient information and medication orders in one system could not be
displayed in the other.
2.3. Data collection
Data collection proceeded in two phases. For the first phase, data
were collected on PEs on the ward over a two-month period prior to the
conduction of the focus groups. The purpose of this phase was to collect
actual cases of PEs that could be used for focus groups discussions. One
Table 1
Description of the two prescribing software systems.
Software Brief description
O-software - Recently introduced (2016)
- Contained all patient's information such as laboratory results,
progress notes, consultation, and imagining
- Contained medication orders except chemotherapy protocols, due
to the software being unable to cope with the complexity of
chemotherapy protocols
- Prescribers in all wards in the hospital had access to and used the
O-software while prescribing
C-software - Had been used since 2015
- Contained some of the patient information
- Used solely for prescribing chemotherapy protocols
- Only oncology prescribers had access to the C-software
D.F. Bannan et al.
researcher (AA), the clinical pharmacist in charge of the ward, reviewed
all medication charts and orders for all patients during normal work-
days (Sunday-Thursday) for any issues (including PEs), as part of her
usual practice. All patient safety problems were addressed with the
prescriber in the usual way. Errors were then recorded on a standar-
dised form, using previously validated criteria.15 The form had two
pages: the first contained patient information such as the name, age,
gender, and condition, which allowed the pharmacist to review patient
notes if needed. The second page described the error(s) in detail. After
both pages were coded with the same identification number, all first
pages were removed and stored in a locked hospital cabinet in order to
protect patient and doctor confidentiality. The researcher (DB) then
used the second pages for data analysis.
As Dean et al.’s definition of PEs was used when collecting errors,
the equivalent error classification system7 was used to categorized er-
rors collected. The system classified errors into five categories: need for
drug therapy; selection of dosage regimen; selection of specific drug;
administration of drug; and provide drug product.
For the second phase, four focus groups were conducted with the
senior doctors described above. One week before the conduct of each
focus group (focus groups 1–3), the department secretary emailed six
prescribers, as evidence suggests that the optimum size for a focus
group is between six and eight.41–43 An important characteristic of the
sample was to have prescribers with different grades (consultants, as-
sistant consultants, staff physicians, and clinical fellows). Therefore, the
secretary randomly emailed two consultants, two assistant consultants,
and two staff physicians currently working on the ward and invited
them to participate in the focus groups. Different assistant consultants
and staff physicians were invited to each focus group. However, the
same consultant could be invited more than once to participate in dif-
ferent focus groups, as only five were working on the ward at that time.
Only one fellow was practicing in the ward. Thus, the fellow was in-
vited once. The number of participants attending the focus groups
ranged from 4 to 6 participants. For the fourth focus group, all pre-
scribers who had been invited to the previous three focus groups were
invited to attend, as this provided respondent validation after pre-
liminary data analysis on the first three focus groups was complete (see
below for more details).
The composition of the focus groups ranged from higher-grade
doctors (i.e. consultants) to lower-grade doctors (i.e. staff physicians
and fellows), although all were considered “senior doctors” in the
hospital. This was important to get the perspective of different doctors
and to understand the causes of errors made by different doctors.
However, this also ran the risk of affecting the openness of either group
to talk freely about the errors or their causes.
The research team chose common examples of each category of
errors, created case scenarios and presented these during focus groups
1–3 (see Appendix A). Different error examples were presented to the
participants in the different focus groups. Examples related to one
classification of error were discussed with each group – examples of
error related to 'need for drug therapy' in focus group 1, 'selection of
dosage regimen' in focus group 2 and 'selection of specific drug' in focus
group 3.
Two topic guides were developed. The first topic guide (see
Appendix B) was used for focus groups 1–3. This topic guide focused on
discussing examples of common errors committed in the ward (based on
the case scenarios) and on identifying contributory factors to PEs. For
the purpose of this study, contributory factors refer to any factor that
prescribers believe could contribute to the occurrence of errors. Parti-
cipants were asked very broad questions such as why they thought that
the errors in the case scenarios had error occurred, what might have
contributed to the occurrence and what they thought might have been
done differently to prevent the errors happening. Errors during these
focus groups discussions were presented without knowing whether the
prescriber who made the errors in question had been invited to the
focus group or not. Unless the prescriber who made the error self-
548
Research in Social and Administrative Pharmacy 15 (2019) 546–557
identified during the discussion (and they were asked not to), neither
the facilitators nor the participants knew who they were.
The fourth focus group was conducted for respondent validation,
after preliminary data analysis on the first three focus groups was
complete. This fourth session (see Appendix C for the fourth focus
group topic guide) began with a presentation of the preliminary ana-
lysis results based on the application of the COM-B model. To add va-
lidity to the researchers' interpretations, participants provided feedback
and any disagreements between the participant meaning from data
collected from earlier focus groups and the researchers' interpretations
or representation were resolved. The second part of this session in-
volved discussing questions that were developed based on the COM-B
model application, explicitly to connect the identified contributory
factors from the first three focus groups to COM-B components.
Two members of the research team ran the focus groups. One (MA)
led the discussions for focus groups 1–3, while the other (DB) assisted
and took notes on the major points that emerged. A different facilitator
(DB) ran the fourth focus group. All focus groups lasted between 60 and
75 min. With participant consent, the discussions were taped and
transcribed verbatim by the main researcher (DB). Participant names
were anonymised using codes (P1, P2, P3, etc.), and any other identi-
fying personal information was removed. The focus groups were num-
bered in the order in which they were conducted: FG1, FG2, FG3, and
FG4. Those focus groups were conducted over a four-month period (one
month between focus groups 1–3 and two-month between focus groups
3–4).
In Saudi hospitals, the formal language of communication between
doctors when discussing cases is English. Nonetheless, it was common
for the participants to switch between English and Arabic or add iso-
lated Arabic words in the middle of a conversation. The original ver-
batim transcripts are available from the authors upon request.
Transcripts with Arabic words were translated into English by the bi-
lingual first author (DB) prior to data analysis.
2.4. Analysis
The translated transcripts were read by the research team; DB, MA
and MT and coded iteratively by DB. The researchers discussed this
analysis repeatedly and any differences of interpretation were resolved.
Transcribed focus groups data were analysed in two ways. First, the-
matic analysis based on the work of Braun and Clarke44 was used to
analyse the focus groups transcripts, identify and code potential
themes. Second, the COM-B model was used to conduct an analysis of
the prescribers' behaviours and what led up to those behaviours.
Thematic analysis was chosen because it provides a ‘thick descrip-
tion’ of data sets and allows researchers to identify themes within
data.44 The transcripts were read and re-read multiple times in order to
be familiar with the dataset. Notes were taken of relevant points from
the focus groups discussions, and points made for further clarification
in subsequent focus groups. Then coding was applied using the com-
ments facility within Word, so that data that were related to the con-
tributory factors for and the causes of PEs could be identified. An ex-
tract could be coded in several ways as appropriate.45 The final step was
searching for themes related to the contributory factors or causes of
PEs. The focus was not on identifying which pattern was the most
frequent, but rather capturing the most meaningful elements necessary
to explore the behaviours that could have caused PEs.
As these factors provided no behavioural insights when considered
in isolation, a second analysis using the COM-B model was used. This
well-established model30 is used in behavioural psychology for ana-
lysing and describing behaviour.31,36,46,47 The identified themes that
emerged from the initial analysis were then mapped against the ap-
propriate COM-B model components. This allowed the researcher to
understand prescribers' behaviours and what could have resulted in
these behaviours. For example, if lack of knowledge was identified as a
contributory factor, prescriber behaviours leading to a PE that indicated
D.F. Bannan et al. Research in Social and Administrative Pharmacy 15 (2019) 546–557

a lack of knowledge (such as Prescribers' inability to prescribe correctly
due to lack of knowledge of dose) were identified and mapped against
the appropriate component of the COM-B model. This process allowed
the researchers to determine how a prescriber behaves when their
capability was affected, when an environmental factor was present, or
when they were motivated/not motivated to do certain behaviour.
Findings were presented using the three COM-B model components.
Quotes were selected to highlight the discussion content. Where quotes
were taking out of context, a few words or phrases were added in
parentheses for the reader to understand the situation.
3. Results
Fifteen doctors (nine male and six female) participated in the focus
groups. Approximately half of the participants were staff physicians
(n = 7); the others were consultants (n = 5), an assistant consultant
(n = 1), and a fellow (n = 1). The number of participants in the first
three focus groups was similar, ranged from 4 to 6 participants - FG1
(n = 4), FG2 (n = 6) and FG3 (n = 5). For the fourth focus groups,
eight participants attended. Those who were invited to attend but could
not gave reasons such as dealing with a patient and being busy in other
meetings.
The focus groups provided rich data about prescriber behaviour and
contributory factors that could have led to the errors in the case sce-
narios. By the end of the third focus group, no new themes had emerged
and a data saturation point was reached. The fourth focus groups pro-
vided validation of the initial analysis and additional rich data about
prescribers' behaviour in relation to capability, motivation, and op-
portunities.
During focus groups 1–3, the error scenarios were presented and
discussed freely. Participants appeared motivated to discuss these er-
rors and other issues related to the overall prescribing process and
software systems. The focus groups environment was friendly and
blame-free. During all of the focus groups, participants were willing to
disclose contributory factors, even those that would cast themselves in a
negative light. For example, when participants believed that one con-
tributory factor to behaviour could be a lack of knowledge, they freely
discussed this topic, even concerning medications that were commonly
prescribed to cancer patients, such as antibiotics and anti-emetics. In
fact, in two instances, the prescribers who had made the error being
discussed identified themselves and explained in greater detail why the
error was made.
Before the start of this study, some ward doctors stated that they
made either very few errors (i.e. one every few months) or none at all,
perhaps because they misunderstood the meaning of the term ‘pre-
scribing error.’ In addition, some participants connected the occurrence
of errors to patient harm. Therefore, for those doctors, errors that were
intercepted before medication was administered or before a patient was
harmed were not considered to be errors at all. However, the more error
cases that were discussed in the focus groups, the clearer the meaning of
‘prescribing error’ became, and the more participants appeared to be-
lieve that errors were indeed common and that the paediatric oncology
ward was no exception. In addition, the more these examples of actual
errors were discussed in the focus groups, the more open the partici-
pants became.
The themes identified from the thematic analysis were: lack of
knowledge, lack of training, lack of integration between EP software
systems, heavy workload, interruption and distraction, low staffing
rate, poor communication between teams, and poor access to guide-
lines. Prescribers' behaviours in the presence of these factors are pre-
sented below using the three main components of the COM-B model:
capability, opportunity, and motivation. The application of the COM-B
model was presented in Table 2. Because the model components in-
teracted with each other, incorrect prescribing (i.e. behaviour) some-
times occurred because one component was affected, such as the pre-
scriber capability due to lack of knowledge, or because two or more
components affected each other. For example, an opportunity in the
environment (e.g., heavy workloads) may have caused prescribers to
focus solely on chemotherapy drugs during medication reconciliations
(i.e. motivation), resulting in omission of other medications (i.e. be-
haviour).
In general, all prescribers recognised the ideal types of behaviours
that should be followed. However, the reality of their workplace often
meant that these ideals could not happen.
“[When I am not busy and I don't have lots of patients], I will think
[about] every patient slowly and I will check all the medication [and] the

Table 2
Application of the COM-B to incorrect prescribing behaviour.
COM-B components

Behavioura Anything a person does in response to internal or external events.

In this study, behaviour was anything a prescriber does that eventually leads to a prescribing error

Capability

Definitiona Knowledge, skills and abilities to engage in the behaviour

Factors - Prescribers' improper use of the software because of Insufficient skills
- Prescribers' inability to prescribe correctly because of lack of knowledge

Opportunity

Definitiona Outside factors which make the behaviour possible

Factors - Prescribers' inability to make an informed decision because patient's information not available, can not be accessed, or can not be retrieved quickly
- Prescribers' inability to double check prescription because of manpower
- Prescribers' inability to complete a task because of workload and interruption
- Prescribers' having to re-check the doses frequently because of frequent change in patients' weight and surface area

Motivation

Definitiona Thought processes and perceptions which direct decisions and behaviours
Factors - Prescribers' following recommendations by specialists because they over-trusted the specialist
- Prescribers' not communicating with other teams because they feared the specialist would get angry, or because the medication was topical (e.g., ear drops) so no
need to communicate
- Prescribers' not completing a certain task because of competing behaviour
- Prescribers' not performing medication reconciliation because they did not want to do extra effort

a
Definition adapted from Michie et al., (2011) (30).

549
D.F. Bannan et al.
interactions. I will double check the doses properly maybe once or twice,
[and] even at the end of the day I will check the patients again, all
medications expired or not. If I am very busy, I have a lot of sick patients,
I don't have time. I will try my best. I am not saying I will intentionally
[make mistakes], but by default because I will be too busy, I will do
mistakes—maybe I will forget to add the medications” P11, FG4
3.1. Capability
Insufficient skills and lack of knowledge were two factors that af-
fected prescriber behaviour and their ability to engage in error-free
prescribing processes. For example, prescriber ability to properly use
the EP software was hindered by prescribers' insufficient skills due to
improper training on the O- software system. The training was im-
proper because it was offered after the launch of the software and it was
described as ‘not structured’. Lack of knowledge was another factor that
affected prescribers' capability in making appropriate medical decisions
when prescribing. For example, lack of knowledge of the correct course
of action or the condition of a patient could have influenced prescriber
behaviour resulting in errors.
Participants discussed their own lack of knowledge as well as the
lack of knowledge of others, such as lower-grade doctors (in compar-
ison to more experienced doctors such as consultants), new staff coming
to the ward, and certain specialists, particularly surgeons and dentists
with ostensibly less knowledge about medications such as antibiotics.
However, the perception that these doctors lacked required knowledge
was based on participants' opinions and was not supported by any other
evidence.
“Unfortunately, many juniors [lower-grade doctors] have not got much
experience compared to you, or compared to others who are working
with you. They [junior doctors] think that it is like what they memorised,
as long as [the patient is] fungal afebrile, neutrophils recover, CT scan
negative, [then] stop antifungal.” P4, FG2
Lack of knowledge accompanied by other factors such as fatigue or
excessive workload (i.e. opportunity afforded in the environment) also
sometimes influenced prescriber behaviour. As new workdays in the
ward began at midnight, ordering processes were found to be particu-
larly problematic at night, when on-call doctors were often tired (i.e.
opportunity), treated multiple patients (i.e. opportunity) and were not
familiar with the condition of every patient (i.e. capability). As one
participant explained, even if a written progress note stated that a
medication had expired on the morning of a given day and thus there
was no need for a renewal, if a nurse asked for a renewal at night when
a doctor was already fatigued and asked, “please renew”, the doctor
might agree and “automatically renew, that's it.”
Apart from a lack of knowledge, some participants pointed out that
prescriber competency levels could cause errors. Regardless of work-
place environment, a few participants stated that errors should not
happen unless there was a problem with the prescriber, their compe-
tency, or their efforts.
“In reality, it [errors] should not happen because when you are trained
as a doctor, you go from junior to senior [and] you are told how to
respect roles, doses, methods of administration and all of that. You
cannot go beyond the dose, unless your training is a big problem, really.
OK. Or you recommend that acetaminophen [is] used once a day, for an
example, unless there is a big problem with you, ok? But paracetamol is
given in this dose 10-15mg/kg orally or as an IV. For example—you give
it every 4–6 hours for pain for fever for whatever. OK. This is the re-
commendation, but if you want to give it for something else, there is a
problem with you. Really. Who would do that? This, in reality, this
should not happen.” P4, FG4
In summary, participants identified that prescriber's ability to en-
gage in error-free prescribing was affected by insufficient skills, and
550
Research in Social and Administrative Pharmacy 15 (2019) 546–557
lack of knowledge among oncology prescribers, new staff, and certain
specialists.
3.2. Opportunity
The results showed that prescriber behaviour can also be influenced
by opportunities afforded in the environment, which helped to perpe-
tuate some behaviours that subsequently led to errors or made it dif-
ficult for less error-prone behaviour to occur. For example, access to
patient information at the time of prescribing is necessary to make an
informed decision. However, in some circumstances, prescribers did not
access information because it was either unavailable or difficult to re-
trieve in a timely fashion.
Participants reported that patient information was sometimes not
available because it was not documented in the first place. Prescribers
were not always able to document this information fully because
healthcare providers sometimes do not communicate with each other.
In addition, sometimes prescribers only partially document information
due to the way in which their software system was configured. For
example, the pre-set problem list in the EP software system offered no
space to add information, which sometimes resulted in a prescriber not
being able to document relevant information that were not included in
this list.
In other circumstances, information was available but prescribers
did not frequently look for it because it was difficult to retrieve in time.
For example, medication orders are available in both the O-software
and C-software systems. However, these orders are often spread across
these two different systems, which require prescribers to open both
systems every time they prescribe a medication, which does not always
happen. Some evidence suggests that when decisions were made based
on no access to information, errors occurred. For example, omission of a
treatment occurred because prescribers assumed that the drug was or-
dered in the other software system when in fact it was not, or dupli-
cation errors occurred because prescribers forgot that a medication had
been ordered in the other software system.
Guidelines were also common sources of information. However, the
content and availability of guidelines had an effect on access to this
information. The participants stated that they always used the custo-
mised chemotherapy protocols and guidelines for children, such as
those incorporated within the C-software system, and were one of many
safety checks that reduced the opportunity for errors. In contrast,
guidelines for supportive care medications were available but were not
commonly accessed by prescribers because these guidelines were lo-
cated in the outpatient clinic and were not incorporated in the O-soft-
ware system. Having to physically go and check these guidelines could
have made it more difficult to access information when needed.
“[Was the treatment of irinotecan induced diarrhoea different?] Yes, of
course. It is complex. Usually, to know what you should [do, as] a
supportive care of irinotecan, you have to read it all [the guideline], and
once you read it, you know how to deal with it [the condition]. They
[doctors] should go and read this, irinotecan supportive care. For me to
access this, I have to print it [out] from the out-patient [clinic].” P5, FG1
In addition, most of these guidelines contained information that was
not specific to children with cancer. This might have resulted in pre-
scribers not commonly accessing these guidelines, as they did not have
specific information for their patients (i.e. motivation).
Even when prescribers had physical access to these guidelines, the
lack of consensus on the use of the same guideline among doctors of
different specialities sometimes led to confusion between prescribers as
to which guidelines should be followed. This often resulted in following
specialist recommendations, even when they were not acceptable. One
observed example concerned a lack of consensus between OB/GYN
doctors and oncologists about the suitable method for menses sup-
pression for an 11-year old girl. The OB/GYN guidelines for menses
suppression did not take into account patients on chemotherapy or
D.F. Bannan et al.
possible drug interactions with chemotherapy drugs. As a result, fol-
lowing the recommendations of OB/GYN doctors based on the OB/GYN
guidelines resulted in prescribing a medication that was not suitable for
a cancer patient.
“We consulted the OB/GYN. The OB/GYN insisted [on giving] this
medication; they even convinced the mother to give it. We asked the
mother—we actually refused to discharge her [the patient] on this patch
[medication]. But they insisted, the OB/GYN, to give [it].” P13, FG2
Participants also stated that having prescriptions double-checking
by another prescriber (to check the work of the first prescriber) was a
recognised behaviour that could potentially reduce errors. However,
prescribers were not always able to perform this behaviour due to lack
of manpower and the frequent unavailability of a second senior doctor
in some areas. One prescriber highlighted that to be able to double
check prescriptions, other factors must also be present at the time of
ordering:
“The culture of double checking before each prescription needs time, less
interruptions, and accessible guidelines. For each medication you need to
prescribe, you need to have, like, easy, easy I mean, easy to search for
whatever you are looking for in the guidelines and you [have to] know
where to look in a quick way, and you [need] enough time to look for
that. And, less interruptions so you are not interrupted while prescribing
specifically highlighted medications or high-risk medications.” P16, FG4
Participants reported that interruptions were another opportunity
for errors, especially if prescribers had many tasks to finish (i.e. high
workload), which could divide their attention or interfere with the
successful completion of tasks. Rather than giving specific details,
prescribers talked about interruption in more general terms.
“There is no specific time, ah, for the physician to sit in peace without
interruption to write these medications—patient medications. It just
comes from the nurses' side that each time you are doing your job, you
are interrupted by something else to do.” P16, FG2
In summary, participants identified that incorrect prescribing and
errors occurred because they were unable to access information, their
high workload, they were frequently distracted, and the frequent un-
availability of a second senior doctor in some areas.
3.3. Motivation
Some of the prescribers' perceptions directed their decisions (i.e.
motivated them) to perform certain behaviours or engage in competing
behaviours. The relationship between these perceptions and prescriber
behaviour were discussed in the focus groups.
Performing medication reconciliation was accepted as an important
behaviour in reducing the opportunity for errors. However, not all
prescribers performed medication reconciliation to the same degree.
Some prescribers thoroughly checked all patients' medications, whereas
others did not ask patients about all medications taken because pre-
scribers either did not have time or did not make an effort to ask pa-
tients. As one participant explained:
The results also showed that it was common practice to ask for
consultations from other specialists when patient conditions were out-
side a doctor's scope of expertise. Prescriber perceptions about the
knowledge of the specialist contacted or the reaction of the specialist
when contacted affected prescriber behaviour. In fact, prescribers
sometimes followed recommendations that were not appropriate for a
patient because they over-trusted a specialist's level of knowledge.
“The problem [the medication involved with error discussed] is that the
medication is their speciality. So, if something interferes with my
551
Research in Social and Administrative Pharmacy 15 (2019) 546–557
speciality, I will know better than them, I will just question it. But if it's
their speciality and they are the one prescribing!” N9, FG3
Over-trusting specialists was influenced by the grade of the spe-
cialist and the type of medication recommended. When the impact of a
specialist's grade was discussed, not all prescribers believed that it af-
fected their decisions. However, one high-grade prescriber believed that
if a peer was consulted, they were required to follow their re-
commendations unless something was obviously wrong with these re-
commendations.
“This will influence me very much, I believe. Not a personal relationship,
some colleagues working with you, you know him, ok. And you know
that he is almost perfect in everything, [so] when he tells me about a
medication [to] do it this way, I do it.” P4, FG4
The type of medication recommended by specialists could also in-
fluence prescriber decisions in terms of whether or not they follow these
recommendations. In a case of a topical medication that was contra-
indicated, yet prescribed, one participant was asked if they would ac-
cept the recommendations by others as guaranteed, they replied:
“If a specialist recommends, for example, systemic steroids, I will not
take it. I will first discuss [it] with them. But, if an ENT prescribes local
[topical medication], definitely I will trust them and I will not go and
check.” P9, FG3
In addition, the reaction of a specialist when contacted by oncology
prescribers had affected prescriber behaviour. Prescribers were hesitant
and sometimes decided not to contact specialists because they felt that the
specialists could become agitated if contacted for further clarification.
“Sometimes they [doctors from other teams] get angry. We wrote our
recommendation, why are you calling me now. You see [it] in the
computer. They will tell you like this.” P11, FG4
Finally, the opportunity to pursuing competing behaviours at the
time of performing the behaviour of interest sometimes motivated
prescribers to prioritise tasks they had to perform within given time
frames. This could have resulted in an original behaviour of interest not
being performed or being only partially performed. For example, pre-
scribers stated that treating patients and stabilising their condition as
their priorities, particularly during on call periods, and shifted the re-
sponsibility of ordering additional medications or completing doc-
umentation to other teams. As one participant explained:
“In admission and discharge, it's probably related to a patient coming
[in] as an emergency or when you are dealing with certain emergency,
and given the medication for a certain emergency, and you think or push
that responsibility to the morning team, which is wrong. OK. But [it is]
like this where a miss can happen. Or you are just dealing with an
emergency right now, [so] admit him, get settled, get over the fever
neutropenia, then all his day medications can happen with the morning
team and they can sort it out. That is pushing [off] the problem.” P1,
FG4
Some evidence showed that heavy workloads and having large
numbers of patients to treat were reasons that prescribers prioritised
necessary tasks. For example, when busy, prescribers stated that they
focused on immediate concerns and left other problems for later:
“You postpone until you are free—then you are a human being. There
are other things to do, you are too busy and you forget to document. You
think you've done it, but you haven't done it.” P4, FG4
In summary, participants' over-trusting other prescribers or specia-
lists, and prioritising tasks to perform were some of the motivations that
could cause incorrect prescribing and errors.
D.F. Bannan et al.
4. Discussion
The COM-B model facilitated an understanding of how prescriber
behaviour resulted in PEs. By using this model, we were able to describe
how prescriber behaviours had been impacted upon by their cap-
abilities, by opportunities in the environment, or by how they were
motivated to behave. This study sheds light on two novel aspects re-
lated to prescribing errors: 1) the use of the COM-B model to under-
stand the causes of PEs, and 2) the behaviours of senior prescribers that
led to PEs.
The COM-B model30 is becoming a well-recognised approach to
understanding and changing behaviour – in mid-2018 it had been cited
over 500 times in articles indexed on PubMed. It has advantages over
other models used to understand causes of PEs, as it enabled the re-
searchers to identify behaviours that led to errors and explain the in-
terrelationship between contributory factors and behaviours expressed.
This in-depth understanding of the sources of the problem allowed the
researcher to understand why errors occurred, which could provide a
foundation for the development of an intervention.
Using the COM-B model also showed that prescribing error is a
complex problem and that often multiple sources of behaviour were
affected. For example, prescriber lack of knowledge of a patient (affects
capability) was sometimes accompanied by prescriber fatigue, heavy
workload, and presence of multiple tasks to complete (opportunities in
an environment). This suggests the importance of using the model in its
entirety, rather than considering only single constructs. Thus, inter-
ventions designed to reduce errors should consider the complexity of
this problem and the different sources of behaviour that could be tar-
geted, as different sources can be targeted by different intervention
functions. For example, to improve prescriber knowledge of patients
(i.e. psychological capability), interventions that include education,
training, or enabling functions could be used.30,48 However, to affect
behaviour when prescribers are busy and fatigued (i.e. physical op-
portunity), interventions that incorporate training, restriction, en-
vironmental restructuring, or enabling functions could be used.30,48
According to the literature, interventions used to reduce PEs often
have similar functions, but little information appears on whether the
selection of intervention functions was made based on a systematic
understanding of sources of behaviour. For example, a systematic re-
view that used the Behavioural Change Wheel as a framework for de-
scribing the content of 17 bundle interventions used to reduce PEs in
hospitalised children found that all 17 bundles contained an interven-
tion with an environmental restructuring function and 16 bundles
contained an intervention with an educational function.49 Environ-
mental restructuring is an intervention function aimed at physical/so-
cial opportunities and automatic motivation, whereas education is a
function aimed at psychological capability and reflective motivation.
Although a bundle of interventions that addressed these two functions
would cover almost all sources of behaviour (except physical cap-
ability), PEs still occur in settings where these interventions were im-
plemented. Thus, one could hypothesise that targeting only relevant
sources of behaviour, rather than a broad range, would have a greater
impact on incorrect prescribing and errors; this hypothesis needs to be
tested. However, to date limited research exists on interventions de-
signed based on identified sources of incorrect prescribing. Under-
standing the causes of errors and then designing interventions based on
behavioural sources to be changed could help to test this hypothesis.
Causes of PEs in hospitalised patients have been investigated, and
the majority of the studies have looked at causes of errors made by
doctors who were one or two year's post-qualification.17,50 Causes of
errors made by senior doctors in specialised areas, such as paediatrics,
have rarely been investigated. While one study in a large UK hospital
552
Research in Social and Administrative Pharmacy 15 (2019) 546–557
did investigate the causes of PEs made by both senior and junior doc-
tors,51 any differences in causes of errors between doctors of different
grades were not reported. The contributory factors to errors identified
in our study are similar to those reported elsewhere in the literature in
terms of prescribing for both adults and children. Examples of con-
tributory factors include lack of knowledge and training, poor com-
munication, poor access to guidelines, heavy workload, and interrup-
tions.14,17,52
On the other hand, the behaviours expressed by prescribers that
result in prescribing errors have rarely been explicitly described in the
literature. It may be possible to infer behaviours from the literature on
causes of errors. However, this may not necessarily be transferrable to
the context of senior prescribers because, as previously stated, such
research has been conducted with the most junior doctors. The effect of
the contributory factors on behaviour may vary for senior and junior
doctors. For example, for both senior and junior doctors, interruption is
a contributory factor that can serve as an opportunity for error.16
However, it is not known if interruptions affect the behaviour of these
two groups to the same extent. In addition, the relative lack of research
focusing on the behaviour of different grades of prescribers that re-
sulted in errors makes it difficult to compare findings from this study to
other studies.
The study has both strengths and limitations. Focus groups sessions
were used to explore the contributory factors of incorrect prescribing
and the associated behaviours that potentially resulted in PEs. The
dynamics and interactions between participants in discussion were
useful in clarifying perspectives, something the researchers would not
have been able to achieve with individual interviews. In addition, it is
possible that presenting examples of real errors that prescribers could
relate to may have improved their engagement in the process and en-
riched the data.
The study also has a few limitations. First, the study was conducted
in a single specialised ward, with only a small group of prescribers from
which to draw the study sample. Second, the study involved only senior
doctors, as junior doctors were not allowed to prescribe. Junior doctors
might be authorised to prescribe in other settings in SA or elsewhere,
and they might express different behaviours than those expressed by
senior doctors. Therefore, these findings may not be generalisable to
that group of doctors. Third, one researcher coded the transcripts.
However, the researchers discussed this analysis repeatedly and any
differences of interpretation were resolved. Fourth, the design of the
focus groups might have affected the openness of the participants to
talk about prescribing behaviours. Although participants were all senior
doctors, they were of different grades, such as staff physicians and
consultants. Having this mixture of doctors with different experience
might have affected the openness of consultants to admit their lack of
knowledge in front of staff physicians, and vice versa. Finally, the
clinical pharmacist who collected error data (AA) was the clinical
pharmacist on charge of monitoring patients' treatment at the ward.
The close relationship between the participants and the clinical phar-
macist (i.e. the clinical pharmacist being part of the oncology team)
might have affected the openness of participants to talk about causes of
errors made. However this effect was not assessed.
In conclusion, we believe that the use of COM-B helped to identify
the causes of PEs from a new perspective. Future work could focus on
mapping identified sources of behaviour and errors against intervention
functions and policies in order to design an effective intervention.
Declarations of interest
None.
D.F. Bannan et al. Research in Social and Administrative Pharmacy 15 (2019) 546–557

Appendix A

Focus groups 1–3 error scenarios ‘cases’

Box 1

Cases presented at focus group 1.

During data collection 44 errors related to the ‘need of drug therapy’ were found. The following are examples. We are interested in
understanding why these errors or errors like these could have occurred and what could be done to prevent their recurrence.
Case 1: An 11-year old male with Hodgkin Lymphoma suffering from distal right internal jugular vein thrombosis admitted for
thrombosis treatment. The patient has a history of insulin resistance, glucose-6-phosphate dehydrogenase deficiency, and obesity. On
admission home medication (Metformin 500 mg twice a day; Budesonide inhaler 1 puff twice a day; and Salbutamol inhaler as needed)
were not prescribed.
Case 2: A 3-year old male with acute lymphoblastic leukaemia was on Bactrim 240 mg orally every 12 h two days/week for pneu-
mocystis pneumonia prophylaxis. The drug was put on hold during high dose Methotrexate treatment. The drug remained on hold for six
weeks.
Case 3: A 12-year old female with diffuse large B-cell lymphoma on percutaneous transhepatic cholangiography drainage with fluid and
wound infection. The antifungal was discontinued based on negative fungal search by Computerized Tomography scan, although patient
has fungal infection with candida glabrata from wound and fluid drain.
Case 4: A 2-year old female with acute lymphoblastic leukaemia suffering from lice was prescribed Permethrin topical once weekly with
no end date or instructions.
Case 5: A 3 and half year old male with monoclonal B-cell lymphocytosis suffering from tooth abscess was given Tazocin and cefazolin.
These antimicrobials have the same coverage.
Case 6: A 12-year old female with Burkitt lymphoma was started on a regimen for non- Hodgkin's lymphoma. Pre-medications (pre
Rituximab) were not included in the patient Performa and no justification for not prescribing was given in the medical record.
Case 7: An 11-year old female with acute lymphoblastic leukaemia (post paediatric intensive care unit discharge) was on
Acetylcysteine. Bronchodilator was not prescribed prior to Acetylcysteine.
Case 8: A 7-year old male with lymphoma on chemotherapy. The patient was on Methylprednisolone 21 mg intravenous every 24 h with
no gastrointestinal prophylaxis.
Case 9: A 10-year old male with pelvic synovial sarcoma on Doxorubicin and Ifosfamide was not prescribed an antiemetic.
Case 10: A 13-year old male with rhabdomyosarcoma suffering from diarrhea (6 days after he received a dose of Voriconazole/
Irinotecan) was given Metronidazole 400 mg intravenous every 8 h.

Box 2

Cases presented at focus group 2.

During data collection 88 errors related to the ‘selection of dosage regimen’ were found. The following are examples. We are interested in
understanding why these errors or errors like these could have occurred and what could be done to prevent their recurrence.
Case 1: A 6-year old male with mediastinal mass (T-cell acute lymphoblastic leukaemia) is on Prednisolone 23 mg orally every 24 h for
acute lymphoblastic leukaemia induction and Ranitidine 36 mg orally every 24 h for gastrointestinal prophylaxis. Doses were divided
incorrectly.
Case 2: A 7-year old female with acute lymphoblastic leukaemia was prescribed Voriconazole 160 mg orally every 24 h as prophylaxis
from fungal infection.
Case 3: A 9-year old male (actual weight.: 51.5 kg, ideal body weight: 32 kg) with acute lymphoblastic leukaemia suffering from febrile
neutropenia and septic shock was prescribed Amikacin 380 mg intravenous every 8 h (correct dose 7.5 mg/kg).
Case 4: A 10-year old male (surface area > 1m2) with acute lymphoblastic leukaemia was prescribed Bactrim 7.5 ml (360 mg) for
pneumocystis pneumonia prophylaxis.
Case 5: A 13-year-old male with acute lymphoblastic leukaemia was prescribed Caspofungin 125 mg intravenous every 24 h as pro-
phylaxis, the dose was then changed to 90 mg intravenous every 24 h, 88 mg intravenous every 24 h, and finally reduced to 64 mg in-
travenous every 24 h (maintenance dose is 50mg/m2/day with maximum of 50 mg).
Case 6: A 12-year old male (surface area = 1.16m2) was prescribed dexamethasone 4 mg intravenous every 8 h as pre-chemo an-
tiemetic. Patient was also on high emetogenic chemotherapy (Aprepitant +5H3 receptor anatagonist + steroid).
Case 7: A 19-month old male (weight = 9.5 kg) with acute lymphoblastic leukaemia suffering from constipation was given Lactulose
2.5 ml.
Case 8: An 11-year old female with osteosarcoma was given Evra 1 patch weekly for menses suppression. Estrogen and patch for-
mulation are not recommended for oncology patient due to side effects and chemotherapy interaction but the gynaecologist insisted on the
patch.

553
D.F. Bannan et al. Research in Social and Administrative Pharmacy 15 (2019) 546–557

Box 3

Cases presented at focus group 3.

During data collection 11 errors related to the ‘Selection of a specific drug’ and 14 errors related to wrong drug choice were found. The
following are examples. We are interested in understanding why these errors or errors like these could have occurred and what could be
done to prevent their recurrence.
Case 1: An 11-year old male with T-cell acute lymphoblastic leukaemia was prescribed Amikacin 249 mg intravenous every 8 h al-
though the patient was labelled Gentamicin allergic. Evidence showed that if it is true allergy for Gentamicin, then Amikacin will have 50%
chnace of cross reactivity as all aminoglycosides shows strong structural similarities.
Case 2: A two and half year old male with langerhans cell histiocytosis was prescribed Waxol 3 drops every 8 h for wax removal after
consultation with an otolaryngologist. The use of Waxol is contra-indicated in case of ear perforation and inflammation.
Case 3: An 8-year-old male with acute myeloid leukaemia was given Etoposide with Fluconazol. This might reduce the execretiion of
Etoposide and increase side effects.
Case 4: A 4-year-old male was given Etoposide with Carbamazepine. This might alter metabolism and decrease level or effect of
Etoposide.
Case 5: An 11-year-old female with acute lymphoblastic leukaemia on Voriconazol was given Omeprazole. This might increase the level
of Omeprazole and side effects.
Case 6: A 9-year-old male with relapsed acute lymphoblastic leukaemia was given Tazocin for Salmonella eradication.
Case 7: A 4-year-old male was prescribed Calcium Carbonate as Phosphate binder. Calcium level was 2.44, Phosphate level was 2.22,
Calcium x Phosphate: 5.4. the chnaces of precipitation and oral calcification was high.

Appendix B

Focus groups 1–3 topic guide

No one knows the real cause of a prescribing error more than the prescribers. The purpose of the focus group is to explore your perspective of the
cause(s) of prescribing errors intercepted in prescriptions written in the ward. The focus group will last approximately 60–90 min and will be audio-
recorded. All information discussed during this focus group will be treated in confidence. No one outside the research team will know what you have
said or that you are participating in the study. Data will be analysed anonymously and any identifying information including names of anyone
mentioned during the focus group will be removed.
Please do not identify yourself as responsible for the error and do not lay blame for the error on your colleagues. Please respect the confidentiality
of your colleagues, as they may be sharing examples of errors that they have been aware of. The purpose is to understand why the errors occur and
learn from the errors so that these errors do not happen again.
The focus group will be divided into three parts: The first part is about the prescribing system in the hospital and the next two parts are about
examples of the common types of prescribing errors on the ward and not the errors themselves.

Part one

The focus of this part is talk in general about the prescribing system in the hospital to get prescribers' view and perception of the prescribing
process (system) and if the process contributes to the occurrence of errors.

• Tell me about the process for medication prescribing at your hospital?

• Do you think the prescribing process contributes to the occurrence of error? [probe as necessary, e.g. In what way?]
• Why else do you think prescribing errors occur?
Part two

This part of the focus group will focus on the error(s) identified in phase 1. The same process in part two will be repeated for each type of
error identified in phase 1.

• I would like to talk to you about an example of an error we identified in one of the prescriptions written in the ward. The error was … … … …..
• From your knowledge and experience, why do you think that error occurred?
• What factors do you think contributed to the occurrence of errors like that?
• What do you think could've been done differently to prevent the error happening?
• Is this type of error something you usually encounter?

554
D.F. Bannan et al. Research in Social and Administrative Pharmacy 15 (2019) 546–557

Further questions will be asked according to the case and the answers of the participants to get as much information as possible about the causes
of prescribing errors.

Part three

This part focuses on safety measures/system improvements prescribers believe could mitigate prescribing errors in their setting.

• What safety measures could be implemented to reduce prescribing errors such as the ones you made?
• What systems improvements can be used to make the prescribing process safer?
• Is there anything else you want to add?
Thank you for participating, your time is appreciated.
(Stop recording)

Appendix C

Focus group 4 topic guide

No one knows the real cause of a prescribing error more than the prescribers. The purpose of the focus group is to explore your perspective of the
cause(s) of prescribing errors intercepted in prescriptions written in the ward. The focus group will last approximately 60–90 min and will be audio-
recorded. All information discussed during this focus group will be treated in confidence. No one outside the research team will know what you have
said or that you are participating in the study. Data will be analysed anonymously and any identifying information including names of anyone
mentioned during the focus group will be removed.
Please do not identify yourself as responsible for the error and do not lay blame for the error on your colleagues. Please respect the confidentiality
of your colleagues, as they may be sharing examples of errors that they have been aware of. The purpose is to understand why the errors occur and
learn from the errors so that these errors do not happen again.
The focus group will be divided into three parts: The first part is a PowerPoint presentation about the preliminary results from focus groups 1–3
and an introduction about the concepts of capability, opportunity, and motivation. The second part is about what enable/disable you from pre-
scribing properly in terms of capability, opportunity and motivation.

Part one
PowerPoint presentation (Prescribing errors in the Paediatric Oncology Ward).

Part two
Q: In terms of individual capability, what do you think are the barriers that affect the ability of doctors on the unit to prescribe properly?
[Prompts: knowledge, physical and psychological skills)
Q: What do you think are the opportunities in the environment that makes errors easy to happen?
[Probes: time, resources, more staff, support from others]
Q: What about motivation? What do you think are the factors that motivate doctors to prescribe the way they prescribe? Or discourage doctors
from doing some tasks?
When we analysed the data, the majority of errors were related to supportive meds rather than chemotherapy meds and the majority
was dosing errors.
Q: Why do you think that is?
One of the contributory factors that physicians talked about was the transition from one electronic prescribing system to another
without being properly trained to use it.
Q: Do you think not knowing how to use the system affected your ability to prescribe properly?
Q: Is there any other example were you were not trained properly?
Q: what other tasks related to prescribing do you think doctors should have training or more training?
You talked about some errors that occurred during admission and discharge because of problems with the medication reconciliation or
patient counselling.
Q: How much do you think it is needed to check all patients' medication (chemo and others) upon admission and discharge?
Q: Do you think this is part of the job of the doctor? [Probes; who's job it is?]
Q: Why do you think this job is not being done?
Q: Doctors may be discouraged to check all patients' medications upon admission or discharge, why do you think that is?
[Prompts: heavy workload, not having enough time]
Q: What other factors affect you?
Q: What would encourage doctors to spend more time on medication reconciliation?
Some of you talked about communication with specialists.
Q: Are there particular problems prescribing after a consultation by someone else?
Q: Does the name or level (e.g., consultant vs. resident) of the person who made the consultation affect your decision to whether you follow the
recommendations or not?
Q: Do you think maintaining the relationship with other doctors has an effect on the doctor's decision to follow the recommendations?

555
D.F. Bannan et al.
Q: How can you improve the communication with doctors from other specialities?
Some of you mentioned the lack of knowledge of doctors from other specialities (e.g., surgeons and dentists) especially when it comes
to antibiotics.
Q: When the other doctor makes the consultation, do you know at that point if the order is correct or has an error?
Q: If you know, do you communicate with the other doctor? Why not?
Q: How much do you think the lack of knowledge of others contributed to errors made in the oncology ward?
Some of you raised the issue of poor documentation.
Q: What hinders your ability to document everything you did?
Q: What sort of an impact does poor documentation have on how you work?
You have said that some of the tasks that the doctors do every day are tedious and time consuming such as re-entering and renewing
orders in the system every day or every few days for all the patients in the ward.
Q: How much do you feel you need to do this task?
Q: How can you improve this process (renewing orders)?
Q: What other tasks you think is time consuming?
Just before we finish, can I ask a few general questions:
Q: What tasks do you think doctors should focus on to prescribe effectively and properly?
Q: What are the environmental factors surrounding doctors that contribute to error occurring or affect the way doctors prescribe?
Q: What do you think can be done to improve the prescribing process?
Thank you for participating, your time is appreciated.
(Stop recording)
References
1. Michie S, Atkins L, West R. The Behaviour Change Wheel: A Guide to Designing
Interventions. London: Silverback Publishing; 2014.
2. Rouse WB. Managing complexity: disease control as a complex adaptive system. Inf
Knowl Syst Manag. 2000;2:143–165.
3. Pediatric pharmacy practice guidelines. Pediatric pharmacy administrative group
committee on pediatric pharmacy practice. Am J Hosp Pharm. 1991;48:2475–2477.
4. Gupta A, Waldhauser LK. Adverse drug reactions from birth to early childhood.
Pediatr Clin. 1997;44:79–92.
5. NHS.England. 2013. [cited 2017 June 10]. Available from: https://www.england.
nhs.uk/wp-content/uploads/2013/06/b15-cancr-chemoth-child-teen-yng-adul.pdf.
6. Aspden P, Wolcott J, Bootman JL, Cronewett LR. Preventing medication errors:
quality chasm series [Internet]. Washington, DC. . [cited 2015 Aug 7]. Available
from:. https://www.nap.edu/read/11623/chapter/2-p20010c2f9980001; 2007.
7. Dean B, Barber N, Schachter M. What is a prescribing error? Qual Health Care.
2000;9(4):232–237.
8. Lewis PJ, Dornan T, Taylor D, Tully MP, Wass V, Ashcroft DM. Prevalence, incidence
and nature of prescribing errors in hospital inpatients: a systematic review. Drug Saf.
2009;32:379–389.
9. Burmester MK, Dionne R, Thiagarajan RR, Laussen PC. Interventions to reduce
medication prescribing errors in a paediatric cardiac intensive care unit. Intensive
Care Med. 2008;34:1083–1090.
10. Alagha HZ, Badary OA, Ibrahim HM, Sabri NA. Reducing prescribing errors in the
paediatric intensive care unit: an experience from Egypt. Acta Paediatr.
2011;100:e169–e174.
11. Aseeri MA. The impact of a pediatric antibiotic standard dosing table on dosing er-
rors. J Pediatr Pharmacol Therapeut. 2013;18:220–226.
12. Al-Jeraisy MI, Alanazi MQ, Abolfotouh MA. Medication prescribing errors in a pe-
diatric inpatient tertiary care setting in Saudi Arabia. BMC Res Notes. 2011;4:294.
13. Reason J. Human Error. Cambridge: Cambridge University Press; 1990.
14. Tully MP, Ashcroft DM, Dornan T, Lewis PJ, Taylor D, Wass V. The causes of and
factors associated with prescribing errors in hospital inpatients: a systematic review.
Drug Saf. 2009;32:819–836.
15. Dornan T, Ashcroft DM, Heathfield H, et al. An in depth investigation into causes of
prescribing errors by foundation trainees in relation to their medical education.
EQUIP study. 2009. [cited 2016 Feb 9]. Available from. http://www.gmc-uk.org/
FINAL_Report_prevalence_and_causes_of_prescribing_errors.pdf_28935150.pdf.
16. Ryan C, Ross S, Davey P, et al. Junior doctors' perceptions of their self-efficacy in
prescribing, their prescribing errors and the possible causes of errors. Br J Clin
Pharmacol. 2013;76:980–987.
17. Lewis PJ, Ashcroft DM, Dornan T, Taylor D, Wass V, Tully MP. Exploring the causes
of junior doctors' prescribing mistakes: a qualitative study. Br J Clin Pharmacol.
2014;78:310–319.
18. Payne CH, Smith CR, Newkirk LE, Hicks RW. Pediatric medication errors in the
postanesthesia care unit: analysis of MEDMARX data. AORN J. 2007;85:41–44 731-
40; quiz.
19. Becker MH. The Health Belief Model and Personal Health Behavior. Thorofare, NJ:
Charles B. Slack; 1974.
20. Janz NK, Becker MH. The health belief model: a decade later. Health Educ Q.
1984;11:1–47.
21. Rosenstock IM. Historical origins of the health belief model. Health Educ Monogr.
1974;2:1–8.
22. Bandura A. Self-efficacy: toward a unifying theory of behavioral change. Psychol Rev.
1977;84:191–215.
556
Research in Social and Administrative Pharmacy 15 (2019) 546–557
23. Bandura A. Social Foundations of Thought and Action: a Social Cognitive Theory.
Englewood Cliffs, NJ: Prentice Hall; 1986.
24. Bandura A. Self-efficacy: the Exercise of Control. New York: Freeman; 1997.
25. Fishbein M, Ajzen I. Belief, Attitude, Intention and Behavior: An Introduction to Theory
and Research. Reading, MA: Addison-Wesley; 1975.
26. Ajzen I, Fishbein M. Understanding Attitudes and Predicting Social Behavior. Englewood
Cliffs, NJ: Prentice-Hall; 1980.
27. Nicholson SL, Donaghy M, Johnston M, et al. A qualitative theory guided analysis of
stroke survivors' perceived barriers and facilitators to physical activity. Disabil
Rehabil. 2014;36:1857–1868.
28. Taylor N, Lawton R, Conner M. Development and initial validation of the determi-
nants of physical activity questionnaire. Int J Behav Nutr Phys Activ. 2013;10:74.
29. Honigh-de Vlaming R, Haveman-Nies A, Heinrich J, van't Veer P, de Groot LC. Effect
evaluation of a two-year complex intervention to reduce loneliness in non-in-
stitutionalised elderly Dutch people. BMC Publ Health. 2013;13:984.
30. Michie S, van Stralen MM, West R. The behaviour change wheel: a new method for
characterising and designing behaviour change interventions. Implement Sci.
2011;6:42.
31. Jackson C, Eliasson L, Barber N, Weinman J. Applying COM-B to medication ad-
herence. Eur. Health Psychol. 2014;16:7–17.
32. Elrouby S, Tully MP. Using the behaviour change wheel to identify interventions to
facilitate the transfer of information on medication changes on electronic discharge
summaries. Res Soc Adm Pharm. 2017;13:456–475.
33. Fulton EA, Brown KE, Kwah KL, Wild S. StopApp: using the behaviour change wheel
to develop an app to increase uptake and attendance at NHS stop smoking services.
Healthcare (Basel). 2016:4.
34. Handley MA, Harleman E, Gonzalez-Mendez E, et al. Applying the COM-B model to
creation of an IT-enabled health coaching and resource linkage program for low-
income Latina moms with recent gestational diabetes: the STAR MAMA program.
Implement Sci. 2016;11:73.
35. Barker F, Atkins L, de Lusignan S. Applying the COM-B behaviour model and be-
haviour change wheel to develop an intervention to improve hearing-aid use in adult
auditory rehabilitation. Int J Audiol. 2016;55(3):S90–S98.
36. Curtis KE, Lahiri S, Brown KE. Targeting parents for childhood weight management:
development of a theory-driven and user-centered healthy eating app. JMIR Mhealth
Uhealth. 2015;3:e69.
37. Dyson J, Lawton R, Jackson C, Cheater F. Does the use of theoretical approach tell us
more about hand hygiene behaviour? the barriers and levers to hand hygiene. J.
Infect. Prev. 2011:12.
38. Jeffe DB, Dunagan WC, Garbutt J, et al. Using focus groups to understand physicians'
and nurses' perspectives on error reporting in hospitals. Joint Comm J Qual Saf.
2004;30:471–479.
39. Kitzinger J. Qualitative research. Introducing focus groups. BMJ.
1995;311(7000):299–302.
40. Kitzinger J. The methodology of focus groups: the importance of interaction between
research participants. Sociol Health Illness. 1994;16:103–121.
41. Morgan DL. The Focus Group Guide Book. London: Sage Publication; 1998.
42. Bloor M, Frankland J, Thomas M, Robson K. Focus Groug in Social Research. London:
Sage Publication; 2001.
43. Stewart DW, Shamdasani PM. Theory and Practice. London: Sage Publication; 1990.
44. Braun V, Clarke V. Using thematic analysis in psychology. Qual Res Psychol.
2006;3:77–101.
45. Braun V, Clarke V. Successful Qualitative Research a Practical Guide for Biginners. UK:
SAGE; 2013.
46. Lorencatto F, West R, Michie S. Specifying evidence-based behavior change
D.F. Bannan et al.
techniques to aid smoking cessation in pregnancy. Nicotine Tob Res.
2012;14:1019–1026.
47. Behaviour Change: Individual Approaches [Internet]. NICE. 2014; 2014.
48. Michie S, editor Behaviour Chnage Techniques and a Framework for Increasing
Physical Activity ISBNPA Satellite Conference; 2013; Cambridge.
49. Bannan DF, Tully MP. Bundle interventions used to reduce prescribing and admin-
istration errors in hospitalized children: a systematic review. J Clin Pharm Therapeut.
2016;41:246–255.
557
Research in Social and Administrative Pharmacy 15 (2019) 546–557
50. Ross S, Ryan C, Duncan EM, et al. Perceived causes of prescribing errors by junior
doctors in hospital inpatients: a study from the PROTECT programme. BMJ Qual Saf.
2013;22:97–102.
51. Dean B, Schachter M, Vincent C, Barber N. Causes of prescribing errors in hospital
inpatients: a prospective study. Lancet. 2002;359:1373–1378.
52. Coombes ID, Stowasser DA, Coombes JA, Mitchell C. Why do interns make pre-
scribing errors? A qualitative study. Med J Aust. 2008;188:89–94.