Medication management errors are defined by Shawahna et al. (6) as deviations from that prescribed in the patient’s medical records, deviations from the manufacturer’s recommendations for storage, expiry date, preparation and administration or deviations from relevant institutional policies. More than 80 per cent of the errors are linked to assembling, preparing and dispensing medications to patients (7).
Medication management errors can be caused by outdated practices, poor routines, the absence of procedures or skills, carelessness or lack of knowledge (8). Simonsen et al. (7) refer to surprising weaknesses in basic knowledge on medication management among nurses in the specialist health service and primary health care. These weaknesses represent a significant potential for error.
Responsibilities in medication management
Healthcare service managers are responsible for development, implementation and compliance in relation to procedures and instructions for medication management. They must also ensure that healthcare personnel have the necessary knowledge and skills within medication management (1). Nevertheless, many studies, both national and international, identify a major need to improve medication management routines (9, 10) and internal controls (11).
Safety procedures are not always possible to carry out in practice, and there is a mismatch between medication competence, work tasks and staffing (11). Healthcare personnel are often interrupted when preparing and administering medication and when monitoring patients’ reactions. The employer’s expectations within medication management are also unclear sometimes (9). Improving competence and dealing with deviations are vital to reducing the risk of errors (7, 11–14).
Pharmaceutical advice and medication audits can reduce the risk of errors in medication management, but Circular IS-7/2015 does not clarify how the advisory aspect should work in practice (10). Role clarification or clarifying responsibilities, establishing interdisciplinary teams and increased knowledge of medication management are among the measures covered.
According to the report on appropriate medication use for elderly patients/residents in nursing homes and in the community nursing service (15), the measures can contribute to optimum use of medicines in primary health care. However, the report has a limited focus on the practicalities entailed in the actual handling of medications in the nursing and care service.
An updated and harmonised list of medications used should always accompany patients when the level of care is changed (3). The literature shows that the manual exchange of information on medications can represent a threat to patient safety (16). An electronic list of medications is a recognised way of safeguarding the quality of information on medications and increasing patient safety (17), but there is still scope for improvement in terms of the electronic exchange of information in Norway (18).
Objective of the study
The objective of the study was to present the practical challenges of medication management in the nursing and care service, and to establish a consensus on how to address these challenges.
The Delphi method involves a consensus processes and associated questionnaire development. This method is a group communication process aimed at conducting detailed surveys and discussions on a particular topic with a view to reaching a consensus (19). The method has been used in a number of surveys, both nationally and internationally (6, 20).
Typical surveys seek to examine ‘what is’, while the Delphi method tries to establish ‘what should or may be’. The Delphi method entails several rounds of questionnaires (19). Researchers collect and analyse data from a questionnaire, and then create a new questionnaire, which is sent to the same respondent group, normally accompanied by the results from the previous round.
The questionnaires are answered anonymously and the respondents do not meet each other, thus preventing group influence during the process. Selecting questions and which indicators should be considered can be a challenge for researchers.
Consensus achieved through this method entails agreement or concurrence in opinions and attitudes between qualified experts within a defined field. Clear criteria should therefore be drawn up for selecting respondents, often referred to as the group or panel of experts (21).
First, we retrieved the e-mail addresses of chief executives and heads of department from the websites of 26 local authorities in one county. They were all sent an information letter about the study and permission was requested to conduct the survey. Those we contacted were also asked to provide the names and e-mail addresses of charge nurses or healthcare service managers at all relevant municipal services, such as nursing homes, sheltered housing and the community nursing service.
We received 17 positive replies, and subsequently sent letters to the healthcare service managers and charge nurses in the 17 local authorities. The letter asked the recipients to convey information about the study to all nurses and social educators at their institution.
We requested permission to conduct the study among staff at a pharmacy with a regulatory role within pharmaceutics in the nursing and care sector. We collected 183 e-mail addresses of nurses, social educators, pharmacists and department heads.
The study was approved by the Norwegian Centre for Research Data, project number 34428. Participation in the study was voluntary. The respondents were informed that they could reserve the right not to receive the questionnaires and withdraw from the study at any time. The data were stored in accordance with research ethics guidelines and the Declaration of Helsinki (22).
Preparing the questionnaire and data collection
The starting point for the questionnaire was a list of the 161 most common medication management problems in nursing homes and in the community nursing service. The list was based on regulatory reports covering more than half of the municipalities in one county in the period 2008–2012. We further refined this list and discussed it with the reference group, which consisted of a researcher and a manager from the municipal nursing and care service.
We formulated 65 statements relating to medication management, which we used to create the questionnaire. Twenty-seven of these statements were used in round one, 21 in round two and 17 in round three. In addition to the initial statements, we devised new statements as a result of the analysis of comments received in rounds one (28 statements) and two (15 statements).
The respondents gave responses to 27, 49 and 32 statements in rounds one, two and three respectively. In the statements, we chose to use both ‘should’ and ‘must’, which is in line with the wording of the regulations on medication management and the associated circular. The respondents were asked to indicate the extent to which they agreed with the statements using a five-point Likert scale with the following values: ‘Completely agree’, ‘Partly agree’, ‘Neither agree nor disagree (neutral)’, ‘Partly disagree’ and ‘Completely disagree’. Alternatively, they could select ‘Don’t know’.
In the first questionnaire round, respondents could comment on each statement, while in the second and third rounds we asked for comments on statements to be assembled thematically. The questionnaire also included questions about work experience, education and function in the organisation. We used the Questback program to design and distribute the questionnaire.
We first conducted a pilot test of the questionnaire in one of the municipalities. The questionnaire was sent to six nurses, and four responded. The respondents were asked to assess whether the statements were understandable and whether potential difficulties could arise when answering the questionnaire.
No relevant issues were identified in the pilot test, and only small, linguistic adjustments were made after consulting the reference group. We conducted the survey in three rounds in the period November 2013 to May 2014. In all three rounds, we sent reminders after one week of sending the first e-mail, and another after two weeks.
We produced descriptive statistics using the analysis tool IBM SPSS Statistics 22.
Since the data were not normally distributed, we used the Mann-Whitney U test (non-parametric test) (23) to compare the responses from different respondent groups. P-value <0.05 was considered to be statistically significant.
Different definitions of consensus have been used in various Delphi studies (24). In our study, we used the interquartile range (IQR) to describe consensus:
- A consensus was considered to be reached in statements with IQR = 0.00.
- No consensus was considered to be reached in statements with IQR > 1.00.
Using the IQR criterion does not allow for distinctions between the different degrees of consensus for statements with IQR = 1.00. The frequency distribution of responses to a statement can be bimodal, meaning that the IQR is greater than 1.00, despite 60 per cent of respondents generally agreeing with the statement. We therefore developed an additional criterion:
- A consensus was considered to be reached in statements with IQR = 1.00 which received generally positive responses (‘completely agree’ and ‘partly agree’) from more than 60 per cent of respondents.
We excluded non-responses and ‘Don’t know’ responses from the analysis.
Respondents’ comments on statements in the questionnaire formed the basis for the qualitative analysis. The comments were entered in Excel and analysed using the framework approach (25, 26). This method enables the researcher to perform a systematic qualitative data analysis and frame the ongoing data collection (25, 26).
The volume of data in our study was relatively small. We therefore used a simplified framework analysis (27). We interpreted the comments after sorting them into categories. The aim of this analysis was also to identify where statements needed to be clarified or reformulated in the next round.
As a result, we changed some statements and added more follow-up statements. Special attention was paid to statements with IQR = 1.00. Here we searched for findings that could undermine the consensus or non-consensus.
We invited a total of 183 nurses, social educators and pharmacists to participate in the study. Eighty-one of these were department heads in municipal nursing and care services. In the first round (n = 183) we received 54 responses (30 per cent). Some of the respondents exercised their right not to receive questionnaires. The number of questionnaires sent out in the second round was therefore 169, and 164 in the third round.
In the second round, we received 46 responses (27 per cent) and in the third round the figure was 43 (26 per cent). In total, 23 respondents answered in all three rounds. Table 1 shows the composition of the respondent group in all rounds. A consensus was reached in the study for 77 standards in medication management (Table 2).