Abstract
Background:
Hospitalized children are at risk of medication errors (MEs) due to complex dosage calculations and preparations. Incident reporting systems may facilitate prevention of MEs but underreporting potentially undermines this system. We aimed to examine whether scenarios involving medications should be reported to a national mandatory incident reporting system and the likelihood of self- and peer-reporting these scenarios among paediatric nurses and physicians.
Methods:
Participants’ reporting of MEs was explored through a questionnaire involving 20 medication scenarios. The scenarios represented different steps in the medication process, types of error, patient outcomes and medications. Reporting rates and odds ratios with 95% confidence interval [OR, (95% CI)] were calculated. Barriers to and enablers of reporting were identified through content analysis of participants’ comments.
Results
The response rate was 42% (291/689). Overall, 61% of participants reported that scenarios should be reported. The likelihood of reporting was 60% for self-reporting and 37% for peer-reporting. Nurses versus physicians, and healthcare professionals with versus without patient safety responsibilities assessed to a larger extent that the scenarios should be reported [OR = 1.34 (1.05–1.70) and OR = 1.41 (1.12–1.78), respectively]; were more likely to self-report, [OR = 2.81 (1.71–4.62) and OR = 2.93 (1.47–5.84), respectively]; and were more likely to peer-report [OR = 1.89 (1.36–2.63) and OR = 3.61 (2.57–5.06), respectively].
Healthcare professionals with versus without management responsibilities were more likely to peer-report [OR = 5.16 (3.44–7.72)]. Participants reported that scenarios resulting in actual injury or incidents considered to have a learning potential should be reported.
Conclusion:
The likelihood of underreporting scenarios was high among paediatric nurses and physicians. Nurses and staff with patient safety responsibilities were more likely to assess that scenarios should be reported and to report. Incidents with actual injury or learning potential were more likely to be reported. The potential for improving reporting rates involving MEs seems high.
Keywords: incident reporting, paediatrics, patient safety, medication errors
Introduction
Medication errors (MEs), resulting from incorrect medication prescribing, dispensing or administration, are associated with increased mortality and cost in hospitalized children.1–4 Children are potentially at greater risk of experiencing medications errors than adults due to multiple factors, such as complex dosage calculations and manipulations of medications in order to administer a correct dose.5,6
Incident reporting is one of the most widespread strategies to facilitate learning and prevention of MEs.7–9 However, underreporting may undermine the value of the incident reporting system.10–14 Barriers to reporting among healthcare professionals involve beliefs that the reporting system is ineffective in bringing about change, fear of being blamed/disciplinary actions, not knowing what to report, reporting being too time consuming and lack of feedback.10,11,15–18 Among the barriers to peer-reporting are that peer-reporting is perceived as tattling, reporting is believed to be outside one’s responsibility, and that reporting should only include one’s own profession.11
The type of profession may affect the likelihood of incident reporting, where physicians are less likely to report near misses or less-serious errors than nurses.18–22 Harmful MEs are likely to be reported by both nurses and physicians.11,21 Further, nurses engage more readily in self-reporting than physicians.11
Self-reported likelihood of incident reporting involving medications in hospitalized children among nurses and physicians has been studied to a limited extent.23,24 Paediatric nurses have shown to be more likely than physicians to report MEs to a local voluntary reporting system.23 Examining the assessed likelihood of reporting and the factors influencing reporting behaviour among nurses and physicians may help identify areas for improving ME reporting.
The aim of this study was to examine whether scenarios involving medications should be reported to a national mandatory incident reporting system and the likelihood of reporting these scenarios among paediatric nurses and physicians.
Methods
In this survey study, nurses and physicians from neonatal and paediatric hospital wards in one of the five regions in Denmark assessed 20 scenarios involving medications in relation to whether the scenarios should be reported and the likelihood of self- and peer-reporting these.
Setting
Denmark was the first country to introduce a national mandatory reporting system, the Danish Patient Safety Database (DPSD), by law.14 In 2016, more than 189,000 reports were submitted to the DPSD from public hospitals and other healthcare settings.25 Healthcare professionals are obliged to report all patient safety-related incidents including MEs to the DPSD.14,26 Reporting involves both incidents in which the reporter was involved (self-reporting) or witnessed (peer-reporting). Incidents which should be reported are events occurring in connection with healthcare activities that are not caused by the patient’s disease. These events can either have caused harm, or could potentially have caused harm but were intercepted or did not occur because of other circumstances.26 Adverse drug reactions (ADRs) that are not a result of MEs are reported elsewhere.26 Further, the system permits incident reporting from patients and relatives.
Reporting of incidents to the DPSD is based on a system perspective. The principle is that to err is human and that incidents are mainly caused by system failures rather than human failures. Learning from preventable incidents and near-misses is the main purpose of the reporting system. Incident reporting is nonpunitive and reports can be submitted anonymously. Reporting and analysis of incidents should be systematic and take place locally.
Participants
All trained nurses and physicians working on hospital neonatal and paediatric wards located in the region of Southern Denmark were invited to participate. Four hospitals were invited, including a university hospital (>1000 beds) and three secondary hospitals (<700 beds) with a total of 22 units specializing in general paediatric care (6), paediatric emergency (3), neonatal and neonatal intensive care (6), paediatric ambulatory care (4), paediatric intensive care (1), paediatric surgery (1) and paediatric haematology and oncology (1).
Questionnaire
The questionnaire involved assessment of 20 scenarios. The scenarios consisted of MEs previously reported to the DPSD, incidents observed by members of the research group and hypothetical scenarios (Table 1; a full version of the scenarios is presented in online supplement 1). Further, the stages of the medication process involving the hospital pharmacy service, prescribing, dispensing and administration were represented in the scenarios.
Table 1.
Description of the 20 scenarios assessed in the online questionnaire.
| Scenario | Description | Medication involved (ATC code) | Route of administration | Process | Harm |
|---|---|---|---|---|---|
| 1* | Delayed administration, busy ward | Metronidazole (J01XD01) | Oral | Administration | No harm |
| 2* | Wrong dose (overdose), calculation error | Vancomycin (J01XA01) | Intravenous | Prescribing | Harm |
| 3* | Wrong dose (overdose), calculation error | Vancomycin (J01XA01) | Intravenous | Dispensing | Harm |
| 4* | Wrong dose, suppository is split in half | Paracetamol (N02BE01) | Rectal | Prescribing | No harm |
| 5* | Expired medication, overlooked medicine expiry date | Meropenem (J01DH02) | Intravenous | Dispensing | No harm |
| 6* | Wrong duration (too long), forgetting to discontinue medicine | Gentamicin (J01GB03) | Intravenous | Prescribing | No harm |
| 7* | Drug interaction, unaware of drug interaction | Cisplatin/tobramycin (L01XA01/J01GB01) | Intravenous | Prescribing | No harm |
| 8* | Misplaced central venous catheter | Numerous medications | Intravenous | – | Harm |
| 9 | Adverse drug reaction subcutaneous | Human papillomavirus (J07BM02) | Subcutaneous | – | Harm |
| 10* | Wrong dose (overdose), intentional prescription | Salbutamol (R03AC02) | Oral | Prescribing | No harm |
| 11* | Wrong formulation, inappropriate formulation | Diclofenac (M01AB05) | Oral instead of intravenous | Prescribing | No harm |
| 12* | Wrong dose (underdose), defect calculator | Cefuroxime (J01DC02) | Intravenous | Dispensing | No harm |
| 13* | Wrong time (late), unknown cause of delayed medicine delivery, wrong time (late) | Carabine (L01BC01) | Intravenous | Hospital pharmacy service | Missing |
| 14 | Omitted dose, refusal of medication treatment due to foul taste | Phenoxymethylpenicillin, (J01CE02) | Oral | Administration | No harm |
| 15* | Wrong patient, missed patient identification, | Dicloxacillin (J01CF01) | Intravenous | Administration | No harm |
| 16* | Wrong drug, mixes up medicines (sound-alikes) | Methylprednisolone (Solu-medrol) (H02AB04) instead of Hydrocortisone (Solu-cortef) (H02AB09) |
Intravenous | Prescribing | No harm |
| 17* | Wrong volume, saline bag contains 10 ml more than specified | Sodium chloride (B. Braun) (B05BB01) | Intravenous | Dispensing | No harm |
| 18* | Infusion infiltration, unknown cause | Missing | Intravenous | Administration | Harm |
| 19* | Wrong dose (overdose), calculation error | Vancomycin (J01XA01) | Intravenous | Prescribing | No harm |
| 20* | Wrong dose (overdose), calculation error | Vancomycin (J01XA01) | Intravenous | Dispensing | No harm |
ATC, Anatomical Therapeutic Chemical Classification System.
Scenarios that should be reported to the national mandatory database.
According to official criteria of incidents to report to the DPSD, 18 of the scenarios were mandatory to report (scenarios 1–8, 10–13 and 15–20).26 Of the scenarios, three involved actual injury (scenarios 2, 3 and 8) and 14 involved no harm, including scenarios where interventions (e.g. observation or measuring serum concentration of a drug) examined potential injury (scenarios 5, 6, 19 and 20) and a near-miss (scenario 11). Two scenarios (scenarios 9 and 14) were not included in the DPSD criteria. In scenario 9, a girl received treatment in the paediatric emergency ward for a potential ADR of an HPV vaccine. In scenario 14, a parent refused medication treatment of his child due to the foul taste of oral penicillin, without any consequences for the child.
The questionnaire consisted of three parts. First, participants were asked about their previous experience with incident reporting (three items). Secondly, they were asked to assess the 20 scenarios from three questions: (a) do you believe that the scenario should be reported; (b) how likely is it that you will report the scenario if you were involved in the incident; and (3) how likely is it that you will report the scenario if you were not personally involved in the incident? Likelihood of self- and peer-reporting was evaluated using a 5-point Likert scale (1 = very likely to report, 5 = very unlikely to report). A text field was added to each scenario where participants could add a comment on their responses if they wanted to. The last part of the questionnaire involved demographic characteristics of the participants (10 items). To help facilitate anonymity, participants were not asked about their sex. The DPSD definition of an incident was deliberately left out of the questionnaire to encourage participants to provide answers based on their own views.
The questionnaire was pilot tested by five paediatric physicians and five nurses not included in the study sample. An email including study information and a link to the questionnaire was sent to participants by a local coordinator at each site. A 2- and 4-week follow-up reminder was subsequently sent. Data collection took place between January and March 2016.
Analysis
Responses from the questionnaire were analysed using descriptive statistics. A total of 131 missing responses involving 15 participants and related to the questions concerning the scenarios were corrected. Corrections were made based on comments in the text fields or patterns of reply, for example, a participant who responded ‘no’ to question 1 and consistently left questions 2 and 3 unanswered was considered very unlikely to self- and peer-report. No additional corrections were made to missing data.
Reporting rates indicating the percentages of respondents who felt that scenarios should be reported and how likely they believed they were to self- and peer-report scenarios were calculated for the total number of participants and for nurses and physicians, respectively. Also, inverse reporting rates are presented in this study. Reporting rates were calculated separately for the 20 scenarios and for the 18 scenarios included in the DPSD criteria involving incidents to report.
A logistic regression model involving the 20 scenarios was analysed for each of the three questions. The analysis was carried out through dichotomization of responses to question 1 (‘yes’ and ‘no/do not know’) and questions 2–3 (‘very likely/likely’ and ‘very unlikely/unlikely/do not know’).
An odds ratio (OR) involving profession (nurses versus physicians) was estimated for each of the three questions. The predictor model for question 1 involved all participants. Participants responding ‘no/don’t know’ to question 1 were excluded from the predictor model of question 2 and question 3 as their responses were assumed to be ‘very unlikely/unlikely/don’t know’. Factors included profession, age of participant, years since graduation, years in workplace (ward), working hours per week, management responsibilities and patient safety responsibilities. A subanalysis was carried out on the 18 scenarios included in the DPSD criteria. The overall performance of the three models was evaluated using Hosmer–Lemeshow χ2 (calibration: overall goodness-of-fit) and C statistics (discriminative power).
Content analysis was performed on the comments from the questionnaire text fields. Comments were read and thematically analysed according to whether the comment addressed reasons to report or not to report the scenario, reasons for missing responses or for responding ‘do not know’ involving lack of information, lack of reporting experience or being in doubt about whether to report or not, or which assumptions the participants made to answer the questions.
Ethics
According to Danish law, no ethics approval was required for this study, as the questionnaire was completed anonymously.27
Results
The response rate was 42% (291 out of 689 invited). In all, 31% (n = 216) of nurses and 44% (n = 75) of physicians completed the questionnaire (Table 2). The majority of participating nurses and physicians (>80%) had previously reported an incident to the national mandatory reporting system, DPSD. Participants who were inexperienced with reporting (n = 36) were generally younger, had completed their education more recently and had been working fewer years in their current workplace compared with participants who had reporting experience. More than half of the participants (66%) had not submitted a report during the last 6 months.
Table 2.
Characteristics of participants.
| Factors | Nurses n (%) (216) |
Physicians n (%) (75) |
|---|---|---|
| Age (years) | ||
| ⩽30 | 30 (13.9) | 1 (1.3) |
| 31–40 | 40 (18.5) | 19 (25.3) |
| 41–50 | 45 (20.8) | 24 (32.0) |
| 51–60 | 47 (21.8) | 9 (12.0) |
| ⩾61 | 6 (2.8) | 7 (9.3) |
| Missing responses | 48 (22.2) | 15 (20.0) |
| Years since graduation | ||
| <1 | 4 (1.9) | 0 (0.0) |
| 1–5 | 26 (12.0) | 7 (9.3) |
| 6–10 | 23 (10.6) | 11 (14.7) |
| 11–15 | 28 (13.0) | 15 (20.0) |
| 16–20 | 25 (11.6) | 8 (10.7) |
| ⩾21 | 62 (28.7) | 19 (25.3) |
| Missing responses | 48 (22.2) | 15 (20.0) |
| Workplace | ||
| Hospital 1: children’s hospital | 69 (31.9) | 31 (41.3) |
| Hospital 1: intensive care unit | 26 (12.0) | 12 (16.0) |
| Hospital 2: general paediatric ward and neonatal ward | 41 (19.0) | 6 (8.0) |
| Hospital 3: maternity and neonatal unit | 15 (6.9) | 0 (0.0) |
| Hospital 3: general paediatric ward | 12 (5.6) | 9 (12.0) |
| Hospital 4: maternity and neonatal unit | 3 (1.4) | 0 (0.0) |
| Hospital 4: general paediatric ward | 1 (0.5) | 2 (2.7) |
| Missing responses | 49 (22.7) | 15 (20) |
| Years in workplace (ward) | ||
| <1 | 17 (7.9) | 13 (17.3) |
| 1–5 | 37 (17.1) | 22 (29.3) |
| 6–10 | 40 (18.5) | 12 (16.0) |
| 11–15 | 22 (10.2) | 7 (9.3) |
| 16–20 | 27 (12.5) | 3 (4.0) |
| ⩾21 | 25 (11.6) | 3 (4.0) |
| Missing responses | 48 (22.2) | 15 (20.0) |
| Work hours per week | ||
| ⩽37 | 158 (73.1) | 10 (13.3) |
| >37 | 9 (4.2) | 50 (66.7) |
| Missing responses | 49 (22.7) | 15 (20.0) |
| Management responsibilities | ||
| No | 165 (76.4) | 34 (45.3) |
| Yes | 2 (0.9) | 26 (34.7) |
| Missing responses | 49 (22.7) | 15 (20.0) |
| Patient safety responsibilities | ||
| No | 155 (71.8) | 53 (70.7) |
| Yes | 13 (6.0) | 6 (8.0) |
| Missing responses | 48 (22.2) | 16 (21.3) |
| Experience of reporting MEs | ||
| Yes | 183 (84.7) | 65 (86.7) |
| No | 28 (13.0) | 9 (12.0) |
| Cannot remember | 5 (2.3) | 1 (1.3) |
| Number of reports submitted to the DPSD within the last 6 months | ||
| None | 143 (66.2) | 50 (66.7) |
| 1–2 | 64 (29.6) | 23 (30.7) |
| 3–5 | 8 (3.7) | 0 (0) |
| 6–10 | 1 (0.5) | 1 (1.3) |
| ⩾11 | 0 (0) | 1 (1.3) |
DPSD, Danish Patient Safety Database, ME, medication error.
Likelihood of reporting the scenarios
The overall reporting rates for whether participants felt that scenarios should be reported and the likelihood of self- and peer-reporting for all 20 scenarios were 61%, 55% and 34%, respectively (Table 3, Supplement 1). Higher overall reporting rates were found for the 18 scenarios included in the DPSD criteria with a reporting rate of 66% for whether scenarios should be reported, and 60% and 37% for likelihood of self-reporting and peer-reporting, respectively. Accordingly, the percentage likelihood of underreporting was 40% for self-reporting and 63% for peer-reporting these scenarios.
Table 3.
Percentages of participants who felt that scenarios involving medication should be reported and their likelihood of self- and peer-reporting.
| Scenario | Reporting rate (%) | |||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Should scenarios be reported?$ | Likelihood of self-reporting‡ | Likelihood of peer-reporting§ | ||||||||||
| Overall | Nurses | Physicians | OR (95% CI) | Overall | Nurses | Physicians | OR (95% CI) | Overall | Nurses | Physicians | OR (95% CI) | |
| 1: Delayed administration | 46 | 53 | 28 | 2.88 (1.58–5.25) | 33 | 38 | 18 | 1.50 (0.55–4.06) | 11 | 13 | 4 | 1.60 (0.43–5.96) |
| 2: Overdose | 100 | 100 | 100 | ! | 97 | 98 | 96 | 2.19 (0.48–10.06) | 71 | 73 | 67 | 1.30 (0.71–2.37) |
| 3: Overdose | 100 | 100 | 100 | ! | 97 | 99 | 90 | 21.80 (2.63–180.75) | 76 | 80 | 65 | 2.20 (1.19–4.05) |
| 4: Dosing error | 27 | 28 | 24 | 1.20 (0.63–2.86) | 18 | 19 | 15 | 1.13 (0.35–3.62) | 8 | 8 | 8 | 0.68 (0.20–2.34) |
| 5: Expired medicine | 83 | 84 | 80 | 1.35 (0.66–2.80) | 73 | 79 | 57 | 4.28 (1.85–9.90) | 41 | 44 | 32 | 1.61 (0.85–3.05) |
| 6: Wrong duration | 81 | 85 | 68 | 2.65 (1.36–5.18) | 69 | 74 | 58 | 1.08 (0.43–2.71) | 41 | 43 | 36 | 0.91 (0.47–1.79) |
| 7: Drug interaction | 81 | 82 | 81 | 1.05 (0.50–2.18) | 74 | 74 | 75 | 0.68 (0.22–2.13) | 42 | 41 | 44 | 0.86 (0.45–1.62) |
| 8: Central venous catheter | 82 | 85 | 71 | 2.39 (1.20–4.76) | 73 | 75 | 68 | 0.16 (0.02–1.23) | 50 | 53 | 41 | 1.20 (0.60–2.39) |
| 9: Adverse drug reaction | 16 | 19 | 8 | 2.71 (1.01–7.28) | 13 | 15 | 7 | 0.81 (0.08–8.35) | 7 | 7 | 5 | 0.39 (0.06–2.68) |
| 10: Overdose (intentional) | 5 | 5 | 7 | 0.69 (0.20–2.39) | 3 | 2 | 5 | 0.33 (0.02–4.74) | 2 | 1 | 5 | 0.05 (0.00–1.04) |
| 11: Wrong formulation | 2 | 2 | 0 | ! | 1 | 1 | 0 | ! | 0 | 1 | 0 | ! |
| 12: Underdose | 73 | 77 | 63 | 1.99 (1.07–3.72) | 64 | 72 | 44 | 3.53 (1.48–8.39) | 32 | 34 | 26 | 1.09 (0.53–2.26) |
| 13: Wrong time (late) | 55 | 60 | 42 | 2.11 (1.16–3.83) | 50 | 55 | 36 | 1.92 (0.54–6.81) | 29 | 31 | 21 | 1.10 (0.47–2.61) |
| 14: Omitted dose | 6 | 8 | 0 | ! | 4 | 6 | 0 | ! | 1 | 1 | 0 | ! |
| 15: Wrong patient | 97 | 97 | 95 | 1.72 (0.40–7.41) | 91 | 94 | 85 | 2.62 (0.84–8.15) | 60 | 62 | 56 | 1.24 (0.67–2.30) |
| 16: Wrong medication | 97 | 99 | 94 | 1.38 (0.76–2.48) | 92 | 95 | 89 | 1.27 (0.32–5.08) | 61 | 62 | 59 | 0.96 (0.51–1.79) |
| 17: Wrong volume | 52 | 54 | 46 | 6.00 (1.07–33.65) | 41 | 50 | 31 | 4.86 (1.66–14.23) | 26 | 29 | 16 | 2.01 (0.84–4.85) |
| 18: Infusion infiltration | 17 | 20 | 8 | 2.74 (1.02–7.37) | 12 | 16 | 8 | ! | 8 | 9 | 5 | 0.49 (0.07–3.34) |
| 19: Overdose | 98 | 98 | 97 | 1.96 (0.32–12.06) | 94 | 95 | 92 | 1.44 (0.35–5.94) | 57 | 59 | 53 | 1.25 (0.68–2.29) |
| 20: Overdose | 98 | 98 | 95 | 3.34 (0.71–26.65) | 93 | 98 | 78 | 11.49 (3.04–43.46) | 57 | 59 | 51 | 1.25 (0.68–2.30) |
| Total (20 scenarios) | 61 | 63 | 55 | 55 | 58 | 48 | 34 | 36 | 30 | |||
| Total (18 scenarios)* | 66 | 68 | 61 | 60 | 63 | 53 | 37 | 39 | 33 | |||
| Total (2 scenarios)** | 11 | 14 | 4 | 9 | 11 | 4 | 4 | 4 | 3 | |||
Percentages of participants answering ‘yes’.
Percentages of participants answering ‘very likely/likely’.
Perfect collinearity prevents statistical comparison.
OR (95% CI), odds ratio (95% confidence interval).
Scenarios included in the DPSD criteria on incidents to report.
Scenarios not included in the DPSD criteria on incidents to report.
Bold material is considered significant findings (95 % confidence interval).
The overall reporting rates for whether scenarios should be reported and likelihood of self- and peer-reporting were higher for nurses than physicians. In total, 63% of nurses considered the 20 scenarios should be reported compared with 55% of physicians. Higher numbers were seen with regard the 18 scenarios for nurses (68%) compared with physicians (61%).
In 17 out of 20 scenarios, more nurses considered that the scenarios should be reported compared with physicians. In only one scenario (scenario 10), in which an overdose of salbutamol was deliberately prescribed and administered to a child, a larger number of physicians considered that the scenario should be reported compared with nurses (6.5% versus 4.6%).
The likelihood of self- and peer-reporting among and between nurses and physicians was lower compared with whether scenarios should be reported. For nurses, the likelihood of self- and peer-reporting the 20 scenarios was found to be 58% and 36%, respectively. Subsequently, physicians’ likelihood of self- and peer-reporting was found to be 48% and 30%.
Nurses were more likely to self-report (nurses: 63% and physicians: 53%) and peer-report (nurses: 39% and physicians: 33%) the 18 scenarios included in the DPSD criteria than physicians. Physicians were more likely to self- and peer-report two scenarios involving omission of drug interaction check during prescribing (scenario 7) and scenario 10 compared with nurses.
The likelihood of self-reporting the 20 scenarios varied widely among and between nurses (1–99%) and physicians (0–96%). Similarly, the likelihood of peer-reporting the scenarios varied among and between the professions (nurses: 1–80% and physicians: 0–67%).
In 6 out of 20 scenarios, the majority of participants felt that the scenarios should be reported (overall reporting rate > 90%) and reported to be more likely to self-report (⩾90%) and peer-report (>50%). Scenarios involved actual injury (scenarios 2 and 3), interventions to examine potential injury (scenarios 19 and 20), an overdose of intravenous medications (scenarios 2, 3, 16, 19 and 20) and omitted patient identification during administration of intravenous dicloxacillin (scenario 15). All nurses and physicians agreed that scenarios where a patient is harmed by an overdose of a medication given intravenously (scenarios 2 and 3) should be reported. However, a lower likelihood of self-reporting and peer-reporting was observed for the two scenarios.
In 3 of the 20 scenarios, a relatively low number of participants felt that the scenarios should be reported and reported as unlikely to self- and peer-report these (>10%). The three scenarios involved a near-miss incident where an oral tablet is prevented from being administered to a child who has undergone throat surgery (scenario 11), an incident where a child is deliberately prescribed and administered an overdose of salbutamol (scenario 10) and an incident where a parent declines treatment due to foul taste of an oral medication (scenario 14).
Scenarios 9 and 14 were not included in the DPSD, however, some nurses felt that both scenarios should be reported (scenario 9: 19% and scenario 14: 8%) and were likely to self-report (scenario 9: 15% and scenario 14: 6%) and peer-report (scenario 9: 7% and scenario 14: 1%) these. On the contrary, none of the physicians felt that scenario 14 should be reported. A lower reporting rate for whether scenario 9 should be reported and the likelihood of self- and peer-reporting this scenario was found for physicians compared with nurses.
In each of the 20 scenarios, approximately 10% of participants indicated that they did not know whether the scenario should be reported or their likelihood of self- or peer-reporting. In 33 of the 60 questions, participants without reporting experience had a higher ‘do not know’ response rate.
Factors associated with reporting of the scenarios
The predictive model identified several factors associated with assessing whether scenarios should be reported and the likelihood of self- and peer-reporting (Table 4). Factors associated with assessing whether scenarios should be reported were nurse profession [OR = 1.34, confidence interval (CI) 1.03–1.73] and having patient safety responsibilities (OR = 1.50, CI 1.16–1.94). Factors associated with the likelihood to self-report were nursing profession (OR = 2.81, CI 1.71–4.61), being in either the age group 41–50 years (OR = 3.62, CI 2.34–5.48) or >50 years (OR = 2.57, CI 1.20–5.51) compared with the age group 40 years or younger and having patient safety responsibilities (2.92, CI 1.46–5.82). Being in the group of 11–20 years since graduation decreased the likelihood to self-report scenarios (OR = 0.29, CI 0.20–0.44) compared with being in the group ⩽10 years since graduation. Factors associated with the likelihood of peer-reporting included nurse profession (OR = 1.51, CI 1.10–2.07), having management responsibilities (OR = 3.42, CI 2.33–5.03) or patient safety responsibilities (OR = 3.50, CI 2.50–4.91). Healthcare professionals older than 50 years had a lower likelihood of peer-reporting scenarios (OR = 0.63, CI 0.42–0.96) compared with those who were 40 years or younger. Factors such as years in workplace and working hours per week were not associated with assessing whether scenarios should be reported (OR = 1.17 and 1.01, respectively) or the likelihood of self-reporting (OR = 1.38 and 1.33, respectively) and peer-reporting (OR = 1.27 and 0.79, respectively). The subanalysis of the 18 scenarios which should be reported to the DPSD showed similar results (not shown).
Table 4.
Multivariate logistic regression analysis of factors associated with whether scenarios should be reported and the likelihood of self-reporting and peer-reporting.
| Variables | Scenarios (20) | |||||
|---|---|---|---|---|---|---|
| Should scenarios be reported?$ | Likelihood of self-reporting‡ | Likelihood of peer-reporting§ | ||||
| OR | 95% CI | OR | 95% CI | OR | 95% CI | |
| Profession | ||||||
| Physician | Ref | Ref | Ref | |||
| Nurse | 1.34* | 1.03–1.73 | 2.81** | 1.71–4.61 | 1.51* | 1.10–2.07 |
| Age (years) | ||||||
| ⩽40 | Ref | Ref | Ref | |||
| 41–50 | 1.00 | 0.80–1.25 | 3.62** | 2.34–5.48 | 0.91 | 0.70–1.19 |
| >50 | 0.93 | 0.66–1.32 | 2.57* | 1.20–5.51 | 0.63* | 0.42–0.96 |
| Years since graduation | ||||||
| ⩽10 | Ref | Ref | Ref | |||
| 11–20 | 0.96 | 0.76–1.21 | 0.29** | 0.20–0.44 | 0.98 | 0.74–1.29 |
| >20 | 1.01 | 0.72–1.42 | 0.62 | 0.29–1.35 | 1.51 | 1.00–2.27 |
| Years in workplace (ward) | ||||||
| ⩽5 | Ref | Ref | Ref | |||
| 6–15 | 1.17 | 0.96–1.42 | 1.38 | 0.94–2.02 | 0.97 | 0.77–1.22 |
| >15 | 1.13 | 0.89–1.43 | 0.72 | 0.43–1.21 | 1.27 | 0.96–1.69 |
| Working hours per week | ||||||
| ⩽37 | Ref | Ref | Ref | |||
| >37 | 1.01 | 0.78–1.30 | 1.33 | 0.80–2.21 | 0.79 | 0.58–1.09 |
| Management responsibilities | ||||||
| No | Ref | Ref | Ref | |||
| Yes | 0.88 | 0.66–1.18 | 1.58 | 0.87–2.89 | 3.42** | 2.33–5.03 |
| Patient safety responsibilities | ||||||
| No | Ref | Ref | Ref | |||
| Yes | 1.50** | 1.16–1.94 | 2.92* | 1.46–5.82 | 3.50** | 2.50–4.91 |
CI, confidence interval; OR, odds ratio; Ref, reference; ROC-AUC, area under the receiver operating characteristics curve.
p < 0.05; **p < 0.01.
Number of observations = 4483, average number of participants = 239.9, calibrationa (Hosmer–Lemeshow goodness-of-fit): χ2 = 7.41 (p value = 0.39), discriminative powerb (ROC-AUC): 0.55 (CI 0.54–0.58).
Number of observations = 2733, average number of participants = 239.1, calibrationa (Hosmer–Lemeshow goodness-of-fit): χ2 = 5.79 (p value = 0.45), discriminative powerb (ROC-AUC): 0.66 (CI 0.62–0.69).
Number of observations = 2720, average number of participants = 238.4, calibrationa (Hosmer–Lemeshow goodness-of-fit): χ2 = 9.81 (p value = 0.20), discriminative powerb (ROC-AUC): 0.63 (CI 0.61–0.66).
A Hosmer–Lemeshow χ2 p value < 0.05 indicates poor model calibration.
Thresholds to determine discriminative power of our models: ROC-AUC 0.50–0.59, poor; 0.60–0.69, moderate; 0.70–0.79, good; ⩾0.80, very good to excellent.
Bold material is considered significant findings (95 % confidence interval).
Performance assessment of all three logistic regression models indicated a good calibration (p value < 0.05). The discriminatory power of the model involving the question ‘should the scenario be reported?’ was poor [area under the receiver operating characteristics curve (ROC-AUC) 0.55, 0.50–0.59]. Models involving the questions ‘how likely are you to self-report?’ and ‘how likely are you to peer-report?’ showed moderate discriminatory power (ROC-AUC > 0.60, 0.60–0.69).
Barriers to, and enablers of, self- and peer-reporting
In total, 46% of participants made comments on the scenarios. From these comments, identified barriers to self-reporting involved attitudes that reporting is too time-consuming, the reporting form is too complicated, reporting adds to workload, that there is no actual injury, the incident has no learning potential, not knowing what to report, reporting is not part of the culture, and whether the act of the healthcare professional was intended (scenarios 4, 10 and 14). In some scenarios, participants commented that harmless scenarios were unlikely to be reported, even if the scenario was considered to have a learning potential. Reasons for this were not included in the comments. Participants did not peer-report because of attitudes that the person involved is responsible for reporting, reporting is not part of one’s job and that the quality of a report would improve if the incident was reported by the person involved. Instead of peer-reporting, participants preferred to encourage colleagues to self-report.
Enablers of reporting included the belief that lessons can be learned, that there is actual injury and repeat incidents. Subjectivity seemed to influence assessment of the learning potential, as some participants would not report a dosing error caused by a defect handheld calculator (scenario 12) because they believed that nothing could be learned. However, others were likely to report the scenario as training or implementing a computerized calculator could possibly prevent the incident from reoccurring:
“The calculator should be replaced and the staff should be made aware that something like this may occur. This is why the computer calculator should be preferred.” (Comment from a nurse.)
Further, participants commented that the incident should be reported to facilitate change when inappropriate working procedures, not involving actual injury, were repeatedly observed and to facilitate examinations of current working procedures, for example examining why delays in medication treatment occurred (scenario 1), to potentially improve these.
Discussion
The overall likelihood of self- and peer-reporting scenarios involving MEs in children to a national mandatory reporting system was found to be to be 60% and 37%, respectively. Large variations in the likelihood of reporting were observed among and between nurses and physicians. Nurses and healthcare professionals with either patient safety responsibilities or management responsibilities were the most likely to self- and peer-report scenarios. Barriers to reporting involved scenarios being too trivial (no injury and no learning potential), lack of knowledge on what to report, existing reporting culture, and difficulty and workload of reporting. Enablers primarily focused on actual injury and subjective assessment of the learning potential.
Factors affecting incident reporting
Several factors affect reporting. Lack of knowledge on what to report may both prevent reporting of relevant incidents and lead to unnecessary reporting of irrelevant incidents. In our study, approximately 1 out of 10 of participants responded that they did not know if the scenarios should be reported. Further, nurses and physicians were likely to report scenarios to DPSD which should not be reported (scenarios 9 and 14). Facilitators such as bridging the communication gap and educating on what and why to report have previously been mentioned by nurses, physicians and pharmacists involved in focus group interviews.29
Recent recommendations to optimise the Danish national reporting system involve refining the definition on which incidents to report and reporting of incidents involving harm or learning potential.14 Changes may lead to greater motivation of the reporter to improving patient safety through reporting and show confidence in healthcare professionals.30 However, feedback is essential to motivate reporting.14
Nurses were more likely to report scenarios regardless of patient outcome compared with physicians. This finding is supported by results reported in other studies.18–21,23 Nurses are likely to report to protect themselves and reporting is more of a habit for them than for physicians.30 Further, nurses more often perceive reporting as part of their professional job.11,26 Similar to this, our results suggested that being appointed patient safety responsibilities will probably increase both self- and peer-reporting.
Like others, we identified that nurses and physicians reported that scenarios involving actual injury should be reported.11,21,28 However, the likelihood to report these was lower in our study. Comments did not specify any reasons for this, but several barriers potentially influence this finding. As already mentioned, lack of a clear definition may have caused underreporting. Also, comments from participants saying that reporting was too complicated and time consuming may affect reporting regardless of the patient outcome. Others have previously described these barriers.28,30 Danish recommendations suggest simplifying the existing reporting system, as the system is generally considered user hostile.14 Simplifying the reporting procedure will reduce the time spent on reporting and potentially increase reporting. An increase in incident reporting was observed in an intensive care unit by focusing on patient safety and making reporting easier.31 Simplifying the reporting procedure is also shown to increase reporting among physicians.32
The existing reporting culture also acted as a barrier to reporting, as comments from participants revealed that reporting was not part of normal practice. Also, fear of blame may be part of the reporting culture.10,11 However, this barrier was not identified in our study. Others have previously addressed the importance of creating an environment that facilitates reporting.10,11,28,33
Further, the existing reporting culture may have discouraged peer-reporting since we identified a relatively low likelihood of peer-reporting compared with self-reporting. Like us, Hewitt and colleagues found that the general attitude among practitioners is that peers are not responsible for reporting.11 In our study, participants preferred to encourage colleagues to self-report rather than engage in peer-reporting. Nurses and physicians are likely to value their relationship with their colleagues and peer-reporting may be looked upon as tattletaling.30 However, physicians are potentially more likely to consider peer-reporting disloyal to their colleagues than nurses, which our results may reflect.30 Still, managers should encourage healthcare professionals to peer-report if they become aware of incidents not being reported that would facilitate learning.
Barriers to reporting involving lack of feedback were not identified through the comments in our study, but have previously been reported by others.11,15–18 Feedback provided to the reporter is a fundamental part of reporting.14,29 The Danish reporting system DPSD has previously been associated with the demotivating effect of not providing feedback, nor taking actions towards improving patient safety.14,29 Reporting has to bring about change in order to continue the facilitation of reporting. Participants may experience this barrier, but might have omitted to comment on this, as we did not encourage participants to comment on specific barriers and enablers. Barriers identified in this study have all previously been described in the literature. This emphasizes that relevant incidents are not reported and barriers to reporting still exist.
Implications to improve incident reporting involving medications
Our data support the recommendations to reduce the reporter burden by refining the criteria on what to report and simplifying reporting. Resources must be allocated to secure training about what and why to report, but also to allow efficient feedback to the reporter and facilitate change to improve patient safety.14,28 Future studies should examine effective forms of feedback that have been studied to a limited extent.34
One of the challenges of reporting to a national database seems to be the lack of ownership among healthcare professionals at the local level, and the expectations of learning and initiatives to change are disseminated from the national level. In Denmark, the mandatory reporting system is managed and supported by The Danish Patient Safety Authority who identifies risk areas from patterns in reports and monitors the system.25 Limited resources at the national level and vast numbers of reports challenge the systematic analysis of reports.14 Recommendations highlight the need for reports to be used locally, as staff has the best prerequisites of implementing change.14 However, lack of authority among personnel at the sharp end may affect local implementation of system changes. Actors at the national and regional level, including hospital management, must support learning when incidents are unsolved at the local level.14 Hence, effective communication between local, regional and national levels must be established.
Strengths and limitations
One of the strengths of this study was the relatively large sample of participants compared with similar studies.23,24 Participants were recruited from university and regional hospital settings. Further, different types of scenarios involving different patient outcomes were included in the questionnaire. This gave a more detailed picture of the variability in the assessment. In addition, the analytical performance of our multivariate analysis was generally satisfactory.
Our study also has some limitations. First, assessing the likelihood of reporting does not necessarily translate into actual reporting behaviour and is possibly influenced by social desirability bias. Semistructured interviews would allow us to gain a deeper understanding of the participants’ beliefs and attitudes towards incident reporting and how to overcome barriers to reporting. Secondly, a suboptimal number of participants responded to the questionnaire, which influences generalizability of the findings. Reporting rates identified in this study may be even lower in reality, since respondents are more likely to have a strong engagement in incident reporting than nonrespondents. Random sampling of participants from all neonatal and paediatric wards in Denmark would have potentially increased representativeness. Participants were enrolled from one of the five Danish regions, and differences in assessing the scenarios identified in this study were likely to be explained by local factors instead of regional factors, such as local reporting culture and the attitudes of the personnel working in Denmark. Further studies are needed to explore potential differences among the wards or hospitals, as we did not include the paediatric and neonatal wards in the logistic regression models. Lastly, ME reporting systems differ between nations, limiting the external validity of studies from a single country.
Conclusion
The likelihood of underreporting MEs was high. Considerable variability in assessing whether to report scenarios and the likelihood of self- and peer-reporting scenarios involving medication in hospitalized children was identified among and between nurses and physicians. Facilitators influencing the assessment of whether the scenarios should be reported and the likelihood to self-report included being a nurse, having patient safety or management responsibilities, incidents involving patient injury or subjectively assessing the incidents to have learning potential. Factors such as years in a workplace and working hours per week did not influence the assessment of whether the scenarios should be reported or the likelihood of self- or peer-reporting. Barriers continue to challenge incident reporting, where relevant incidents are underreported, but the potential for improvement is high. We suggest that developing a clear definition on what to report, simplifying the reporting procedure, and creating a reporting culture, with education and empowerment of healthcare professionals on what and why to report may result in higher and more correct ME reporting rates. Actions must be taken at local and national levels to optimize ME reporting.
Supplementary Material
Acknowledgments
The authors would like to thank all nurses and physicians who contributed to the pilot test and who took time to complete our questionnaire. We thank our dedicated local project managers who enabled us to recruit our participants and we thank the team from Odense Patient data Explorative Network at Odense University Hospital for making REDCap available for us to use and assisting us in designing our questionnaire.
Footnotes
Author Note: Rikke Mie Rishoej, is also affiliated to OPEN, Odense Patient data Explorative Network, Odense University Hospital, Odense, Denmark.
Funding: This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.
Conflict of interest statement: The authors declare that there is no conflict of interest.
Contributor Information
Rikke Mie Rishoej, Clinical Pharmacology and Pharmacy, Department of Public Health, University of Southern Denmark, J. B. Winsløws Vej 19, 2., 5000 Odense C, Denmark.
Jesper Hallas, Clinical Pharmacology and Pharmacy, Department of Public Health, University of Southern Denmark, Odense, Denmark.
Lene Juel Kjeldsen, Amgros I/S, Copenhagen, Denmark.
Henrik Thybo Christesen, Hans Christian Andersen Children’s Hospital, Odense University Hospital, Odense, Denmark Department of Clinical Research, University of Southern Denmark, Odense, Denmark.
Anna Birna Almarsdóttir, Social and Clinical Pharmacy, Department of Pharmacy, University of Copenhagen, Copenhagen, Denmark.
References
- 1. Kaushal R, Bates DW, Landrigan C, et al. Medication errors and adverse drug events in pediatric inpatients. JAMA 2001; 285: 2114–2120. [DOI] [PubMed] [Google Scholar]
- 2. Kunac DL, Kennedy J, Austin N, et al. Incidence, preventability, and impact of adverse drug events (ADEs) and potential ADEs in hospitalized children in New Zealand: a prospective observational cohort study. Paediatr Drugs 2009; 11: 153–160. [DOI] [PubMed] [Google Scholar]
- 3. Agarwal S, Classen D, Larsen G, et al. Prevalence of adverse events in pediatric intensive care units in the United States. Pediatr Crit Care Med 2010; 11: 568–578. [DOI] [PubMed] [Google Scholar]
- 4. Tundia NL, Heaton PC, Kelton CML. The national burden of E-code-identified adverse drug events among hospitalized children using a national discharge database. Pharmacoepidemiol Drug Saf 2011; 20: 866–878. [DOI] [PubMed] [Google Scholar]
- 5. Kearns GL, Abdel-Rahman SM, Alander SW, et al. Developmental pharmacology: drug disposition, action, and therapy in infants and children. N Engl J Med 2003; 349: 1157–1167. [DOI] [PubMed] [Google Scholar]
- 6. Richey RH, Shah UU, Peak M, et al. Manipulation of drugs to achieve the required dose is intrinsic to paediatric practice but is not supported by guidelines or evidence. BMC Pediatr 2013; 13: 81. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Spigelman AD, Swan J. Review of the Australian incident monitoring system. ANZ J Surg 2005; 75: 657–661. [DOI] [PubMed] [Google Scholar]
- 8. Williams SK, Osborn SS. The development of the National Reporting and Learning System in England and Wales, 2001–2005. Med J Aust 2006; 184: S65–S68. [DOI] [PubMed] [Google Scholar]
- 9. Cheng L, Sun N, Li Y, et al. International comparative analyses of incidents reporting systems for healthcare risk management. J Evid Based Med 2011; 4: 32–47. [DOI] [PubMed] [Google Scholar]
- 10. Pfeiffer Y, Manser T, Wehner T. Conceptualising barriers to incident reporting: a psychological framework. Qual Saf Health Care 2010; 19: e60. [DOI] [PubMed] [Google Scholar]
- 11. Hewitt T, Chreim S, Forster A. Sociocultural factors influencing incident reporting among physicians and nurses: understanding frames underlying self- and peer-reporting practices. J Patient Saf 2017; 13: 129-137. [DOI] [PubMed] [Google Scholar]
- 12. Zhan C, Smith SR, Keyes MA, et al. How useful are voluntary medication error reports? The case of warfarin-related medication errors. Jt Comm J Qual Patient Saf 2008; 34: 36–45. [DOI] [PubMed] [Google Scholar]
- 13. Cullen DJ, Bates DW, Small SD, et al. The incident reporting system does not detect adverse drug events: a problem for quality improvement. Jt Comm J Qual Improv 1995; 21: 541–548. [DOI] [PubMed] [Google Scholar]
- 14. The Danish Society for Patient Safety. Optimisation of the Danish incident reporting system. 2016. https://patientsikkerhed.dk/content/uploads/2016/09/optimisationofthedanishincidentreportingsystem.pdf (accessed 23 August 2017).
- 15. Noble DJ, Pronovost PJ. Underreporting of patient safety incidents reduces health care’s ability to quantify and accurately measure harm reduction. J Patient Saf 2010; 6: 247–250. [DOI] [PubMed] [Google Scholar]
- 16. Brubacher JR, Hunte GS, Hamilton L, et al. Barriers to and incentives for safety event reporting in emergency departments. Healthc Q 2011; 14: 57–65. [DOI] [PubMed] [Google Scholar]
- 17. Hewitt T, Chreim S, Forster A. Incident reporting systems: a comparative study of two hospital divisions. Arch Public Health 2016; 74: 34. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Evans SM, Berry JG, Smith BJ, et al. Attitudes and barriers to incident reporting: a collaborative hospital study. Qual Saf Health Care 2006; 15: 39–43. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Wolf ZR, Serembus JF, Smetzer J, et al. Responses and concerns of healthcare providers to medication errors. Clin Nurse Spec 2000; 14: 278–287. [DOI] [PubMed] [Google Scholar]
- 20. Alrwisan A, Ross J, Williams D. Medication incidents reported to an online incident reporting system. Eur J Clin Pharmacol 2011; 67: 527–532. [DOI] [PubMed] [Google Scholar]
- 21. Sarvadikar A, Prescott G, Williams D. Attitudes to reporting medication error among differing healthcare professionals. Eur J Clin Pharmacol 2010; 66: 843–853. [DOI] [PubMed] [Google Scholar]
- 22. Lawton R, Parker D. Barriers to incident reporting in a healthcare system. Qual Saf Health Care 2002; 11: 15–18. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Taylor JA, Brownstein D, Christakis DA, et al. Use of incident reports by physicians and nurses to document medical errors in pediatric patients. Pediatrics 2004; 114: 729–735. [DOI] [PubMed] [Google Scholar]
- 24. Stratton KM, Blegen MA, Pepper G, et al. Reporting of medication errors by pediatric nurses. J Pediatr Nurs 2004; 19: 385–392. [DOI] [PubMed] [Google Scholar]
- 25. Dansk Patientsikkerhedsdatabase. Årsberetning 2016, http://viewer.ipaper.io/styrelsen-for-patientsikkerhed/dpsd-aarsberetning-for-2016/ (2017, accessed 7 July 2017).
- 26. Sundhedsloven. 2010. https://www.retsinformation.dk/Forms/R0710.aspx?id=130455 (accessed 30 August 2017).
- 27. Persondataloven. Lov om behandling af personoplysninger, 2000. https://www.retsinformation.dk/forms/r0710.aspx?id=828 (accessed 25 November2015).
- 28. Hartnell N, MacKinnon N, Sketris I, et al. Identifying, understanding and overcoming barriers to medication error reporting in hospitals: a focus group study. BMJ Qual Saf 2012; 21: 361–368. [DOI] [PubMed] [Google Scholar]
- 29. Patientombuddet og Ministeriet for Sunhed og Forebyggelse. Rapport om serviceeftersyn af rapporteringssystemet for utilsigtede hændelser, 2014. http://www.sum.dk/~/media/Filer%20-%20Publikationer_i_pdf/2014/Rapport-om-serviceeftersyn-af-rapsystem-utilsigtede-haendelser-juli-2014/Rap-om-serviceeftersyn-rapporteringssystemet-utilsigtede-haendelser.ashx (accessed 5 July 2017).
- 30. Kingston MJ, Evans SM, Smith BJ, et al. Attitudes of doctors and nurses towards incident reporting: a qualitative analysis. Med J Aust 2004; 181: 36–39. [DOI] [PubMed] [Google Scholar]
- 31. Harris CB, Krauss MJ, Coopersmith CM, et al. Patient safety event reporting in critical care: a study of three intensive care units. Crit Care Med 2007; 35: 1068–1076. [DOI] [PubMed] [Google Scholar]
- 32. King ES, Moyer DV, Couturie MJ, et al. Getting doctors to report medical errors: project DISCLOSE. Jt Comm J Qual Patient Saf 2006; 32: 382–392. [DOI] [PubMed] [Google Scholar]
- 33. Braithwaite J, Westbrook MT, Travaglia JF, et al. Cultural and associated enablers of, and barriers to, adverse incident reporting. Qual Saf Health Care 2010; 19: 229–233. [DOI] [PubMed] [Google Scholar]
- 34. Benn J, Koutantji M, Wallace L, et al. Feedback from incident reporting: information and action to improve patient safety. Qual Saf Health Care 2009; 18: 11–21. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.