Implementing an intervention to improve adverse incident reporting in the hospital setting: A pilot study

Objective: This pilot study evaluated an intervention designed to increase error reporting among physicians. Research was conducted at University Hospital A, where data were collected from April 2017 to March 2019. The intervention began in April 2018 and consisted of the four following steps: dene reporting standards, improve the incident reporting system, have the hospital administrators set clear goals and begin a visualized feedback process, and achieve support and appropriate feedback from the hospital administrator. Results: Physicians’ reporting rates were higher in FY 2018 than 2017. Particularly, differences began to occur in November of FY 2018 (p < 0.05, analysed using Fisher’s exact test). Further, the number of reports submitted by non-physicians increased by 900 in FY 2018 compared to those in FY 2017. Based on these results, the intervention effectively increased incident reporting rates among not only physicians, but also other staff members. In this regard, reporting barriers were broken when hospital administrators encouraged employees to submit incident reports.


Introduction
Fatal accidents resulting from medical errors range from < 5% to > 30% [1], with estimates from developed nations suggesting that between 7.5% and 10.4% of patients in the acute care setting experience adverse drug events [2]. A pilot study revealed 11.3% of hospitalized cases experienced adverse events in Japan [3]. Using eight years of medical deathrate data from the United States, patient safety experts at Johns Hopkins calculated that more than 250,000 deaths per year were due to medical errors [1]. The number of patient deaths caused by adverse events annually was estimated to be between 1,326 and 1,433 in Japan [4]. To improve patient safety, it is needed to collect information related to adverse events and near-miss reports. This practice constitutes one of the strategies hospital managers use to understand the kinds of errors that occur at their hospitals.
Previous studies have focused on reporting barriers. Kaldjian et al. surveyed physicians to improve knowledge about error reporting [5]. Only about 55% of respondents knew how to report errors and only 40% knew which errors to report; hence, they recommended that institutions teach physicians ways to report errors and which errors to report. Poorolajal et al. reported that the main reasons mentioned for underreporting were: lack of effective medical error reporting systems, proper reporting forms, peer support for a person who has committed an error, and personal attention for the importance of medical errors [6]. The attitude of administration is also related to barriers to incident reporting [7]. Evans et al. reported that the most frequently stated barrier to reporting for doctors and nurses was lack of feedback [8].
The current pilot study evaluated an intervention to increase incident reporting rates. The intervention was conducted in collaboration with hospital administrators to remove barriers to reporting incidents.

Study setting
This study was conducted at University Hospital A in northern Japan from scal year (FY) 2017 to FY 2018. The hospital contained approximately 1,160 beds and 2,650 staff members including 530 physicians (some of whom were dentists), 1,260 nurses, and 80 pharmacists. The hospital used an electronic reporting system that required all staff members to report all types of errors and near misses. As such, this study de ned the scope of incident reports to include these. The reporting system operated through a local broadband network from within the hospital, with all reports reviewed by a medical safety management department (MSMD).

Intervention methods
Randomised control trials are likely to be challenging to conduct in this area, because nding su cient units to randomise and suitable control groups will be challenging [9]. This pilot study used a time-series design with comparisons before and after the intervention. The intervention examined in this pilot study consisted of four steps, the rst of which was implemented in April 2018.
Step 1: De ne reporting standards beginning in April 2018 The patient safety manual used at University Hospital A required staff to report all incidents resulting from negligence, error, or failure. However, there was no standard requiring them to submit incident reports related to sentinel events. We, therefore, collaborated with MSMD members to de ne reporting standards for 33 topics and identi ed references related to adverse events [3], used Clavien-Dindo classi cation for complications [10], and investigated the reporting standards of several hospitals on their homepages. As shown in Table 1, these were distributed across three categories Part 1 was related to surgery; Part 2 was related to medical events other than surgery; and Part 3 included all other events. These reporting standards were then described in the manual and announced during a patient safety meeting attended by approximately 120 risk managers, who were then required to work as departmental educators. Monthly meetings were held with these risk managers to report on their experiences in that regard. Step 2: Improvements to create a su cient incident reporting system since September 2018 University Hospital A was using its own electronic reporting system. However, clinical staff members asked the MSMD to improve this system because reporting forms were considered too long. We therefore analysed retrospective data and altered the question format from text input to select input, offered optional inputs rather than required inputs, and removed questions that were considered less important, reducing the total questions from 78 to 35. These efforts were designed to alleviate the burdens on incident reporters. These improvements reduced the time usually needed to enter one report from 20 minutes to 5-10 minutes. MSMD announced improvements and ways to use this system in the risk managers' meeting, which was also useful for re-notifying them about ways to access and use it.
Step 3: Visualized feedback implemented by the hospital administrator since October 2018 The MSMD prepared and published monthly documents concerning the submission status of incident reports from physicians inside the hospital, while the director referred to these documents when meeting with representatives of the clinical department.
Step 4: Support and appropriate feedback from the hospital administrator throughout FY2018 The deputy director (also one of the risk managers) explained the importance of incident reporting for preventing similar incidents during one of the risk-manager meetings. He speci cally explained that staff members who reported incidents would not receive inappropriate feedback from their bosses consequently. This was also established as a hospital rule and listed in the patient safety manual.

Data source and analysis
The MSMD collected data from April 2017 to March 2018. In this context, the administrator of individual information deleted personal identi ers such as patient names, identi cation numbers, and the names of hospital staff members.
Data were then given to this study's researchers. Data were varied, including reporting data (year/month/day), related personnel's occupations, impact on the patient outcomes due to the event, and type of reports (medication errors, transfusion errors, etc.). Differences in the reporting rates before and after the intervention each month were estimated with the use of Newcombe method [11] and compared with the use of chi-square tests. Additionally, we compared the number and types of reports. All analyses were performed using SAS statistical software, version 9.4 and JMP version 12.0 (SAS Institute, Cary, NC).

Ethics approval
This study was approved by Ethics Committees at Iwate Medical University School of Medicine. We also obtained comprehensive agreements from participating hospital staff members and posted an o cial notice on the hospital's website. The recruitment period lasted from 30 September to 31 October 2018, during which participating hospital staff members could refuse the use of their data. Table 2 shows the characteristics of incident reports between FY 2017 and FY 2018. Incident reports of drugs were most frequently reported by staff, while those of treatment/procedure were most frequently reported by physicians. There were differences in reporter's occupation (staff, physicians) between FY 2017 and FY 2018. Among incident reports by staff, drug report rate of FY 2018 was higher than that of FY 2017. Intervention term and reporting numbers Table 3 shows the numbers and rates of incident reports from all staff members and physicians, with side-by-side comparisons for FY 2017 and FY 2018. Before conducting the intervention (FY 2017), physicians typically submitted around 30-40 reports per month, with the fewest in December and January (24 each; Table 3). The fewest overall number of physician-submitted reports (14) occurred in April of FY 2018, which was the same month the risk managers held their rst meeting to reveal the new reporting standards for distribution to staff members. After this, reporting numbers substantially increased. Next, improvements to the incident reporting system were o cially released in October, which is also when the hospital director began the visualised feedback process among physicians, whose incident reporting rates increased by 10% and higher since November of FY 2018. In particular, the highest monthly number of physician-submitted reports (79) was found in December, with the total reaching 641 in FY 2018. As shown in Table 3 and Fig. 1, differences have occurred in all reporting rates since November of FY 2018. The differences in the physicians'

Discussion
Our pilot intervention was associated with an increase in the number and rate of incident reports by physicians.
Comparison from FY 2017 to FY 2018 revealed that the number of incident reports by both physicians and staff increased.
Physicians tended to report far fewer minor than major incidents [12], but near-miss reports were increasingly submitted, from 16 in FY 2017 to 106 in FY 2018. Reporting standards were focused on severe cases, but they had related treatment/procedure and drug, so that not only error cases but also near-miss cases were reported. Reasons for underreporting may include lack of knowledge, which includes how to report errors, what should be reported, and what kinds of errors to report [5]. As such, reporting criteria were de ned; staff members were informed about which incidents should be reported; and the MSMD could easily educate staff accordingly. Staff members were also pleased that incident reporting times were reduced. Announcement of reporting standards in our intervention brought knowledge and information about e cient reporting systems to physicians and all staff; this intervention led to the increase in the number of incident reports. However, physician submissions only increased to constitute 2.6% (95% CI, 1.7 to 3.5) more of the total in FY 2018, thus reaching 9%. The pilot intervention effectively increased the overall number of incident reports submitted by physicians but did not reach the aforementioned 10% goal. Nevertheless, physicians submitted 241 more reports in FY 2018, with all employees submitting 871 A calculation of physician reporting rates revealed an increase in the number of all incident reports as denominators and the number of physicians reported as numerators.
Hence, the overall physician reporting rates seemed small.
The pilot intervention also entailed a visualised condition feedback conducted on a monthly basis. However, effective safety feedback does not solely depend on publicised incident rates, but also requires timely, visible, and repeatable corrective actions and quality improvement processes [8]. Concerning having hospital administrators provide support and feedback, previous studies have also shown that a patient safety manager can effectively reduce the fear of reporting [13]. Most physicians believed that reduction in medical errors should be a national priority, but physicians believed that fear of medical malpractice is a barrier to reporting errors, and that greater legal safeguards are necessary for the success of mandatory reporting systems [14]. Further, this study's intervention involved hospital managers, including the director and deputy director. Finally, it seemed that employees did not feel there was inappropriate feedback (including on nancial and legal matters) when reporting incidents due to the direct involvement of the hospital administrator.

Conclusion
This pilot study helped remove incident reporting barriers to increase patient safety. Issues included the establishment of (1) clear reporting standards, (2) an effective medical error reporting system, (3) feedback on what actions are/were taken, and (4) peer support/appropriate feedback from hospital administrators. A before-after intervention comparison revealed that all incident reporting rates increased (i.e., from physicians and all other staff members). Hospital administrators also encouraged employees to submit incident reports, thus effectively removing reporting barriers.

Limitations
Its primary outcome was based on measuring the number of incident reports by physicians. Hence, it did not measure factors related to reporting barriers, including non-blaming, non-punitive, and  Figure 1 The rate of incident reports by physicians

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download. consort1.jpeg