Healthcare Worker Usage of Large-scale Health Information Exchanges in Japan: User-level Audit Log Analysis Study

Background: A health information exchange (HIE) system enables healthcare providers and patients to exchange clinical data electronically across multiple healthcare institutions. Sharing medical information appropriately using HIE has several benefits in clinical practice. More than 200 HIEs have been implemented in Japan and are currently operational. However, in Japan, little academic research has been conducted on whether healthcare workers use HIE in the institutions where it was introduced. The primary objective of this study was to clarify the usage rate, usage frequency, and usage bias of large-scale HIEs in Japan by analyzing audit logs. Objective: The primary objective of this study was to clarify the usage rate, usage frequency, and usage bias of large-scale HIEs in Japan by analyzing audit logs. Methods: We conducted a retrospective cohort study. The research subjects were HIEs joined by over 100 facilities and 10,000 patients. Each healthcare worker’s profile and audit log data for HIE were collected. The following four analyses were conducted: First, we counted the annual number of days of HIE use by each hospital doctor in the financial year (FY) 2021/22 and calculated the Gini coefficient. Second, we calculated monthly HIE use frequency distribution for each facility type in FY2021/22. Third, we calculated each facility type's monthly HIE usage rate in FY2021/22. Fourth, we compared the monthly HIE usage rate by medical institution for each HIE and the proportion of man-days of HIE use by occupation for each HIE. Results: Twenty-four HIEs were identified as candidates for data collection, and we analyzed data from seven HIEs. Among the hospital doctors, 93.5% had never used HIE during the available period in FY2021/22. The Gini coefficient for each hospital doctor's estimated annual HIE usage days was 0.984. The monthly HIE usage rates were 0.482 (0.470-0.487) for hospitals, 0.243 (0.230-0


Preprint Settings
1) Would you like to publish your submitted manuscript as preprint?Please make my preprint PDF available to anyone at any time (recommended).
Please make my preprint PDF available only to logged-in users; I understand that my title and abstract will remain visible to all users.Only make the preprint title and abstract visible.No, I do not wish to publish my submitted manuscript as a preprint.2) If accepted for publication in a JMIR journal, would you like the PDF to be visible to the public?
Yes, please make my accepted manuscript PDF available to anyone at any time (Recommended).
Yes, but please make my accepted manuscript PDF available only to logged-in users; I understand that the title and abstract will remain v Yes, but only make the title and abstract visible (see Important note, above).I understand that if I later pay to participate in <a href="http

Introduction Background
A health information exchange (HIE) is an electronic mobilization system of clinical data across entities such as institutions or organizations, or an organization that controls it [1][2][3].The appropriate sharing of medical information using HIEs enables fewer duplicated procedures, duplicated imaging, and total orders.HIE usage is also associated with improved medication reconciliation and immunization [1,4,5].HIEs have several major features.The first feature enables sending and receiving secure information electronically between care providers to support coordinated care, known as directed exchange [1,6].The second enables remote, on-demand viewing or searching of aggregated patient health data from multiple healthcare institutions.This feature is known as query-based exchange or query-based HIE [1,6,7].In addition to these two features, the Office of the National Coordinator for Health Information Technology (ONC) proposes the concept of consumer mediated exchange, which allows patients to aggregate and control the use of their health data among providers [1].
In Japan, the concept of HIE is not widely used.Instead, systems called "Chiiki iryo joho renkei nettowa-ku" have provided features equivalent to HIEs [8][9][10][11].In some literature, it is also called "regional healthcare network (RHN)" in English [10].Past surveys show that over 200 RHNs of various sizes operate in Japan [9].investigates the services provided by each RHN, revealing that "sharing of medical data" was the most common service at 83.0%, followed by "sharing of medical images" at 81.6%.These features are equivalent to query-based exchange.Approximately 28.8% provide email services and 19.7% provide electronic patient referral documents, which is equivalent to directed exchange.Only 3.9% provide self-management systems by patients, which is equivalent to consumer mediated exchange.
In other words, query-based exchange is the most common feature of Japanese HIEs.Since HIE and RHN essentially refer to systems with the same features, we will refer to RHN as an HIE in the subsequent paragraphs.To avoid confusion between computer systems themselves and organizations, we refer to the organization that promotes HIE as regional health information organization (RHIO), a term adopted in most literature [7,12].
It is crucial to evaluate the benefits of HIEs for individual healthcare workers and institutions.
Reviews on HIEs have highlighted the importance of understanding whether the system is used [13].To study the actual use of HIEs, audit logs have been frequently analyzed [14][15][16][17][18][19][20][21][22][23][24][25].The Japanese Association of Healthcare Information Systems Industry (JAHIS) published a technical document called "JAHIS's Guide Ver.1.0on Evaluation Indicators for Regional Medical Collaboration" [26], which emphasizes the importance of evaluating HIE systems using audit log analysis.However, most previous reports [8,9,27] or peer-reviewed journal articles [28][29][30] about HIE in Japan either did not include audit log analysis or were limited to simple analyses.MHLW conducted a survey [9] on access to HIEs in 2019.The average monthly active institution ratio based on the MHLW report was 38.1% for HIEs connected to 100 or more institutions (Table S1 in Multimedia Appendix 2), meaning that more than half of the connected institutions did not access HIEs.While the MHLW report suggests low utilization of large-scale HIEs, it does not include userlevel analysis.Since there are no studies analyzing the audit logs of multiple HIEs in Japan at the user-level, the usage of HIE by individual users or medical institutions remains unknown.
The primary objective of this study was to clarify the extent to which query-based exchange is used by individual healthcare workers and institutions in large-scale HIEs in Japan by analyzing audit logs at the user-level.One reason for investigating only query-based exchange is that, as already mentioned, it is the most common style of the system in Japan.The other reason is that while directed exchange has alternatives such as patient referral letters on paper, and consumer mediated exchange has alternatives such as prescription records on paper, query-based exchange can only be achieved through HIE.Therefore, analyzing the usage of this feature indicates the significance of HIE.There are two reasons for investigating only large-scale HIEs.First, it is not realistic to investigate the audit logs of all 200 or more HIEs.Second, large-scale HIEs appear to have spread as they are accepted by many medical institutions and many patients in the region.In the beginning of this study, we thought that by investigating large-scale HIEs, we would obtain suggestions for the further spread of other small-scale HIEs.

Study design and data collection
In this study, we collected data on HIEs that met the following inclusion criteria: (1) they must be included in the list of the survey report, "About the current situation of regional healthcare network" [9] published by MHLW; (2) each HIE must be connected to more than 100 institutions and with more than 10,000 patients according to the report above.We asked all RHIOs operating HIEs that met the inclusion criteria to cooperate in this study.When requesting data from each RHIO, we promised to conceal the identity of the RHIO that provided the data in this study.We also promised to limit our published analysis results to such that the RHIO which provided the data, would be concealed.These promises were made to avoid any effects public disclosure of the usage status of each HIE would have on its operation.We obtained data from RHIOs that provided informed consent.
The profile and audit log data of healthcare workers enrolled in the HIE were collected.In this study, we did not collect patient data in the HIE.The following data (Textbox 1) were requested for each HIE: Consequently, datasets and data representation formats differ among HIEs.We also obtained data on the number of connected institutions per month or year for each RHIO.The maximum period of the audit log data was five years.We aimed to acquire audit log data from April 1, 2017, to March 31, 2022, but if there were no accumulated data for that period, we asked the RHIO to provide data for the period that could be extracted.
This study was approved by the ethics committee of Kyoto University Graduate School and the Faculty of Medicine.The accession number was R3266-7.

Measures and data analyses
We refer to "viewing patient medical data using query-based exchange" as "HIE use" in this study.Patient medical data viewed by healthcare workers is obtained from multiple storage such as hospital electronic medical records (EMR).If a patient agrees to the disclosure of their medical data stored at the institution, the institution or the RHIO office will take steps to release the stored medical data to HIEs.The types of medical data disclosed from storage to HIEs vary by institution.Patients can also choose the facilities to which their medical data should be disclosed.Several models have been proposed by the MHLW for viewing the medical data disclosed in this way [11].For example, doctors at the clinic can see what medical tests a patient has previously undergone when visiting the clinic for the first time.Patients referred from a clinic to a hospital can later check at the clinic the kind of treatment they will receive at the hospital to which they are referred.We investigated how often these types of use cases presented by the MHLW occur by analyzing audit logs.Audit logs for logging into the HIE system are not subject to analysis.Furthermore, the sending and receiving of documents between medical workers using the directed exchange is not included in the analysis.
As a unit for measuring access, one man-day was defined as HIE use on one day with one user account.The cumulative man-day for an institution is the cumulative number of days of HIE use by each user belonging to the institution.For example, consider the use of a virtual clinic within a given month.At that clinic, one doctor used HIEs for three days, and one nurse used HIEs for two days, which was the clinic's total use of HIEs for that month.In this case, the clinic's usage for that month was five man-days.If multiple users share a common account, this aggregation methodology may underestimate HIE usage.However, it is generally not recommended in HIEs.
We classified the institutions enrolled in HIEs as hospitals, medical clinics, dental clinics, pharmacies, visiting nursing stations, or nursing facilities.The institutions that could not be classified into these categories were excluded from the analysis.For example, this study did not analyze public institutions such as fire departments, public health centers, local medical associations, or vendors that developed HIEs.
In this study, the financial year (FY) is from April 1 of one year to March 31 of the next year.For example, FY2021/22 started on April 1, 2021, and ended on March 31, 2022.We used R (version 4.3.1) to perform the analysis.

Percentage of days of HIE use by each hospital doctor in FY2021/22
For each user account of doctors affiliated with hospitals, we calculated the ratio of the number of days of active HIE use.We analyzed the data of HIEs that met the following two criteria: (1) Audit log data for all periods in FY2021/22 should be available; (2) The "date of account registration and account deletion in HIE" of each user should be available (Textbox 1).For all HIEs that met the criteria, the annual number of days of HIE use by each hospital doctor's account in FY2021/22 was counted.Next, we calculated the number of days that the doctor could use HIE in FY2021/22.For each doctor's account that was registered with the HIE during FY2021/22, we subtracted from 365 the number of days from April 1, 2021 to the day before the account registration date.For accounts that were removed from the HIE during FY2021/22, we subtracted the number of days from the day after the account deletion date to March 31, 2022 from the remaining number of days.The number of days remaining after these subtractions is the number of days that the doctor can use HIE in FY2021/22.For accounts that were able to use the HIE on all days in FY2021/22, we used 365 as the number of days that the account could use HIE.
We then calculated the ratio of days of HIE use by each hospital doctor.The ratio of days of HIE use was defined as follows: Docto r ' s annual number of days of HIE use ∈FY 2021/22 Number of days that the doctor canuse HIE∈FY 2021/ 22 (1 )

Man-days for monthly HIE use by each institution in FY2021/22
We calculated the man-days for monthly HIE use by each institution for each month and aggregated them by facility type.We analyzed the data of HIEs that met the following two criteria: (1) the audit log data for all periods in FY2021/22 should be available; (2) data should be available on the number of participating institutions by facility type, matching our facility classification.For each institution belonging to any HIEs that meet the criteria, man-days for monthly HIE use in FY 2021/22 were aggregated.Next, by each facility type, we tallied the number of months for each manday group divided into 5 or 10 increments.Finally, for each facility type, the percentage of each man-day group was calculated.

Monthly active institution ratio in FY2021/22
We calculated the monthly active institution ratio for each facility type, defining it as follows: Number of institutions any of which members used HIEs during the month Number of institutions participating ∈ HIEs at the end of the month .
(2) We analyzed the data of HIEs that met the following two criteria: (1) Audit log data for all periods in FY2021/22 were available; (2) Data on the number of participating institutions by facility type, which could match our facility classification, were available.For each facility type, the total number of participating institutions for the last day of each month in FY2021/22 was aggregated in all HIEs that met the criteria.This corresponds to the denominator of Eq. ( 2).Next, each month in FY2021/22, and each institution was flagged as to whether it used HIEs.If at least one account of the institution used HIEs for at least one day a month, the institution was deemed to have used HIE that month.Subsequently, for each facility type, we calculated the sum of the institutions that used HIEs for each month.This corresponds to the numerator in Equation (2).Finally, for each facility type, we calculated the active institution ratio for each month using Equation (2): Among the facility type, hospitals were further classified based on the number of beds.Hospitals were divided into three categories: those with 200 or more beds, those with 100 to 199 beds, and those with 99 or fewer beds, and Equation (2) was calculated for each classification.

Monthly active institution ratio of medical institutions and man-days of HIE use for each user type
We compared the monthly active institution ratios at medical institutions for each HIE and the proportion of man-days of HIE use by user type.Equation (2) defines the monthly active institution ratio."Medical institutions" are facility types of "hospital," "medical clinic," and "dental clinic."We analyzed all HIEs during all periods for which data could be obtained.
The total number of medical institutions participating in each HIE on the last day of each month was calculated.Next, for each HIE in each month, each medical institution of each HIE was flagged as to whether it used HIE.If at least one account of the institution used the HIE for at least one day a month, the institution was deemed to have used HIE that month.Subsequently, for each HIE in each month, we calculated the sum of the institutions that used it.We then calculated the active institution ratio of medical institutions for each HIE for each month, according to Equation (2).
Next, we classified all user occupation data into eight user types.That is, "doctor," "nurse," "rehabilitation staff," "pharmacist," "dental profession," "nursing care staff," "other medical professions," and "type unknown."We aggregated the total number of man-days of HIE use by user type and calculated the proportion of man-days by user type.

Data collection
The MHLW report listed 218 HIEs.However, from number 57 to 60 in the report are federated, and the aggregated statistics are shown as number 61.After removing this duplicate, there were 214 HIEs listed [9].Of the 214 HIEs, 36 HIEs were connected to more than 100 institutions and 45 were connected with more than 10,000 patients.Overall, 21 HIEs met both inclusion criteria.Considering number 61 to be 4 HIEs, 24 HIEs were considered candidates for data collection.
Initially, we requested research cooperation from each RHIO administrator via email.The data request document is shown in Multimedia Appendix 1.Thereafter, we requested each RHIO to provide data through a web conference once we were able to hold detailed negotiations.We obtained research data from eight HIEs.Although these tasks were sometimes performed free of charge, the extraction of access logs from the two HIEs was performed for a fee.One of the eight HIEs was unable to extract comprehensive audit log data of query-based exchange and was removed from the final analysis.The flow diagram is depicted in Figure 1.Among twenty four HIEs that met the inclusion criteria for this study, we calculated the average and SD of the monthly active institution ratio based on the MHLW report separately for the HIEs that were included in the final analysis and those that were not.The results are shown in the Table 1.The MHLW report lists the "number of participating medical institutions" and the "number of medical institutions that accessed HIE" for each HIE.According to the MHLW report, the "number of medical institutions that accessed HIE" is the number of institutions that used HIE during the month covered by the survey.Referring to the report, we divided "the number of medical institutions that accessed HIE" by "the number of participating medical institutions" in each HIE and named this number "monthly active institution ratio based on MHLW reports."b Among the HIEs listed in the MHLW report, those connected to more than 100 institutions and with more than 10,000 patients were included in this analysis.
To avoid identification, each RHIO and HIE was assigned a pseudonym using letters from A to G.
We Of the data items we attempted to obtain in advance (Textbox 1), the "type of device used for access" was not accumulated in any HIEs.Consequently, we could not analyze the type of device from which the HIE was accessed.We obtained users' "date of account registration and account deletion in HIE" from only three RHIOs.For the remaining four HIEs, we could not determine the exact number of registrants in each period.

Characteristics of HIEs included in the final analysis
All seven HIEs began operations between 2010 to 2015.Of the 24 HIEs that met the inclusion criteria, 9 did not set membership fees, and of the 7 HIEs included in the final analysis, 3 did not set participation fees for connected institutions.Overall, 5 RHIOs adopted ID-Link for the HIEs they operate, 3 RHIOs adopted HumanBridge, and 2 RHIOs employ products other than these.The reason that the sum of these products is more than 7 is that some RHIOs operate multiple products in parallel.For HIE A to G, the patient consent rate for HIE in 2022 was 21.2%, 5.6%, 69.5%, 1.9%, 2.0%, 4.0%, 6.5%.Patient consent rate is obtained by dividing the number of patients connected with the HIE by population of the area.All six HIEs except HIE C required patients to complete a paper consent form for their medical data to be viewed by healthcare workers using HIE.HIE C has a similar operation using paper consent forms, but in addition, HIE C operates the "patient demographic data synchronization feature" provided by ID-Link.For institutions that disclose patient data to HIE C, basic patient profiles such as name, date of visit, and public insurance data are automatically accumulated in the HIE.This feature allows healthcare workers to use query-based exchange to obtain patient data in the event of emergency treatment, even if the patient cannot explicitly consent to the use of HIE in advance.Patient enrollment in HIE using this feature is optedout and is considered implicit consent to participate in the HIE unless explicitly declined.Therefore, the apparent consent rate of HIE C is extremely high.

Percentage of days of HIE use by each hospital doctor in FY2021/22
Three HIEs met the criteria; HIE A, HIE B, and HIE C. The number of hospital doctor accounts registered in HIEs operated by these three HIEs in FY2021/22 was 7,833.The overall results are shown in Figure 2.

Man-days for monthly HIE use by each institution in FY2021/22
Different HIEs met the criteria for each facility type as shown in Table S2 in the Multimedia Appendix 2. The cumulative number of months of hospital participation in HIEs in FY2021/22 was 3,434.The distribution of man-days for monthly HIE use by hospitals is shown in Figure 3. Table 2 shows the analysis results for facility types other than hospitals.The number of institutions connected to the HIEs included in the analysis for each month in FY2021/22 is shown in Table S3 in Multimedia Appendix 2.

Monthly active institution ratio in FY2021/22
Table S2 in Multimedia Appendix 2 lists the HIEs that met the criteria in this section.The median monthly active institution ratios were 0.482 (0.470-0.487) in hospitals, 0.244 (0.231-0.247) in medical clinics, 0.030 (0.024-0.048) in dental clinics, 0.202 (0.188-0.216) in pharmacies, 0.307 (0.301-0.325) at visiting nursing stations, and 0.197 (0.185-0.204) in nursing facilities.We illustrate the monthly active institution ratios using boxplots in Figure 4. Table 3 shows the monthly active hospital rate for each category subdivided by the number of hospital beds.HIE F data could not be combined with hospital bed data and was therefore excluded from the analysis.

Monthly active institution ratio of medical institutions and man-days of HIE use for each user type
We analyzed all seven HIEs.As the period of audit log data obtained differs for each HIE, the analysis period also differs.The five HIEs included all types of medical institutions.However, no dental clinics participate in HIE D and HIE E. Regarding user type data, precise data for HIE D were not available.Occupational data for HIE F could only be obtained for "doctors" and "other medical professions."Some HIEs have restrictions on the types of users that can use the HIE.In the two HIEs, users affiliated with medical clinics could only use the HIEs if they were doctors.In one HIE, users who were affiliated with hospitals that did not disclose patient data to the HIE could only use the HIE if they were doctors.Six HIEs had no restrictions on the type of data that authorized healthcare workers could view.One HIE was set so that people other than doctors affiliated with the hospital could not view outpatient treatment data.
We illustrated the monthly active institution ratio of the HIE and the proportion of man-days of HIE use by user type in medical institutions using boxplots and bar graphs (Figure 5).Monthly active institution ratios of HIE in medical institutions are also shown in Table S4 in the Multimedia Appendix 2. The proportion of man-days of HIE use by user type in medical institutions are also shown in Table S5 in the Multimedia Appendix 2.

Overview of HIE use
As mentioned in the introduction, the low utilization of large-scale HIEs had been already suggested by MHLW reports (Table S1 in Multimedia Appendix 2).Our analysis included data from only about 20% of such HIEs.However, among HIEs that met the inclusion criteria, monthly active institution ratios based on the MHLW report were similar for HIEs included in the final analysis and those not included (Table 1).This suggests that the results of this study apply to some extent to large-scale HIEs in general.However, HIEs with fewer participating institutions have a higher monthly active institution ratio based on the MHLW report than large-scale HIEs (Table S1 in Multimedia Appendix 2).Therefore, it is possible that small-scale HIEs are used more actively than the results of this study suggest.However, the MHLW report includes 13 HIEs where the number of participating medical institutions is 1 or 2, and the number of participating institutions is equal to the number of institutions that have accessed HIEs.The active institution ratio based on the MHLW report should be interpreted with caution because extremely small-scale HIEs are driving up the ratio.
Across the large-scale HIEs analyzed in this study, many healthcare workers and institutions did not use query-based exchange.These results are consistent with MHLW reports (Table S1 in Multimedia Appendix 2).Exactly 93.5% of the doctors at hospitals registered with HIE do not use it even once a year (Figure 2).Approximately 50% of hospitals did not use query-based exchange even once a month (Figures 3 and 4).This is lower than those reported in previous studies conducted in other countries [20,32].The monthly active institution ratio increases for hospitals with more beds.The monthly active institution ratio for hospitals with 99 beds or fewer is over 30%, but over 70% for hospitals with 200 or more beds (Table 3).Among facilities other than hospitals, the monthly active institution ratio is even lower than hospitals with 99 beds or fewer.The monthly active institution ratio for visiting nursing stations has reached 30%, but only approximately 20% for medical clinics, pharmacies, and nursing care facilities (Table 2).As for dental clinics, the monthly active institution ratio is less than 5%.Previous studies outside Japan have also revealed that HIEs tend not to be used by dental practices [33].This is the first study to reveal the active institution ratio of HIEs by facility type in Japan.
Where query-based exchange was used, most people and institutions only used it for a limited number of days.Previous reports have not provided user-level analysis, therefore this study is the first to reveal the total number of days of HIE use by healthcare workers and institutions.Of the 507 hospital-affiliated doctor accounts who actively used query-based exchange in FY2021/22, 81.3% used it for 5% or fewer days (Figure 2).In other words, assuming the average doctor works 20 days a month, 80% of these doctors use query-based exchange less than once a month.As the percentage of days of HIE use increases, the number of corresponding hospital doctor accounts tends to decrease.This trend is reflected in man-days for monthly HIE use of hospitals.Exactly 90.8% of all hospitals use HIE for 30 or fewer man-days per month (Figure 3).In these hospitals, query-based exchange is used by less than one user daily.The number of man-days of HIE use in hospitals also shows a tendency for the number of applicable months to decrease as the number of man-days increases.This trend remains true for man-days for monthly HIE use at facilities other than hospitals (Table 2).However, some institutions and users use query-based exchange for many days.Of the hospital-affiliated physician accounts, 19 users used query-based exchange for 30% or more days.Of the cumulative months of hospital participation in HIEs in FY2021/22, 3.0% had over 101 man-days for HIE use (Figure 3).This exceeds the number of months when man-days of monthly HIE use is 91 to 100.For visiting nursing stations and nursing care facilities, the number of months in which the number of man-days for HIE use exceeds 26 is greater than the number of months in which the number of man-days is 21 to 25 (Table 2).That is, there are significant disparities in HIE use across institutions and users.
Monthly active institution ratios for medical institutions vary widely by HIE.HIE A, which has the highest monthly active institution ratio, has a rate of approximately 50%, but some HIEs have a rate of over 10% (Figure 5).The proportion of man-days of HIE use by each user type is not constant for each HIE.However, regarding the HIEs that could be confirmed, the number of man-days of HIE use was low for dental professionals, pharmacists, and rehabilitation workers.

Possible factors influencing HIE use
Many other factors could have influenced monthly active institution ratios and man-days of monthly HIE use, for example, whether there are membership fees and usage restrictions based on use type, and the system used.The data viewed by healthcare professionals when using HIEs is also extremely important.None of these can be shown as individual HIE data due to privacy considerations, therefore they cannot be discussed in relation to the results shown in Figure 5.This is a significant limitation of this study and indicates the need for further research into why there are such large disparities in demand for HIEs in Japan.
While a detailed elucidation must be reserved for future research, two factors may have influenced the monthly active institution ratio of facilities in the HIEs in our analysis.One is the consent rate of patients to participate in HIEs.HIE A, which had the highest monthly active institution ratio among medical institutions in this analysis, also had a relatively high patient consent rate of over 21%.HIE C, which had the second highest monthly active institution ratio, has a partial opt-out policy and a very high consent rate.However, HIEs G, D, B, where the monthly active institution ratio was 20% or less, had patient consent rates of 7% or less.However, the patient consent rate in HIEs B and E, where the monthly active institution ratio was in the 20% range, was less than 6%, and therefore lower than in G, D, and B. High consent rates may have contributed to the high active institution ratios for HIEs A and C, but this study cannot determine whether these factors are causally related.
Another factor that may have influenced the monthly active institution ratio is the number of staff at each participating institution.As already shown, the active institution ratio was higher in hospitals with 99 or fewer beds than in medical clinics, and higher in hospitals with 199 or more beds than in hospitals with 99 or fewer beds.It is natural to assume that this difference is caused by the absolute number of staff working at each institution.Therefore, when considering the active institution ratio for a given HIE, the value is likely to be high if a large proportion of the institutions participating in the HIE are large hospitals.

Audit log for further research analysis
To analyze HIE use in detail, proper design of the audit log is extremely important.When analyzing the audit logs in this study, two characteristics of some audit logs posed obstacles.One was that the extracted audit logs are not comprehensive.In HIEs that were configured using products from multiple vendors [34], each product generally has its own unique audit log design and storage.To perform detailed analysis of the usage of such HIEs, it was necessary to extract the logs from each product.This was difficult when each product was controlled by a different institution.Specifically, the HIE platform system is managed by the RHIO, but the EMR data viewing system may be managed by each hospital.In this case, we need to obtain consent for research collaboration from each institution to perform overall log extraction.If user access to individual systems via the platform system is recorded in the audit log, analyzing the general usage status may be possible by extracting only the platform system's audit log.However, in practice, access to individual systems is not necessarily recorded in the platform system's audit log.In HIEs configured using products from multiple vendors, careful attention must be paid to facilitating comprehensive log extraction.
The second was the incompatibility of the institution IDs and user IDs used in HIEs.To clarify the factors that create the disparities in HIE usage across users and across institutions, more detailed data on users and medical institutions, such as medical specialty and whether they provide acute or chronic care, is required.However, such data are not generally included in the audit log itself, therefore log data needs to be cross-referenced with institution IDs compiled in the master data by the government or the detailed user data of each institution.As the institution IDs used in the audit log data extracted in this study did not necessarily correspond to the institution IDs assigned by MHLW, it was difficult to cross-reference log data with other data sets.To investigate HIE usage in greater depth than this study, it is recommended to adopt master data that can be matched with EMR and official datasets when designing audit logs.

Limitations
This study had several limitations.The most significant limitation was the need to maintain the anonymity of the HIEs included in the analysis.Therefore, important data, such as the systems employed by individual HIEs and the types of medical data disclosed, was either kept private or disclosed anonymously.Consequently, it was almost impossible to analyze the causes of differences in active institution ratios for individual HIEs from the data.We also attempted to evaluate the viewing situation for each data, such as images and prescriptions.However, it was difficult to perform a comprehensive analysis because the data storage format was not standardized for each HIE.The list of HIEs included in the analysis differed for each analysis because the data items that could be obtained differed for each HIE.Data regarding the type of device from which the access was made could not be obtained from any HIE.Some HIEs have features other than viewing patient medical data, such as sending and receiving documents or messages [8].Previous reports indicate that some HIEs actively used these additional features when treating COVID-19 patients [27].As this study focused on query-based exchange, we did not perform a quantitative analysis of the usage of other features.Another study is required on the actual usage of features other than query-based exchange.This study revealed that most users do not use query-based exchange or use it infrequently, but it is impossible to prove whether this is due to a lack of patients' medical data in the HIE repository or a lack of need to view data.As mentioned above, HIEs vary widely in both patient consent rates and the types of data that medical workers can view.Therefore, it is difficult to provide a single answer to this remaining question.To answer this question, a deeper investigation of each HIE is required, using more detailed audit log analysis, system descriptions, and qualitative research.

Conclusions
In the large-scale HIEs surveyed in this study, the overall usage of the on-demand patient data viewing feature was low, consistent with past MHLW reports.User-level analysis of audit logs revealed large disparities in the number of days of HIE use among healthcare workers and institutions.There were also large disparities in HIE use by facility type or HIE, and the percentage of cumulative HIE usage days by user type also differed by HIE.This study indicates the need for further research into why there are large disparities in demand for HIEs in Japan, and the need to design comprehensive audit logs that can be matched with other official datasets.

Figure 2 .
Figure 2. Percentage of days of HIE use by doctors affiliated with the hospital.Shown as a semilogarithmic bar graph.

Figure 3 .
Figure 3. Distribution of man-days for monthly HIE use by hospitals.

Figure 4 .
Figure 4. Monthly active institution ratio of HIE categorized by facility type.

Figure 5 .
Figure 5. Monthly active institution ratio and proportion of man-days of HIE use by user type for each HIE.
[8]t of these RHNs are sponsored by governments or local authorities.According to the Ministry of Health, Labour and Welfare (MHLW) report, 27 RHNs cover entire prefectures, 104 RHNs are within the secondary medical area, 32 are the size of a municipality, and 15 are smaller than a municipality[9].Item 2.10.2 of JMARI's 2021 survey[8]

we
received only the available data each RHIO could provide.

Table 1 .
Average monthly active institution ratio based on the MHLW report. a

Table 2 .
Distribution of man-days for monthly HIE use by medical care institutions other than hospitals.

Table 3 .
Monthly active hospital ratio subdivided by the number of hospital beds.