EHR and EDC Integration in Reality

November 16, 2009
Stefan Paepke

,
Nicole Harzendorf

,
Ulrike Shwarz-Boeger

,
Nadia Harbeck

,
Marion Kiechle

,
Markus Schmidt

,
Gudrun Zahlmann

Applied Clinical Trials

An evaluation of the Munich Pilot and the affects of the EHR-EDC integration solution.

Previous publications described the necessity of integrating data capture processes in a medical environment for clinical trials by bridging the two IT worlds between the pharmaceutical industry and health care providers [1-6]. Still there is a major gap between clinical care and clinical trial processes at the health care provider site. No EDC system of a trial sponsor is really integrated nor are all these systems ‘electronic’ for the trial data providers. At HIMSS 2007, CDISC and IHE for the first time jointly demonstrated possibilities of how to re-use EHR data for biosurveillance, trial registry, clinical trials, and safety [7].

The EHR–EDC integration for clinical trials in the HIMSS scenario (initiative of Novartis, Siemens, and SAS) was built as a demonstration project using the IHE (Integrating the Healthcare Enterprise) profile RFD (Retrieve Form for Data capture). In this profile the data request is directly sent to the EHR system and converted into an EHR form for data capture in the care provider’s own process environment.

Based on long-standing experience with EDC at Bayer Healthcare, Johann Pröve, Head of data management at Bayer Schering Healthcare, describes his vision regarding the ultimate solution for data capture in clinical trials without the need of EDC systems, while directly feeding a CDMS by EHR systems for all trial participants [10]. Between this future goal and today’s situation we face many steps for improving clinical trial operations. What are the current obstacles?

So far, only limited integration pilots demonstrated integrated solutions: the STARBRITE project [6] and the here described Siemens integrated clinical trial solution that is running now for three years, which is discussed below. Both integration scenarios fulfill the eSDI group's recommendation [9]. All of these activities prove that the base technology can bridge both worlds.

To make this an industry success story, data regarding the benefit and business model of EDC in the pharmaceutical industry are needed. Those data are rare but describe some hard and soft achievements. However, the industry’s overall adoption rate of EDC solutions is still limited, but is expected to grow until 2011 by a double digit rate [11].

Currently, there is no study published describing the implication of EHR–EDC integration activities on trial processes, involved personnel, time effort, data quality, and health care provider site costs as well as implications for the overall trial process in its communication between trial data provider and trial sponsor. The pharmaceutical industry will never push for an integrated solution unless proof exists that such an activity is financially beneficial. Due to the current pressures regarding efficiency improvements of the pharmaceutical industry, especially in the clinical test phase, there are activities to investigate this further [12].

To overcome some of these shortcomings Siemens conducted in collaboration with the Frauenklinik of the Technical University of Munich an evaluation study that describes the effect of an integrated clinical trial solution. We started with a ‘traditional’ approach of paper-based clinical trial processes for an investigator-initiated trial and compared these with the parameters after switching over to the use of a fully integrated EHR–EDC solution for the same trial. This article will describe the method as well as provide results and discussion, respectively, that deal with the implications on the investigator’s side as well as the effects on the overall trial process. These implications hold true independently of the sponsor of the clinical trial.

Integrated EHR–EDC solution: The Munich pilot
The Siemens Integrated Clinical Trial Solution bridges the clinical care and clinical trial world by re-using electronically available data at the care site that are relevant to a clinical trial. No data search (“where are the patient records?”) and no data re-entry (“this is already in the lab system-why do I have to re-enter the data into another system”) are necessary. Technically, the Integrated Clinical Trial Solution consists of a Portal, an integration engine, an adjusted electronic data capture system, and the clinical IT systems (HIS, PACS, LIMS) (see Figure 1) [8].

The user enters the integrated study platform via a log-in to the portal. This is a single sign on to all underlying applications-the entry to the care and the research world. This is done in a patient aware mode enabling to work on patient specific data in care and research without any additional scrolling, searching, etc. If a trial patient shows up at the doctor’s office for a scheduled visit in a trial, data are documented in the care system (HIS=IS-H Med, Lab, PACS) and from there communicated to the integration engine. This engine uses the information in both the care and trial worlds (e.g., Jane Edwards = 123456789 (clinical systems) = 345 (clinical trial)) to filter the relevant data out of the care data. This works for any vendor’s systems.

The care IT systems in Munich are capable of sending standardized HL7 messages. The integration engine translates the data into the CDISC ODM model. The trial relevant data are sent to the validation buffer. The 21 CFR part 11 compliant integration engine is the interface between the regulations of both the care world and the clinical trials. The validation buffer requires a human interaction of qualified personnel (according to their specific role and system privileges) to review the data, check the match with the patient as well as the source data at the clinical documentation site and then manually confirm the data transfer into the EDC system. This solution represents one out of the four electronic source data alternatives as described by the CDISC eSDI initiative [9]. The EDC system provides edit and consistency checks to the data and the possibility of data cleaning. Images used in the clinical trial are stored in a separate Research PACS and interlinked with the EDC system. InferMed Ltd provided the EDC system used in Munich.

Evaluation study
The evaluation study was designed as a prospective “before and after” study with respect to introduction and use of the above described integrated EHR–EDC solution. The study was designed to answer the following two questions:
• What is the impact of the Integrated EHR–EDC system on trial quality, efficiency, and costs?
• What are the process step differences between the ‘traditional’ trial management and the Integrated EHR–EDC system?


The prospective “before and after study” took 12 months. Phase I of the study included performing one clinical trial at TUM as a paper-based trial (‘traditional way’). After five months, there was a cross-over into the ‘new’ way of conducting the trial based on a process change including implementation of the Integrated Clinical Trial Solution. This study period took five months. Due to missing data from other studies in the provider market segment, the evaluation study was designed as an explorative pilot study.

To evaluate the effects of the Integrated Clinical Trial Solution on the clinical trial process design and on data quality and costs, we used a real life clinical trial conducted following a specific clinical trial protocol. In our case, we evaluated the HEDON (Herceptin, Docetaxel Neoadjuvant) Phase II non-randomized trial, which is an evaluation of response rates to pre-operative Docetaxel and Docetaxel + Herceptin (stratified by HER2-Status) in locally advanced breast cancer patients. The rate of pathologic complete remission following primary systemic therapy in group A (HER2+) with Docetaxel and Herceptin, is assessed and compared to the rate of group B (HER2-) with Docetaxel. On top of this clinical trial, the evaluation study was performed as described in the following figure (Figure 2).

Data Collection
Questionnaires tailored to the individual study tasks of different study personnel were the basis for the evaluation study result assessment. The participants completed their questionnaires regarding the parameters to be quantified. The data gathered by the questionnaires were collected in a database and then analyzed statistically. Some additional data were extracted from the Integrated Clinical Trial Solution's log files. The following time parameters in Table 1 were considered within this evaluation study for the included patients only.

Personnel were specifically trained with regard to the evaluation study. There were two training sessions, one before the start of evaluation period and one after three months of ongoing evaluation. Individual time is given as estimates by the participants.

Process steps
A process description is provided prior to the start of the trial and the start of this evaluation study. This is specifically adapted to the individual trial. During the study, this assessment is validated and the number of steps that are necessary to enter data and manage the trial are accounted for regarding the ‘traditional’ as well as for the ‘new’ trial management arm. Process description is the result of observation and documentation by process analysts and their thorough discussion with the personnel involved in the clinical trial.

A project executed by KKS Düsseldorf on behalf of Siemens provided a generic process analysis for conducting clinical trials based on the current practice at the KKS Düsseldorf [13]. Each of the basic processes is divided into different sub-processes. The data entry (as a part of data management) is the most important one for the subsequent process change analysis.

Out of several different process analysis methods we have chosen the process-organization method that illustrates best the business process relationship between stakeholders, involved systems (e.g. IT), and the sub-processes [14]. Within the HEDON trial, the main focus is on the processes of the initial screening visit and the subsequent chemotherapy visits for neoadjuvant as well as adjuvant chemotherapy. The data entry process steps of the 'traditional' way are compared to them using the integrated EHR–EDC solution.

The number of personnel (investigators, project managers, data managers, monitor, study nurse, IT support staff, and other staff) involved in the direct trial process at the beginning and at the end of the study was counted. Using these staff categories, the average cost estimate for one person in a specific category is measured.

Data Analysis
The main part of the statistical analysis was descriptive. Number and quality of the individual process steps and the measured parameters between the ‘traditional’ and ‘new’ way of conducting the clinical trials were compared (Table 2). Time measurements are described as mean values (MV) and standard deviations (SD) per trial phase (visit type). Statistical methods used depend on the number of available data points. We used a descriptive method and mean value for less than 15 data points per observation. For N>= 15, we used a t-test for continuous variables and for qualitative data chi-squared analysis was applied. All analyses were performed at a significance level of p

Process steps were described in terms of number and mode. The analysis itself was performed by the University of Düsseldorf; this center had not been involved in the original data capture in order to avoid potential bias [15].

Cost assessments were done taking into account all involved personnel as well as estimated costs per personnel category and time.

Results
Nineteen patients participating in the HEDON trial were followed in the evaluation study within a 12 month period. All patients were screened at the Department of Obstetrics and Gynecology of the Technical University of Munich. Nine patients started within the traditional paper-based trial process and were then switched over to the new process, whereas ten patients were completely managed under the new conditions. Between three and 38 visits were conducted for each of the patients, adding up to 293 visits in total. The total number of visits split into 38 screening visits, nine staging visits, nine pre-surgery visits, and 237 chemotherapy visits. Staging and pre-surgery visits were documented only in two cases and are therefore excluded from this analysis. We received 224 evaluation forms for the chemotherapy visits, and seven for the screening visits.

The number of data items requiring assessment influence both manual and electronic data capture. This depends on the visit type. Due to the nature of clinical research compared to clinical care, it is almost impossible to find 100% of all relevant trial data in the clinical documentation. The coverage of trial data by the available care related IT systems depend on the specific trial. We investigated the number of data items directly filled from the care documentation into the eCRF and those data items that are derived from such data and automatically filled into the eCRF. Derived data can be calculations of score values, age calculation out of date of birth, etc.

Between 48 and 69% of the eCRF data could be filled in automatically using the integrated EHR–EDC solution (Table 3). The screening visit with the broad assessment of the health status and disease stage had rather high need for manual data entry, meaning that the majority of the data was not available in the electronical clinical care documentation.

Data analysis and comparison is provided on a per visit basis due to the differences in the number of data items that were captured for the different visit types.

In spite of the training of the personnel regarding the evaluation study, it was difficult in the normal clinical environment to get feedback on the time effort for screening visits (Table 4). Comparing the time effort for the traditional way of data capture with looking into care documentation and searching for the relevant screening data plus entry into a paper form is significantly longer than using the integrated system.

The effect of automatic data transfer and online validation of automatically transferred data in the integrated EHR–EDC solution was superior regarding time effort compared to the conventional data entry procedure of typing from a sheet of paper into EDC (44.6 vs. 36.5 min).

Again, there is a statistically significant difference between the time effort needed for the paper-based process versus the use of the integrated EHR–EDC solution. The integration advantage is the same as for the screening visit. The time effort for the routine chemotherapy visit is significantly lower than for screening (343 data items vs. 161 data items for the visits respectively) (Table 5).

Using the integrated EHR–EDC solution, all visit data was available within 24 hours after the respective visit had started.

All personnel engaged in the clinical trial filled in forms regarding the time spent on the specific trial per week. Table 6 shows the result across all personnel groups and all time periods of the evaluation study.

On average, the personnel spent 375 minutes (6h 15min) per week on the HEDON trial in the paper-based process. The integrated solution required 98 minutes (1h 38min) which is a statistically significant difference.

The paper CRF design was done within 4560 minutes (76 hours or 9.5 working days). Having the pCRF in place, it took an additional 7860 minutes (131 hours or 16 days and 3 hours) to develop the eCRF for the trial in the EDC system. The integrated system was set up and tested for this specific trial after five working days. Training of two persons at the trial site took 50.5 hours. We used the train the trainer concept.

The processes of the data entry were documented and analyzed twice, first during the 'traditional' time using paper based CRF (pCRF) and second after switching over to the electronic CRF (eCRF). The process steps and changes of the screening visit procedure are similar to those of the chemotherapy visits. Due to the different preparations which may differ from patient to patient, it is difficult to sho

w all possibilities within one diagram. Therefore, process changes will be demonstrated in detail for the chemotherapy visits.

Paper-based vs electronic
The 'traditional' visit for chemotherapy consists of six process steps. After checking if all necessary information is available and which parameters still have to be collected, the physician documents the results of the examination in the patient records. Ideally the study nurse collects the records within five days after the visit and enters the required data in the pCRF. All pCRFs are stored in a folder for later data entry into the EDC system and for analysis at the end of the trial.

Compared to the 'traditional' way the basic processes of the 'new' one are nearly the same in the hospital in Munich. Instead of the paper-based CRF the electronic one is filled within process step six, 'Data Capture'. The ideal process of the visit for chemo therapy using the integrated EHR–EDC solution has five steps (see Figure 3). Data are entered directly into the eCRF by the chemo nurse or study nurse/physician depending on availability. Finally, after the visit the study nurse or physician has to validate the data coming directly from the laboratory. The data is available the same day by automatic transfer in a complete fashion.

There was no change in the overall number of the involved personnel throughout the clinical trial, yet roles were changing with the switch over to the integrated EHR–EDC solution. The data manager was more engaged in the EDC system, the project manager function was split into the trial specific tasks and the new technology oriented activities such as training on the new technology. The time spent on the trial per week by the different trial personnel can be calculated into a cost reduction per role. Since for the Munich pilot gained time did not translate into personnel reduction, this study demonstrated that the same number of personnel could then deal with more trials in the same time period.

During the five months of conducting the HEDON trial paper-based, there were two monitoring visits. Two queries needed resolution, which was then finished within five hours. There was no monitoring during the use of the integrated EHR–EDC system.

Discussion
The integrated EHR–EDC solution has now been successfully implemented and running for three years in a real life clinical setting at the Frauenklinik in Munich. It was proven that there is an immediate effort reduction due to using such a novel approach for conducting clinical trials. Regular clinical trial personnel can administer the stable technical solution without the help of IT programmers. The integration between the hospital IT systems and the EDC solutions can be used for any type of EHR or EDC system. One of the solution's benefit is its adherence to existing/emerging standards in HL7 and CDISC ODM. The mapping between both data worlds has to be done once and is then re-usable for other trials. To comply with the recommendations of CDISC eSDI the development of a validation buffer was needed. This buffer was especially developed by InferMed Ltd. since such solutions are not available with any of the commercial EDC systems.

The HEDON clinical trial is a very complex trial setting and was chosen as a challenge for this new technology. Therefore, the eCRF setup was done under supervision of experienced eCRF designers of InferMed Ltd. The breast cancer scenario was chosen due to its relevance and number of clinical trials that are conducted on this disease type in order to improve therapies. For the integrated EHR–EDC system, this is translated into a design of re-usable modules for the integration engine and the eCRFs. This will allow reducing the effort required to set up the eCRF for the next breast cancer trial. Furthermore, some of the lab oriented eCRF were designed as basic forms that will be suitable for any cancer related lab data capture.

The train the trainer concept was very successful. After the initial training of the trial management, further training of trial personnel was completely managed by the clinical partners at the Technical University of Munich. This requires specialized and experienced personnel that ideally have both clinical knowledge and IT experience. Such complex solutions can thus be more easily communicated having such personnel available.

Generating an eCRF is a time consuming process. It is important to adhere to good practices while going from pCRF to eCRF. It is essential to allow a separate validation step for the pCRF by a qualified data manager before the ‘translation’ into an eCRF.

A further effort reduction for new trials can be achieved by re-using a large fraction of the eCRF structures defined in the current trial. Therefore it is important to allow enough time for the design phase of the eCRF for a specific disease area if more than one trial is planned to be conducted. In addition to setting up the eCRF for a new trial with the integration, interfaces of the system have to be adapted re-using the existing interfaces and thereby reducing the effort dramatically.

Even with a high percentage of automatically filled in data items into the eCRFs, there is almost never 100% of data at the clinical care site for specific clinical trials available and this can hardly be expected. From today’s perspective in Munich, the percentage of re-usable data can be increased by going deeper into the clinical documentation system (e.g., utilization of newly introduced databases as additional data sources) than it was possible three years ago. For every trial that utilizes the integrated solution, it is necessary to make an assessment of whether the expected percentage of automatically filled in data items is sufficient enough.

Study nurses play a vital role in the data entry processes. Looking at the times for data entry (Table 4 and 5) and comparing the times to enter data into the pCRF versus eCRF, a difference is noticeable and this can be explained by two factors:

a) Automatic transfer of the lab values: All other data has to be manually looked up in the clinical documentation
b) Utilization of the patient aware portal that allows the user to automatically launch the patient’s EHR record

In the case of pCRF, an additional time effort is required for the data transfer from the pCRF into the database. Including all users who worked with the system and investigating significant differences in the weekly efforts, we found that this time difference can be explained largely by a reduction of the effort for searching for the patient’s paper files.

The automatic transfer of the lab values was one of the most convincing arguments for the study nurses to use such a system. This is easily explained considering the large time efforts spent on searching for data in the clinical documentation that are relevant for the clinical trial in order to fill the pCRF properly plus the effort needed for manual data entry into en EDC system.

The availability of all visit data within 24 hours was one of the biggest achievements of the Munich pilot. This is the prerequisite for timely reactions of trial sponsors, interim data analysis, and to modify adaptive trial designs under real life conditions.

Monitoring activities such as comparing the source data with the eCRF entries is an easy task in this integrated solution due to the context sensitive access to the clinical and trial data with a single sign on technology. Should this procedure be used for monitors of a pharmaceutical company, negotiations (data security, data transfer, etc.) with the clinical site are necessary.

There were several new activities related to the implementation of the integrated EHR–EDC solution that were never done before by the engaged personnel. One result was a change in definitions of roles, like the data manager. The Munich personnel was adapting to those changes in a professional way. Therefore a direct comparison is difficult due to those fundamental changes. However, there is potential for further improvement especially regarding the data entry and data management processes using the new technology. In a clinical environment, where most of the documentation at the physician’s workplace is done on paper, the benefits of an integrated, but finally electronic system can not be fully exploited. Due to the dominance of the clinical care process it would require to change this process first or to define separate research processes with newly defined links into the care process plus additional personnel. Under these circumstances we can only imagine how much better the final solution could be in a truly electronic environment.

Overall this was an extremely successful pilot application that is running in its current form until the two clinical trials are finished. The technological solution is that well accepted now by all clinical trial stakeholders not just in the Frauenklinik but the entire Hospital that a roll-out to other clinical departments is planned. This will be one step forward to a more scalable solution. At a European scale there is an EFPIA (European Federation of Pharmaceutical Industries Associations) initiative together with the European Commission and the health care IT industry to bridge health care and clinical research. We see the described work as one milestone on this road.

Acknowledgement
The excellent collaboration with the investigators of HEDON trial, D. Paepke, MD and N. Gottschalk, MD, the study coordinator, monitor, and study nurses Ms. Röthling, Ms. Schuster, Ms. Lindner-Pöppl, Ms. Markhof is highly appreciated. The authors are grateful for the assistance of the Munich Study Center (MSZ).


Prof. Dr. med. Marion Kiechle is director of the gynecology department; Dr. med. Dipl. med.Stefan Paepke is head of diagnostic and operative senology; and Dr. med. Ulrike Schwarz-Boeger is senior physician medical informatics, controlling and coding, Technical University Munich, Klinikum Rechts der Isar, Frauenklinik, Ismaninger Str. 22, München, Germany. Prof. Dr. med. Nadia Harbeck, Head of the Interdisciplinary Breast Center, University Hospital of Cologne, Germany, email: nadia.harbeck@uk-koeln.de. At the time of submitting the publication, Nicole Harzendorf was a student at Siemens AG, Healthcare sector, Image and Knowledge Management, Henkestr. 127, Erlangen, Germany; she has since completed her Bachelor's Degree, email: nicoleharzendorf@web.de. Gudrun Zahlmann, PhD, Manager Imaging Infrastructure, F. Hoffman-La Roche Ltd., Basel, Switzerland, email: gudrun.zahlmann@roche.com. Markus Schmidt, Project Manager, Siemens AG, Healthcare sector, Magnetic Resonance, Alle am Röthelheimpark 2, Erlangen, Germany.
 

References
1. Convergence of Electronic Health Records and Clinical Trials, Frost & Sullivan Report Mo 42-48, (Nov. 2006).

2. S. Lusignan, C. van Weel, "The use of routinely collected computer data for research in primary care: opportunities and challenges," Fam Pract 2005;23:253-63.

3. J. Powell, I. Buchan, "Electronic health records should support clinical research," J Med Internet Res, 2005;7:e4.

4. P. Bleicher, "Integrating EHR with EDC: When Two Worlds Collide," Applied Clinical Trials, March 2 2006.

5. H. Donovan, "EHR & EDC: Tomorrow’s Technology Today," Applied Clinical Trials, Feb 1 2007, 54-62.

6. R. Kush, et al, "Implementing Single Source: The STARBRITE Proof-of-Concept Study," JAMIA, (14) 662-673, (2007).

7. T. Mason, "CDISC marketing material for HIMSS 2007," CDISC (2007).

8. M. Schmidt, et al, Integrated study platform, IJHTM in review.

9. http://www.cdisc.org/eSDI/eSDI.pdf.

10. J. Pröve, Personal information (2006).

11. C. Connor, "U.S. Electronic Data Capture 2006-2011 Spending Forecast and Analysis," http://www.healthindustry-insights.com (May 2007).

12. I, De Zegher, "EHR usage in Clinical Research draft recommendations from EFPIA task force," Session on EHR and its use in clinical research Barcelona, DIA Europe (2008).

13. C. Ohmann, et al, "Prozessbeschreibung zur Durchführung klinischer Arzneimittelstudien im Krankenhaus (Metastudie)," KKS Projekt: 04P1_017, (2007).

14. H.J. Schmelzer, W. Sesselmann, "Geschäftsprozessmanagement in der Praxis," Hanser Verlag München, Wien, p.123 ff, (2006).

15. Q. Yang; C. Ohmann, "Prospective before and after eClinical Trial Management Study–Evaluation Study–Auswertung," KKS Projekt, Internal report.

Related Content:

Online Extras