Honoraria - Leveraging Enhanced EDC Integrated CTMS
Functions

Issues surrounding honoraria paid in clinical studies increase exponentially with the size and scope of the research. Payments for screening, enrollment, testing, data collection et cetera that are manageable in a small feasibility study or Investigator Initiated Trial, become a time-consuming, high overhead task with dozens or hundreds of sites and an equal or greater number of investigators and patients. In addition, anti-kickback statutes and their interpretation by the Department of Justice and Office of Inspector General dictate that sponsor organizations remain in the safe-harbor of paying only for work performed. From the investigative site's perspective, they wish to be reimbursed at the earliest available opportunity for the expenses they have incurred in participating in the research, and may be intolerant of delays associated with ensuring regulatory compliance.

This paper discusses two approaches taken by clientele of MedNet Solutions in leveraging enhanced electronic data capture (EDC) to reduce the burden surrounding honoraria compliance and increase investigative site satisfaction - in effect, leveraging the EDC system to provide AutoHonoraria CTMS benefits.

Background on system characteristics enabling AutoHonoraria vs. traditional methodologies
Workflow processes in making honoraria payments in clinical studies/initiatives are essentially the same irrespective of the system used. In the simplest rendering, the process may be described as activity>validation of activity>payment generation. A simple concept rapidly becomes complex within the clinical world as compliance issues, SOPs, GCPs and other elements exert influence on the processes. A traditional methodology such as paper CRFs or fax submissions adds a number of steps within each process element, as can be seen in figure 1. Each step in and of itself is not particularly difficult or complex, but the process when taken as a whole can certainly be so - in this example , the whole is greater than the sum of its parts, especially when a manual process necessarily involves multiple personnel.

Figure 1. Process steps in traditional methodology
(Click image for larger image)

The steps outlined in figure 1 are essentially manual processes, and as such are 1) time-consuming, 2) prone to human error/mis-keying and 3) relatively inefficient, and thus present compliance and site satisfaction issues.

Utilization of enhanced EDC can, as an essential product of operation, address all three problem areas - rapid real-time processing, removal of human interaction in many instances, and significantly improved efficiency. Concomitantly, compliance is enhanced through a reliable replicable process with automatically generated audit trails and on-demand visibility/reporting, and site satisfaction increased through timeliness of reimbursement and reduction in error rates.

Implementation approach one - articulation with manual process
CellTech deployed a 370 patient 57 site hypothesis generating registry in 2004, and retained MedNet Solutions to provide enhanced EDC with integrated CTMS functions to support the initiative. It was determined that efficiencies could be achieved through reduction in process steps and articulation with the existing system. Essentially, this approach allowed the system to manage data submission and capture, validate that data, and then produce a series of reports designed to support a manual process of financial system check request and generation. Compare and contrast the fully manual process in figure 1 with that representative of the CellTech/MedNet EDC/manual articulation process depicted in figure 2 below (note: italics indicate system (non-human) process).

Figure 2: Enhanced EDC articulation with manual process
(Click image for larger image)

Reports were run on a monthly basis and served up summary information for processing in the manual system. Online "drill down" capabilities were provided for dispute resolution and error checking. EDC systems inherently populate their databases directly from data entry (unchangeable once signed by the investigator) eliminating double data-entry and opportunities for mis-keying. Other altered process steps include elimination of the human validation elements; those steps as seen in figure 1 were eliminated and replaced with a system level check - replicating the human work of checking for completion and appropriateness at computer level speeds of millions of instructions per second (MIPS). As with any enhanced EDC, the validations were derived from a data dictionary describing field, variable type (alpha/numeric, characters), high/low cutoffs, cross-form/visit delta variance checks, etc. and may be of hard (i.e. preclude entry) or soft (i.e. prompt user for confirmation at time of entry) types. In some instances, apart from speed and accuracy, advanced elements such as delta variance* checks cannot be performed by any but the most highly skilled human operators, and, even if detected, require significant time to resolve: in this implementation such checks are performed real-time and error rates are reduced.

Figure 3: Monthly Honoraria report for manual processing
(Click image for larger image)

Implementation approach two - fully automated process
During 2003, Guidant Corporation implemented enhanced EDC throughout their post-market research program. This program includes a mixture of randomized multi-arm studies, observational single arm studies, and registries ranging in size from 75-200 investigative sites and 500-2000 patients each. At any given point in time, up to 10 of these initiatives are actively enrolling or following patients. Honoraria for data collected and submitted was paid by visit, e.g., screening/enrollment, treatment, follow-up, as well as for conduct of specified testing and upload of resultant data, adverse event reporting, death/LTFU reporting etc., and mailed quarterly.

Guidant had maintained a paper-based reimbursement system for post-market studies prior to 2003. This mirrored, in essence, the process steps described in figure 1. Before 2003 this was manageable as few studies were conducted at relatively few sites - with the significant increase in investment in post-market studies, however, and in anticipation of the rapid enrolment rates and speed of "clean" data acquisition with enhanced EDC, Guidant management determined that maintaining the prior system of reimbursement represented a high-risk of requiring major human resource investment while not decreasing overall error rates.

The specifications delivered to MedNet by Guidant for development of an AutoHonoraria system included -

• Automated reporting of all activities creating payment liability
  
  
• Tracking
  • All payments due in a given period
  • All payments made to date
  • Payments becoming due in future periods
    • Site, study, multi-study level
• Ability to vary payments from site to site
  • Accounting for university overhead etc.
  • Milestone achievement payments
• Ability to pay in local currency
  
  
• Ability to serve up a single consolidated file to SAP financial accounting system
  • Must be in human-readable form for authorization purposes
  • Must include payee Tax ID/SSN, cost-center details etc.
  • Must integrate directly into SAP without modification
• Automatic cover letter production detailing
• Activities, payments associated with activity, when activity performed
• Activities incomplete/not eligible for payment – with option to not include this data as desired
  • e.g. records not locked, data incomplete

The desired outcome of a system conforming to the specifications was reduction in time and human resource burden and increase in error-free production. This was accomplished by further reducing the process steps involved in honoraria fulfillment as can be seen in figure 4, where italics represent a non-human system process.

Figure 4: AutoHonoraria Process
(Click image for larger image)

In essence, the Guidant implementation of AutoHonoraria permitted a single payment file to be produced on demand, authorization of that file, and check generation as a result.

Automated reporting and tracking tools were developed using a variety of color and symbol/icon codes to designate status. while cover-letters that detailed exactly which payments were being made reduced enquiries dramatically as compared to the previous system, phone calls from sites did occur, and quick resolution of concerns could often be achieved utilizing reports of the type shown in figure 5:

Figure 5: Site payments by activity, patient & visit
(Click image for larger image)

Consolidated payment reports by center were also provided, which proved especially useful where a center or investigator participated in more than one research initiative during a reporting period – see figure 6:

Figure 6: Consolidated roll-up reports
(Click image for larger image)

In addition to performing the functions detailed in the specifications sections above, further efficiencies were gained through utilization of mailing address/payee information for auto- generation of mailing labels and/or use of mail-house services (where volume was especially high). In keeping with the core functionality of the AutoHonoraria system, the utilization of this information, already captured within the integrated CTMS function of the enhanced EDC system, took the form of production of a standard comma separated values (CSV) file that can be read by any word-processing/mail-merge program. Additionally, automatic emails were sent to field personnel responsible for given sites with summary level information, dates of payment processing, items outstanding etc.

At time of writing, Guidant has implemented this EDC system throughout all studies. In effect, it disburses multi-million dollar honoraria payments across over a dozen initiatives in a variety of countries with differing local currencies and payment schedules. Where it was estimated that 4.5 FTE was required to manage these payments across the research department and the finance group, the function has been well-executed at a burden of 0.25 FTE – the finance group, recognizing the efficiencies, are actively encouraging other Guidant departments to investigate the technology and implement wherever possible.

Discussion
Investigators and their institutions are under pressure from administrators and others to show profitability in research. Debate over this stance is a philosophical issue worthy of consideration, but beyond the scope of this paper. What is certain, however, is that there is increasingly less tolerance for lengthy reimbursement periods and errors in payment than existed in the recent past: Investigators expect to get promptly paid for work performed, as well they should, and at the same time, sponsors need to reduce their burden in effecting payment – these two needs should not be seen as oppositional, as can be seen from the examples above. In other words, the needs of both parties to the transaction can be met through leveraging enhanced EDC integrated CTMS functions.

Supporting considerations, such as regulatory compliance concerns on the sponsor-side, are also addressed – full audit trails of who made what payment when and to whom and for what activity; and reporting on all payment activities is available literally at the press of a button: affording management teams strategic insight hitherto only available through implementation of notoriously difficult to use and problematic stand-alone CTMS products. The degree to which efficiencies may be gained and reporting may be automated is dependent on attention to re-engineering of processes and change management in the sponsor organization – that technology can assist in attaining these benefits is indisputable.

* Delta variance check refers to the ability of the system to check the entered value against previously recorded values for that field, and alert the user to probable or real error. For example, a weight field might have a low cutoff of 50 pounds, and a high cutoff of 500 pounds. If a patient had 3 monthly visits and had 152, 158, & 153 pounds recorded, a 4th visit value of 255 pounds would not be flagged as erroneous, where to the human observer it is obviously mis-keying. Delta variance would, for example, use logic such as “use a mean of all prior values for this field to establish a baseline and if the entered value on this visit is >20% divergent from that baseline, prompt user to confirm”. If applied in the example described above such programming would flag the user as the delta variance is +/- 65% divergent from the mean for this field and almost certainly incorrect.

© Timothy Pratt 2005