19 February 2018
As is well known, clinical studies cannot get along without conducting laboratory tests. Laboratory data is a significant tool for selecting study subjects, monitoring treatment compliance, evaluating study drug distribution in the body, its safety and efficacy
When the developing market of contract research organizations services raised the issue of the laboratories use, central laboratories were the only solution for uniformity and quality of biomaterials processing in multicenter studies. Central laboratories are not always economically effective because of high logistics cost, transportation of biomaterials can be difficult as well, therefore the capabilities of local laboratories are increasingly being used both in Russia and abroad.
What is the complexity of local laboratories data processing?
Each local laboratory, striving to correspond to the "gold standard" in multicenter clinical trials, uses modern equipment and reagents, regularly calibrates and verifies measuring instruments. The completed work positively characterizes local laboratories, however, abundance of local practices, difference in management and work experience, availability of numerous equipment modifications cause variety in the test sets conducted and format of results provided, which leads eventually to data inconsistency
This problem is the main one for organizations providing data management and statistical data analysis services.
DM 365 company regularly faces the need to organize data collection into the Electronic Data Capture (EDC) system and to adjust the system to a certain clinical study features. Requirements are often challenging, as the client draws on the capabilities of the laboratories of the contracted clinical sites, and provision of a database that meets all the protocol requirements fit into clinical environment is the main goal of DM 365.
How do we cope with it?
Laboratory test results constitute the lion's share of the collected data, so it is very important to make the laboratory forms logical and ergonomic. The process of entering test results should be convenient and comprehensible for the investigator; Clinical Research Associate should have an opportunity to easily verify entered values, and Data Manager – to quickly and efficiently validate the data using automatic and manual data cleaning methods.
When creating forms in EDC, DM 365 is guided primarily by standard procedures and guidelines, both international and internal. Automated checks are programmed on the fields of the forms to compare entered results with the normal ranges submitted by the local laboratories. Normal ranges are age- and gender-specific reference values for assay parameters and measurement units determined by the analyzer. Unlike central laboratory, local reference ranges and laboratory assessment techniques are not unified, have different names for one and the same parameter that leads to automated checks complexity.
DM 365 developed a whole range of solutions for the most difficult problems related to the topic. Below are some of them:
1) Problem. Equipment calibration and reference values updating schedule are specific for each institution, and we are obliged to ascertain the relevance of all applicable normal ranges.
Decision.Therefore an expiry date of the reference values is set up in the EDC for each laboratory. Thus, the system refers only to the ranges that are currently valid.
2) Problem. A challenging task for DM 365 is to solve the problem of processing qualitative analysis results where the results are presented differently by different laboratories: for example, as text, using words like "negative", "not found" or symbols like "0", "-". According to the requirements of Good Clinical Practice, investigator is obliged to enter data into the eCRF exactly as they are recorded in the source documents.
Decision. To simplify processing of such data, it was decided to develop an algorithm for "conversion" of text results to numerical ones, informing investigators about that both at training sessions and in EDC, placing a brief instruction directly on the form. For example, "not found", "-" or "negative", investigators should write as "0". Values conversion instruction is developed for each study individually, based on the submitted laboratory normal ranges. As a comparison: Quantitative analysis data, originally represented in numeric format, undergo an automated standardizing process during export. The process is supported by implementation of a validated set of value conversion rules.
3) Problem.There is a chance to encounter such parameters as Urine Color and its Transparency when processing data from urinalysis. In case various local laboratories are involved in the study and each laboratory has its own idea of these parameters, it may be impossible to create automated checks that will compare entered data with the laboratory normal ranges. The Color can be yellow, amber-yellow and straw-yellow, Transparency is normally characterized as complete, but such definitions as "cloudy", "nebula" and others can be seen, too. Abbreviations and misprints are not rare.
Decision. In these instances to provide biostatisticians with clean data, laboratory results are validated manually; queries and obvious corrections are applied, where necessary for data correction and unification. It should be noted that free text data are less structured and less manageable, so Data Managers pay special attention to them. Unlike other medical terms, laboratory data cannot be coded with special dictionaries yet.
4) Problem. Delayed submission of laboratory reference values by local laboratories slows down the process of eCRF building, thus affecting project timelines. Inclusion of new sites to the study after its start may lead to the design modification (adding new measurement units to the drop-down lists).
Decision. At the kick-off meeting with the client, special attention is paid to the importance of timely provided laboratory reference values for all participating sites. Project Manager, in conjunction with Data Manager monitors receipt of the normal laboratory ranges and promptly processes the information. EDC system can be configured such that it pulls relevant units from the chosen site-specific analyzer.
5) Problem. Non-uniform reference ranges representation by local laboratories. Diverse structure of the laboratory documentation slows down the process of extracting required parameters from the lists provided. Obviously, manual processing implies the risk of human errors and its mitigation takes extra resources.
Decision. DM 365 team developed a universal document-template for laboratory reference ranges collection. A standardized document allows you to quickly collect necessary parameters, enter their normal ranges into the system and make corrections to them, if required.
Of course, the list of the problems described above may be either appended or shortened, as time goes by.
DM 365 Data Managers team is always ready to face new challenges and find the best ways to solve them. If you encounter similar problems or need advice, contact us, and we will be happy to help you!