bm980611 _______________________________________________________________________ Tema: Data Management in Clinical Trials (Gestión de datos en ensayos clínicos) Imparte: Nancy A. Hardy, M.P.H., M.S. Epidemiologa. Facultad de Farmacia, Universidad de Minesota. USA _______________________________________________________________________ *** MJesus changes topic to "CHARLA: Data Management in Clinical Trials.(SLIDES ON: http://www.pharmacy.umn.edu/seoan001/esv/Data/sld001.htm)" [22:44] 4 atencion [22:44] 4 atencion [22:44] 4 atencion [22:44] 4 dentro de breves momentos va a comenzar la conferencia [22:45] 4 correspondiente al tercer tema del curso [22:45] 4 curso de metodología en investigación con [22:45] 4 medicamentos [22:46] 4 que dirige la Dra. Jimenez Buil, de BootsHealtcare [22:46] 4 en colaboracion con la Unidad de Investigacion del Hospital Genenral Yagüe de Burgos [22:47] 4 en esta ocasion esta con nostros la Dra. Nancy A. Hardy, M.P.H., M.S. [22:47] 4 epidemiologa de la universidad de Minnesota, Facultad de Farmacia [22:47]  [22:47] 4 para tratr el tema "Gestion de datos en Ensayos Clinnicos" [22:48] 4 bueno, ¡¡gracias Nancy! [22:48] 4 como siempre, primero expone ella, y al final se abre el coloquio [22:48] 4 en esta ocasion la charla es en ingles.... ahora la Dra Buil les presentara esto en la lengua de Shakespeare. [22:49] We are very delighted to introduce to [22:49] Mrs Nancy A. Hardy, epidemiologyst from [22:49] Minnesota University. She has been working [22:49] about clinical trials for a lot of years and we [22:49] are sure that we will take advantage of her [22:49] experience. Nancy is going to speak about [22:49] Management of Clinical Trials. [22:49] Thanks a lot Nancy for your effort to speak [22:50] in this channel. Welcome and you can begin [22:50] when do you want..... [22:50] here we go... [22:50] Web for this lecture in: http://www.pharmacy.umn.edu/seoan001/ curso.htm [22:51] Thank you for the nice introduction. [22:52] I do not know what experience you all have in this area, so I start with the simplest. [22:52] Please do not be insulted. [22:53] hhhehhehehehheheehe [22:54] When you start thinking about your clinical trial - its important to think about what you wish to accomplish. [22:54] For example, what do you wish to accomplish? Who will benefit and whsat do you want your end result to be. [22:55] Data systems should be developed with the 'end' in mind. So often people do not consider the data system and its design before a clinical trial starts. [22:56] When you know what your 'end' product will look like it is easier to design a data system that will [22:56] help you accomplish your goals. [22:57] you first need to know 'what the unit of analysis' is. [22:57] for example are you interested in finding out about improvement in a 'person', a 'body system' or a group of people. [23:00] If you are interested in persons, your data system should be set up so that one case or row is a person or can become a person. [23:01] Know what type of data you are collecting on your 'unit of analysis'. Is it nominal data, ordinal data, or continuous data. [23:02] At all times make sure that you keep all of your data, regandless of type, in its rawest form. [23:03] So many times poeple have given me data that they have spent hours calculating...when I could have calculated it for them in seconds. [23:03] Their calculations are full of mistakes that could hae been avoided. [23:04] When you know the unit of analysis, and type of data you must next consider the type of software you need into which to enter the data. [23:05] What will you do with the data? If you simply want to keep records... a database program will be helpful. [23:05] If you are going to perform statistical analysis, a statistical software should be used. [23:06] If you are going to perform financial analyses, a spreadsheet may be the best. [23:07] Most clinical trials that I have been involved in have used several types of software to complete their studies. [23:08] for example, a database is used to track patients, hours, clinical examination rooms. A statistical software is used to test results. [23:09] It is important to insure that you 1) have the correct software for each task and that 2) you can put the data you collect into many type of software packages. [23:10] For example, Microsoft uses "DDE" dynamic data exchange which allows data to be easier transferred between software packages. [23:11] Almost any data can be transferred to an ASCII file. Most software packages will write to ASCII formats. [23:12] Next you want to set up all the documentation you will need. Every study should havbe a data code book. [23:13] In it you should record every variable you are collecting data on. All the variable names and the value labels associated with them. [23:14] let me give an example. If I am collecting data such as height, weigth, sex and education level, [23:14] hello celeste and peio: see the topic (slides) [23:14] I will name my variables ht for height, wt for weight... et cetera. I will document that sex has values 1 and 2. [23:14] hello [23:15] 1 will indicate 'Male' and 2 will indicate 'Female'. I will document that the only acceptable values are 1 or 2. [23:16] I will also document that a 'blank' indicates 'unknown' or refused to answer. [23:17] In this code book I will also keep the known ranges of each variable. To follow my original example, [23:17] I will not accept a person's weight if it is a zero or a negative number. I will not accept a weight if it is over 1000. [23:18] When each variable is documented, I will know its name, values, and acceptable ranges. [23:18] Lastly it is important to document the logic rules associated with each variable. [23:19] If a person is a Male, I will make a rule that birthcontrol pills cannot be marked "yes". [23:20] there are some very good data entry software programs that automate the information in a code book. [23:21] they will keep all variable labels and values, ranges and logic rules for you. [23:21] It is important to print these out so that you have a printed copy of them. [23:22] Along with a code book, you should also keep a study book. This book includes such items as your study budget, correspondence, [23:22] a list of involved people and other pertinent data information. [23:23] Most importantly this study book should contain your research question(s) or hypotheses. You must have at least one of these [23:24] to do a clinical trial. It is helpful to have a time-line in this study book also. It will keep you on track. [23:26] Lastly...it is important to have a programming log. In this book you keep all the analyses you do on you clinical trial. [23:27] You must be able to document how you got to the results. The programming log should be kept in such a way that [23:28] each program is names with the same name as the output. For example, a program describing age should be named [23:28] age.prg and its output can be names age.out --- this way you can easily match the two. Each entry into the program log should be dated. [23:29] 4 I have been 'yaking' about the basics for quite a while. I will stop for a moment for questions. [23:30] ok Nancy!! [23:30] ok [23:30] aha [23:30] Nancy please.. [23:31] do you know any benefit whith Microsoft batabase over unix system? [23:32] Both are good. Unix is an operating system with some excellent software available for it. [23:33] how do you manage in clinical trials adverse events?I mean what kind of sofwtware do you use? [23:33] It does have some very sophisicated abilites that Microsoft does not have. If you need to use a mainframe (i.e. you have a very large database [23:34] you might want to use Unix, however if you have a small clinical trial - Microsoft is made for a personal computer [23:34] and it should suit your needs nicely. [23:35] If there are no more questions, I will continue. [23:35] yes nancy.. [23:35] vidal? [23:35] I have a question.. [23:35] Ok [23:36] how do you manage in clinical trials adverse events...Imean [23:36] how kind of sofwarw doyou use?? [23:36] wait [23:37] ORACLE? [23:37] Oracle is a good choice. I have always kept the adverse events in a database program---Oracle is a very sophicated [23:38] mainframe database program. Keep in min d that if you wish to do statistical analyses on the adverse events [23:39] Nancy, have u tested qnx? for getting data? [23:39] ok [23:39] you will want to transfer the data into the statisical software. I suggest you do a trial run on this before the [23:40] clinicla study begins. some statistical programs can be fussy about how your data is formatted. [23:41] they can treat missing data differently in different programs. Check the manual or have your computer person [23:42] confirm that you can transfer the data. [23:42] also make certain that the unit of analysis is the same. you may have set up your data entry for an eye, but [23:43] you will want to analyze adverse events for a person, and you must manipulate your data accordingly. [23:45] On the qnx question. I have not used that particular software, however I believe that it is a very good program, expecially for data entry. [23:45] I believe it has good logic rules and data validation. [23:45] sasha traduce. [23:46] no entendeis? [23:46] Estoy de acuerdo que a la hora de preparar toda gestion de datos, es importante el crear un pequenia maqueta [23:46] this brings me to my next topic: Quality Insurance and quality assurance. [23:46] no, lo mio, es que no se como se dicen algunas palabras ( lo siento) [23:47] sacha ¿puedes traducir eso?? [23:47] ok...vidal says.. [23:47] he agrees to do gestion of data.. [23:48] pero como puedes crear una maqueta de un estudio sin que ello te lleve a hacer un mini studio?, o tener ya al menos algun dato sobre el estudio? [23:48] it's very important to create a plan.. [23:48] 4 Enrique is here and will translate for me from Spanish to English. [23:48] o sencillamente sin tener claro que tipo de equipamiento hard vas a utilizar para el analisis [23:50] Que cantidad de tiempo deberiamos gastar en evaluar soluciones mas optimas sin probar / o soluciones probadas anteriormente con exito? [23:50] you do know what your study is about and what data you are going to collect. [23:50] Therefore, you can create a small 'fake' dataset of data that you think you will get. [23:51] Most clinical trials start with 'pilot data'. [23:51] this pilot data is actually a small clinical trial -- to make sure everthing will run smoothly in the actual trial. [23:52] hello Aurora, luis.... see the topic! [23:52] hi pilar! [23:52] it can help you find out if there are mistakes in your design and in your data management plans that you could [23:52] not otherwise predict. [23:53] Back to quality insurance and quality assurance. [23:53] the plans you have to insure that your data is accurate should be planned before the study begins. [23:54] Quality assurance are those things you plan that prevent you from making mistakes, such as logic rules or data validation. [23:55] software can be set up to prevent you from entering data incorrectly, such as entering a 3 fro sex, when 1=male and 2=female. [23:56] there is no such thing as a sex coded 3. the software will 'beep' at you if this happens. [23:57] this is perfect!isn't it? [23:59] yes. On the same token, quality insurance measures should be taken. these measures make sure you correct mistakes after they happen. [23:59] a pesar de todo?, deberia de incluirse algun tipo de proteccion, como eliminar ese registro, si por error en la manipulacion posterior de los datos the sex =3 ? [0:00] so even with the quality assurance measures, I go back in afterward and make sure that only values 1 and 2 zre entered the [0:00] the data for the variable 'sex'. [0:01] We should ionclude protect and check thereafter. There are many applications of this in clinicla trials. [0:02] Almost every step of the trial should have a quality assurance and quality insurance step attached to it. [0:02] Accuracy is so important. If your data is incorrect, even slightly, it can effect the results and the people [0:03] who will use these results. Another example of quality insurance is double entry of data. [0:03] Data should be entered either by two people, or twice by the same person. [0:04] the ideal is to have two people enter the same data and then compare each entry for accuracy. If this is not [0:05] possible, one person can enter the data into a software package twice and compare both entries. [0:06] Some software packages will do this automatically... you enter the data, then the screen is blanked and then you enter the data again. [0:07] It will stop you every time you do not agree with the original data and will ask you for verification. [0:07] When your data is accurately entered, you may still find errors. It is essential that you monitor the data over time. [0:08] This way you can do two things: 1) catch additional data entry errors and 2) watch and make sure that your subjects [0:09] are not getting worse and they are using the drugs or implement. [0:09] How often you monitor depends on what you are testing, however always monitor and generate reports periodically [0:10] to insure that you are 'doing no harm'. [0:10] If you must change data, keep a log of data editing. This is usually a list a changes with initials next to them, [0:11] verifying who made the changes. [0:11] also make sure that the measures you are taking, be they blood levels etc., are valid and reliable. [0:11] Validity referes to the ability to measure accurately. [0:12] Reliability refers to the ability to measure the same over time. [0:13] I think I have talked enough! i hope you enjoyed what I had to say. [0:14] yes [0:14] Thanks for being such a good audience. this was my first time in the 'chat'. [0:14] very interesting [0:14] 4plas 5plas 6plas 7plas 8plas 9plas 10plas 11plas 12plas 13plas [0:14] 4plas 5plas 6plas 7plas 8plas 9plas 10plas 11plas 12plas 13plas [0:14] 4plas 5plas 6plas 7plas 8plas 9plas 10plas 11plas 12plas 13plas [0:14] 4plas 5plas 6plas 7plas 8plas 9plas 10plas 11plas 12plas 13plas [0:14] 4plas 5plas 6plas 7plas 8plas 9plas 10plas 11plas 12plas 13plas [0:14] 4plas 5plas 6plas 7plas 8plas 9plas 10plas 11plas 12plas 13plas [0:14] 4plas 5plas 6plas 7plas 8plas 9plas 10plas 11plas 12plas 13plas [0:14] 4plas 5plas 6plas 7plas 8plas 9plas 10plas 11plas 12plas 13plas [0:14] devotee I have a question for you [0:14] Of course, ask! [0:14] plas,plas,plas [0:14] your fist time in the chat is very well!! [0:15] plas plas [0:15] Thank you! [0:15] thank you v very much!! [0:15] :) [0:16] Nancy you have done a great job at dealing with two big issues...not just clinical trials but the evaluation of data using current technology [0:17] Good - bye everyone [0:17] *** Nancy (orbital@x171-157.pharmacy.umn.edu) has left #biomedicina [0:17] Nancy..... Magnolia is in New Zealand, Celeste in Brasil (pathologist), Peliroja in Miami (Microbiologist), Luis in Argentina.... [0:17] Great!! Nancy.. [0:17] se fue [0:17] :( [0:18] anda..... y eso? [0:18] buaaaaaaa [0:18] sniff [0:18] Nancy, good explains [0:18] hehehe [0:18] intensivist [0:19] job done, out of here [0:19] *** Enrique (orbital@x171-157.pharmacy.umn.edu) has joined #biomedicina [0:19] hola a todos [0:19] enrique... esta ahi Nancy? [0:19] Nancy sigue aqui, le he dicho que no se vaya [0:19] si, esta aqui [0:19] hola!! [0:19] el comentario que le iba a hacer a Nancy era....que me gusto que hiciera referencia a hablar con un Statistician para evaluar mejor la data [0:19] dile que ha estado de maravilla, por favor [0:19] Nancy..... Magnolia is in New Zealand, Celeste in brasil (pathologist), Peliroja in Miami (Microbiologist), Luis in Argentina.... los demas, en españa [0:20] jose, de argentina [0:20] dile que ha hecho un trabajo genial!!! [0:20] bueno.... me gustaria saber una cosa. [0:20] ¿por que no se hace el mismo protocolo para ensayos en usa-canada y en europa? [0:21] a ver preguntas para nancy [0:21] Nancy, I thougth it was very wise of you to refer to a Statistician to help out in the evaluation of data [0:22] I wish the regulations were the same. but different countries have different regulations. [0:22] it is all political! [0:22] jajajajaja blame it to politics! [0:23] but...it is true! [0:23] but are very similar!!! [0:23] hahhahahahahha [0:23] in the other countries... [0:24] protocols, rules etc.. [0:24] The next coference on this, maybe we can get more sophisicated. [0:25] Enrique told me it is very late in Spain and I should tell you good-bye! I think he's hungary. [0:25] Sorry- he's hungry. [0:25] but, tuday mornin, we are evaluating a protocol for ....europan countries. It was diffrent for the usa'protocol [0:26] bye nancy /enrique [0:26] Yes, the regulation is different. [0:26] heehhe hungary or zchec? [0:26] Maybe it has to do with the american society, overall, it is highly litigious [0:27] bueno, ya se ha ido Nancy [0:27] re henry [0:27] enrique y nancy ..........the great disappearing act! [0:27] ha sido muy estpendo [0:27] ha estado genial! [0:27] jajajjaja [0:27] gracias [0:27] :) [0:27] bueno.... esto supero las previsiones [0:27] le he pedido a Nancy que hicera algo introductorio [0:27] pleasure! [0:28] y que en la proxima conferencia que nos de haga algo mas sofisticado [0:28] ha cubierto muy bien un tema que es tan extenso [0:28] y el management de bases de datos, very well [0:28] que pena no pudo quedarse un poquitin mas para hacerle mas preguntitas [0:29] se las podiamos decir a Enrique? [0:29] si, pero es que ya es demasiado tiempo, y es hora de comer [0:29] talvez puede hacer el mas sofisticado en partes.... [0:29] pero eso si, podeis decirme las preguntas y os prometo que las hare llegar a nancy [0:30] que hora tienen alli? [0:30] aqui son las 2.28 de la tarde [0:30] 5.28 [0:30] hora de comer [0:31] entonces sera cenar [0:31] Pregunta a Nancy: Do you or Information specialists design the database? [0:32] Yes, she designs the databases [0:33] does she prefer to work with oracle or which other database? [0:34] bueno... ¿cuantos ensayos hay en estos momentos activos en la mayo clinic?.... aproximadamente? [0:35] bueno, gracias por la invitacion MJesus, muy interesante la metodologia, haremos algo asi en mintensiva? ojala que si, saludos nos vemos despues [0:35] creo que desean irse a comer [0:36] por supuesto luis! [0:36] hay que montarlo ya [0:36] does she prefer to work with oracle or which other database? [0:36] It depends on how large the study is [0:37] ¿cuantos ensayos hay en estos momentos activos en la mayo clinic?.... aproximadamente? [0:37] grx [0:37] miles de ensayos clinicos [0:37] entre los que ellos hacen [0:37] y los que participan [0:37] miles??? [0:37] Nancy says "hundreds" [0:37] estan con algun tema [0:38] en mi hospital tenemos unos 50 activos en estos momentos [0:38] ¿estas seguro de que son miles? [0:39] eso iba a preguntar [0:39] tantos? [0:39] quien financia? [0:40] supongo que la industria no? [0:40] bueno, chicos/as [0:40] un beso a todas y un saludo muy cordial a los demas [0:40] *** Enrique (orbital@x171-157.pharmacy.umn.edu) has left #biomedicina