显示分类
最新发表
最后回复
最热评论
精华内容
2.7k
0
StatX 发表于 2014-08-08 10:15 pm

会上某位同仁提出的问题。

这其实是一个没有答案的问题。因为没有一个客观的标准,可以评价一个EDC系统是否“好用”。

我接触的EDC并不多,国外的,OC用过,inform进入系统粗略了解过,Rave看过它的界面和功能,没登入过它的系统。国内几家做EDC的也或使用或见识过,从感受上来讲,我觉得,一个EDC系统,在满足法规及数据管理本身的要求基础上,功能越简单越好,菜单越少越好,页面上的各种线框及表单越简洁越好。

一个让人眼花缭乱的EDC系统,不会是一个好用的EDC系统。

一个EDC,或者CTMS,从设计之初就要明确其核心和重点,层次要清晰分明,不能贪多求全,或者说,就算要开发一个万能的EDC系统,也不要把全部功能全部摆在那里,而应该是渐进式的,针对不同的系统角色,突出不同的功能和界面,比如对于研究者,它就应该是一个简单的录入界面和答疑界面,至多再增加个进度的显示和本中心的概况。

3.6k
0
StatX 发表于 2014-08-08 9:36 pm

今天会上有人提问,专家们也给出了回答。这里只谈下我的看法。

从理论上讲,数据录入的双录入方式,通过最后的一致性核验,来发现录入错误从而保证录入的质量,也即录入后的数据与“数据源”的一致性。很显然,无论是纸质的CRF,还是原始的数据形式,双录入都可以提高录入质量,但不能提高数据记载的准确性

为什么在实践当中,EDC均采用一次录入呢?个人认为,最主要的原因,是基于纸质CRF的数据管理过程,与完全基于EDC的数据管理过程,并不完全相同。EDC中的数据,在录入之后进行原始数据核查(source data validation,SDV),就是说它的大体流程是:

数据录入(数据获取过程)-->人工核查(SDV)-->疑问与解决-->锁定

而基于纸质CRF的数据管理流程,大体是:

完成CRF(数据获取过程)-->人工核查(SDV)-->CRF的双录入与一致性校验-->锁库

从流程上可以看出,基于EDC的数据管理过程,数据的载体就是它的各种原始形式,数据录入的过程,同时也是数据获取的过程,加上后面的SDV过程,可以看作是两遍“录入”,基本保证了录入的数据与原始数据的一致性,所以双录入不是必须的;纸质CRF作为数据的载体时,SDV是针对纸质CRF的,而纸质CRF需要通过人工录入(当然OCR也是一种方式,限于当前的技术,识别率肯定不会比人工的高)再转变为电子数据,由于此时没有了人工校验(100%,非抽查),所以双录入并进行比对检验是必须的。

12.6k
0
StatX 发表于 2014-06-18 9:03 am

Detail

Source Data Validataion (SDV)

Document

  • Documentation of Informed Consent
  • Adverse event source documentation
  • study termination source document

Case Report Form v/s Source Documents

  • CRF not intended to replace Source Documents
  • Source data items to be recorded directly into the CRF must be documented

    • CRF REVIEW

    • Performing an overall review of each CRF for: internal consistency, completeness, ogic & legibility

    • Without reference to the corresponding source documentation

    • SDV

    • Verifying CRF data against information in supporting source documents held at the site

    • electronic records are the same as for paper records
    • Monitors may perform SDV by viewing computer screens

IMPORTANCE OF SDV

  • required to provide confidence in any data reported
  • protects study staff and sponsors from accusations of fraud, misconduct

SOURCE DOCUMENT VERIFICATION

  • Refers to data being verifiable from source documents
  • Unclear with regard to process of SDV, frequency & extent
  • Process to ensure that source data reported by an investigator be:

    • complete
    • accounted
    • follows a logical sequence of events
    • support entries in CRF
  • WHEN IT STARTED

    • Preferable undertaken before data are retrieved from investigator site
    • Soon after first subject is enrolled
    • Data forwarded to data management thus valid & accurate
  • WHAT TO CHECK Defined in sponsors’ SOPs

  • Documenting SDV

    • CRFs that were checked
    • The data items checked for each CRF
    • The types of source data/ documents examined
    • Nature & frequency of any errors/problems
    • Corrective action taken
  • Methods of SDV

    • Direct Access

      • Preferred choice
      • Investigators excluded who do not allow direct access
    • INDIRECT ACCESS

      • Time consuming
      • Ensures subject confidentiality
  • Common problems encountered during SDV

    • Missing , scant data
    • High error rate

SOURCE DATA

  • information in original records & certified copies of original records
  • Information of clinical findings, observations; or other activities in a clinical trial
  • Necessary for the reconstruction & evaluation of the trial
  • SOURCE DATA EXAMPLES Demographic data: DOB, gender, age, race, height, weight

SOURCE DOCUMENT

  • ACCORDING TO ICH E6: 1.52 Source Documents ARE Original documents, data, and records
  • ICH E6: 8.3.13 Source documents are one of the “Essential Documents”

    • document existence of patients
    • substantiate integrity of trial data collected
  • IND Regulations 21 CFR 312

    • Investigator record keeping and record retention
    • Case Histories- must be adequate and accurate
  • IND Regulations 21 CFR 11 Electronic Records

  • IMPORTANCE OF good source documents

    • Decrease generation of data queries
    • Record all events that transpired while subject was in the study
    • Include statements describing informed consent process & copy of ICF given to subjects
    • Site specific, protocol specific
  • Source Document Template

    • Speed data collection process
    • Guide the investigator to collect all required study data
    • Should allow for additional comments for unplanned/ unexpected situations
    • Help diminish mistakes
  • SOURCE DOCUMENT EXAMPLES

    • Physician Progress Notes
    • Nurse’s Notes
    • Diagnostic Test Reports
    • Medication Records
    • Laboratory Reports
    • Participant Diaries
    • Informed Consent Forms
  • Source Document Agreement

    • Agreement between investigator & sponsor
    • Made at or prior to study initiation
    • Type of documents/documentation to be
    • done for each data item is defined
    • Place of filing of each document is defined
    • CRF, if source for any data, is also defined
  • Custody & Archiving of Source Documents

    • Should be retained until at least 2 years after the last approval of a marketing application
    • measures to prevent accidental or premature destruction

From:https://www.mindomo.com/zh/mindmap/source-data-validation-sdv-7de3cc980ff449d08809279e1ccec109

5.3k
0

第一种方法:利用sort过程步的dupout和nodupkey,原始的test数据集将输出为_unique和_dups两个数据集。*

proc sort

data = test

out=_unique

dupout = _dups

nodupkey ;

by var;

run ;

第二种方法:在数据步中利用by和first.var and Last.var。注意:此_dups和上面的_dups得到的结果并不相同,实际操作就知道差异在哪里。*

proc sort

data = test;

by var;

run;

data _dups ;

set test;

by var;

if not (first.var and Last.var) then output ;

run;

第三种方法:利用SQL过程步的having count(*),设置条件为>1即可输出重复值;如果要输出没有重复值的,当然是having count(*)=1;此方法得到的_dups与第二种方法相同。*

proc sql;

create table _dups as

select *

from test

group by var

having count(*)> 1

order by var;

quit;

3.6k
0
StatX 发表于 2014-05-26 10:45 am

plotmath中的表达式及其对应的公式如图:

http://imglf1.ph.126.net/apNs1u8uaUSn8oWR11RgYg==/2568177687526077055.png

http://imglf2.ph.126.net/6T6umtpp4ho-idcXKA_XRQ==/1291125718189839707.png

http://imglf1.ph.126.net/VdRRVoLt4Y1UzCiSkhTj6w==/2197475143280955823.png

http://imglf1.ph.126.net/KUzcfPbYBY5Scs9nDE0CaQ==/6598293523100142880.png

http://imglf1.ph.126.net/0rAjQ-H8ixcbDqz9tIwecw==/3238651082133284095.png

例:

formula = expression(italic(Prob) == bgroup("(",atop(n, x),")")%*%p[1]^x%*%(1-p[1])^(n-x))

把上述formula插入到plot中,显示如下:

http://imglf0.ph.126.net/Fen2wJRSylT6-Cw-qwiF4Q==/1816358024814196824.png

虽然比不上LaTex中公式那么漂亮,但在R中无需LaTex支持,这也是不错的选择。

该主题下的所有帖子将 已删除 ?
前一页
1 ... 4567
下一页
选择 取消选择主题
待审的草稿 ... 点击恢复编辑
放弃草稿