培訓(xùn)建庫(kù)使用文件2edc23cdash介紹

CDISC標(biāo)準(zhǔn)在醫(yī)學(xué)研究中的應(yīng)用莊永龍博士北京百奧知信息科技有限公司（微信）信息技術(shù)提升合規(guī)、效率和質(zhì)量——臨床研究信息化與知識(shí)分享大講堂臨床數(shù)據(jù)采集的目的數(shù)據(jù)標(biāo)準(zhǔn)化的目的Objective藥物臨床研究Clinical

trial提交FDA評(píng)審FDA

data

submission

與國(guó)際同行交換數(shù)據(jù)

Data

exchangeeg:

STS

EACTS臨床數(shù)據(jù)的應(yīng)用流程EDC電子病歷EMR源數(shù)據(jù)Source

DataRemote

backup統(tǒng)計(jì)師Statistics心電圖ECG實(shí)驗(yàn)室Labs研究者InvestigatorDatabaseEDC

和臨床數(shù)據(jù)的現(xiàn)狀EDC是目前最流行的數(shù)據(jù)采集方法Electronic

Data

Capture

(EDC)

is

the

commonlyused

method

to

collect

clinical

data有很多EDC系統(tǒng)，每個(gè)的數(shù)據(jù)結(jié)構(gòu)都不一樣

There

are

so

many

EDC

vendors

now,each

ofthem

has

different

database

structure,

setup,different

variable

naming

convention

and

labelingand

data

attributes有的界面和CRF一致some

have

the

data

entry

interface

similar

tothe

CRFs有的不一致，完全是列表式Some

have

a

listing-form

of

data

entry

screenEDCRemote

backupDatabaseEDCEDC

和臨床數(shù)據(jù)的現(xiàn)狀提供給統(tǒng)計(jì)師的數(shù)據(jù)各不一樣Data

delivered

to

statistician

are

very

different!數(shù)據(jù)呈遞給FDA也各不一樣Data

submitted

to

FDA

are

very

different!由于這種非標(biāo)準(zhǔn)化情況，F(xiàn)DA很難橫向的比較不同廠家，同類(lèi)藥的療效和安全性Because

of

non-standard

and

non-exchangeable,there

was

no

way

for

FDA

to

evaluate

drugapplications

cross

different

sponsorsEDCRemote

backupDatabaseEDC樣本數(shù)據(jù)示例Grace研究項(xiàng)目問(wèn)題大部分臨床醫(yī)學(xué)研究Most

clinical

trials…沒(méi)有采用數(shù)據(jù)標(biāo)準(zhǔn)化

don’t

employ

a

standard

for

data

exchange沒(méi)有使用標(biāo)準(zhǔn)化的工具和技術(shù)don’t

use

standardized

analytic

tools

ortechniques后果Result其結(jié)果是造成了臨床數(shù)據(jù)分析中效率和系統(tǒng)化的困難，特別是對(duì)多個(gè)臨床試驗(yàn)Analyzing

clinical

trial

data

efficiently

and

systematically

is

difficult

andtime

consuming,

especially

across

many

trials不如這樣一個(gè)簡(jiǎn)單的問(wèn)題：有多少女性參與了GRACE的臨床試驗(yàn)？e.g.How

many

women

participate

in

clinical

trials

with

GRACE?如何計(jì)算？面對(duì)這樣的一個(gè)簡(jiǎn)單的問(wèn)題，我們往往會(huì)感到無(wú)能為力Facing

on

such

a

simple

question,

we

often

feel

lack

of

power沒(méi)有標(biāo)準(zhǔn)的數(shù)據(jù)集名No

standard

file

names怎么可能自動(dòng)地用計(jì)算機(jī)找出包含要找數(shù)據(jù)的數(shù)據(jù)集How

does

a

computer

find

the

file

that

contains

the

data

?沒(méi)有標(biāo)準(zhǔn)的變量名No

standard

variable

names怎么可能自動(dòng)地用計(jì)算機(jī)找出包含性別的數(shù)據(jù)名How

does

a

computer

find

the

column

that

contains

sex

information沒(méi)有標(biāo)準(zhǔn)的數(shù)據(jù)編碼No

standard

terminology怎么可能自動(dòng)地用計(jì)算機(jī)告知1還是2代表女性Howdoes

a

computer

know

which

code

represents

which

sex?臨床數(shù)據(jù)標(biāo)準(zhǔn)化非常關(guān)鍵和重要數(shù)據(jù)的可交換性定義Definition數(shù)據(jù)的可交換性是它具有的屬性，它使得數(shù)據(jù)可以以各種方式傳送和使用而不會(huì)失去它的內(nèi)容和意義Data

exchangeability

is

the

attributes

of

data

that

enable

the

data

to

becommunicated

and

used

in

many

ways

without

loss

or

alter

the

informationcontent

or

meaning

assigned

to

the

data.只有標(biāo)準(zhǔn)化的數(shù)據(jù)才具有可交換性O(shè)nly

standardized

data

are

exchangeableCDISC是什么?CDISC:臨床數(shù)據(jù)交換標(biāo)準(zhǔn)協(xié)會(huì)Clinical

Data

Interexchange

Standards

Consortium,

started

with

smallgroup

of

volunteers

in

1997是一個(gè)全球的、開(kāi)放的、多學(xué)科的、非營(yíng)利的組織，創(chuàng)建了支持臨床研究數(shù)據(jù)和元數(shù)據(jù)采集、交換、提交和存檔的標(biāo)準(zhǔn)It

is

a

global,

open,

multidisciplinary,

non-profit

organization

that

hasestablished

standards

to

support

the

acquisition,

exchange,

submissionand

archive

of

clinical

research

data

and

metadata.主要的模型There

are

two

major

models

in

CDISC:SDTM:針對(duì)臨床數(shù)據(jù)Standard

Data

Tabulation

Model

for

clinical

trial

dataADaM:針對(duì)分析數(shù)據(jù)Analysis

Data

Model其他正在建立的標(biāo)準(zhǔn)Other

still-developing

standard

CDASH:針對(duì)CRF表Clinical

Data

Acquisition

StandardsHarmonization

for

CRF

standardsSDTM的基本模型Observation觀察Generic

structure（衍伸結(jié)構(gòu)）Unique

identifiersTopic

variable

or

parameterTiming

VariablesQualifiers.Interventions干預(yù)Findings發(fā)現(xiàn)Events事件General

classes

（一般分類(lèi)）SubjectCMSDTM

Domains（SDTM域）(dataset

structures)EXEGIELBPEAEDSThe

patient/subject

focused

information

model

ofthe

clinical

‘reality’

(general

classesofobservations

on

subjects:

interventions,

findings,

events).

This

model

has

beendevelopedby

CDISC/SDS

team.Interventions干預(yù)Events事件ConMedsExposureAEMedHistDispositionFindings結(jié)果ECGPhysExamLabsVitalsDemogOther其他Subj

Char*SubstanceUseIncl

Excl*RELATESSUPPQUALStudy

SumStudy

DesignQS*,

MB*CommentsCP*,

DV*CDISC

SDTM

域CDISC

SDTM域注解分類(lèi)解釋EX試驗(yàn)用藥記錄CM伴隨用藥記錄和其它物理治療SU物質(zhì)使用（煙、酒、咖啡等）事件EventsAE不良反應(yīng)事件DS試驗(yàn)完成、中止、脫落報(bào)告MH病史（試驗(yàn)藥物相關(guān)疾病病史及其它病史）干預(yù)Interventions

結(jié)果Findings分類(lèi)解釋EG心電圖IE入選、排除標(biāo)準(zhǔn)LB實(shí)驗(yàn)室檢查QS調(diào)查問(wèn)卷（HAMILTONDEPRESSION

SCLAE,SF36,ADAS.）PE體格檢查SC個(gè)體特征（比如患者姓名、眼睛顏色、生育狀態(tài)、過(guò)敏狀態(tài)）VS生命體征注：HAMILTONDEPRESSIONSCLAE指漢密頓抑郁量表SF36

簡(jiǎn)明健康狀況問(wèn)卷ADAS

阿爾茨海默病評(píng)分其他OtherDM受試者人口學(xué)資料CO非異常描述、其它注釋申報(bào)數(shù)據(jù)進(jìn)入FDA的流程Janus

System審辦方SponsorSDTMADaM統(tǒng)計(jì)評(píng)審員FDA

Statistical

Reviewer醫(yī)學(xué)評(píng)審員FDA

Medical

Reviewer電子申報(bào)接口

ElectronicSubmissionGateway

(ESG)轉(zhuǎn)換為SDTM和ADaM存在的問(wèn)題目前的SDTM轉(zhuǎn)換是單向的，沒(méi)法溯源It

is

an

one-way

mapping

(i.e.

RawSDTM)

without

traceability.

E.g.

age

wascalculated

variable. From

the

mapped

SDTM,

it

didn’t

tell

how

age

wascalculated太多的數(shù)據(jù)庫(kù)外的手工程序，非常容易出錯(cuò)!Too

much

outside

database

programming，easy

to

make

mistakes

in

theconversion

processes增加了大量的質(zhì)控和驗(yàn)證工作Increased

a

lot

of

validation

and

QC

work不能保證申報(bào)數(shù)據(jù)的質(zhì)量和溯源性Lack

of

insurance

of

quality

and

traceability

for

submitted

data

package沒(méi)有有用的軟件或工具No

powerful

tools

for

automatic

SDTM

mapping解決方案對(duì)一個(gè)新study，數(shù)據(jù)標(biāo)準(zhǔn)化應(yīng)從數(shù)據(jù)庫(kù)的設(shè)計(jì)開(kāi)始

For

new

study,

data

standardization

can

be

started

with

thedatabase

design利用CDASH的標(biāo)準(zhǔn)表建庫(kù)

Use

CDASH

standard

forms用SDTM給變量命名和標(biāo)識(shí)，但不是所有的變量都能這樣做Use

SDTM

to

name

and

label

variables

(not

all

variables

can

bedirectly

named

in

SDTM,

e.g.

date

field用SDTM域來(lái)分類(lèi)

Group

data

using

SDTM

domain即使這樣，一個(gè)完全的SDTM標(biāo)準(zhǔn)庫(kù)也會(huì)面臨挑戰(zhàn)Even

you

setup

database

in

CDASH,

it

is

still

not

possible

to

havea100%

SDTM

database.

However,

it

is

a

SDTM-like

database對(duì)以前的歷史遺留研究，用SAS轉(zhuǎn)換仍然是唯一的辦法For

legacy

studies,

mapping

through

SAS

is

still

the

only

wayCDISC

總結(jié)CDISC是用于數(shù)據(jù)申報(bào)