Transcript Session 2 - ICMART Home Page

How to Organize Data
Collection For Registers
on ART
What Data and Why?
Istanbul. March 21, 2009
David Adamson, MD
Director, Fertility Physicians of Northern California
Clinical Professor, Stanford University
Associate Clinical Professor, UCSF
International Committee Monitoring ART
What Do We Need to
Know?
1. Country/Region
2. ART Clinic
3. Patient demographics
4. Treatment
5. Outcome
Country/Region
Why does it matter?
1.
2.
3.
4.
5.
Race
Socioeconomic status
Cultural/religious uniqueness
Political agenda
Healthcare
1. Quality
2. Access
6. Submission to ICMART
ART Clinic
Why does it matter?
• Public vs. private
• Academic vs. non-academic
• Large vs. small
• Urban vs. rural
• Types of services provided
ART Patient Demographics
Why does it matter?
• Types of patients using ART
• Factors that affect pregnancy
rates
• Types of interventions in
different types of patients
• Different outcomes in different
types of patients
ART Treatment &
Outcomes
Why does it matter?
• Need to know which
interventions are successful
• Need to link intervention with
patient type and outcome to
determine benefit of
intervention
• Link with complications
What is Success?*
Percentage=
Numerator
Denominator
*ICMART Glossary does not include specific measures of
“success”, which should take into consideration the
wellbeing of babies as well as their mothers.
Possible Numerators
1. Number of oocytes
2. Number of 2 pronuclear zygotes
3. Number of healthy embryos
4. Number of embyros transferred
5. Number of embryos cyropreserved
6. Chemical pregnancy
7. Clinical pregnancy
8. Fetal heartbeat
9. Viable (ongoing) pregnancy
10. Third trimester pregnancy
11. Term pregnancy
12. Live birth(s)
13. Healthy child(ren)
14. Healthy child(ren) including frozen embryos
15. Healthy adult(s)
Possible Numerators
The numerator that is chosen usually
depends on the interest of the
person doing the choosing:
• Patient
• Infertility specialist
• Scientist
• Obstetrician
• Pediatrician
• Policy-maker
Most appropriate numerator:
Healthy Singleton
•Best for patient and family
•Best for child
•Best for obstetrician
•Best for pediatrician
•Best for policy-maker
•? Best for infertility physician
Problems With Healthy
Singleton as Measure of
Outcome
•Who collects the data?
•What is definition of healthy?
•Who evaluates baby?
•When is evaluation done?
•How long is the follow-up?
•What about healthy multiples?
Possible Denominators
• ART is not a single procedure
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
Decision to pursue ART
Ovarian stimulation (“Intention to treat”)
Egg retrieval
Fertilization
Embryo transfer
Implantation (chemical pregnancy)
Fetal development (clinical vs. heartbeat)
Viable pregnancy (>20 wks vs. > 26 wks)
Live-birth delivery
Patient (i.e. more than one cycle, + frozen)
Problems Using “Patient” as
Denominator
•
•
•
•
•
Women do different numbers of cycles
Some have frozen embryo cycles, some not
Variation in decision to pursue additional cycles
Different populations in each round of cycles
Cycles performed different
– Years
– Intervals
– Clinics
• Important laboratory data may be overlooked
USA Definition
Pregnancy Rate=
Live Births per number of
ovarian stimulation
procedures initiated
USA Dataset
Clinic
Clinic name, address, number
Name of laboratory used*
SART member ?
Services for single women ?
Donor egg available ?
Donor egg sharing ?
Total number of ART cycles
USA Dataset
Patient Information
Ethnicity, birth date, address
U.S. Resident
Prior pregnancy history
Sterilization
Months of infertility
Prior ART cycles
FSH/E2 levels
USA Dataset
ART Cycle Information
•Reason(s) for ART
•Cycle start date
•Suppression with GnRHa
•Stimulation drugs & dosage
•Intended treatment procedures
–Fresh/frozen
–Patient/donor eggs/surrogacy
–IVF, GIFT, ZIFT,TET
–Cycle for embryo banking, research
USA Dataset
ART Cycle Information
•Did cycle occur as intended?
–Cancelled, date, reason
–Complications
–Hospitalization
•Date of oocyte retrieval
•Number of oocytes
•Semen source, collection
method
USA Dataset
ART Cycle Information
•Use of ICSI, hatching
•Transfer attempt & date
•Number of fresh embryos
transferred, cryopreserved
•Number of thawed embryos
transferred, re-frozen
USA Dataset
ART Cycle Outcome
Information
•Outcome
–Not pregnant
–Type of pregnancy
•Ultrasound
–Date
–Number of fetal hearts
•Induced reduction
USA Dataset
•
•
•
•
•
ART Cycle Outcome
Information
Outcome of pregnancy
Source of information
Number of infants born
Birth weight
Neonatal morbidity/mortality
USA Dataset
ART Cycle Outcome Information
Birth Defects
• For each infant
–
•
•
•
•
Pregnancy termination
• Spontaneous
• Elective
• Induced reduction (multifetal reduction)
– Stillbirth
– Live birth
Categorization of defect
Ascertainment issues and bias
Birth certificates unreliable*
Later death and diagnosis ?
*Gore et al. J Reprod Med 2002;47:297-302
ART Outcomes
Birth Defects
Confidentiality issues (e.g., HIPAA)
Linkage to other registries (e.g. birth registry)
Definition inconsistency
Data poor quality
Methodology weaknesses
Ascertainment bias
Lack of controls
Statistics complicated
Wide confidence intervals
Statistical vs. clinical significance
Summary
What Data and Why?
•Many factors influence what data
should be collected
•Need system to collect good data
(GIGO!)
•STANDARDIZATION
–Definitions (ICMART/WHO glossary)
–Collection
–Analysis
The Fertility Clinic Success Rate
And Certification Act (FCSRCA) of
1992 (Wyden Law)
• Passed with support of SART and ASRM
• Required:
– Annual reporting clinic-specific success rates
– Listing of clinics that do not report
– Development of model program for certification of
embryo laboratories
– Promulgation of criteria and procedures for
approval of accreditation programs to inspect and
certify labs
• First report published under the law in 1997 for 1995
cycles
FCSRCA
Clinic-SART-CDC Relationship
• CDC contracts with SART to obtain annually a copy of
SART database
• Clinics submit data to SART
• 380 of 400 (95%) clinics are SART members
• All clinics that submit data to this CDC-supported SART
system in compliance with FCSRCA
• SART keeps database of clinics (openings and closings)
• Clinics must notify SART of personnel, address etc.
changes
• Requirements published in the Federal Register
2000;65:53310-53316.
FCSRCA
Data Collection
• SART distributes Clinical Outcomes Reporting System
and instructions (SART-CORS) annually to clinics
• Clinics abstract data from clinic records Jan 1-Dec 31
• Data entered using SART software
• ART cycle starts with
– ovarian stimulatory drugs, or
– ovarian monitoring with the intent of having embryos
transferred
• Data file organized with one record per cycle
• Multiple cycles from one patient not linked (HIPAA)
FCSRCA
Data Quality During Collection
• Medical Director responsible
– Submit data on time (one year following year being
reported)
– Verify by signature that data are accurate
– All patient identifiers removed before
submission to SART
• Prospective data submission
– Started with Year 2000 data
– Within 3 days of cycle start (i.e. before outcome
known)
– Demographic data
– Internet based
FCSRCA
Data Quality During Collection
• SART compiles data and submits to CDC
– CDC cannot identify individual patients
• SART & CDC review, resolve inconsistencies
• SART compiles and submits final, corrected
dataset to CDC
• Individual cycle data compiled and analyzed
– Report clinic-specific results
– Aggregated for national report
FCSRCA
Data Quality Validation
• Sample of 8-10% of clinics chosen for on-site
validation
• Randomly, and
• Pre-selected variables
– 50 randomly selected charts reviewed by 2-person
team
– CDC representative attends some visits as observer
– Error rates calculated
– SART & CDC review findings
FCSRCA
Data Quality Validation
• Data validation process
– Primarily educational
– Identify problem areas in data-collection process
– Correct general data collection problems
– Requires individual clinics to correct problems
– Funded by government as of 1999
• Validation results to date
– All error rates for all clinics acceptable
– Majority of errors minor
– Errors have minimal impact on success rates
FCSRCA
Data Analysis and Publication
• CDC primarily responsible
• Report co-authored by ASRM, SART, RESOLVE
• Results published hard copy and website
– National report
– Individual fertility clinic tables
– Appendices of associated information
• Detailed internal analysis and statistical
validation
• CDC editorial staff for final format and proofing
• Publication by CDC
Additional Recent Advances
in USA Registry
• Model program for certification of
laboratories
• Publish scientific articles based on
registry data (SART & CDC)
• Obtain highest possible Confidentiality
Status (308D): limits data access to
CDC & SART
• Bridge to SART CORS from clinic IT
systems
What Have We Learned?
• ART surveillance system is not static
• Challenges
– Data collection
• Large number of clinics, coordination difficult
• Deadlines necessary
• Individual clinic variability
• Standardization of definitions and practice
difficult
– Data presentation
• Focus groups
• Other feedback
• Simplification and explanation
The Future
• Surveillance system will continue to evolve
• Clarification of definitions and ambiguities (ICMART)
• Collect better data on outcomes and confounding
•
•
•
•
variables
Stop collecting non-useful variables
Link ART cycles in same patient=cumulative live births
Prospective reporting, validation, analysis, audit,
investigation, sanctions
Focus on
– Accuracy
– Appropriate context for use of data
• Fairness to clinics
• Clarity to patients
• Continued collaboration ASRM, SART, Clinics, RESOLVE,
CDC
THANK
YOU!