Transcript DHB Collections - Centre for Public Health Research

National Health Data Collections
– completeness, quality, timeliness, availability
Presentation to Massey University’s Centre for Public Health Research
Simon Ross
Information Group, National Health Board
8 May 2012
1
Overview
1. What are the National Collections
2. Where National Collections sit in the current MoH structure
3. Purpose and characteristics
4. Types of collections – high level overview
5. Completeness, quality, timeliness and availability
6. Who to contact for data requests and queries about the data
2
What are the National Collections
•
A national repository of health information collected and maintained by the
Ministry of Health
•
Split into ~ 14 individual collections
•
Held in the Ministry of Health’s data warehouse and accessible to some
users directly and to a much wider group by request
•
Often the initial rationale for a collection was for a payment, funding or
monitoring purpose, but the information collected serves many purposes
including research
•
Information can be linked to the same patient across collections
• Not included – Health Survey data
3
Structural change – from NZHIS to NCR
New Zealand Health Information
Service (NZHIS)
•
disestablished 2008
Public Health Intelligence (PHI)
National Collections and Reporting
(NCR)
•
Part of the Information Group in
the National Health Board (NHB)
Health and Disability Intelligence (HDI)
4
National Collections & Reporting (NCR)
Group Manager – Tracey Vandenberg
5 Teams:
1. Data Management, National Collections
2. Classification & Terminology
3. Analytical Services
4. Statistics & Reporting
5. Projects
5
The 6 uses of data principle
Collect once, use many times:
• Supporting self-management
• Supporting clinical intervention
• Clinical governance
• Administration (in all parts of health)
• Strategy and policy development
• Research
6
National Collections - characteristics
Person-centred – NHIs on all records
Multiple uses – (‘collect once, use many times’)
A mix of information available
•
Administrative
•
Demographic
•
Geographical
•
Clinical
•
Financial
7
National Collections – here they are:
DHB Collections
Primary Care Collections
•
National Minimum Dataset (NMDS)
•
Laboratory Claims Collection
•
National Booking Reporting System (NBRS)
•
Pharmaceutical Collection
•
National Non-Admitted Patient Collection
(NNPAC)
•
General Medical Subsidy Collection
•
Primary Health Organisation Enrolment
Collection
•
PRIMHD – mental health data
Registries
Other
•
New Zealand Cancer Registry (NZCR)
•
National Maternity Collection
•
National Immunisation Register
•
Medical Warning System
•
Mortality Collection
•
National Health Index
•
Health Practitioners Index
8
National Minimum Dataset (NMDS)
Hospital discharge event data from all DHBs (~1,000,000 events per annum)
Hospital events from many private hospitals (130,00 events per annum)
Clinical coding applied to all events (ICD-10-AM)
Coded diagnosis, procedure and external cause detail
Up to 99 codes able to be reported per event
Coded data augmented with free text in some cases
Year ends 30 June
9
Private Hospitals data
Discharge event data from >300 private hospitals/facilities
Reporting not mandatory (except publicly funded events)
•
data are incomplete
•
some large surgical hospitals don’t report
Quality of diagnosis information report often poor – procedures information is better
Data loaded into NMDS
Availability
•
Affected by completeness
•
published along with public hospital NMDS data
10
Mortality Collection – information sources
Data from 1988 (but statistics from earlier years are available)
BDM Death and Stillbirth registrations – core datasets
Causes of death information
• Medical certificates of cause of death
• Coroners reports
• Postmortem reports
• Hospital events in NMDS
• New Zealand Cancer Registry (NZCR)
• Land Transport NZ, Water Safety NZ
11
Mortality Collection – continued
Underlying cause of death – on all records
Specific contributing causes:
• Diseases including diabetes mellitus, alcoholism, HIV & others
• Injuries (from 1999 onwards)
• All causes for 0-24 years (from 2010)
Dynamic database
• Each year’s data is published once a determination is made that
most salient data has been received
• Updates are applied if subsequent relevant information is
received
• Coroner’s decisions are the primary reason for updates
12
New Zealand Cancer Registry (NZCR)
Data from 1948, Cancer Registry Act 1993 & Regulations 1994
All new cancers diagnosed in NZ
Information sources:
•
Pathology & haematology reports from Labs
•
Other National Collections (NMDS / Mortality Collection)
ICD-10-AM cancer ‘site’ codes, ICD-O morphology
Timeliness:
•
Specialist ‘sites’ – coded within 3 months of notification (respiratory, breast,
melanoma, prostate, cervix, colorectal, haematology/lymphatic, 0-24 yrs)
•
General release ~18 months after year of reference
13
Collection – who provides the data?
Local
• GPs, pharmacies, laboratories, NGOs, LMCs, private
hospitals
Regional
• DHBs, PHOs
National (government agencies)
• Department of Internal Affairs, Coronial Services
14
Examples
NMDS
•
DHBs, private hospitals
PRIMHD
•
DHB secondary mental health services, NGOs
Maternity
•
LMC claims, NMDS
•
mother-baby links from up to three sources (hospitals, claims, registrations)
Mortality
•
Registrations – Department of Internal Affairs
•
Cause of death – coroners, death certificates, post mortem reports, NMDS, NZCR, more
15
What do the collections contain?
• A patient identifier (NHI numbers)
• Demographics
• Geographic locators (meshblocks, domicile codes, TLA, DHB)
• Dates of service
• Clinical information (varying levels of clinically relevant data)
• Administrative data
• Financial data (varying levels and sources)
16
Contents discussion (examples)
Varying levels of clinical information
•
NMDS vs NNPAC
•
Pharms: medications but not conditions
•
Labs: tests but not test results
•
PRIMHD: services provided / team information but limited diagnosis and outcomes
information at this point
Varying levels and sources of financial information
•
NMDS vs NNPAC
•
Pharms vs Labs (estimates)
•
PHO (capitation), GMS (fee for service)
17
Completeness
Variable and collection specific
Completeness does affect our release policy for certain collections
For example:
• NMDS (public vs private)
• Pharms (community dispensed and subsidised vs hospital)
• Maternity (LMC claims data vs DHB provided services)
• NHI reporting to labs and pharms – improvements over time
18
Completeness – example
NHI reporting (pharms)
NHI reporting (labs)
Claim year
Year HCU %
Claim year
Year HCU %
2001
0.0%
2001
66.7%
2002
25.6%
2002
73.5%
2003
43.7%
2003
82.0%
2004
63.9%
2004
87.9%
2005
86.5%
2005
90.9%
2006
92.2%
2006
92.1%
2007
94.3%
2007
93.9%
2008
95.4%
2008
95.5%
2009
95.8%
2009
96.8%
19
Quality - general
Quality and completeness are closely related
Quality can vary based on many factors, for example:
• The source of the data
• The maturity of the collection
• The method and location of data collection, coding and entry
This is not a exhaustive list
20
A selection of quality-related concepts
•
Compliance
•
Business rules
•
Opportunities for re-submission
•
Master NHIs: merge, unmerge, overlays
•
Geocoding
•
Applying aggregate measures to individuals: NZDep
•
Challenges of using claims data – the impact of purpose of collection on the quality of
information submitted
•
The effect of incentives on patterns of coding and data submission
•
Examples: NMDS coding (public vs. private), maternity data quality
21
Timeliness
Submission times
•
DHB collections – monthly
•
Claims collections – ad hoc (but with limits)
•
Mortality – dependent on the data source
•
Cancer – dependent on the source of diagnosis and the data element
22
Availability
Controlled release collections
•
Mortality and cancer
Provisional data
Identifiable > encrypted > non-identifiable > aggregate
Who to contact?
•
[email protected]
•
Team Leader, Analytical Services, 04 816 2893
23