2. Ontario Team:
Dr. Murray Krahn (PI)
Dr. Claire de Oliveira
Karen Bremner
Dr. Kelvin Chan
Dr. Jeff Hoch
Funded by the Canadian Cancer Society
Research Institute
All inferences, opinions, and conclusions
drawn in this presentation are those of the
authors, and do not reflect the opinions or
policies of the Data Steward(s).
2
Acknowledgements:
BC Team:
Dr. Stuart Peacock
Dr. Paulos Teckle
Dr. Winson Cheung
3. • Objective
– To estimate direct medical costs and predictors of these
costs for 21 common types of cancer and 4 phases of care
(pre-diagnosis, initial, continuing care, and terminal), using
linked administrative data in BC and Ontario
• Cases from BC Cancer Registry and Ontario Cancer
Registry diagnosed in 1997-2007
• Follow up to the end of 2009
BC-ON Costing Study
3
4. Phase-based costing
• Course of disease divided into distinct phases
• Observation time after diagnosis is allocated to phases
hierarchically:
Terminal Phase (12 mo) > Initial Phase (6 mo) > Continuing Care Phase
4
time (months)
-3 3 6 9 12 15 18 21 24 … 60 63 66 69 72 75 …
A 1
B 2
C 3 … 4
D … …
6. 6
British Columbia Ontario
Hospitalization Discharge Abstract Database
(Hospital Separations)
Discharge Abstract Database
Physician
services
Medical Services Plan (MSP)
Payment Information File; BCCA
Schedule data
Ontario Health Insurance Plan
(OHIP) claims
Outpatient
prescription
drugs
PharmaNet Ontario Drug Benefit (ODB)
Program
Chemotherapy BCCA Provincial Systemic Therapy
Program
New Drug Funding Program, ODB,
National Ambulatory Care
Reporting System (NACRS)
Radiotherapy BCCA Radiotherapy database Cancer Care Ontario Activity Level
Reporting System
Home and
Community Care
Home and Community Care
(Continuing Care) Data
ON Home Care Database,
Continuing Care Reporting System,
ODB database
Diagnostics MSP Payment Information File,
BCCA Schedule and Document
data
OHIP claims
7. • Registry data
– Diagnosis date definition: clinical vs. pathological
• Oncologist services:
– ON: claims or shadow billing in OHIP, except for RadOncs
– BC: no fee-for-service; used BCCA appointment data
• Diagnostics
– ON: claims in OHIP; professional and technical components
– BC: some claims in MSP; limited data for lab/imaging services from budgets
• Radiotherapy
– ON: NHPIP workload measurement
– BC: estimated from BCCA radiotherapy data and unit cost per fraction
Key Differences
8. • Chemotherapy drug cost:
• Chemotherapy delivery cost:
– ON: visits identified in OHIP and NACRS; costs from NACRS
– BC: delivery at BCCA sites is in data, others missing
• Hospitalization cost:
– Same method, but total cost is very sensitive to assumptions about cost per
weighted case
Key Differences
Ontario BC
I.V. drugs
New, $ NDFP BCCA
Old, $ NACRS (RIW) BCCA
Oral drugs
< 65 yrs No data BCCA
≥ 65 yrs* ODB BCCA
17. • Same sites consistently high
– myeloma, pancreas, brain
• …or low across phases
– melanoma, cervix, prostate
• High initial-phase costs for aggressive cancers but
not necessarily terminal phase
– Cost of dying is consistently high
• Phases not appropriate for all cancers
Patterns by phase and site
17
18. Patterns in BC and Ontario
• Cost estimates tend to be lower for BC
• Unavoidable differences in costing methods
– Some can be accommodated by comparing
component costs
• Separating true differences across provinces
from differences in methods is challenging
18
19. • Population-based admin data well-suited to costing
studies
– Analysis of cost trajectories, patterns
– Inputs for future economic evaluation
• Aligning service definitions and component costs is a
foundational step
• Familiarity with the data is essential!
– ICES, Population Data BC
Conclusions
20. Advancing Health Economics, Services, Policy and Ethics
Thank you
• Contact: rpataky@bccrc.ca
• Ontario Team: Dr. Murray Krahn (PI), Dr. Claire de Oliveira, Dr.
Jeff Hoch, Karen Bremner, Dr. Kelvin Chan
• BC Team: Dr. Stuart Peacock, Dr. Paulos Teckle, Dr. Winson
Cheung
• Funding provided by the Canadian Cancer Society Research
Institute (grant #020200)
20
Editor's Notes
Outline:
BC Cancer Agency data
BC-ON costing study
Data sources and key differences
Preliminary phase-based cost results
comparative analysis using admin data from BC and ON to highlight the issues/challenges, strengths and limitations around the standardization of the existing costing methods. I think this would be really important to focus on as it will be the main challenge we will face when we try to replicate our work for other provinces.
Include Popdata/MoH disclaimer
By “cost” we mean direct medical costs, not direct non-medical (e.g. travel cost) or indirect (e.g. lost productivity)
Thankfully, even though costs change over time, they follow a predicable pattern
Phases correspond to major changes in cost function, but are also clinically relevant
4 phases:
1 – Pre-diagnosis phase (green) – 3 months prior to diagnosis
2 – Initial phase (navy) – up to 6 month following diagnosis
3 – Continuing phase (light blue) – length varies
4 – Terminal care phase (purple) – up to last 12 months of life
We calculated average cost per phase by cancer site, with non-parametric bootstrapping for 95% confidence intervals.
All differences are statistically significant. BC is slightly older, slightly higher proportion of males.
Registry audience probably knows more about these differences than I do...
In BC:
BC Cancer Agency, BC Ministry of Health, PharmaNet, and BC Vital Statistics Agency approved data access and use facilitated by Population Data BC
In Ontario:
Data from Cancer Care Ontario and the Ontario Ministry of Health and Long-Term Care
Linkage and analysis conducted by Institute for Clinical Evaluative Sciences
… So using these data sources and our own costing methods, we went away and calculated our own costs in both provinces and came back with estimates that were dramatically different.
Pre-diagnosis phase – cost expressed for 90 days before diagnosis
Top three for each province and sex marked in yellow, bottom three in blue
~6 fold difference between top ($4660 for myeloma in Ontario, vs. $733 for testicular cancer in BC)
Values for BC tend to be lower; but ranks are similar
The biggest difference here is likely to do differences in diagnostics – there are differences in two main areas: physician services and diagnostics. They act in opposite directions, but the net effect is higher costs in Ontario. We need to unpack this more:
Initial care phase – cost expressed for 6 months after diagnosis (~180 days)
Top three for each province and sex marked in yellow, bottom three in blue
12-fold difference between top ($43,400 for esophageal cancer in Ontario, vs. $3400 for melanoma in BC)
Values for BC tend to be lower; but ranks are similar
Continuing care phase – cost expressed per year (~360 days)
Top three for each province and sex marked in yellow, bottom three in blue
9 fold difference between top ($19,300 for myeloma in Ontario, vs. $2000 for testicular cancer in BC)
Values for Ontario tend to be higher; but ranks are similar
-This is also where sex differences appear, although not shown. Costs for women tend to be higher than for men
This is phase where (I think) control matching with have the greatest effect – we’re dealing with an older population with relatively high service use regardless of cancer
Big mix of services – Myeloma highest by far
Driven by maintenance therapy with lenalidomide
Highest cost drugs in BC data:
Myeloma: pamidronate, thalidomide, lenalidomide, bortezomib
Leukemia: imatinib, rituximab, fludarabine, dasatinib
Terminal phase – cost expressed for last year of life
Top three for each province and sex marked in yellow, bottom three in blue
Less than a 3-fold difference between top ($78,200 for brain cancer in Ontario women, vs. $34,100 for melanoma in men in BC)
A note re: testicular cancer: there are few deaths in both cohorts; 95% CI’s (not shown) are wide
Values for BC tend to be lower; but ranks are similar
Mostly inpatient – more home and community care
These are also not selected for cause of death – mix of cancer and non-cancer
Part of the challenge in interpreting “phases” for all cancers is that sometimes they’re not appropriate. Eg. For aggressive cancers, few patients survive to contribute time to “continuing care” phase (e.g brain). These are probably atypical cases, so it’s hard to draw any conclusions. For cancers with low mortality, most patients we capture in the terminal phases are dying of other causes (e.g melanoma)
Caution that we haven’t done bootstrapping to compare 95% CI’s – results for BC show that intervals generally narrow for common cancer sites (as we’d expect) but for less common cancers (e.g. testicular, terminal phase) 95% CI’s are wide
Not just differences in methods – there are structural differences in the organization and delivery of care that influence the data and cost available.