WALTER WODCHIS - Canadian Association for Health Services and

Healthcare Costs for 17 Chronic Condi3ons in Ontario Walter P Wodchis, PhD CAHSPR May 26, 2015 Overview • 
• 
• 
• 
• 
• 
Background Purpose Methods Results Strengths and Limita3ons Implica3ons & Next Steps 2 Research Team YuQing Bai, Susan Bronskill, Andrea Gruneir, Anna Kone, Natasha Lane, Colleen Maxwell, Yelena Petrosyan, Kednapa Thavorn, Walter Wodchis 3 Background •  The high and growing cost burden of chronic condi3ons is a highly cited feature of most western countries. •  There is an increasing apprecia3on of the magnitude of costs associated with caring for individuals with common chronic condi3ons. •  However, es3mates to date were derived for specific condi3ons with varied methods precluding direct comparison of costs. 4 Background •  …or, in the nearest approxima3on to the ideal, es3mates of burden were based on Ambulatory Care Group (ACG®) weights •  See: Broemeling, Watson, Black. Chronic condi3ons and co-­‐
morbidity among residents of Bri3sh Columbia. Centre for Health Services and Policy Research. 2005 5 Background High Impact/Prevalence:
Hypertension
Depression
90
hypertension, depression, diabetes,
ischemic heart disease, asthma,
cancers, cardiac arrhythmia, COPD,
congestive heart failure, stroke
Diabetes
35
Ischemic heart disease
30
Degenerative
joint disease
25
Cancers
Prevalence
] $FOUSFGPS)FBMUI4FSWJDFTBOE1PMJDZ3FTFBSDI
'JHVSF)JHIJNQBDUBOEPSIJHIQSFWBMFODFDISPOJDDPOEJUJPOT
Asthma
Cardiac Arrhythmia
20
Thyroid Dis
Dis Lymphoid Metab.
Allergic Rhinitis
15
Congestive
heart failure
COPD
Autoimmune /
conn. tissue
Cerebrovascular disease
Deafness / Hearing loss
10
Prostatic
hypertrophy
Malignant
neoplasm
of skin
Diverticular
dis. of colon
Schizophrenia/
affective psychosis
Obesity
Chronic Cystic
Dis Breast
5
Psoriasis
Glaucoma
Osteoperosis
IBD
Parkinson’s
MS
ADD
0
Congenital
anom: limbs
2
Peripheral vascular
disease
Cardiac valve dis
Chronic liver dis
Generalized
atherosclerosis
Personality
disorders
Endometriosis
0
Dementia /
delerium
Seizure
Chronic skin ulcer
HAnem
Devt
dis
CHD
CF
Chromosomal
HIV /AIDS
Chronic renal
failure
Cardiomyopathy
Blindness
Aplastic anemia
MD
4
Kyphoscoliosis
6
8
10
Impact
6 Purpose •  The objec3ve of the present study was to es3mate the annual incremental costs associated with 17 common chronic condi3ons using a common methodology to measure the compara3ve costs associated with each condi3on. 7 Methods: Data Sources Ontario has rich health administra3ve data held at the Ins3tute for Clinical Evalua3ve Sciences (ICES) that are linkable at the individual level using coded iden3fiers: –  Hospital Data from Canadian Ins3tute for Health Informa3on (CIHI) Discharge Abstract Database, Na3onal Rehabilita3on Repor3ng System, Con3nuing Care Repor3ng System, Mental Health Repor3ng System, Ambulatory Care Repor3ng System –  Physician, pharmaceu3cal and home care billing databases maintained in Ontario –  Also hospital cost databases from the Management Informa3on System (MIS) 8 Methods: Popula3on •  Cross sec3onal 20% sample of Ontario residents aged 0-­‐105 as of the index date of the study (April 1, 2012) diagnosed with one of the following 17 condi3ons: Cardiac arrhythmia Osteoporosis Acute myocardial infarc3on Rheumatoid arthri3s Hypertension Osteo-­‐ and other arthri3s Chronic coronary syndrome Mood disorders Conges3ve heart failure Demen3a Stroke Other mental health Asthma Cancer Chronic obstruc3ve pulmonary disorder Diabetes Renal failure 9 Methods: Popula3on •  Validated ICES derived chronic disease databases were used to iden3fy prevalent cases of: –  Hypertension, –  Conges3ve Heart Failure, –  Asthma, –  Chronic Obstruc3ve Pulmonary Disorder, –  Diabetes, –  Acute Myocardial Infarc3on (April 2011 to March 2012) 10 Methods: Popula3on All other diagnoses: •  One acute care code in an acute care episode (Canadian Ins3tute for Health Informa3on Discharge Abstract Database, CIHI-­‐DAD), •  OR 2 relevant ICD9 codes in OHIP physician billing records … within 2 years prior to the index date (April 1, 2012) This is generally the approach of the ICES derived databases. For details, see: Koné Pefoyo AJ, Bronskill SE, Gruneir A, Calzavara A, Thavorn K, Petrosyan Y, Maxwell CJ, Bai Y, Wodchis WP. The increasing burden and complexity of mul3morbidity. BMC Public Health. 2015 Apr 23;15(1):415., 11 Methods: Cost Es3ma3on Individual pa3ent-­‐specific total direct cost es3ma3on based on methods developed for use with administra3ve data: •  Mostly bojom-­‐up with some top-­‐down algorithms to determine encounter-­‐specific costs. •  Generally = Sum over one year of all Use x Cost/Use (all encounters, not those specific to one condi3on) •  U3liza3on includes all hospital, home care, long term care, physician, pharmacy for 65+ and all receiving provincial government income support payments 12 Methods: Cost Es3ma3on cont’d •  Encounter-­‐specific costs equal to amounts paid for physician, drugs, home care (plus allocated case-­‐management) •  Hospital costs allocated to cost per weighted case (using CIHI Resource Intensity Weights) from hospital financial submissions using Management Informa3on System (MIS) and related sources from Ontario Ministry of Health and Long Term Care (MOHLTC) Health Data Branch, •  Other costs from MOHLTC (e.g. OHIP Architected Payment Database) For details, see: Wodchis WP, Bushmeneva K, Nikitovic M, McKillop I. Guidelines on Person-­‐Level Cos3ng Using Administra3ve Databases in Ontario. Working Paper Series. Vol 1. Toronto: Health System Performance Research Network; 2013. 13 Methods: Cost Es3ma3on cont’d For each condi3on: •  Find person in Ontario popula3on without condi3on but matched on : Age +/-­‐ 3 months, Sex, LTC residence •  Then propensity match on age, sex, LTC, rurality, 12 Collapsed ADG scores, neighborhood income quin3le, 14 provincial regions (caliper of 0.2 s.d. of propensity score) •  Match within 20% sample of residents except for Arthri3s, Hypertension and Asthma where we use en3re popula3on to find matches (large number of unmatched in 20% sample). •  Incremental cost = difference cost w/ vs w/o condi3on •  Provincial total cost = popula3on x incremental cost 14 Results: Study Popula3on Cardiac arrhythmia Acute myocardial infarc3on Hypertension Chronic coronary syndrome Conges3ve heart failure Stroke Asthma Chronic obstruc3ve pulmonary disorder Osteoporosis Rheumatoid arthri3s Osteo-­‐ and other arthri3s Mood disorders Demen3a Other mental health Cancer Diabetes Renal failure ON Popula*on Study 20% Sample 375,946 75,223 11,994 2,430 2,755,432 551,899 684,762 137,071 228,084 45,691 145,156 28,758 1,902,121 380,953 272,186 54,732 324,166 64,436 163,151 32,934 3,186,543 637,718 1,388,188 278,103 175,082 35,300 1,775,706 355,454 981,778 196,146 1,199,808 240,005 194,674 38,907 15 Cost (Incremental and Per Year Alive) ($)
(2,000)
12,000
Physician
10,000
Hospital
Osteoporosis
4,000
Osteoarthritis
6,000
Asthma
8,000
Rheumatoid arthritis
Cancer
Depression
Hypertension
Coronary syndrome
Arrhythmia
Other mental health
Diabetes
Stroke
COPD
AMI
Dementia
CHF
Renal
Results: Incremental Cost per year Drug/Device
Long Term Care
Other
2,000
0
16 (0.5)
3.5
Physician
3.0
Hospital
2.5
Drug/Device
Osteoporosis
Osteoarthritis
1.5
Asthma
2.0
Rheumatoid arthritis
Cancer
Depression
Hypertension
Coronary syndrome
Arrhythmia
Other mental health
Diabetes
Stroke
COPD
AMI
Dementia
CHF
Renal
Total System Cost (Billion $)
Results: Total Direct Cost per year Long Term Care
Other
1.0
0.5
0.0
17 Results: Incremental vs Total Costs 3.5 Total System Cost (Billion $) 3.0 2.5 Hypertension Diabetes Other mental health 2.0 1.5 Osteoarthri3s Depression Renal CHF COPD Demen3a 1.0 Asthma Cancer Coronary syndrome Stroke Arrhythmia 0.5 Rheumatoid AMI arthri3s (1000) Osteoporosis 1000 2000 3000 4000 5000 6000 7000 (0.5) Total Incremental Cost ($) 8000 9000 18 Strengths and Limita3ons •  Study popula3on and costs representa3ve of Ontario popula3on (> 95% sample matched). •  Robust incremental cost es3mates. •  Condi3ons are those present at start of year and do not include new incident cases so cannot say “annual cost for all diabetes care”. •  Some condi3ons are episodic and episode-­‐
based costs may be more appropriate (AMI, cancer) 19 Implica3ons and Next Steps •  We want to understand how quality relates to costs. Do people who get bejer care have lower costs? Over what 3me horizon? (some early evidence for diabetes) •  We have examined mul3morbidity as incremental number of condi3ons but some clusters might have par3cularly important cost implica3ons. 20