cost-effectiveness of detection and treatment of anxiety and depression in frequent attenders of the...

14

Upload: dulcie-barber

Post on 29-Dec-2015

213 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Cost-effectiveness of detection and treatment of anxiety and depression in frequent attenders of the GP Judith Bosmans, Frans Smits, Veerle Coupé, Jacob
Page 2: Cost-effectiveness of detection and treatment of anxiety and depression in frequent attenders of the GP Judith Bosmans, Frans Smits, Veerle Coupé, Jacob

Cost-effectiveness of detection and treatment of anxiety and depression in

frequent attenders of the GP

Judith Bosmans, Frans Smits, Veerle Coupé,Jacob Mohrs, Aart Schene, Henk van Weert,

Gerben ter Riet

[email protected]

Page 3: Cost-effectiveness of detection and treatment of anxiety and depression in frequent attenders of the GP Judith Bosmans, Frans Smits, Veerle Coupé, Jacob

Frequent Attenders• 90-100th centile of attendance frequency in 4 age categories

for men and women separately

• Somatic, psychological & social problems

• Referral and costs

• Heartsink for (some) GPs

• Why embark on RCTs?

Page 4: Cost-effectiveness of detection and treatment of anxiety and depression in frequent attenders of the GP Judith Bosmans, Frans Smits, Veerle Coupé, Jacob

Questions• What would happen if

– We detected and treated all depression and anxiety in:• one-year frequent attenders (1yFA)• two-year frequent attenders (2yFA)

• In terms of– years spent as non-FA without depression or anxiety– quality-adjusted life years

• Compared to– Usual care in Amsterdam SE region, The Netherlands 2007 – 2011– Δ cost strategy x vs usual care / Δ effects strategy x vs usual care

Page 5: Cost-effectiveness of detection and treatment of anxiety and depression in frequent attenders of the GP Judith Bosmans, Frans Smits, Veerle Coupé, Jacob

QALY: Quality Adjusted Life Years

Usual care

Strategy x

gained by the intervention

Quality improvement

Quantity improvement

Page 6: Cost-effectiveness of detection and treatment of anxiety and depression in frequent attenders of the GP Judith Bosmans, Frans Smits, Veerle Coupé, Jacob

α

α = cost-effectiveness ratio = Δ cost / Δ effect

.

Page 7: Cost-effectiveness of detection and treatment of anxiety and depression in frequent attenders of the GP Judith Bosmans, Frans Smits, Veerle Coupé, Jacob

Methods: components of CEAComponent Source or number Comment

States (non)FA; depression, anxiety; success of treatment

See next slide

Transitions Simplifications necessary

Transition probabilities Cohort of 1yFAs from HAG-net-AMC Markov model

Utilities of states Literature; FA anxiety 0.55; nonFA 0.85 0=death; 1=full health

Cycle length 1 year Assess states each yr

Time horizon 5 years

Strategies (treatments) RR={0.90, 0.80, 0.60, 0.00} ± spill-over effects

Costs states Achmea dbase; € 2400 – 12400 per year real

Cost psychotherapy € 500 (NL costs 6 x 80) real

Outcomes Years spent as non-FA without depression or anxiety; QALYs

Uncertainty Monte Carlo simulation; 10,000 x Sample from distributions

Perspective Health care Societal better?

Page 8: Cost-effectiveness of detection and treatment of anxiety and depression in frequent attenders of the GP Judith Bosmans, Frans Smits, Veerle Coupé, Jacob

1yFA + depression

2yFA + depression

2yFA + anxiety

2yFA + unknown or no morbidity

2yFA + Tx success

pFA + depression

pFA + anxiety

pFA + unknown or no morbidity

pFA + Tx success

nonFA + depression

nonFA + anxiety

nonFA + unknown or no morbidity

1yFA + depression

1yFA + depression

0.02

0.31

0.10

0.06

0.39

0.12

N=49

Example (1 out of 5 cycles): 1yFA with depression and transition probabilities under usual care

Page 9: Cost-effectiveness of detection and treatment of anxiety and depression in frequent attenders of the GP Judith Bosmans, Frans Smits, Veerle Coupé, Jacob

Results: Usual Care

• 70 % percent of 1yFA and 2yFA with depression or anxiety were free of those after one year.

• In pFAs 50% stayed depressed and 60% stayed anxious after one year.

• 1000 1yFAs spent 86% of time in a nonFA state without depression or anxiety (4322/5000 person-years)

Page 10: Cost-effectiveness of detection and treatment of anxiety and depression in frequent attenders of the GP Judith Bosmans, Frans Smits, Veerle Coupé, Jacob

Results: 1000 1yFAsTreatment effect

(RR) Δ with usual care

- 330.710/44.1 = - 7.504

Δ with usual care

Page 11: Cost-effectiveness of detection and treatment of anxiety and depression in frequent attenders of the GP Judith Bosmans, Frans Smits, Veerle Coupé, Jacob

Interpretation

• Detection and treatment of depression cost- effective at RR <= 0.6, unless spill-over

• Detection and treatment of anxiety not cost-effective, even with spill-over

• Detection and treatment of depression and anxiety cost effective at RR <= 0.6

Page 12: Cost-effectiveness of detection and treatment of anxiety and depression in frequent attenders of the GP Judith Bosmans, Frans Smits, Veerle Coupé, Jacob

Cost-effectiveness of detection and treatment of anxiety and depression in

frequent attenders of the GP

Judith Bosmans, Frans Smits, Veerle Coupé,Jacob Mohrs, Aart Schene, Henk van Weert,

Gerben ter Riet

[email protected]

Page 13: Cost-effectiveness of detection and treatment of anxiety and depression in frequent attenders of the GP Judith Bosmans, Frans Smits, Veerle Coupé, Jacob

Results: 1000 2yFAsTreatment effect

(RR) Δ with usual care

Δ with usual care

Page 14: Cost-effectiveness of detection and treatment of anxiety and depression in frequent attenders of the GP Judith Bosmans, Frans Smits, Veerle Coupé, Jacob

Strengths & limitationsStrengths Limitations

Realistic data for transition probabilities 1 year 5 years

Realistic cost data (Achmea) Limited data to populate the model (N=692, but “zeroes”)

Many scenarios Simplifications: number of states

Effect of spill-over analysed Simplifications: possible transitions not considered

PHQ as the yardstick PHQ as the yardstick

Monte Carlo (uncertainty) Pertinent utility weight literature scarce

Usual care quite good

Confidence intervals were wide