Can%20paying%20for%20outcomes%20be%20effective%20in%20target%20subgroups?%20%20A%20cluster%20controlled%20study%20of%20smoking%20cessation%20pilots%20in%20the%20NHS

1
Can paying for outcomes be effective in target
subgroups? A cluster controlled study of
smoking cessation pilots in the NHS

• Deirdre OBrien, Research Fellow, Health
  Economics Unit
• Co-Authors Hugh McLeod, Steven Wyatt, Mohammed
  A. Mohammed
• Date 26.06.2013

2
NHS Stop Smoking Services

• Effective cost-effective
• Provide behavioural pharmaceutical support
• Widely rolled out since 2001
• Help reduce health inequalities
• Commissioned through fixed contracts
• Currently undersupplied
• Only accessed by 6 of smokers 5 of pregnant
  smokers

3
Rational for change
4
The Intervention

• First tariff of this kind in the NHS
• All income linked to outcomes
• Enhanced tariffs for target populations
• No income cap
• Introduced Any Qualified Provider (AQP)
  Regulation
• Encourage new market entrance

5
Tariffs for reported Quits
Standard Tariffs Standard Tariffs Tariff for pregnant women before 24 weeks Tariff for pregnant women before 24 weeks
Non-Target Target Non-Target Target
4 week No Prescribing Costs 94 136 216 425
12 week No Prescribing Costs 129 271 395 566

• Higher tariffs when prescribing charges incurred
• In pregnancy, standard tariffs apply after 24th
  week

6
Implementation
7
Research Question
Can paying for outcomes be effective in target
subgroups?

• Effective at aggregate level
• Led to two-fold increase in quits/population
• Examine effect on non-white British smokers
• Smoking rates lt 30 in black Caribbean,
  Bangladeshi Chinese males
• Examine effect on pregnant women
• High risk of poor maternal infant outcomes
• Smoking rates vary between 4 -30

8
Design Matched cluster controlled study
Intervention

• 8 volunteer PCTs

Intervention

• 64 control PCTs

Control
2010

• Difference in rate of change in 4 week quits
• non-white British populations
• Pregnant women

Outcome measures
Control

• Routine NHS SSS data
• ONS population statistics

Data Sources

• Mixed effect Poisson regression models
• Incident rate ratio (IRR)

Analysis
9
Controls derived using ONS clusters
ONS subgroup Cluster No of control PCTs No of Intervention PCTs
Centres with industry A 1 9 2
Centres with industry B 2 7 1
Industrial Hinterland A 3 14 1
Manufacturing Towns A 4 13 1
Prospering Smaller Towns B 5 12 1
Prospering Smaller Towns C 6 9 2

• Rational
• Similar local population demographics
• 1000 population used as a denominator

10
NWB and pregnant women represent a small
proportion of quits
4 week quits in intervention PCTs in 2011/12
11
Results at Aggregate Level (submitted for
publication)

• Two-fold increase in the change in 4 week quits
  per 1000 population
• 11 per year in intervention PCTs, compared to 5
  per year in controls
• Incident rate ratio 1.056, p0.006, CI 1.016 to
  1.098

12
Non-white British smokers
13
Non-white British smokers

• Outcome similar to aggregate results
• Positive effect on NWB smokers
• 26 per year in intervention PCTs
• 14 per year in controls
• Incident rate ratio 1.104, p0.093, CI 0.983 to
  1.240

14
Pregnant Women
15
Pregnant Women

• Little difference for pregnant women
• 13 per year in intervention PCTs,
• 11 per year in controls
• Incident rate ratio 1.019, p0.784, CI 0.888 to
  1.170

16
Enhanced tariffs had different effects on the two
target subgroups

• Appeared to increase supply effectiveness of
  NHS SSS in NWB smokers
• No evidence of effect in pregnant women

16
17
Possible explanations

• Non-white British
• Better Reporting
• Enhanced tariffs incentivised targeting
• Any Qualified Provider regulation
• New providers able to reach niche groups

• Pregnant women
• Outcome measured did not capture full effect
• Stage of pregnancy if sustained
• Barriers to market entry
• Enhanced tariffs not high enough
• Targeting of this group in controls

18
Opportunities for further research..

• Change in other outcomes
• 12 week quits
• Effects on other target groups
• Pregnant women
• Exploring reasons for lack of effect
• Quits in first second trimesters
• Quits sustained until birth

19
Questions