Using behavioural insights in health

Transcription

1 Using behavioural insights in health Andy Hollingsworth & Eva Kolker

2 What are behavioural insights? Psychology Evaluation methods Behavioural Insights Public Policy Economics Understanding how people behave in practice so that we can design policy better

3 We need to think differently about behaviour System 1 Fast thinking/automatic intuitive, effortless 2x2 Taking your daily commute System 2 Slow thinking/reflective deliberate, analytic 24x17 Planning a trip overseas It turns out that the environmental effects on behavior are a lot stronger than most people expect Daniel Kahneman, Nobel Laureate

4 Randomised controlled trials

5 EASY ATTRACTIVE TIMELY SOCIAL

6 Easy Simplify Friction costs Defaults

7 Increasing tax returns by removing friction

8 We ran a trial with HMRC to encourage people to get their tax returns in order Webpage Form Direct to form

9 Removing small frictions seems to make a disproportionate difference 19.2% 23.4% Webpage Direct to Form

10 How can you reduce prescribing errors in clinical settings?

11 Imperial College collaborated with us to simplify hospital prescription charts

12 The revised chart reduced prescribing errors 95% 100% 84% 96% 32% Dose entered correctly Source: King et al., 2014 Prescriber's contact number entered 11% Frequency of medications entered correctly

13 We can make it more difficult too Changing packaging led to fewer deaths Hawton, et al. (2013) Long term effect of reduced pack sizes of paracetamol. British Medical Journal.

14 EASY ATTRACTIVE TIMELY SOCIAL

15 Attractive Attract attention Personalise Rewards & incentives

16 We wanted to find out whether appealing to peoples concerns about their health, or their appearance was more effective at engaging individuals who visit the drinkaware website.

17

18 Likelihood of completing Audit C after visiting homepage n= 104, % 6.65% Health Appearance

19 Likelihood of completing Audit C after visiting homepage Health Appearance n= 104, % 6.43% 6.74% 5.81% Male Female

20 This difference is entirely driven by women s preference for appearance messaging- ~15% relative increase

21 Reducing missed appointments Trial partners Michael Hallsworth, Dan Berry, Michael Sanders, Anna Sallis, Dominic King, Ivo Vlaev, Ara Darzi (2015) Stating appointment costs in SMS reminders reduces missed hospital appointments: Findings from two randomised controlled trials. PLOS One /journal.pone

22 Solution: Trial 1 Messages Name Message Control Number Norm Costs Appt at [hospital] on [Sep 16] at [10:00am]. To cancel or rearrange call the number on your appointment letter. Appt at [hospital] on [Sep 16] at [10:00am]. To cancel or rearrange call We are expecting you at [hospital] on [Sep 16] at [10:00am]. 9 out of 10 people attend. Call if you need to cancel or rearrange. We are expecting you at [hospital] on [Sep 16] at [10:00am]. Not attending costs NHS 160 approx. Call if you need to cancel or rearrange.

23 The best performing message reduced missed appointments by 25% % of appointments recorded as missed Total n = 10,111 ** 11.1% 9.8% 10.0% 8.5% Control Number Norms Costs * p < 0.05, ** p < 0.01

24 This result would lead to 5,800 fewer missed appointments if applied over one year in same location 406,740 fewer if applied in England Benefits x5 with full mobile coverage

25 Adapt: Trial 2 Messages: Best-performing message now the default Name Message Costs specific Costs general Fairness Recording We are expecting you at [hospital] on [Sep 16] at [10:00am]. Not attending costs NHS 160 approx. Call if you need to cancel or rearrange. We are expecting you at [hospital] on [Sep 16] at [10:00am]. Not attending wastes NHS money. Call if you need to cancel or rearrange. We are expecting you at [hospital] on [Sep 16] at [10:00am]. Please be fair to others waiting and call if you need to cancel or rearrange. We are expecting you at [hospital] on [date] at [time]. Please attend or call to cancel/rearrange, or we will record as a missed appt.

26 Specific costs work better than general costs % of appointments recorded as missed ** * Specific costs ( 160 ) are more effective 11.1% 10.6% 9.6% 9.8% 8.2% Control (from previous trial) Fairness Recording General costs Specific costs Total n = 9,862 * p < 0.05, ** p < 0.01

27 EASY ATTRACTIVE TIMELY SOCIAL

28 Social Social norms Commitments Reciprocity

29 Nine out of ten people pay their tax on time.

30 Reducing antibiotic prescribing Hallsworth, et al. (2016) Provision of social norm feedback to high prescribers of antibiotics in general practice: a pragmatic national randomised controlled trial. The Lancet, 387 (10029),

31 GP prescribing data is online

32 Chief Medical Officer letter to 3,227 GPs 3 clear actionable steps The great majority (80%) of practices in London prescribe fewer antibiotics per head than yours. Influential messenger [Presenter Name]

33 Antibiotic items dispensed per 1000 weighted population Effect of Chief Medical Officer letter on antibiotic prescribing, Control Treatment n=1, September (pre) October November December January February March April (post) Month

34 Antibiotic items dispensed per 1000 weighted population Effect of Chief Medical Officer letter on antibiotic prescribing, Control Treatment n=1, September (pre) October November December January February March April (post) Month

35 Antibiotic items dispensed per 1000 weighted population The gap disappeared when the control group also received a letter Control Treatment n=1, September (pre) October November December January February March April (post) Month

36 The letter saved 73,406 doses across 791 practices Direct saving of 92,356 in public sector prescription costs Possible to achieve a 0.85% reduction in England s prescribing

37 Some GPs under-use urgent cancer referral, missing opportunities for early diagnosis. This may be because it s hard to gauge what the right level of referral is.

38 We used social norm feedback 1) Anticipated feedback I will be sending you feedback on how your practice s urgent GP referral compares to other practices in your area. 2) Social norm feedback Low referrers: The great majority (70%) of GM practices have a higher urgent GP referral than yours. Medium referrers: Many practices in GM. Local cancer lead endorsement Suggested easy actions 3) Reminder Data on urgent GP referral in XX will continue to be collected.

39 Results Feedback increased urgent GP referrals by 10%... 10% + Crude rate (per 100,000)

40 Results During the trial alone, feedback resulted in 90 earlier cancer diagnoses 1, Projected impact 11, Across Greater Manchester over 6 months, this could result in: 154 additional early cancer diagnoses Saving 1.2 million Across England over 6 months, this feedback could result in: 2,500 more early cancer diagnoses Saving 20 million Patients referred Without feedback Patients diagnosed With feedback

41 EASY ATTRACTIVE TIMELY SOCIAL

42 Timely Timely prompts Intertemporal choice Implementation intentions

43 How can we spread patients across hospitals with shorter waiting times?

44 The old referrals system Before

45 The new referrals system After

46 % of patients referred to BHRUT services out of all acute referrals Patients were less likely to be referred to hospitals with long waiting times when a red alert was used Without red alerts *** 7.1 With red alerts Red alerts reduced referrals by 38% * p<0.05, ** p<0.01, *** p< n=5,563

47 Further reading

48 Questions?