SoGive Cost-effectiveness analysis, StrongMinds (in progress)

	A	B	C	D	E
1	Estimates by SoGive regarding psychotherapy in general		Sources and explanations
2	Effect size of therapy (SDs), average RCT patient, before adj for number of sessions	0.41	see article: What's the effect size of therapy? We ran trim and fill publication bias correction after removing outliers on a pooled effect size of all studies in METAPSY which passed 4 risk of bias criteria i
3	Adjustment for 6 sessions (RCT average is 9)	79%	See article: what's the effect size of therapy? We ran a meta-regressions on number of sessions
4	Duration multiplier for treated patients who fully recovered	0.625	See article: Duration multiplier (title pending). We made an estimate by combining spontaneous remission rates with relapse rates for treated recovered patients.
5	Subjective duration multipier adjustment for the average treated patient	0.9	We used relapse data from treated remission patients to set an upper bound on the duration, but this may be too optimistic because not all patients will recieve full remission, so we made a 90% downwards adjustment
6	length of 6 session intervention (years)	0.1153846154	Some of the improvement happens before the intervention ends
7	SD-years averted per therapy for client	0.2004401905	SD-years represent the area integrated under the curve when standard deviations of improvement attributed to psychotherapy are plotted over time after the intervention is over
8
9	PHQ-9 mean score for pretreatment depression RCT participants	13.77738449	We took means and standard deviations from the metapsy database. See article: Duration multiplier (title pending) for further explanation
10	PHQ-9 score SD for pretreatment depression RCT population	4.211710187	We took means and standard deviations from the metapsy database. See article: Duration multiplier (title pending) for further explanation
11
12	Estimates by GiveWell incorporated into our analysis
13	Number of family members	5.3	Taken from GiveWell's evaluation of HLI's evaluation of StrongMinds
14	Spillover per family member	0.15	Taken from GiveWell's evaluation of HLI's evaluation of StrongMinds
15	Family multiplier	1.645
16	Cost to double income with GiveDirectly	$344.48	we took this number from GiveWell's CEA of GiveDirectly, which says the average income is $285.82 / 88% of every transfer goes to recipients
17	Cost to save a life with AMF	$6,592	Taken from GiveWell's Cost Effectiveness Analysis of AMF
18
19	Conclusions regarding psychotherapy (6 sessions)
20	SD-years depression averted per person treated	0.3297241133
21	QALYs gained / DALYs averted per person treated	0.06037836552
22
23	SoGive Benchmarks
24	How many adjusted life years count for 1 life saved?	25	SoGive benchmarks 25 years of severe depresion to 1 life saved. We defined "severe depression" as "bad enough that you'd rather experience nothing"
25	How many income doulings count for 1 subjective life year?	4
26	At what PHQ-9 score would most people "rather experience nothing"?	23	See article: "How much is psychotherapy worth morally". We used time-trade-off and rating scale data to estimate DALYs averted per SD-year averted
27	How many SD-years of improvement count for 1 subjective life year	5.460964544
28	Gold standard, willingness to pay per 1 DALY of depression averted (USD)	256.18	200 GBP as of Jul 7, 2024.		267.040524
29	Silver standard, willingness to pay per 1 DALY of depression averted	$1,377.94	Charities that meet the Silver standard are approximately as good as cash transfers
30	Bronze standard		Charities that meet the Bronze standard are those we subjectively think are worth watching
31
32	Estimates by SoGive regarding StrongMinds Cost Per Person (CPP)		Some of these subjective judgements are not taken directly from, but are informed by, internal strongminds data which is currently confidential. (Document can be unlocked once StrongMinds okays releasing it, possibly at end of year)
33	Cost per person (CPP) for StrongMinds	33	Average of current estimate and goal in strongminds q2 report. SM has historically met or exceeded its goals.
34	Percent of current budget spent on SMUS and SMG	30%
35	Percent of impact attributable to partners	0.17	See expanded CEA for more details - in summary, about 2/3rd of the program is partner based, we think strongminds basically funds half of it, and half the remaining can attributed to strongminds as leverage.
36	Estimated cost per person	37.60914939	Total increase:	114%
37
38	Conclusions
39	Cost per 1 DALY depression averted	622.8911477
40	How good relative to Against Malaria Foundation?	$0.42	is not more cost effective than amf
41	How good relative to Gold Benchmark?	0.4112757116	does not meet gold benchmark
42	How good relative to cash? (Silver benchmark)	$2.21	is approximately as good as cash
43
44	Going forward - What is needed for psychotherapy to be competitive by SoGive benchmarks?
45	How many times better would the intervention have to be to meet the gold standard?	2.431458926	Given that we think the true costs are 114 of the reported costs, the reported costs would be:
46	What price would make the intervention equal to cash?	$83.20	$73.00
47	What cost per person would have met the Gold Standard?	$15.47	$16.28
48	What cost per person would have beat AMF?	$15.92	$13.97
49
50	how good could a volunteer-led mental health program at <$15 CPP be relative to cash?	5.55
51	Robustness legend
52	Robust estimate - we hit diminishing returns on further analysis and stand behind this estimate
53	Fragile estimate - we put effort into finding this number, but more analysis would help
54	Quick estimate - We spent 0.5-3 hours doing quick numerical calculations with low confidence
55	Complete guess - we made this subjective guess in under half an hour
56	Calculations, moral weights, and numbers taken from other orgs are not colored
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100