WEBVTT

1
00:00:02.430 --> 00:00:06.210
<v ->Prior to joining the faculty at YSPH,</v>

2
00:00:06.210 --> 00:00:07.650
Dr. Cameron was a member

3
00:00:07.650 --> 00:00:10.080
of the Global Health Cost Consortium,

4
00:00:10.080 --> 00:00:11.190
an initiative funded

5
00:00:11.190 --> 00:00:13.620
by the Bill and Melinda Gates Foundation

6
00:00:13.620 --> 00:00:15.780
to provide decision makers

7
00:00:15.780 --> 00:00:17.370
with improved resources

8
00:00:17.370 --> 00:00:18.960
to estimate the cost

9
00:00:18.960 --> 00:00:21.450
of HIV and tuberculosis programs,

10
00:00:21.450 --> 00:00:26.450
and to improve the availability, quality, timing,

11
00:00:26.550 --> 00:00:30.150
and policy relevance of cost information.

12
00:00:30.150 --> 00:00:32.760
He holds an MA in international development

13
00:00:32.760 --> 00:00:37.020
from American University and a PhD in health policy

14
00:00:37.020 --> 00:00:38.850
from the University of California Berkeley

15
00:00:38.850 --> 00:00:40.620
School of Public Health.

16
00:00:40.620 --> 00:00:42.157
The title of his talk is

17
00:00:42.157 --> 00:00:44.510
"Costing and Economic Evaluations

18
00:00:44.510 --> 00:00:46.410
in Implementation Science."

19
00:00:46.410 --> 00:00:48.300
He'll speak for about 45 minutes,

20
00:00:48.300 --> 00:00:52.050
and then we'll open the floor for questions and discussion.

21
00:00:52.050 --> 00:00:53.193
There you go, drew.

22
00:00:54.120 --> 00:00:56.220
<v ->Thanks very much for that introduction, Debbie,</v>

23
00:00:56.220 --> 00:00:59.340
and thank you to those who were kind enough

24
00:00:59.340 --> 00:01:00.573
to join the call today.

25
00:01:01.590 --> 00:01:03.930
I won't introduce myself any further

26
00:01:03.930 --> 00:01:05.820
than to say as a member

27
00:01:05.820 --> 00:01:07.680
of the Center for Methods

28
00:01:07.680 --> 00:01:09.600
and Implementation and Prevention Science,

29
00:01:09.600 --> 00:01:12.600
this is a pretty sort of exciting opportunity

30
00:01:12.600 --> 00:01:14.280
to share some of the work

31
00:01:14.280 --> 00:01:16.440
that I've been involved with in the past.

32
00:01:16.440 --> 00:01:20.760
This talk though, I'm gonna focus largely on, you know,

33
00:01:20.760 --> 00:01:24.780
what might be for many of you a review of topics

34
00:01:24.780 --> 00:01:28.110
as they relate to economic evaluations and costing.

35
00:01:28.110 --> 00:01:30.870
So please do bear with me on that front.

36
00:01:30.870 --> 00:01:33.660
I'll try to get to some novel research findings

37
00:01:33.660 --> 00:01:35.640
towards the end of the talk,

38
00:01:35.640 --> 00:01:39.900
but this is sort of designed as a, anyone can participate.

39
00:01:39.900 --> 00:01:41.970
You don't have to have much of any background

40
00:01:41.970 --> 00:01:43.590
in economic evaluation or costing.

41
00:01:43.590 --> 00:01:46.770
So without much more,

42
00:01:46.770 --> 00:01:50.790
let me get into the topics here.

43
00:01:50.790 --> 00:01:52.740
Okay, so this talk is divided

44
00:01:52.740 --> 00:01:54.060
into three main sections.

45
00:01:54.060 --> 00:01:56.700
First, I'm gonna discuss some key economics concepts

46
00:01:56.700 --> 00:01:59.490
that kind of lay the groundwork for the rest of the talk.

47
00:01:59.490 --> 00:02:01.290
I'll briefly cover some different types

48
00:02:01.290 --> 00:02:04.650
of economic evaluation and decision analyses

49
00:02:04.650 --> 00:02:06.840
and how costing

50
00:02:06.840 --> 00:02:08.400
in particular is an important

51
00:02:08.400 --> 00:02:12.150
and often overlooked component of implementation science.

52
00:02:12.150 --> 00:02:13.740
Second, we'll get into the nitty gritty

53
00:02:13.740 --> 00:02:15.090
of costs and costing.

54
00:02:15.090 --> 00:02:17.340
So what are they, who uses them?

55
00:02:17.340 --> 00:02:18.510
How do you do it?

56
00:02:18.510 --> 00:02:20.370
And what are some of the key considerations

57
00:02:20.370 --> 00:02:22.230
that you have to make along the way

58
00:02:22.230 --> 00:02:25.953
when you are conducting a costing study?

59
00:02:26.790 --> 00:02:29.100
If you have a great deal of experience with costing already,

60
00:02:29.100 --> 00:02:31.050
some of you may find some of this to be familiar,

61
00:02:31.050 --> 00:02:32.760
so I'll apologize in advance,

62
00:02:32.760 --> 00:02:35.460
but I'm still eager for your feedback.

63
00:02:35.460 --> 00:02:36.293
And then finally,

64
00:02:36.293 --> 00:02:38.736
I'm gonna briefly discuss the efforts

65
00:02:38.736 --> 00:02:39.600
of the Global Health Cost Consortium,

66
00:02:39.600 --> 00:02:41.190
which includes a website

67
00:02:41.190 --> 00:02:43.440
with a ton of great costing resources.

68
00:02:43.440 --> 00:02:44.400
So if you're more interested

69
00:02:44.400 --> 00:02:46.530
in costing and economic evaluation,

70
00:02:46.530 --> 00:02:49.620
I can direct you towards those at the end of the talk.

71
00:02:49.620 --> 00:02:50.940
And as part of that effort,

72
00:02:50.940 --> 00:02:52.140
I'll share some of the results

73
00:02:52.140 --> 00:02:53.670
of a somewhat recent publication,

74
00:02:53.670 --> 00:02:56.610
a systematic review of costing literature on HIV

75
00:02:56.610 --> 00:02:58.230
in Sub-Saharan Africa,

76
00:02:58.230 --> 00:03:01.500
and another study that's in progress.

77
00:03:01.500 --> 00:03:04.560
So let me move ahead here.

78
00:03:04.560 --> 00:03:06.030
So there are four main takeaways.

79
00:03:06.030 --> 00:03:09.510
You know, if you take nothing else from this talk,

80
00:03:09.510 --> 00:03:11.523
and I wouldn't blame you if you didn't.

81
00:03:13.350 --> 00:03:17.070
Four things that I'd like you to walk away from this talk,

82
00:03:17.070 --> 00:03:20.460
considering like number one is perspective.

83
00:03:20.460 --> 00:03:23.940
I'll get into what perspective is, and timeframe.

84
00:03:23.940 --> 00:03:26.130
These are two concepts that matter,

85
00:03:26.130 --> 00:03:30.360
I think the absolute most in terms of economic evaluations

86
00:03:30.360 --> 00:03:33.359
and costing and how they're conducted and the findings,

87
00:03:33.359 --> 00:03:34.740
you know, that you were able to generalize

88
00:03:34.740 --> 00:03:36.513
from these studies.

89
00:03:37.440 --> 00:03:39.780
I wanna also leave you with the impression

90
00:03:39.780 --> 00:03:42.270
that although expenditures are real,

91
00:03:42.270 --> 00:03:43.380
they're real things,

92
00:03:43.380 --> 00:03:44.790
things you can quantify,

93
00:03:44.790 --> 00:03:46.830
costs really aren't.

94
00:03:46.830 --> 00:03:48.840
They're conceptual in nature,

95
00:03:48.840 --> 00:03:51.810
and they depend on a lot of different assumptions.

96
00:03:51.810 --> 00:03:55.110
So, you know, we'll get into the nitty gritty of that.

97
00:03:55.110 --> 00:03:57.150
I'd also like to leave you with the idea

98
00:03:57.150 --> 00:03:59.520
that perfect is often the enemy of the good.

99
00:03:59.520 --> 00:04:01.260
I think that's a concept that's very familiar

100
00:04:01.260 --> 00:04:04.470
to anybody who's ever published a paper, for example.

101
00:04:04.470 --> 00:04:06.150
But on the flip side,

102
00:04:06.150 --> 00:04:09.537
little details in your decision making and costing

103
00:04:09.537 --> 00:04:12.510
and data collection can matter a lot.

104
00:04:12.510 --> 00:04:15.540
So there's a balance that we try to strike

105
00:04:15.540 --> 00:04:17.610
when we do these types of studies.

106
00:04:17.610 --> 00:04:18.600
And then finally,

107
00:04:18.600 --> 00:04:21.660
reporting standards and quality are two areas

108
00:04:21.660 --> 00:04:23.940
that have a lot of room to improve in this field.

109
00:04:23.940 --> 00:04:26.193
And I'll show you some examples of why.

110
00:04:27.300 --> 00:04:29.310
So what is economic evaluation?

111
00:04:29.310 --> 00:04:32.820
Economic evaluation is an often sort of misunderstood term.

112
00:04:32.820 --> 00:04:34.740
You've probably heard of monitoring and evaluation,

113
00:04:34.740 --> 00:04:37.180
which typically refers to monitoring processes

114
00:04:37.180 --> 00:04:40.830
for a program's implementation and ongoing operation.

115
00:04:40.830 --> 00:04:42.330
If you work in the social sciences,

116
00:04:42.330 --> 00:04:44.160
you've probably heard of impact evaluation,

117
00:04:44.160 --> 00:04:47.790
which typically refers to evaluating, say, health, social,

118
00:04:47.790 --> 00:04:50.370
or economic impacts of the program.

119
00:04:50.370 --> 00:04:54.390
These are all separate but related fields and methods.

120
00:04:54.390 --> 00:04:57.750
Economic evaluation is sort of an umbrella term.

121
00:04:57.750 --> 00:05:00.300
This includes a broad range of comparative methods

122
00:05:00.300 --> 00:05:01.260
to help to evaluate

123
00:05:01.260 --> 00:05:04.650
choices or trade-offs between costs and benefits.

124
00:05:04.650 --> 00:05:08.013
Another familiar term might be decision analysis.

125
00:05:08.013 --> 00:05:10.208
At the heart though of economic evaluation

126
00:05:10.208 --> 00:05:12.930
and decision analysis is a concept that, you know,

127
00:05:12.930 --> 00:05:17.580
we spoke about or that I'll speak about a little further

128
00:05:17.580 --> 00:05:18.450
into the talk.

129
00:05:18.450 --> 00:05:20.370
And that is opportunity costs.

130
00:05:20.370 --> 00:05:22.620
So when we talk about economic evaluation,

131
00:05:22.620 --> 00:05:24.120
what we typically wanna understand

132
00:05:24.120 --> 00:05:26.310
is what's the best value for money?

133
00:05:26.310 --> 00:05:27.920
And in the most basic sense,

134
00:05:27.920 --> 00:05:31.140
cost is what you pay and value is what you get.

135
00:05:31.140 --> 00:05:31.990
It's the benefit.

136
00:05:32.850 --> 00:05:35.880
So it won't always indicate a clear choice,

137
00:05:35.880 --> 00:05:39.630
but it will help to evaluate options quantitatively

138
00:05:39.630 --> 00:05:42.093
based on a defined set of decisions.

139
00:05:43.020 --> 00:05:46.590
These processes, they aim to capture opportunity costs

140
00:05:46.590 --> 00:05:48.300
of doing one thing over another.

141
00:05:48.300 --> 00:05:49.200
So for example,

142
00:05:49.200 --> 00:05:51.750
do we invest in home visits

143
00:05:51.750 --> 00:05:54.750
or do we make mass media platforms

144
00:05:54.750 --> 00:05:56.940
to disseminate behavior change communication

145
00:05:56.940 --> 00:06:01.650
for improved, say infant and young child feeding practices,

146
00:06:01.650 --> 00:06:04.980
or do we promote home gardens or vitamin A capsule

147
00:06:04.980 --> 00:06:09.783
supplementations to address micronutrient deficiencies?

148
00:06:10.620 --> 00:06:12.000
So lastly, I'll leave you

149
00:06:12.000 --> 00:06:13.650
with the thought that economic evaluations

150
00:06:13.650 --> 00:06:15.270
are comparative by nature.

151
00:06:15.270 --> 00:06:18.420
So not only do they compare costs and benefits,

152
00:06:18.420 --> 00:06:19.980
but a comprehensive evaluation

153
00:06:19.980 --> 00:06:22.830
will compare two or more different options.

154
00:06:22.830 --> 00:06:24.690
So let's look at a couple of different types

155
00:06:24.690 --> 00:06:26.280
of economic evaluations.

156
00:06:26.280 --> 00:06:29.520
So you may have come across these in the past.

157
00:06:29.520 --> 00:06:30.930
There's really a large family

158
00:06:30.930 --> 00:06:33.180
of economic evaluation methods,

159
00:06:33.180 --> 00:06:35.580
but it's important to note that all of these different types

160
00:06:35.580 --> 00:06:38.730
of analyses and tools at their core

161
00:06:38.730 --> 00:06:40.410
are measuring costs in dollars.

162
00:06:40.410 --> 00:06:42.870
So that that column down the middle.

163
00:06:42.870 --> 00:06:46.110
And that's the main focus of the rest of this talk.

164
00:06:46.110 --> 00:06:49.920
But a key difference across these different types

165
00:06:49.920 --> 00:06:53.850
of evaluations is gonna be whose costs it is

166
00:06:53.850 --> 00:06:55.230
that we're taking into account

167
00:06:55.230 --> 00:06:57.600
as well as how we're measuring benefits.

168
00:06:57.600 --> 00:07:00.120
So all decision analyses are comparative,

169
00:07:00.120 --> 00:07:01.260
as I mentioned,

170
00:07:01.260 --> 00:07:03.060
we're gonna compare one or more options

171
00:07:03.060 --> 00:07:04.200
in relation to another.

172
00:07:04.200 --> 00:07:05.940
So that other option could just be

173
00:07:05.940 --> 00:07:06.900
what the current standard

174
00:07:06.900 --> 00:07:09.000
of care is or the status quo.

175
00:07:09.000 --> 00:07:10.800
It could be doing nothing

176
00:07:10.800 --> 00:07:13.020
or it could be another active intervention.

177
00:07:13.020 --> 00:07:14.850
So one of the first things that we need to do

178
00:07:14.850 --> 00:07:15.960
in any such analysis

179
00:07:15.960 --> 00:07:18.496
is to understand where a comparator is

180
00:07:18.496 --> 00:07:21.188
and the most basic type of economic evaluation

181
00:07:21.188 --> 00:07:22.830
is the one on the top there.

182
00:07:22.830 --> 00:07:25.140
We'll start with cost efficiency analysis,

183
00:07:25.140 --> 00:07:27.690
which is sort of a partial form of economic appraisal

184
00:07:27.690 --> 00:07:29.030
'cause it looks only at the costs

185
00:07:29.030 --> 00:07:30.660
of the program and it doesn't

186
00:07:30.660 --> 00:07:33.900
really provide any information on let's say agriculture,

187
00:07:33.900 --> 00:07:36.570
nutrition, or health outcome of interest.

188
00:07:36.570 --> 00:07:39.600
This typically reports out on the cost per output achieved.

189
00:07:39.600 --> 00:07:41.130
So another example of this might be

190
00:07:41.130 --> 00:07:43.650
like a budget impact analysis where you're looking

191
00:07:43.650 --> 00:07:46.860
at the cost per output, per patient served,

192
00:07:46.860 --> 00:07:49.923
or per vaccine administered.

193
00:07:50.850 --> 00:07:53.880
Now the more common type of economic evaluation

194
00:07:53.880 --> 00:07:55.920
for evaluating health interventions

195
00:07:55.920 --> 00:07:58.170
is cost effectiveness analysis.

196
00:07:58.170 --> 00:08:02.250
And CEAs, they compare the cost and outcomes of two or more

197
00:08:02.250 --> 00:08:05.610
alternatives or compare a new intervention

198
00:08:05.610 --> 00:08:07.410
or treatment with the status quo.

199
00:08:07.410 --> 00:08:09.900
CEAs relate the cost associated

200
00:08:09.900 --> 00:08:13.920
with the health outcome such as cost per disease avoided,

201
00:08:13.920 --> 00:08:15.180
cost per death avoided

202
00:08:15.180 --> 00:08:19.020
or additional expected life year for example.

203
00:08:19.020 --> 00:08:21.030
And then there's some other types of analysis

204
00:08:21.030 --> 00:08:22.710
that we won't get into too much.

205
00:08:22.710 --> 00:08:24.870
Cost utility analysis,

206
00:08:24.870 --> 00:08:28.650
you know, includes analyses with say disability

207
00:08:28.650 --> 00:08:30.550
or quality adjusted life years

208
00:08:31.950 --> 00:08:36.950
or other standardized measures of preference.

209
00:08:37.230 --> 00:08:38.970
And then there's cost benefit analysis,

210
00:08:38.970 --> 00:08:42.210
which is a method in which both the costs and the benefits

211
00:08:42.210 --> 00:08:45.209
of the intervention are expressed in monetary terms.

212
00:08:45.209 --> 00:08:47.490
So in agriculture for example,

213
00:08:47.490 --> 00:08:50.700
this might be the value of increased crop production.

214
00:08:50.700 --> 00:08:54.180
While in health it would be the dollar value associated

215
00:08:54.180 --> 00:08:55.770
with the life years that were gained

216
00:08:55.770 --> 00:08:57.120
because of an intervention.

217
00:08:57.960 --> 00:09:00.900
There's several other types of decision analysis

218
00:09:00.900 --> 00:09:03.180
that I haven't included in this table

219
00:09:03.180 --> 00:09:05.880
and they exist, so sorry.

220
00:09:05.880 --> 00:09:09.630
So those include return on investment analysis,

221
00:09:09.630 --> 00:09:12.063
cost of illness analysis and others.

222
00:09:13.140 --> 00:09:14.220
So as I've suggested,

223
00:09:14.220 --> 00:09:18.180
different sectors kind of use different of these approaches.

224
00:09:18.180 --> 00:09:19.013
So for example,

225
00:09:19.013 --> 00:09:21.090
global health and education disciplines

226
00:09:21.090 --> 00:09:24.540
have promoted the use of cost effectiveness analysis.

227
00:09:24.540 --> 00:09:27.690
Implementation science probably naturally falls

228
00:09:27.690 --> 00:09:29.820
under this category

229
00:09:29.820 --> 00:09:32.580
or in the cost efficiency analysis category

230
00:09:32.580 --> 00:09:34.101
with a little bit more emphasis

231
00:09:34.101 --> 00:09:36.210
on perhaps cost efficiency

232
00:09:36.210 --> 00:09:39.420
depending on the scope of the evaluation taking place.

233
00:09:39.420 --> 00:09:41.820
And I'll get into that in a moment.

234
00:09:41.820 --> 00:09:44.640
Meanwhile, you might see agriculture

235
00:09:44.640 --> 00:09:46.380
or large scale investment

236
00:09:46.380 --> 00:09:48.780
in infrastructure being evaluated using

237
00:09:48.780 --> 00:09:50.433
say a cost benefit analysis.

238
00:09:52.860 --> 00:09:57.810
So zooming out to sort of the level that we often interact

239
00:09:59.878 --> 00:10:02.130
with these types of evaluations,

240
00:10:02.130 --> 00:10:05.370
say an aim three of your NIH proposal,

241
00:10:05.370 --> 00:10:09.390
depending of course on your desired evaluation type,

242
00:10:09.390 --> 00:10:10.710
you'll often wanna arrive

243
00:10:10.710 --> 00:10:12.720
at really a simple division problem.

244
00:10:12.720 --> 00:10:15.900
The cost of a program or policy or intervention

245
00:10:15.900 --> 00:10:18.450
divided by the outcome in question.

246
00:10:18.450 --> 00:10:20.100
For cost effectiveness analysis,

247
00:10:20.100 --> 00:10:22.830
the outcome is some measure of effectiveness,

248
00:10:22.830 --> 00:10:26.820
particularly a health outcome.

249
00:10:26.820 --> 00:10:29.460
For budget impact analysis and cost efficiency.

250
00:10:29.460 --> 00:10:31.530
This might be an intermediate outcome

251
00:10:31.530 --> 00:10:35.670
per person vaccinated or per patients served.

252
00:10:35.670 --> 00:10:36.660
But at its core,

253
00:10:36.660 --> 00:10:38.543
all the economic evaluations boil down

254
00:10:38.543 --> 00:10:40.920
to this simple division problem.

255
00:10:40.920 --> 00:10:43.350
And I wanna mention one more

256
00:10:43.350 --> 00:10:44.640
slightly more complicated

257
00:10:44.640 --> 00:10:46.560
but still simple division problem

258
00:10:46.560 --> 00:10:50.508
that you may have come across at some point, you know,

259
00:10:50.508 --> 00:10:54.330
in the process of of putting together a proposal.

260
00:10:54.330 --> 00:10:58.230
And that is one that builds very simply on the previous,

261
00:10:58.230 --> 00:11:00.750
and that's for cost effectiveness analysis

262
00:11:00.750 --> 00:11:02.760
that compares the cost effectiveness

263
00:11:02.760 --> 00:11:04.620
of the intervention and question

264
00:11:04.620 --> 00:11:06.030
to some particular alternative.

265
00:11:06.030 --> 00:11:08.190
So that's, that's the ICER.

266
00:11:08.190 --> 00:11:11.253
And this stands for incremental cost effectiveness ratio.

267
00:11:12.480 --> 00:11:14.910
So usually the comparator here is a status quo

268
00:11:14.910 --> 00:11:16.143
or a standard of care,

269
00:11:17.040 --> 00:11:20.683
but it might be some new alternative you're considering too.

270
00:11:20.683 --> 00:11:23.863
<v Person>That is very useful for developers.</v>

271
00:11:23.863 --> 00:11:26.520
<v ->I dunno if I'm getting a question here or not,</v>

272
00:11:26.520 --> 00:11:28.923
but I'll just continue about that.

273
00:11:30.360 --> 00:11:31.920
You can also use ICERs to compare

274
00:11:31.920 --> 00:11:33.060
between multiple different

275
00:11:33.060 --> 00:11:35.580
alternative interventions or configurations.

276
00:11:35.580 --> 00:11:38.430
You might end up with like a list of different interventions

277
00:11:38.430 --> 00:11:40.380
that you want to choose between

278
00:11:40.380 --> 00:11:44.580
using a process like a shopping spree scenario.

279
00:11:44.580 --> 00:11:45.930
I won't get into the details of that,

280
00:11:45.930 --> 00:11:47.010
but there's lots of different ways

281
00:11:47.010 --> 00:11:48.360
that you can use this tool.

282
00:11:51.840 --> 00:11:52.800
Lemme move on.

283
00:11:52.800 --> 00:11:53.850
In this talk though,

284
00:11:53.850 --> 00:11:55.650
I'm mostly concerned with the numerator.

285
00:11:55.650 --> 00:11:58.650
So how do we choose what counts and what doesn't count?

286
00:11:58.650 --> 00:12:00.630
What goes into the process of collecting data

287
00:12:00.630 --> 00:12:02.790
to inform these numbers

288
00:12:02.790 --> 00:12:04.530
and how might we begin to distinguish

289
00:12:04.530 --> 00:12:07.500
between studies that do that well or do that poorly?

290
00:12:07.500 --> 00:12:10.230
So I wanna get through a couple of key concepts here

291
00:12:10.230 --> 00:12:12.150
before we get started.

292
00:12:12.150 --> 00:12:15.090
And those are scarcity, opportunity costs,

293
00:12:15.090 --> 00:12:17.700
types of efficiency and perspective.

294
00:12:17.700 --> 00:12:20.460
And I'll linger on perspective a bit longer

295
00:12:20.460 --> 00:12:21.480
than the others.

296
00:12:21.480 --> 00:12:24.450
Scarcity is one of the most important concepts in economics

297
00:12:24.450 --> 00:12:27.480
and that's simply stated, resources are limited.

298
00:12:27.480 --> 00:12:29.940
So whether we have relatively more resources

299
00:12:29.940 --> 00:12:33.390
as within the US or say other OECD countries,

300
00:12:33.390 --> 00:12:35.040
or relatively fewer resources

301
00:12:35.040 --> 00:12:39.000
as many low and middle income countries find,

302
00:12:39.000 --> 00:12:41.970
there's always trade offs to be made in terms of investments

303
00:12:41.970 --> 00:12:44.040
and competing programs and activities.

304
00:12:44.040 --> 00:12:45.993
So it's a major consideration.

305
00:12:46.980 --> 00:12:49.650
The most powerful implication of the concept of scarcity

306
00:12:49.650 --> 00:12:51.630
is of course opportunity costs.

307
00:12:51.630 --> 00:12:55.050
So this refers to the simple truth that the use of resources

308
00:12:55.050 --> 00:12:58.410
for any one purpose precludes their use for another.

309
00:12:58.410 --> 00:13:01.440
And I have to say that opportunity cost is is a really great

310
00:13:01.440 --> 00:13:05.100
concept for focusing the mind by counteracting

311
00:13:05.100 --> 00:13:07.590
our own personal tendencies towards inertia,

312
00:13:07.590 --> 00:13:11.580
which is to say the tendency to say unthinkingly continue

313
00:13:11.580 --> 00:13:15.360
this year, what worked or what we did last year,

314
00:13:15.360 --> 00:13:16.193
or for that matter,

315
00:13:16.193 --> 00:13:18.960
what didn't work particularly well last year.

316
00:13:18.960 --> 00:13:21.810
So thinking about opportunity costs stimulates

317
00:13:21.810 --> 00:13:24.630
some creativity about the best use of resources

318
00:13:24.630 --> 00:13:28.590
given the needs and alternatives that exist.

319
00:13:28.590 --> 00:13:30.330
And the problem of resource allocation

320
00:13:30.330 --> 00:13:32.520
then becomes where can we get

321
00:13:32.520 --> 00:13:36.000
the greatest benefit for a given expenditure.

322
00:13:36.000 --> 00:13:38.400
And cost effectiveness analysis provides

323
00:13:38.400 --> 00:13:40.560
the sort of conceptual framework and methods

324
00:13:40.560 --> 00:13:42.902
to help make that determination.

325
00:13:42.902 --> 00:13:44.070
Efficiency.

326
00:13:44.070 --> 00:13:46.050
I'll just touch on briefly,

327
00:13:46.050 --> 00:13:47.250
there's two major types.

328
00:13:47.250 --> 00:13:48.270
Allocative efficiency,

329
00:13:48.270 --> 00:13:50.850
which con concerns the division of funds

330
00:13:50.850 --> 00:13:53.160
among different interventions.

331
00:13:53.160 --> 00:13:56.310
So HIV prevention versus treatment.

332
00:13:56.310 --> 00:13:59.010
This often involves complex choices,

333
00:13:59.010 --> 00:14:02.040
even if the health gains achievable in one area are greater

334
00:14:02.040 --> 00:14:04.950
than in another because costs also differ.

335
00:14:04.950 --> 00:14:06.510
And then there's technical efficiency,

336
00:14:06.510 --> 00:14:08.490
which is about how we implement an intervention

337
00:14:08.490 --> 00:14:10.290
that we've decided upon.

338
00:14:10.290 --> 00:14:11.280
So for example,

339
00:14:11.280 --> 00:14:13.710
what's the best program designed

340
00:14:13.710 --> 00:14:16.530
to deliver ARV therapy?

341
00:14:16.530 --> 00:14:19.530
Both kinds of efficiency require good cost data

342
00:14:19.530 --> 00:14:21.600
in order to properly assess them.

343
00:14:21.600 --> 00:14:24.210
However, technical efficiency is only about costs,

344
00:14:24.210 --> 00:14:25.530
whereas allocative efficiency

345
00:14:25.530 --> 00:14:28.293
concerns health impacts as well.

346
00:14:29.280 --> 00:14:30.357
And then finally, perspective,

347
00:14:30.357 --> 00:14:31.487
and I'll spend a little more time here

348
00:14:31.487 --> 00:14:33.900
'cause I think this is a really key thing

349
00:14:33.900 --> 00:14:37.230
that we all need to be on the same level with.

350
00:14:37.230 --> 00:14:39.993
This is perhaps the most important concept.

351
00:14:40.890 --> 00:14:42.180
So different stakeholders

352
00:14:42.180 --> 00:14:44.610
have different experiences of cost.

353
00:14:44.610 --> 00:14:46.170
There's the societal perspective,

354
00:14:46.170 --> 00:14:48.960
which is what we consider the most inclusive

355
00:14:48.960 --> 00:14:53.670
and is often recommended for most policy analyses

356
00:14:53.670 --> 00:14:55.420
in low and middle income countries.

357
00:14:56.610 --> 00:14:57.600
In this case,

358
00:14:57.600 --> 00:15:00.550
all costs are considered regardless of who incurs them

359
00:15:01.470 --> 00:15:02.940
within a society.

360
00:15:02.940 --> 00:15:05.790
However, considering the perspective

361
00:15:05.790 --> 00:15:09.270
of specific stakeholders is particularly useful

362
00:15:09.270 --> 00:15:12.810
in helping to address those stakeholder concerns.

363
00:15:12.810 --> 00:15:16.710
So will a payer or an insurance provider lose money

364
00:15:16.710 --> 00:15:19.980
even though a society gains in net savings?

365
00:15:19.980 --> 00:15:22.320
Or should we align financial incentives

366
00:15:22.320 --> 00:15:24.963
and how do we go about accomplishing that?

367
00:15:26.490 --> 00:15:28.560
So here's another sort of useful graphic

368
00:15:28.560 --> 00:15:30.000
for thinking about perspectives

369
00:15:30.000 --> 00:15:33.060
and in the context of a cost effectiveness analysis,

370
00:15:33.060 --> 00:15:35.100
these are the areas in which you might consider your choice

371
00:15:35.100 --> 00:15:36.870
of inputs for costs and benefits

372
00:15:36.870 --> 00:15:38.370
for a programmer intervention.

373
00:15:39.570 --> 00:15:43.080
So you'll see that in the rows of the table here.

374
00:15:43.080 --> 00:15:45.720
So many decision analyses are gonna be taking a societal

375
00:15:45.720 --> 00:15:47.763
perspective to start,

376
00:15:48.840 --> 00:15:51.870
but let's drill down a little bit more

377
00:15:51.870 --> 00:15:55.020
where we're talking about the hospital perspective

378
00:15:55.020 --> 00:15:56.700
or a payer perspective.

379
00:15:56.700 --> 00:16:00.303
So this is sort of the narrowest definition.

380
00:16:01.170 --> 00:16:03.480
They'll have their own particular interests

381
00:16:03.480 --> 00:16:05.520
and we can also think about this being done

382
00:16:05.520 --> 00:16:07.323
from the perspective of an employer,

383
00:16:08.340 --> 00:16:11.240
how we might think about the healthcare system as a whole.

384
00:16:12.300 --> 00:16:17.300
The next level out would be the payer perspective,

385
00:16:18.720 --> 00:16:22.140
and that captures the insurer

386
00:16:22.140 --> 00:16:24.690
and that that may be a government entity.

387
00:16:24.690 --> 00:16:26.373
So Medicare, Medicaid,

388
00:16:27.210 --> 00:16:30.090
and then zooming out even further

389
00:16:30.090 --> 00:16:32.010
the societal perspective.

390
00:16:32.010 --> 00:16:36.270
And that captures, you know, this broader range of costs.

391
00:16:36.270 --> 00:16:37.103
Okay?

392
00:16:40.140 --> 00:16:44.430
I wanna mention then the question arises,

393
00:16:44.430 --> 00:16:47.310
which of these perspectives should we choose?

394
00:16:47.310 --> 00:16:49.833
So in implementation science,

395
00:16:51.180 --> 00:16:56.180
it often depends on the priorities

396
00:16:56.874 --> 00:17:00.750
of the those on the ground partners

397
00:17:00.750 --> 00:17:02.400
and the institutions that you're working with

398
00:17:02.400 --> 00:17:05.220
when you roll out one of these studies.

399
00:17:05.220 --> 00:17:07.920
So in low and middle income country settings,

400
00:17:07.920 --> 00:17:12.660
the guidance from traditional reference case examples,

401
00:17:12.660 --> 00:17:14.940
and I'll touch on those in a bit,

402
00:17:14.940 --> 00:17:16.920
tend to encourage including

403
00:17:16.920 --> 00:17:19.590
this broader set of perspectives.

404
00:17:19.590 --> 00:17:20.730
So the societal

405
00:17:20.730 --> 00:17:22.380
or the health system perspective.

406
00:17:23.550 --> 00:17:25.320
However, narrower perspectives

407
00:17:25.320 --> 00:17:28.650
are usually these are more directed at the interests

408
00:17:28.650 --> 00:17:30.390
of key stakeholders,

409
00:17:30.390 --> 00:17:33.240
and that may be the number one priority

410
00:17:33.240 --> 00:17:37.350
in any one of these examples.

411
00:17:37.350 --> 00:17:38.910
So it really, you know,

412
00:17:38.910 --> 00:17:41.370
in short it really depends on the priorities

413
00:17:41.370 --> 00:17:43.720
of your partner agencies

414
00:17:44.670 --> 00:17:46.710
and those who are rolling out

415
00:17:46.710 --> 00:17:50.133
whatever this intervention happens to be.

416
00:17:51.420 --> 00:17:53.040
You know, there's a really great quote

417
00:17:53.040 --> 00:17:53.873
from this paper

418
00:17:53.873 --> 00:17:54.900
by Eisman and colleagues

419
00:17:54.900 --> 00:17:56.580
that just came out last year,

420
00:17:56.580 --> 00:17:58.807
and really it's a pragmatic answer.

421
00:17:58.807 --> 00:18:01.380
"The perspective of the stakeholder and the decision makers

422
00:18:01.380 --> 00:18:03.570
who will be informed by the analysis

423
00:18:03.570 --> 00:18:05.880
should probably be prioritized."

424
00:18:05.880 --> 00:18:09.810
But you can also conduct analysis at multiple levels,

425
00:18:09.810 --> 00:18:10.950
you know, as long as you're able

426
00:18:10.950 --> 00:18:13.200
to distinguish between those different levels

427
00:18:14.220 --> 00:18:15.390
when you write things up.

428
00:18:15.390 --> 00:18:18.570
So just keep in mind in this process,

429
00:18:18.570 --> 00:18:21.750
the organizations that are adopting evidence-based policies,

430
00:18:21.750 --> 00:18:23.850
they really need to know what it's gonna cost them

431
00:18:23.850 --> 00:18:26.103
in their setting, given their resources.

432
00:18:28.080 --> 00:18:29.310
But then there's a trade off

433
00:18:29.310 --> 00:18:31.740
and that is whose reality counts

434
00:18:31.740 --> 00:18:33.030
and whose being missed

435
00:18:33.030 --> 00:18:34.800
by these types of analyses?

436
00:18:34.800 --> 00:18:39.270
I don't know if those on the call have ever read anything

437
00:18:39.270 --> 00:18:41.280
by Robert Chambers, but you know,

438
00:18:41.280 --> 00:18:42.430
I think a lot about

439
00:18:44.250 --> 00:18:48.120
one of his early books, "Whose Reality Counts,"

440
00:18:48.120 --> 00:18:49.470
when it comes to development interventions

441
00:18:49.470 --> 00:18:52.743
when I think about this idea of perspective.

442
00:18:55.050 --> 00:18:57.600
Another consideration that's closely related

443
00:18:57.600 --> 00:19:00.150
to perspective is timeframe.

444
00:19:00.150 --> 00:19:02.550
And so that is not just what costs and benefits

445
00:19:02.550 --> 00:19:05.160
should we collect, but like when and where

446
00:19:05.160 --> 00:19:07.290
in the project life cycle itself

447
00:19:07.290 --> 00:19:09.480
is it appropriate to collect 'em?

448
00:19:09.480 --> 00:19:11.520
So one option, you know, and we'll start

449
00:19:11.520 --> 00:19:13.170
with the first in that list there

450
00:19:13.170 --> 00:19:16.440
is to focus solely on the short term healthcare costs,

451
00:19:16.440 --> 00:19:18.750
which could include the costs of the intervention,

452
00:19:18.750 --> 00:19:22.110
costs of downstream stream healthcare.

453
00:19:22.110 --> 00:19:23.910
This approach is consistent with,

454
00:19:23.910 --> 00:19:24.743
I think, you know,

455
00:19:24.743 --> 00:19:27.600
most budget impact analyses, frameworks,

456
00:19:27.600 --> 00:19:29.970
particularly those mentioned

457
00:19:29.970 --> 00:19:32.580
in the CHEERS guidelines,

458
00:19:32.580 --> 00:19:35.610
which I'll provide some resources for later.

459
00:19:35.610 --> 00:19:37.770
And in the context of costing,

460
00:19:37.770 --> 00:19:40.770
this is usually on the order of about six months

461
00:19:40.770 --> 00:19:42.660
after a program's been implemented.

462
00:19:42.660 --> 00:19:45.990
So it's once that program is reached maturity,

463
00:19:45.990 --> 00:19:50.220
you know, what does it cost to continue to implement?

464
00:19:50.220 --> 00:19:52.200
But that ignores another set of questions,

465
00:19:52.200 --> 00:19:54.540
which is, you know, number two here,

466
00:19:54.540 --> 00:19:56.910
the intervention startup.

467
00:19:56.910 --> 00:19:58.440
And it might be extremely valuable,

468
00:19:58.440 --> 00:20:00.150
the organizations and key stakeholders

469
00:20:00.150 --> 00:20:01.080
who are carrying out

470
00:20:01.080 --> 00:20:03.510
that evidence-based program

471
00:20:03.510 --> 00:20:04.830
to include these costs

472
00:20:04.830 --> 00:20:07.080
in order to capture some sense

473
00:20:07.080 --> 00:20:08.520
of what it would take for a system

474
00:20:08.520 --> 00:20:10.200
to start to integrate new practices,

475
00:20:10.200 --> 00:20:11.424
how long it might take

476
00:20:11.424 --> 00:20:14.223
for those systems to reach maturity.

477
00:20:15.360 --> 00:20:16.193
Another option,

478
00:20:16.193 --> 00:20:18.030
the third in that list

479
00:20:18.030 --> 00:20:20.640
is to include additional research costs,

480
00:20:20.640 --> 00:20:23.752
research costs included in the design and development

481
00:20:23.752 --> 00:20:28.380
of implementing a new evidence-based practice.

482
00:20:28.380 --> 00:20:32.640
So that might be adapting that evidence-based practice

483
00:20:32.640 --> 00:20:34.800
to a local setting or an intervention.

484
00:20:34.800 --> 00:20:38.707
And it could also include some of the costs of say,

485
00:20:40.800 --> 00:20:42.360
monitoring and getting feedback

486
00:20:42.360 --> 00:20:43.860
from participants in order

487
00:20:43.860 --> 00:20:48.780
to adapt and to improve,

488
00:20:48.780 --> 00:20:51.810
I guess the implementation of that process.

489
00:20:51.810 --> 00:20:52.643
And then finally,

490
00:20:52.643 --> 00:20:53.476
there's this question

491
00:20:53.476 --> 00:20:55.290
of future downstream costs and benefits.

492
00:20:55.290 --> 00:20:56.790
And that's particularly to patients

493
00:20:56.790 --> 00:20:58.170
and to program beneficiaries.

494
00:20:58.170 --> 00:21:01.050
So that's number four in the image that you see.

495
00:21:01.050 --> 00:21:03.360
And this is often included, you know,

496
00:21:03.360 --> 00:21:04.193
as I mentioned before,

497
00:21:04.193 --> 00:21:05.280
in perspectives that are taken

498
00:21:05.280 --> 00:21:06.513
from the societal level,

499
00:21:07.530 --> 00:21:11.580
but all too often they fall outside

500
00:21:11.580 --> 00:21:13.500
of the scope of interest

501
00:21:13.500 --> 00:21:14.850
for implementing partners

502
00:21:14.850 --> 00:21:16.650
and particularly for payers.

503
00:21:16.650 --> 00:21:17.520
You know, and I'll just mention

504
00:21:17.520 --> 00:21:21.210
that this is one of the main major downsides to many,

505
00:21:21.210 --> 00:21:23.700
if not most economic evaluations

506
00:21:23.700 --> 00:21:25.380
that are conducted domestically.

507
00:21:25.380 --> 00:21:26.370
So when it, you know,

508
00:21:26.370 --> 00:21:27.750
this is a setting where the payer

509
00:21:27.750 --> 00:21:30.780
is often a private health insurer who has no monetary stake

510
00:21:30.780 --> 00:21:32.340
in the future health state

511
00:21:32.340 --> 00:21:34.140
of their clients because

512
00:21:34.140 --> 00:21:38.640
in most cases those customers might age out

513
00:21:38.640 --> 00:21:40.200
and move into a new health insurer,

514
00:21:40.200 --> 00:21:42.783
so Medicare, for example, when you turn 65.

515
00:21:43.920 --> 00:21:46.470
So I'll just mention that an analyst may choose

516
00:21:46.470 --> 00:21:48.600
to include any number of these different components

517
00:21:48.600 --> 00:21:50.700
of an intervention or these different parts

518
00:21:50.700 --> 00:21:51.900
of the timeframe.

519
00:21:51.900 --> 00:21:52.733
But in any case,

520
00:21:52.733 --> 00:21:54.870
it's really essential to be able to distinguish

521
00:21:54.870 --> 00:21:56.850
between these different sets of costs and benefits

522
00:21:56.850 --> 00:21:59.610
and to be explicit about the scope,

523
00:21:59.610 --> 00:22:01.830
particularly for generalizability

524
00:22:01.830 --> 00:22:03.393
and understanding limitations.

525
00:22:05.130 --> 00:22:06.870
So lemme go into costs and costing,

526
00:22:06.870 --> 00:22:09.660
I've kinda droned on for a long time here,

527
00:22:09.660 --> 00:22:12.693
but let's talk about what costs are.

528
00:22:14.820 --> 00:22:15.930
So there's definitely,

529
00:22:15.930 --> 00:22:19.080
there's many different sources of costs

530
00:22:19.080 --> 00:22:21.390
from donors and government providers,

531
00:22:21.390 --> 00:22:23.070
international and local NGOs,

532
00:22:23.070 --> 00:22:25.773
participant costs from households and individuals.

533
00:22:26.670 --> 00:22:28.710
There's also many different ways to describe

534
00:22:28.710 --> 00:22:30.303
and categorize costs,

535
00:22:31.410 --> 00:22:32.607
say by the type of costs.

536
00:22:32.607 --> 00:22:35.360
Are they real world costs or are they costs

537
00:22:35.360 --> 00:22:37.440
of a randomized control trial?

538
00:22:37.440 --> 00:22:39.123
Are they full or incremental?

539
00:22:40.560 --> 00:22:42.930
Or by cost categories,

540
00:22:42.930 --> 00:22:44.970
so different types of inputs

541
00:22:44.970 --> 00:22:46.443
or organized by activities.

542
00:22:47.589 --> 00:22:50.243
Or, you know, as I mentioned even by the timing of costs.

543
00:22:51.810 --> 00:22:53.490
If you remember my first major takeaway

544
00:22:53.490 --> 00:22:54.750
from the presentation,

545
00:22:54.750 --> 00:22:56.160
you should be starting to get an idea

546
00:22:56.160 --> 00:22:56.993
of why I say

547
00:22:56.993 --> 00:22:59.763
that there was my second takeaway, rather,

548
00:23:00.780 --> 00:23:02.550
you start to get an idea here

549
00:23:02.550 --> 00:23:05.850
of the notion that expenditures exist,

550
00:23:05.850 --> 00:23:07.200
but that costs,

551
00:23:07.200 --> 00:23:08.760
they're a little bit fuzzier.

552
00:23:08.760 --> 00:23:10.380
They're frequently abstract

553
00:23:10.380 --> 00:23:12.680
and they depend on a lot of different factors.

554
00:23:13.950 --> 00:23:17.310
So two different kinds of costs to be able

555
00:23:17.310 --> 00:23:21.540
to distinguish between are economic and financial costs.

556
00:23:21.540 --> 00:23:25.230
So financial costs are those appearing potentially

557
00:23:25.230 --> 00:23:29.610
in the program expenditures documents.

558
00:23:29.610 --> 00:23:33.090
Meanwhile, economic costs are costs

559
00:23:33.090 --> 00:23:34.980
that are included or that include

560
00:23:34.980 --> 00:23:36.900
the value of all resources,

561
00:23:36.900 --> 00:23:38.160
regardless of who's paying.

562
00:23:38.160 --> 00:23:42.090
So the difference here is resources contributed

563
00:23:42.090 --> 00:23:42.990
by other entities.

564
00:23:42.990 --> 00:23:45.600
So for example, the local hospital,

565
00:23:45.600 --> 00:23:47.430
the clinical staff,

566
00:23:47.430 --> 00:23:50.610
volunteers or subsidized costs,

567
00:23:50.610 --> 00:23:55.233
such as from an international agency paying for commodities.

568
00:23:56.400 --> 00:24:01.200
So note that when you see an economic costing exercise

569
00:24:01.200 --> 00:24:03.270
referred to in the literature,

570
00:24:03.270 --> 00:24:06.365
it typically subsumes financial costing

571
00:24:06.365 --> 00:24:11.043
when it's, you know, thought of in this Venn diagram.

572
00:24:13.260 --> 00:24:15.720
So we have a sense of what costs are

573
00:24:15.720 --> 00:24:16.740
and why they're important,

574
00:24:16.740 --> 00:24:19.410
but what exactly is costing.

575
00:24:19.410 --> 00:24:22.397
So we use this term costing as a shorthand

576
00:24:22.397 --> 00:24:25.230
to describe the estimation of the costs

577
00:24:25.230 --> 00:24:29.220
of producing health services, for example.

578
00:24:29.220 --> 00:24:32.400
So costing places of value on the total resources used

579
00:24:32.400 --> 00:24:34.140
to produce a good or service.

580
00:24:34.140 --> 00:24:37.380
It requires measurement of the amount of each resource

581
00:24:37.380 --> 00:24:39.480
as well as information about the price

582
00:24:39.480 --> 00:24:41.640
of each of those resources.

583
00:24:41.640 --> 00:24:44.520
So costs can vary by context.

584
00:24:44.520 --> 00:24:47.670
So when one is costing,

585
00:24:47.670 --> 00:24:50.160
it's important to gather information about location,

586
00:24:50.160 --> 00:24:53.823
about time period, population, and a host of other factors.

587
00:24:54.990 --> 00:24:56.670
Some of the different types of resources

588
00:24:56.670 --> 00:24:59.250
that are used to produce health services

589
00:24:59.250 --> 00:25:02.640
might include human resources,

590
00:25:02.640 --> 00:25:04.470
drugs and supplies,

591
00:25:04.470 --> 00:25:07.710
medical or other non-medical equipment

592
00:25:07.710 --> 00:25:09.213
or even patient costs.

593
00:25:10.380 --> 00:25:12.393
So what exactly do we use costing for?

594
00:25:13.260 --> 00:25:14.220
Well, costing is used

595
00:25:14.220 --> 00:25:18.450
in a range of research evaluations, programs,

596
00:25:18.450 --> 00:25:19.950
and in planning,

597
00:25:19.950 --> 00:25:22.680
primarily it's in the areas of financial planning

598
00:25:22.680 --> 00:25:24.900
and budgeting and priority setting

599
00:25:24.900 --> 00:25:28.833
that cost data is the most sort of visible we'll say.

600
00:25:29.670 --> 00:25:30.503
So for example,

601
00:25:30.503 --> 00:25:33.960
cost data are used to evaluate the effectiveness

602
00:25:33.960 --> 00:25:36.150
of an intervention or to understand

603
00:25:36.150 --> 00:25:38.550
the efficiency of health delivery,

604
00:25:38.550 --> 00:25:40.000
but what are the cost drivers

605
00:25:40.890 --> 00:25:42.570
and how do these vary over time?

606
00:25:42.570 --> 00:25:44.160
So in in ART,

607
00:25:44.160 --> 00:25:46.983
which is the example in the slide here,

608
00:25:47.940 --> 00:25:50.700
this figure is from a systematic review

609
00:25:50.700 --> 00:25:52.163
of the costing literature.

610
00:25:52.163 --> 00:25:54.390
This is one that I'll touch on later

611
00:25:54.390 --> 00:25:55.920
for antiretroviral treatment

612
00:25:55.920 --> 00:25:57.670
in low and middle income countries,

613
00:25:58.590 --> 00:26:00.750
it shows the different drivers

614
00:26:00.750 --> 00:26:04.893
of the average unit costs across five different settings.

615
00:26:05.910 --> 00:26:07.470
And this helps to show

616
00:26:07.470 --> 00:26:11.940
which cost inputs contribute most to the overall unit cost.

617
00:26:11.940 --> 00:26:13.080
So this figure I think does

618
00:26:13.080 --> 00:26:16.260
a really great job of showing the absolute difference

619
00:26:16.260 --> 00:26:18.450
in unit costs between settings.

620
00:26:18.450 --> 00:26:20.340
And in this case it's it's by country

621
00:26:20.340 --> 00:26:22.140
and it includes a lot of different settings

622
00:26:22.140 --> 00:26:27.140
within countries and doing so over time.

623
00:26:28.260 --> 00:26:29.103
Oops, excuse me.

624
00:26:33.690 --> 00:26:35.850
So cost data are also used, you know,

625
00:26:35.850 --> 00:26:38.550
as I mentioned in efficiency analyses.

626
00:26:38.550 --> 00:26:39.383
So for example,

627
00:26:39.383 --> 00:26:42.570
this might be to examine how costs vary

628
00:26:42.570 --> 00:26:45.120
with different levels of service delivery.

629
00:26:45.120 --> 00:26:48.900
So understanding unit costs at different coverage levels

630
00:26:48.900 --> 00:26:51.960
is really important to predicting costs as programs tend

631
00:26:51.960 --> 00:26:53.580
to scale up or to grow.

632
00:26:53.580 --> 00:26:57.120
So scale is a really important component here.

633
00:26:57.120 --> 00:27:00.000
Now these dynamics can be explored

634
00:27:00.000 --> 00:27:01.830
through the use of cost functions.

635
00:27:01.830 --> 00:27:03.630
So here the figure shows

636
00:27:03.630 --> 00:27:06.180
this sort of interesting dynamic relationship

637
00:27:06.180 --> 00:27:10.230
between scale and unit cost for antiretroviral treatments

638
00:27:10.230 --> 00:27:11.670
across a large number of sites

639
00:27:11.670 --> 00:27:13.950
in sub-Saharan African countries.

640
00:27:13.950 --> 00:27:17.910
So you'll note that the relationship here isn't linear.

641
00:27:17.910 --> 00:27:19.380
It's a little noisy,

642
00:27:19.380 --> 00:27:21.990
but that makes it important to estimate cost functions

643
00:27:21.990 --> 00:27:24.120
and then to utilize different costs

644
00:27:24.120 --> 00:27:26.160
for different levels of coverage.

645
00:27:26.160 --> 00:27:28.350
So in this case we can see

646
00:27:28.350 --> 00:27:31.620
some suggestive evidence of economies of scale,

647
00:27:31.620 --> 00:27:36.060
which is to say that per patient ART services

648
00:27:36.060 --> 00:27:38.970
within facilities with larger numbers

649
00:27:38.970 --> 00:27:40.170
of annual patients tend

650
00:27:40.170 --> 00:27:42.450
to be slightly cheaper than facilities

651
00:27:42.450 --> 00:27:44.073
with fewer annual patients.

652
00:27:45.930 --> 00:27:48.210
So one thing to remember here

653
00:27:48.210 --> 00:27:50.130
is that although there's a wide range

654
00:27:50.130 --> 00:27:52.350
of different analyses that need cost data,

655
00:27:52.350 --> 00:27:57.350
they don't always use or need the same types of data.

656
00:27:57.690 --> 00:27:59.370
The purpose of the analysis

657
00:27:59.370 --> 00:28:01.440
in this case sort of dictates

658
00:28:01.440 --> 00:28:03.300
the type of cost that is required.

659
00:28:03.300 --> 00:28:04.780
And so it's really important

660
00:28:05.640 --> 00:28:08.163
that the purpose be identified at the outset.

661
00:28:09.510 --> 00:28:10.650
So, you know,

662
00:28:10.650 --> 00:28:13.950
we have to consider budgeting, forecasting,

663
00:28:13.950 --> 00:28:18.060
efficiency analysis, priority setting, among others.

664
00:28:20.641 --> 00:28:22.200
Move forward.

665
00:28:22.200 --> 00:28:24.630
So before we go into costs further,

666
00:28:24.630 --> 00:28:27.530
I think it's also important to get some terminology right.

667
00:28:28.710 --> 00:28:31.710
In addition to the types of costs,

668
00:28:31.710 --> 00:28:34.020
there's also different measures of cost.

669
00:28:34.020 --> 00:28:36.633
So being clear about what each means, I think,

670
00:28:37.470 --> 00:28:40.710
is pretty central to understanding

671
00:28:40.710 --> 00:28:44.340
and interpreting cost information better as a field.

672
00:28:44.340 --> 00:28:46.140
So here are four different types

673
00:28:46.140 --> 00:28:49.170
of costing terms that you're probably familiar with

674
00:28:49.170 --> 00:28:51.960
or, you know, you may have encountered some,

675
00:28:51.960 --> 00:28:53.370
but what do they mean?

676
00:28:53.370 --> 00:28:55.470
So the first is total cost,

677
00:28:55.470 --> 00:28:57.900
which is as you'd imagine the total cost

678
00:28:57.900 --> 00:28:59.073
of producing a service.

679
00:29:00.030 --> 00:29:03.330
And in healthcare this is often presented as an annual cost.

680
00:29:03.330 --> 00:29:07.237
For example, the total annual cost of VMMC services

681
00:29:07.237 --> 00:29:09.813
at a clinic in a particular setting.

682
00:29:11.040 --> 00:29:13.530
The second term is average unit cost.

683
00:29:13.530 --> 00:29:16.860
And so this is the total cost per unit of output.

684
00:29:16.860 --> 00:29:18.840
And you'll notice that the units of output

685
00:29:18.840 --> 00:29:20.540
can be measured in different ways,

686
00:29:21.480 --> 00:29:22.800
even within the same service.

687
00:29:22.800 --> 00:29:23.820
So for example,

688
00:29:23.820 --> 00:29:25.950
there's a cost per person contacted

689
00:29:25.950 --> 00:29:27.900
or a cost per person treated.

690
00:29:27.900 --> 00:29:30.540
And I'll return to this a little later.

691
00:29:30.540 --> 00:29:32.460
The third term is marginal cost,

692
00:29:32.460 --> 00:29:35.490
which is a concept that's frequently used in economics

693
00:29:35.490 --> 00:29:38.246
and it's really critical to efficiency analysis.

694
00:29:38.246 --> 00:29:40.020
And this is the additional cost of producing

695
00:29:40.020 --> 00:29:41.880
just one more unit of output.

696
00:29:41.880 --> 00:29:43.557
So for example,

697
00:29:43.557 --> 00:29:45.870
the cost of testing one more person

698
00:29:45.870 --> 00:29:48.033
or vaccinating one additional patient.

699
00:29:49.140 --> 00:29:52.530
And then the final term here is incremental cost,

700
00:29:52.530 --> 00:29:54.840
which also examines change.

701
00:29:54.840 --> 00:29:57.690
In this case, however, the focus is on the cost

702
00:29:57.690 --> 00:29:59.190
of adding a completely new level

703
00:29:59.190 --> 00:30:00.360
or type of service rather

704
00:30:00.360 --> 00:30:02.820
than adding one more unit of output

705
00:30:02.820 --> 00:30:04.320
to an existing service.

706
00:30:04.320 --> 00:30:05.190
So for example,

707
00:30:05.190 --> 00:30:08.070
the additional cost of adding counseling

708
00:30:08.070 --> 00:30:09.210
and linkage to care

709
00:30:09.210 --> 00:30:10.653
to current HIV testing.

710
00:30:12.180 --> 00:30:15.810
Most costing exercises though that you'll encounter tend to

711
00:30:15.810 --> 00:30:18.423
refer to either average or to unit costs.

712
00:30:21.009 --> 00:30:23.070
I wanna give you just a quick example here

713
00:30:23.070 --> 00:30:26.970
of what a program costing table might look like.

714
00:30:26.970 --> 00:30:30.960
So this slide comes from a presentation by Dr. Jim Kahn,

715
00:30:30.960 --> 00:30:35.940
who is a mentor of mine at Cal.

716
00:30:35.940 --> 00:30:40.860
The example here is a small rural HIV clinic

717
00:30:40.860 --> 00:30:45.860
that was set up by a colleague of Dr. Khan at UCSF,

718
00:30:45.990 --> 00:30:46.920
which I'll have to say

719
00:30:46.920 --> 00:30:51.030
is where he spent most of his career.

720
00:30:51.030 --> 00:30:54.720
And the two in this case conducted

721
00:30:54.720 --> 00:30:57.600
a costing analysis of this facility.

722
00:30:57.600 --> 00:30:58.433
So as you can see,

723
00:30:58.433 --> 00:31:00.000
the summary divides costs

724
00:31:00.000 --> 00:31:02.160
into four different categories.

725
00:31:02.160 --> 00:31:03.723
There's personnel at the top,

726
00:31:04.620 --> 00:31:06.210
antiretroviral medications,

727
00:31:06.210 --> 00:31:07.800
which are the dominant supply

728
00:31:07.800 --> 00:31:09.300
that were used by the program,

729
00:31:10.260 --> 00:31:14.040
lab tests, which is the main service that was provided

730
00:31:14.040 --> 00:31:15.300
by the program,

731
00:31:15.300 --> 00:31:18.420
and then there's this fourth sort of residual category

732
00:31:18.420 --> 00:31:20.190
of less expensive items.

733
00:31:20.190 --> 00:31:23.340
So administration, general supplies,

734
00:31:23.340 --> 00:31:25.890
vehicles, storage, utilities.

735
00:31:25.890 --> 00:31:27.440
You might find pencils in here.

736
00:31:28.740 --> 00:31:31.320
So the standard input categories here

737
00:31:31.320 --> 00:31:32.490
were adapted to conform

738
00:31:32.490 --> 00:31:35.460
to how clinicians think about major components

739
00:31:35.460 --> 00:31:36.963
of antiretroviral treatment.

740
00:31:37.800 --> 00:31:39.630
In this case, the columns show the units

741
00:31:39.630 --> 00:31:41.250
that were tested,

742
00:31:41.250 --> 00:31:44.040
the costs per unit,

743
00:31:44.040 --> 00:31:47.460
the resulting monthly cost for the entire activity

744
00:31:47.460 --> 00:31:50.340
and the entire costs per patient

745
00:31:50.340 --> 00:31:51.333
for each line item.

746
00:31:52.320 --> 00:31:53.700
So in this case, for example,

747
00:31:53.700 --> 00:31:58.290
the full-time doctor costs $1083 per month

748
00:31:58.290 --> 00:32:00.453
and $52 per patient year.

749
00:32:02.400 --> 00:32:04.110
So one interesting takeaway here

750
00:32:04.110 --> 00:32:06.030
is that the authors found

751
00:32:06.030 --> 00:32:09.660
that once the clinic had reached fully operational capacity,

752
00:32:09.660 --> 00:32:11.580
it was able to deliver an ART

753
00:32:11.580 --> 00:32:16.110
at it looks like about $550 per patient per year.

754
00:32:16.110 --> 00:32:18.330
And they also found, which isn't shown here,

755
00:32:18.330 --> 00:32:21.180
that the cost could drop by about a third

756
00:32:21.180 --> 00:32:23.580
with less expensive medications

757
00:32:23.580 --> 00:32:25.320
as well as lower wage scales

758
00:32:25.320 --> 00:32:27.990
and modest increases in patient load.

759
00:32:27.990 --> 00:32:29.490
So just an interesting example

760
00:32:30.750 --> 00:32:32.853
to kind of put some of this into context.

761
00:32:34.290 --> 00:32:36.810
A couple of key costing principles here.

762
00:32:36.810 --> 00:32:38.220
And these are major takeaways

763
00:32:38.220 --> 00:32:40.980
I think from Jim as well.

764
00:32:40.980 --> 00:32:44.610
But in theory you can do some really solid costing work

765
00:32:44.610 --> 00:32:48.420
if you adhere to really these three main principles.

766
00:32:48.420 --> 00:32:50.100
So the most important is that your goal

767
00:32:50.100 --> 00:32:52.440
is to quantify the resources required

768
00:32:52.440 --> 00:32:55.530
for program operation and the associated cost,

769
00:32:55.530 --> 00:32:59.880
and that means developing an inventory of resources

770
00:32:59.880 --> 00:33:01.590
and assigning appropriate costs.

771
00:33:01.590 --> 00:33:05.040
So this task is quite different

772
00:33:05.040 --> 00:33:07.080
than simply accepting a program's budget

773
00:33:07.080 --> 00:33:08.970
or their expenditure report

774
00:33:08.970 --> 00:33:10.650
because that report might include costs

775
00:33:10.650 --> 00:33:13.350
that aren't required for routine program operations,

776
00:33:13.350 --> 00:33:15.483
like special evaluation needs of a funder.

777
00:33:17.400 --> 00:33:19.980
Also, even if the costs are accurate,

778
00:33:19.980 --> 00:33:21.330
you've likely learned nothing

779
00:33:21.330 --> 00:33:23.160
about how the program operates.

780
00:33:23.160 --> 00:33:26.010
So specifically, what types of inputs might be available?

781
00:33:27.450 --> 00:33:29.220
The number two point, you know,

782
00:33:29.220 --> 00:33:33.390
that Jim makes often is that you should be systematic

783
00:33:33.390 --> 00:33:35.130
and thoughtfully thorough.

784
00:33:35.130 --> 00:33:37.950
And by systematic, you know, we mean

785
00:33:37.950 --> 00:33:39.630
that the search for resources

786
00:33:39.630 --> 00:33:42.240
should include all of the usual input categories

787
00:33:42.240 --> 00:33:43.803
like personnel and equipment,

788
00:33:44.790 --> 00:33:46.740
the range of activities, such as marketing

789
00:33:46.740 --> 00:33:47.793
and service delivery.

790
00:33:48.840 --> 00:33:52.620
But, by being thoughtfully thorough, you know,

791
00:33:52.620 --> 00:33:56.580
what Jim refers to is that all of the significant resources

792
00:33:56.580 --> 00:34:00.150
should be included, quantified and costed.

793
00:34:00.150 --> 00:34:01.800
But minor costs,

794
00:34:01.800 --> 00:34:05.220
truly minor things like routine stationary supplies

795
00:34:05.220 --> 00:34:09.570
or routine services, pencils and erasers,

796
00:34:09.570 --> 00:34:13.170
they can often be safely ignored if the data collection

797
00:34:13.170 --> 00:34:16.200
in order to obtain those is too onerous.

798
00:34:16.200 --> 00:34:18.780
So of course that's like a judgment call.

799
00:34:18.780 --> 00:34:21.120
And one suggestion, you know,

800
00:34:21.120 --> 00:34:24.240
that Jim makes is to use a table

801
00:34:24.240 --> 00:34:25.860
and kind of order resources

802
00:34:25.860 --> 00:34:27.510
from the most to the least expensive

803
00:34:27.510 --> 00:34:29.850
so that you can empirically identify,

804
00:34:29.850 --> 00:34:32.610
say a cutoff for each of the resources

805
00:34:32.610 --> 00:34:34.770
that are considered maybe too minor

806
00:34:34.770 --> 00:34:38.850
or that may not play into the overall costs

807
00:34:38.850 --> 00:34:40.620
as much as you might be concerned about.

808
00:34:40.620 --> 00:34:44.070
So, you know, this is really getting into fine, fine detail.

809
00:34:44.070 --> 00:34:46.270
Do you really need to cost that last pencil?

810
00:34:47.940 --> 00:34:50.700
And then finally, you know, as we mentioned before,

811
00:34:50.700 --> 00:34:52.740
adopt the appropriate perspective.

812
00:34:52.740 --> 00:34:56.103
So I've talked about, you know, this at some length,

813
00:34:58.320 --> 00:34:59.670
but it's worth reiterating.

814
00:35:03.750 --> 00:35:06.480
I'll go over, I think I might skip past some

815
00:35:06.480 --> 00:35:10.380
of the typical cost and categories except to say, you know,

816
00:35:10.380 --> 00:35:13.620
these are some examples they may often

817
00:35:13.620 --> 00:35:16.023
go as far as to include patient costs.

818
00:35:18.030 --> 00:35:23.030
They may also vary by program and over time.

819
00:35:26.100 --> 00:35:29.340
The timeframe, you know, which I mentioned earlier,

820
00:35:29.340 --> 00:35:32.220
typically when we think about a costing exercise,

821
00:35:32.220 --> 00:35:35.670
it's about 12 months or it's multiples thereof.

822
00:35:35.670 --> 00:35:37.170
And you know, the reason for this

823
00:35:37.170 --> 00:35:38.400
for annual costing

824
00:35:38.400 --> 00:35:41.940
is that it has the hallmark of, you know,

825
00:35:41.940 --> 00:35:45.000
conforming to our pervasive and programmatic record keeping,

826
00:35:45.000 --> 00:35:46.470
which is really nice,

827
00:35:46.470 --> 00:35:49.500
but it can also help to deal with distortions,

828
00:35:49.500 --> 00:35:51.420
seasonal effects and shocks.

829
00:35:51.420 --> 00:35:54.180
It's smooths as a lot of random variation.

830
00:35:54.180 --> 00:35:56.010
So that can be for like the calendar year

831
00:35:56.010 --> 00:35:57.510
or the fiscal year, typically.

832
00:35:58.620 --> 00:36:00.510
Portraying costs by month,

833
00:36:00.510 --> 00:36:01.343
if it's easier,

834
00:36:01.343 --> 00:36:04.050
might indicate how costs vary over the course of a year.

835
00:36:04.050 --> 00:36:05.650
So that's another consideration.

836
00:36:07.470 --> 00:36:08.730
And you know, again,

837
00:36:08.730 --> 00:36:12.000
if we're talking about fully operational programs,

838
00:36:12.000 --> 00:36:15.780
we wanna make sure that the program is actually matured,

839
00:36:15.780 --> 00:36:18.910
but again, that might ignore some of the startup costs

840
00:36:20.210 --> 00:36:22.353
or other costs that I discussed earlier.

841
00:36:23.340 --> 00:36:26.640
One other little piece of terminology that I want to cover

842
00:36:26.640 --> 00:36:29.640
is gross versus macro, and top down versus bottom up.

843
00:36:29.640 --> 00:36:32.010
So you might have come across this a number of times,

844
00:36:32.010 --> 00:36:33.990
you know, in proposals or mentioned

845
00:36:33.990 --> 00:36:35.740
sort of casually in the literature,

846
00:36:36.600 --> 00:36:39.840
micro-costing focuses on really granular accounting

847
00:36:39.840 --> 00:36:44.840
of inputs as opposed to to macro-costing,

848
00:36:45.420 --> 00:36:50.420
which is, or sorry, gross costing.

849
00:36:50.430 --> 00:36:51.780
I always get them confused.

850
00:36:52.852 --> 00:36:54.930
A gross costing approach that might simply estimate

851
00:36:54.930 --> 00:36:56.790
all of the relevant costs,

852
00:36:56.790 --> 00:36:59.250
typically from program expenditure data,

853
00:36:59.250 --> 00:37:02.070
and then dividing the associated outputs.

854
00:37:02.070 --> 00:37:04.083
So like patient episodes.

855
00:37:05.280 --> 00:37:08.250
In contrast, we've got the bottom up or top down,

856
00:37:08.250 --> 00:37:12.390
which refers to the way in which each resource is allocated

857
00:37:12.390 --> 00:37:14.280
to the total unit cost being estimated.

858
00:37:14.280 --> 00:37:17.370
So top down costing divides

859
00:37:17.370 --> 00:37:19.680
overall program costs or expenditures,

860
00:37:19.680 --> 00:37:21.780
often including those,

861
00:37:21.780 --> 00:37:26.780
which are above the service delivery level

862
00:37:26.820 --> 00:37:28.290
by the number of outputs in order

863
00:37:28.290 --> 00:37:29.943
to calculate the unit cost,

864
00:37:30.870 --> 00:37:33.180
whereas bottom up costing

865
00:37:33.180 --> 00:37:37.500
measures input quantities at the client or activity level.

866
00:37:37.500 --> 00:37:39.390
So gross costing, for example,

867
00:37:39.390 --> 00:37:42.183
is commonly done top down, for example.

868
00:37:44.640 --> 00:37:46.990
And then micro-costing kind of has a bottom up.

869
00:37:49.590 --> 00:37:50.760
I'll skip over.

870
00:37:50.760 --> 00:37:55.323
Well, let's see, Debbie, we're at 43 after.

871
00:37:57.300 --> 00:38:00.750
I wanna mention pricing and valuation very briefly

872
00:38:00.750 --> 00:38:04.830
before moving into some of our other findings

873
00:38:04.830 --> 00:38:09.300
simply because I think there's one example

874
00:38:09.300 --> 00:38:11.640
in the popular press right now

875
00:38:11.640 --> 00:38:13.170
that's really worth consideration,

876
00:38:13.170 --> 00:38:14.610
particularly if you're operating

877
00:38:14.610 --> 00:38:15.840
in low and middle income countries.

878
00:38:15.840 --> 00:38:20.840
And I'll just point to the fourth bullet here, inflation.

879
00:38:20.970 --> 00:38:22.780
So inflation comes up a lot

880
00:38:24.600 --> 00:38:25.650
in the popular press

881
00:38:25.650 --> 00:38:27.750
because we're experiencing rapid inflation

882
00:38:27.750 --> 00:38:28.983
in the US right now,

883
00:38:29.850 --> 00:38:33.960
but I wanna linger on it for a second, simply to say,

884
00:38:33.960 --> 00:38:38.960
so the operation that you use

885
00:38:40.020 --> 00:38:43.020
in order to account for inflation

886
00:38:43.020 --> 00:38:46.620
in different country settings matters a great deal.

887
00:38:46.620 --> 00:38:48.270
So the example here is,

888
00:38:48.270 --> 00:38:49.820
let's say that you've collected

889
00:38:50.915 --> 00:38:51.748
a piece of critical cost data

890
00:38:51.748 --> 00:38:53.610
from a program in Tanzania,

891
00:38:53.610 --> 00:38:55.433
the unit cost of a service

892
00:38:55.433 --> 00:38:58.353
in the year 2000 to say a $100.

893
00:38:59.400 --> 00:39:02.040
So, you know, the typical strategy

894
00:39:02.040 --> 00:39:07.040
is to use a US price inflation tool,

895
00:39:08.220 --> 00:39:10.720
so typically the GDP price deflator

896
00:39:11.700 --> 00:39:15.300
and convert that into US dollars

897
00:39:15.300 --> 00:39:17.820
in order to inflate over time.

898
00:39:17.820 --> 00:39:21.390
There's some recommendations in the literature

899
00:39:21.390 --> 00:39:22.223
that this should be done

900
00:39:22.223 --> 00:39:24.150
in a different order of operations,

901
00:39:24.150 --> 00:39:26.420
and that would be to actually inflate

902
00:39:26.420 --> 00:39:30.720
in local inflation before converting into US dollars.

903
00:39:30.720 --> 00:39:32.010
And the implications for that.

904
00:39:32.010 --> 00:39:35.700
So in the next slide here are pretty substantial.

905
00:39:35.700 --> 00:39:38.850
So, you know, in Tanzania over that same period,

906
00:39:38.850 --> 00:39:42.630
inflation was quite rampant and so it it really changes

907
00:39:42.630 --> 00:39:45.900
the overall dollar amount

908
00:39:45.900 --> 00:39:46.860
that you're looking at

909
00:39:46.860 --> 00:39:49.770
in the last period of the cycle there.

910
00:39:49.770 --> 00:39:51.720
And it's even more volatile

911
00:39:51.720 --> 00:39:53.550
when you include other countries

912
00:39:53.550 --> 00:39:55.653
in sub-Saharan Africa, period.

913
00:39:57.180 --> 00:40:02.180
So, your choice of inflating first before exchanging

914
00:40:02.340 --> 00:40:04.830
into US dollars or exchanging into US dollars

915
00:40:04.830 --> 00:40:07.800
and then using a US inflation process

916
00:40:07.800 --> 00:40:10.100
can have some really substantial implications.

917
00:40:11.970 --> 00:40:13.350
I just wanted to highlight that.

918
00:40:13.350 --> 00:40:18.350
I'm gonna push through some of these final notes here

919
00:40:18.780 --> 00:40:23.010
because I'd like to get to some of our findings.

920
00:40:23.010 --> 00:40:26.040
I do wanna leave you with three key resources at the end,

921
00:40:26.040 --> 00:40:28.950
and those are the consolidated health economics

922
00:40:28.950 --> 00:40:30.450
evaluation reporting standards,

923
00:40:30.450 --> 00:40:31.320
or CHEERS standards.

924
00:40:31.320 --> 00:40:33.180
There's a new update to those standards

925
00:40:33.180 --> 00:40:35.370
that was published in 2022.

926
00:40:35.370 --> 00:40:36.570
And then for those working

927
00:40:36.570 --> 00:40:39.060
in low and middle income country settings,

928
00:40:39.060 --> 00:40:42.090
Vasal and colleagues reference case

929
00:40:42.090 --> 00:40:44.064
for estimating the cost

930
00:40:44.064 --> 00:40:45.630
of global health services and interventions

931
00:40:45.630 --> 00:40:49.110
and the Bill and Melinda Gates Foundation 2014 guidelines

932
00:40:49.110 --> 00:40:51.570
that come through their reference case.

933
00:40:51.570 --> 00:40:54.900
And I'm happy to share those resources after the talk,

934
00:40:54.900 --> 00:40:56.610
but let me get into costing and practice here

935
00:40:56.610 --> 00:40:59.250
because I am running quite low on time.

936
00:40:59.250 --> 00:41:01.860
So the Global Health Cost Consortium,

937
00:41:01.860 --> 00:41:03.230
as Debbie mentioned at the top

938
00:41:03.230 --> 00:41:04.920
of the hour was funded

939
00:41:04.920 --> 00:41:06.970
by the Bill and Melinda Gates Foundation.

940
00:41:08.550 --> 00:41:11.670
I was involved with this project for about four years,

941
00:41:11.670 --> 00:41:15.600
and two of the outputs from the project

942
00:41:15.600 --> 00:41:18.450
were our unit cost study repository,

943
00:41:18.450 --> 00:41:21.060
which I'd invite you to visit.

944
00:41:21.060 --> 00:41:24.360
And that's really a one-stop shop for all

945
00:41:24.360 --> 00:41:25.770
of the published cost data

946
00:41:25.770 --> 00:41:27.690
from HIV and TB interventions

947
00:41:27.690 --> 00:41:28.523
that were conducted

948
00:41:28.523 --> 00:41:30.390
in low and middle income countries settings,

949
00:41:30.390 --> 00:41:32.223
really over the last 20 years or so.

950
00:41:33.150 --> 00:41:36.450
So we went through a pretty long painstaking process

951
00:41:36.450 --> 00:41:39.033
of extracting primary cost data,

952
00:41:39.033 --> 00:41:41.710
that is costs that were not modeled

953
00:41:42.840 --> 00:41:44.700
from the published literature,

954
00:41:44.700 --> 00:41:47.670
standardizing them across

955
00:41:47.670 --> 00:41:51.150
several hundred different variables,

956
00:41:51.150 --> 00:41:55.560
and then making them sort of available to the public

957
00:41:55.560 --> 00:41:57.630
at large in a pretty easy

958
00:41:57.630 --> 00:42:01.050
to use dropdown type of database.

959
00:42:01.050 --> 00:42:04.650
So if you're ever doing any program planning

960
00:42:04.650 --> 00:42:06.570
for HIV or TB interventions,

961
00:42:06.570 --> 00:42:09.630
I'd invite you to visit that site

962
00:42:09.630 --> 00:42:11.400
and see if you can get any utility

963
00:42:11.400 --> 00:42:13.740
from some of the efforts that we put forward.

964
00:42:13.740 --> 00:42:15.840
Another big output was the reference case

965
00:42:15.840 --> 00:42:17.700
that I mentioned before,

966
00:42:17.700 --> 00:42:19.439
guidelines for estimating costs

967
00:42:19.439 --> 00:42:21.333
for Global Health Interventions.

968
00:42:22.170 --> 00:42:24.520
And again, happy to share those after the talk.

969
00:42:26.100 --> 00:42:28.740
Another major output here was we were able

970
00:42:28.740 --> 00:42:33.120
to produce several different publications.

971
00:42:33.120 --> 00:42:36.330
So I'll present a little bit of the findings from that.

972
00:42:36.330 --> 00:42:38.129
So over several years, our team

973
00:42:38.129 --> 00:42:39.900
conducted the systematic search

974
00:42:39.900 --> 00:42:43.200
and screening of peer reviewed and grey literature.

975
00:42:43.200 --> 00:42:45.660
We had a few different goals here.

976
00:42:45.660 --> 00:42:48.390
We wanted to describe the quantity and characteristics

977
00:42:48.390 --> 00:42:50.820
of public published costing data

978
00:42:50.820 --> 00:42:54.270
to identify production patterns over time,

979
00:42:54.270 --> 00:42:55.770
so geographic location,

980
00:42:55.770 --> 00:42:58.260
publication venues like journals

981
00:42:58.260 --> 00:43:01.830
and authorship and to look into key methods

982
00:43:01.830 --> 00:43:04.560
and reporting standards to try to identify gaps

983
00:43:04.560 --> 00:43:06.540
in the costing literature.

984
00:43:06.540 --> 00:43:09.150
So I'll go over some of those findings.

985
00:43:09.150 --> 00:43:12.210
So figure one here shows the results

986
00:43:12.210 --> 00:43:15.180
of that search and screening process.

987
00:43:15.180 --> 00:43:17.700
So we covered articles that were published starting

988
00:43:17.700 --> 00:43:22.700
in January, 2006 through October, 2017.

989
00:43:22.710 --> 00:43:24.273
So it's a couple years old now.

990
00:43:25.770 --> 00:43:27.270
We use six databases,

991
00:43:27.270 --> 00:43:30.243
so PubMed, M Base, Web Science, Cochrane,

992
00:43:32.454 --> 00:43:34.350
and a number of others.

993
00:43:34.350 --> 00:43:36.930
I'm gonna give you high level content here.

994
00:43:36.930 --> 00:43:39.600
So searching also included grey literature,

995
00:43:39.600 --> 00:43:41.700
which you see on the right hand side.

996
00:43:41.700 --> 00:43:45.540
And we identified over 23,000 possible titles,

997
00:43:45.540 --> 00:43:48.360
including nearly 500 known costing studies

998
00:43:48.360 --> 00:43:49.770
that we had started with.

999
00:43:49.770 --> 00:43:51.600
So from these we were able to screen down

1000
00:43:51.600 --> 00:43:53.824
to 217 relevant titles

1001
00:43:53.824 --> 00:43:58.140
that got included in that unit cost study repository,

1002
00:43:58.140 --> 00:44:00.810
that database that I mentioned before.

1003
00:44:00.810 --> 00:44:05.673
And of those we published one study among 159

1004
00:44:10.560 --> 00:44:12.060
published studies that had taken place

1005
00:44:12.060 --> 00:44:13.203
in Sub-Saharan Africa.

1006
00:44:14.070 --> 00:44:17.610
So I'm gonna discuss really briefly two different studies

1007
00:44:17.610 --> 00:44:18.660
that came out of the effort.

1008
00:44:18.660 --> 00:44:22.230
And the first was that 2019 study that I just described.

1009
00:44:22.230 --> 00:44:24.060
And the second is a study that's in progress

1010
00:44:24.060 --> 00:44:26.340
where we're reviewing findings

1011
00:44:26.340 --> 00:44:28.020
from the detailed extraction

1012
00:44:28.020 --> 00:44:31.000
of the 243 studies that we started with

1013
00:44:32.010 --> 00:44:33.750
in the lines above.

1014
00:44:33.750 --> 00:44:35.820
So I might jump a little bit between findings

1015
00:44:35.820 --> 00:44:37.020
in these two different venues.

1016
00:44:37.020 --> 00:44:40.920
So the first study focused on that evidence

1017
00:44:40.920 --> 00:44:42.450
from Sub-Saharan Africa.

1018
00:44:42.450 --> 00:44:43.830
This gave us a really nice picture

1019
00:44:43.830 --> 00:44:46.830
of where costing data is and isn't collected

1020
00:44:46.830 --> 00:44:48.213
on the continent.

1021
00:44:49.050 --> 00:44:51.840
The studies we identified took place in 25 countries,

1022
00:44:51.840 --> 00:44:55.170
so mostly East Africa and Southern Africa.

1023
00:44:55.170 --> 00:44:57.870
And although this sort of geographic variation

1024
00:44:57.870 --> 00:44:59.400
is broad, you know,

1025
00:44:59.400 --> 00:45:00.990
we found that the quality of evidence

1026
00:45:00.990 --> 00:45:03.357
within many of these countries is,

1027
00:45:03.357 --> 00:45:05.160
or the quantity rather of evidence within these countries

1028
00:45:05.160 --> 00:45:06.690
is quite sparse.

1029
00:45:06.690 --> 00:45:10.200
So west Africa doesn't have

1030
00:45:10.200 --> 00:45:12.535
a great deal of primary cost data

1031
00:45:12.535 --> 00:45:14.190
in the published literature.

1032
00:45:14.190 --> 00:45:16.503
A lot of it, you know, is often modeled.

1033
00:45:18.720 --> 00:45:20.970
In terms of who is publishing studies

1034
00:45:20.970 --> 00:45:23.970
that report on HIV intervention costs.

1035
00:45:23.970 --> 00:45:25.890
There's a few notable peer-reviewed journals.

1036
00:45:25.890 --> 00:45:30.843
So PLoS One, J AIDS and AIDS are kind of our leaders.

1037
00:45:33.360 --> 00:45:34.710
But I think the big takeaway here

1038
00:45:34.710 --> 00:45:36.000
is that there's an awful lot

1039
00:45:36.000 --> 00:45:38.010
of potential destinations

1040
00:45:38.010 --> 00:45:41.050
for published primary cost data

1041
00:45:42.000 --> 00:45:45.000
and more so than I think we realized

1042
00:45:45.000 --> 00:45:47.760
when we set out in this exercise.

1043
00:45:47.760 --> 00:45:50.430
So you can see the large list

1044
00:45:50.430 --> 00:45:53.010
of different published studies on the left

1045
00:45:53.010 --> 00:45:55.440
or sorry journal destinations

1046
00:45:55.440 --> 00:45:57.090
for published studies on the left

1047
00:45:57.930 --> 00:45:59.280
and the table in the right.

1048
00:46:00.210 --> 00:46:01.820
These are grey literature resources.

1049
00:46:01.820 --> 00:46:05.340
So these are the sources of grey lit,

1050
00:46:05.340 --> 00:46:08.880
a few key players here, USAID, PEPFAR,

1051
00:46:08.880 --> 00:46:11.970
the Futures Group, Health Policy Project

1052
00:46:11.970 --> 00:46:15.240
and PANGEA and Bill and Melinda Gates

1053
00:46:15.240 --> 00:46:16.410
had their fingers

1054
00:46:16.410 --> 00:46:19.080
in a lot of these different published studies.

1055
00:46:19.080 --> 00:46:21.123
But these are the destinations.

1056
00:46:22.710 --> 00:46:24.570
One thing that I do wanna mention here

1057
00:46:24.570 --> 00:46:26.220
is that although we're examining

1058
00:46:26.220 --> 00:46:29.340
this subset of primary cost data that was collected

1059
00:46:29.340 --> 00:46:31.860
from studies in Sub-Saharan Africa,

1060
00:46:31.860 --> 00:46:33.270
there's a little bit of a disconnect here

1061
00:46:33.270 --> 00:46:37.050
in terms of where we find that primary cost evidence

1062
00:46:37.050 --> 00:46:39.690
is being published and where leaders in the field,

1063
00:46:39.690 --> 00:46:41.954
at least nine years ago

1064
00:46:41.954 --> 00:46:45.003
when the first CHEERS guidelines paper was published,

1065
00:46:46.020 --> 00:46:47.760
that there's, you know,

1066
00:46:47.760 --> 00:46:52.760
in that paper there was an explicit effort

1067
00:46:52.920 --> 00:46:56.820
to identify leading journals as landing places

1068
00:46:56.820 --> 00:46:59.190
for health economic evaluations.

1069
00:46:59.190 --> 00:47:02.520
But what we find is that of the 10 that they identified

1070
00:47:02.520 --> 00:47:04.830
in that CHEERS statement paper,

1071
00:47:04.830 --> 00:47:07.680
only three were we able to find

1072
00:47:07.680 --> 00:47:10.893
in our larger systematic search of the HIV literature.

1073
00:47:11.760 --> 00:47:14.220
And this is, you know, beyond Sub-Saharan Africa.

1074
00:47:14.220 --> 00:47:16.620
This is that paper too that I mentioned.

1075
00:47:16.620 --> 00:47:19.740
So this isn't necessarily an indictment of the list

1076
00:47:19.740 --> 00:47:23.160
that was generated by the CHEERS authors,

1077
00:47:23.160 --> 00:47:24.240
but it does highlight,

1078
00:47:24.240 --> 00:47:27.270
I think a large and varied number of journal sources

1079
00:47:27.270 --> 00:47:28.950
where these data are published.

1080
00:47:28.950 --> 00:47:30.480
And it emphasizes the need to be sure

1081
00:47:30.480 --> 00:47:32.040
that these standards, you know,

1082
00:47:32.040 --> 00:47:35.113
like the ones laid out by the CHEERS document are being

1083
00:47:35.113 --> 00:47:38.373
upheld across different publications.

1084
00:47:39.420 --> 00:47:43.680
So, you know, a large growth in publications over time,

1085
00:47:43.680 --> 00:47:46.170
a difference in the time to publication.

1086
00:47:46.170 --> 00:47:51.150
So peer reviewed studies tended to take 2.8 years on average

1087
00:47:51.150 --> 00:47:52.590
to get published,

1088
00:47:52.590 --> 00:47:55.980
whereas those in the grey literature 1.4 years.

1089
00:47:55.980 --> 00:47:59.710
So this is much more quickly disseminated

1090
00:48:00.900 --> 00:48:03.540
into the zeitgeist, you know,

1091
00:48:03.540 --> 00:48:06.150
the ability for folks to actually start

1092
00:48:06.150 --> 00:48:07.410
to use these data

1093
00:48:07.410 --> 00:48:11.013
for their own program planning purposes.

1094
00:48:12.900 --> 00:48:14.250
Scale, I'll mention.

1095
00:48:14.250 --> 00:48:16.680
This is probably our most substantial finding

1096
00:48:16.680 --> 00:48:17.580
and you don't need to spend

1097
00:48:17.580 --> 00:48:19.230
a lot of time pouring over this table,

1098
00:48:19.230 --> 00:48:21.030
but this was actually quite damning.

1099
00:48:22.440 --> 00:48:23.970
If you don't take anything else away

1100
00:48:23.970 --> 00:48:24.990
from this presentation,

1101
00:48:24.990 --> 00:48:28.323
I think, I hope that that this resonates,

1102
00:48:29.310 --> 00:48:30.840
that reporting of scale

1103
00:48:30.840 --> 00:48:34.080
is, you know, it's a really critical, important component

1104
00:48:34.080 --> 00:48:36.180
of any costing exercise.

1105
00:48:36.180 --> 00:48:40.020
But what we found is that in, you know,

1106
00:48:40.020 --> 00:48:43.620
the six of the intervention types of all the studies

1107
00:48:43.620 --> 00:48:46.440
through which we were able to collect data,

1108
00:48:46.440 --> 00:48:48.610
the average number of unit costs

1109
00:48:50.190 --> 00:48:52.623
in those published studies,

1110
00:48:54.060 --> 00:48:57.243
scale was reported below 50% of the time.

1111
00:48:58.230 --> 00:49:03.230
And in some areas as low as 35 or 26% of the time.

1112
00:49:04.320 --> 00:49:06.303
So in total, I think the average, you know,

1113
00:49:06.303 --> 00:49:07.530
that we found was something

1114
00:49:07.530 --> 00:49:10.800
like 44% of unit costs have

1115
00:49:10.800 --> 00:49:13.865
explicit measurement of scale associated

1116
00:49:13.865 --> 00:49:16.020
with those reports.

1117
00:49:16.020 --> 00:49:18.540
So that's a problem,

1118
00:49:18.540 --> 00:49:20.723
that's something that needs to improve in the field.

1119
00:49:21.780 --> 00:49:24.090
And then I'll also mention perspectives.

1120
00:49:24.090 --> 00:49:28.987
So despite I think recommendations and urging from say,

1121
00:49:30.000 --> 00:49:31.890
the Gates Foundation's reference case,

1122
00:49:31.890 --> 00:49:33.453
which I cited earlier,

1123
00:49:34.290 --> 00:49:36.150
which emphasizes that authors should focus

1124
00:49:36.150 --> 00:49:38.850
on societal perspectives as much as possible,

1125
00:49:38.850 --> 00:49:40.710
we actually find here in terms

1126
00:49:40.710 --> 00:49:45.710
of what folks in these articles report

1127
00:49:45.750 --> 00:49:47.790
as their given perspective

1128
00:49:47.790 --> 00:49:51.210
and when we clean it up and we, you know,

1129
00:49:51.210 --> 00:49:53.280
we figure out among those who didn't report

1130
00:49:53.280 --> 00:49:54.630
what their perspective was, by the way,

1131
00:49:54.630 --> 00:49:57.840
15.6% of of studies

1132
00:49:57.840 --> 00:50:00.180
that we uncover don't report a perspective,

1133
00:50:00.180 --> 00:50:03.003
which is again, a problem for reporting standards,

1134
00:50:04.680 --> 00:50:08.190
only seven or about 2.9% of the studies

1135
00:50:08.190 --> 00:50:11.610
that we identified take a societal perspective.

1136
00:50:11.610 --> 00:50:13.320
So, you know, most of this is done

1137
00:50:13.320 --> 00:50:15.360
at the level of the provider.

1138
00:50:15.360 --> 00:50:17.310
So, you know, it gives you a little bit of context,

1139
00:50:17.310 --> 00:50:19.983
I think for where there's room for growth.

1140
00:50:21.420 --> 00:50:24.090
So in conclusion, on these points, you know,

1141
00:50:24.090 --> 00:50:27.330
there's, I wanna say that there's, you know,

1142
00:50:27.330 --> 00:50:28.833
a few final recommendations.

1143
00:50:30.570 --> 00:50:32.280
We recommend that that future costing

1144
00:50:32.280 --> 00:50:35.430
and cost effectiveness studies closely follow

1145
00:50:35.430 --> 00:50:38.133
the Gates and the GHCC reference case.

1146
00:50:40.186 --> 00:50:42.090
And you know,

1147
00:50:42.090 --> 00:50:44.850
that there needs to be more detailed reporting

1148
00:50:44.850 --> 00:50:47.490
on a number of different items

1149
00:50:47.490 --> 00:50:49.680
that I wasn't able to touch on in this talk.

1150
00:50:49.680 --> 00:50:51.120
But, you know, most of all,

1151
00:50:51.120 --> 00:50:53.580
I think scale is something that's really missing

1152
00:50:53.580 --> 00:50:54.480
in the literature.

1153
00:50:55.650 --> 00:50:59.790
So I won't touch on this too much

1154
00:50:59.790 --> 00:51:01.440
other than to say there's some future research

1155
00:51:01.440 --> 00:51:02.580
that we're working on

1156
00:51:02.580 --> 00:51:05.250
to determine the quality of some of these cost data.

1157
00:51:05.250 --> 00:51:09.660
So be on the lookout for publication on that in the future.

1158
00:51:09.660 --> 00:51:13.337
And now that I've ran through all of our Q and A time,

1159
00:51:13.337 --> 00:51:14.790
I do wanna thank you for attending

1160
00:51:14.790 --> 00:51:16.590
and I'm happy to take any questions.

1161
00:51:20.490 --> 00:51:21.323
<v ->Yeah.</v>

1162
00:51:22.530 --> 00:51:24.210
Thank you for the great talk.

1163
00:51:24.210 --> 00:51:25.290
So I had a question,

1164
00:51:25.290 --> 00:51:28.170
comment about, you know, the societal perspective.

1165
00:51:28.170 --> 00:51:31.380
I think in certain scenarios or situation,

1166
00:51:31.380 --> 00:51:32.850
this could be a little bit tricky.

1167
00:51:32.850 --> 00:51:34.860
For example, if you take vaccine

1168
00:51:34.860 --> 00:51:38.133
for some infectious disease in low income countries,

1169
00:51:39.030 --> 00:51:41.670
usually the price is heavily negotiated

1170
00:51:41.670 --> 00:51:46.527
between the pharmaceutical companies and donors and NGOs.

1171
00:51:46.527 --> 00:51:49.170
And if you take the societal perspective,

1172
00:51:49.170 --> 00:51:52.050
when you are doing the cost effectiveness analysis,

1173
00:51:52.050 --> 00:51:55.050
including, you know, the productivity loss,

1174
00:51:55.050 --> 00:51:59.820
the income loss, the vaccine becomes extremely,

1175
00:51:59.820 --> 00:52:02.610
you know, cost effective or even cost saving.

1176
00:52:02.610 --> 00:52:04.710
So even if they price the vaccine

1177
00:52:04.710 --> 00:52:06.960
to $50, $100,

1178
00:52:06.960 --> 00:52:10.080
still it becomes cost saving, which is, you know,

1179
00:52:10.080 --> 00:52:12.720
it becomes an excuse in hands of, you know,

1180
00:52:13.710 --> 00:52:16.680
the pharmaceutical companies to jump up the price of vaccine

1181
00:52:16.680 --> 00:52:19.140
because such a cost effective strategy.

1182
00:52:19.140 --> 00:52:20.910
So I dunno if you have thought about that.

1183
00:52:20.910 --> 00:52:23.700
I usually, I have purposefully avoided

1184
00:52:23.700 --> 00:52:27.270
taking societal perspective in those contexts

1185
00:52:27.270 --> 00:52:29.820
because I just don't want to put these numbers out,

1186
00:52:29.820 --> 00:52:34.050
but I don't know if you have comments, thoughts.

1187
00:52:34.050 --> 00:52:36.540
<v ->Yeah, no, I mean that's a great point.</v>

1188
00:52:36.540 --> 00:52:39.720
I often don't think far enough along the causal chain

1189
00:52:39.720 --> 00:52:43.470
to consider sort of the potential negative externalities

1190
00:52:43.470 --> 00:52:48.470
of price gouging, you know, in those contexts.

1191
00:52:49.020 --> 00:52:51.960
I guess I'll say, you know,

1192
00:52:51.960 --> 00:52:54.930
at least in the context of what we found, you know,

1193
00:52:54.930 --> 00:52:57.150
in our search of the literature and you know,

1194
00:52:57.150 --> 00:52:59.190
there were a few of those different intervention types

1195
00:52:59.190 --> 00:53:01.650
that dealt with pharmaceuticals,

1196
00:53:01.650 --> 00:53:03.900
you know, and the costs of those being, you know,

1197
00:53:03.900 --> 00:53:07.740
not insignificant drivers of overall program

1198
00:53:07.740 --> 00:53:10.600
and unit costs, you know, for HIV care

1199
00:53:11.940 --> 00:53:15.853
that most of the studies that we identified,

1200
00:53:20.100 --> 00:53:22.140
you know, really do focus in, you know,

1201
00:53:22.140 --> 00:53:24.660
don't tend to focus in on the societal level.

1202
00:53:24.660 --> 00:53:26.670
So it's not, you know,

1203
00:53:26.670 --> 00:53:29.580
perhaps that that type of crisis

1204
00:53:29.580 --> 00:53:31.380
hasn't borne out necessarily

1205
00:53:31.380 --> 00:53:33.510
as a result of some of this in the literature,

1206
00:53:33.510 --> 00:53:35.708
but it's definitely not something that's discussed

1207
00:53:35.708 --> 00:53:37.290
as often as I think it should be.

1208
00:53:37.290 --> 00:53:39.027
So you raise an excellent point and, you know,

1209
00:53:39.027 --> 00:53:40.770
and one that I hope, you know,

1210
00:53:40.770 --> 00:53:43.167
gets added to further discussions

1211
00:53:43.167 --> 00:53:46.833
in published literature and forums like this.

1212
00:53:48.629 --> 00:53:49.462
So I don't have a good, you know,

1213
00:53:49.462 --> 00:53:54.270
I don't have a good anecdotal example to share,

1214
00:53:54.270 --> 00:53:56.370
but I really, yeah, I like the point.

1215
00:53:56.370 --> 00:53:58.020
Yeah. Jeremy.

1216
00:53:58.020 --> 00:53:59.100
<v ->Yeah, Drew, thanks so much.</v>

1217
00:53:59.100 --> 00:54:01.053
This was really informative.

1218
00:54:01.950 --> 00:54:03.390
I actually wonder maybe building

1219
00:54:03.390 --> 00:54:05.040
on the previous question,

1220
00:54:05.040 --> 00:54:08.430
if you could go back to the slide with the table

1221
00:54:08.430 --> 00:54:10.170
from that rural Uganda study,

1222
00:54:10.170 --> 00:54:14.790
because my question was about sort of was this a financial

1223
00:54:14.790 --> 00:54:17.730
or economic perspective that was being presented

1224
00:54:17.730 --> 00:54:20.070
and I was honing in on the ART

1225
00:54:20.070 --> 00:54:21.660
because presumably, you know,

1226
00:54:21.660 --> 00:54:25.290
through PEPFAR from the programs perspective,

1227
00:54:25.290 --> 00:54:28.516
ART should have been without cost

1228
00:54:28.516 --> 00:54:31.230
and also without cost to the patient.

1229
00:54:31.230 --> 00:54:32.700
So then, so which made me think,

1230
00:54:32.700 --> 00:54:35.790
well maybe this is an economic perspective,

1231
00:54:35.790 --> 00:54:36.750
but then there are other,

1232
00:54:36.750 --> 00:54:39.300
all those other pieces of deliver of getting the medicine

1233
00:54:39.300 --> 00:54:40.133
to the patient,

1234
00:54:40.133 --> 00:54:41.790
the whole supply chain and whatnot.

1235
00:54:41.790 --> 00:54:44.490
And so I was just wondering if you have a sense of like,

1236
00:54:44.490 --> 00:54:47.350
was this trying to describe programmatic

1237
00:54:49.398 --> 00:54:51.461
or economic perspectives

1238
00:54:51.461 --> 00:54:53.940
and, you know, when, if this is economic,

1239
00:54:53.940 --> 00:54:57.360
should we truly be thinking about like that whole process?

1240
00:54:57.360 --> 00:55:01.260
Like where in the supply chain do you start counting?

1241
00:55:01.260 --> 00:55:03.660
<v ->Yeah, you know, I wasn't involved</v>

1242
00:55:03.660 --> 00:55:04.710
in this particular study

1243
00:55:04.710 --> 00:55:07.470
and so I'm happy to field that question to Jim

1244
00:55:07.470 --> 00:55:09.860
'cause he uses this example in his talks.

1245
00:55:09.860 --> 00:55:13.350
My sense is that this is programmatic, you know.

1246
00:55:13.350 --> 00:55:15.720
It does seem fairly narrow in scope,

1247
00:55:15.720 --> 00:55:17.190
but you do, you raise a really good point

1248
00:55:17.190 --> 00:55:19.440
I think, you know, that Reza raised as well,

1249
00:55:19.440 --> 00:55:21.210
which is, you know,

1250
00:55:21.210 --> 00:55:23.670
it depends on who's providing that,

1251
00:55:23.670 --> 00:55:25.520
where the supply chain is coming from

1252
00:55:26.634 --> 00:55:27.720
and you know, and that obviously

1253
00:55:27.720 --> 00:55:29.070
has downstream implications

1254
00:55:29.070 --> 00:55:34.070
for, you know, how you deal with the costing procedures

1255
00:55:34.860 --> 00:55:37.593
in, you know, these different income country settings,

1256
00:55:39.120 --> 00:55:42.330
how you'd inflate and how you'd adjust.

1257
00:55:42.330 --> 00:55:46.620
And so supply chains don't often make it into the discussion

1258
00:55:46.620 --> 00:55:49.170
in these write-ups, but they need to, right?

1259
00:55:49.170 --> 00:55:51.320
Because that's a really critical component.

1260
00:55:52.920 --> 00:55:55.530
I don't know if that answers your underlying question.

1261
00:55:55.530 --> 00:55:58.830
<v ->Yeah, I think it gets to it for sure.</v>

1262
00:55:58.830 --> 00:55:59.663
<v ->Okay.</v>

1263
00:55:59.663 --> 00:56:00.496
Yeah.

1264
00:56:00.496 --> 00:56:01.329
And you know, like I said,

1265
00:56:01.329 --> 00:56:02.580
I'm happy to reach out and and find out

1266
00:56:02.580 --> 00:56:05.040
a little this particular example,

1267
00:56:05.040 --> 00:56:06.270
but it's question.

1268
00:56:06.270 --> 00:56:07.103
<v ->Great, thanks.</v>

1269
00:56:08.124 --> 00:56:09.360
<v ->Yeah.</v>

1270
00:56:09.360 --> 00:56:11.100
I do realize we're at the top of the hour.

1271
00:56:11.100 --> 00:56:14.520
I'm not able to see the chat very well.

1272
00:56:14.520 --> 00:56:17.460
So if there's anyone that would like to raise anything

1273
00:56:17.460 --> 00:56:18.870
from there that I'm missing,

1274
00:56:18.870 --> 00:56:21.000
I'm happy to field those.

1275
00:56:21.000 --> 00:56:22.106
<v ->Drew.</v>

1276
00:56:22.106 --> 00:56:25.860
I would just say that before David Palteo ran away,

1277
00:56:25.860 --> 00:56:28.590
he said that for what it's worth,

1278
00:56:28.590 --> 00:56:32.040
his concern is the mirror opposite of Reza's.

1279
00:56:32.040 --> 00:56:34.740
When we abandon the societal perspective,

1280
00:56:34.740 --> 00:56:37.800
interventions that have broad public benefit

1281
00:56:37.800 --> 00:56:39.570
can be greatly undervalued

1282
00:56:39.570 --> 00:56:41.490
because they're not germane to the narrow

1283
00:56:41.490 --> 00:56:45.330
decision maker and real economic resource use

1284
00:56:45.330 --> 00:56:46.920
such as volunteer time and effort

1285
00:56:46.920 --> 00:56:49.350
can be ignored because it looks free

1286
00:56:49.350 --> 00:56:51.060
to the narrow stakeholder.

1287
00:56:51.060 --> 00:56:53.940
So he's, and I would also say

1288
00:56:53.940 --> 00:56:55.077
that my question,

1289
00:56:55.077 --> 00:56:57.780
and this is one not as an economist

1290
00:56:57.780 --> 00:56:59.160
that I've wondered about

1291
00:56:59.160 --> 00:57:00.780
for a couple of years,

1292
00:57:00.780 --> 00:57:03.510
is that from an equity perspective,

1293
00:57:03.510 --> 00:57:07.440
I actually wonder about the household level perspective

1294
00:57:07.440 --> 00:57:10.623
as opposed to the individual level perspective.

1295
00:57:12.690 --> 00:57:14.880
I've never seen one of these presentations

1296
00:57:14.880 --> 00:57:17.970
that looks at the household level

1297
00:57:17.970 --> 00:57:20.130
and yet it's at the household level

1298
00:57:20.130 --> 00:57:22.770
that cushioning of resources

1299
00:57:22.770 --> 00:57:26.100
and equity balances often happen

1300
00:57:26.100 --> 00:57:27.987
not at the individual level.

1301
00:57:27.987 --> 00:57:31.440
And so I'm just curious from your experience,

1302
00:57:31.440 --> 00:57:36.440
have you ever seen work done at the household cost level?

1303
00:57:36.720 --> 00:57:37.740
And what do you think

1304
00:57:37.740 --> 00:57:41.250
about those as an equity perspective?

1305
00:57:41.250 --> 00:57:42.510
<v ->Yeah, I mean I think there's,</v>

1306
00:57:42.510 --> 00:57:44.790
I think there's some downstream costs there that, you know,

1307
00:57:44.790 --> 00:57:45.900
in theory would get

1308
00:57:45.900 --> 00:57:50.040
some subsumed into a societal perspective.

1309
00:57:50.040 --> 00:57:53.070
But I struggle to think of specific examples

1310
00:57:53.070 --> 00:57:56.010
that kind of cast the umbrella wide enough

1311
00:57:56.010 --> 00:57:57.150
or cast a net wide enough

1312
00:57:57.150 --> 00:58:00.750
to include other individuals within the household.

1313
00:58:00.750 --> 00:58:03.210
So I think that's a really,

1314
00:58:03.210 --> 00:58:04.890
that's a really great point.

1315
00:58:04.890 --> 00:58:05.723
You know, and it speaks

1316
00:58:05.723 --> 00:58:10.670
to the demand side part of this question

1317
00:58:12.480 --> 00:58:13.770
in a way that's really, you know,

1318
00:58:13.770 --> 00:58:15.363
potentially quite meaningful.

1319
00:58:16.410 --> 00:58:21.410
And I would tend to think that as you suspect, you know,

1320
00:58:23.220 --> 00:58:27.750
we'd be moving towards even greater downstream benefits

1321
00:58:27.750 --> 00:58:28.680
at the household level.

1322
00:58:28.680 --> 00:58:30.600
So I am working on one study right now

1323
00:58:30.600 --> 00:58:31.980
in Uganda actually

1324
00:58:31.980 --> 00:58:35.359
with Jeremy who I think might have jumped off

1325
00:58:35.359 --> 00:58:40.200
where we are investigating patient costs

1326
00:58:40.200 --> 00:58:42.960
as a component of that intervention

1327
00:58:42.960 --> 00:58:46.170
and we do tend to ask in the scope

1328
00:58:46.170 --> 00:58:51.170
of collecting patient costs data about opportunity costs

1329
00:58:51.540 --> 00:58:53.370
that are germane to the rest of the household.

1330
00:58:53.370 --> 00:58:57.550
So we'll often ask questions about household expenditures

1331
00:58:58.770 --> 00:59:01.660
and household income generating activities

1332
00:59:03.180 --> 00:59:07.260
as well as durable assets of the household

1333
00:59:07.260 --> 00:59:09.480
and ways that households can sort of deal

1334
00:59:09.480 --> 00:59:11.640
with shocks over time.

1335
00:59:11.640 --> 00:59:13.350
'Cause those are all really important questions

1336
00:59:13.350 --> 00:59:16.110
I think that play into, you know,

1337
00:59:16.110 --> 00:59:17.910
what the experience of a patient is like,

1338
00:59:17.910 --> 00:59:20.370
particularly in a lower middle income country setting

1339
00:59:20.370 --> 00:59:21.420
where one of these interventions

1340
00:59:21.420 --> 00:59:24.270
might have a really meaningful downstream impact

1341
00:59:24.270 --> 00:59:26.700
on, you know, on the rest of the household.

1342
00:59:26.700 --> 00:59:29.370
So it's something that I spend a lot of time thinking about

1343
00:59:29.370 --> 00:59:31.290
'cause I kind of started in that world I think

1344
00:59:31.290 --> 00:59:34.260
before coming into the economic evaluation world.

1345
00:59:34.260 --> 00:59:37.350
And it's one that I think doesn't currently receive

1346
00:59:37.350 --> 00:59:39.930
a great deal of attention, but ought to.

1347
00:59:39.930 --> 00:59:41.910
But again, to, you know, to Reza's warning,

1348
00:59:41.910 --> 00:59:45.240
I think you have to be careful about, you know,

1349
00:59:45.240 --> 00:59:47.523
how much generalizability you pull,

1350
00:59:49.020 --> 00:59:51.480
and try to tout from different studies like this

1351
00:59:51.480 --> 00:59:55.140
because, right, they have different implications

1352
00:59:55.140 --> 00:59:56.280
for different parties

1353
00:59:56.280 --> 00:59:57.117
And so, you know,

1354
00:59:57.117 --> 00:59:59.550
I'd like to see a standard,

1355
00:59:59.550 --> 01:00:01.440
not where one perspective is chosen

1356
01:00:01.440 --> 01:00:03.840
over another ad nauseum,

1357
01:00:03.840 --> 01:00:05.490
but where they're both incorporated

1358
01:00:05.490 --> 01:00:07.410
and where there's a societal

1359
01:00:07.410 --> 01:00:11.160
and a financial or a programmatic perspective presented

1360
01:00:11.160 --> 01:00:14.070
within the same paper, you know.

1361
01:00:14.070 --> 01:00:17.490
And readers of those pieces of evidence

1362
01:00:17.490 --> 01:00:18.870
can then decide for themselves,

1363
01:00:18.870 --> 01:00:20.070
like which is the perspective

1364
01:00:20.070 --> 01:00:21.390
that's the most meaningful?

1365
01:00:21.390 --> 01:00:24.300
What are the trade-offs, you know, along the way?

1366
01:00:24.300 --> 01:00:27.930
'Cause it, you know, the value add of that is substantial,

1367
01:00:27.930 --> 01:00:30.240
to add in a societal perspective piece

1368
01:00:30.240 --> 01:00:33.060
to a publish economic evaluation.

1369
01:00:33.060 --> 01:00:35.490
<v ->So I think we do need to wrap up.</v>

1370
01:00:35.490 --> 01:00:36.870
But thank you Drew.

1371
01:00:36.870 --> 01:00:38.910
I think that's very helpful

1372
01:00:38.910 --> 01:00:42.993
and we will have the video available for people.

1373
01:00:44.970 --> 01:00:47.550
And really appreciate your time

1374
01:00:47.550 --> 01:00:50.490
in preparing all of this material

1375
01:00:50.490 --> 01:00:53.283
and sharing it with all of us.

1376
01:00:54.420 --> 01:00:55.500
<v ->Appreciate the invitation</v>

1377
01:00:55.500 --> 01:00:58.260
and thanks to those attended today.

1378
01:00:58.260 --> 01:00:59.660
It's nice to see some faces.

1379
01:01:01.260 --> 01:01:02.093
<v ->Okay.</v>