2018 Cost-Effectiveness Analysis Changelog

This page provides details about changes that were made to our cost-effectiveness analysis (CEA) in 2018. Below each change, we share a table indicating how the change impacted our cost-effectiveness estimates for each charity. (If a charity doesn't appear in the table, it was not impacted by the change.)

Version 16 — Published November 27, 2018

Link to the cost-effectiveness analysis file: 2018 CEA — Version 16

Change 1: Adjusted one individual's input values on the "Moral weights" tab

One individual in the CEA accidentally misentered some values in his column on the "Moral weights" tab. We updated the CEA to correct this error. The median results of the CEA were not affected.

Version 15 — Published November 26, 2018

Link to the cost-effectiveness analysis file: 2018 CEA — Version 15

Change 1: Started accounting for possible negative spillovers in the GiveDirectly model

We began making a 5% discount in the GiveDirectly CEA to account for potential, negative spillover effects. The rationale for this change is described in detail here.

Charity	Median [Charity] vs. cash before	Median [Charity] vs. cash after	Percent change
Deworm the World	17.4x	18.3x	5.3%
Schistosomiasis Control Initiative	8.9x	9.4x	5.3%
Sightsavers	9.3x	9.8x	5.3%
Against Malaria Foundation	5.6x	5.9x	5.3%
Malaria Consortium	8.9x	9.3x	5.3%
Helen Keller International	5.8x	6.1x	5.3%
The END Fund	5.2x	5.5x	5.3%

Version 14 — Published November 25, 2018

Link to the cost-effectiveness analysis file: 2018 CEA — Version 14

Change 1: Removed one individual from the model and added five

We removed one individual from the cost-effectiveness model and added five others. This reflects a change in the composition of GiveWell's research team.

Charity	Median [Charity] vs. cash before	Median [Charity] vs. cash after	Percent change
Deworm the World	21.4x	23.8x	11.4%
Schistosomiasis Control Initiative	14.8x	16.5x	11.2%
Sightsavers	7.9x	8.8x	11.2%
Against Malaria Foundation	5.7x	5.6x	-1.4%
Malaria Consortium	8.2x	8.0x	-1.9%
Helen Keller International	3.5x	3.5x	0.0%
The END Fund	5.6x	6.2x	11.6%

Change 2: Began accounting for potential long-term benefits of vitamin A supplementation

Previously, we weren't including possible long-term development benefits from vitamin A supplementation due to a lack of strong evidence. We decided to include a speculative adjustment to ensure consistency in comparisons of cost-effectiveness with our other priority programs (which have stronger evidence for long-run development effects).

Charity	Median [Charity] vs. cash before	Median [Charity] vs. cash after	Percent change
Helen Keller International	3.5x	5.2x	49.9%

Change 3: Updated other parameter values

Individuals included in the model updated the values used for other parameters found on the "User inputs" tab at their discretion.

Charity	Median [Charity] vs. cash before	Median [Charity] vs. cash after	Percent change
Deworm the World	23.8x	23.2x	-2.7%
Schistosomiasis Control Initiative	16.5x	15.9x	-3.4%
Sightsavers	8.8x	8.6x	-3.1%
Against Malaria Foundation	5.6x	5.7x	1.5%
Helen Keller International	5.2x	5.2x	-1.2%
The END Fund	6.2x	6.0x	-3.6%

Change 4: Updated moral weights

Individuals included in the model updated the values used for parameters found on the "Moral weights" tab at their discretion.

Charity	Median [Charity] vs. cash before	Median [Charity] vs. cash after	Percent change
Against Malaria Foundation	5.7x	5.6x	-1.4%
Malaria Consortium	8.0x	8.9x	10.5%
Helen Keller International	5.2x	5.8x	12.7%

Change 5: Updated worm intensity adjustments

We made a major change to our process for calculating worm intensity in the regions in which our charities work. The spreadsheet where the new values are calculated is here.

Charity	Median [Charity] vs. cash before	Median [Charity] vs. cash after	Percent change
Deworm the World	23.2x	17.4x	-24.9%
Schistosomiasis Control Initiative	15.9x	8.9x	-44.2%
Sightsavers	8.6x	9.3x	8.1%
The END Fund	6.0x	5.2x	-13.3%

Version 13 — Published November 20, 2018

Link to the cost-effectiveness analysis file: 2018 CEA — Version 13

Change 1: Removed No Lean Season from the CEA

We removed No Lean Season from the cost-effectiveness model. We did this because No Lean Season is no longer a top charity. For more information, please read our blog post detailing the decision.

Version 12 — Published November 18, 2018

Link to the cost-effectiveness analysis file: 2018 CEA — Version 12

Change 1: Updated the cost per supplement in the vitamin A supplementation model

We recently incorportated new coverage information into our 2018 analysis of Helen Keller International's cost per supplement. The overall cost per supplement changed by less than one cent, but changes in country-specific cost estimates were larger.

Charity	Median [Charity] vs. cash before	Median [Charity] vs. cash after	Percent change
Helen Keller International	3.5x	3.5x	-0.3%

Change 2: Updated Deworld the World's cost per treatment

We updated Deworm the World's overall cost per treatment (including costs paid by Deworm the World and other entities) from $0.77 to $0.61. We made this update to incorporate the results of our 2018 analysis of Deworm the World's cost per treatment. Most of the change in the overall cost per treatment came from a reduction in our estimate of Deworm the World's costs. The change in estimated cost-effectiveness was substantial due to the way we model leverage and funging.

Charity	Median [Charity] vs. cash before	Median [Charity] vs. cash after	Percent change
Deworm the World	13.6x	21.4x	57.8%

Change 3: Corrected an error in SCI's cost per treatment

We updated SCI's overall cost per treatment from $1.00 to $0.99. This change was made to correct a small error in our 2018 analysis of SCI's cost per treatment.

Charity	Median [Charity] vs. cash before	Median [Charity] vs. cash after	Percent change
Schistosomiasis Control Initiative	14.6x	14.8x	1.5%

Change 4: Updated the "Inclusion/Exclusion" tab

We made updates to the "SMC" and "HKI" sections of the "Inclusion/Exclusion" tab. These changes did not alter the results of our model.

Version 11 — Published November 9, 2018

Link to the cost-effectiveness analysis file: 2018 CEA — Version 11

Change 1: Updated expectations about the duration of long-term benefits of deworming

We updated the value of the parameter labeled "Duration of long-term benefits of deworming (in years)" from 30 to 40. We made this update to reflect our latest thinking which was impacted by a recent follow-up survey.

Charity	Median [Charity] vs. cash before	Median [Charity] vs. cash after	Percent change
Deworm the World	12.0x	13.6x	13.3%
Schistosomiasis Control Initiative	9.5x	10.8x	13.8%
Sightsavers	7.0x	8.0x	14.0%
Against Malaria Foundation	4.8x	5.0x	4.3%
Malaria Consortium	7.2x	7.4x	2.8%
Helen Keller International	3.1x	3.1x	0.0%
The END Fund	4.9x	5.6x	13.7%

Change 2: Updated SCI's cost per treatment

We updated SCI's overall cost per treatment from $1.13 to $1.00. We made this update to incorporate the results of our 2018 analysis of SCI's cost per treatment.

Charity	Median [Charity] vs. cash before	Median [Charity] vs. cash after	Percent change
Schistosomiasis Control Initiative	10.8x	14.7x	36.6%

Change 3: Updated baseline mortality rates used in the vitamin A supplementation model

We revised the baseline mortality rates used in the vitamin A supplementation model. We now slightly adjust rates published by the Global Burden of Disease Project to estimate counterfactual mortality rates in absence of vitamin A supplementation programs.

Charity	Median [Charity] vs. cash before	Median [Charity] vs. cash after	Percent change
Helen Keller International	3.1x	3.5x	14.4%

Change 4: Revised our process for estimating the long-term impact of malaria prevention with nets

We made a major change to our method for estimating the long-term economic impact of malaria prevention during childhood. We now extrapolate from the results of two studies assessing the long-term impact of large-scale malaria-prevention interventions, Bleakley 2010 and Cutler et al. 2010. We expect to publish a detailed write-up about this work sometime in the next several months.

Charity	Median [Charity] vs. cash before	Median [Charity] vs. cash after	Percent change
Deworm the World	13.6x	13.6x	0.0%
Schistosomiasis Control Initiative	14.7x	14.6x	-0.6%
Sightsavers	8.0x	7.9x	-0.9%
Against Malaria Foundation	5.0x	5.7x	13.4%
Malaria Consortium	7.4x	7.4x	0.3%
The END Fund	5.6x	5.6x	-0.2%

Change 5: Revised our process for estimating the long-term impact of malaria prevention with SMC

We made a revision to our process for estimating the long-term effects of malaria prevention via seasonal malaria chemoprevention. This revision was similar to the change in the nets model described above.

Charity	Median [Charity] vs. cash before	Median [Charity] vs. cash after	Percent change
Malaria Consortium	7.4x	8.2x	10.2%

Version 10 — Published September 28, 2018

Link to the cost-effectiveness analysis file: 2018 CEA — Version 10

Change 1: Updated Sightsavers' cost per treatment

We updated the cost per treatment (which includes both Sightsavers and government costs) in the Sightsavers' CEA from $1.22 to $0.95. We made this update to incorporate the results of our 2018 analysis of Sightsavers' cost per treatment. The increase in cost-effectiveness was mostly driven by a 33% reduction in Sightsavers costs, resulting in a 49% increase in cost-effectiveness. The remainder was driven by a smaller decrease in government costs.

Charity	Median [Charity] vs. cash before	Median [Charity] vs. cash after	Percent change
Sightsavers	4.7x	7.0x	50.4%

Change 2: Corrected an error in the END Fund's leverage calculation

In version 8 of the 2018 CEA, we failed to update some cost data in the leverage calculations for the END Fund when updating END Fund's cost per treatment. This error has been corrected.

Charity	Median [Charity] vs. cash before	Median [Charity] vs. cash after	Percent change
The END Fund	4.8x	4.9x	2.7%

Change 3: Updated the probability of some scenarios considered in the "Leverage/funging" tab

In the "Leverage/funging" tab, we consider how funding our top charities may alter the incentives other entities have to fund programs. We adjusted the probability of scenarios considered for Deworm the Wold, Malaria Consortium, and the Against Malaria Foundation to reflect our latest thinking.

Charity	Median [Charity] vs. cash before	Median [Charity] vs. cash after	Percent change
Deworm the World	12.0x	12.0x	0.1%
Against Malaria Foundation	4.7x	4.8x	2.3%
Malaria Consortium	7.7x	7.2x	-6.1%

Version 9 — Published September 26, 2018

Link to the cost-effectiveness analysis file: 2018 CEA — Version 9

Change 1: Added benefits for untreated individuals to the SMC CEA

We added a section to the season malaria chemoprevention (SMC) CEA to account for reductions in malaria incidence experienced by individuals who do not receive treatment.

Charity	Median [Charity] vs. cash before	Median [Charity] vs. cash after	Percent change
Malaria Consortium	6.9x	8.0x	14.8%

Change 2: Removed several countries from the SMC model

We updated portions of the SMC model to reflect our expectations about areas where Malaria Consortium will run programs in the future. Mortality estimates from Guinea, Mali, Niger, and The Gambia are no longer included in the model. We expect that Malaria Consortium will continue to work in Burkina Faso, Nigeria and Chad.

Charity	Median [Charity] vs. cash before	Median [Charity] vs. cash after	Percent change
Malaria Consortium	8.0x	7.7x	-4.0%

Change 3: Updated mortality estimates in the VAS model

We updated mortality estimates found in the Vitamin A supplementation (VAS) model to reflect our expectations about where Helen Keller International (HKI) will run programs going forward. The previous version of our model drew on mortality estimates from Burkina Faso, Mali, Guinea, Cote d'Ivoire, and Niger. As a part of this change, mortality estimates from Cameroon, DRC, Kenya, Mozambique, and Nigeria were added. The calculations behind the updated estimates can be seen here.

Charity	Median [Charity] vs. cash before	Median [Charity] vs. cash after	Percent change
Helen Keller International	5.1x	4.1x	-19.5%

Change 4: Updated the external validity parameter in the VAS model

We introduced country-specific values for the VAS external validity parameter (which reflects different levels of vitamin A deficiency and causes of child mortality) and revised the default value to reflect our latest expectations about where HKI is likely to run programs. The parameter was moved from the "User inputs" tab to the "Country selection" tab. Calculations behind the latest figures can be seen in this workbook.

Charity	Median [Charity] vs. cash before	Median [Charity] vs. cash after	Percent change
Helen Keller International	4.1x	3.1x	-25.1%

Change 5: Additions to the nets portion of the "Country selection" tab

We added an adjustment to the nets portion of the "Country selection" tab to account for the share of all-cause mortality coming from malaria in different countries. This change only affects our model's results when a specific country is selected on the "Country selection" tab.

Version 8 — Published September 6, 2018

Link to the cost-effectiveness analysis file: 2018 CEA — Version 8

Change 1: Updated the adjustment for insecticide resistance in the nets model

We changed the value for the parameter labeled Efficacy reduction due to insecticide resistance from 26% to 35%. This updated value reflects our latest views on insecticide resistance.

Charity	Median [Charity] vs. cash before	Median [Charity] vs. cash after	Percent change
Deworm the World	12.2x	12.2x	0.0%
Schistosomiasis Control Initiative	9.7x	9.7x	0.4%
Sightsavers	4.8x	4.8x	0.6%
Against Malaria Foundation	5.4x	4.7x	-12.1%
Malaria Consortium	7.0x	6.9x	-0.9%
The END Fund	4.2x	4.2x	0.2%

Change 2: Removed short-term health benefits of deworming from the model

We removed the short-term health benefits of deworming from our model. Short-term health benefits represented only a small portion of the overall benefits of deworming according to our model. We felt that our method for assessing short-term health benefits had too many issues to be worth keeping in the CEA. The short-term health benefits of deworming are now listed as "Excluded" in the "Inclusion/Exclusion" tab.

Charity	Median [Charity] vs. cash before	Median [Charity] vs. cash after	Percent change
Deworm the World	12.2x	12.0x	-2.1%
Schistosomiasis Control Initiative	9.7x	9.5x	-2.6%
Sightsavers	4.8x	4.7x	-2.2%
The END Fund	4.2x	4.1x	-2.6%

Change 3: Updated END Fund's cost per treatment

An error was found in our 2017 analysis of END Fund's cost per treatment. We corrected this error in our updated analysis. The cost per person deworming moved from $0.94 to $0.81.

Charity	Median [Charity] vs. cash before	Median [Charity] vs. cash after	Percent change
The END Fund	4.1x	4.8x	17.6%

Change 4: Added a parameter to account for long-term benefits of Vitamin A supplementation

We added a parameter labeled Value of development benefits from a year of VAS coverage relative to a year of deworming in Miguel and Kremer. By default, this parameter has been set to a value of zero for all individuals. Accordingly, this addition has not yet had any effect on the model's results. We expect that individuals contributing inputs to the CEA may change the values used for this parameter in the future. We added this parameter so that individuals have the option to include potential development benefits of VAS; this makes the VAS CEA more consistent with our CEAs for interventions related to malaria and deworming, but we are highly uncertain whether VAS leads to development benefits.

Change 5: Updated the "Inclusion/Exclusion" tab

We made several additions and adjustments to content in the "Inclusion/Exclusion" tab. These changes did not affect the model's results.

Version 7 — Published August 24, 2018

Link to the cost-effectiveness analysis file: 2018 CEA — Version 7

Change 1: Added a "Country selection" tab to the CEA

We added a new "Country selection" tab to the CEA spreadsheet. If users select specific countries from drop-down menus on this new tab, country-specific parameter values will be used in the CEA.

By default, the CEA will continue to use overall figures that are calculated by aggregating across countries. Selecting specific countries will lead to changes in CEA results. When the default figures are used, results in this version of the CEA are identical to results in the previous version.

Version 6 — Published August 10, 2018

Link to the cost-effectiveness analysis file: 2018 CEA — Version 6

Change 1: Updated the cost per child covered in Helen Keller International's vitamin A supplementation program

We updated the cost per child covered in each round of Helen Keller International's vitamin A supplementation program from $0.75 to $1.35. We made this update to incorporate the results of our 2018 analysis of Helen Keller International's cost per supplement.

Charity	Median [Charity] vs. cash before	Median [Charity] vs. cash after	Percent change
Helen Keller International	8.7x	5.1x	-41.6%

Version 5 — Published July 23, 2018

Link to the cost-effectiveness analysis file: 2018 CEA — Version 5

Change 1: Updated the cost per child targeted in Malaria Consortium's SMC program

We updated the cost per child targeted in Malaria Consortium's seasonal malaria chemoprevention (SMC) program from $4.33 to $4.54. We made this update to incorporate the results of our 2018 analysis of Malaria Consortium's cost per treatment.

Charity	Median [Charity] vs. cash before	Median [Charity] vs. cash after	Percent change
Malaria Consortium	7.1x	6.8x	-3.8%

Change 2: Updated estimates of coverage rates in Malaria Consortium's SMC program

We updated estimates of coverage rates in Malaria Consortium's SMC program. We made this update to incorporate the results of coverage surveys that occurred in 2017.

Charity	Median [Charity] vs. cash before	Median [Charity] vs. cash after	Percent change
Malaria Consortium	6.8x	7.0x	2.9%

Version 4 — Published July 16, 2018

Link to the cost-effectiveness analysis file: 2018 CEA — Version 4

Change 1: Updated the Cost per ITN parameter

We updated the Cost per ITN parameter in the "Nets" tab from a value of $4.22 to $4.53. We made this update to incorporate the results of our 2018 analysis of AMF's cost per net.

Charity	Median [Charity] vs. cash before	Median [Charity] vs. cash after	Percent change
Schistosomiasis Control Initiative	9.7x	9.7x	0.3%
Sightsavers	4.7x	4.8x	0.3%
Against Malaria Foundation	5.7x	5.4x	-6.5%
Malaria Consortium	7.1x	7.1x	-0.5%
The END Fund	4.2x	4.2x	0.1%

Change 2: Made a new explanatory document available

We added links in the CEA spreadsheet to a document that includes:

• A high-level overview of why we use CEAs and how we structure them
• Conceptual guidelines on how to create CEAs
• Technical guidelines related to CEAs

While the target audience of the document is members of the GiveWell research team, the content may be useful to others. The new document is intended to replace the out-of-date FAQ & Parameter Details document.

Version 3 — Published May 3, 2018

Link to the cost-effectiveness analysis file: 2018 CEA — Version 3

Change 1: Updated the Transfers as a percentage of total cost parameter

We updated the Transfers as a percentage of total cost parameter in the "Cash" tab from a value of 81.5% to 83.0%. We made this update to incorporate information about GiveDirectly's spending from August 2017 through February 2018. The calculations behind the new value can be seen here.

Charity	Median [Charity] vs. cash before	Median [Charity] vs. cash after	Percent change
Deworm the World	12.4x	12.2x	-1.8%
Schistosomiasis Control Initiative	9.8x	9.7x	-1.8%
Sightsavers	4.8x	4.7x	-1.8%
Against Malaria Foundation	5.8x	5.7x	-1.8%
Malaria Consortium	7.2x	7.1x	-1.8%
Helen Keller International	8.8x	8.7x	-1.8%
The END Fund	4.2x	4.2x	-1.8%
No Lean Season	5.3x	5.2x	-1.8%

Version 2 — Published April 26, 2018

Link to the cost-effectiveness analysis file: 2018 CEA — Version 2

Change 1: Renamed the "Parameters" tab to "User inputs"

This change did not affect the model's results.

Change 2: Replaced items on the "User inputs" tab with default values

We removed the majority of items from the "User inputs" tab and began using a single default value for these parameters. The parameters and default values are listed below, separated by intervention. Items marked "Removed" were deleted from the CEA. Removing values was mathematically equivalent to setting their values to 100%. Our reasoning behind the default values can be found in the cell notes associated with each item. These changes are intended to make the CEA easier to engage with.

Cash parameters

• Percent of investment returned when benefits end [20%]

Charity	Median [Charity] vs. cash before	Median [Charity] vs. cash after	Percent change
Against Malaria Foundation	6.0x	6.1x	0.8%
Malaria Consortium	6.9x	6.9x	0.6%
Helen Keller International	8.9x	8.9x	0.2%

Migration subsidies parameters

• Internal validity adjustment — Migration subsidies [80%]
• External validity adjustment — Migration subsidies [80%]
• Further adjustment for uncertainty of evidence for benefits in the second year of remigration [Removed]
• Adjustment for uncertainty of evidence regarding benefits in lesser lean seasons [Removed]
• Adjustment to the utility of migration, migrants' nonmigrating household members [100%]

Charity	Median [Charity] vs. cash before	Median [Charity] vs. cash after	Percent change
No Lean Season	4.9x	5.3x	8.2%

VAS parameters

• Relative risk of all-cause mortality for children aged 6- to 59-months participating in VAS programs [0.76]
• Internal validity adjustment — VAS [95%]

Charity	Median [Charity] vs. cash before	Median [Charity] vs. cash after	Percent change
Helen Keller International	8.9x	8.8x	-0.5%

SMC parameters

• Internal validity adjustment — SMC [95%]
• External validity adjustment — SMC [100%]
• Ratio of the reduction in malaria mortality to the reduction in malaria incidence [100%]

Charity	Median [Charity] vs. cash before	Median [Charity] vs. cash after	Percent change
Malaria Consortium	6.9x	7.2x	4.2%

Nets parameters

• Relative efficacy of ITNs for reducing mortality of individuals age 5 and older [80%]
• Net use adjustment [90%]
• Proportion of mortality attributed to malaria in areas AMF works vs. the contexts of trials in Lengeler 2004 [100%]
• Internal validity adjustment — Nets [95%]

Charity	Median [Charity] vs. cash before	Median [Charity] vs. cash after	Percent change
Deworm the World	12.4x	12.4x	0.0%
Schistosomiasis Control Initiative	9.8	9.8x	0.4%
Sightsavers	4.8x	4.8	0.5%
Against Malaria Foundation	6.1x	5.8x	-3.6%
Malaria Consortium	7.2x	7.2x	0.5%
The END Fund	4.2x	4.2x	0.2%

Deworming parameters

• Treatment effect of deworming on ln(consumption) [0.14]
• Multiplier for resource sharing within households [2.0]
• Duration of long-term benefits of deworming (in years) [30]
• Proportion of dewormed children that benefited from long-term gains in Baird et al. [Removed]
• Adjustment for varying treatment frequency (all deworming charities) [Removed]

Changes to deworming parameters moved results for some individuals, but did not alter the median CEA result.

Change 3: Added "Inclusion/Exclusion" tab

We added a tab titled "Inclusion/Exclusion" to the CEA. This tab lists different items that we considered accounting for in our cost-effectiveness analysis. Each item is evaluated with a rubric assessing how easy it would to include the item in our model and how much including an item might move our bottom line estimates.

This change did not affect the model's results.

Change 4: Removed leverage/funging adjustments from the "User inputs" tab

This change was intended to make the CEA easier to engage with. The change did not affect the model's results.

Change 5: Added a set of "Conventional" moral weights to the CEA

We added a column labeled "Conventional" to the "Moral weights" tab of the CEA. This set of moral weights is an attempt to represent normative values commonly used in analyses conducted by other organizations in the global health and development field; see more information here. We don't believe there's a single set of normative values that can accurately represent the typical view in the global health and development community, so these moral weights should be interpreted cautiously. The new set of moral weights is combined with the median values for items on the "User inputs" tab to arrive at final cost-effectiveness figures in the "Conventional" column. These final figures were not accounted for when calculating the median results among individuals included in our CEA.

Change 6: Revised content on the "Notes" tab

Content on the "Notes" tab was revised to reflect the changes above. This change did not affect the model's results.

Version 1 — Published January 24, 2018

Link to the cost-effectiveness analysis file: 2018 CEA — Version 1

Change 1: Rolled out a new update process

With this update, we transitioned to a new process for releasing updates to the cost-effectiveness model. The new process involves both internal changes in how GiveWell works on updates and an external change to the structure of release notes. We hope that this change will make updating the CEA easier.

This change did not affect the model's results.

Change 2: Changed the summary effect in the seasonal malaria chemoprevention CEA

We moved the parameter labeled "Relative risk for malaria cases, intention to treat effect" from the "Parameters" tab to the "SMC" tab. The parameter was set to a default value of 0.25, the relative risk suggested by our meta-analysis of seasonal malaria chemoprevention trials.

Charity	Median [Charity] vs. cash before	Median [Charity] vs. cash after	Percent change
Malaria Consortium	6.6x	6.9x	4.2%