Summary of Cost per Treatment Estimates

Grey text color indicates formula does currency conversion.

Cost per treatment

% of costs (for reference across estimates)

SCI costs, excluding drugs

Government costs

Total treatments

SCI costs, excluding drugs

Government costs

SCI costs, excluding drugs

Government costs

In the below sections, bold lines are numbers that are referenced in the summary table above

To calculate drug costs, we focus on the data we have available, primarily purchases by Crown Agents as part of the ICOSA program funded by DFID. Note, we use the same value for all programs, though the price has increased over the last 10 years.

SCI stated that 20 pence per treatment, which lines up with our analysis when excluding wastage, is a reasonable estimate. "Praziquantel leaving the factory gate was as low as 7 cents per tablet 10 years ago. Now it is 10 cents per tablet. That is cents not pence. With insurance and shipping the cost at the receiving country is about 12 cents - or taking a 1.66 exchange rate say 8 pence per tablet = 20 pence per child at 2.5 tablets on average." Alan Fenwick, SCI Director, email to GiveWell, October 23, 2014

Crown Agents Total ICOSA Procurement Spend (updated 2014)

The units in the source file are not labelled. We have assumed the figures are in GBP.

Total number of tablets procured and delivered by March 2014

SCI ICOSA Mid-Year Report 2014, Pg 12 (Table 3).

This includes a large number of tablets held in reserve and not yet delivered as of March 2014.

GiveWell calculation.

For reference, today this is about: $0.13.

GiveWell judgment.

This is the percentage of drugs we assume are wasted. We do not have data on actual wastage so are intuitively hypothesizing what seems reasonable. We note that a similar assumption was discussed in an paper we found (Utroska et al. 1989), though this is quite dated and we do not know how the authors arrived at 10%.

Drug costs per tablet, accounting for wastage

GiveWell calculation.

SCI ICOSA Mid-Year Report 2014, Pg 12.

This is consistent with a variety of other sources.

Drug costs per treatment, accounting for wastage

GiveWell calculation.

For all of our analysis, we use a uniform percentage of total costs to estimate the share of the costs that the government incurs.

Sources and notes: See "Shortcomings on our analysis" section section of review on SCI: http://www.givewell.org/node/2365#Shortcomingsofouranalysis

This uses all of the ICOSA programs, which have been primarily funded by DFID.

Time period: Focuses on October 2010 (start of DFID grant) through March 2014 (last key reporting date; also prior to 2nd grant being issued). See notes for exceptions.

Programs included: Cote d'Ivoire, Liberia, Malawi, Mozambique, Niger, Tanzania, Uganda, Zambia, and Zanzibar

GiveWell calculation.

For simplicity, taking the simple average of yearly average GBP/USD rate from 2011 through 2013.

FY1-3 from SCI ICOSA Mid-Year Report 2014, Pg 9 (Table 1). FY4 from SCI treatment numbers (October 2014); GiveWell calculated total of the 9 ICOSA programs for "Apr 13 - Mar 14" period, and added this to FY1-3.

Though the label on table 1 is not clear, from the context of the report, we infer that this is the number of schistosomiasis treatments (though there may have been STH or other treatments delivered in combination); e.g. "The outcome of the project will be to contribute to the WHO global strategic plan for SCH (2012-2020) by providing a total of 75 million treatments." (Pg 7). We are using a more recent source for FY4 as we believe it is more accurate (it shows about 2 million additional treatments, which may not have been completed at time of Mid-Year report). Both sources have a breakdown available by country and by period.

SCI advisory board financial report (June 2014), Slide 3

£10.5 million is reinforced by other documents. For example, see SCI ICOSA Mid-Year Report 2014, Page 4.

SCI advisory board financial report (June 2014), Slide 11

This appears to be the amount of the first DFID grant award that has not been spent as of May 2014. SCI ICOSA Mid-Year Report 2014 does not report total actual spending.

GiveWell calculation

This is the implied amount of the £10.5 million grant from DFID that SCI had spent as of May 2014. Ideally, we would exclude April and May 2014 from this figure.

Unrestricted funding, ICOSA programs, program spending

GiveWell summary of SCI finances (October 2014), Sheet Combined with previous updates. GiveWell calculated the total from each of the 9 relevant program lines for each of the 3 time periods November 2011 through March 2014 and then converted from USD to GBP (for comparison with DFID spending).

We ignore unrestricted funds spent prior to November 2011, which we believe to be small (see details on unrestricted funding in SCI review).

GiveWell summary of SCI finances (October 2014), Sheet Combined with previous updates. GiveWell calculated the total from the "operating costs" line item for each of the 3 time periods November 2011 through March 2014 and then converted from USD to GBP (for comparison with DFID spending).

It is possible that operating costs are just one part of HQ costs, but we don't know of other line items that clearly make sense to include.

GiveWell summary of SCI finances (October 2014), Sheet Combined with previous updates. GiveWell calculated the total from the "total spending" line for each of the 3 time periods November 2011 through March 2014 and then converted from USD to GBP (for comparison with DFID spending).

Unrestricted funding, % non-HQ costs spent on ICOSA programs

GiveWell calculation

For context, a large amount of the remaining 70% of unrestricted spending was spent in Ethiopia (about £1 million).

GiveWell calculation

Allocating HQ costs to ICOSA under rough assumption that every £ of unrestricted (non-HQ) spending requires the same amount of HQ support.

GiveWell calculation

It is possible this includes some drug purchases, which we'd ideally exclude here to avoid double counting. We expect it is unlikely any unrestricted funds have been used for this purpose given that DFID has allocated additional funding for drug purchases.

SCI advisory board financial report (June 2014), Slide 3

Vitol funded mapping in Cote d'Ivoire; we use "Expended (incl. commitments)" since (according to the same source) this mapping has been completed (so we don't expect there are outstanding commitments). Other than this, we are not aware of any other sources of funding going towards these programs.

GiveWell calculation

SCI shared detailed monthly expenses with us (unpublished), which show totals of £8.5 and £8.7 million spent in ICOSA countries through April 2014. It is not clear to us what explains the difference. However, it appears that a spreadsheet error double counts some line items, and correcting this reduces the total to £7.9 million (of which 0.1 million is in April 2014, which we'd exclude for comparison). If these inferences are incorrect and we are actually missing ~£1 million in costs, our cost per treatment is too low by about $0.11. Either way, the details give us some confidence that large buckets (e.g. of SCI salaries and spending through sub-contracts) are not excluded from our analysis. Note, It is possible that some of the difference includes costs we would want to exclude here (e.g. drug purchases, or research programs unrelated to these treatments).

GiveWell calculation

For reference, SCI reported £0.40 per treatment (which we believe includes little or no drug costs) on the first three years of the ICOSA program (with ~12 million treatments delivered at that point): "This gives a direct financial cost per treatment of 40 pence. These figures are broadly representative of other implementation projects in SCI but do not include funds allocated to advocacy and fundraising." SCI report to GiveWell (September 2013), Pg 1.

During the ICOSA programs, there has been additional spending for operational research. We are not certain that these expenses did not lead to some of the treatments counted in the ICOSA programs (either directly, as part of the research projects, or indirectly through improved operations due to the research), so we look at these as possible additional costs to get a more conservative estimate.

SCI advisory board financial report (June 2014), Slides 5, 17, 21. Sum of "actual spend" for SCORE and CIFF funding (from bar charts on slides 17 and 21) and "expended (incl. commitments)" (from table on slide 5) for Comic Relief (we did not have an estimate of how much of this was committed but not yet spent).

For reference, using current exchange rates this is about $1.8 million. We exclude the "Control of Cysticercosis" grant as it seems unlikely it led to deworming treatments. SCORE studies are in Niger and Mozambique, CIFF grant is in Liberia and Uganda, and Comic Relief funding is in Malawi.

Costs for SCI, excluding drugs, including research

GiveWell calculation

DFID spending, ICOSA programs, overseas spending

GiveWell calculated the total from country figures in SCI advisory board financial report (June 2014), Slide 8.

We are not confident in this estimate (it seems likely that some of the funds were transferred to the country but not yet spent such that it is over-stating the actual cost) and see an alternative that would be substantially different (explained in rest of note), but this is simply used to look at breakdown between program and HQ buckets and does not affect overall cost per treatment. The Mid-Year Report shows "direct financial cost per treatment" data by program, which implies a total of £2.7 million spent through March 2014 (details in unpublished GiveWell spreadsheet using data from Pgs 8-9 (for treatments by country) and Pg 18 (for costs per treatment by country)). This suggests this number might be overstated by GBP2.1 million.

GiveWell calculation

GiveWell calculation

This includes all of the "other funding" category.

GiveWell calculation

For comparison, SCI reported to us in 2013 that HQ costs were about 48% of their total costs. See http://www.givewell.org/international/top-charities/schistosomiasis-control-initiative#BreakdownofSCIsspending

Time period: 2014-2018. We believe this is likely to be April 2014 - December 2018, though it is not always clear and we could be relying on cost and treatment data with some timing inconsistencies.

Programs included: includes initial ICOSA programs (see above) and also Ethiopia and DRC.

Google (accessed November 4, 2014)

Using recent exchange rate as best guess of future.

SCI ICOSA Mid-Year Report 2014, Pgs 30-31. GiveWell calculated sum across programs and years.

Though the label on table 1 is not clear, from the context of the report, we infer that this is the number of schistosomiasis treatments (though there may be STH or other treatments delivered in combination); e.g. "The outcome of the project will be to contribute to the WHO global strategic plan for SCH (2012-2020) by providing a total of 75 million treatments." (Pg 7).

SCI ICOSA Mid-Year Report 2014, Pgs 30-31. GiveWell calculated sum across programs and years.

Using this data (instead of the estimate of 2.5 tablets per treatment used in the rest of the analysis) since it is available. It implies 2.53 tablets per treatment so this is not very impactful to the analysis.

SCI ICOSA Mid-Year Report 2014, Pgs 30-31. GiveWell calculated sum across programs and years.

In source, this is labelled "Estimated total cost (excluding drug costs)" though given the context it seems clear it is program costs (excluding HQ costs)

SCI advisory board financial report (June 2014), Slide 11

It is unclear why project expenses that feed into this are lower than the program projections used above, but it is plausible that the difference (about £2 million) is funding from other (non-DFID) sources.

SCI advisory board financial report (June 2014), Slide 11

It is unclear, but we are assuming that "Total Fees" (which includes management and technical fees) is roughly what we'd want as HQ costs.

GiveWell calculation

GiveWell calculation

Assuming that non-DFID funding will have the same HQ costs % of total as the DFID-funding.

GiveWell calculation

This estimate relies primarily on: SCI, Summary sheet of treatments instigated and overseen by SCI.

http://www.givewell.org/files/DWDA%202009/SCI/SCI%20country%20table.xls

Time period: It shows data from 2003 (roughly the start of SCI programs) through about 2009 (the end date isn't explicit). We are assuming that SCI is reporting consistent funding and treatment data (i.e. that this funding, and only this funding, was spent to deliver these treatments).

Programs included: Burkina Faso, Burundi, Cote D'Ivoire, Ghana, Malawi, Mali, Mozambique, Niger, Rwanda, Tanzania, Uganda, Yemen, Zambia, and Zanzibar.

GiveWell calculation.

For simplicity, taking the simple average of yearly average GBP/USD rate from 2003-2009.

SCI Summary sheet of treatments instigated and overseen by SCI

We are not including treatments "for all NTDs." See review for further explanation.

Funding invested in country through SCI, from Gates

SCI Summary sheet of treatments instigated and overseen by SCI

We assume all of the "funding invested in country" was spent to deliver the treatments that are reported. If some was transferred to country for treatments that had not yet been reported, then our costs are overstated.

SCI Summary sheet of treatments instigated and overseen by SCI

Funding invested in country through SCI, from Geneva Global and others

SCI Summary sheet of treatments instigated and overseen by SCI

SCI Summary sheet of treatments instigated and overseen by SCI

It is possible (though we believe unlikely) that these managements costs are a subset of the "funding invested in country" counted above. We would be double counting these if that were the case.

SCI Summary sheet of treatments instigated and overseen by SCI

It is possible (though we believe unlikely) that these managements costs are a subset of the "funding invested in country" counted above. We would be double counting these if that were the case.

SCI management costs and fees, from Geneva Global and others

SCI Summary sheet of treatments instigated and overseen by SCI

It is possible (though we believe unlikely) that these managements costs are a subset of the "funding invested in country" counted above. We would be double counting these if that were the case.

SCI Summary sheet of treatments instigated and overseen by SCI

This number appears to be the number of tablets.

SCI Summary sheet of treatments instigated and overseen by SCI

This number appears to be the number of tablets.

GiveWell calculation

Valuing Pzq purchases at current estimate. Costs have risen about $0.03 per tablet over the last 10 years (see note in drug cost section). While some of the intermediate numbers will shift, this ultimately only affects how we're valuing the donated drugs (since we are only shifting around the actual SCI costs b/w non-drug and drug expenses), so is a relatively small effect. If we reduced drug costs by ~$0.03 (which seems extreme, since we want the average cost over the period, not the minimum), our cost/treatment would drop about $0.04.

GiveWell calculation

This is an estimate the drug costs that are reported in SCI's total cost, which we use to infer SCI's total costs excluding drugs. In our final calculation, we calculate total drug costs off actual treatments rather than starting with this value and adding an estimate of donated drug costs; the alternative may be reasonable, but it does not appear to result in a materially different cost per treatment estimate and our approach has the advantage of being more consistent (e.g. in regards to our wastage assumption).

GiveWell calculation

For reference only. In our final calculation, we calculate drug costs off actual treatments.

GiveWell calculation

For reference. Normally, we expect this to be 2.5. The fact that it is higher could be a sign that we are undercounting treatments or that some drugs were wasted.