A | B | C | D | E | F | G | H | I | J | K | L | M | N | O | P | Q | R | S | T | U | V | W | X | Y | Z | |
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
1 | New Incentives Sheet | Median | Andrew | Chelsea | Chris | Elie | Isabel | James | Josh | Natalie | Sophie | Anyone | Notes | |||||||||||||
2 | calculation | results | less subjective input | input to be updated by more data | subjective input | imptnt uncertain input | ||||||||||||||||||||
3 | ||||||||||||||||||||||||||
4 | Summary & Results | |||||||||||||||||||||||||
5 | ||||||||||||||||||||||||||
6 | New Incentives vs AMF | 1.43 | 1.40 | 2.04 | 1.81 | 1.30 | 1.21 | 2.02 | 0.96 | 1.43 | 2.03 | |||||||||||||||
7 | New Incentives vs GiveDirectly | 10.31 | 7.79 | 23.18 | 26.54 | 11.76 | 7.96 | 14.75 | 5.77 | 10.31 | 2.46 | |||||||||||||||
8 | Cost per equivalent outcome to an under-5 death averted | $1,101 | $1,024 | $1,039 | $1,178 | $1,037 | $1,153 | $1,101 | $1,673 | $1,132 | $926 | |||||||||||||||
9 | ||||||||||||||||||||||||||
10 | Percent benefit driven by preventing under-5 deaths | 98% | 99% | 99% | 99% | 98% | 99% | 97% | 98% | 90% | ||||||||||||||||
11 | Percent benefit driven by consumption benenfits | 2% | 1% | 1% | 1% | 2% | 1% | 3% | 2% | 10% | ||||||||||||||||
12 | ||||||||||||||||||||||||||
13 | Value Assignments | |||||||||||||||||||||||||
14 | ||||||||||||||||||||||||||
15 | Value assigned to averting the death of an under-5 — Immunization | 8.00 | 80.00 | 50.00 | 15.00 | 36.00 | 30.52 | 10.00 | 50.00 | 0.14 | - | |||||||||||||||
16 | Value assigned to increasing ln(consumption) by one unit for one person for one year | 0.38 | 1.44 | 0.72 | 0.33 | 1.44 | 0.72 | 0.43 | 1.44 | 0.03 | - | |||||||||||||||
17 | Ratio: averting the death of an under-5 is X times as good as increasing ln(consumption) by one unit for one person for one year | 21.05 | 55.56 | 69.44 | 45.45 | 25.00 | 42.39 | 23.26 | 34.72 | 4.67 | Illustrative. Not used. | |||||||||||||||
18 | ||||||||||||||||||||||||||
19 | Cost per outcome as good as: averting the death of an individual under 5 — AMF | $1,429 | $2,119 | $2,137 | $1,351 | $1,395 | $2,219 | $1,602 | $1,615 | $1,884 | BED NETS | |||||||||||||||
20 | Cost per outcome as good as: averting the death of an individual under 5 — AMF | $7,977 | $24,091 | $31,261 | $12,198 | $9,177 | $16,230 | $9,649 | $11,663 | $2,281 | CASH | |||||||||||||||
21 | ||||||||||||||||||||||||||
22 | Parameters | |||||||||||||||||||||||||
23 | ||||||||||||||||||||||||||
24 | Use which incentive size? | Smaller (500&2000) | Smaller (500&2000) | Smaller (500&2000) | Smaller (500&2000) | Smaller (500&2000) | Smaller (500&2000) | Smaller (500&2000) | Smaller (500&2000) | Smaller (500&2000) | Smaller (500&2000) | Updated 05 June 2017 for all columns. Our current understanding is that New Incentives plans to offer 500 Naira incentives for the first four immunization visits (including birth/BCG), and is still determining the incentive size for measles. IDinsight "Proposed Evaluation Design 01Jun17". | ||||||||||||||
25 | Use which vaccination rates? | North West | North West | North West | North West | North West | North West | North West | North West | North West | North West | Vaccination rates (without the program) vary by region in Nigeria. Our source for vaccination rates is the 2013 Demographic and Health Survey. In late 2016 through mid 2017, New Incentives is conducting pilot activities in states representing three regions: Federal Capital Territory and Nasarawa (North Central), Anambra (South East), and Akwa Ibom (South South). Pilot sites in the Federal Capital Territory were discontinued sometime around the new year. In 2017, New Incentives is exploring learning sites in the North West, where vaccination rates are particularly low. We are uncertain at this stage where the program will scale: this will be determined by feasibility and cost-effectiveness. Our previous estimate of cost-effectiveness used the 2013 DHS results on vaccination rate and mortality from the national middle wealth quintile. Because SMART 2015 does not report vaccination rate by wealth characteristics, selecting "Middle Wealth Quintile" and "SMART 2015" will result in #N/As. | Please note that we expect this model to be less accurate/useful/plausible for the states with lower overall mortality. We extrapolate overall mortality with a trend line. Mortality decreasing with time, is of course _due_ to cause-specific mortality decreasing with time. So if you select a demographic with lower overall mortality, consider adjusting disease-specific incidence and fatality downwards as well. There is conditional formatting on this sheet that will highlight cells in dark red when they imply that a cause-specific mortality exceeds overall mortality. Consider adjusting the relevant disease parameters downwards until cells are no longer red. This is a rough fix; we have not at this time built an intuition-checker that allows users to adjust disease parameters until diseases comprise an intuitively reasonable proportion of all-cause deaths. The "scratch intuition checks" sheet is a start at this. | |||||||||||||
26 | Prefer which vaccination rate data source? | SMART 2015 | DHS 2013 | DHS 2013 | SMART 2015 | DHS 2013 | SMART 2015 | DHS 2013 | DHS 2013 | SMART 2015 | SMART 2015 | Recommended: SMART 2015. (More recent than DHS 2013.) We have not closely reviewed the difference in methodology between SMART 2015 and DHS 2013. Note that SMART 2015 does not include data on BCG vaccination rate, so DHS 2013 is used for that parameter. Concern about SMART: One organization has indicated to us that it does not include UNICEF SMART surveys from Nigeria in its analyses because it is generally agreed that these surveys' results do not line up with results from other surveys across a variety of health indicators, suggesting that they are not accurate. Once available, the recommended option will be to use data from New Incentives' program or pilot activities. | ||||||||||||||
27 | Year of analysis | 2020 | 2020 | 2020 | 2020 | 2020 | 2020 | 2020 | 2020 | 2020 | 2020 | We extrapolate trends in immunization and trends in child mortality to this year of evaluation. Immunization rates are increasing with time. Child mortality is decreasing with time. This extrapolation is likely to be less appropriate and less accurate for years further into the future. | ||||||||||||||
28 | Cap, vaccination rate without the program | 95% | 95% | 95% | 95% | 95% | 95% | 95% | 95% | 95% | 95% | We extrapolate trends in immunization to this year of evaluation. Immunization rates are increasing with time. This parameter caps how high extrapolated vaccination rates may go. | TODO: cap, vaccination rate with the program. FYI this is currently hardcoded in. | |||||||||||||
29 | By default, assume what effect of the program (percentage point increase in vaccination rate)? (Non-measles) | 30% | 30% | 30% | 30% | 30% | 30% | 30% | 30% | 30% | 30% | In the absence of data about the effect of the program, and subject to vaccine-specific vaccination caps, in general assume how much increase in vaccination rate (percentage points) due to the program? It is possible that we should have different expectations of program effect in different contexts: for example, in the North West, where baseline immunization rate is very low, there may be far more room (in terms of percentage points) for incentives to have an effect. On the other hand, it is possible that low immunization rates are due to factors that are not overcome by incentives. Users may wish to experiment with choosing the North West context and setting this value high -- e.g. 30% or 40%. | TODO: also mention NI's effect on facility delivery. Natalie wrote in slack: "in NI's facility delivery RCT midline, incentives increased facility delivery by 21 percentage points from 27% to 48% (significant). There was a smaller effect on (and smaller incentives for) bringing babies in for HIV tests weeks after birth: 32% to 46%, not stat. sig." | https://givewell.app.box.com/files/0/f/9360366589/1/f_78038662097 | Sato and Takasaki 2016. RCT of CCTs for adult TT in Nigeria. https://givewell.box.com/s/mlf39lhvntdyfpvmzw1cudikszue54xn | |||||||||||
30 | By default, assume what effect of the program (percentage point increase in vaccination rate)? (Measles) | 30% | 30% | 30% | 30% | 30% | 30% | 30% | 30% | 30% | 30% | See note above. As measles vaccination occurs 9 months after birth, and after a potentially longer period without contact between mothers and health workers, we expect it to be more difficult to have a large effect on routine measles immunization rate. Note that measles immunization is assigned a larger incentive than other vaccination visits. | ||||||||||||||
31 | ||||||||||||||||||||||||||
32 | Disease & Vaccine Parameter Calculations | |||||||||||||||||||||||||
33 | Average age at Visit 1, in weeks, without program | 2 | 2 | 2 | 2 | 2 | 2 | 2 | 2 | 2 | 2 | To update with empirical data. BCG is recommended at birth or as soon as possible after birth. Our understanding is that it is administered as late as 12 months, but not after 12 months. We are uncertain about at what age women in the New Incentives population present to vaccinate neonates with BCG. | TODO. Varying the date of Visit 1 is NOT supported at this time. In general, varying the age at immunization has not been stress tested and may lead to errors. | |||||||||||||
34 | Average age at Visit 2, in weeks, without program | 6 | 6 | 6 | 6 | 6 | 6 | 6 | 6 | 6 | 6 | To update with empirical data. Visit 2 is scheduled for 6 weeks. | ||||||||||||||
35 | Average age at Visit 3, in weeks, without program | 10 | 10 | 10 | 10 | 10 | 10 | 10 | 10 | 10 | 10 | To update with empirical data. Visit 3 is scheduled for 10 weeks. | ||||||||||||||
36 | Average age at Visit 4, in weeks, without program | 14 | 14 | 14 | 14 | 14 | 14 | 14 | 14 | 14 | 14 | To update with empirical data. Visit 4 is scheduled for 14 weeks. | ||||||||||||||
37 | Average age at Visit 5, in weeks, without program | 39 | 39 | 39 | 39 | 39 | 39 | 39 | 39 | 39 | 39 | To update with empirical data. Visit 5, measles vaccination, is scheduled at 9 months. | Duplication here is an model-building harmless mistake, difficult to track down right now without trace dependents. TODO. | |||||||||||||
38 | Average age at Visit 5, in weeks, without program | 39 | 39 | 39 | 39 | 39 | 39 | 39 | 39 | 39 | 39 | To update with empirical data. Visit 5, measles vaccination, is scheduled at 9 months. | ||||||||||||||
39 | ||||||||||||||||||||||||||
40 | Average age at Visit 1 (BCG) in weeks with program | 2 | 2 | 2 | 2 | 2 | 2 | 2 | 2 | 2 | 2 | To update with empirical data. BCG is recommended at birth or as soon as possible after birth. Our understanding is that it is administered as late as 12 months, but not after 12 months. We are uncertain about at what age women in the New Incentives population present to vaccinate neonates with BCG. | TODO. Varying the date of Visit 1 is NOT supported at this time. | |||||||||||||
41 | Average age at Visit 2, in weeks, with program | 6 | 6 | 6 | 6 | 6 | 6 | 6 | 6 | 6 | 6 | To update with empirical data. Visit 2 is scheduled for 6 weeks. | ||||||||||||||
42 | Average age at Visit 3, in weeks, with program | 10 | 10 | 10 | 10 | 10 | 10 | 10 | 10 | 10 | 10 | To update with empirical data. Visit 3 is scheduled for 10 weeks. | ||||||||||||||
43 | Average age at Visit 4, in weeks, with program | 14 | 14 | 14 | 14 | 14 | 14 | 14 | 14 | 14 | 14 | To update with empirical data. Visit 4 is scheduled for 14 weeks. | ||||||||||||||
44 | Average age at Visit 5, in weeks, with program | 39 | 39 | 39 | 39 | 39 | 39 | 39 | 39 | 39 | 39 | To update with empirical data. Visit 5, measles vaccination, is scheduled at 9 months. | Duplication here is an model-building harmless mistake, difficult to track down right now without trace dependents. TODO. | |||||||||||||
45 | Average age at Visit 5, in weeks, with program | 39 | 39 | 39 | 39 | 39 | 39 | 39 | 39 | 39 | 39 | To update with empirical data. Visit 5, measles vaccination, is scheduled at 9 months. | ||||||||||||||
46 | ||||||||||||||||||||||||||
47 | All-cause mortality | In the model below, we assume for simplicity that mortality that is not vaccine-preventable (all-cause mortality minus mortality due to diseases relevant to this CEA) is similar for vaccinated children compared to unvaccinated children. In reality, we expect that vaccination rate and mortality from non-vaccine-preventable causes are both correlated with factors such as poverty and distance from a health clinic. Additionally, note that we expect that there is some effect of co-infection on mortality; e.g. if a child infected with both malaria and measles is recorded as having died of measles, it is possible that the child would have survived if not for the malaria co-infection. Thus we expect that reducing disease incidence has some effect beyond reducing disease-specific mortality. The best type of evidence about the effect of vaccination is evidence about the effect on all-cause mortality. We have seen this type of evidence only for PCV; for other vaccinations we model disease-specific mortality. | ||||||||||||||||||||||||
48 | Percentage point annual decrease in neonatal mortality | 0.06% | 0.06% | 0.06% | 0.06% | 0.06% | 0.06% | 0.06% | 0.06% | 0.06% | 0.06% | Child mortality has been decreasing with time. We use the 2015 report of Levels and Trends in Child Mortality (pp. 22-23) in order to estimate a trend between 2000-2015. We expect that decreases in child mortality will level off nonlinearly. The graph of empirical data of under-5 mortality in Nigeria shown on p. 14 suggests a roughly linear and rather steep decrease in child mortality between 2000-2010. Because of our expectation of nonlinearity, our current estimate of mortality trend in Nigeria 2008-2018 was made as follows: for the North West province, which has higher child mortality, we assume that child mortality in this period continues to decreases at the national rate for 2000-2015 (in effect, we are assuming that child mortality decreases in the North West are similar to, but lagging, the country as a whole). For other states, and the national middle wealth quintile, we use Ghana's 2000-2015 rate of mortality decrease as a proxy for Nigeria's rate of decrease in 2008-2018. Ghana was selected as the nearby West African country with 2000 child mortality most similar (10.1%) to Nigeria's in 2015 (10.7%). Hence, we are assuming that after steeper earlier declines in mortality, Nigeria's trend line in the present levels off to be more like Ghana's in the near past. | ||||||||||||||
49 | Percentage point annual decrease in infant mortality | 0.23% | 0.23% | 0.23% | 0.23% | 0.23% | 0.23% | 0.23% | 0.23% | 0.23% | 0.23% | Child mortality has been decreasing with time. We use the 2015 report of Levels and Trends in Child Mortality (pp. 22-23) in order to estimate a trend between 2000-2015. We expect that decreases in child mortality will level off nonlinearly. The graph of empirical data of under-5 mortality in Nigeria shown on p. 14 suggests a roughly linear and rather steep decrease in child mortality between 2000-2010. Because of our expectation of nonlinearity, our current estimate of mortality trend in Nigeria 2008-2018 was made as follows: for the North West province, which has higher child mortality, we assume that child mortality in this period continues to decreases at the national rate for 2000-2015 (in effect, we are assuming that child mortality decreases in the North West are similar to, but lagging, the country as a whole). For other states, and the national middle wealth quintile, we use Ghana's 2000-2015 rate of mortality decrease as a proxy for Nigeria's rate of decrease in 2008-2018. Ghana was selected as the nearby West African country with 2000 child mortality most similar (10.1%) to Nigeria's in 2015 (10.7%). Hence, we are assuming that after steeper earlier declines in mortality, Nigeria's trend line in the present levels off to be more like Ghana's in the near past. | ||||||||||||||
50 | Percentage point annual decrease in under-5 mortality | 0.52% | 0.52% | 0.52% | 0.52% | 0.52% | 0.52% | 0.52% | 0.52% | 0.52% | 0.52% | Child mortality has been decreasing with time. We use the 2015 report of Levels and Trends in Child Mortality (pp. 22-23) in order to estimate a trend between 2000-2015. We expect that decreases in child mortality will level off nonlinearly. The graph of empirical data of under-5 mortality in Nigeria shown on p. 14 suggests a roughly linear and rather steep decrease in child mortality between 2000-2010. Because of our expectation of nonlinearity, our current estimate of mortality trend in Nigeria 2008-2018 was made as follows: for the North West province, which has higher child mortality, we assume that child mortality in this period continues to decreases at the national rate for 2000-2015 (in effect, we are assuming that child mortality decreases in the North West are similar to, but lagging, the country as a whole). For other states, and the national middle wealth quintile, we use Ghana's 2000-2015 rate of mortality decrease as a proxy for Nigeria's rate of decrease in 2008-2018. Ghana was selected as the nearby West African country with 2000 child mortality most similar (10.1%) to Nigeria's in 2015 (10.7%). Hence, we are assuming that after steeper earlier declines in mortality, Nigeria's trend line in the present levels off to be more like Ghana's in the near past. | ||||||||||||||
51 | Neonatal mortality (under 4 weeks; under 1 month) | 3.63% | 3.63% | 3.63% | 3.63% | 3.63% | 3.63% | 3.63% | 3.63% | 3.63% | 3.63% | 2013 Nigeria Demographic and Health Survey, p. 120 reflects mortality rate for the 10 years preceeding the survey (2003-2012). Child mortality has been decreasing with time. We use the 2015 report of Levels and Trends in Child Mortality (p. 22) in order to extrapolate child mortality (from an average year of 2008) into the year of evaluation. | ||||||||||||||
52 | Infant mortality (under 52 weeks; under 12 months) | 6.16% | 6.16% | 6.16% | 6.16% | 6.16% | 6.16% | 6.16% | 6.16% | 6.16% | 6.16% | 2013 Nigeria Demographic and Health Survey, p. 120 reflects mortality rate for the 10 years preceeding the survey (2003-2012). Child mortality has been decreasing with time. We use the 2015 report of Levels and Trends in Child Mortality (p. 22) in order to extrapolate child mortality (from an average year of 2008) into the year of evaluation. | ||||||||||||||
53 | Under-5 mortality (under 260 weeks; under 60 months) | 12.26% | 12.26% | 12.26% | 12.26% | 12.26% | 12.26% | 12.26% | 12.26% | 12.26% | 12.26% | 2013 Nigeria Demographic and Health Survey, p. 120 reflects mortality rate for the 10 years preceeding the survey (2003-2012). Child mortality has been decreasing with time. We use the 2015 report of Levels and Trends in Child Mortality (p. 22) in order to extrapolate child mortality (from an average year of 2008) into the year of evaluation. | ||||||||||||||
54 | Weekly mortality risk, first month (28 days) | 0.92% | 0.92% | 0.92% | 0.92% | 0.92% | 0.92% | 0.92% | 0.92% | 0.92% | 0.92% | |||||||||||||||
55 | Mortality between 1-12 months (period of 11 months) | 2.44% | 2.44% | 2.44% | 2.44% | 2.44% | 2.44% | 2.44% | 2.44% | 2.44% | 2.44% | If 7.1% of liveborn children die before 12 months, and 3.9% die before 1 month, then 3.2% of liveborn children die between 1 and 12 months. We renormalize this to be out of children who survive to 1 month. | ||||||||||||||
56 | Weekly mortality risk, 1-12 months | 0.05% | 0.05% | 0.05% | 0.05% | 0.05% | 0.05% | 0.05% | 0.05% | 0.05% | 0.05% | |||||||||||||||
57 | Mortality between 12-60 months | 6.50% | 6.50% | 6.50% | 6.50% | 6.50% | 6.50% | 6.50% | 6.50% | 6.50% | 6.50% | Same method as used for mortality between 1-12 months, above. | ||||||||||||||
58 | Weekly mortality risk, 12-60 months | 0.03% | 0.03% | 0.03% | 0.03% | 0.03% | 0.03% | 0.03% | 0.03% | 0.03% | 0.03% | |||||||||||||||
59 | ||||||||||||||||||||||||||
60 | Tuberculosis (disease) | Tuberculosis is an airborne bacterial infectious disease generally affecting the lungs. Most infections ("latent tuberculosis") do not have symptoms and do not transmit. Some latent infections progress to active disease. Interaction between tuberculosis and HIV is a significant factor in the epidemiology and mortality of tuberculosis. | ||||||||||||||||||||||||
61 | Tuberculosis: Chance of infection if susceptible, yearly | 1.15% | 1.15% | 1.15% | 1.15% | 1.15% | 1.15% | 0.15% | 1.15% | 1.15% | 1.15% | Trunz, Fine and Dye 2006, a meta-analysis and modeling study. This paper presents estimated annual chance of infection in several countries in east Africa (p. 1174) based on review of tuberculin surveys in children: 0.9 and 0.68 in Tanzania, 1.0 in Malawi, 1.1 in Kenya, 1.2 in Madagascar. The authors estimate annual chance of infection at 0.13-1.95 in Africa (low HIV) contexts, and 0.57-2.16 in Africa (high HIV) contexts. Believing that HIV prevalence in Nigeria is middling (https://www.cia.gov/library/publications/the-world-factbook/rankorder/2155rank.html), we take the middle of the entire range for our estimate of annual chance of infection in Nigeria. This is the chance of contracting tuberculosis; the chance of progression to active tuberculosis and death is represented in the case fatality rate. | http://www.thelancet.com/pdfs/journals/lancet/PIIS0140-6736(06)68507-3.pdf | |||||||||||||
62 | Tuberculosis: Chance of infection if susceptible, weekly | 0.02% | 0.02% | 0.02% | 0.02% | 0.02% | 0.02% | 0.00% | 0.02% | 0.02% | 0.02% | |||||||||||||||
63 | Tuberculosis: Case fatality rate | 27% | 27% | 27% | 27% | 27% | 27% | 27% | 27% | 27% | 27% | We are highly uncertain about tuberculosis case fatality rate, especially restricted to infant case fatality rate. We are highly uncertain about the time period represented by a case fatality rate; i.e. time between infection and death. We expect that case-fatality rate in the NI context depends on the extent to which infants successfully access diagnosis and treatment for tuberculosis. In their cost-effectiveness analysis, Trunz, Fine and Dye 2006 assume 100% case-fatality for childhood tuberculosis: "The costs per case and death prevented are regarded as the same, on the assumption of a case fatality rate of 100% as in the pre-antibiotic era, and in the absence of any more recent data for the outcomes of treatment." (p. 1176). We are skeptical that this is appropriate in the present Nigerian context. Dye et al. 1999, an expert panel review, estimate tuberculosis case fatality rate in Nigeria at 27% (p. 682), noting that their estimates reflect "the growing conviction among TB experts that CFRs have previously been overestimated. Inferior drug treatment may not definitively cure patients or significantly reduce transmission, but it will often prevent death. We have also assumed lower CFRs for smear-negative disease (average, 20%) than some previous authors." (p. 685) Coldiz et al. 1994, a meta-analysis of BCG efficacy studies (in developed countries; not restricted to infants/young children) found a 61% protective effect against TB death (51% reduction in TB cases). The 7 studies reporting on death were dated 1948-1974; we have not traced these studies or identified the case fatality rates in these studies; we doubt that these reflect the mortality patterns in current-day Nigeria. | http://www.thelancet.com/pdfs/journals/lancet/PIIS0140-6736(06)68507-3.pdf | http://jamanetwork.com/journals/jama/article-abstract/191271 | ||||||||||||
64 | Tuberculosis: Uncertainty adjustment for case fatality rate | 100% | 100% | 80% | 100% | 80% | 80% | 100% | 100% | 100% | 100% | We are particularly uncertain about tuberculosis case fatality rate (see above), and would not be surprised of our best-guess estimate were off by 50%. Users may choose to discount this parameter for weakness of evidence. Users may prefer to eschew this sort of more arbitrary discount in the model, and consider strength of evidence separately from cost-effectiveness. | ||||||||||||||
65 | Tuberculosis: Adjusted case fatality rate | 27% | 27% | 22% | 27% | 22% | 22% | 27% | 27% | 27% | 27% | |||||||||||||||
66 | Proportion of live neonates who are have been infected by and survived tuberculosis prior to Visit 1: BCG immunization | 0.00% | 0.00% | 0.00% | 0.00% | 0.00% | 0.00% | 0.00% | 0.00% | 0.00% | 0.00% | We are highly uncertain about the weekly chance of infection and the case-fatality of tuberculosis for infants prior to their first vaccination visit. It may be difficult to ascertain cause of death for young infants. Given that we estimate weekly chance of infection to be low, and we would guess that case fatality rate for neonates to be relatively high, we believe it is reasonable to consider this factor negligible for the purposes of this CEA. | ||||||||||||||
67 | ||||||||||||||||||||||||||
68 | BCG (vaccine against tuberculosis) | WHO recommends BCG vaccination as soon as possible after birth. New Incentives has told us that in clinics with which it works, BCG is often not given during facility delivery due to the need to open a new vial of vaccine; it is more efficient to open vials of vaccine when many infants can be vaccinated at once. Hence, women who deliver in a facility must return to the clinic post-birth in order to receive BCG. Most women do not deliver in a facility anyway. BCG is recommended for all infants under 12 months; BCG is not recommended for unvaccinated infants over 12 months. | ||||||||||||||||||||||||
69 | Vaccination rate without the program: BCG | 24% | 24% | 24% | 24% | 24% | 24% | 24% | 24% | 24% | 24% | 2018 IDinsight Baseline Report, Table 2a, p. 22," Immunization Coverage for 12 to 16-month olds across Katsina and Zamfara" | https://givewell.box.com/s/kc7mccqa3dbp00ock0en3m33364msp6y | |||||||||||||
70 | Vaccination rate with the program: BCG | 54% | 54% | 54% | 54% | 54% | 54% | 54% | 54% | 54% | 54% | This value should be updated based on preliminary evidence from New Incentives pilot sites. | ||||||||||||||
71 | Vaccine efficacy in preventing disease | 59% | 59% | 59% | 59% | 59% | 59% | 59% | 59% | 59% | 59% | Mangtani et al. 2013, a review of RCTs of the efficacy of BCG, found high variation in results. This review included the 4 identified RCTs of infant BCG effectiveness on TB cases identified in Coldiz et al. 1995, all taking place in the US/Canada 1933-1941, and also included a 1976 study in Bombay, India. (p. 473), for an overall risk reduction of .41 (random effects). We have not examined in detail why the effect estimate differs between Mangtani et al. 2013 and Coldiz et al. 1995, although we observe that efficacy in the Bombay trial was lower than in US/Canada trials. Colditz et al. 1995, a review and meta-analysis of newborn BCG efficacy, finds 74% efficacy in preventing infection based on four RCTs, and 65% efficacy in preventing death based on 5 trials. (abstract, p. 29) This analysis notes that "Three trials and six case-control studies provided some age-specific information that allowed us to examine the duration of BCG efficacy. Most of this evidence suggested that BCG efficacy may persist through 10 years after infant vaccination." (p. 29) As shown in Table 1, p. 31, the four RCTs reporting on TB cases were published in 1948-1961. While we do not necessarily expect BCG efficacy to have changed with time, we are concerned that these studies might reflect outdated reporting techniqies and study implementation. We have not individually reviewed these studies. | https://academic.oup.com/cid/article/58/4/470/347668/Protection-by-BCG-Vaccine-Against-Tuberculosis-A | http://pediatrics.aappublications.org/content/96/1/29.short | ||||||||||||
72 | Uncertainty adjustment: vaccine efficacy | 100% | 100% | 90% | 100% | 95% | 100% | 90% | 90% | 100% | 100% | Probabilty that the estimated clinical efficacy of the vaccine is accurate. We have not carefully revieved the evidence for vaccine efficacy. Evidence for the efficacy of common vaccines is usually strong. As noted above, the evidence we rely on for the efficacy of BCG is somewhat dated. Users may choose to discount this parameter for weakness of evidence. Users may prefer to eschew this sort of more arbitrary discount in the model, and consider strength of evidence separately from cost-effectiveness. | ||||||||||||||
73 | Vaccine efficacy, adjusted for uncertainty | 59% | 59% | 53% | 59% | 56% | 59% | 53% | 53% | 59% | 59% | |||||||||||||||
74 | ||||||||||||||||||||||||||
75 | Pneumonia (disease) | Pneumococcal diseases are diseases caused by the bacterium Streptococcus pneumoniae. The most common pneumococcal disease is pneumonia. "In developing countries, the disease is common in children under two years, including newborn infants; rates of the disease in the elderly population are largely unknown." There are 3 pneumococcal conjugae vaccines (PCV) licensed for use in children under 2 and marketed interenationally. They cover 7, 10, and 13 serotypes. Pneumococcal conjugate vaccines (PCVs) are considered safe in all target groups for vaccination, also in immunocompromised individuals. | http://www.who.int/immunization/topics/pneumococcal_disease/en/ | http://www.who.int/wer/2007/wer8212.pdf | http://www.who.int/bulletin/volumes/86/5/07-048769/en/ | |||||||||||||||||||||
76 | There is evidence about the effect of PCV vaccine on all-cause mortality, hence we are not modeling the intermediate steps of pneumonia infection and progression. | |||||||||||||||||||||||||
77 | ||||||||||||||||||||||||||
78 | PCV (vaccine against pneumonia) | We use PENTA coverage as a proxy. | ||||||||||||||||||||||||
79 | Vaccination rate without the program: PCV 1 | 21% | 21% | 21% | 21% | 21% | 21% | 21% | 21% | 21% | 21% | 2018 IDinsight Baseline Report, Table 2a, p. 22," Immunization Coverage for 12 to 16-month olds across Katsina and Zamfara" | https://givewell.box.com/s/kc7mccqa3dbp00ock0en3m33364msp6y | |||||||||||||
80 | Vaccination rate without the program: PCV 2 | 13% | 13% | 13% | 13% | 13% | 13% | 13% | 13% | 13% | 13% | 2018 IDinsight Baseline Report, Table 2a, p. 22," Immunization Coverage for 12 to 16-month olds across Katsina and Zamfara" | https://givewell.box.com/s/kc7mccqa3dbp00ock0en3m33364msp6y | |||||||||||||
81 | Vaccination rate without the program: PCV 3 | 6% | 6% | 6% | 6% | 6% | 6% | 6% | 6% | 6% | 6% | 2018 IDinsight Baseline Report, Table 2a, p. 22," Immunization Coverage for 12 to 16-month olds across Katsina and Zamfara" | https://givewell.box.com/s/kc7mccqa3dbp00ock0en3m33364msp6y | |||||||||||||
82 | Vaccination rate without the program: exactly 0 doses of PCV | 79% | 79% | 79% | 79% | 79% | 79% | 79% | 79% | 79% | 79% | Illustrative. These are not used in the model. | ||||||||||||||
83 | Vaccination rate without the program: exactly 1 dose of PCV | 8% | 8% | 8% | 8% | 8% | 8% | 8% | 8% | 8% | 8% | Illustrative. These are not used in the model. | ||||||||||||||
84 | Vaccination rate without the program: exactly 2 doses of PCV | 8% | 8% | 8% | 8% | 8% | 8% | 8% | 8% | 8% | 8% | Illustrative. These are not used in the model. | ||||||||||||||
85 | Vaccination rate without the program: exactly 3 doses of PCV | 6% | 6% | 6% | 6% | 6% | 6% | 6% | 6% | 6% | 6% | Illustrative. These are not used in the model. | ||||||||||||||
86 | Vaccination rate with the program: PCV 1 | 51% | 51% | 51% | 51% | 51% | 51% | 51% | 51% | 51% | 51% | To update with empirical data. | ||||||||||||||
87 | Vaccination rate with the program: PCV 2 | 43% | 43% | 43% | 43% | 43% | 43% | 43% | 43% | 43% | 43% | To update with empirical data. | ||||||||||||||
88 | Vaccination rate with the program: PCV 3 | 36% | 36% | 36% | 36% | 36% | 36% | 36% | 36% | 36% | 36% | To update with empirical data. | ||||||||||||||
89 | Vaccination rate with the program: exactly 0 doses of PCV | 49% | 49% | 49% | 49% | 49% | 49% | 49% | 49% | 49% | 49% | Illustrative. These are not used in the model. | ||||||||||||||
90 | Vaccination rate with the program: exactly 1 doses of PCV | 8% | 8% | 8% | 8% | 8% | 8% | 8% | 8% | 8% | 8% | Illustrative. These are not used in the model. | ||||||||||||||
91 | Vaccination rate with the program: exactly 2 doses of PCV | 8% | 8% | 8% | 8% | 8% | 8% | 8% | 8% | 8% | 8% | Illustrative. These are not used in the model. | ||||||||||||||
92 | Vaccination rate with the program: exactly 3 doses of PCV | 36% | 36% | 36% | 36% | 36% | 36% | 36% | 36% | 36% | 36% | Illustrative. These are not used in the model. | ||||||||||||||
93 | ||||||||||||||||||||||||||
94 | Efficacy of 1 dose of PCV against all-cause mortality | 3% | 3% | 3% | 3% | 3% | 3% | 3% | 3% | 3% | 3% | Arbitrary guess. We are highly uncertain about the effect of 1 or 2 doses of PCV. We note that Scott et al. 2010, a SAGE review of PCV schedules, notes (p. 3) that "The clinical relevance of differences in immunogenicity between PCV schedules is not known because of limitations in understanding the correlation between antibody concentrations and pneumococcal disease." | http://www.who.int/immunization/sage/1_SAGE_PCV_review_Executive_summary_v3_101020.pdf | |||||||||||||
95 | Efficacy of 2 doses of PCV against all-cause mortality | 12% | 12% | 12% | 12% | 12% | 12% | 12% | 12% | 12% | 12% | Arbitrary guess. We are highly uncertain about the effect of 1 or 2 doses of PCV. We note that Scott et al. 2010, a SAGE review of PCV schedules, notes (p. 3) that "The clinical relevance of differences in immunogenicity between PCV schedules is not known because of limitations in understanding the correlation between antibody concentrations and pneumococcal disease." | http://www.who.int/immunization/sage/1_SAGE_PCV_review_Executive_summary_v3_101020.pdf | |||||||||||||
96 | Efficacy of 3 doses of PCV against all-cause mortality | 16% | 16% | 16% | 16% | 16% | 16% | 16% | 16% | 16% | 16% | Cutts et al. 2005, an RCT of PCV efficacy in the Gambia found a 16% (per-protocol) reduction in all-cause mortality in infants under 30mo treated with 3 doses of PCV compared to placebo. See Table 5, p. 1144. This study was not designed to be powered to find a significant effect on mortality, but regardless found a barely-significant effect on mortality. We have not examined this study closely. We have not conducted a review searching for other evidence of PCV effectiveness. Cutts et al. 2005 is the source of efficacy evidence cited (p. 391) by Sinha et al. 2007, a cost-effectiveness analysis of adding PCV to the immunization schedule in Gavi-eligible countries. | http://www.thelancet.com/journals/lancet/article/PIIS0140-6736(05)71876-6/abstract | https://www.ncbi.nlm.nih.gov/pubmed/17276779 | ||||||||||||
97 | Uncertainty adjustment for PCV vaccine efficacy | 100% | 100% | 80% | 100% | 80% | 80% | 80% | 80% | 100% | 100% | Probabilty that the estimated efficacy of the vaccine is accurate. We have not carefully revieved the evidence for vaccine efficacy, and are currently basing this value on one RCT with a mortality endpoint. Users may choose to discount this parameter for weakness of evidence. Users may prefer to eschew this sort of more arbitrary discount in the model, and consider strength of evidence separately from cost-effectiveness. | ||||||||||||||
98 | Adjusted efficacy of 1 dose of PCV against all-cause mortality | 3% | 3% | 3% | 3% | 3% | 3% | 3% | 3% | 3% | 3% | |||||||||||||||
99 | Adjusted efficacy of 2 doses of PCV against all-cause mortality | 12% | 12% | 10% | 12% | 10% | 10% | 10% | 10% | 12% | 12% | |||||||||||||||
100 | Adjusted efficacy of 3 doeses of PCV against all-cause mortality | 16% | 16% | 13% | 16% | 13% | 13% | 13% | 13% | 16% | 16% |