Table 2.
One-way sensitivity analyses of different COVID-19 vaccine characteristic and epidemic growth scenarios in South Africa.
| Parameter/value | SARS-CoV-2 infections averted, compared with no vaccination | COVID-19 deaths averted, compared with no vaccination | Years-of-life saved, compared with no vaccination | Change in health care costs, compared with no vaccination, USD | ICER, compared with no vaccination, USD per YLSa |
|---|---|---|---|---|---|
| Vaccine effectiveness in preventing SARS-CoV-2 infection, % | |||||
| 20 | 5,466,500 | 71,600 | 1,254,900 | −166,032,500 | Cost-saving |
| 40 (Base case) | 10,427,000 | 74,600 | 1,299,100 | −428,052,700 | Cost-saving |
| 50 | 12,758,000 | 77,500 | 1,349,700 | −554,501,500 | Cost-saving |
| 75b | 16,067,300 | 82,000 | 1,429,400 | −750,946,700 | Cost-saving |
| Vaccine effectiveness in preventing mild/moderate COVID-19, %c | |||||
| 29 | 8,310,500 | 74,000 | 1,298,900 | −377,101,700 | Cost-saving |
| 51 (Base case) | 10,427,000 | 74,600 | 1,299,100 | −428,052,700 | Cost-saving |
| 67 | 10,625,200 | 76,200 | 1,332,200 | −410,883,200 | Cost-saving |
| 79 | 10,722,500 | 75,300 | 1,316,800 | −399,131,600 | Cost-saving |
| Vaccine effectiveness in preventing severe or critical COVID-19 requiring hospitalization, %d | |||||
| 40 | 10,659,300 | 65,800 | 1,180,100 | −80,901,300 | Cost-saving |
| 86 (Base case) | 10,427,000 | 74,600 | 1,299,100 | −428,052,700 | Cost-saving |
| 98 | 10,690,200 | 77,500 | 1,341,700 | −545,358,200 | Cost-saving |
| Vaccine acceptance among those eligible, % | |||||
| 50 | 10,026,700 | 71,100 | 1,251,600 | −272,592,000 | Cost-saving |
| 67 (Base case) | 10,427,000 | 74,600 | 1,299,100 | −428,052,700 | Cost-saving |
| 90 | 10,562,000 | 79,200 | 1,360,000 | −526,334,700 | Cost-saving |
| Vaccination cost per person, USD | |||||
| 9 | 10,427,000 | 74,600 | 1,299,100 | −656,846,300 | Cost-saving |
| 14.81 (Base case) | 10,427,000 | 74,600 | 1,299,100 | −428,052,700 | Cost-saving |
| 25 | 10,427,000 | 74,600 | 1,299,100 | −26,778,000 | Cost-saving |
| 26 | 10,427,000 | 74,600 | 1,299,100 | 12,601,200 | 10 |
| 35 | 10,427,000 | 74,600 | 1,299,100 | 367,014,600 | 280 |
| 45 | 10,427,000 | 74,600 | 1,299,100 | 760,807,300 | 590 |
| 75 | 10,427,000 | 74,600 | 1,299,100 | 1,942,185,200 | 1500 |
| Re | |||||
| 1.1 | 2,640,400 | 6600 | 98,000 | 299,493,000 | 3050 |
| 1.4 (Base case) | 10,427,000 | 74,600 | 1,299,100 | −428,052,700 | Cost-saving |
| 1.8 | 5,955,700 | 110,500 | 1,957,700 | 129,359,500 | 70 |
| Two-wave epidemice | 13,696,300 | 62,700 | 1,072,500 | −682,063,700 | Cost-saving |
| Prior immunity to SARS-CoV-2, % of population | |||||
| 10 | 8,025,900 | 147,200 | 2,581,000 | 85,889,700 | 30 |
| 20 | 9,087,700 | 119,000 | 2,168,000 | 55,790,700 | 30 |
| 30 (Base case) | 10,427,000 | 74,600 | 1,299,100 | −428,052,700 | Cost-saving |
| 40 | 7,127,300 | 18,000 | 279,500 | −252,757,900 | Cost-saving |
| 50 | 608,300 | 1500 | 24,300 | 545,399,700 | 22,460 |
| Initial prevalence of active COVID-19, % of population | |||||
| 0.05%f | 12,247,900 | 70,300 | 1,269,000 | −557,621,500 | Cost-saving |
| 0.1% (Base case) | 10,427,000 | 74,600 | 1,299,100 | −428,052,700 | Cost-saving |
| 0.2% | 8,403,300 | 72,300 | 1,288,700 | −180,874,600 | Cost-saving |
| 0.5% | 6,028,800 | 64,100 | 1,119,800 | 51,633,800 | 50 |
ICER incremental cost-effectiveness ratio, Re effective reproduction number, USD United States dollars, YLS year-of-life saved.
aIn these scenario analyses, the reference vaccination program (67% supply, 150,000 vaccinations per day) is compared with no vaccination program under different scenarios. Displayed life-years and costs are rounded to the nearest hundred, whereas ICERs are calculated based on non-rounded life-years and costs, and then rounded to the nearest ten. Cost-saving reflects more years-of-life (greater clinical benefit) and lower costs, and therefore ICERs are not displayed.
bIn the scenario analysis of a vaccine with 75% effectiveness in preventing SARS-CoV-2 infection, the effectiveness in preventing mild/moderate COVID-19 disease was adjusted to avoid a scenario in which a vaccine has higher effectiveness in preventing infection than it does in preventing symptomatic disease.
cVaccine effectiveness in preventing mild/moderate COVID-19 (apart from severe/critical disease) has minimal impact on the number of deaths. Therefore, seemingly counterintuitive results are due to stochastic variability in the microsimulation. In the analysis of a vaccine that is 29% effective in preventing mild/moderate COVID-19, the vaccine effectiveness in preventing SARS-CoV-2 infection was adjusted to avoid a scenario in which a vaccine is more effective in preventing infection than in preventing symptomatic disease.
dVaccine effectiveness in preventing severe/critical COVID-19 itself has minimal impact on transmission and the number of infections. Therefore, seemingly counterintuitive results are due to stochastic variability in the microsimulation. In the analysis of a vaccine that is 40% effective in preventing severe COVID-19 requiring hospitalization, the vaccine effectiveness in preventing mild/moderate COVID-19 was adjusted to avoid a scenario in which a vaccine is more effective in preventing symptomatic disease than in preventing severe disease requiring hospitalization.
eIn the analysis of an epidemic with periodic surges, the basic reproduction number (Ro) alternates between low and high values over time, and the Re changes day-to-day as the epidemic and vaccination program progress and there are fewer susceptible individuals. For most of the simulation horizon, Ro is 1.6 (equivalent to an initial Re of 1.1). However, during days 90–150 and 240–300 of the simulation, Ro is increased to 2.6. This results in two epidemic waves with peak Re of ~1.4–1.5.
fWhen the initial prevalence of active SARS-CoV-2 infection is 0.05%, the epidemic peak occurs more than 180 days into the simulation. As our modeled time horizon only considers outcomes occurring through day 360, delaying the epidemic peak leads to a small decrease in the number of infections and deaths that are recorded in the scenario without vaccines. As a result, the absolute number of deaths prevented by vaccination decreases slightly as initial prevalence of active infection is changed from 0.1% to 0.05%, even though a greater proportion of deaths are prevented.