As the second wave of COVID is beginning to ebb, many in India are worrying about a future third wave. The best way we know to protect against the next wave is to vaccinate the population. In previous posts we explored (1) the progress being made to vaccinate India and (2) the supply of vaccines to the Centre and then states. In this post we will examine (3) the administration of vaccine doses, including the problem of wastage of doses.
Before we jump in, we repeat here our main take-aways from these three posts:
India is vaccinating at a daily rate that is short of what would be required to meet its stated goal of vaccinating 20% of the population by July.
There is a lot of variation in how different states are performing. Interestingly, some Southern states that are typically higher performing lag behind other states.
India does not appear to have enough planned supply through July to meet its vaccination target.
The supply of vaccines from the Centre to states is mainly a function of their population. It does not appear to be affected by political circumstances.
Even though supplies are constrained, states with greater demand seem to administer a greater share of the doses they are provisioned.
States with greater demand have also have less wastage via expiration of doses. Interestingly, we find evidence that administration of Covaxin is associated with greater wastage, consistent with news stories that the larger 20-dose vials of Covaxin increase the risk of doses going unused before the vial expires.
This leaves us with the following three recommendations.
India should increase the production or procurement of vaccinations. If this is not feasible, it should explore whether solutions, like de-prioritizing those people who already have anti-COVID antibodies and fractional dosing, would be effective at increasing the value of limited vaccine doses.
There may be value in switching from allocation to states in proportion to population to allocation to states (or even districts) in proportion to future risk-based so that we allocate limited doses as impactfully as possible.
Even though supply is constrained, efforts to encourage demand (via marketing or lotteries) will reduce the risk of wastage. India may also want to consider making it easier to access doses by increasing the number of vaccination sessions it holds at each clinic.
The government should focus on improving rural distribution chains and building more durable adult vaccine infrastructure. Because the risk of future mutation may require future shots with reformulated vaccines, overcoming these logistical challenges should be viewed as a long-term investment in public health.
Going forward, we will work to keep people abreast of vaccination progress through our Twitter bot @covidmetrics.
III. The distribution of vaccines to people
A. Distribution of available doses
Let’s take supply to states as given and consider the administration of that supply by states to their populations. We already reviewed the slow progress and large variance in distribution of supplied doses by states. What determines which states are doing better and worse? As we analyze the rate of administration, it is important to note that this quantity will reflect both how well states have handled logistics and how much demand there is for the vaccine.
We measure progress on vaccine administration by the total number of doses (or first doses) divided by the eligible population. We explore the relationship between administration rate and urbanization rate, rural health facilities, per capita income and COVID exposure using linear regression (Table 3).
Our main takeaway is that, despite supply constraints, demand likely plays an important role in distribution. We see that each 10% increase in state level per capita income is associated with 8.9% greater vaccine administration. This does not indicate that states are trying to vaccinate higher income persons more: we are not showing that higher income individuals within states are vaccinated more than lower income individuals within states. Instead, our regression compares states with higher income and those with lower income per capita. Nor is it likely this correlation mainly shows that more wealthy status are more equipped to distribute vaccines. There is surely some truth to that explanation, but we observe a correlation with income even after controlling for distribution capacity (i.e., controlling for rural health facilities and urbanization rate). Instead, we think the most plausible explanation is that wealthier states have greater demand for and thus take up of available vaccines.
Likewise, we find that each 10% increase in confirmed cases is associated with 4% less vaccine administration. This is initially a bit surprising. One might expect that greater COVID burden should either increase the urgency to state logistical activities or demand for available vaccines. But instead we find burden has a negative association with administration. Two possible explanations are that (a) states with high burden are occupied with lockdown or treatment rather than vaccination or (b) individual demand for vaccines falls with greater naturally-acquired immunity. We discount the first explanation because the negative relationship between burden and distribution was the same before the second wave and during it. Because the relationship does not change when lockdowns and current infections rose, we think it more likely that demand drives the relationship.
In particular, we think that the number of reported confirmed cases may have been interpreted by individuals as evidence that they already have immunity and thus reduced demand for vaccine-acquired immunity. We think reported cases matter because we do not see a similarly strong relationship between distribution and either total positivity rate or the number of people who were previously infected (accounting for serological studies).
Demand may be the primary driver of within-state distribution, but logistics also matter. A 10% increase in rural health facilities in a state is associated with a 3.4% greater administration rate.
Table 3: Determinants of vaccine distribution rates.
This table does not include Meghalaya. Meghalaya is a massive outlier in terms of population density. That said, the regression results do not meaningfully change if we include Meghalaya in the regression analysis.
B. Wastage of vaccine doses
Wastage is the loss of doses -- a reduction in supply -- due to expiration. We interpret wastage as a measure of how long it takes from receipt of vaccines to administration; the longer that period, the more likely a dose is to expire.
We use regression analysis to explore the main correlates of wastage. We draw three primary conclusions.
First, greater demand for vaccines reduces wastage. Wealthier states, i.e., states with higher per capita income, have significantly lower levels of wastage. We think this is likely a demand story: the wealthy are more likely to value and take up vaccines. It is unlikely to be evidence in favor of more capable states having less wastage because our analysis already controls for logistics (via rural health facilities and urbanization).
Second, we find some evidence that there is more wastage associated with the Covaxin vaccine. The hypothesized reason is that, while Covishield is supplies in 10 dose vials, Covaxin is supplied in 20 dose vials. An entire vial must be used in 4 hours. If one is unable to obtain enough arms to jab in this time period, then the remaining doses are wasted. Thus, Covishield, by lowering the demand for patients, lowers the risk of wastage. While our results are not significant, the coefficient is quite high. Each 1% increase in the share of vaccines administered that are Covaxin is associated with more than an 11% increase in wastage.
Third, better logistics is likely associated with lower wastage, but the effect is not statistically significant. More rural health facilities, which are critical to the vaccine distribution infrastructure, are loosely associated with lower wastage.
Table 4: Determinants of wastage
References
Asher , S., Lunt , T., Matsuura , R., & Novosad, P. (2021). DDL COVID India. Retrieved from Data Development Lab: http://www.devdatalab.org/
covid19india.org. (2021). Retrieved from covid19india.org
Government of India. (2021). Co-WIn. Retrieved from Dashboard: https://dashboard.cowin.gov.in
IDFC Institute. (2021). The Indian OVID-19 Alliance, TICA. Retrieved from https://www.idfcinstitute.org/projects/state-capacity/the-indian-covid-19-alliance-tica/
Ministry of Health and Family Welfare. (2020, October). Press Release: Dr Harsh Vardhan interacts with social media users during Sunday Samvaad-4. New Delhi.
Ministry of Health and Family Welfare. (2021, May 11). More than 18 crore vaccine doses provided to States/UTs Free of Cost by Govt. of India, so far. Delhi.
Ministry of Health and Family Welfare. (2021, May 17). More than 20 crore vaccine doses provided to States/UTs Free of Cost by Govt. of India, so far . Delhi.
Office of the Registrar General & Census Commissioner, India. (n.d.). 2011 Census Data. Retrieved from Census India: https://censusindia.gov.in/2011-common/censusdata2011.html