COVID-19 vaccine effectiveness in Malaysia

Basically, what it means to be vaccinated.

Saiful Safuan Md Sani
4 min readSep 2, 2021
Sihat Milikku — KKM

Malaysia started its vaccination programme on February 24, 2021. On August 22, 40% of its population have been fully vaccinated.

A recent post suggested that there is an important and rising proportion of COVID-19 cases who are fully vaccinated individuals — breakthrough infections. Figures provided were that this group made up almost a fifth of the total number of cases.

Are the vaccines NOT effective?

First Impression

Let’s pull data from Malaysia’s CITF — COVID-19 Immunisation Task Force who runs the National COVID-19 Immunisation Programme and figure this out.

Indeed, breakthrough infections comprised 19.5% of the country’s COVID-19 cases:

Data up to August 31, 2021.

And similarly for severe infections:

Data up to August 31, 2021.

However, these do not correctly inform us of how effective the vaccines are in reducing the number of cases and severe infections, or how ineffective they are, if vaccines are truly ineffective. These raw counts merely show the proportions relative to one another.

Adjust for Vaccination Rates

The first step to the right direction is to adjust for vaccination rates. We do this by normalising the raw counts relative to the population group (fully vaccinated group and unvaccinated/partial group). Since we are dealing with population, we usually compute per 100K or per million people in that group. Here, we will use per 100K.

e.g.

Here is the full calculation for COVID-19 cases:

And for severe infection:

Rate Ratio and Vaccine Effectiveness

The next proper step is to compute the rate ratio and vaccine effectiveness value. Comparing the rate of infection (or severe infection) in the unvaccinated/partially vaccinated individuals against fully vaccinated people, yields the rate ratio.

*Use corresponding data for severe infections

Vaccine effectiveness may even be more intuitive and is calculated as such,

*Use corresponding data for severe infections

So, the rate of COVID-19 cases is 3.5x higher in partially/unvaccinated individuals than fully vaccinated people, and the vaccines have averted 71.5% cases, which would have occurred without/partial vaccination.

The rate of severe COVID-19 is 4.1x higher in partially/unvaccinated individuals, and 75.8% of severe cases have been averted with vaccination.

Visualising Rate Ratio and Vaccine Effectiveness

The series of graphs below illustrate these two parameters. In the first two, cumulative raw counts of cases were used in the calculation.

MA7 7-days moving average.

Both parameters show lower values in the early part, suggesting misleadingly, poor vaccine performance. This illustrates the importance of having adequate data. Once data were adequate, the true values were observed.

The last two graphs below explored the parameters when daily raw counts were used to calculate them.

MA7 7-days moving average.

Vaccine Efficacy

The table above shows the vaccine efficacies found in clinical trials, of three of the vaccines used in Malaysia.

The Delta

In the weeks of May 3–17, 2021, the proportion of the Delta variant in Malaysia exceeded 5% and continued to rise. Between July 26-August 9, only the Delta variant were isolated — concluding that, in Malaysia, the Delta comprised 100% of cases.

Conclusion

The vaccines are effective and slightly more so in preventing severe COVID-19 infections in Malaysia. However, in comparison to vaccine efficacy found in clinical studies, vaccines effectiveness are lower, plausibly due to the rampaging Delta. Another crucial point: in comparing vaccine effectiveness between groups, use rate ratio rather than raw counts or proportions, as the latter would be misleading.

Caveats:

This analysis gives the apparent real-world values from real-world data. The variables are limited such that confounders, e.g. immunity obtained from natural infection, were not controlled. Furthermore, even natural immunity wanes, and that makes it a varying parameter which in turn will need to be estimated. A deeper analysis would certainly include stratification by age, gender, etc. A well-designed study will be required to be conducted to control for confounders and the performance of more comprehensive analyses. Business intelligence needed for decision-making, often for better or worse, cannot afford this in a timely fashion.

--

--