COVID-19, Vaccination Rates, Infection Rates, Death Rates, and Behavioral Interventions: It’s Time to Cut the Bull!


 Efficacy of SARS-CoV-2 (COVID-19) vaccines was assessed by comparing infection and death rates in the eight most- and eight least-fully-vaccinated German states. Infection and death rates were measured on three dates in fall 2021. Less-vaccinated states had substantially higher infection and death rates than more-vaccinated states. These effects occurred in the context of a time main effect and a vaccination rate x time interaction: Infection and death rates increased substantially over time, particularly in less-vaccinated states. These increased infection and death rates over time may be due to Germany having few measures other than vaccination in place for controlling the spread of COVID-19, despite the high community prevalence of the Delta variant. The results are discussed in the context of anti-vax rhetoric, policy, strategies for quelling the pandemic, the lack of a synergistic research relationship between the social and medical sciences, and issues related to vaccine equity.


Introduction
Whereas clinical trials, which constitute the gold standard for assessing vaccines' e cacy in preventing infection, have shown that COVID-19 vaccines signi cantly reduce the likelihood of infection (see, e.g., Polack et al., 2021), eld studies are important for several reasons. First, sample sizes in clinical trials are too small to reliably estimate e cacy for low-frequency outcomes such as hospitalizations and deaths. Second, it is important to examine how vaccines perform in the context of other factors that in uence the spread of COVID-19, such as the emergence of new, more infectious variants, differences in the nature of non-pharmaceutical interventions (NPIs) such as mask mandates or physical distancing requirements in different locales, seasonal changes that in uence behavior related to viral spread (e.g., cold weather driving people indoors), or relevant demographic features of different countries such as the median age of their populations (Centers for Disease Control and Prevention, 2021;Caruso, 2021;Coronavirus Cases, 2021;Kam & Reynolds, 2020;NewIndianXpress, 2021;Pratt, 2019;Melton & Sinclair, under review a). Even if more highly vaccinated groups retain relative protection compared to less-vaccinated groups as circumstances change, absolute rates of infection or other adverse outcomes may change with those changed circumstances in both the vaccinated and the unvaccinated.
We (Melton & Sinclair, under review a) have shown that the timing and scope of public health measures aimed at limiting COVID-19 infection rates, such as mask mandates, is very strongly associated with COVID-19 infection and death rates, which vary across countries by a factor of several hundred. In fact, we argued, it is likely that variations in social behavior, including both differences in governmental policy (e.g., border lockdowns, mask mandates, and physical distancing requirements) and differences in individuals' conduct (e.g., compliance with mask mandates), have a greater immediate impact on infectious disease rates than vaccines-and, moreover, dropping other policies aimed at controlling the spread of COVID-19 and relying exclusively on vaccines when there is still a very large unvaccinated segment of the population can be expected to back re and result in infection spikes even with moderately high vaccination rates. Thus, given the fact that the Delta variant of SARS-CoV-2 is more than twice as contagious as the original virus (and the Omicron variant is approximately twice as contagious as Delta), and the exponential nature of viral spread, it is unsurprising that countries that dropped most other protective measures have been hit hard by spikes in cases over the past few months. As we write this, several European countries are facing their highest infection rates of the entire pandemic, despite having the majority of their populations vaccinated.
Given that so many factors other than vaccines have such a substantial impact on COVID-19 case rates, effects of vaccines may very well be obscured in comparisons of regions differing widely in ways other than their vaccination rates (climate, median age, policy, culture, etc.) that substantially impact infection or death rates. As our own recent research [REDACTED, under review b] suggests, that appears to be precisely what happened in a recently published study by Subramanian and Kumar (2021), which found no differences in COVID-19 infection rates as a function of vaccination rates across 68 countries. This resulted, unfortunately but predictably, in their study being cherry-picked by anti-vaxxers as "proof" that vaccines do not work. Accordingly, to better assess the role of vaccination rates in rates of COVID-19 infections and deaths, we examined the relationship between the full vaccination rates and infection and death rates of the 16 German states during the current European surge. We expected that, given that we were examining different regions of a single country that does not cover a particularly large geographical area, does not have wide variation in relevant demographic factors such as age, has many of its public health policy decisions made on a national rather than state by state level, and is relatively culturally homogeneous, we would nd a substantial relationship between vaccination rates and COVID-19 infection and death rates, even though the difference between the vaccination rate of the most-vaccinated state (Bremen) and the least-vaccinated state (Saxony) was less than 20% at the time of our study.
We split the 16 German states into eight higher and eight lower full vaccination rate groups, and measured infection and death per hundred thousand rates on three dates in the fall of 2021. Given that Germany, like many other European countries, removed most NPIs in the fall (Johnson, 2021;Kekatos, 2021;Kingsley, 2021;Xinhua, 2021) and relied almost exclusively on vaccines to limit COVID-19 spread despite the prevalence of the highly infectious Delta variant, we expected a large main effect for time, with infection and death rates increasing for both groups of states. However, we expected the higher vaccination rate states to substantially outperform the lower rate states, and expected this difference to increase over time.

Methods
We performed a median split on the rates of fully vaccinated people in all 16 German states, resulting in eight less-vaccinated and eight more-vaccinated states (Bundesministerium für Gesundheit, n.d.).
[1] We measured infection and death per hundred thousand rates on October 18, November 18, and December 1, (New York Times, 2021. [1] Note that by having only eight cases per group, we are stacking the deck against ourselves. However, given that the differences between German states that impact COVID-19 rates other than their vaccination rates appear to be considerably smaller than what we (REDACTED, under review a) found to be the case for differences across countries that impact COVID-19 rates, we were con dent that we would nd large effect sizes. and .03, respectively) than in the lower-rate states (Ms = 0.07, .28, and .58 for times 1, 2, and 3, respectively; s = 0.03, 0.14, and 0.37, respectively). LSD tests indicated that all means differed at the p < .05 level except for the higher-rate time 2 and lower-rate time 1 means, which differed at the p < .10 level. In short, as was the case with infections, lower-rate states performed signi cantly more poorly than higher-rate states, and this difference increased over time. At the same time, both higher-rate and lowerrate states exhibited increased death rates over time, consistent with there being few mitigations in place other than vaccines in the context of a prevalent, extremely infectious variant.

Results
[2] All tests were 2-tailed unless otherwise noted. The same pattern of effects was obtained using regression analyses with vaccination rates as a continuous variable. We chose to present the data in the present manner both for ease of understanding and because the groupings of states (high vs. low) remained constant over time, whereas vaccination rates increased over time and it is di cult to say which date would be the most appropriate one to choose.

Discussion
Our results provided strong evidence that COVID-19 vaccines substantially reduce infection and death rates. We found signi cant main effects for vaccination rate: The higher-rate states had higher infection ( p 2 s = .43) and death ( p 2 s = .44) rates than the lower-rate states. We also found a signi cant rate x time interaction, with the gap between higher-rate and lower-rate states increasing over time for both infections and deaths, providing further evidence of vaccine e cacy. The large effects we obtained are striking given that we only had 16 data points for each date for our dependent variables, and given that the mean vaccination rate of the high-rate group (72%) was less than 9% greater than the low-rate group's (63.7%).
Thus, although a double-blind clinical trial with random assignment is a "cleaner" test of vaccine e cacy than comparing regions differing in vaccination rates, German states are su ciently similar in terms of demographics, governmental policies, and conduct-as well as geographically proximate-that the "signal" of vaccines' impact on infection and death rates generally comes through loud and clear above the "noise." In contrast, in a previous study [REDACTED, under review a] that compared countries with higher and lower vaccination rates, our effect size did not become strong until we covaried out other factors strongly associated with infection rates. There are many variables-cultural individualism vs.
collectivism, obesity, cultural tightness/looseness, the presence or lack of mask mandates, median age, etc.-that are related to COVID-19 infection and death rates and differ substantially between countries [REDACTED, under review a], making it more di cult to "see" the powerful impact of vaccines than it is in a country like Germany.
Although there is considerable evidence on the individual level (i.e., in comparisons of vaccinated and unvaccinated individuals) that COVID-19 vaccinations signi cantly reduce rates of infection (e.g., Lopez Bernal et al., 2021), transmission of the virus by infected individuals (e.g., Van Naarden Braun et al., 2021), and severe illness or death (e.g., Gri n et al., 2021), there has been comparatively little evidence at the population level (i.e., involving comparisons of regions differing in vaccination rates). Our results are an important addition to that body of evidence.
We also found signi cant main effects for time, with states' infection and death rates increasing substantially between the rst and last dates for which we collected infection and death rate data. The implications of this nding are in our view equally as important as our other effects. Germany, like many other countries in Western Europe (Johnson, 2021;Kekatos, 2021;Kingsley, 2021), eliminated most measures to limit the spread of COVID-19 other than vaccination once it reached a moderately high vaccination rate (65.3% fully vaccinated as of October 18; Ritchie et al., 2020).
[3] For instance, masks were not mandated in schools even though few children were vaccinated (Xinhua, 2021), and restaurant reservations were running close to pre-pandemic levels throughout the fall (OpenTable, n.d.). In short, whether because they deemed it better for corporate pro ts, because they overestimated the capacity of the vaccines to limit spread in the face of more contagious variants of the virus, because they failed to acknowledge that SARS-CoV-2 is an airborne virus (Greenhalgh et al., 2021), or for some combination of reasons, government o cials throughout the West relied almost exclusively on mass vaccination to control the spread of COVID-19, and largely reverted to a "business-as-usual (pre-pandemic)" state of affairs, whereas this is less the case throughout Asia and the Paci c Islands (REDACTED, under review a).
As we write this, the skyrocketing number of cases resulting from the dominance of the ultra-contagious Omicron variant, estimated to be the second-most contagious virus in history (Naquin, 2021), has made clear that this approach has failed miserably. Even Spain and Portugal, with more than 80% of their populations vaccinated, are having record numbers of cases.
Of course, community prevalence of infectious diseases depends on vaccination rates, as our data show.
We believe that it is an important milestone in the battle against the pandemic that many countries have mandated COVID vaccinations in many public indoor spaces, and some are mandating them for all citizens (Reuters, 2021), because a vital aspect of using vaccines to control the spread of infectious diseases is reaching a very high population level of vaccination, something unlikely to otherwise be achieved in countries with high levels of vaccine hesitancy. That will have to be achieved on a worldwide basis, which has not happened so far because pharmaceutical companies have limited access of poor countries to vaccines through their patent protections ("A patent waiver on Covid vaccines is right and fair," 2021). Steps are being taken to address this issue: Peter Hotez and his team at Baylor University have developed an inexpensive vaccine that is being made available on a not-for-pro t basis with no patent protections ("Texas Children's Hospital and Baylor College of Medicine covid-19 vaccine technology…", 2021), and China recently announced that it was making 2 billion vaccines available to other countries cheaply or for free (Wee, 2021).
But vaccines alone are not going to "solve" the pandemic. In fact, behavioral interventions are arguably more important for bringing infectious diseases under control than vaccines are. Evidence suggests that lockdowns (broadly de ned as measures to limit public indoor social contact) substantially decrease the number of new con rmed infections (Alfano & Ercolano, 2020), and countries or regions that implemented lockdowns shortly after their rst case (i.e., early 2020, prior to the availability of vaccines) had fewer deaths and hospitalizations than those that implemented them later (Amuedo-Dorantes et al., 2020;Huber & Langden, 2020). So successful were countries such as China and New Zealand at bringing COVID-19 infection rates under control through border controls and quarantines, effective contact tracing, closures of non-essential businesses (with adequate nancial compensation for those affected), and universal, mandatory mask-wearing that they have been able to live relatively normal lives, without fear of the virus, the majority of the time since these early lockdowns (Coronavirus Cases, 2021;Kam & Reynolds, 2020;NewIndianXpress, 2021). As a result, not only was their economic performance better (Cheng, 2021), but their rates of infections and death have been dozens or hundreds of times lower than many Western countries (Coronavirus Cases, 2021).
Ending a pandemic involving a virus as highly contagious as SARS-CoV-2 variants is a daunting challenge and will require both the steps toward near-universal vaccination outlined above [4] and a recognition that a panoply of infection control measures (e.g., Melton & Sinclair, 2021;Melton & Sinclair, under review b;Melton & Sinclair, under review c;Sinclair & Kumar, 2020) will be needed to rein COVID-19 infections in (Noh et al., 2020). History has shown that highly contagious and deadly diseases such as polio and smallpox can be brought under control. Whether and when we do so with COVID-19 is up to us: Let's cut the bull.
[3] This changed when the Omicron variant arrived.
[4] As well as updating vaccines designed to immunize against the original SARS-CoV-2 virus to re ect more recent, immunity-evading variants such as Omicron.