026 - VII. Report on the Prophylactic Vaccination of 1536 Persons Against Acute Respiratory Diseases, 1919–1920

Considering the difficulties of previous prophylactic vaccine studies, especially during the 1918 flu pandemic, there was something of a movement to improve the quality of future studies. Two things were the focus in particular:

• previous studies had immunized during the pandemic, so there was a question of whether immunity had been established in time before the subject was exposed; it would be better to immunize long before the expected exposure, and
• previous studies' control groups hadn't always been very similar to the experimental groups in size or demographics, so it was difficult to compare the populations.

So some tried to do better. In the current study, vaccines were made to try to protect against all known causes of respiratory disease (which known causes, at the time, were all bacterial), such as pneumococcus. The subjects were employees at the Metropolitan Life Insurance Company (MetLife) in New York City. Those who volunteered, 1536 in total, received a vaccine; 1293 got three doses, the rest only one or two (probably due to side effects). There were another 3025 employees as controls, who got nothing. In order to ensure that the two groups were similar, the scientists administered questionnaires about personal history, home conditions, and habits, and the groups did seem pretty similar in sex, age, and habits.

In terms of side effects, there were the common local (redness, soreness) and general (headache, malaise) ones; a few people got fever and such, but nothing was very alarming.

For results, one impressive thing was that vaccinated employees were absent about half a workday less than controls, on average.

Other than that, the results were not very impressive: for almost all diseases, the inoculated group had more cases than the control group, sometimes more than twice as many. The main exception to this was pneumonia: there was one case in the vaccinated group (0.05%) in a girl who had only one of the three recommended doses, compared to 11 cases (0.29%) in the control group. That's a five-fold difference.

The authors said, "One might infer from the tabulated figures that it were wiser not to be inoculated." They make a good point, though, by reminding the reader that the vaccinated subjects had volunteered for it, and were thus more likely to be more susceptible to respiratory infections. Indeed, according to questionnaires, about 22% of the controls said had been fairly free of infection, while only 10.5% of the volunteers said the same.

So overall, it seemed that the vaccine helped against pneumonia and not much else, not surprisingly (since many respiratory infections are viral, not bacterial). This seems like a pretty good example of a negative study. With its accounting for previous history and attempting to ensure similarity between groups, this study seems better than the ones I've written about previously. Clinical trials seem to have taken a step forward at this point. But there was still no blinding, placebo, or randomization, so there's a lot of improvement yet to be made. Right now it's still pretty much correlational (an observation of people who choose to be vaccinated compared to those who don't). So we'll see.

Citation: Von Sholly, A. I. & Park, W. H. VII. Report on the Prophylactic Vaccination of 1536 Persons Against Acute Respiratory Diseases, 1919–1920. J Immunol 6, 103–115 (1921).

025 - Mills-Reincke Phenomenon and Typhoid Control by Vaccine

This article wasn't really about the safety or effectiveness of a particular vaccine, but it does relate to the question of whether it's worthwhile to vaccinate the whole population (and also another concern people sometimes raise), so I thought it was interesting.

First off, a definition: The Mills-Reincke phenomenon (and similarly, Hazen's theorem) was, hypothetically at least, the idea that typhoid fever could be prevented by cleaning up a population's water supply (since typhoid is spread through fecal contamination of water and such), and that for every typhoid death prevented by cleaning the water supply, several deaths from other causes would also be prevented by the same sanitation efforts. Presumably these other deaths might be caused by other water-borne infectious agents, or a general weakening due to any kind of water-borne infection.

At the time, public health officials in the US were considering vaccinating the population, or at least certain parts of it, against typhoid. The author of this paper warns against that, saying that this approach may mask the true problem (contamination of the water supply). Instead, water should be made as clean as possible, and typhoid cases may be used as an indication of contamination.

To support this position, he presents evidence from Alpena, Michigan, a town situated near Lake Huron. This town took its water from the lake and also dumped its sewage therein, and not surprisingly suffered from typhoid fever. But in the years between 1900 and 1918, a few changes occurred in the water supply: first, in 1907, the water intake was moved to a more polluted section of the lake; then in 1915, improvements were made in the sterilization of the water.

As a result of these changes, the annual mortality rates (that is, number of people dying each year) from typhoid or all causes changed. The average mortality from typhoid between 1908 and 1915 was higher by about 36 deaths per 100,000 people than the average between 1900 and 1907; in contrast, the mortality from all causes was higher by almost 150 deaths per 100,000. So for every one extra typhoid death, there were about 3 more deaths from other causes, in line with Hazen's theorem.

And when the water was cleaned better in 1915, there were about 4 fewer non-typhoid deaths for each one typhoid death prevented, though the data for this period is limited to only two years (1916 to 1917; the 1918 flu epidemic kinda skewed the all-cause mortality data for that year).

So it does seem that, given a choice between vaccinating against typhoid vs. cleaning up the water supply, the latter is the better option. Not surprising. Lest anyone conclude that this makes vaccines less valuable, though, the average mortality from typhoid between 1916 and 1918 was still about 13 deaths per 100,000 people, so there was still room for vaccinating to be useful. At least back then. Since we don't really do typhoid vaccination much in the US today, I suppose it turned out not to be helpful. This study doesn't say much about any other kind of vaccine though.

Citation: McGee, H. G. Mills-Reincke Phenomenon and Typhoid Control by Vaccine. Am J Public Health (N Y) 10, 585–587 (1920).

024 - Studies on Experimental Pneumonia V. Active Immunity Against Experimental Pneumococcus Pneumonia in Monkeys Following Vaccination with Living Cultures of Pneumococcus

This is very similar to the last post, by the same authors and in the same journal around the same time, but instead of vaccinating monkeys with dead bacteria, this time they vaccinated with living ones, subcutaneously (right under the skin).

The numbers of subjects in each experiment were still quite small, only a few each time. They tried two different strains of pneumococcus, one very virulent and the other very non-virulent (henceforth known as "attenuated"). Not surprisingly, vaccinating with the virulent strain often caused significant illness in the monkeys, sometimes fatal, though sometimes not very serious.

The attenuated strain seemed to be pretty mild in all cases though, only causing mild local reactions (in the two monkeys that received it).

Both live vaccines appeared to confer much more protection against pneumonia (transmitted the natural way, not injected) than the dead vaccines had. The monkeys even seemed to have some cross-protection against some other types of pneumococcus.

So, if one accepts the results of such tiny experiments, it seems that infection with live organisms, even minimal infection in the case of attenuated, induces better immunity than simple exposure to dead pathogen, at least in monkeys. And the researchers observed that immunity to pneumococcus didn't seem to correlate with the presence or levels of antibodies in the monkeys' blood. We'll see how these findings play out in the future.

Citation: Cecil, R. L. & Blake, F. G. Studies on Experimental Pneumonia V. Active Immunity Against Experimental Pneumococcus Pneumonia in Monkeys Following Vaccination with Living Cultures of Pneumococcus. J Exp Med 31, 657–683 (1920).

023 - Studies on Experimental Pneumonia IV. Results of Prophylactic Vaccination Against Pneumococcus Pneumonia in Monkeys

Disappointed by previous ambiguous results (here and here), Russell Cecil and colleagues decided to refine their attempts using model animals—Philippine monkeys, in this case. These animals show symptoms similar to those in humans when infected with pneumococcus, so it seemed like they would make a good model organism.

In a series of small experiments, the authors tried out different doses, types of vaccine, and routes of infection. Most of them had up to only 3 vaccinated monkeys (often with different doses of vaccine and/or infectious bacteria) and up to 3 controls. They tried making vaccines with bacteria suspended in vegetable oil or in salt solution; they tried high doses (the same amount as would be given to an adult man) or lower doses (like a human dose but proportional to the monkey's weight); and they tried infection by squirting pneumococcus into the monkeys' airways, or by putting them in a cage with other infected monkeys, or by injecting bacteria directly into the blood.

Overall, the results were disappointing: when exposed to bacteria by aerosol, almost all of the monkeys, vaccinated or not, got pneumonia. 42% of the vaccinated died, compared to 76% of the controls (that's something, at least, but not great at all). They tried testing for antibodies, but rarely found any.

The authors speculated about why the vaccine worked so poorly: possibly the dose was incorrect, or the strain used for infection was too virulent and overcame any immunity, or the monkeys were just too susceptible to pneumococcus. They pointed out that others had tried testing in apes the typhoid vaccine (so successful already in humans), and found no protection, so maybe monkeys aren't such a great model.

The one good and interesting result they got in this study was the observation that, when they vaccinated monkeys and infected them by infection directly into the bloodstream, they seemed to be protected much more effectively than when infected by aerosol. The two vaccinated monkeys survived with hardly any symptoms, while the one control died quickly. This fit in with results of the earlier experiments showing relatively sterile blood in the vaccinated monkeys (who nonetheless still suffered lung infections). The implication is that there is a distinction between humoral (that is, bloodstream-based) immunity and immunity in the lung (mucosal, perhaps?), and one did not necessarily translate into the other. That's an interesting observation, one worth considering in other experiments in which subjects may be exposed to infection via a route that doesn't accurately represent the normal route. On the other hand, there were only three subjects, so maybe it's meaningless. Too soon to say.

Citation: Cecil, R. L. & Blake, F. G. Studies on Experimental Pneumonia IV. Results of Prophylactic Vaccination Against Pneumococcus Pneumonia in Monkeys. J Exp Med 31, 519–553 (1920).

022 - Distribution of Vaccines and Serums by the State Laboratories

This is a short post about a short article, but I thought it was interesting, considering all the people who are suspicious of vaccines because they're made by huge greedy pharmaceutical corporations. See what you think.

C.A. Shore, the author, was the director of the North Carolina State Laboratory of Hygiene back in 1919. Back then, the lab had been trying to decide what its mission included: definitely diagnostic work, possibly some education and research (though those were secondary), but what about vaccine production?

Considering the success of typhoid vaccinations, which in 1913 had cost $2.50 per three injections (too expensive!), the NC lab started producing their own typhoid vaccines and distributing them to registered physicians for free! This worked so well that at least 1/5th of the state had been vaccinated since 1913. The bill for this was charged to the state, $15000.

That worked pretty well, so the lab expanded its production services to pertussis and smallpox vaccines and diphtheria antitoxin (antitoxin is a solution of antibodies that bind to and inactivate the bacterial toxin, used to treat people who are infected with toxin-producing bacteria). They could buy the antitoxin for $8 per 5000 units, but decided to make their own, and charge only $0.25 per package (regardless of whether it contained 1000 or 10000 units), just the cost of syringe and packaging. This small fee was to discourage waste that might occur if it were completely free.

The budget for all of this production and distribution was $25000 in 1918, but the lab hoped to get $35000 for the next year to be able to produce tetanus antitoxin and anything else that was proven to be effective.

Why couldn't the state do this kind of thing today, you may ask? Well...
1) For one thing, inflation: $15000 in 1919 was about the same as $209,000 today. Which still doesn't seem like a lot...
2) Also because producing vaccines in 1919, before very many pharmaceutical regulations were in place, would've been easier and cheaper. Just grow up the cells and kill them, in the case of typhoid. No  purification or recombinant technology or anything like that. But...
3) Nevertheless, some do! Such as this and this. Probably because, relative to other medical treatments, vaccines are cheap, and worthwhile.

Citation: Shore, C. A. Distribution of Vaccines and Serums by the State Laboratories. Am J Public Health (N Y) 9, 106–107 (1919).

020 and 021 - Results of Prophylactic Vaccination Against Pneumonia at Camp Wheeler

In terms of time, people, and purpose, this post is a continuation of 017: published a year later, performed by at least one of the same authors, and as a follow-up to the previous. Also subject to many of the same comments and criticisms.

So, the Cecil & Vaughan article is the study itself, and the other is basically a review of it with some policy recommendations. While 017 took place in Camp Upton, NY, this took place at Camp Wheeler, Georgia. This camp had been having serious trouble with pneumonia, and 34% of it was from pneumococcus types I-III, against which Cecil & Co could vaccinate. And they did.

The vaccine in the previous study was a suspension of killed bacteria in salt solution, which seemed to cause more irritation and side effects than the authors felt it should, so in this study they tried to improve it by instead suspending the bacteria in vegetable oil. Apparently others had good results with this method, decreasing the dose volume and number of injections while maintaining good immune response, with lower irritation to the recipient.

So Cecil and Vaughan gave this vaccine to a total of about 13460 men in the camp, leaving 3415 unvaccinated. It did seem to be less mild, with only 37 out of 100,000 recipients suffering the strange infiltrations that were common with the saline vaccine (1214 of 100,000). Only 7 in 1000 men had to stay in the hospital for a few days, and 77% of these had received another bacterial vaccine at the same time as the pneumococcal one. Still, this seemed more severe than it needed to be.

Now the results. I'm going to share the graph they put in the paper, since I can:

This shows that the vaccination rate rose over time (as they vaccinated more of the troops), and partway through they got a surge of new recruits. Then there was a spike of pneumonia in the unvaccinated troops, but much less in the vaccinated.

Here are graphs I made, similar to those in 017 except without a breakdown of different causative agents:

There were many fewer cases of both pneumococcus and overall pneumonia. Actual numbers were 42 of types I-III in the 3415 unvaccinated troops, vs. 8 in the 13460 vaccinated (actually there were 32 cases, but 24 of those happened within a week after vaccination, and work by another author showed that immunity is not present until 8 days after vaccination, so those were excluded).

Overall, for all types of pneumonia, there were 327 cases in 3415 unvaccinated, and 155 in 13460 vaccinated.

Death rates are proportionally similar:

Numbers for types I-III are 1 in 13460 vaccinated vs. 14 in 3415 unvaccinated.

This study was complicated by a few things, though, which make the results more difficult to interpret. Most importantly, the 1918 flu pandemic arrived about the middle of this study, making the risk of secondary pneumonia much greater. Second, the number of men in the camp was changing, with new recruits (more susceptible to pneumonia) coming in and other troops leaving, so it was difficult to say how much effect vaccination had on each group. Still, the populations did seem decently equivalent.

And so, like in 017, it's difficult to explain why the vaccinated troops seemed to be well-protected even from types of pneumonia against which they weren't vaccinated. The authors speculate there could be cross-protection, providing some protection against multiple different pathogens, but it is not clear. And again, how long the protection against pneumococcus might last after vaccination was not determined, so it might not be very long. So again, medium quality (relative to other studies I've written about so far, at least).

Russell Cecil seems to agree, acknowledging that his pneumococcus vaccine was not good enough for a population-wide vaccination effort (because of its limited cross-protection, still-too-unpleasant side effects, and the fact that pneumonia doesn't really cause awful outbreaks like smallpox did). But he did recommend it for certain high-risk populations, such as soldiers, miners, people in institutions, and those individuals that seemed especially susceptible. But I believe there is more improvement to be made.

Citations:
Cecil, R. L. Present Status of Pneumococcus Vaccine. Am J Public Health (N Y) 9, 589–592 (1919).

Cecil, R. L. & Vaughan, H. F. Results of Prophylactic Vaccination Against Pneumonia at Camp Wheeler. J Exp Med 29, 457–483 (1919).

019 - The Use of a Vaccine in the recent Epidemic of Influenza

Another interesting thing that was going on in 1918 is something you might have heard of: the worst influenza pandemic in recorded history, sometimes called Spanish flu, which killed up to 100 million people (~5% of the world’s population) in all parts of the world.³

It doesn't seem like the influenza virus had been discovered at the time, not until the 30s, but it did seem like health officials at the time understood that the nature of the disease was not bacterial. Nevertheless, they recognized that a bacterial respiratory infection and pneumonia often accompanied cases of influenza, and that such an infection was responsible for much of the mortality associated with flu.

So today’s study describes the arrival of 1918 flu in Winnipeg, Canada in October of 1918, and the attempt to create a vaccine against the bacteria that caused secondary pneumonia in patients.¹

After swabbing and sampling more than 100 cases, they determined which bacteria seemed most closely associated with the disease: some kinds of streptococcus, pneumococcus, and something called “Bacillus influenzae.” So they made a vaccine in the typical way at the time: grow these up, kill them with tricresol and heat, and inject people.

Then they inoculated soldiers in Winnipeg, 4842 of the 7600 present, leaving 2578 unvaccinated.

Of the vaccinated, about 6% were hospitalized with flu, 0.35% got pneumonia, and 0.1% died. Though of the 5 deaths, 3 had only been vaccinated after being admitted to the hospital, and the other two had been vaccinated 3 or 10 days before, so it might've been too late for them anyway.

Of the unvaccinated, about 9% were hospitalized with flu, 1.5% got pneumonia, and 0.62% died. The average hospital stay of the unvaccinated was twice as long as the vaccinated, and the captain of the soldiers said it seemed like the unvaccinated had more severe cases.

So that’s a 33% reduction in flu, 77% reduction in pneumonia, and 84% (or probably more) reduction in deaths from vaccination.

It seemed like the populations were mostly equivalent in exposure, except for some of the flu cases being from troops coming in on trains, rather than being in the camp, so that complicates things a bit.

Then there was another experiment, this time with civilians: 52,999 were vaccinated and another 85,941 not. About half who were received one dose, while the rest received two.

Number of doses didn’t affect number of flu cases much (9.7% for 2 vs. 9.8% for 1), but pneumonia cases decreased 18% and deaths 57%. Not super impressive.

But compared to unvaccinated, there seemed to be up to 61% reduction in flu cases, 77% in pneumonia, and 86% in deaths. And of doctors surveyed, 93% said the vaccine seemed to have an effect, and 86% said it seemed to help prevent fatalities in pregnant women. However, precautions were not taken to make sure these populations were equivalent, and the types of bacteria infecting people seemed to vary between different places and/or times, so it’s not a very good study overall.

"Contemporary medical opinion generally agreed that while the right vaccine would be invaluable, it probably had not been found yet...the consensus was that vaccine should only be used on an experimental basis, its results being unknown and perhaps dangerous."²

References:

1. Cadham, F. T. The Use of a Vaccine in the recent Epidemic of Influenza. Can. Med. Assoc. J. 9, 519–527 (1919).

2. McGinnis, J. P. D. The Impact of Epidemic Influenza: Canada, 1918-1919. Hist. Pap. 12, 120 (1977).

3. Wikipedia.

018 - Observations on Prophylactic Inoculation Against Pneumococcus Infections, and on the Results Which Have Been Achieved by It.

Now for the study that preceeded that of my last post (017) because I encountered that one before this one: this one was by the famous Almroth E. Wright who developed the typhoid vaccine and hated biostatistics. He undertook a fairly large vaccination program against pneumococcal pneumonia in South African mining camps.¹

Actually he tried vaccine therapy as well, but it didn’t work very well, so he could not recommend its practice. But the prophylactic  use of vaccines seemed much more effective. The study described 6 mass experiments of mining workers, each with a slightly-to-very different purpose.

In all of them, a vaccine made of killed pneumococci was injected into subjects.

The first was a study of 11634 men, about half of whom were vaccinated. These were enrolled and vaccinated, then hung around the camp for about a month, and then headed off to the mines. In that month, about 2.6% of the vaccinated and 3.5% of the unvaccinated caught pneumonia, and 83 out of 10,000 vaccinated vs. 153 out of 10,000 unvaccinated died. In total in the first 2 months after vaccination (one month in camp and one in mines), 1% of vaccinated and 2% of unvaccinated died of pneumonia. After the second month, the difference between groups was not so great; the number of cases in the unvaccinated group decreased until it was about the same as the vaccinated.

The second studied as many as 30,000 subjects, but the record-keeping was so bad that the data was unusable. Hopefully it helped some people survive pneumonia though. In the third, about 10,000 men were vaccinated and another 10,000 not, and vaccination reduced the morbidity (number of cases) from 2.05% to 1.1% and the mortality (number of deaths) from 38 in 10,000 to 21 in 10,000.

The fourth mass experiment was smaller but mainly similar to the first three; I’m not sure what set it apart. But the fifth was intended to determine the optimum dosing and growth medium for vaccine preparation, in which the authors found that they had been giving too small a dose. In terms of side effects, none of the doses produced worse side effects than a slight fever and malaise. So for the 6th experiment they increased the dose.

So in the sixth, there was a reduction in mortality of about 60%, from 1.77% unvaccinated dying to 0.7% vaccinated dying, and this was the best result of the six experiments. There is a summary table showing the trend in the camps over the three years this study occupied; as higher proportions of workers were vaccinated, the incidence, death rate, and number of working days lost decreased.

 The authors speculated that the decline in the difference between vaccinated and unvaccinated after the first 2 months could be due to the latter population gaining some natural immunity to the disease, rather than the former losing their protection, and that idea does seem to have some merit. They also point out that it’s quite possible that the rates of pneumonia in unvaccinated workers could be lower than they would be if not for the 50-90% of the population that were vaccinated; in a sort of herd immunity, the presence of vaccinated individuals could limit the spread of disease among the unvaccinated.

However, there are significant limitations to this study, as later reviews point out.  For example, the study I reviewed in the previous post points out that Wright neglected to distinguish between different strains of pneumococcus in his vaccine, thus confusing the interpretation.²

Other criticisms:

"Unfortunately, the results he obtained with pneumococcal vaccine did not convince the scientific community of its efficacy. The problem lay in the failure to include both pneumococcal serotypes known at that time and in the use of an inadequate vaccine dosage because of the discomfort associated with the injection of relatively large inocula of whole killed pneumococci."³

"Although the results of these trials indicated that the vaccine could prevent pneumonia, the many flaws in design, particularly the lack of statistical soundness and microbiological evaluations, precluded any conclusions."⁴

"The vaccines employed were of unknown composition with regard to the pneumococcal type or types included in them, and in retrospect contained too few organisms to be fully immunogenic to man. Although Wright, who had a profound antipathy toward biostatistics, concluded that the trials demonstrated the efficacy of vaccination, careful analysis of his data provides little support for this view."⁵

So overall, the results are not great, even compared to the study in my previous post. But better seems yet to come.

1. Wright, A., Parry Morgan, W., Colebrook, L. & Dodgson, R. W. Observations on Prophylactic Inoculation Against Pneumococcus Infections. and on the Results Which Have Been Achieved by It. The Lancet 183, 87–95 (1914).

2. Cecil, R. L. & Austin, J. H. Results of prophylactic inoculation against pneumococcus in 12,519 men. J. Exp. Med. 28, 19–41 (1918).

3. Watson, D. A., Musher, D. M., Jacobson, J. W. & Verhoef, J. A Brief History of the Pneumococcus in Biomedical Research: A Panoply of Scientific Discovery. Clin. Infect. Dis. 17, 913–924 (1993).

4. Bruyn, G. a. W. & Furth, R. van. Pneumococcal polysaccharide vaccines: Indications, efficacy and recommendations. Eur. J. Clin. Microbiol. Infect. Dis. 10, 897–910 (1991).

5. Austrian, R. The Development of Pneumococcal Vaccine. Proc. Am. Philos. Soc. 125, 46–51 (1981).