Vaccines-related Podcast Episode - BacterioFiles 196

This blog is not my only project, nor even my primary: I also have a podcast, called BacterioFiles, which in some ways is sorta the opposite of this blog. It is about how microbes (bacteria, viruses, archaea, fungi) are awesome and useful, whereas this blog is more about how they are deadly and unpleasant.

But sometimes, in happy coincidence, they overlap, and the latest episode was one of those times:

BacterioFiles 196 - Flagellin Facilitates Flu-shot Function

Gut bacteria are important for a good immune response to unadjuvanted influenza vaccines!

It's a much more recent study than the ones on this blog have been so far, but interesting context to keep in mind while reading (or in my case, writing) the entries for this blog, so I thought I would share. Enjoy!

085 - The clinical severity of diphtheria in certain cities in Great Britain

So what is the reason all these people were trying to develop all these vaccines? This study is one that provides some justification, which is that these diseases can be pretty deadly. Hedley D. Wright looked at the clinical severity (that is, case-fatality rates, number of cases that ended in death) of diphtheria in seven cities of Great Britain over a number of years.¹

So Wright looked at the data of diphtheria cases and deaths in Liverpool, Birmingham, Manchester, Glasgow, Leeds, Hull, and Edinburgh from 1911 to 1935. Keep in mind that diphtheria antitoxin/serum treatment had been around at least 20 years in this case.

One difficulty in such a study is the issue of reporting: you don't really see things if you're not looking. And diagnostic criteria matter too: if something is not diagnosed as diphtheria, it's not reported as such, even if it should be (the same is true in reverse). Wright estimated that there could be as few as 40% of actual cases being reported, so more than half went unreported. Since deaths are more likely to be reported, this means the case-fatality rate appears higher than it actually is. But it's still possible to use it to compare cities, and also to get an idea of deadliness by death rates overall.

And this is how it seems: the average case-fatality rates ranged from 6% (Hull) to 10.1% (Manchester): this means that of 100 people who got diphtheria, 6-10 of them died. That's not great. On the other hand, looking at death rates per 100,000 people (based on 1931 census data), the rankings are almost reversed: Manchester is the lowest, with about 10 people out of each 100,000 dying (so in a population of about 800,000 that's 80 deaths), while Hull has a rate of almost 14 (so in 300,000, that's 42 deaths). The highest rate is Liverpool, with 17 (in 850,000, that's 144-145 deaths). So it seems like Hull might have a pathogen that spreads better between people but is not as deadly, compared with Manchester.

There was change over time though. Early in the data, around 1915, there was a pretty bad epidemic, and then cases and deaths fell off. But in the last five years of observation, cases and deaths picked up again. This trend was broadly similar over the seven cities, though not exactly. The new epidemic seemed to have a lot more cases but not as much death. Possibly the ability to diagnose had improved, though not everywhere.

Also important was the analysis of different age groups. In almost every case, death and case-fatality rates were worse in younger children than in older. This is consistent with what we've seen before; youth makes this disease deadlier. So ages 0-4 have case-fatalities from around 10-20%, 5-9 years have 5-10%, and 10-14 have 1-5%. 20% is a pretty high severity.

There's some good news though: the deadliness for the youngest age group went down over the period in question. It started out around 20%, and went down to about 10%. However, in the 5-9 years group, it dropped initially, after the 1915 epidemic, and then rose again. So it seemed like the later epidemic was worse for older children.

Wright and others speculated that this pattern could be due to children going to school together more, or better treatment for younger patients somehow, or changing sizes of families. Not sure. But this is the kind of disease that is why people were trying to develop vaccines.

There aren't really many studies citing this one that I can quote from, but here's one that's also by Wright, sort of an update on the Liverpool situation.²

References:
1. Wright, H. D. The clinical severity of diphtheria in certain cities in Great Britain. Journal of Pathology and Bacteriology 49, 135–155 (1939).
2. Wright, H. D. Diphtheria in Liverpool during the years 1937-40. Journal of Pathology and Bacteriology 52, 283–294 (1941).

084 - A study in active immunization against pertussis

In the early 20th century, lots of people were working on developing a vaccine against whooping cough. Makes sense, because it was one of the biggest causes of death in young children. But arguably two of the most important of these researchers were Pearl Kendrick and Grace Eldering.

I've done a couple posts about studies by them before (072 and 079); the former of those was a progress report, and this post is the full report on the first large pertussis vaccine trial that they undertook in Grand Rapids, Michigan.¹

The study started in late 1933 and went for 44 months, following thousands of subjects. Soon after the progress report came out, another study (actually another progress report) came out that seemed to have negative results for pertussis vaccination (065), which motivated Kendrick and Eldering to be extra-careful in their own final report.

It was a pretty big effort, not just these two; many nurses and public health workers in Michigan were involved, though supervised by the authors. It took place in Grand Rapids, as I mentioned, and the final count involved 1,815 subjects in the vaccinated group and 2,397 unvaccinated controls. These were children with no history of whooping cough (so, presumably susceptible) that lived nearby and could be followed over the course of the study.

Not all of them remained in the study for the whole 44 months, of course, because once they actually caught pertussis, they were presumably not susceptible anymore. Or if they moved away or something. Or if they grew out of (or into) the age range, which was 8 months old to 5 years old. But the results were corrected for how long each was followed.

The groups were selected by families presenting themselves at clinics to receive the vaccine. These were the vaccinated subjects; others were selected from the same districts as controls. So it wasn't randomized or blinded at all, which is a limitation. Kendrick and Eldering recognized this, and took special care to try to make the groups as equivalent as possible:

-The average time they were part of the study was 15 months for vaccinated, 11.6 months for controls. This was corrected for though, and could be due to some of the controls getting vaccinated and thus being removed from the study, which is something that couldn't happen to those already vaccinated. The proportions that moved away weren't significantly different.
-The proportion of each sex in each group wasn't significantly different.
-The proportions of ages weren't significantly different.
-The proportions in each district of the city weren't significantly different.
-The proportions of family sizes weren't significantly different.
-The proportions of each group getting measles and scarlet fever were equivalent, so it didn't seem like either was healthier or less exposed than the other.
-The average interval between nurse visits for each group was the same.
So overall, the groups seemed equivalent, at least in these characteristics.

As mentioned in 072, the vaccine was made of freshly isolated and lab-cultured bacteria, killed with phenol and/or merthiolate (AKA thimerosal) in small amounts. It was produced continuously on small scales, so none of it got older than about a year. It was injected under the skin of the arms.

In terms of reactions, most of the ones they observed were local—soreness, etc.—and only slight otherwise. One report of the 1815 was of convulsions, and 2 had high fever and vomiting, though it is only correlative because there wasn't a placebo control. Mostly it seemed ok.

Finally, a word on diagnosis and severity ratings: diagnoses were made based on cough plates (that is, culturing the organism), clinical symptoms, and history of exposure. Severity was rated somewhat arbitrarily, based on frequency of whooping and vomiting or the occurrence of complications/weight loss.

Results
The results were corrected for amount of time each subject was participating in the study, and how many subjects there were in each group, so they're reported as annual attacks per 100 subjects. So with that in mind, the incidence of whooping cough overall was:
2.3 annual attacks per 100 vaccinated subjects
vs.
15.1 annual attacks per 100 controls.
This is a significant difference, which could be expected to occur by random chance only once if they repeated the trial millions of times.

Compared to other reports of the rates of pertussis in Grand Rapids, the control group followed the same up-and-down trends, but had a higher incidence, probably because the closer observation detected cases that would've gone undetected otherwise, due to low severity. And speaking of severity:

Even of the vaccinated subjects that did get whooping cough, the severity was much lower. 73% were rated as light or very light severity, compared to 27% in the controls. And only 4% were severe in the vaccinated, and these didn't have serious complications, only frequent coughing/vomiting; that's compared to 13% severe in the controls. And considering that the "very light" cases were questionable about whether they could even be considered cases at all, by removing them from both groups, the difference in incidence increases even more.

Kendrick and Eldering also looked specifically at their data on known exposures of subjects to the disease, and found similar patterns. The vaccinated group actually had a higher number of exposures than the controls, but many fewer cases from them. Calculating the number of cases expected (based on the cases in the controls and the exposures in the vaccinated), it appears that the vaccine prevented about 81.3% of cases in the test subjects. Not excellent, but pretty good.

The differences were not as big when the exposures were more intimate, like within a household. The attack rates were about 35% vaccinated vs. 90% controls. Still, that's significant protection.

Another thing they controlled when watching exposures was coughs that weren't diagnosed as pertussis, of which there were more in the vaccinated group. These were mild, taking place right around an exposure incident, and could be classified as slightly less than "Very light" whooping cough. So if they added these coughs to the numbers, incidences overall would be 24% vaccinated vs. 72% controls, which means only 67% protection. Which is still pretty good, especially considering that these extra cases are barely cases at all.

So overall, this study is very good compared to others at the time or before, though not quite good judging by modern standards: not randomized, not placebo-controlled, though more controlled than it could've been. And the protection seemed pretty good, though arguably it would be better to have higher complete protection rather than just reducing the severity of cases.
It isn't a study that looked at the duration of immunity, or whether it reduced transmission at all (thus providing herd immunity), or if the vaccine was safe, especially in the long run, or if it could be effective in more than just this population. But what it does look at is how well it protects young children over at least a few years after it's given.

Oh, and it's not funded by any pharmaceutical company.

It's hard to make any judgments overall, so I'll just cite some comments from later publications that cited this one, both positive and negative:

"There is ample evidence in the literature now that individuals inoculated with suitable doses of a proper vaccine have a high degree of immunity against whooping cough."²

"The Sargent-Merrell method of evaluating the success of an immunization program has been applied to data covering a 6-year period in the city of Grand Rapids...the proportion of cases prevented is 84%. The validity of the result has been verified on the basis of controlled field data."³

"The controversy dates back to the first trials of pertussis vaccines, which were carried out during the 1930s. These were criticized as biased in favor of the vaccines because they were not randomized; vaccinated volunteers were compared with unvaccinated 'nonvolunteers.'"⁴

"Methodologically, the original field trial design was flawed. The experimental group was self-selected and only control subjects were randomly chosen. Despite careful attention paid to case detection and diagnosis, 1603 observations from the study's early years had to be excluded from the final analysis. Several featuers of the trial nonetheless make it an important contribution, not simply to the development of an effective pertussis vaccine, but to the history of controlled trials: ... 2) The trial was unusual for the level of attention given to case diagnosis and follow-up, and to the discussion of unknown factors which might have biased the results; 3) a similar level of detail was given in reporting the analysis and the methodological limitations of the field trial"⁶

Also, if you want a really detailed historical account of this study and everything that went into it, before and after, check out Shapiro-Shapin 2007.⁵

References:

1.

Kendrick, P. & Eldering, G. A study in active immunization against pertussis. Am. J. Hyg. 29, 133–153 (1939).

2.

Cohen, P. & Scadron, S. J. The placental transmission of protective antibodies against whooping cough: By inoculation of the pregnant mother. JAMA 121, 656–662 (1943).

3.

Weiss, E. S. & Kendrick, P. L. The Effectiveness of Pertussis Vaccine: An Application of Sargent and Merrell’s Method of Measurement. Am. J. Epidemiol. 38, 306–309 (1943).

4.

Fine, P. E. M. & Clarkson, J. A. Reflections on the Efficacy of Pertussis Vaccines. Reviews of Infectious Diseases 9, 866–883 (1987).

5.

Shapiro-Shapin, C. G. ‘A Whole Community Working Together’: Pearl Kendrick, Grace Eldering, and the Grand Rapids Pertussis Trials, 1932-1939. Michigan Historical Review 33, 59–85 (2007).

6.

Marks, H. M. The Kendrick-Eldering-(Frost) pertussis vaccine field trial. J R Soc Med 100, 242–247 (2007).

083 - Immunity and susceptibility to disease in early infancy

Similar to the last poat (082), this one is about how newborns are surprisingly immune to certain diseases, because they receive disease-targeting antibodies from their mothers. Charles McKhann and Israel Kapnick wrote this review, and sadly there's no new data, but it's a good summary of the topic (for the time).¹

So most interesting is that newborns are immune to several important diseases—measles, scarlet fever, polio, and diphtheria—for up to half a year after birth. This is especially true if their mothers had immunity to those diseases. The duration is impressive because when physicians attempted to produce immunity in people by injecting them with serum from recovered patients (so-called passive immunity, because it is not derived from the patient's own immune system), it only lasted a few weeks at most. Somehow the infant is able to maintain the passive immunity from the mother.

This is not the case with every disease though. Stuff that causes fevers or gut infections and diarrhea, and whooping cough, are not prevented in these newborns quite so well. Other things are intermediate, like chickenpox and pneumonia.

The duration of the immunity might seem to suggest that the antibodies are coming repeatedly from the mother, perhaps from breastmilk, but McKhann and Kapnick say that seems only to happen in cattle, not humans. The evidence seemed to show that infants' passive immunity came through the placenta in the womb. Though it's possible that some components of immunity come through breastmilk and others don't. But it seems possible that antibodies from the placenta get stored up somewhere in the infant.

So the importance of this study as pertains to vaccines is two-fold: first, as discussed in the last post, it may be possible to make the infant immune for some time by vaccinating the mother. Apparently this is pretty helpful with tetanus, as it prevents neonatal tetanus which is common in places in Africa.

Second, it is important for determining when the first vaccines should be given. With stuff like measles, which is an attenuated live virus, if there are already antibodies present, the vaccine won't produce as much of a good response. McKhann and Kapnick recognized this with certain things. So knowing when infants will become susceptible is important for immunizing them at the right time, not too early or too late. For some things, this may be late in the first year of life.

However, immunology is pretty complicated, with things we didn't understand until recently, and things we still don't understand. So it's not clear (to me, anyway) how much of this review is accurate. A later publication made these comments about older reviews, including this one:

"A large and useful part of the data bearing on the immunology of the newborn infant has come from clinical studies of immunizing schedules. There have been many careful reviews of this subject [like this one]...[But] quantitative interpretation of these studies and separation of the factors involved has often been difficult for reasons such as the following: 1. Antigen-antibody reactions may have been used for which there were no accurate methods of titration. Only rough, qualitative conclusions could be drawn. 2. Some studies used antigen-antibody reactions that require complement. The blood of the newborn infant has a low level of complement. This may have caused an undetermined error in the antibody estimation."²

I expect to have more to say in later posts.

References:
1. McKhann, C. F. & Kapnick, I. Immunity and susceptibility to disease in early infancy. The Journal of Pediatrics 13, 907–918 (1938).
2. Osborn, J. J., Dancis, J. & Julia, J. F. Studies of the Immunology of the Newborn Infant 1. Age and Antibody Production. Pediatrics 9, 736–744 (1952).

082 - An Attempt to Increase Resistance to Pertussis in Newborn Infants by Immunizing Their Mothers During Pregnancy

With vaccine-preventable diseases, young children are often most at risk of serious health impacts or death; the younger, the higher the risk. At least, this is true of whooping cough, which had mortality rates of 26-55% in the 1930s among infants less than a year old.

However, some observed that newborns up to six months old seemed to have some resistance to some diseases—diphtheria, polio, measles, and scarlet fever, for example—especially when the mother had resistance of her own, such as immunity from having had the disease. So it seemed like the mother was transferring her immunity to the infant, probably through the placenta.

So John Lichty, Betty Slavin, and William Bradford thought it might be wise to take advantage of this transfer to give newborns more resistance until they could be vaccinated themselves around 6 months. In this study, they try immunizing mothers during pregnancy and then observing the immune response in mother and infant.¹ This was building on previous work in humans and animals with the same or other diseases, to some extent.

So they selected healthy women with normal pregnancies in obstetrics departments of Rochester hospitals, assigned them randomly to be immunized or be a control, using Sauer's whole-cell pertussis vaccine from Eli Lilly. No placebos, so no blinding of patients. They did separate observations of mothers and infants based on whether mothers had had pertussis before; i.e. history or no history. So they had four study groups: no history or vaccine, history but no vaccine, vaccine but no history, and both history and vaccine.

The way they measured immunity was a bit unusual: opsono-cytophagic index. They took blood from subjects, mixed it with dead pertussis bacteria, and observed how many dead cells the white blood cells gobbled up. They compared subjects based on the number of white cells that ate at least 20 dead bacteria; the "index" value. Presumably the immune status would affect how likely the white cells were to eat the bacteria. There was blinding in this test somewhat: the examiner counting the index didn't know the status of the subject from whom the blood was taken, so as not to be biased in counting.

Results

In total, there were 28 women immunized and 22 as controls. They observed in most groups, most infants had a lower index than their mothers; the exception was the vaccine+history group, in which a third of infants had a higher index.

I made a graph showing the values for the four groups, mothers and infants (in mothers' cases, after the vaccine, when relevant):

Opsono-cytophagic index for mothers (post-vaccine) and their infants; error bars are standard deviations reported in the study.

Overall, looking at the error bars, it doesn't seem like there's much significant difference anywhere. But looking at trends, two things stand out: infants from mothers with no history or vaccine seem lower than from mothers with either, and with both it's highest. Second, the difference between mothers and infants is largest with neither history nor vaccine, lower and similar for history or vaccine, and mothers and infants are closest with history plus vaccine.

The authors looked at a couple other things too. They observed some of the infants before they had nursed and then again after nursing for one week, to see if the colostrum affected the immunity at all. It didn't seem to make a difference.

Secondly, they looked at some clinical data for other patients; specifically, of 31 infants that died from pertussis. Eighteen of them died before 6 months of age, and 13 after. Of those that died younger, 28% of their mothers had had pertussis before giving birth; of those that died older, the number was at least 54%. Sample sizes were pretty small, but it suggests that mother's immunity does have a protective effect for the infant up to 6 months. Seems like just correlation though.

The authors concluded that vaccinating mothers seemed to help. Comparing each infant's index to mother's, the group with neither immunity had only 50% the index in infants compared to the mother; with either vaccine or history, that number went up to 75%; and with both, 100%, almost identical index. So perhaps an additive effect.

Overall, not a very rigorous study, but suggestive. Later articles were somewhat critical of the study, perhaps explaining the weak results:

"Lichty, Slavin, and Bradford attempted, as they put it, to increase resistance against pertussis in newborn infants by immunizing the mother during pregnancy. They confessed their failure. An analysis of the data revealed the following facts: The injections were given at two week intervals in the last six weeks of pregnancy. The total dose administered was 20-25 billion [cells]. Thus the dose was inadequate and too late for antibody formation which reaches its climax between one and two months after the last inoculation. The test for immunity which they employed, cytophagocytosis of the blood, has distinct limitations and has been abandoned by them in favor of mouse tests. Their figures showed no increase in cytophagocytosis of the inoculated mother's blood. Granted the validity of the test, they found no increased immunity in the mother, so that there were no antibodies transferable to the baby through the placenta."²

Even later, though, most studies citing this one seemed to focus on the safety aspect (which I forgot to mention above): of the mothers in the vaccinated group, almost the only side effect was a sore arm that wasn't bad enough to interfere with daily life. One woman had a systemic reaction with nausea and vomiting. Here's an example of a mention:

"Although phase 1 studies of maternal immunization with Tdap are in progress, studies many decades ago with whole-cell pertussis vaccine administration late in pregnancy resulted in high levels of pertussis-specific antibodies in infants and no safety concerns."³

Sometimes I wonder if people actually read old studies before citing them, but I guess usually it doesn't make much difference.

References:

1. Lichty, J. A., Slavin, B. & Bradford, W. L. An Attempt to Increase Resistance to Pertussis in Newborn Infants by Immunizing Their Mothers During Pregnancy. J Clin Invest 17, 613–621 (1938).

2. Cohen, P. & Scadron, S. J. The placental transmission of protective antibodies against whooping cough: By inoculation of the pregnant mother. JAMA 121, 656–662 (1943).

3. Healy, C. M., Rench, M. A. & Baker, C. J. Importance of Timing of Maternal Combined Tetanus, Diphtheria, and Acellular Pertussis (Tdap) Immunization and Protection of Young Infants. Clin Infect Dis. 56, 539–544 (2013).

081 - Measles in Detroit, 1935 I. Factors Influencing the Secondary Attack Rate Among Susceptibles at Risk

Before you can say much about a disease or treatments for it, it's useful to actually understand how it works. One aspect of that is attack rates: When a number of susceptible people are exposed to the disease, how many actually get it?

This was discussed for measles a bit in entry 060, concluding that about 95% of people in cities caught measles by age 15. Here, Franklin H. Top looked at attack rates on a smaller scale, following a number of families in Detroit that experienced one or more cases of measles and seeing how many other susceptible individuals in each family caught it.

Measles is one of the most contagious diseases we know about; this figure compares it to other diseases we know about:

Source: NPR

This is probably why it's showing up so much in the US, despite the very effective preventative measure we have (vaccination).

Franklin had observed high attack rates in Detroit, so much that he says, "parents often consider it unfortunate when all susceptible children in the family do not take the disease at one time, reasoning that it is more convenient to be done with it at once."

So he looked at more than 27,000 cases reported in Detroit in 1935, focusing on families that didn't have any kind of prophylaxis against measles (not that there was anything particularly effective, but we'll talk about that in a later post). He selected a fifth of these families for study, excluding those with no susceptible children or with difficulty in keeping records.

Then he looked at the influence of the sex and age of the primary case of measles, the sex and age of the susceptibles, the period in the seasonal measles cycle in which the case happened, the number of primary cases in a family, and the frequency and intensity of exposures, to see if any of these influenced the attack rate.

Some definitions: the primary case is the first case of measles in a family, so the exposure must've come from somewhere outside. Secondary cases presumably came from the primary case, infecting susceptible contacts in the family. It was assumed (based on previous study) that each case was contagious for 8 days: the exposure period.

There were 1,253 families in the study with one primary case, and 1,380 secondary cases resulting from them. The effect of the sex of the primary case was negligible: the attack rate from male cases was 84.3%, and 84.7% from females. Broken down by age groups, sometimes the male rate was slightly higher, sometimes female.

Looking at age groups of the primary case specifically, cases aged 5-9 years had by far the most susceptible contacts. Those under 1 year or over 15 had too few to make good conclusions. But in the middle, the attack rate was consistently 70-90%. So it didn't seem to matter much; if anything, 5-9 years was the highest, but not statistically significant.

For secondary cases, again sex didn't really matter (84.4% vs. 84.7%). In this case though, age did, a little: contacts between 1-9 years had significantly higher attack rates than those under 1 year, about 85% vs. 45%. I wonder if this is because of maternal antibodies.

Regarding time of year/period within seasonal disease cycle, rates stay constant from January through April at around 86%, then fall off by July until the next winter. Makes sense; measles might transmit better in cold weather, like influenza, or because people are huddled together inside more than when it's warm.

Now about intensity of exposure: comparing families with just one primary case to those with more than one, there didn't seem to be a significant difference in attack rate. It was all maxed out around 80-85%. If a second primary affected anything it would only be another 15%, and those are possibly already somewhat resistant somehow; otherwise they would've caught it from the first primary.

The same is mostly true for those exposed more than once: Not many who have been exposed once will escape, and the attack rate for them when exposed again is much lower. This makes sense; if they escaped the first time, they're more likely somewhat resistant (or records of past illness are incorrect, and they're immune). Still, the attack rate got up to 63% in contacts aged 1-4 exposed twice, so just being exposed once doesn't mean there isn't anything to worry about from later exposures.

Lastly, the influence of intensity of exposure: apparently exposure to multiple cases at the same time doesn't significantly increase the attack rate either.

So overall, it seems like time of year and age of the susceptible person matter the most, while sex and intensity of exposure don't affect much, though repeated exposures can increase the number of victims. And more than 80% of susceptible children exposed to a case of measles will catch it. Pretty hard to prevent.

Reference: Top, F. H. Measles in Detroit, 1935 I. Factors Influencing the Secondary Attack Rate Among Susceptibles at Risk. Am J Public Health Nations Health 28, 935–943 (1938).

080 - Whooping-Cough or Pertussis

As mentioned before, whooping cough can be pretty hard on children, especially young ones. In this article from 1938, Robert Cruickshank discusses whooping cough and how it compares to some other diseases in the UK at the time.

What Should We Call...

"Whooping cough" is the common term, referring to the shrill intake of air after a bout of intense coughing, but Cruickshank pointed out that even in severe cases of the infection, not all patients actually whoop. And since just "cough" or maybe "whooping and/or non-whooping cough" don't work too well, he suggests "pertussis" as a good alternative. On the other hand, as I discussed in 079, this could cause some confusion too, since not all cases of coughs with whooping are caused by B. pertussis. But obviously both names have stuck with us throughout the years.

Mortality

Different people had different estimates of how many people died from whooping cough. The case-fatality rate seemed to be between 1 and 8.5%, generally higher for younger patients. Though in Glasgow, the reported rate was 27%, and up to 44% for those less than a year old. Pretty bad.

For comparison, the rates for measles, diphtheria, and scarlet fever were 5%, 4%, and 0.4% respectively. So pertussis was the fourth leading cause of death in London ages 0-5 years, killing 434 people per year. The three leading causes were congenital causes, pneumonia, and diarrhea (presumably infections of unknown etiology). Measles was 5th.

Though despite these numbers, the death rates for these diseases had actually been decreasing over the past 70 years, at least for younger children. Cruickshank doesn't discuss why this might be. Could be better treatments, supportive care, immunization (at least for diphtheria), increasing public health in general... not clear.

Prevalence

Keeping track of cases of whooping cough wasn't mandatory throughout the UK at this time, though some areas did so. So it was only possible to estimate the prevalence. Some estimated that 44% of children in London got pertussis before age 5, and 60% by age 10. Measles was similar, diphtheria and scarlet fever less so.

In England, it seemed like pertussis came in two-year intervals, though it seemed different in other countries. This seemed to be because of the addition of susceptible people to the population (newborns), but could also because immunity after infection didn't last too long (possibly only a year; I wasn't clear on this part).

Lab Tests

It was pretty clear at this point that B. pertussis caused whooping cough (most of the time), not some virus. People infected with these bacteria developed antibodies, and antibodies produced from vaccination correlated with immunity to infection. Cruickshank discusses methods for diagnosis, those that work and those that don't.

Treatment and Prevention

Cruickshank says: "Pertussis is a disease of which it may be said that the multiplicity of remedies is an index of therapeutic failure." I think what he means is there a lot of suggestions but not many that actually seem to work. Probably like what I discussed in 078. Supportive care is good, of course, and anything that helps children breathe better. Some thought vaccine therapy or antiserum worked well, especially in the early stages of disease, but it didn't seem clear.

For controlling spread, Cruickshank mainly recommended keeping infected patients away from susceptible children, which makes sense. Pertussis isn't as contagious as measles or chickenpox, for example, so it wouldn't be too hard, even in hospitals. He thought that patients shouldn't be contagious anymore after the 4th week of disease.

If isolation of cases were impossible in any situation, he recommended vaccination as something that seemed effective. He cited Madsen's data (069) and Sauer's, showing effectiveness of their vaccines. But in the interest of more solid data, he recommends a more controlled study, and possibly a program similar to the one in place for diphtheria at the time.

Overall, not much new here, but an interesting perspective.

Reference: Cruickshank, R. Whooping-Cough or Pertussis. The Lancet 232, 33–37 (1938).

My Methods

It occurs to me that, if I were reading this blog instead of writing it, I would wonder, "how is this guy finding these studies and choosing which to blog about? And what isn't he telling me?" And these are good questions, worth asking of anyone who claims to be any kind of knowledgeable. So I thought it'd be worth going into my methods for this blog a little.

I started out by going to PubMed and searching for "vaccine", then saving all of the results as far back as they went, so far up to 1940. Each time I completed a decade of studies, I could go to PubMed for another. I save these studies in my favorite reference software, Zotero.

This method of searching is likely to miss out on a great deal, if not the majority, of potentially interesting studies: those that don't have "vaccine" in the title, for example, but instead say "immunization" or whatever. For this reason, I also gather studies from other sources on the web: pro-vaccine ones, like Science-Based Medicine, I Speak of Dreams, or this one; and also, perhaps more especially, anti-vaccine ones, like IMCV (now there's a place where it's important to ask yourself the questions mentioned above!). I also look at what the scientific journal Vaccine puts out, to see what's going on in that field.

And finally, for each study I blog, using Web of Science, I look up all the later studies that cite that study and save them, as well as any interesting studies cited by the blogged study. This helps give me an idea of what later scientists thought of the studies I'm reading, and also broadens my net to catch things that I might not otherwise encounter. With these methods, I've accumulated more than 2,000 articles I might someday blog about, and I'm sure I'll find many more as time passes.

How I Choose
You might've noticed that I don't blog about every study I've found. Part of this is practical: it takes a while to write a blog post, and I have a lot of other stuff going on.

The other part is that a lot of the studies just aren't that relevant to my purpose. I want to research vaccines' safety and effectiveness, and a lot of studies I've found are about basic bacteriology or virology, methods or vaccines that have been abandoned to history and thus aren't relevant today, or just don't add anything new to what I've already blogged about. For example, almost 400 articles cited reference 1 in 040, but the majority of those are about an experimental form of multiple sclerosis in animals, not very related to vaccine safety or effectiveness. So those I put aside.

I try to be careful not to be dismissive of articles that seem to show some problems with vaccines; I hope that's apparent. Those are the ones most likely to be interesting (aside from large, well-done trials of safety and efficacy, of course), so I give them some priority. So my goal for the answer to the question, "what isn't he telling me" is "not much."

I must reiterate, though, that I'm not an expert on this subject, so instead of telling you what is true, I'm trying to show you where to find that information. If anyone wants the full list of my references, blogged or not, shoot me an email and I'd be happy to share.

079 - Bacillus para-pertussis: A Species Resembling Both Bacillus pertussis and Bacillus bronchisepticus but Identical with Neither

This is a topic I've heard mentioned before as related to the effectiveness (or lack thereof) of the pertussis vaccine. And perhaps more distantly related to other vaccines. The idea is that there are other, rare, related pathogens or mutants of common pathogens that may take the place of pathogens made rare by vaccination, essentially refilling the niche. 

Or, in a more conspiracy-prone vein, that after a vaccine is widespread, cases of the vaccine-preventable disease are recorded instead as caused by a different strain, species, whatever; essentially changing the definition of disease so that artificially it looks like the rate of the disease declines drastically post-vaccine.

Personally I haven't found much evidence to bear out the latter idea yet. Definitions do change, a notable example being the broadening definition of autism artificially increasing the number of cases, but to say it's as simple to change a case rate as to reassign a set of symptoms to a different cause is to ignore the fairly modern ability to identify pathogens using very precise molecular techniques and such.

The former is an interesting idea, but I don't think it has borne out well in history. Though it may sometimes appear that there is a parallel increase in a parallel disease as one declines due to vaccination, often that's just because the vaccine-preventable was so common before, it masked instances of the parallel one, and now that's not so common, the parallel one becomes more apparent.

Anyway, the specific case today is infection with the bacterium Bordetella parapertussis as a rare alternative cause of whooping cough. Normally the disease is caused by Bordetella pertussis, which seems generally more severe than its cousin. Grace Eldering and Pearl Kendrick worked on pertussis a lot in the early 20th century, and in this study they identified B. parapertussis as similar but distinct from its cousins.¹

These researchers had collected almost 1500 isolates from whooping cough patients over 5 years. But 10 of them seemed unusual: on agar, colonies grew larger than expected over time, and could grow without blood in the medium. Another species, B. bronchisepticus, also caused disease, but was also different in some ways from this new isolate (chiefly, motility).

Of the cases from which these isolates came, half were less than moderately severe, but almost all of them whooped, so it seemed like regular whooping cough. Apparently at least one had a co-infection with this isolate and regular B. pertussis.

Then Eldering and Kendrick did a bunch of biochemical bacteriological tests on one of the isolates, called 309, and compared it to B. pertussis and B. bronchisepticus. This was neat because I remember learning about most of these in basic microbiology lab, and here I see them applied, many decades ago. Overall, 309 was similar to B. pertussis in some ways and similar to B. bronchisepticus in others, but to neither in all.

Finally they tested the immunological characteristics of the three strains. They found when antibodies were generated to any one of them, they cross-reacted somewhat with each of the others, but not completely. So they seem to share some antigens.

So apparently B. parapertussis is a separate agent, that occasionally (around 0.7% of cases) causes whooping cough in humans. One important question related to vaccines is the one raised at the beginning (if B. parapertussis could come fill a niche left if B. pertussis is eliminated by vaccines); more on that later, but J.J. Miller Jr. mentioned this in a study about a decade later:

"Some apparent failures of immunization in vaccinated children are due to infections with the Bacillus parapertussis. These infections will seldom be diagnosed correctly, as the cough is clinically indistinguishable from that of pertussis: in other words they are cases of whooping cough but not of pertussis."²

Another question, kind of on the other side, is whether a vaccine against B. pertussis could give cross-protection against B. parapertussis (or vice versa). Since they do share antigens, it's not unlikely. Today's study doesn't say much about this, but I think Eldering and Kendrick did address this question in later studies.

A final note on taxonomy: it seems like the taxonomy of Bordetella was pretty confused at least until the 1950s, so B. pertussis was referred to as Bacillus pertussis, Hemophilus pertussis, and probably others until people finally settled on Bordetella. So if you see those other names, you know.

References:

1.

Eldering, G. & Kendrick, P. Bacillus para-pertussis: A Species Resembling Both Bacillus pertussis and Bacillus bronchisepticus but Identical with Neither. J. Bacteriol. 35, 561–572 (1938).

2.

Miller, Jr., J. J. Immunization procedures in pediatrics. JAMA 134, 1064–1069 (1947).

078 - Treatment of Whooping Cough with Vitamin C

Another thing I've seen from those who think vaccines are not useful is that vaccines are unnecessary, because large doses of vitamins are cheap and adequate to treat or prevent the relevant diseases. Especially vitamin C, which somehow got a reputation as a disease-busting super-drug. So today I'll be looking at one study that I've seen cited on multiple articles, as well as a couple other related studies.

A. Ascorbic Acid (Vitamin C) Treatment of Whooping Cough¹

Interestingly, Ormerod and Unkauf start off their paper by saying that vaccination so far was the only thing that seemed effective against whooping cough:

"While some protection has been afforded against it by vaccination, treatment of the active disease has not progressed as has treatment of other infectious diseases such as scarlet fever and diphtheria. Madsen reports that, of 1842 vaccinated children, about 25% escaped infection, while of 446 non-vaccinated children less than 2% escaped."

But for those who caught the disease anyway somehow, it'd be very helpful to have a way to treat it, to remove its status as a scary childhood disease:

"In the years 1932-34 there were 45,755 cases of whooping cough reported to the Dominion Bureau of Vital Statistics, with 1982 deaths. Of the fatal cases over 50% occur in the first year of life."

Others had studied the effects of ascorbic acid (vitamin C) on other things, like seeing if it could inactivate diphtheria toxin or inhibiting bacteria directly, but it wasn't clear that its effects weren't only due to its acidity. Even if it can inhibit bacteria in culture, anyway, doesn't mean it can do so in the human body.

Some thought they observed cases of scurvy in pertussis or pneumonia patients, and hypothesized that the body might use vitamin C as a defense mechanism somehow, thus using up its stores. And children seemed to need more than adults. So Ormerod and Unkauf decided to test ascorbic acid's effectiveness in treating whooping cough.

It's actually not so much a study as a series of case reports, since there was no control group. The authors treated pertussis patients with a synthetic form of vitamin C, sold by Roche. All but two of the patients were seven or under; one was 12, and the last was 22. The children all had the characteristic whoop from under a week up to 6 weeks, and starting at various points in the disease, the authors gave them between 125 and 500 mg vitamin C daily. It didn't seem very standardized. Their coughs all disappeared in from 3 days up to 15 days.

Ormerod and Unkauf admit that the data isn't robust enough for good conclusions, but claim that vitamin C seemed to reduce the duration from weeks to days. They don't have any controls though, so it's not really known how the patients would have done without the treatment. Is this study worth anything?

B. Pertussis Endotoxin in the Treatment of Whooping-Cough²

As a comparison, let's look at another study done around the same time, using a treatment that everyone has pretty much discarded as worthless (and had even been discredited at the time, in 1936): vaccine therapy. This is the idea of inducing a better immune response against pathogens currently infecting a person, so the body can better fight it off, rather than producing immunity before an infection is present.

In this study, Thompson wanted to try to induce an immune response against bacterial toxins produced in whooping cough patients by injecting purified toxin. Others had tried this, but the studies weren't done well, so it was hard to tell if there was really an effect.

Thompson used a procedure similar to Krueger's (076) for producing the toxin. He found that it had little effect when injected into rodents, but it often produced an allergy-like reddening in children. Possibly an immune reaction. In whooping cough patients, it seemed to exacerbate the symptoms sometimes, which doesn't seem surprising. So he tried to use lower doses to avoid this reaction.

He studied a group of 403 children admitted to the hospital with whooping cough. He gave 132 of them daily injections with the toxin, and the others were controls. Some of the controls even got placebo in the form of saline injections. The children received some other treatments too, seemingly to treat symptoms.

But even so, it was tough to compare the two groups, because there weren't specific endpoints. It didn't seem like severity was different, but maybe the duration was shorter, like the treatment accelerated the disease process, making it end sooner.

Possibly there was an effect in those treated early in their infection, shortening the case and reducing severity. But Thompson tried to replicate the result in another 71 patients, treating them early, and again it seemed to reduce the duration but not severity. So of questionable use overall.

C. Vitamin C in Treatment of Whooping-cough³

Responding to both of the above studies, apparently Douglas Gairdner was dubious:

"It also seems true to say that although 'nearly every newly discovered remedy in ancient and modern times has at one time been advocated for treatment of this disease', there is not one that has been proved to have an appreciable effect on its course. Of vaccine or endotoxin therapy, lately so enthusiastically recommended, the conclusion...'that the more carefully the results are controlled the less impressive do they become' seems inescapable in light of recent reports."

Ormerod's results from North America using vitamin C showed a much shorter whooping cough duration than what was common in the UK, so Gairdner wanted to try to replicate those results. He used naturally derived ascorbic acid, donated by Roche, to avoid a possible difference in effect from the synthetic (like Ormerod used); a few younger children got the synthetic kind because it was less bulky. He tried to use doses similar to Ormerod, several hundred mg, much more than the daily requirement for normal children (about 25 mg). Control subjects were given a placebo, cod liver oil and a couple other things that were known not to affect the disease.

In total, Gairdner treated 21 children, and kept another 20 as controls. 20% of each dropped out. Cases were confirmed by culturing the bacteria and hearing the cough. All cases had been coughing for less than 3 weeks when they started treatment. He followed them by getting the mothers to record the number of coughs each day and night until they stopped coughing at night. He also weighed the children to see if the disease was impairing their growth.

In Ormerod and Unkauf's study, they started treating an average of 9 days after disease started, and it lasted an average of 14 days longer, for 23 days total. 

Gairdner started treating his patients 10 days after the disease started on average (14 days for controls), and the disease lasted an average of 25 days after treatment (27 days for controls), for a total of 35 days (or 41 for controls). Apparently there was no statistically significant difference between the groups, nor for the difference in weight gained during the disease period (0.01 lb difference). Severity wasn't really measured.

So both treated and controls had much longer disease than in Ormerod's study, at least 12 days. Gairdner's conclusions:

"The assertion of Ormerod and UnKauf that the paroxysmal period of the disease is shortened 'from a matter of weeks to a matter of days' was not confirmed."

"As there were no controls in the Canadian series, however, it is impossible to judge whether the natural course of the untreated disease varies in the two countries, or whether the considerable difference in the course of the disease in the present and in the Canadian series is due, for instance, to the application of a more rigorous standard of cure in the former series."

"It is considered that the statement that the administration of vitamin C in whooping-cough has an effect upon the course of the disease is at present unproven."

It's possible that synthetic vitamin C worked better than naturally-derived kind, though these studies can't really be used to conclude that either. But it would be interesting.

Later studies commented on the vitamin C studies discussed:

"Gairdner has reported a failure to affect the course of whooping cough using the method of Omerod (sic), but Gairdner also failed to test for a deficiency before instituting treatment."⁴

"Gairdner in a controlled experiment found that the duration of illness in a group receiving vitamin C was shorter than in controls. The difference in the two groups was not a significant one, and he considered that the alleged benefits of vitamin C in whooping cough were unproven."⁵

The first statement, by J.B. Youmans, is slightly puzzling to me, because the original paper doesn't indicate that Ormerod and Unkauf tested for deficiency either. All just assumed that the doses they were giving were high enough to correct any deficiency. If it is a legitimate criticism, it seems to show that vitamin C would only be beneficial if there is a real deficiency.

References:

1.

Ormerod, M. J. & Unkauf, B. M. Ascorbic Acid (Vitamin C) Treatment of Whooping Cough. Can Med Assoc J 37, 134–136 (1937).

2.

Thompson, A. R. Pertussis Endotoxin in the Treatment of Whooping-Cough. The Lancet 230, 733–736 (1937).

3.

Gairdner, D. Vitamin C in Treatment of Whooping-cough. British Medical Journal 2, 742–744 (1938).

4.

Youmans, J. B. The Influence of Vitamin Deficiencies on Other Diseases. Ann Intern Med 13, 980–986 (1939).

5.

Glazebrook, A. J. & Thomson, S. The Administration of Vitamin C in a Large Institution and Its Effect on General Health and Resistance to Infection. The Journal of Hygiene 42, 1–19 (1942).

077 - Reinfection (Second Attack) in Experimental Poliomyelitis

A common thing I hear from those that don't like vaccines is that coming down with the "natural" version of the disease gives a much stronger, even life-long immunity to it, while vaccine-derived immunity only lasts a few years. The truth of this depends on the vaccine, the disease, and the person in question, of course, but it's worth asking if it's true that getting a disease makes one immune thereafter.

This study, by Dr. Simon Flexner, investigated whether monkeys that had recovered from a "natural" (though experimental) infection of polio could be reinfected with the same or related virus—a second attack. Others had previously observed second attacks in children and monkeys infected before with polio, but not in such a formal setting.

Polio tends to be more severe in monkeys, often paralyzing and killing them, but they're not as susceptible to it, so it needs to be introduced to them experimentally, in the lab, with larger doses of virus than people encounter. Because of this, it's a bit less "natural" than human infections, but easier to work with. Those that survive have high levels of anti-polio antibodies.

So Flexner took monkeys that had recovered from polio and tried to infect them again, using the same strain of virus as the first time, or a different strain. He found that it wasn't too hard to reinfect these monkeys, even with the same strain of virus. The disease sometimes was just as severe as the first time, even in those that had severe disease the first time. It was even possible in monkeys that had been hyperimmunized through a kind of vaccination.

The final question was whether monkeys could even be reinfected a third time. There had been a case report of a third attack in a child, but none in monkeys yet. Flexner took the monkeys he had left after the second attack and tried reinfecting them, but none of them got sick that time.

So "natural" immunity is not some magical shield, at least not in lab monkeys with polio.

Citation: Flexner, S. Reinfection (Second Attack) in Experimental Poliomyelitis. J Exp Med 65, 497–513 (1937).

076 - Effect of Preservatives on Undenatured Bacterial Antigens

A big concern among parents seems to be preservatives used in vaccines, especially mercury-containing thimerosal. This study was a short one that looked at different preservatives and their effect on the immunity-producing components of bacteria.

Often it is desirable to separate out just the part of a pathogen that an immune response targets best and make a vaccine out of that, because that way you get a good, focused immune response against the important part and don't have anything else that could produce unwanted reactions. So Krueger and Nichols had developed a method to purify bacterial antigens, removing all whole bacterial cells from them, resulting in a product called Undenatured Bacteria Antigen, or UBA.

Using preservatives in vaccine preparations was often a good idea, because a bacterial contamination could lead to pretty bad outcomes in the recipients who received vaccines by injection. So this study tested how harsh different preservative candidates were on the UBA.

The ones they tested were tricresol, phenol, and merthiolate (aka thimerosal or thiomersal). The concentrations they used of the first two were much higher than of the third, more than 10 times higher, presumably because they didn't inhibit bacterial growth very well at lower concentrations. But they tried three different concentrations of merthiolate, going as low as 20 parts per million, or 0.002%. Then they tested how much UBA was denatured over time.

Denaturation by all of them was as complete as it was going to be by one week, and didn't increase beyond that. UBA with no preservative remained mostly undenatured, 95%, while that with phenol or tricresol was nearly half denatured. Merthiolate was intermediate, the lower concentration the better, between 24 and 37% denaturation.

So it seemed that merthiolate was the best preservative to use for preserving UBA.

Citation: Krueger, A. P. & Nichols, V. C. Effect of Preservatives on Undenatured Bacterial Antigens. Exp Biol Med 34, 335–337 (1936).

075 - Epidemiological Studies in Influenza

People still hadn't worked out exactly what influenza was or wasn't. Was it a disease caused by a single virus, or a collection of symptoms that could be caused by multiple viruses? Part of that was an arbitrary definition. If just one virus, were there different strains? They also strongly suspected that bacterial secondary infections could play an important role sometimes.

So Thomas Francis attempted to define influenza as well as possible at the time, and give some other information about it.¹ He defined what flu is not (common colds, pneumonia, sore throat, diarrhea) and described a typical case. 

He discussed Shope's studies (049) distinguishing the swine flu virus from Haemophilus influenzae suis bacterial infection, and other animal studies in ferrets and mice (such as 074). Also important research growing flu virus on tissue culture or in eggs.

More importantly, he discussed the issue that some had suggested that people don't form immunity against the flu. It's understandable how people could get that impression, considering how the flu virus mutates enough pretty much yearly to be able to reinfect even people who had it the previous year, so we need a new flu vaccine every year. But the presence of at least some immunity is important; otherwise every flu infection could be as scary as avian flu is supposed to be.

Part of the problem was that the techniques for identifying and distinguishing strains of virus weren't very developed at the time. They could try to infect animal models with samples to isolate virus from them, but if the virus were a type that didn't infect such animals very well, it would give a false negative. This happened to Francis: he observed an epidemic of influenza in California in 1936 with all the usual clinical symptoms, but hardly any patient samples gave infectious virus. As others noted:

"Although it has been suggested by Stuart-Harris et al. that in the presence of an epidemic of respiratory disease a certain symptom complex may serve to differentiate influenza from similar but etiologically different diseases, the California epidemic studied by Francis makes this possibility seem unlikely."²

Later, people realized that this was the first identifiable observation of an outbreak of Influenza B.³ Previous studies apparently had focused mainly on Influenza A.

Lastly, Francis discussed attempts to immunize people against the flu. Mostly it was similar to the results in 074: they saw a rise in antibodies against flu for at least a few months, but didn't actually test if it were protective against infection. Interestingly, some tried inoculation with live virus, subcutaneously, and didn't see any evidence of respiratory infection or serious side effects. Looking at antibodies in people who had just recovered from flu (the way they did this was to inject mice with human serum and see if it protected them against viral infection; another technique with questionable assumptions), they found good antibodies in about 30-60%, and also found strong antibodies in about 30% of people who gave negative histories of flu. So, more work to be done.

References:

1.  Francis, T. Epidemiological Studies in Influenza. Am J Public Health Nations Health 27, 211–225 (1937).

2.  Horsfall, Jr., F. L., Hahn, R. G. & Rickard, E. R. Four Recent Influenza Epidemics: An Experimental Study. J Clin Invest 19, 379–392 (1940).

3.  Burnet, F. M., Stone, J. D. & Anderson, S. G. An Epidemic of Influenza B in Australia. The Lancet 247, 807–811 (1946).

074 - Influenza: Further Experiments on the Active Immunization of Mice

Andrewes and Smith were some of the researchers working on creating a flu vaccine, especially since people had discovered that influenza was caused by a virus, not bacteria.

The vaccine they were developing was made from infected mouse lungs, and mice were the model animal they focused on mostly, especially in this study. Mouse lungs produced a lot of virus, but it wasn't the cleanest, so what they were attempting in this study was to produce a cleaner version.

Actually there were three main goals:
1) Try to get as much virus as possible,
So less volume is needed for the same dose

2) Purify the virus as much as possible without reducing its immunizing ability,
So there aren't contaminants that could cause unnecessary reactions

3) and if possible, inactivate the virus (so it can't infect) without reducing its immunizing ability.
So that it can't possibly infect and cause disease.

For objective 1, they tried filtering the virus with membranes that the viruses were too large to pass through, but that didn't really seem to help. At some point though, their virus densities increased 10 to 100 times spontaneously, maybe through some mutation, so that worked out.

For objective 2, they wanted to remove mouse proteins from the preparation, so they tried adsorption/elution, in which they could stick the virus to something and wash everything else off, but they lost a lot of virus with this method too so it wasn't great. Filtering seemed to help though.

For objective 3, they tried inactivating the virus with formaldehyde. A solution of 0.01% could inactivate almost completely in 5 days at -2°C, and 0.02% could completely. This inactivated virus couldn't infect mice when put into their nose.

Immunization Experiments
Then they tested these preps in mice, to see which gave the best immunity against flu virus challenge. What they found was that washed live virus immunized about as well as unpurified virus (when inoculated into the skin or body cavity), and inactivated virus seemed almost as good, though it seemed like the dose they gave of this was higher than the dose of live. Even 0.1% formaldehyde-inactivated gave good immunity. Virus-free filtrate didn't help at all, so the antigen is not soluble.

They did find that virus that had been washed and then inactivated (or the reverse) didn't have much immunizing power in mice. That was unfortunate.

The immunity from each vaccine seemed to fade in mice after 6 weeks. However, this was similar to how long mice had immunity when they had gotten sick with the flu and recovered, so the vaccine was as long-lasting as natural immunity (especially considering that most of the infected mice died from the disease).

Preliminary Human Trial
Finally, they tested inactivated virus in a few human volunteers. They didn't want to use live virus, considering how others had seen what seemed like flu outbreaks from live virus vaccines (049). So 5 volunteers got washed and inactivated virus, and two more got inactivated unwashed virus. They also added 0.01% merthiolate (thimerosal) to prevent bacterial contamination just in case.

The first two had some pain, maybe from excess formaldehyde, so for the others Andrewes and Smith changed the pH to convert the formaldehyde to something else, which worked better. They didn't see any serious reactions to any version, though the ones getting unpurified virus had more tenderness (possibly sensitivity to mouse proteins).

What they saw was that in all but one volunteer, levels of antibodies against the virus rose after the first dose (not much after the second dose for some reason). This was heartening, especially with the washed+inactivated virus that hadn't worked well in mice. Even better, the levels seemed higher than in other people who had recently recovered from the flu! (Though I'm not sure the flu the people had would be the same antigenically as the virus used in this study.) And the levels still seemed high after 2.5 months. So they might be on the way to a good flu vaccine, but they weren't sure yet if antibody levels correlated well with immunity. More work to be done.

Citation: Andrewes, C. H. & Smith, W. Influenza: Further Experiments on the Active Immunization of Mice. Br J Exp Pathol 18, 43–55 (1937).

073 - Vaccination Against Acute Anterior Poliomyelitis

I've talked about John A. Kolmer and his polio vaccine before (047, 048, and 063), but I wanted to touch on it once more.¹

Kolmer's vaccine was a "live" but partially inactivated virus. He took infected monkey spinal cords, treated them with sodium ricinoleate, added some phenyl-mercuri-nitrate as a preservative to prevent bacterial contamination, and injected them subcutaneously. These things should prevent infectivity in humans, he thought. And it seemed to work well in monkeys, though it could still paralyze if injected into the brain. The reason he wanted it partially "alive" was that he thought completely inactivated virus was unable to immunize, for some reason.

By this point, more than 12,000 people had received Kolmer's vaccine. None seemed to have severe reactions, like encephalomyelitis, though some that received the version without preservative had abscesses temporarily.

However, there were 10 cases Kolmer knew of in which the subject seemed to get sick with polio soon after receiving the vaccine (soon meaning 1-6 days later). Usually it was after the second dose, never after the third, but five of the 10 (50%) actually died from their illness, from paralysis.

In this paper, Kolmer thought it unlikely that the polio had come from his vaccine, considering the many that received the same lot without getting sick, and how no one receiving all three doses got sick. However, he was unable to explain where the virus had come from for some of the cases, since there wasn't an outbreak in their areas. It was a mystery.

It seems like later, though, he does conclude that the vaccine is not safe enough to use in people, especially because he hadn't been able to establish its efficacy in preventing any disease.

"It was my hope that this strain of virus had lost infectivity for human beings by reason of its long adaptation to the monkey, and especially after treatment with sodium ricinoleate and when given by subcutaneous injection, but the occurrence of nine cases of poliomyelitis among 10,725 individuals given the vaccine in 1935 has indicated that the virus apparently possesses infectivity for human beings and that this vaccine as well as the formalized vaccine of Park and Brodie is too dangerous for use."²

So, I'm not sure whether it was the correct decision or not, but that's why we don't use Kolmer's vaccine these days.

References:
1. Kolmer, J. A. Vaccination Against Acute Anterior Poliomyelitis. Am J Public Health Nations Health 26, 126–135 (1936).
2. Kolmer, J. A. The Present Status of Methods for the Prophylaxis of Acute Anterior Poliomyelitis. Ann Intern Med 12, 95–105 (1938).

072 - Progress Report on Pertussis Immunization

Other than Louis Sauer and Madsen and others, Pearl Kendrick and Grace Eldering were trying to develop a good pertussis vaccine. They did a trial of the one they made in Grand Rapids, Michigan, which from what I can tell was pretty historic. This report was not their final conclusion, but they had enough data to make an interesting preliminary report.¹

They used a virulent strain of B. pertussis for the vaccine, growing it on sheep’s blood, then killing it with merthiolate (AKA thiomersal or thimerosal) or phenol. In terms of dose, they used closer to Sauer’s preferred dose, much higher than Madsen’s, but tried a few different doses (though this didn’t seem to make much difference).

The subjects in the study were children, 8 months to 5 years old, especially 1-2 years. As controls, they followed children the same age, preferring children in the same families as those who received the vaccine (to ensure similar circumstances).

They worked with nurses from the city’s public health department to follow up with the subjects at 3- or 4-month intervals, verifying reports of pertussis cases or exposures. Exposures were classified as definite (coming from someone in the home), indefinite (coming from elsewhere), or unknown (a case of whooping cough from an unknown source). Cases were rated by severity based on quantitative definitions of duration of disease, cough, whether there were complications, etc.

So what did they find? It was pretty good. By the time of publication, sixty vaccinated subjects had been exposed, but only four got sick. That’s less than 7% (or 0.5% of the total group, including non-exposed subjects, 712 subjects total).

In contrast, there were 63 cases in the control group, from 84 exposures (880 subjects total). That’s 75% infection rate (or 7% overall). So the vaccine seemed to protect more than 90%.

And even considering the four vaccinated subjects that got sick, their disease was either light or very light, while 85% of the control cases were of medium severity or higher.

The researchers warn that the numbers might be too small to make any solid conclusions, and the trial is not up to today’s standards (in terms of blinding, randomization, placebos, etc) but for what it’s worth it seems impressive.

A later historical account gives the following description of the response to this report. Keep in mind that requirements for trials in the 1930s were much less rigorous than they are now, and that yet Frost basically admits he can’t find much fault with the design: 

"Although it was well controlled, the 1934-1936 field trial had to surmount considerable initial skepticism within the national public health community. Shortly after Kendrick and Eldering announced their results, James Doull, a prominent Cleveland epidemiologist, reported that children received no protection from a vaccine he had designed and produced (065). The APHA subcommittee on whooping cough, which included both Kendrick and Doull, evaluated the contradictory results of the two studies but was unable to explain why they differed. The committee then asked Wade Hampton Frost, a Johns Hopkins epidemiologist and head of the APHA, to review both studies.
     Frost was predisposed to find fault with Kendrick's study. In a "personal" note...Frost voiced his doubts: 'I very strongly suspect that Miss Kendricks field studies are not set up in such a way as to give a really good control. My reason for this suspicion is that, as you know well enough, the satisfactory set-up of such an experiment is an exceedingly difficult matter. Not 1 out of 10—perhaps not 1 out of 50—attempts is successful and as a mere matter of probability the odds are strongly against Miss Kendrick's experiments being sound.'
     Unable to find the faults he expected in Kendrick and Eldering's study, Frost journeyed twice to Grand Rapids to examine their data and recommend suggestions for improving statistical accuracy and coding. In the end, Frost said to Kendrick, 'I think it may be assumed, not as a conclusion but merely as a working hypothesis, that your data when finally analyzed are likely to show some protection in the vaccinated group.'"²

Citations:

1.  Kendrick, P. & Eldering, G. Progress Report on Pertussis Immunization. Am J Public Health Nations Health 26, 8–12 (1936).

2.  Shapiro-Shapin, C. G. ‘A Whole Community Working Together’: Pearl Kendrick, Grace Eldering, and the Grand Rapids Pertussis Trials, 1932-1939. Michigan Historical Review 33, 59–85 (2007).

O722 - Safeguards in the Publicity Use of Vital Statistics

Not specifically about vaccines, but interesting nonetheless, because vital statistics are often used (or misused) in support of (or opposition to) vaccines. Also the author was the one who wrote the article for post 060 about measles attack rates.

Anyway, it's a good lesson and a good reminder about how it is difficult to convey accurate information with statistics, especially if you have some personal interest in what is being conveyed.

"To reason correctly from such complex data is not easy, and even the most acute and careful persons will at times fall into error in the process. But no confusion can be quite so profound as that of the individual who rushes into this maze with the intention of finding support of a preconceived idea."

Hedrich gives several examples of how statistics can go wrong that are worth checking out. It reminds me a bit of the book How to Lie with Statistics, though not quite as in-depth or entertaining.

But for we who want to examine the truth of a matter without bias, it is important to be twice as careful with our own statistics as we are when examining others', especially those on the other side of the issue.

"Public health certainly aspires to as high an ideal as is professed by business: Truth in Advertising. To that end, the first safeguard proposed in the publicity use of statistics is to try to be more critical of the materials and of the reasoning which favor our cause, than of those which oppose it—more critical, because otherwise we shall certainly not be critical enough; the instinct to defend one's beliefs and interests is so strong that desire too often pulls the wool over the eye of reason."

And keep in mind that finding the truth is not as simple as just finding a reasonable hypothesis and sticking with it:

"It is a fundamental theorem of scientific procedure that a given explanation cannot logically be accepted as the preferable one, unless all other reasonable hypotheses have been eliminated."

Anyway, it's a good article. Go check it out! It's even open access.

Reference: Hedrich, A. W. Safeguards in the Publicity Use of Vital Statistics. Am J Public Health Nations Health 24, 336–341 (1934).