The Keyworth study of Whooping Cough
1974 to present
Here I describe the study that started in 1974 and still continues.
The full story will soon appear as a book to be published by Springer Nature.
Some of the information on this website is based on the study of whooping cough that I have undertaken as a family doctor over 40 years in Keyworth. Much of the material has been published in medical journals. Some is unpublished, and some is my opinion based on experience. I believe this is a unique study and that I have a contribution to make to the understanding of this unpleasant and sometimes lethal disease.
I wish to make my data available to the public so they may judge its value for themselves. This page outlines the main findings.
Keyworth is a village about 5 miles south of Nottingham in the East Midlands of England. It has a population of 8,000. There are several smaller nearby villages and all together make up a community of about 11,000, that all come under the care of 8 family doctors working from a single medical centre. 30 years ago there were 11,800 patients and 4 doctors)
I have worked at Keyworth Health Centre since 1974, when I started as the most junior partner after returning from 3 years in central Africa where my research interests developed. Since 1977 I have made a special study of whooping cough in this small population ( 744 cases). I have developed the ability to recognise cases that most other doctors would miss, simply because of my intense interest in this disease and being on the lookout for it all the time. Because of the way healthcare is organized in England, with single medical records and patients registering with one medical centre only, it is possible for me to be confident that what I observe about whooping cough in Keyworth, is as complete as possible, accurate, and most of all, consistent.
I retired from the partnership in 2011 but was able to reliably follow the incidence until 2013. Since then it has not been possible to continue the study with the previous thoroughness and therefore the study offiicially ended then, but the doctors in the practice continue to diagnose it competently and the numbers recorded continue to determine the current pattern of disease just as before.
It has become even more important to continue in the same way because serological diagnosis has become mandatory as Public Health England (previously the Health Protection Agency) now uses laboratory confirmed cases for its statistical baseline. As the age of victims has climbed to an adult age, so the ease of getting a blood test increases. Blood tests have only been available since 2002 in the UK and only widely used since 2006. Increasing familiarity with the need for the test and increased awareness and self diagnosis by the now adult victims by means of the internet (and previously this site in particular), has led to a large rise in the proportion of suspected cases that are getting confirmed, and number of tests being done. Previously they would not have been tested at all, or by pernasal swab which is difficult to arrange and uncomfortable, as well as usually being negative because it was too late in the illness. So they would hardly ever be notified.
Because we are just an average medical practice, what I have observed in Keyworth is also probably representative of what happens in the rest of the United Kingdom. It may also be quite similar to what happens in other developed countries with similar immunization practices (for example: USA, Canada, Australia, New Zealand and the countries of the European Union).
What have I concluded?
Whooping cough has been to a great extent ignored and forgotten for half a century or more, because immunization has been so successful at reducing the number of cases of the disease. It did, however, not completely go away, and people are now realising that it is still about and causing quite a lot of trouble. Some people think it is making a comeback. It is doubtful if this is true if the Keyworth data are correct. They seem to indicate that the amount of trouble whooping cough has been causing has been much the same for 30 years, although there are some quite interesting changes in the ages of people it attacks.
Why is this relevant?
There is currently discussion in the media about whooping cough making a comeback, particularly in adults. I think much of this is apparent rather than real. Recent research has shown that many adults with persistent coughs do indeed had whooping cough. This is not new information if the Keyworth study is representative. Looking for it is what is new. The Keyworth data show that the incidence in adults has remained constant since 1986, and it is the others that have fallen.
Since immunization came in the 1950s, doctors have seen less and less whooping cough and modern doctors may never have seen a case, let alone heard the cough. I believe that much of the drop in notifications has simply been a reflection of the poorer diagnostic skills of modern doctors in respect to whooping cough. Now that some people are looking for it with more sophisticated tests such as PCR, blood antibody, and recently oral fluid antibody tests, they are finding it, but notifications are still low, as the average doctor is still reluctant to diagnose it. This changing however and an apparent resurgence in the USA, Australia and the UK in 2011-12 or thereabouts has raised the numbers notified in these countries, and the numbers have only dropped a little since then. Most of this is in my view due to increased recognition but some of it may be due to acellular vaccine performing poorly compared with the older one which came into use around the millennium.
There is a a new factor operating now that is likely to further inflate the pertussis statistics in developed countries. That is the practice of using PCR for primary diagnosis. This test is positive in the early stages of infection irrespective of whether it develops into clinical whooping cough. Early and sensible testing of contacts of index cases in order to manage the infection better (with prophylactic antibiotics for example), will identify infections that would never previously have been even suspected.
There are now inexpensive PCR point of care tests available for B. pertussis.
If the numbers recorded are to be interpreted accurately there is a need to record clinical whooping cough separately from B. pertussis infection.
Raw data (anonymous) from this Keyworth study along with tables and graphs can be made available on email request in order that health care workers, epidemiologists and interested others can study the detail.
Immunization was introduced in the UK between 1952 and 1957.
Between 1974 and 1994 the immunization acceptance rate in England and Wales fell to 31% and then slowly rose. This was the result of a ‘scare’ about vaccine mediated brain damage that turned out to be false.
This histogram is the strongest evidence that doctors ceased diagnosing whooping cough in the mid nineties and started again in the mid noughties.
It was this failure to diagnose that I recognised in the late nineties that led to the launch of this website in 2000 to help people diagnose themselves.
The correspondence I received subsequently confirmed what I suspected, that it was a problem not just in the UK, but the USA, Canada and Australia too, and probably many others.
For many years this was the only website with sound files that enabled sufferers to recognise their own condition, and I believe this site made significant contribution to the re-recognition of the disease.
Nowadays there are many excellent websites informing people about this disease.
Numerically most visitors were and still are from the USA.
My published work on whooping cough includes the following most relevant papers briefly summarised
Outbreak of whooping cough in general practice. Jenkinson D. British Medical Journal 1978;277:896.
In 1977-8, 191 cases of whooping cough occurred in the Keyworth practice (11,800 patients then). This was at a time when the national immunization rate had fallen dramatically as a result of fears about the safety of the vaccine. There was general skepticism about the effectiveness of the vaccine. 126 cases were in the under fives. Because the numbers affected and unaffected were known it was possible to calculate the vaccine protection. This was 84% if those too young to be immunized were excluded. This was the first information of this sort for several decades and was soon confirmed in other studies. It was welcome news and helped with the decision to continue to recommend the vaccine as part of the national programme.
Whooping cough: what proportion of cases is notified in an epidemic? Jenkinson D. British Medical Journal 1983;287:185-6.
September 1982 had the greatest number of notifications in the 1982-3 epidemic in England and Wales at the time whooping cough had made a big come-back because of a low immunization rate. A postal survey asked all family doctors in Nottingham how many cases of whooping cough they had seen in September. The number (620) was compared with the number notified (116). This is 18.7%. The response rate was 83.6%. The conclusion was that even at a time of high awareness of the disease the probable real number of cases diagnosed could be at least 5 times the number notified. One might suppose that at times of low awareness, the ratio would be even higher (present times for example).
A search for subclinical infection during a small outbreak of whooping cough: implications for clinical diagnosis. Jenkinson D, Pepper JD. Journal of the Royal College of General Practitioners 1986;36:547-8.
At the start of the 1985 outbreak in Keyworth we took pernasal swabs from all suspected cases of whooping cough and any of their contacts with any cough. 102 were taken in all. Of all these, 39 were clinically diagnosed as whooping cough and 17 of them has positive swabs. No swabs were positive in the ones without clinical whooping cough. We concluded that there was no evidence of substantial subclinical infection. We also asked about catarrhal symptoms in those with whooping cough. Only one third had catarrhal symptoms.
Duration of effectiveness of pertussis vaccine: evidence from a ten year community study. Jenkinson D. British Medical Journal 1988;296:612-4.
I was able to analyze the cases I had seen over 10 years in a way that permitted a calculation of the effectiveness of whooping cough vaccine at different ages. The results, based on 326 cases in the 1 to 7 year olds gave the following results. 1 year olds 100%, 2 year olds 96%, 3 year olds 89%, 5 year olds 52%, 6 year olds 54% and 7 year olds 46% protection.
Many assumptions were made for the calculation. For instance, it was assumed that the population moving in and out had suffered from whooping cough in the same way as the population in whom it had been counted. It also assumed that the number of missed cases was low, and equal in immunized and unimmunized subjects.
This paper was the subject of a paper by Connor Farrington in which he calculated the size of the possible errors. His arguments did not invalidate the result of my study. He showed the possible flaws inherent in working out vaccine effectiveness from such a simple model. In 2002 a fourth dose of pertussis vaccine was recommended in the UK in the preschool booster in order to increase immunity. This brought the UK more in line with other countries.
Natural course of 500 consecutive cases of whooping cough: a general practice population study. Jenkinson D. British Medical Journal 1995;310,299-302.
The average number of paroxysms was 13.5 per 24 hours. 11 in immunized, 15 in unimmunized.
The average duration was 52 days. 49 in immunized, 55 in unimmunized. The range was 2 to 164.
The more paroxysms, the longer the illness lasted.
The younger the patient, the longer it lasted.
57% vomited. (49% in immunized, 65% in unimmunized).
49% whooped, (39% in immunized, 56% in unimmunized).
11% had significant cessation of breathing (enough to go blue) 8% in immunized, 15% in unimmunized.
Females were affected slightly more often in childhood but twice as often in adulthood.
Females had more severe disease.
Swabs were positive in 25% of immunized, 52% of unimmunized.
5 patients developed pneumonia.
Reviewed 22 May 2020