An Enigma: Transmission Of Epidemic Influenza (part 2)

I attempt to shed light on the riddle that is seasonal influenza using my bag of spanners. Today I take a look at the range of available diagnoses 1901 - 2000

It made an awful lot of sense to me to turn to the Office for National Statistics to see what data they held on influenza, so I headed for the main landing page and stuffed INFLUENZA into the new beta search results page. I came away with 166 data sets that I carefully worked through to see if any met my basic criteria of offering a time series (daily, weekly, monthly, quarterly or annual) of counts of something to do with influenza that dated back at least 50 years.

My eyeballs met with a lot of death series amongst those 166 data sets, most of which were concerned the COVID era, as may be expected. A paltry table of annual counts going back from 2000 to 1994 was the best I could scrape together, with another pathetic spreadsheet offering annual counts by age band for the period 2001 – 2021. Harumph!

It was then that I stumbled across a beefy spreadsheet of registered deaths by ICD10 code, age, sex and year but only for the period 2001 – 2022. Still, that will come in very handy indeed a little down the line. All this rummaging for the grail was rather frustrating for a man with a big spanner (ARIMA), but my leery gaze accidentally fell upon a most promising hyperlink that led me to this page within the ONS National Archives. A bit of button pressing later and I was sitting on 11 Excel files offering annual death counts by sex, age group and underlying cause within England & Wales for the period 1901 – 2000 - yaroo! Now that’s something I can get my teeth into; and what’s more I can bolt that bit extra on the back end to take that series through to 2022, which is well good enough for my purposes.

Oh Dear!

OK, so that felt good up to that point and I retired for a light lunch. I’m glad it was light for I then discovered that the underlying cause of death coding framework differed with each file. For example, back in 1901 – 1910 the code 0010 denoted smallpox – vaccinated¹; whereas in 1994 – 2000 it denoted cholera due to Vibrio choleae. Ouch! Back in 1901 – 1910 there were a total of 192 unique codes under the framework that is labelled ‘icd1’; whereas in 1994 – 2000 there were no less than 5,292 under the framework labelled ‘icd9’. Double ouch! How did death become so complicated?

What this meant was I had to devise a customised coding frame that would bridge differences across ICD revisions and align diagnoses as best as I could muster. The prospect of assigning upwards of 5,292 bridging codes by hand almost put me off continuing with the project, but it was then that I remembered the tin of shortbread.

But that wasn’t the end of my troubles, for in the very first file some bright spark had decided to start each numerical field value with an apostrophe, rendering it as a text value. Quite possibly the same person also decided to add the letter ‘T’ as a prefix to some age groups but not others to indicate 10-year spans. Yes, well, I shall say no more but surely it’s not beyond the reach of the ONS to do something altogether rather sophisticated and smartly polished!

And so I started out on my grand journey of re-coding the codes only to find terminology that no longer exists. For example, back in 1901 – 1910 there was such a thing as cerebro-spinal fever that no longer features in the modern ICD10 framework in this form of wording. Today we know it as meningitis, of which there are bacterial, viral, fungal and parasitic variants.

Working my way through several thousand diseases one disease at a time whilst pausing to evaluate archaic terminology was not an option, and so I opted to code diseases in block form, as per the modern ICD10 chapters. Here’s a reminder of this structure:

This block approach to coding offered promise… and then I came a cropper! Sitting with gonorrhoea, syphilis, anthrax and other very sensible chapter I rather nasty infectious diseases was Phlegmasia alba dolens (colloquially known as milk leg owing to the condition tending to arise in pregnant and breast feeding mothers). Today this diagnosis sits with a group of diseases that go by the name of deep vein thrombosis, and we’d give it an ICD10 code of I82 that places it in chapter IX: Diseases of the circulatory system. The word is WERGH!

I looked further down the list and groaned for England. In between Puerperal fever (not otherwise defined) and Pneumonia (lobar) we had Infective endocarditis. I considered swearing a lot but opted to go make a Grumpy Mule in that new mug my niece had got me for Christmas. Manual coding and cross-checking of a few hundred diagnoses I could handle in the space of a few days or so but processing several thousand was out of the question. There was no quick and easy solution that I could see since all was higgledy-piggledy.

As my coffee brewed I realised the effort needed to rationalise the full range of diagnoses made from 1901 to 2022 was astronomical, and well out of my reach even if I coded at lightning speed. Besides which, what the heck was glanders? It turns out that glanders and meliodosis (A24) is a nasty we can get from infected horses. So much for lightning speed... we could say I fell at the first hurdle!

Just One Cornetto

This left only one remaining course of action and that was to focus on re-coding influenza alone. In doing so I thought it prudent to additionally code major respiratory conditions such as pneumonia, bronchitis, tuberculosis, pleurisy, emphysema, asthma etc. Wading through 192 entries in order to do this wouldn’t take that long – less than the time it would take to chomp through a small packet of biscuits – but going through 5,292 disease descriptions would be prohibitive, onerous and boring. Hence my reliance on Excel’s find feature, for which the following search string stubs were used across all 11 files to ensure consistency of search: pneu, pleu, bronch, emp, pulm, asth, influ, resp & lung. These nine stubs weeded out everything I wished to nab as far as I could ascertain.

Several hours of re-coding later and I was sitting on 9 fascinating lists covering trends in disease and diagnosis terminology over the period 1901 – 2000. These are worth mulling over in their own right, and so I present them in A-Z sort order of description. Not only do the lists lengthen over time as the internationally agreed system of disease classification progresses but so do the descriptions, these reflecting the steady progress of medical science. I suggest you make a fresh pot of tea, open a decent-sized packet of something yummy, then make yourself exceeding comfortable:

We could spend hours going through this!

What we have in front of us is effectively the medical history of England & Wales in terms of trends, fashions and fads as well as technological breakthroughs and disease profiles. There are a few fascinating finds that have found my eye that I shall summarise in pithy bullet point format:

• Pneumonia was the prime coding frame for the period 1901 - 1910 just prior to the infamous ‘Spanish Flu’ outbreak of 1918 - 1919.

• Influenza came into its own during 1921 - 1930 with the introduction of four levels of differentiation instead of just one simple entry. I am going to suggest that influenza wasn’t a big deal back before 1921, but then again we’ll need to see some numbers. In this regard what do you think will happen to the statistics when the authorities go from one simple code for physicians to a posse of four? Exactly. I guess we better check this!

• Pneumonic plague caught my eye for 1921 - 1930. The next time you see a similar entry is for 1979 - 2000. Just what the heck is this all about?!

• We see a surge in descriptions during WWII, as may be expected. What is rather interesting is the dramatic emphasis on asthma, with no less than 6 entries. We also start to see occupational pneumoconiosis for the first time (coal mining and other heavy industries). Something else that catches my eye is Haemorrhagic infarction of lung. Anybody who has followed the story of aspirin will figure why my interest has been aroused.

• Occupational lung conditions flourish during 1950 – 1957; though to be picky we are talking here about coding terminology expansion rather than incidence of cases. Influenza now takes a leap to 5 diagnostic entries, and tuberculosis seems to be another concern of health bosses.

• 1968 – 1978 provides us with a hefty list of occupational and bacterial nasties. If I take a step back and try to suck the list in as a whole through a small straw I end up thinking in terms of the respiratory environment. A legacy of the UK’s massive industrial base, if you will, combined with the push for puffing ciggies, and a disregard for housing conditions.

• The industrial legacy is very much evident in the list for 1979 – 2000 as former workers in the steel mills and factories reach retirement age, and the bodies of rest of the population react to decades of various shades of pollution. But it isn’t just coal, iron and steel-working that have taken their toll: pretty much any industrial dust hasn’t gone down well with our delicate alveolar surfaces. What stands out more than any other condition is that of pneumonia whose diagnostic character now reaches a staggering 21 entries. Then we have the appearance of 11 entries for tuberculosis, which appears to be lingering sufficiently for them to worry about coding it in detail. And what the hecky-decky is plague doing there? It would seem we are going backwards in terms of the respiratory health of the nation despite all the regulations and advances, not to mention the raft of jabby jabs, drugs and pills that are supposed to be making us better! It will be interesting to see how the counts pan out.

OK, so there are a few of my first thoughts to get the ball rolling. We should note that the expansion of these lists is a most complex matter that embraces changes in medical thinking, government policy, technological advancement, resourcing, service provision, education, and economics as well as trends in the diseases themselves. Back in 1901 we started off with pneumonia differentiated 4 ways, and come 2000 we ended with pneumonia differentiated 21 ways. I’d like to think the move toward greater sophistication over time didn’t impact on counts made but the monkey on my shoulder is laughing.

A little Can Of Worms

From 1979 onward we can trace the rise and fall of Maltworkers' lung. I’m pretty sure Maltworkers were working malt prior to 1979 so assume this condition was coded as something else back then. If we call it hypersensitivity pneumonitis, then we realise we are talking about snorting in a whole raft of troubling agents for those who are particularly susceptible (molds, bacteria, bird droppings, bird feathers, agricultural dusts, bioaerosols, paints, plastics). Plastics were not around in 1901 but bird droppings certainly were, so hypersensitivity pneumonitis would also have been a thing back then, likely coded as Other diseases of respiratory system, though some physicians may have opted for asthma or even bronchitis – we cannot say for certain. This is my little can of worms for we should realise that some diseases are going to wriggle about.

With this in mind I devised a crude boiling plan, with all respiratory diseases and conditions being thrust into one of 9 major categories:

• Influenza (with or without mention of pneumonia)

• Pneumonia (viral, fungal, bacterial)

• Tuberculosis

• Asthma/Emphysema/COPD

• Bronchitis/bronchiolitis/bronchiectasis

• Pneumoconiosis (arising from industrial dusts and irritants)

• Pneumonitis (arising from inhalation of fumes, vapours, food)

• Parasites/Infectious diseases (measles, anthrax, plague, mycobacteria)

• Other respiratory conditions (pleurisy, alveolar/parietoalveolar pneumopathy, lung diseases)

This first boiling of the raw ingredients should enable me to track trends over time with reasonable felicity, though I appreciate further frying may be necessary. At this point I think we all need a cuppa…

Kettle On!

1. Yes indeed, folk were still dying of smallpox after being given a jibby jab.