Some Evidence on Non-Transmissibillty of Acute Upper Respiratory Disease and Related Matters

F. Hoyle, N.C. Wickramasinghe & J. Watkins

Abstract

Acute upper respiratory tract infections account for a significant level of morbidity and sometimes mortality throughout the world. The reported incidence of morbidity ranges from 23-63% of the population being affected in any six-month period (1). Such infections are mainly of viral origin, the principal viruses implicated being Influenza A, Respiratory Syncytial Virus, Parainfluenza, Adenovirus, Rhinovirus, Enterovirus, Coxsackie and Echo virus (2,3). These viruses are isolated the world over and show well-defined patterns of global incidence. Despite a coordinated international endeavour directed at tracking down both the temporal and spatial behaviour of such viruses, there seems at the present time to be little understanding with regard to their origin and mode of propagation. We show here that a wide range of observational data fit well to a model involving atmospheric incidence and transport of these viruses.

1. Introduction

It is generally thought that acute upper respiratory tract infections are caused by the intake of viral particles that were previously exuded by some other person, or in rare cases by some other animal. Yet little or no evidence capable of standing up to critical analysis has ever been presented in support of this widespread opinion, which appears to have arisen through historical accident rather than through accurate observation and experiment. Following Pasteur's classic experiments on alcoholic fermentation and silkworm diseases it became established that some human diseases arise from the transmission of bacteria from person-to-person, and since in the later decades of the 19th century there was no appreciation of the difference between viral and bacterial diseases the concept of infection by person-to-person transmission became applied to all diseases. This point of view appeared to gain support when in 1892 Pfeiffer (3) mistakenly implicated the bacillus H. Influenzae as the causative agent of influenza. A few epidemiologists, notably Charles Creighton in Britain, continued to protest that the evidence contradicted the rising tide of medical opinion, but in an age when few students have the leisure, affluence and inclination to examine the facts for themselves, an age in which individual enquiry has become increasingly replaced by the authority of teacher and textbook, the 19th century belief became set rigid in the modern educational system.

Once a false belief becomes established it is very difficult to get it out, essentially because the system invents supposed facts in order to support it. Two of the present authors (4) had an experience of this process in action following the peculiar epidemic outbreaks of influenza A in the winter of 1977-78, peculiar because of the return of the influenza subtype H1N1 with a variant dating apparently from the year 1950. It is commonly stated that the person-to-person transmissibility of influenza is proved by very high attack rates in institutions such as barracks and boarding schools. Yet a survey of boarding schools in 1977-78 involving a total of more than twenty thousand pupils with a number of victims estimated to be some 8880 for an average attack rate of about 30%, yielded the distribution of attack rates shown in the histogram of Fig.1.

Figure 1. Histogram showing distribution of influenza attack rate among independent schools in England and Wales during the 1977-78 pandemic.

In fact, only three schools out of more than a hundred at the extreme upper end of an approximately exponential distribution had the very high attack rates which have been claimed to be the norm.

All the diagnoses involved in these data were made by school medical staffs in advance of our enquiries. Possibly other respiratory infections became associated with influenza in the diagnoses, but since January and February 1978 were months of influenza epidemics, and since children of school age had no established immunity against influenza H1N1, the bulk of the reported cases were very likely of this disease. And even if, in the absence of isolates or serological tests, one were sceptical of explicit diagnoses, the cases were certainly of acute upper respiratory infection, to which just the same remarks and conclusions would apply regardless of the explicit viruses involved. The schools in question were fee-paying, all with boarders sleeping together in dormitories. The degrees of association of pupils in dormitories, classes and at meal times were not much different from one school to another, and if the virus or viruses responsible for the 8880 cases were passed from pupil to pupil, much more uniformity of behaviour would have been expected. Already in Fig. 1 we have evidence of great diversity, with a hint that the attack rate experienced by an individual school depended on where it was located, with some schools being in fortunate places and some in unfortunate places.

The alternative to the person-to-person transmission of a virus is that it falls from the air. For semantic convenience we refer to falling from the air as vertical incidence and to person-to-person transmission as horizontal transmission. Although in this article we are concerned to argue the case for vertical incidence as the cause of most acute upper respiratory infections, it is to be emphasised that we are not making this claim for all viral diseases. While we think that all viral diseases arise in the first place by vertical incidence, it is possible for a virus to establish a reservoir in the human population such that the chance p of contracting the associated disease by human contact is greater than the chance q of contracting the disease through vertical incidence. Normalising so that p+q = 1, there are the possibilities p >> q, p @ 1; p @ q @ 1 ; p << q, q @ 1. Diseases in the first of these categories are truly- infectious and can be moderated greatly through the old-fashioned method of isolating victims. Indeed one could say that it is just those diseases, as for example, smallpox and scarlet fever, which the medical profession found to be successfully treated by isolation, that constitute the truly infectious category p >> q, p @ 1. In this article we are concerned with the opposite more numerous category, p << q, q @ 1, which includes most acute upper respiratory tract illnesses. Data for measles, the discussion of which goes beyond the scope of this article, suggests that measles belongs to the intermediate category p @ q @ 1 . We suspect it is the intermediate nature of measles, which explains why the medical profession is divided in its opinions on whether the isolation of victims would or would not effectively stamp out the disease. If our assignment of measles is correct, isolation would appreciably reduce the number of cases but would not stamp out the disease. To stamp out a disease q must be strictly zero, requiring that the input of a virus to the Earth's upper atmosphere shall have ceased.

As regards the input of viruses to the Earth's atmosphere, the particles responsible for the strong ultraviolet component of the zodiacal light must have radii of order 30 nm, the scale of viruses. The density of such particles necessary to explain the observed strength of the zodiacal ultraviolet is remarkably high, implying an addition of ~ 10⁴ tons per year to the Earth's atmosphere, a total ~ 10²⁶ particles added annually. This number may be compared with an epidemic of disease in which each of ~10⁹ humans sheds ~ 10¹¹ viral particles for a total shedding of ~10²⁰ particles. If only a small fraction of the small zodiacal particles are viruses, if only a small fraction maintain viability, and if only a small fraction interact pathogenically with terrestrial plants and animals, the incident number would nevertheless be so vast that it could dominate horizontal transmission, even under extreme epidemic conditions.

It is commonly assumed that viral diseases are caused by the input to a victim of particles that are substantially identical to the output of viruses from the victim. This assumption is a necessary concomitant of horizontal transmission, but it is not necessary for vertical incidence. All our data and all our arguments require a causative agent or trigger to fall from the air, but the resulting disease could be caused by the association of the causative agent with dormant viral particles present already in the victim. Or the whole virus could be involved as with horizontal transmission. The evidence we shall present does not distinguish these possibilities.

Bacterial diseases can also be thought of in terms of the categories p >> q, p @ 1; p @ q @ 1/2; p << q @ 1, but with the last category, less common than for viral diseases, i.e. dominant vertical incidence being less common. One bacterial disease that is difficult to explain except by vertical incidence, however, is whooping cough. Pertussis has for long been known to occur in cycles of about 3.5 years, which used to be explained on the density of susceptibles theory, namely that after children susceptible to the disease become exhausted by a particular epidemic it was then supposed to take about three and a half years for new births to rebuild the density of susceptibles to the level at which a further epidemic would run. Thus the periodicity on this theory should have been a function of population density, with the shorter periods being found in inner city areas of very high density and either long periods or no periodicity at all in lightly populated country areas. But the periodicity was found to be everywhere the same, in town and country alike, and from one country to another. Figure 2 shows the record of notifications for England and Wales over the period 1940-82. If the theory had been correct, the sudden reduction in the density of susceptibles brought about in the 1950s by the introduction of an effective vaccine should have greatly disturbed the periodicity, or even destroyed it altogether. Yet the periodicity persisted exactly as before, but with the total number of cases much reduced.

Figure 2. Whooping cough notifications in England and Wales from 1940-1982.

2. General Evidence Against the Horizontal Transmission of Certain Diseases

If infectious diseases were propagated from person-to-person according to the commonly-held view then people living in high-density city areas should be significantly more subject to disease than people living in lightly-populated areas. From normalised attack rates plotted as a function of population density it would be possible therefore to prove the correctness or otherwise of this point of view. The circumstance that such data do not appear to exist, despite the cogency they would have, is interesting psychologically. Whereas people are avid to collect the slightest scraps of information that support conformist opinion, they are unremitting in their determination not to collect, or even to notice when collected, data which prove the opposite. It really needs no more than the absence of this simple but critical information to see that the commonly-held view must be wrong.

One can say in general terms that if any major discontinuity existed between town and country the population at large would easily be aware of it. Also in general terms, one of the present authors has shifted on occasions from general practice among a major city population to standing locum in highly rural areas, without any difference in morbidity patterns being qualitatively apparent. Thousands of general practitioners must have experienced similar comparisons without any discontinuity of pattern being emphasised and reported. On a more quantitative level, Fig.3 shows data collected by Dr.P..Jenkins, the Community Health Officer for the City of Cardiff. It gives data covering the three diseases of so-called infective jaundice, whooping cough and measles, obtained quarterly from the heavily-populaled Cardiff city area (after normalising to 100, 000 population) and from the Vale of Glamorgan, much of which is very rural. Thus each disease in each quarter of a

Figure 3. Quarterly incidence of whooping cough, measles and infective jaundice in the City of Cardiff vs The Vale of Glamorgan.

period of three years yields a point in Fig. 3. This is except for measles which was so prevalent in one particular quarter that the corresponding point for that quarter could not be plotted without prejudicing the scale of the figure. The one missing point lies on the line defined by the other eleven points, but far away to the right of the figure. Such bias as one can see in Fig. 3 goes the wrong way for horizontal transmission. It is the lightly populated Vale of Glamorgan that on a normalised basis appears worse affected. Of course one can always invent the hypothesis that standards of reporting are higher in country practices than in the cities, but one of us as a general practitioner in a city area would naturally dispute this suggestion. At all events, general experience, together with the data of Fig. 3, suggests that there is no marked difference between town and country, as one would expect for vertical incidence but not as one would expect for horizontal transmission.

Ockham's razor warns us against inventing a 'multiplicity of hypotheses', a warning which some have seen fit to interpret as an edict proscribing the consideration of new ideas. What the warning really means is that we should be on our guard against the invention of a multiplicity of unsubstantiated hypotheses in order to defend conformist views against awkward facts. For example, it is in our opinion a vacuous hypothesis to suppose that city populations possess greater immunity against infectious diseases than rural populations, and to such an extent that the greater exposure which city populations experience with respect to person-to-person transmission is almost precisely compensated by their greater immunity. A similar vacuous hypothesis would also be required to explain why individuals whose occupations involve exceptional hazards with respect to person-to-person transmission, for instance dentists and cashiers in banks, newsagents and large stores, nevertheless have records of upper respiratory infections that are not noticeably abnormal.

Looking over the case notes dating from 1970 in the practice of one of us (J.W.) 16 pairs of twins with ages between 6 months and 14 years were identified. Of the 118 instances in which one twin was consulted for acute upper respiratory infection the corresponding twin succumbed to a similar infection in 28 instances. Since twins in the age range in question are found almost perpetually together, the opportunity for person-to-person transmission would be maximal in these twin-to-twin relationships Yet in only 24 percent of instances did the second twin become a victim, which is not an impressive fraction, particularly as attack rates during epidemics of upper respiratory infections tend to run typically at about the 25 percent level among the general population. An epidemic will not run according to horizontal transmission unless each victim infects at least one other victim, thereby establishing a supercritical chain relation. A transmission probability as low as 0.24, which this data for twins yields on the horizontal transmission hypothesis, would therefore be quite insufficient to establish a supercritical chain. Only if a stricken twin contrived to infect others much more readily than his or her own twin could adequate transmission be attained.

Following in the steps of Charles Creighton, Edgar Hope-Simpson was the first person in recent years to bring the hypothesis of person-to-person transmission rigorously under the hammer. Hope-Simpson had the idea of defining a set of households by the condition that one member succumbs initially to Influenza A. He then observed the subsequent fates of other members of the households thus defined, finding them to develop no greater proportion of attacks than would be expected for the population at large (5,6). Figure 4 gives Hope-Simpson's results for epidemics of H3N2 in 1968/69 and 1969/70, shown in the histograms as I and II respectively. Besides the fraction of subsequent cases being normal for the population at large, no well-defined subsidiary peak occurred 1-2 days after the first cases were reported, as would be expected from incubation if horizontal transmission had been occurring. Hope-Simpson's results have been fully confirmed by Mann et al., (8) and by Philip et al., (7)

The unusual circumstances in 1984-85 that few true cases of Influenza A were reported anywhere in the world up to mid-February 1985 (INFLU Centre, London, (9)) permitted the behaviour of other sources of upper respiratory infections to be examined in a manner similar to that used by Hope-Simpson. Starting in May 1984, a total of 80 households were defined, again by the criterion that one member presented themselves

Figure 4. Percentage attack rates in households where one member succumbs to influenza in the epidemics of 1968/69 and 1969/70 at Cirencester, England according to data from Hope-Simpson.

with an acute upper respiratory infection, and then subsequent histories of other household members were studied by one of us (J.W.), with the results shown in Fig. 5. The interesting point emerges that upper respiratory infections quite generally are like influenza, without evidence of person-to-person transmission, which if it had occurred would have caused an incubation peak of cases to occur two or three days after the initial attacks on day zero.

Figure 5. Percentage cases of acute upper respiratory infections in households where one member succumbs to influenza on day 0, from data collected by Dr. J. Watkins in 1984/85.

A fraction of the fee-paying schools in the survey already mentioned had both day pupils and boarders. The boarders were exposed to close person-to-person contacts for 24 hours a day, whereas the day pupils were only some 8 hours at school, with the remaining 16 hours spent under non-institutional conditions, conditions having fewer person-to-person contacts generally. If there were any substance to the claim of high attack rates in institutions, used to bolster the person-to-person transmission hypothesis, the overall attack rates on boarders should have been significantly higher than it was on the day pupils. With each of the schools in question represented by a point in Fig. 6, the results gave essentially a scatter diagram.

Figure. 6. Correlation of attack rates for Day pupils and Boarders for schools which had a mixture of both during the 1977/78 influenza pandemic in England and Wales.

Whatever slight bias there is about the 45° line in this diagram disappears for a line of slope 40°, and this is within the expected statistical fluctuation. There are many instances in which the day pupils experienced considerably higher attack rates than boarders, a situation that defies the imagination to explain according to person-to-person transmission, for we would have to suppose that after leaving school the day pupils encountered more seriously infective contacts than were present at school, and that they did so systematically in order to explain high attack rates above 70 percent for the day pupils in some of the cases.

In the next section, we shall see that vertical incidence is expected to lead to intricate patchy details in attack rates, with some localities relatively safe from attack and others relatively dangerous. On this view, schools that happened to be in relatively safe areas would have their boarders staying comparatively safe the whole time, whereas the day pupils would go out from the school into comparatively dangerous areas and so would experience significantly higher attack rates. And of course the opposite situation would occur for other schools, thereby producing the scatter shown in Fig. 6.

Figure 7. Deviations of attack rates of influenza above the mean attack rate for the 25 school houses at Eton College during the 1978/79 pandemic. The deviations are relative to the standard deviation computed house by house.

Attack rates of around 30 percent were found most useful for studying variations within school boundaries, since very high attack rates evidently preclude variations being found, while low attack rates gave inadequate statistical weight. Eton College had 441 victims among 1248 pupils, for an attack rate of 35 percent, with high statistical weight because of the large number of pupils. We were fortunate that Dr. .J.Briscoe, the Medical Officer at Eton, had for long been puzzled to understand how his observations could be explained in terms of pupil-to-pupil transmission. Consequently, Dr. Briscoe had collected comprehensive information giving the distribution of victims in some 25 school houses. The houses averaged about 50 pupils each, with about 17 cases expected as the mean number of victims. Such numbers were very suitable for computing standard deviations, with the results shown in Fig. 7. Two houses had excess morbidities of 4 standard deviations, two had deficits of about 4 standard deviations, while one house (COLL) had a remarkable deficit of 6 standard deviations. Since pupils in the different houses were mixed in classes and at games these enormous fluctuations from a random distribution are quite inexplicable it seems to us in terms of horizontal transmission. The Eton College results imply that the school was hit vertically by the influenza virus during the night hours, or possibly at a weekend, and that the vertical incidence was patchy enough to distinguish between the locations of the various houses, some houses happening to lie in safe areas and others in dangerous areas. Dr. Briscoe informed us that similar effects had occurred in other influenza epidemics, with the identities of the lucky and unlucky houses being different from the situation in 1978. A patchy vertical incidence would of course not be reproducible in its details from one epidemic to another, so this too would accord with the vertical incidence hypothesis.

We end this section with a somewhat different issue. Younger children have sometimes been found to be more susceptible to influenza than older children. Usually it is not possible to distinguish how far the greater resistance of older children is inherent and how far due to already established immunities. Since no children of school age in 1978 had any established immunity to the H1N1 type, and since some schools in our survey had both junior and senior pupils, it was possible to compare attack rates that gave information largely free of the immunity factor. Results are shown in Fig. 8 where each point refers to a school having both junior and senior pupils. These data suggest that inherent resistance has little to do with age, implying that differences observed in other years were related to the immunity factor. Such asymmetry as one sees in Fig.8 about the 45° line (after noting the two very low points marked heavily to catch the eye) would be removed by increasing the 45° slope to a slope of 50°. A slight bias in this sense could have arisen from a minority of cases where the virus type was still H3N2 rather than H1N1, with older children having better immunity to H3N2.

Figure 8. Correlation of influenza attack rates for Junior and Senior pupils (Junior, 5-13 years; Senior ³ 14 years) during the 1978/79 pandemic.

3. The Vertical Incidence Theory

According to medieval lore diseases come from comets, and according to our view this is true, but only in a broad sense. We cannot maintain the dramatic position that ferocious new diseases come from spectacular comets, because for every spectacular comet there are almost certainly very many smaller ones. The smaller comets may not only evaporate more material collectively than large ones but the effect of the Earth crossing almost precisely the track of a small comet would lead to a greater addition of evaporated particles to the terrestrial atmosphere than would a more distant relationship to a large comet, as for instance a distant relationship to Halley's comet. Support for this position comes from an analysis by Z. Sekanina (10) of some 20,000 orbits of meteors, meteors being small particles typically with sizes ~ 0.1-1 mm also evaporated from comets. A minority of the orbits could be associated with known comets but the majority could not. It is perhaps possible to understand both the origin and the demise of the medieval lore in these terms. Over a time scale of historic length there must have been special situations with the Earth in close proximity to a large comet. If following such special situations serious attacks of disease occurred, particularly if entirely new diseases appeared, their association with the comets would be an obvious and natural deduction. But then as other less close comets appeared in the sky, and were not followed by spectacular outbreaks of disease, the belief would be thrown first into doubt and then into ridicule.

Figure 9. The orbits of known short-period comets lying mostly in the region between Mars and Jupiter.

Figure 9 gives a rather mild idea of the complexity of the situation. It shows only the orbits of the so-called Jupiter family of comets. One has to imagine thousands upon thousands of orbits of smaller comets added to Fig. 9 in order to appreciate the real position. At all events, we can see that the Earth is perpetually embedded in a halo of evaporated cometary material, some of the material newly evaporated, without much in the way of exposure to solar ultraviolet light. Following the development of the panspermia theory by Svante Arrhenius (11) early in the twentieth century, it became fashionable to discount the theory by claiming that micro-organisms in space would be destroyed by UV, by X-rays, by low pressure, by temperature etc. The circumstance that all these claims have turned out to be untrue is indicative of the correctness of the space-borne theory. All that was wrong with the panspermia theory was that it did not go nearly far enough. While data on the radiation hardiness of viruses is less complete than for other kinds of micro-organism, on the general argument that simpler systems are usually more hardy than complex ones, viruses can be expected to withstand conditions in space at least as well as bacteria and algae. The following is a quotation from R.B.Hoover, F. Hoyle, N.C. Wickramasinghe, M.J.Hoover, and S. Al-Mufti (12):

The discovery of such properties reverses sharply the logic used formerly against the panspermia theory. Instead of the problem being survival in a space environment, a seemingly insoluble problem now confronts conventional theory, namely to understand how micro-organisms that are supposed never to have been outside the terrestrial environment could have acquired properties so remarkable and so profoundly suited to a space environment. If biologists were to apply with any conviction their theories on the necessity for selection to produce positive properties in evolution, they would see that conventional theory is destroyed.

Because the Earth and the finely-divided cometary material in the halo around the Earth are not comoving, the high terrestrial atmosphere is essential if micro-organism are to make a soft landing here. Speeds relative to the Earth are so high that micro-organisms would be destroyed by hitting a hard surface, for instance the surface of the Moon. The maximum size for the safe entry of biological material is ~0.1 mm. This is under the most favourable conditions of speed and geometry. For smaller sizes, those of bacteria and virsuses, the situation is not restrictive, however. Particles of the sizes of bacteria and viruses, indeed particles not larger than 0.01 mm, land 'soft' in the high atmosphere, permitting them to retain viability, and subsequently to fall gently downwards as active agents.

4. The Descent of Small Particles Through the Terrestrial Atmosphere

The lower atmosphere, the troposphere, has a height which varies from about 18 km in the tropics to 10 km in temperature latitudes to 7 km in polar regions. Small particles over the whole size range from viruses with diameters ~100 nm up to colonies of bacteria descend comparatively quickly from the top of the troposphere to ground-level. The troposphere is a region of falling temperature with increasing height, a physical condition permitting vertical air movements to occur readily, thereby causing water vapour to be carried upwards from the surface regions to essentially the top of the troposphere. The falling temperature with increasing height makes the water vapour supersaturated, but the temperature is not usually so low that the supersaturated water vapour condenses spontaneously into ice crystals. Initially-existing nuclei are required for the water to condense around, and the small particles in question provide such nuclei. Thus small particles on reaching the troposphere from above become condensation centres around which much larger ice crystals form. Because they are less impeded by air resistance than the original small particles themselves, such ice crystals fall with much increased speeds. As they descend into warmer air the ice crystals usually become melted and the resulting water droplet may either fall to the ground as rain, or it may become partially evaporated, and as a smaller droplet remain suspended in the air. Normal precipitation rates are such that this process, often involving repeated cycles of condensation and evaporation of the ice crystals and water drops, serves to wash-out the troposphere of small particles in a time-scale of a few weeks. Exceptional conditions may extend the time-scale, however, for example in desert regions or as in the unusually cold weather experienced in northern latitudes during January and February 1985.

Above the top of the troposphere, the tropopause, the temperature undergoes inversion through the stratosphere, rising from ~ -55°C at the tropopause to a maximum of ~ 3°C at a height of ~50 km. The physical reason for this temperature inversion is that the ozone in the height range in question, the stratosphere, absorbs solar ultraviolet light very strongly shortward of 3000 A, thereby giving an energy input into the stratosphere. The dynamical effect of the temperature inversion is to inhibit greatly the generally free movement of air such as occurs in the troposphere. Travellers by air will be familiar with the difference between the clarity of the lower stratosphere into which airplanes normally climb and the cloudy turbulent troposphere below. Free air movement in the stratosphere is limited to west-to-east movements along parallels of latitude, of which the most violent are the jet streams. The effect of the free west-to-east movement is to produce a general uniformity with respect to longitude in the stratospheric distribution of small atmospheric particles. If the Earth were smooth at its surface, we would therefore expect any incoming pathogens from space to arrive at ground level at more or less the same times along a given parallel of latitude (although local weather patterns could still introduce modest fluctuations). But because the troposphere has a marked dependence of height on latitude, we would not expect different latitudes to behave similarly, unless the particles happened to be so large that they were able to fall rapidly without much regard to air resistance.

Since the surface of the Earth is actually not smooth, in particular the Himalayas project about halfway up to the stratosphere, the rule of contemporaneous incidence along parallels of latitude is likely to be inappropriate in regions of high relief. Nevertheless, with the exception of the Rocky Mountains of North America there is a belt around the Earth from about 45°N to 60°N where the land is not much above sea-level and where the rule should be applicable. Prague, the capital city of Czechoslovakia lies a little north of 50° and Cirencester in England has a similar latitude within about 1°. Hope-Simpson has noted the similarity shown in Fig. 10 between his influenza records for the Cirencester district and the Czech records, (13).

Figure 10. The attack rates of influenza in Prague and Cirencester after Hope-Simpson

A particle of radius 10 m m falls through the lowest ten kilometres of the stratosphere at a speed of ~ 1 cm/s and thus takes only a few days to cover what for smaller particles is the slowest part of their downward journey. All particles fall comparatively rapidly through the upper atmosphere above the stratosphere, and then more and more slowly down through the stratosphere. A particle with the size of a typical bacterium, ~1 m m, falls through the lowest ten kilometres of the stratosphere at a speed of about 2.10^-2cm s^-1 and thus falls in a time-scale of ~5.10⁷ s, i.e. about 2 years. Because there is more of the stratosphere through which such a particle must fall in high latitudes than in the tropics (recalling that the tropopause is higher in the tropics) the slow part of the journey is more extended the higher the latitude. A bacterium falling in ~ 1 year in the tropics would fall in ~ 2 years in temperate latitudes and in ~ 2 to 3 years towards the poles, a situation that is broadly consistent with historical records of the dates of outbreak of the Black Death in various latitudes, as shown in Fig. 11. It is interesting to notice the perturbation of contours of contemporaneous outbreak produced by the Alps, which mountains rise to about half the height of the tropopause.

If a particle of the size of a typical virus, a particle say with diameter ~0.1 m m, fell under gravity through still air the time-scale for the slowest part of the journey through the bottom ten kilometres of the stratosphere would be ~ 10⁹ s, i.e. about 30 years. This is so slow that other means of descent involving large scale air movements in the stratosphere have to be considered. Although vertical mass movements of air are feeble compared to those in the troposphere, some vertical stratospheric movement takes place despite the inhibiting effect of the inverted temperature gradient. The physical cause of mass stratospheric movements is the equator to pole temperature difference which is available to work a heat engine crossing parallels of latitude, a heat engine that operates more strongly the larger the temperature difference - i.e. much more strongly in winter than in summer. A similar consideration applies also in the troposphere, where an engine crossing parallels of latitude transfers heat from tropical regions towards the poles, again more in winter than in summer. The heat engine in the troposphere is that which we experience in cyclonic storms.

Figure 11. The spread of the Black Death through Europe

Ozone measurements can be used to trace the mass movements of air in the stratosphere. Such measurements show a winter downdraft that is strongest over the latitude range from 40° to 60°. Taking advantage of this annual downdraft, individual viral particles incident on the atmosphere from space would therefore reach ground-level generally in temperate latitudes, which therefore emerge from these considerations as the regions of the Earth where upper respiratory infections are likely to be most prevalent, once again on the supposition that the Earth is smooth. The exceptionally high mountains of the Himalayas, rearing up through most of the height range to the stratosphere, introduce a large perturbation on the smooth condition, which may be expected to affect adversely this particular region of the Earth, especially regions lying downwind of the Himalayas, particularly China and S.E. Asia. In effect, the Himalayas are so high that they could act as a drain plug for most of the viruses incident on the atmosphere at latitude ~30°N, the large population of China being inundated by this drainage effect, making China the quickest and worst affected region of the Earth. Concomitantly, other parts of the Earth at ~30°N should be largely free of viral particles, unless it happens that such particles are incident as components within larger particles.

A direct demonstration that the general winter downdraft in the stratosphere occurs strongly over the latitude range 40° to 60° was given by Kalkstein (14). A radioactive tracer, Rh-102, was introduced into the atmosphere at a height above 100 km and the incidence of the tracer was then measured year by year through airplane and balloon flights at altitudes ~20 km. The tracer took about a decade to clear itself through repeated downdrafts of the form shown in Fig. 12. Noting that the ordinate scale is logarithmic, the incidence of the Rh-102 is seen to be much greater in temperate latitudes than elsewhere, with the period January to March the dominant months.

Figure 12. The fall out of Rh-102 at various latitude intervals from the HARDTACK atmospheric nuclear bomb which was exploded on 11^th August 1958.

The observed incidence of a radioactive tracer agrees closely with the well-known winter season of the viruses responsible for the majority of upper respiratory infections, including influenza. Figure 13, (taken from Communicable Disease Report CDR 83/149, Public Health Laboratory Service,) shows the year-by-year incidence of respiratory syncytical virus, demonstrating a remarkable temporal concurrence with the radioactive data of Fig. 12. How we wonder is the almost clockwork regularity of RS infections to be explained otherwise? Unfortunately so little has been understood of the mode of attack of so-called infectious diseases that almost any form of hypothesis has come to be accepted in the past as an answer to questions of this sort. The truth is that, although the world may be extremely complex it is nevertheless extremely precise, with explanations every bit as clear-cur as that of the quantum mechanical analysis of the energy levels of the hydrogen atom being ultimately available for every phenomenon we observe.

5. Details of Vertical Incidence

It was remarked above that the world is extremely complex, and it is interesting to notice how true this can be, even for the first steps in the acquisition of small cometary particles by the Earth, for the Earth could cut just once on a unique occasion through a trail of evaporated particles, or it could cut periodically or irregularly at both closely and widely spaced time intervals. Besides which a pathogenic agent could be carried by a distribution of particles with varying sizes that descended through the terrestrial atmosphere in quite different intervals of time according to the discussion of the previous section. To recapitulate, particles with sizes ~10 m m fall under gravity everywhere over the Earth in only a few weeks. Particles with sizes ~1 m m fall in a few years, sooner in low latitudes than in high, while particles with sizes ~ 0.1m m have a winter season from about December to March (in the northern hemisphere) when they are carried down through the lower stratosphere by mass movements of air, a process that occurs dominantly over the latitude range ~40° to ~60° and which would lead to a cyclically regular disease pattern like that of Fig. 13. To these previous considerations we must note that particle sizes can change not only due to water drop formation in the troposphere but to the aggregation of one particle with another produced by sticking effects should the particles become coated by acid as they pass through a sulphur layer of volcanic origin at heights of ~ 20 km. Thus time intervals could be appreciably affected by the varying exudations of SO₂ from volcanoes, as well as by the already mentioned effects of high mountain ranges. Finally, we have to consider the complexities that can arise at ground-level itself, complexities giving rise to the local variability in attack rates of a disease such as for Eton College in Fig. 7.

Figure 13. Incidence of RS infections in England and Wales (CDR reports)

The epidemiologist observes the net outcome of all these complexities, with the situation so scrambled together as to present an almost impossible problem of unscrambling at any rate when the situation is treated empirically. Only with the aid of a model allied to observation can progress be made, as for instance the model of the winter downdraft in the stratosphere (Fig.12) leading to an understanding of the pattern of RS infections (Fig. 13). It is a further consequence of this model that similar effects should occur in the southern hemisphere, but six months different in phase because of the alternation of summer and winter in the two hemispheres. We do not have comparable data for RS infections, but influenza behaves similarly and data for influenza confirms the prediction of the model. Thus Hope-Simpson noted the phase variation in the occurrence of influenza across the continent of Africa over the period 1950-51, while we ourselves have assembled in Fig. 14 data issued by public health authorities in Sweden, Sri Lanka and Melbourne, Australia over periods ranging from 5 years to more than a decade. The global inference from the model is thus confirmed.

Figure 14. Incidence of Influenza A in three separate countries

We have chosen to show the data for Sweden rather than Britain, not because there is any important difference between Sweden and Britain, but to bring out the point that the simple physical cold of winter is not a relevant factor. Sweden has a really cold winter, whereas Australia has a clement winter not much cooler than a Swedish summer. If simple exposure to cold were important, the effect would long ago have been demonstrated under controlled conditions in the laboratory, which it has not been.

Influenza A pandemics, following changes of virus type, do not fit the annual winter cycle in the manner of Fig. 14. Influenza pandemics fit readily into the model, however, with each major influenza type assigned to a separate accretion of virus from space, and with the viral particles being present in larger aggregates as well as individually by themselves. The larger particles fall through the atmosphere under gravity, those with sizes ~10 m m in a matter of weeks, those with sizes of ~1 m m in a year or two, and those existing by themselves with sizes ~0.1 m m taking a decade or more to reach ground-level. It is the latter extended period, characterised by so-called antigenic drift, possibly caused by solar ultraviolet light, that displays the effect of Fig. 14, after the earlier accretion of larger particles is over and done with.

Exactly where on the Earth the larger particles, responsible on this view for the initial outbreaks of an influenza pandemic, first reach ground-level is a matter of the vagaries of wind and weather, with the possibility that the larger particles reach ground-level more or less contemporaneously at widely-separated localities. If in accordance with the mistaken person-to-person transmission hypothesis, one describes the first place of incidence as the 'focus' of the pandemic, the disease would appear to spread from the initial focus to other foci, perhaps separated geographically by large distances, with an amazing and quite inconceivable rapidity. Thus in the 1918/1919 pandemic the first outbreaks occurred within hours of each other in Boston, USA and Bombay, India, an impossibility for person-to-person transmission in days before air travel (14a).

Also indicative of the short fall times of larger particles was the near contemporaneous outbreaks of the so-called Asian flu' pandemic of 1957-58, with similarities of timing that are again beyond reasonable possibility on the person-to-person transmission hypothesis, for example, the data in the following Table:

Table 1

Data on the 1957-58 'Asian' Influenza Pandemic

The localities in the above Table are at nearly the same latitude. As expected, the effect of the lower height of the tropopause at higher latitudes caused a delay of some weeks, with the initial outbreaks in Alaska occurring at St. Paul (57°N) on Oct. 11 and at Campbell Island (63°N) on Oct. 30, (20).

Consider next the situation at ground-level itself. It is a matter of experience that we do not normally snuffle raindrops up into the nose or gulp them into the mouth. So if the viruses causing upper respiratory infections fall down through the troposphere inside raindrops or snowflakes it may be wondered how we contract these diseases at all? Rain which impacts the face tends to drip off the end of the nose instead of entering the respiratory tract, possibly the reason for the possession of a nose. But rain does not end because all the water has fallen out of the atmosphere. Rain ends because falling droplets evaporate before reaching the ground. Droplets evaporating immediately in front of one's face, releasing viral particles into the air, would not be harmless because the released particles could then be breathed into the respiratory tract. It is therefore the end of a shower of rain that is dangerous, with the details highly local and irregular, thereby explaining why the incidence of breathable viruses at ground-level is often confined to small patches such as those responsible for the varying behaviours of school houses with respect to influenza, as in Fig. 7. The houses of Eton College which experienced unusually high attack rates were the ones where it happened that pupils were in exposed positions at just the moment when a shower of rain dried up, and conversely for the houses where attack rates were markedly low. It is also apparent why there can be no reproducibility from one epidemic to another in the identities of the fortunate and unfortunate houses, the relation of particular places on the ground to the turbulent swirl of falling drops being essentially a matter of chance on distance scales of the order of the precincts of a school. But not entirely a matter of chance on a scale of miles or on the scale which separates town and country. The efflux of heat from a large city could play a significant role in evaporating water droplets, and if the evaporation occurs high enough above the heads of city people they would be less at risk than country folk since country folk have no such self-protecting source of heat at their disposal. This may explain why attack rates of many diseases appear to be higher in the countryside than in neighbouring cities, as indicated in Fig. 3.

It must be rare for snowflakes to evaporate, because at the temperatures < ~0°C at which snow is able to reach the ground without melting into rain, evaporation rates are low. Hence cold conditions with precipitation falling as snow should on the vertical incidence model be almost free from the danger of upper respiratory infections. Neither is heavy rain a dangerous condition. It is misty, drizzly weather that provides incoming viruses with the opportunity to become dispersed in the air close to ground-level. These expectations agree very well with popular lore, according to which damp is unhealthy but sharp cold weather is healthy. Shakespeare expressed the general lore when he wrote

"……the winds…….have suck'd up from the sea, contagious fogs……".

In the exceptionally cold weather of January and February 1985 it was widely noticed that Influenza A was essentially absent throughout the world, and that in the U.K. there was an atypical absence of upper respiratory infections generally. We ourselves predicted that once damper, warmer weather set in, as it did in late February, there would be a sharp rise of such infections, as indeed actually happened, with Influenza A at last appearing over the whole latitude belt ranging from the U.S.S.R. to the western states of America. In effect, the exceptionally cold weather held off the infections which normally occur in January and February.

The above discussion has been biased to suit the situation in northern temperate latitudes. If one lives in desert conditions, other factors would be seen to be important. Precipitation in deserts tends to be one thing or the other, either heavy or absent, with little of the damp, drizzly weather of northern latitudes. Viruses falling from the atmosphere would mostly reach ground-level therefore without constituting a serious immediate threat to a desert population. Once on the ground they would largely stay there, however, instead of being washed away in streams and rivers, to be stirred-up into the atmosphere again by winds. Windy periods with sandstorms would thus be the times when upper respiratory infections appear, an expectation which according to our somewhat fragmentary knowledge of desert conditions seems to be correct (23).

The circumstance that in a vertical incidence model viruses could come upwards from the ground as well as downwards from the atmosphere raises still further complications. While most of the falling viruses to reach ground-level in regions other than deserts will be washed away by streams and rivers, some viral particles remain suspended in the subsurface water table, and some would accumulate in freshwater lakes. Evaporation provides a channel whereby water-embedded viruses could subsequently be released into the atmosphere near ground-level. There are many subchannels whereby evaporation occurs, indoors from the domestic water supply at all times of the year, and outside mostly in spring and summer. A high volume subchannel lies in evaporation from the leaves of trees, This particular subchannel is especially interesting because of the tendency of people to plant trees around their houses, and because the evaporation process cools the air around trees due to the absorption of heat necessary to supply the latent energy of evaporation. The cooled air, being denser in summer than warm surrounding air, falls immediately to ground-level thereby dumping any viruses it may contain on to unsuspecting folk living or picnicing in the shelter of trees.

The biological system of a tree probably filters out most viruses, particularly those with an affinity for cellulose, but a minority may well be expelled both by trees and foliage of all kinds into the atmosphere. The minority that are not filtered then appear as typical summer diseases. Figure 15, taken from Communicable Disease Report 83/24, shows reported cases of parainfluenza from 1980 to 1983.

Figure 15. Incidence of Parainfluenza type 3 (CDR reports)

Each year cases begin a steep rise in April, attain a maximum in June, and fall essentially to zero in September, in correspondence with the annual growing cycle of plants. It is hard to imagine any other process that could give a cycle so repeated and so perfectly timed.

6. Conclusions

On the vertical incidence theory localities of exceptionally high incidence are to be expected at ground-level. Ironically, such regions appear at first sight to give credence to the person-to-person transmission theory, because people living in a locality of high vertical incidence have an impression that they are infecting each other. Likewise whenever any of us contracts an upper respiratory infection after being in contact with an early victim of a similar infection we tend to believe we have 'caught' the infection from the earlier victim. Such perceptions lack quantitative support, however, for whenever they are examined with a little care discrepancies for the person-to-person theory soon emerge. Illustrative of this lack of quantitative support, one of us (J.W.) while working as a General Practitioner in Newport, Gwent, U.K., took the opportunity to issue a questionnaire to patients attending surgery with acute upper respiratory disease. Of a total of 705, only 179 could recall having had previous contact with a person suffering from a similar affliction, leaving 526 without having had any such contact. While there is no difficulty in seeing how in the vertical incidence theory there will be clusters of victims, in the person-to-person transmission theory it is indeed difficult to see how the bulk of cases could have arisen without any previous infectious contacts having taken place. Facts such as these lead supporters of the person-to-person theory to invent excuses, for example, to invent the postulate that the 526 cases without previous contacts with visible sufferers came from invisible contacts with latent carriers^†. This never-ceasing invention of excuses is what Ockham's razor enjoins us not to do.

Data collected on all scales, ranging from the individual surgeries of general practitioners to worldwide patterns of disease show overwhelmingly that most acute upper respiratory infections arise from vertical incidence. The data could be extended greatly at comparatively little effort and cost, as for instance the determination of the attack rates of particular diseases as a function of population density. The reason why such data is not precisely analysed is we think psychological, because it would instantly destroy conformist opinion, which in all branches of science avoids as far as possible confrontations with awkward facts. Yet the benefits to public health from a clear understanding of the cause of acute upper respiratory infections could be very great. With the modes of incidence known for various diseases only quite simple precautionary measures could well serve to reduce morbidities very appreciably, with consequent benefits economically for the community at large as well as for individuals personally.

References

Site maintained by N.C. Wickramasinghe - e-mail wickramasinghe@cf.ac.uk.
This website is supported by Fields Associates – experts in data recovery and computer forensics.
Last updated 26 October, 2005