key: cord-0862799-9s7y8guc
authors: Brundage, John F
title: Interactions between influenza and bacterial respiratory pathogens: implications for pandemic preparedness
date: 2006-04-20
journal: Lancet Infect Dis
DOI: 10.1016/s1473-3099(06)70466-2
sha: dec1cade89375bf8f754ecc6507502786a151e06
doc_id: 862799
cord_uid: 9s7y8guc

It is commonly believed that the clinical and epidemiological characteristics of the next influenza pandemic will mimic those of the 1918 pandemic. Determinative beliefs regarding the 1918 pandemic include that infections were expressed as primary viral pneumonias and/or acute respiratory distress syndrome, that pandemic-related deaths were the end states of the natural progression of disease caused by the pandemic strain, and that bacterial superinfections caused relatively fewer deaths in 1918 than in subsequent pandemics. In turn, response plans are focused on developing and/or increasing inventories of a strain-specific vaccine, antivirals, intensive care beds, mechanical ventilators, and so on. Yet, there is strong and consistent evidence of epidemiologically and clinically important interactions between influenza and secondary bacterial respiratory pathogens, including during the 1918 pandemic. Countermeasures (eg, vaccination against pneumococcal and meningococcal disease before a pandemic; mass uses of antibiotic(s) with broad spectrums of activity against common bacterial respiratory pathogens during local epidemics) designed to prevent or mitigate the effects of influenza-bacterial interactions should be major focuses of pandemic-related research, prevention, and response planning.

The infl uenza pandemic of 1918 accounted for an estimated 40-100 million deaths worldwide. 1, 2 The emergence and international spread of coronavirusassociated severe acute respiratory syndrome (SARS), 3 of H5N1 infl uenza among domestic and wild avian species, [4] [5] [6] [7] [8] and of avian infl uenza among human beings have alarmed public health and clinical professionals. [4] [5] [6] [7] [8] Plans for responding to the next infl uenza pandemic are guided by beliefs regarding the 1918 pandemic including that the pandemic was caused by a highly virulent, effi ciently transmitted infl uenza virus that evolved from an avian strain; 2,6-10 that infections with the pandemic strain were expressed as primary viral pneumonias and/or acute respiratory distress syndrome (ARDS); [6] [7] [8] 11 that pandemic-related deaths were the end states of the natural progression of disease caused by the pandemic strain; [6] [7] [8] 11, 12 that bacterial superinfections caused relatively fewer deaths in 1918 than in subsequent infl uenza epidemics; [6] [7] [8] 13 and that the clinical and epidemiological characteristics of the next pandemic will closely mimic those of the 1918 pandemic. 6, 7 In turn, vaccines and antivirals (which are in short supply and/or may not be eff ective) are the mainstays of prevention planning, and the availability of intensive care beds and mechanical ventilators to treat rapidly progressing viral pneumonias and ARDS are central to clinical preparations. [6] [7] [8] 14 However, outside of highly controlled laboratory settings, the clinical and pathophysiological characteristics of most infectious diseases are much more complex than implied by a simple "one germ, one disease" model because the pathophysiological eff ects of many infectious agents-and particularly infl uenza viruses-modify the eff ects of coinfecting agents. [15] [16] [17] [18] [19] [20] [21] [22] [23] [24] [25] [26] [27] [28] [29] The mechanisms involved in such interactions include breakdowns of physical barriers to tissue invasion; decreased mucociliary clearance activity; destruction, depression, and/or dysregulation of immune system components; enhanced expressions of receptors by, and/or increased adherence to, epithelial cells; increased aerosolisation and dispersion of coinfecting agents; production of antibodies that block immune responses to other agents; up-regulation of expressions of genes that code for toxins; and so on [15] [16] [17] [18] [19] [20] [21] [22] [23] [24] [25] [26] [27] [28] [29] (fi gure 1). In turn, interactions among infl uenza and coinfecting bacterial respiratory pathogens, for example, often determine clinical attack rates, the nature and severity of clinical expressions, and the mode, direction, and velocity of spread of coinfecting agents in various populations, locations, and settings.

I review some of the clinically and epidemiologically signifi cant interactions between infl uenza and common respiratory bacterial pathogens, particularly in relation to the 1918 pandemic. Contemporaneous reports regarding the 1918 pandemic bacteriological fi ndings, and interpretations of their clinical and epidemiological relevance, must be interpreted cautiously because, for

Febrile respiratory illness (self-limited) Viral pneumonia Acute respiratory distress syndrome 

In the winter of 1916-17, a highly virulent respiratory illness ("purulent bronchitis") attacked a large British army camp at Etaples in northern France. The clinical presentation was "so distinctive as to constitute a defi nite clinical entity". 30 At the height of the epidemic, purulent bronchitis accounted for nearly half of all necropsies that were done in the aff ected area. Cultures of 20 sputum specimens from purulent bronchitis cases revealed Bacterium infl uenzae (now Haemophilus infl uenzae; 90%), pneumococcus (65%), streptococcus (25%), and staphylococcus (15%). Of note, specimens that yielded high numbers of B infl uenzae were nearly always coinfected with pneumococci. 30 About the same time, a similar illness was aff ecting soldiers in large numbers at the British army camp in Aldershot, England. The clinical course was characterised by profound respiratory distress, "a peculiar dusky heliotrope type of cyanosis" 31 (fi gure 2) and high case fatality (approximately 50%). In eight consecutive cases, seven had "copious growths of B infl uenzae" associated with pneumococci. 31 Medical offi cers at the camp concluded that "these patients suff er from a primary invasion of their lung tissues by the B infl uenzae, and that pneumococci, present at the same time, are at fi rst of low virulence … Exaltation of the virulence of the pneumococci by symbiotic growth with B infl uenzae would appear to follow." 31 The emergence of a distinctive, highly virulent disease in military camps in France and England suggested to Abrahams and colleagues 31 that purulent bronchitis "is more widespread than has hitherto been recognized".

Beginning in October 1917, there was a substantial rise in infl uenza-related morbidity among American troops in France. After his review of clinical and necropsy fi ndings from US Army hospitals in France, MacNeal 33 had "little, if any, doubt" that the disease that attacked American soldiers late in 1917 was essentially the same as the purulent bronchitis that attacked British camps in France and England earlier that year.

In 1919, Abrahams and colleagues 32 reviewed their extensive experience during the 1918-19 infl uenza pandemic. They reported that Streptococcus pyogenes longus (36%), pneumococcus (29%), and B infl uenzae (25%) were recovered relatively frequently from cultures (n=28) of lung tissue of fatal "infl uenzal pneumonia" cases. They noted that "in essentials the infl uenzapneumococcal 'purulent bronchitis' that we and others described in 1916 and 1917 is fundamentally the same condition as the 'infl uenzal pneumonia' of this present pandemic." In both conditions, the virulence of the secondary organisms appears "to be exalted by the initial infl uenzal infection".

Recently, Oxford and colleagues 34 postulated that the outbreaks of purulent bronchitis in military camps in France and England in 1916-17 were progenitors of the nearly simultaneous outbreaks of infl uenza that aff ected countries throughout the world in 1918-19.

In the USA, the infl uenza pandemic was fi rst manifested on a large scale in late winter of 1918 at Camp Funston, KS. In March, approximately 1100 soldiers assigned to the camp were hospitalised for infl uenza-approximately 22% developed pneumonias (>90% lobar), of which approximately 20% were fatal. Opie and colleagues 35 reported that "the greatest incidence of pneumonia aff ecting troops in this camp occurred … coincident and immediately following the outburst of infl uenza, the maximum of pneumonia being fi ve days after the maximum for infl uenza". 

In the autumn of 1918, "an epidemic of epidemics" attacked military camps in general. 36 Within 8 days of the start of the index epidemic at Camp Devens, MA, 11 other camps had been aff ected; by the end of September, 31 camps had been attacked; and by the end of October, all large camps in the continental USA had been aff ected (fi gure 3). 36 The natures, dynamics, and eff ects of the epidemics that aff ected the widely separated camps were remarkably similar. In general, they had explosive onsets, nearly as rapid declines, and durations of 3-4 weeks (fi gure 4). Within approximately 1 week after the start of each epidemic, there was an "ominous prevalence of pneumonia. The pneumonia [did] not exist as a separate epidemic, but it [was] always a follower of infl uenza". 36 At the various camps, the epidemic curves of pneumonias and deaths lagged by approximately 7-10 days behind those of infl uenza (fi gure 4). Overall, there was a remarkably strong relation between cases of infl uenza each week and pneumonia-related deaths the following week (pneumonia-related deaths each week=0·0635 × infl uenza cases the previous week; R 2 =0·98). 36 Soper 36 concluded that "the infl uenza paves the way for the pneumonia, if it does not actually produce it".

Descriptions of epidemics at various camps-and impressions of medical offi cers who responded to them-were also remarkably similar. For example, at Camp Devens, in 1 month beginning September 12, approximately one-quarter of all troops were diagnosed with infl uenza; of them, approximately 17% developed pneumonias, of which approximately 35% were fatal. 36, 40 B infl uenzae were found in pure culture in at least one lobe in 43% of 37 necropsies; and pneumococci (65%), Streptococcus haemolyticus (7·5%), H infl uenzae (2·5%), and Staphylococcus aureus (1·3%) were recovered from 80 post-mortem cultures of heart blood. Medical offi cers felt that "considerable importance must be placed on the secondary invaders, the pneumococcus and in this hospital rarely the S hemolyticus, which are found in such a large percentage of cases examined, both during life and post mortem". 40 At Camp Logan, TX, from September 13 to October 8, there were 2487 hospitalisations for infl uenza; of them, approximately 17% developed pneumonias, of which approximately 4% were fatal. 36, 41 Of 302 sputum cultures of infl uenza and pneumonia patients, the predominant organism by far was pneumococcus. In nine postmortem examinations, pneumococci were the most frequently recovered organisms from lungs (44%), pleural cavities (67%), and heart blood (33%). 41 At Fort Riley, KS, between September 15 and the end of October, approximately one-quarter of the camp's population were aff ected by infl uenza; of them, approximately 17% developed pneumonias, of which approximately 36% were fatal. 36, 42 Medical offi cers reported that there were "no deaths during the epidemic except from pneumonia or its complications". 42 They concluded that "infl uenza bacillus or some unknown Historical Review virus merely served to predispose to pulmonary infection with organisms commonly known to produce pneumonia such as the pneumococcus and Streptococcus hemolyticus". 42 At Camp Jackson, SC, from approximately September 15 through mid-October, about one-fi fth of the camp's population were hospitalised with infl uenza; of them, about 17% developed pneumonias, of which approximately 31% were fatal. 36, 43 The most frequent isolate from sputum of pneumonia cases was Streptococcus pneumoniae; remarkably, nearly half of 312 post-mortem lung cultures revealed S aureus. Medical offi cers postulated that "the primary epidemic infection depresses the hematopoietic system to such a degree that it is unable to react normally to infection by bacteria". 43 At Camp Pike, AR, between September 20 and October 19, nearly one-quarter of the camp's population were treated for infl uenza; of them, approximately 12% developed pneumonias of which approximately 31% were fatal (fi gure 4). 36, 38 From clinical and autopsy fi ndings, Opie and colleagues 38 concluded that "pneumonia follows infl uenza because the disease renders the air passages susceptible to the invasion of a variety of micro-organisms, among which those commonly found are pneumococci, hemolytic streptococci, and staphylococci".

At Camp Dix, NJ, from mid-September through October, approximately one-fi fth of the troops were hospitalised for infl uenza; of them, approximately 18% developed pneumonias, of which approximately 50% were fatal. 36, 39 There were 20 cases of frank empyema and several lesions of pleura that would have resulted in empyema had the patients lived longer. Medical offi cers noted that "a large variety of organisms has been encountered in cultures and smears from the lung substance, from the bronchial mucous membrane, and from the sputum. Streptococci and pneumococci have been most frequently found". 39 At Camp Custer, MI, from late September through October, more than one-quarter of all troops were diagnosed with infl uenza; of them, approximately 21% developed pneumonias, of which approximately 28% were fatal. 36, 44 Of 740 cultures of sputum of pneumonia patients, 26% and 17% were positive for pneumococcus and haemolytic streptococcus, respectively. Post-mortem cultures of lung and heart blood specimens from 280 fatal pneumonia cases revealed that 28% of each were positive for pneumococcus, and 27% and 22%, respectively, were positive for haemolytic streptococcus. The authors noted that "without exception the deaths from this respiratory epidemic have been due to secondary pneumonia. In no instance has a case come to necropsy in which death occurred from infl uenza infection alone". 44 They concluded that bacteria recovered from pneumonic lungs were "secondary invaders, the fi eld being prepared by the lowering of resistance incident to a preceding disease, which in this epidemic was in most instances infl uenza". 44 At Camp Grant, IL, in approximately 1 month following September 21, about one-quarter of all troops were hospitalised for infl uenza; of them, approximately 22% developed pneumonias, of which approximately 46% were fatal (fi gure 4). 36, 45 Post-mortem cultures of exudates of consolidated lungs and heart's blood often revealed "purely fi ne green colonies containing gram positive, lancet shaped diplococci". 45 45 of 90 blood cultures of living patients revealed "a gram-positive diplococcus in pure strain without exception … all the morphologic and cultural charac teristics of a pneumococcus". 45 The authors concluded that "the epidemic of bronchopneumonia at Camp Grant is due to infection by a virulent strain of pneumococcus. The virulence of this organism exceeds greatly that of strains usually identifi ed in pneumonia". 45 At Camp Fremont, CA, in 6 weeks beginning October 8, there were 2418 hospitalisations for respiratory diseases; of them, approximately 17% developed pneumonias, of which approximately 36% were fatal. 36, 46 Cultures of nasopharynges or sputum of 158 pneumonia cases revealed B infl uenzae (38%), pneumococci (41%), and streptococci (29%). Post-mortem cultures of lungs in 20 lobar pneumonia cases revealed pneumococci (45%) and staphylococci (10%). Post-mortem examinations of bronchopneumonia cases revealed that "pus streamed from their tracheas when the lungs were removed. Small areas … of consolidated tissue were scattered uniformly throughout the lungs … a drop of pus could be expressed from the center of each." 46 The authors concluded that the primary infection caused lower resistance to certain secondary organisms; secondary lobar pneumonias were due to pneumococci and streptococci-pneumococci predominating; and bronchopneumonias were due to B infl uenzae. 46 In his review of clinical reports from 72 US Army hospitals throughout the USA, Conner 47 noted "the multiplicity of types of the infecting micro-organisms" and emphasised the "close relation which exists between the purely clinical aspects of the disease and the nature of the dominating organism in the lung. This relation seems to be especially important in the case of Streptococcus hemolyticus and of Staphylococcus aureus."

In his review of the eff ects of the pandemic at 37 large army camps between September 12 and October 31, Soper 36 estimated that 22% of all soldiers were diagnosed with infl uenza; of them, approximately 17% developed pneumonias, of which approximately 34% were fatal. Across all camps, the mean infl uenza attack rate was 23% (median 22%; quartiles 15-28%), the mean percentage of infl uenza cases that developed pneumonias was 16% (median 17%; quartiles 10-20%), and the mean percent of pneumonia cases that were fatal was 34% (median 33%; quartiles 22-41%). 36 Finally, in his review of the pandemic's eff ects on the American Expeditionary Forces in Europe, MacNeal 32 reported that "infl uenza bacilli, pneumococci of various types, hemolytic and non-hemolytic streptococci have occurred most frequently in the infi ltrated lungs … In many cases, two or more of these organisms were isolated from the same tissue." He concluded that "the disease has been essentially due to an invasion of the respiratory tract by infl uenza bacilli, followed by and associated with other pharyngeal organisms, and the fatal outcome, in most instances, has been brought about particularly by these secondary invaders, in some instances streptococci, in others pneumococci".

Descriptions of the pandemic's eff ects at US Navy installations were similar. For example, in the annual report of the Surgeon General of the US Navy for 1918, 48 the epidemic that aff ected the submarine base in San Pedro, CA, was described as "infl uenza complicated by streptococcus infection". In the same report, Borden and Leopold noted that "no case of pandemic pneumonia … failed to present the clinical signs of infl uenza well before any pneumonic symptoms were recognized, thus demonstrating a distinct relationship between the causative factor in the uncomplicated infl uenza and the post infl uenzal pneumonia".

During the epidemic at the Puget Sound Navy Yard, WA, in the fall of 1918, streptococci were recovered from 44% of 52 blood cultures of living cases. During 20 postmortem examinations, haemolytic streptococci were recovered from heart blood (85%), the lungs (70%), and the pericardium (60%). The authors concluded that at the Puget Sound Navy Yard, the disease called infl uenza was due to an organism that "should be classed as a hemolytic streptococcus". 49 At the US Naval Hospital at League Island, PA, during 1 week in mid-September, 600 patients were hospitalised with infl uenza; of them, approximately 28% developed pneumonias, of which 29% were fatal. Pneumococci were recovered from "the majority of cultures" of sputum, whereas B infl uenzae, streptococci, and staphylococci were recovered less often. Devers and colleagues 50 noted that "the presence of these organisms, especially the pneumococcus and streptococcus … is interesting in view of the very common complication of pneumonia".

At the First Naval District in Massachusetts, Rapaport 51 found that 54·5% of 295 convalescent infl uenzal pneumonia patients (versus 9·6% of 300 controls) had antibodies to B infl uenzae.

In the US Naval Medical Bulletin, Hare 52 reported that "in the vast majority of cases, the illness was not the result of infection by one pathogenic organism, but was a multiple infection in which one of several organisms was the chief agent or in which all were approximately equally responsible … the physical signs of disease, the symptoms, and the lesions recognized at autopsy were much more those of the Streptococcus hemolyticus than of any of the other associated organisms".

In summary, medical offi cers at military installations throughout the USA and Europe were consistent in their Historical Review impressions that primary infections with infl uenza increased susceptibility to bacterial respiratory pathogens, that secondary bacterial infections were common and were clinically expressed with unusual virulence, and that bacteria accounted for many, if not most, of the pneumonias and deaths. By contrast with many recent reports, 6, 7, 13 there were few, if any, contemporaneous descriptions of epidemics at military camps in 1918 that did not emphasise the importance of secondary bacterial infections.

In this regard, it is worth noting Conner's mention 47 of a "clinical type which, in a few camps, notably at Camp Sherman, was suffi ciently common to form an outstanding feature of the epidemic. The patients … showed extreme cyanosis, high fever, and intense air hunger, and died in from 24 to 48 hours. The chest was fi lled with coarse, bubbling rales, and pinkish, frothy serous fl uid often poured from the mouth and nostrils. The postmortem fi ndings in the lungs were those of intense congestion and edema, without actual pneumonia. Friedlander compares the clinical picture to that seen after severe exposure to chlorin gas." Such cases have become the focus of most current pandemic preparedness planning.

The pandemic's eff ects in civilian populations were generally similar to those in the military. For example, in December 1918, a house-to-house survey of 112 958 people in eight locations throughout the USA found that approximately 28% were aff ected by infl uenza and approximately 30% of associated pneumonias were fatal. 53 Between September 23 and October 29, in Chicago, more than 2000 patients were hospitalised for infl uenza at Cook County hospital-nearly one-third of them died. Most fatalities were among young adults between 25 and 35 years old. Pneumococci were isolated from 70% of sputum, 74% of throat, and 73% of 15 lung puncture specimens of living patients. Pneumococci were also isolated from 75% of lung cultures post-mortem. Nuzum and colleagues 54 concluded that "the high percentage of pneumococci obtained during life and at necropsy and predominating in the sputum, tracheal mucosa, and lung tissue both early and late in the course of the disease suggest that this organism is at least the most important secondary invader and is responsible for many of the rapidly fatal pneumonias".

Reports of associations between infl uenza and bacterial pneumonias of unusual virulence are not limited to the 1918 pandemic. For example, Burgess and Gormley 55 reported three cases of staphylococcal pneumonia that rapidly progressed to death during an infl uenza epidemic in January 1929. S aureus was recovered from the sputum of all three patients (and from the lungs of the only case examined post-mortem). 55 In December through January 1941, during an epidemic of infl uenza A in Boston, Finland and colleagues 56 reported 66 cases (32% fatal) of S aureus pneumonia.

During the same period, 22 patients were admitted to the same hospital with pneumococcal lobar pneumoniasall had onsets of pneumonia 1-10 days after symptoms of infl uenza. 56 During the same epidemic, Pearson and colleagues 57 found high titres of antibodies against infl uenza A virus (suggestive of recent infection) in 34 of 82 (41%) patients hospitalised with pneumococcal pneumonia and seven of nine (78%) patients hospitalised with staphylococcal pneumonia. Of patients who had paired serum samples, fi ve of 21 (24%) patients with pneumococcal pneumonia and all six with staphylococcal pneumonia had changes in titres indicative of recent infl uenza A infection. The authors noted the possibility that "in certain of these patients, at least, infl uenza represented a factor predisposing to the pneumonia". 57 During the same period, Michael 58 described fi ve staphylococcal pneumonia cases with fulminating clinical courses. In three of the cases, serological tests confi rmed that infl uenza A infection immediately preceded the pneumonia. Michael concluded that "S aureus pneumonia often occurs as a complication of infl uenza" and suggested that "chemotherapy might have little eff ect unless given prophylactically during infl uenzal epidemics".

In 1944, Dingle and colleagues 59 observed that "outbreaks of pneumonia in camps and institutions were usually preceded by waves of infl uenza-like infection". They postulated that infl uenza infections increased susceptibility to pneumococcal infections and/or enhanced the transmission of pneumococci in aff ected populations. To clarify the relation, they studied acute and convalescent sera of patients involved in a pneumonia outbreak in a small town in New York and found that infl uenza B pre-existed in the community. They concluded that some localised outbreaks of acute bacterial pneumonia were secondary manifestations of infl uenza epidemics and that an increased prevalence of pneumonia may be a useful clue to identifying and studying outbreaks of infl uenza.

During the 1957 Asian infl uenza pandemic in the USA, there were substantial increases in hospitalisations and deaths from respiratory illnesses in general and pneumonias in particular. A case series analysis revealed that the most frequently isolated bacteria from patients with post-infl uenzal pneumonias were S pneumoniae, H infl uenzae, and S aureus. 60 During the same period in the Netherlands, Hers and colleagues 61 studied 158 fatal cases of Asian infl uenza. S aureus and pneumococci were recovered from 59% and 15% of lung cultures, respectively. Approximately two-thirds (n=56) of the pneumonia-related deaths attributable to S aureus were in individuals between 6 and 40 years old. "Pure" infl uenza pneumonia was considered the cause of death in 20% of the fatal cases. 61 During the 1968 Hong Kong infl uenza pandemic in the USA, there was an estimated threefold increase in the incidence of staphylococcal pneumonias and a strong Historical Review correlation between staphylococcal pneumonia risk and prior infl uenza infection. Much of the excess mortality during the pandemic was attributed to the increased incidence of bacterial pneumonias. 62 In Rochester, MN, infl uenza A Hong Kong/68 virus was isolated from 127 patients; of these, 16% developed pneumonias, of which 40% were fatal. S aureus or Pseudomonas aeruginosa was recovered from six of eight fatal cases. 63 In 1996, Kim and colleagues 64 reported results of a community-wide surveillance of respiratory diseases in Houston. Among adults, they found signifi cant correlations between the occurrence of pneumococcal disease and the isolation of infl uenza, respiratory syncytial virus, and other viruses (p<0·001). Recently, in a controlled trial among South African children, Madhi and colleagues 65 found that pneumococcal vaccine prevented 31% of all pneumonias associated with a variety of respiratory viruses. They concluded that pneumococci were important in the pathogenesis of viral pneumonias, and that viruses were important in the pathogenesis of bacterial pneumonias.

In summary, over decades in various populations and settings worldwide, there have been strong and consistent relations between infl uenza and secondary pneumonias, particularly due to pneumococcus, S aureus, S pyogenes, and H infl uenzae.

Interactions between infl uenza and Neisseria meningitidis have also been well documented. For example, during mobilisation for World War 1 (which included the infl uenza pandemic period), there were historically high numbers and rates of meningococcal disease in US civilian and military populations. In the US Army, recruits in initial training camps and soldiers who had recently disembarked from troop ships were at highest risk. 66 In October 1918, the weekly bulletin of the American Expeditionary Forces in Europe documented a sharp rise in reports of cerebrospinal meningitis approximately 1 week after sharp increases in reports of infl uenza and pneumonia. The bulletin noted that "it has been a usual observation that when infections of the upper respiratory tract prevail, the incidence of meningitis in the community increases soon after and this rule prevails at present". 67 Since 1918, there have been numerous reports of clusters of meningococcal disease that occurred during or shortly after outbreaks of infl uenza or infl uenza-like illnesses. For example, in 1972, Young and colleagues 68 described 11 cases (three fatal) of serogroup B meningococcal meningitis among 55 residents of a mental institution during a community outbreak of infl uenza A. Both nasopharyngeal carriage and systemic infections with N meningitidis were signifi cantly associated with serological evidence of recent infl uenza infections (p=0·054 and p=0·029, respectively). In 1976, Mackowiak and colleagues 69 reported four cases of meningococcaemia that occurred simultaneously in a family of six. The family members were aff ected by infl uenza-like illnesses before the outbreak of meningococcemia. 69 In 1986, Schubiger and colleagues 70 reported an outbreak of group B meningococcal disease among boarding school students. Each aff ected student had serological evidence of concomitant infl uenza infection. The authors concluded that "the viral infection made way for the outbreak of the meningococcal disease and for the high rate of secondary meningococcal infection". In February 1986, within 2 days, fi ve children who rode the same school bus developed group C meningococcal disease. All of the aff ected children had recent infl uenza-like illnesses, and cases had higher titres of antibodies to infl uenza B than non-aff ected students on the same bus. 71 In January 1996, ten cases of group C meningococcal disease occurred within 6 days among 1034 air force recruits in Greece. The peak of the outbreak was approximately 3 days after the peak of a large outbreak of infl uenza B among the recruits. 72 Relations between infl uenza and meningococcal disease incidence have also been documented at population levels. For example, in November-December 1989, during an infl uenza outbreak in the UK, Cartwright and colleagues 73 noted a striking increase in the number of meningococcal strains submitted to the national reference laboratory. Sera from 28% of 43 meningococcal disease cases (compared with 9% of 67 other cases) had antibodies to the epidemic strain of infl uenza A. During the same period, rates of meningococcal disease were increased and case fatality rates were exceptionally high throughout southwest England. 73 In 1957 and 1976, in England and Wales, approximately 2 weeks separated large outbreaks of infl uenza A from increases in reports of meningococcal meningitis. 74 From 1985 through 1990 in France, the incidence of meningococcal disease during a given week correlated with rates of infl uenzalike syndrome during the preceding (but not the following) 5 weeks. Also, there were signifi cant spatiotemporal associations between the spread of infl uenza-like syndrome throughout the country and subsequent increases in meningococcal disease rates (p<0·05). Finally, meningococcal cases had more severe clinical outcomes during and up to 2 months following outbreaks of infl uenza-like syndrome compared with other times. 75 In summary, in many locations, populations, and settings over many decades, important relations have been documented between infl uenza incidence and subsequent rates and severities of meningococcal disease.

Infl uenza has been associated with increased risk of staphylococcal toxic shock syndrome. In 1987, Sperber and Francis 76 reported a fatal case of staphylococcal toxic shock syndrome in an 18-year-old boy with bilateral

S aureus pneumonitis and ulcerative tracheobronchitis following a 3-day history of infl uenza-like illness. The fatal illness occurred during an outbreak of infl uenza B in the boy's community. In the same year, Macdonald and colleagues 77 reported nine cases of staphylococcal toxic shock syndrome during infl uenza outbreaks (primarily type B) in Minnesota during one winter season. Eight of the cases occurred within 1-4 weeks of peaks of infl uenza activity in the same communities; and seven of the cases had S aureus isolates that produced toxic shock syndrome toxin 1 or enterotoxin B. The authors surmised that toxic shock syndrome can occur in patients "who have a toxigenic strain of S. aureus in their respiratory tracts during infl uenzalike illness".

In 1960, Eichenwald and colleagues 78 reported that newborns whose noses were colonised with S aureus dispersed large numbers of bacteria ("cloud baby") and were highly contagious ("superspreaders") when coinfected with common respiratory viruses. In 1972, Gwaltney and colleagues 79 documented 25 transmissions of S pneumoniae between family members. 14 (56%) of these donors had symptomatic upper respiratory infections around times of transmission. In 1996, Sherertz and colleagues 80 documented the "cloud" phenomenon in adults. Investigation of an outbreak of MRSA in a surgical intensive care unit revealed that one of 64 workers in the unit was a nasal carrier of the outbreak strain. The carrier had a minor upper respiratory illness during the outbreak; and after experimental inoculation with a rhinovirus, he had a 40-fold increase in dispersion of the carriage strain of S aureus. In 2004, Bassetti and colleagues 81 reported that university students with persistent nasal carriage of S aureus increased dispersal by twofold (with peak increases up to 34-fold) after experimental infection with a rhinovirus.

During outbreaks of SARS coronavirus in 2003, some infected individuals transmitted the virus much more effi ciently than others. The few superspreaders were epidemiologically important. 82 For example, during the outbreak in Hong Kong, the index case-a superspreader-presented with a runny nose, a relatively uncommon clinical manifestation of SARS coronavirus. Bassetti and colleagues 83 postulated that the transmissibility of SARS coronavirus may be substantially enhanced by coinfections with other respiratory viruses.

Studies in animals and natural experiments in human beings suggest that infl uenza spreads from person-toperson through aerosols of droplet nuclei ("virus clouds"). 84 It is likely (but not documented) that infl uenza increases the transmissibility of bacteria and/or viruses that colonise or coinfect the respiratory tracts of coinfected hosts; and that infections with other viruses and/or bacteria (eg, pertussis) may increase the aerosolisationand hence the transmissibility-of coinfecting infl uenza.

The epidemiology and clinical expressions of respiratory infectious diseases depend on characteristics of and interactions among co-circulating infectious agents, infected and at-risk human hosts, and the environments in which they interact. Narrow epidemiological models that do not account for agent, host, and environmental interactions unnecessarily restrict opportunities for prevention, treatment, and control.

During the 1918 pandemic (and subsequent pandemics and epidemics) of infl uenza, a large proportion of deaths were likely attributable to bacterial respiratory infections. Unlike 1918, however, we now have safe and eff ective vaccines against the most prevalent strains of S pneumoniae and N meningitidis. In addition, we now have antibiotics that have been used safely and eff ectively to prevent severe bacterial respiratory illnesses (eg, invasive group A streptococcal disease, acute rheumatic fever, pneumonia, empyema, cerebrospinal meningitis) in individuals, groups, and settings considered to be at high risk. [85] [86] [87] [88] [89] [90] [91] [92] [93] During future pandemics, virtually all members of general populations will be at high risk of infl uenza infection, and all those infected (regardless of age or prior health) will be at exceptionally high risk of fulminant clinical expressions of secondary bacterial respiratory infections.

The prevention of secondary bacterial infections with currently available vaccines and antibiotics would seem to be "low hanging fruit" for pandemic preparedness. Fock and colleagues 94 have noted that "even if suitable infl uenza vaccines and virostatic agents are not suffi ciently available at the start of a pandemic, it is still possible to at least prevent an outbreak of two of the most feared secondary infections that accompany infl uenza: pneumococcal pneumonia or meningitis and illnesses resulting from Haemophilus infl uenzae".

Yet, there is relatively little knowledge regarding-and little research specifi cally focused on defi ning-the public health and clinical implications of interactions between infl uenza viruses in general (or H5N1 specifi cally) and bacterial respiratory pathogens. Vaccines and antibiotics may be useful adjuncts to current infl uenza epidemic countermeasures. Research on prevention and treatment measures specifi cally related to bacterial infections that occur secondary to infl uenza should be a high priority.

I declare that I have no confl icts of interest.

Updating the accounts: global mortality of the 1918-1920 "Spanish" infl uenza pandemic

Infl uenza A pandemics of the 20th century with special reference to 1918: virology, pathology and epidemiology

Severe acute respiratory syndrome

WHO Global Infl uenza Program Surveillance Network. Evolution of H5N1 avian infl uenza viruses in Asia

Re-emergence of fatal human infl uenza A subtype H5N1 disease

Preparing for the next pandemic

Infl uenza A (H5N1): will it be the next pandemic infl uenza? Are we ready?

Writing Committee of the WHO Consultation on Human Infl uenza A/H5. Current concepts: avian infl uenza A (H5N1) infection in humans

Transmissibility of 1918 pandemic infl uenza

Characterization of the 1918 infl uenza virus polymerase genes

Enhanced virulence of infl uenza A viruses with the haemagglutinin of the 1918 pandemic virus

Characterization of the reconstructed 1918 Spanish infl uenza pandemic virus

Fever of war: the infl uenza epidemic in the US army during World War 1

Avian fl u: isolation of drug-resistant H5N1 virus

Interaction between viral and bacterial infections in the respiratory tract

Viral infections predisposing to bacterial infections

Infl uenza virus neuraminidase contributes to secondary bacterial pneumonia

Bacterial-viral interrelations in respiratory infections of children

Host defense impairments that may lead to respiratory infections

Immunosuppression in viral infections

Viral-bacterial synergistic interaction in respiratory disease

Lethal synergism induced in mice by infl uenza type A virus and type Ia group B streptococci

Eff ect of infl uenza viral infection on the ingestion and killing of bacteria by alveolar macrophages

Mechanisms of bacterial superinfections in viral pneumonias

Pulmonary antibacterial defenses during mild and severe infl uenza virus infection

Immunoepidemiology of meningococcal disease in military recruits. II. Blocking of serum bactericidal activity by circulating IgA early in the course of invasive disease

Epidemic meningococcal disease: synthesis of a hypothetical immunoepidemiologic model

Rapid identifi cation and strain-typing of respiratory pathogens for epidemic surveillance

A cloud adult: the Staphylococcus aureus-virus interaction revisited

Purulent bronchitis: a study of cases occurring amongst the British troops at a base in France

Purulent bronchitis: its infl uenza and pneumococcal bacteriology

A further investigation into infl uenza pneumococcal and infl uenza-streptococcal septicaemia: epidemic infl uenzal "pneumonia" of highly fatal type and its relation to "purulent bronchitis

The infl uenza epidemic of 1918 in the American expeditionary forces in France and England

World War I may have allowed the emergence of "Spanish" infl uenza

Pneumonia at Camp Funston-report to the Surgeon General

The pandemic in the Army camps

In: Siler JF, ed. The Medical Department of the United States Army in the World War. Volume IX: communicable and other diseases

Pneumonia following infl uenza (at Camp Pike, Ark.)

The infl uenza epidemic at Camp Dix

A bacteriologic study of the infl uenza epidemic at Camp Devens, Mass

The epidemic of pneumonia following infl uenza at Camp Logan, Texas

Infl uenza and infl uenzal pneumonia at Fort Riley

Staphylococcus aureus pneumonia

A recent epidemic of acute respiratory infection at Camp Custer, Mich

Epidemic of bronchopneumonia at Camp Grant, Ill

Pandemic infl uenza and secondary pneumonia at Camp Fremont, Calif

The symptomatology and complications of infl uenza

Annual report of the Surgeon General, US Navy for the fi scal year 1918

Infl uenza as seen at the Puget Sound navy yard

Infl uenza in the United States naval hospital

The complement fi xation test in infl uenzal pneumonia. Studies with serum from convalescent patients, the infl uenza bacillus being used as antigen

Section II (a). The clinical and therapeutic standpoint

The epidemiology of infl uenza

Pandemic infl uenza and pneumonia in a large civilian hospital

Pneumonia in relation to an epidemic of "mild" infl uenza, with report of three fulminating cases apparently due to Staphylococcus aureus

Staphylococcic pneumonia during an epidemic of infl uenza

A study of infl uenza in Boston during the winter of 1940-1941

Staphylococcal aureus pneumonia with special reference to its occurrence as a complication of infl uenza

The relation between epidemics of acute bacterial pneumonia and infl uenza

Pulmonary infections complicating Asian infl uenza

Bacteriology and histopathology of the respiratory tract and lungs in fatal Asian infl uenza

Bacterial pneumonia during the Hong Kong infl uenza epidemic of 1968-1969

Kong infl uenza: clinical, microbiologic, and pathologic features in 127 cases

Association of invasive pneumococcal disease with season, atmospheric conditions, air pollution, and the isolation of respiratory viruses

Vaccine Trialist Group. A role for Streptococcus pneumoniae in virus-associated pneumonia

Evolution of meningococcal disease epidemiology in the U.S. Army. In: Vedros NA, ed. Evolution of meningococcal disease

American expeditionary forces

A simultaneous outbreak of meningococcal and infl uenza infections

Acute meningococcemia without meningitis in association with infl uenza-like illness

Meningococcal epidemic in a boarding school: a rifampicin-resistant secondary case while under chemoprophylaxis

A cluster of meningococcal disease on a school bus following epidemic infl uenza

Outbreak of meningococcal disease after an infl uenza B epidemic at a Hellenic Air Force recruit training center

Infl uenza A and meningococcal disease

Infl uenza A and meningococcal disease

Meningococcal disease and infl uenza-like syndrome: a new approach to an old question

Toxic shock syndrome during an infl uenza outbreak

Toxic shock syndrome. A newly recognized complication of infl uenza and infl uenzalike illness

The "cloud baby": an example of bacterial-viral interaction

Spread of Streptococcus pneumoniae in families. Relation of transfer of S. pneumoniae to incidence of colds and serum antibody

A cloud adult: the Staphylococcus aureus-virus interaction revisited

Dispersal of Staphylococcus aureus into the air associated with a rhinovirus infection

Cluster of SARS among medical students exposed to single patient

Are SARS superspreaders cloud adults?

Transmission of viral respiratory infections in the home

Chemoprophylaxis for bacterial infections: principles of and application to meningococcal infections

Epidemiology and control of acute respiratory diseases with emphasis on group A beta-hemolytic streptococcus: a decade of U.S. Army experience

Randomized, placebocontrolled clinical trial of oral azithromycin prophylaxis against respiratory infections in a high-risk, young adult population

An outbreak of pneumococcal pneumonia among military personnel at high risk: control by lowdose azithromycin postexposure chemoprophylaxis

Halting a pneumococcal pneumonia outbreak among United States Marine Corps trainees

Global strategies to prevent bacterial pneumonia in adults with HIV disease

House-to-house, community-wide chemoprophylaxis for meningococcal disease: an aggressive approach to disease prevention

Broad and persistent eff ects of benzathine penicillin G in the prevention of febrile, acute respiratory disease

Evaluation of the use of mass chemoprophylaxis during a school outbreak of enzyme type 5 serogroup B meningococcal disease

Conceptional considerations for a German infl uenza pandemic preparedness plan

I thank Gregory C Gray, John D Grabenstein, and Charles H Hoke for providing contemporaneous reports regarding the 1918-19 pandemic and/or comments regarding an earlier version of this paper.