Bacterial Pneumonia and Pandemic Influenza Planning

Prevention and treatment of secondary bacterial complications are important but neglected areas of planning.

site, such as blood or pleural fl uid. This defi nition therefore underestimates pneumococcal pneumonia where isolation of the organism is not possible (13). Notwithstanding, a number of studies have documented the temporal association between infl uenza and invasive pneumococcal disease, which suggests synergism. Grabowska et al. (14) recently used 2 epidemiologic methods based on Swedish surveillance data to estimate the excess cases of invasive pneumococcal pneumonia associated with seasonal infl uenza at 12%-30%.
HIV-infected children have a 40× greater risk than HIV-noninfected children for invasive pneumococcal disease and account for most cases of invasive pneumococcal disease in certain sub-Saharan African countries (13,15). HIV-infected children and adults would likely be more severely affected by an infl uenza pandemic.

Staphylococcus aureus (Methicillin Sensitive and Methicillin Resistant)
A retrospective study of infl uenza-related childhood deaths in the United States in the 2003-04 season found S. aureus to be the most common bacterial agent, accounting for 46% of isolates, >50% of which were methicillin-resistant strains (5). Surveillance for severe infl uenza-related S. aureus community-acquired pneumonia in the United States during the 2003-04 season recorded 17 cases (88% methicillin-resistant S. aureus [MRSA]) and 5 deaths (4 with MRSA) and a median age of 21 years (16); laboratory evidence of infl uenza infection was available for ≈75%. More recently, 10 cases of severe community-acquired MRSA pneumonia in children (6 of whom died) from 2 southern states in a 2-month period were reported (17). For 30% of those patients, MRSA was recovered from sputum only, and 4 had a documented recent history of MRSA skin infection in themselves or in a close contact. Preceding staphylococcal skin disease in persons with staphylococcal pneumonia was described by Goslings et al. (18) during the 1957-58 pandemic. In the context of emerging community-acquired MRSA skin infection in persons without traditional risk factors, this association has substantial implications for possible emergence of MRSA pneumonia in a future pandemic (19).

Other Pathogens
A recent study from New Zealand (7) that aimed to characterize viral causes of community-acquired pneumonia reported viral-bacterial co-infection in 45 (15%) of 304 hospitalized patients. S. pneumoniae (67%) and Haemophilus infl uenzae (11%) were the 2 pathogens most commonly associated with infl uenza A infection; atypical microbes (Chlamydia pneumoniae, Mycoplasma pneumoniae, and Legionella pneumophila) were also well represented (22%). These fi gures are generally consistent with other published data; group A streptococci are a rare but serious cause of community-acquired pneumonia (20) and have been associated with fatal cases of infl uenza (5).

Stockpiling and Strategic Use of Antimicrobial Drugs
In most modern healthcare systems, which increasingly emphasize just-in-time supply chains, shortages of antimicrobial drugs may occur rapidly unless more are stockpiled. These shortages would limit the treatment of secondary bacterial infections in the middle and the later stages of a pandemic. For this reason a range of antimicrobial drug options have been suggested, taking into account the likely limitations of availability in diagnostics for communityacquired pneumonia and the fact that, because of the sheer number of patients, therapy is likely to be empirical. Clinical management guidelines for pandemic infl uenza have recommended amoxicillin + clavulanate or doxycycline (21); third-generation cephalosporins or respiratory fl uoroquinolones (22); and second-generation cephalosporins, macrolides, doxycycline, or co-trimoxazole (23) as fi rstline empirical therapies for community-acquired pneumonia associated with pandemic infl uenza. Dependent on the extent of any stockpile, shortages of these preferred agents might occur fi rst during a pandemic.
In the United States, the emergence of communityacquired MRSA has prompted revision to include vancomycin and other agents as empirical therapy for severe cases (21,22). The demand created by empirical use of vancomycin in such cases, the limited number of alternative agents, and the limited global production capacity of this drug are likely to lead to its shortage. Other treatment possibilities include linezolid, quinopristin/dalfopristin, and tigecycline.
Fortunately, in the United Kingdom most MRSA isolates are sensitive to doxycycline (95% of respiratory isolates; Health Protection Agency [HPA], unpub. data) and rifampin (97%; HPA, unpub. data); fewer are sensitive to trimethoprim (72%). Less severe MRSA infections treated with these widely available and inexpensive drugs would be expected to respond. Rifampin and cotrimoxazole are widely produced in developing countries, where the prevalence of tuberculosis and HIV infection are high.

Real-time Surveillance of Pathogen Resistance
After a country has committed to acquiring a stockpile of antimicrobial drugs, several important practical and logistic issues arise. The fi rst is deciding on the range of antimicrobial drugs to be stockpiled. After the World Health Organization declares a global pandemic phase 5 alert, antimicrobial drug supplies will be quickly depleted as countries scour the global market to build up stocks. The choice of available agents may be limited by this stage; therefore, procuring in advance is sensible, although this involves predicting which bacterial agents will be of greatest importance. The UK HPA has developed a program of real-time surveillance of antimicrobial susceptibility for the 3 most likely infl uenza-related bacterial pneumonia pathogens: S. pneumoniae, H. infl uenzae, and S. aureus. Contemporaneous data are available for each pathogen, enabling recommendation of antimicrobial drugs on the basis of the proportion of respiratory tract isolates likely to be susceptible at a particular point in time. Such real-time data may be useful for guiding the evolution of pandemic antimicrobial drug treatment policy in order to optimize the use of scarce antimicrobial drugs by drawing on a range of different agents according to national stock availability at the time. The surveillance program may also provide early warning of likely clinical failures caused by emerging resistance.

Size, Storage, and Turnover of Stockpiles
Decisions about pandemic stockpiles, procurements, and size depend primarily on fi nancial considerations. Decision-makers must bear in the mind the need not only to purchase the initial stockpile but also to maintain it, perhaps for a sustained period. In most circumstances, stockpiles of vaccines for infl uenza virus subtype H5N1 and neuraminidase inhibitors are reserved exclusively for use during or immediately before a pandemic; they are not intended for day-to-day use on the same scale. In contrast, antimicrobial drugs are widely used every day. This difference means that antimicrobial drugs could act as buffer stock (conceptually similar to vendor-managed inventory) in most healthcare systems, rather than a true stockpile. Indeed, the word stockpile may be a misnomer in relation to increased stores of antimicrobial drugs because these drugs can be channeled into day-to-day use and replaced through fresh procurement. Thus, over time the amount, proportion, and range of these agents held can be slowly altered. These 2 mechanisms, ongoing interpandemic use and restocking, make such a stockpile far less vulnerable than antiviral drugs to expiration before use and far more responsive to changes in antimicrobial drug sensitivity detected between the date of procurement and the onset of the next pandemic.
Further considerations relate to storage. Whereas antiviral drugs and vaccines essentially need to be held in secure centralized storage (the latter within the cold chain) until eventual deployment, antimicrobial drugs can be held, at least in part, lower down the supply chain by wholesalers and community pharmacies or their equivalent.
Additionally, the proportion of pandemic infl uenza cases that will progress to bacterial complications needs to be estimated. The diffi culty in making such an estimate relates partly to the paucity of contemporary data that specifi cally describe the incidence of bacterial complications after infl uenza and partly to the fact that widespread use of neuraminidase inhibitors, rarely used for seasonal infl uenza, might reduce the development of antimicrobial drugrelated complications by 25%-40% (24,25). Data from the extensive reviews by Brundage and Soper suggest that in the 3 pandemics of the 20th century, bacterial pneumonia developed in 15%-20% of infl uenza patients (2,26); some estimates for seasonal infl uenza are far higher (27). It can be argued that in 1918 the primary viral infection was so virulent that it caused the premature demise of some patients who might otherwise have survived long enough for bacterial pneumonia to develop; i.e., the reported frequency of bacterial complications was spuriously low. Coupled with a population clinical attack rate that will most likely lie in the range of 25% to 50%, an antimicrobial drug stockpile is likely to be needed for a minimum of 10% of the population. This fi gure does not account for wastage, misdiagnosis (if, as is most likely, prescribing is based on clinical suspicion alone), or a higher rate of secondary bacterial complication than expected; it is also based on a strategy of treatment only.

Treatment and Prophylaxis Strategies
Alternative strategies include offering antimicrobial drug prophylaxis at the same time as antiviral treatment to patients with conditions that put them at high risk, such as chronic obstructive pulmonary disease. Antimicrobial drugs (or a prescription for them) could be issued to high-risk patients at the same time as antiviral treatment. The ability to start antimicrobial drug therapy with minimal delay and without the need for repeat consultation if antiviral drugs alone are not effective might be advantageous in an already overstretched health system. Both of these strategies incorporate further uncertainty because the high-risk groups in a pandemic are unknown and may not correspond to those currently recognizable for seasonal infl uenza; if anything, the high-risk groups are more likely than not to be larger in a pandemic. This might increase the requirement for antimicrobial drug stockpiling to 25% population coverage.
Each country should estimate its own needs. Countryspecifi c factors to take into account include treatment strategy (treatment alone or treatment and prophylaxis), health service confi guration, historical use of antimicrobial drugs, physician behavior, inappropriate prescribing linked to misdiagnosis, and the availability of antiviral drugs.

Quinolone Stockpiles
A large number of countries hold stockpiles of quinolones, in particular ciprofl oxacin, as a contingency against bioterrorist threats. In the United Kingdom, ciprofl oxacin is active against all H. infl uenzae isolates (≈100% of recent UK respiratory tract isolates susceptible; HPA, unpub. data), most methicillin-sensitive S. aureus isolates (≈82%), and atypical organisms. Therefore, if these bacterial pathogens were known or suspected to predominate in infl uenzarelated pneumonia associated with a future pandemic, the use of ciprofl oxacin might be justifi ed, and agents effective against MRSA would be reserved for severe cases and those with culture-confi rmed MRSA (99% of UK respiratory MRSA isolates, most of which are hospital acquired, are quinolone resistant; HPA, unpub. data).
However, ciprofl oxacin activity against S. pneumoniae (28) is only intermediate, and a signifi cant number of bacterial pneumonias complicating infl uenza may not respond to empirical treatment. This fact is well supported by evidence from mouse models; more modern "respiratory" fl uoroquinolones such as gatifl oxacin demonstrate good results (29) against S. pneumoniae, which was not always so for ciprofl oxacin. Therefore, in a pandemic empirical ciprofl oxacin use could be justifi ed only if all other more suitable antimicrobial drug supplies were exhausted.
Given ciprofl oxacin's weak activity against pneumococci, reserving its use in a pandemic to empirical treatment of persons previously vaccinated against pneumococcal infection, who would be at reduced risk for co-infection with this particular organism, would be reasonable. Theoretical support for this hypothesis comes from the United States, where use of a 7-valent conjugate vaccine since 2000 has resulted in declining invasive pneumococcal disease (30) and relatively infrequent infl uenza-related deaths caused by pneumococci in children (5). A strategic approach might involve the use of ciprofl oxacin in fully immunized persons.

Pneumococcal Vaccination Strategies
Including a vaccination strategy in pandemic planning would potentially reduce the amount of disease caused by secondary S. pneumoniae bacterial pneumonia. We have already described this pathogen's role in community-acquired pneumonia and infl uenza complications. The public health benefi t from vaccination could be substantial.
Pneumococcal polysaccharide vaccine (PPV) is currently recommended in many countries for persons >65 years of age and for high-risk groups of all ages. Few specifi c data exist on the effectiveness of PPV for reducing pneumococcal pneumonia-associated illness and death after infection with infl uenza A virus. Furthermore, in the context of pneumococcal disease not specifi cally associated with infl uenza, use of PPV has protected against invasive pneumococcal disease but not against pneumococcal pneumonia in the absence of bacteremia (31). Therefore, on the basis of current evidence, prior PPV administration could not reliably be used to identify persons who could receive empirical ciprofl oxacin therapy for bacterial pneumonia as a complication of infl uenza. It could, however, be used as a large-scale preventive measure against invasive pneumococcal disease in adults.
The protective effi cacy of a 9-valent pneumococcal conjugate vaccine (PncCV) against nonbacteremic pneumonia as well as invasive pneumococcal disease has been demonstrated in 37,107 children from South Africa among whom the prevalence of HIV infection was 6.5% (32).The vaccine also substantially reduced the incidence of fi rst episodes of invasive pneumococcal disease that were resistant to penicillin or trimethoprim-sulfamethoxazole.
PncCV may have more greatly reduced the incidence of pneumonia in children when a virus was isolated (33). This effect was more pronounced when infl uenza A was isolated; protective effi cacy was 41% (95% confi dence interval 13%-60%). The study provided indirect evidence of the frequency of pneumococcal superinfection of viral pneumonias in children in this setting. If similar results could be achieved through vaccination before an infl uenza pandemic, the benefi ts of preventing pneumococcal complications could be substantial. The introduction of conjugate vaccine in the United States in 2000 has led to a decline in invasive pneumococcal disease in not only children but also adults; reduction was 32% for those 20-39 years of age and 18% for those >65 years (30). Therefore, vaccination of children might be the most cost-effective policy. In September 2006, the United Kingdom started vaccinating children from the age of 2 months; early unpublished data (minutes from the Joint Committee on Vaccination and Immunisation meeting on February 14, 2007, available from www.advisorybodies.doh.gov.uk/jcvi/mins140207.htm) suggest that invasive pneumococcal disease in children <2 years of age is already reduced.
The use of PncCV in children and 23-valent PPV in adults as part of a pandemic strategy would be consistent with recommendations resulting from current published data. However, such use may still not allow for ciprofl oxacin stockpiles to be reliably targeted for specifi c populations, given the lack of protection against nonbacteremic pneumoccal pneumonia associated with PPV. If conjugate vaccine were used in all patients (although no convincing data exist to support effi cacy of conjugate vaccine in adults), ciprofl oxacin might be more reliably targeted at a group more likely to have a nonpneumococcal pneumonia. However, a conjugate vaccine is likely to be expensive and limited in serotype coverage, and approval for its use in adults will take time.

Discussion
Substantial laboratory and epidemiologic evidence shows that infl uenza A and bacterial pathogens often participate in the pathogenesis of pneumonia. Several issues need to be considered with regard to antimicrobial drug treatment for large numbers of patients who have secondary bacterial infection during a pandemic. Real-time antimicrobial drugresistance surveillance programs could be incorporated into preparedness frameworks; information from such networks could result in stockpiling of inexpensive, generically manufactured antimicrobial drugs. Vaccination against pneumococcal disease, particularly vaccination of HIV-infected persons, potentially will save lives in the short term as well as provide protection in the event of a pandemic.
Dr Gupta is an infectious diseases physician undergoing postgraduate specialist medical training. He is currently conducting research on HIV resistance in developing-world settings and has an interest in pandemic infl uenza preparedness.