Pathogenicity of Influenza Viruses

Influenza virus is a single-stranded RNA virus of the family Orthomyxoviridae It affects birds and mammals and includes three genuses: influenza virus A, B, and C, based upon their matrix proteins.^1,4 Influenza A virus is typically the most virulent, has pandemic potential, and leads to the most severe disease. Based upon the antibody response to two major antigenic proteins on the outside of virus, hemagglutinin (HA) and neuraminidase (NA), influenza A is subdivided into different serotypes including: H1N1 (responsible for Spanish flu in 1918, in addition to the 2009 flu pandemic); H2N2 (Asian flu of 1957); H3N2 (Hong Kong flu of 1968); H5N1 (the avian flu, often sited as the most recent pandemic threat), and a number of others currently less relevant to humans (H7N7, H1N2, H9N2, H7N2, H7N3, H10N7). The two other forms of influenza include B (which almost exclusively infects humans but is less common) and C (affecting humans, dogs, and pigs), which only rarely cause severe illness and epidemics in humans.⁵

A notable characteristic of influenza virus is the genomic structure consisting of eight separate single-strand segments, each encoding a single major protein to complete the synthesis of the mature virus. The RNA-based genome provides a high background mutation rate and gives the virus genetic plasticity. The multiple genome segments provide the substrate for reassortment of large sequences of RNA and permit hybrid viruses to form in hosts infected simultaneously by more than one virus strain. These events lead to whole-scale recombination of entirely novel hybrid viruses with new antigenic constituents (antigenic shift). As an example, the novel swine-origin influenza A/Mexico City/4/2009 (H1N1) outbreak strain was a quadruple-reassorted virus derived from gene segments originating from ducks, Eurasian swine, North American swine, and human-adapted influenza virus.⁶

Avian-adapted viruses can occasionally be transmitted to mammals, causing outbreaks in animals or giving rise to disease in human pandemics. The pig is an important “mixing vessel” host in shuttling avian influenza viruses to humans, as they can carry both avian and human influenza viruses.¹ Porcine mucous membranes express a mixture of sialic acid–coated glycopeptides linked in a favorable conformation to bind both avian and human-adapted viruses. This is vitally important in the biology of influenza viruses, as the initial event in influenza infection is interaction of the hemagglutinin receptor to binding sites on host epithelial tissues. Avian species express α2,3-linked sialic acid–galactose disaccharides on their epithelial surfaces, and avian-adapted influenza preferentially binds to this linkage pattern. Human upper respiratory airways primarily express α2,6-linked sialyl-galactose surface receptors, and seasonal influenza strains in humans bind readily only to α2,6 linkages. Pigs, in contrast, normally express both α2,3- and α2,6-linked disaccharides on their mucous membranes, facilitating the opportunity for dual infections with avian- and human-adapted viruses.^1,6,7

The lower airways and alveolar pneumocytes of humans actually express α2,3-linked sialylated glycopeptides, and viruses that bind efficiently to α2,3 linkages can cause severe pneumonia if deposited into the distal airways. Most seasonal influenza strains bind preferentially to α2,6-linked disaccharide hemagglutinin (HA) binding sites found in human upper airways. This usually leads to high transmission frequency by the airborne droplet nuclei deposited upon the upper airways, but a low risk of primary influenza pneumonia.⁸ The avian strain of H5N1 preferentially binds to α2,3 linkages and therefore is poorly transmissible from person to person, but it has the potential to cause severe pneumonia if delivered to the lower airways. Poultry workers in Asia in close proximity to infected livestock can occasionally receive enough viruses deposited into the distal airways to cause severe influenza pneumonia with a high mortality rate (60% to 70%).^9,10

One of the explanations for the severity of the 1918 pandemic of H1N1 influenza was its HA that could bind with high affinity to both α2,6- and 2,3-linked sialyl-galactose moieties.^11,12 The result of this unusual HA binding affinity was a highly transmissible virus with the capacity to replicate and cause severe disease in the lower airways. Disturbingly, the hemagglutinin of the 2009 outbreak strain of novel swine origin also bound with high affinity to both α2,6 and α2,3 linkages. Fortunately, influenza A Mexico City 4/2009 (H1N1) virus lacked the full complement of other known virulence factors of the influenza virus (Table 140-1), resulting an overall low case-fatality rate (<0.1 %). A further mitigating factor against mortality in older populations during the 2009 outbreak was the presence of already-existing memory cells with B-cell and T-cell epitope recognition sites in humans born before the early 1950s, induced by H1N1 viruses circulating in the first half of the 20th century.¹³

TABLE 140-1 Pathogenicity Traits and Virulence Factors of Influenza Viruses

H274Y, histidine substitution for tyrosine at amino acid at position 274; HA, hemagglutinin; M, matrix protein; NA, neuraminidase; NS-1, nonstructural protein; PB, polymerase basic; Pol, polymerase; S31N, serine substitution for asparagine at amino position 31.

Clinical Manifestations and Complications of Influenza

Classical seasonal influenza in adults is typified by a 4- to 5-day period of sudden-onset fever, chills, upper respiratory tract symptoms, headache, muscle pain, and weakness. Rhinitis is relatively uncommon and diarrhea is more common with influenza than with most rhinovirus upper respiratory tract infections. Severe complications and death can occur, especially in infants, the elderly, and individuals with chronic medical conditions. Among the most severe complications are primary influenza pneumonia and secondary bacterial infection leading to respiratory failure.^14,15 Influenza can also cause central nervous system, cardiac, skeletal muscle, kidney, and hepatic complications.^5,15 Underlying pulmonary disease is a frequent risk factor, occurring in 18% of patients, most commonly asthma (7%), followed by neurologic disease (12%), hematologic or oncologic (9.9%), and cardiac conditions (4.6%).¹⁶ However, approximately half of those hospitalized (rates ranging from 1-5/1000) for influenza are otherwise healthy.^14–16

In the absence of a pandemic, 11% to 19% of patients hospitalized with laboratory-confirmed influenza require treatment in the intensive care unit (ICU).¹⁵ The mean duration of mechanical ventilation is approximately 5 days; the sickest patients require treatment with advanced techniques for the treatment of hypoxemia, such as high-frequency oscillatory ventilation (HFOV), extracorporeal membrane oxygenation (ECMO), prone positioning, and nitric oxide. These patients have an attendant increase in length of stay, duration of ventilation, and mortality.^14,16,17

An estimated 50 to 100 million people died during the 1918 pandemic. Death followed from aggressive secondary bronchopneumonia, influenza-related lung disease with associated hypoxemia, and cardiac collapse.^18,19 During the 1918 pandemic, there was unexplained excess influenza mortality in persons 20 to 40 years of age. This mortality increase may have been due to limited native immunity and/or a vigorous immune response directed against the virus in healthy young persons.¹⁸ Today the high mortality rate observed in the 1918 pandemic would almost certainly be reduced because of the availability of ICUs, vaccines, antibacterial agents, and antiviral medications. However, the cost would be a dramatic increase in critical care admissions and length of stay, assuming that this surge capacity is available. Long-stay ICU patients have significantly higher critical care and hospital mortality rates compared to short-stay patients, occupy a disproportionate number of critical care bed-days,⁴ and consume even greater resources.⁸ Sophisticated ICU care is often unavailable in developing countries today, and the case-fatality rates in these countries will probably be regrettably similar to the 1918 pandemic.²⁰

Influenza A 2009 H1N1-Related Epidemiology and Clinical Manifestations

Since March 2009, influenza A 2009 H1N1 has spread from Mexico to virtually all countries of the world. By September 27, 2009, there were over 340,000 cases with 4100 deaths worldwide.^7,21 The World Health Organization issued the first phase 6 pandemic alert of the century, anticipating substantial influenza transmission and related disease. Over the period of June to September 2009, there were dramatic spikes in H1N1-related disease in Australia, New Zealand, and South America that breached the capacity for ICU care in some regions. In Australian provinces, approximately 5% of the population developed H1N1-related illness, 0.3% of infected patients were hospitalized, and 20% of hospitalized patients required ICU care.²² In the Northern Hemisphere, an early and severe influenza outbreak occurred that was blunted in part by widespread deployment of an effective inactivated monovalent influenza vaccine program.²³

The events that transpired in Canada were illustrative of the influenza situation in much of the Northern Hemisphere in 2009. Among 168 critically ill Canadian patients with influenza A 2009 H1N1, the mean age has been 32 years, with a possible predilection for more severe disease in women (67% of patients).²⁴ Pregnant women in particular suffered from a disproportionate high level of influenza disease severity.^25,26 Nosocomial transmission was the mechanism of acquisition in approximately 10% of patients. Hospital-acquired transmission to healthcare workers occurred early in the outbreak, but healthcare-related infection occurred at a low incidence rate once the pandemic was recognized and appropriate infection-control safeguards were instituted. One or more comorbidities were observed in nearly all patients, most commonly chronic lung disease such as asthma, chronic obstructive pulmonary disease, bronchopulmonary dysplasia (41%), obesity (33%, mean body mass index of 34.6 kg/m²), hypertension (24%), history of smoking (23%), and diabetes (21%). Similar clinical findings and predisposing illnesses were reported in other regions of the world during the 2009 outbreak.^21,22,27,28 Serious comorbid illness was observed in only 30% of patients. Notably, aboriginal Canadians have thus far been over-represented (26% of patients). A summary of clinical risk factors and comorbidities associated with severe influenza complications is found in Table 140-2.

TABLE 140-2 Prognostic Indicators and Risk Factors for Severe Influenza Complications

The most common specific symptoms with influenza A 2009 H1N1 have included fever and respiratory symptoms in greater than 90% of patients, and less commonly weakness and myalgias. Several severe clinical syndromes associated with influenza A 2009 H1N1 infection may be seen, including:

The typical clinical syndrome requiring ICU care among all age groups appeared to be a diffuse bilateral four-quadrant pneumonitis that was often rapidly progressive. This process accounted for over 80% of ICU admissions in Canada and elsewhere and often necessitated advanced ventilatory/oxygenation modalities including HFOV, inhaled nitric oxide, and/or ECMO therapy.^24,29,30

Patients who subsequently developed critical illness generally presented to the hospital within 4 days of symptom onset and required ICU admission within 1 day of hospital presentation for bilateral pulmonary infiltrates and hypoxic respiratory failure. The mean Acute Physiology and Chronic Health Evaluation (APACHE) II score was 20. Notable laboratory findings have included elevated creatine kinase levels and normal white blood cell counts. Concomitant presenting conditions included possible bacterial pneumonia (32.1%), hypotension requiring vasopressors (13.7%), and asthma or chronic obstructive pulmonary disease exacerbation (13.7%).

Over 80% of patients with H1N1-related acute lung injury (ALI) received mechanical ventilation; very few patients were successfully managed with noninvasive ventilation strategies alone. Oxygenation support included high concentrations of inspired oxygen (mean admission PaO₂/FIO₂ 147 mmHg), positive end-expiratory pressure (PEEP), frequent use of HFOV (12%), nitric oxide (14%), neuromuscular blockade (30%), prone ventilation (5%), and occasionally ECMO (7%). Medical therapies included neuraminidase inhibitors (90.5%), antibacterial agents (98.8%), and, despite uncertain efficacy, corticosteroids (50.6%).²⁴

Secondary bacterial pneumonia following ICU admission was found in 24% of cases, most commonly due to S. aureus and S. pneumoniae. The frequency of secondary bacterial infection was difficult to accurately determine owing to the widespread use of empirical antibacterial therapy in influenza patients with rapidly progressive respiratory failure. Overall mortality among critically ill patients at 90 days was 17.3% (similar to that reported from Australia).²² The median duration of ventilation was 12 days. The most common cause of death was severe acute respiratory distress syndrome (ARDS) and hypoxemia, complications thereof, secondary infection, sepsis, or multiorgan dysfunction syndrome. Characteristic radiographic changes of severe primary influenza pneumonia are shown in Figure 140-1, A and B.

Figure 140-1 A, Chest radiograph of a 70-year-old male with B-cell lymphoma and hypogammaglobulinemia, with primary influenza pneumonia at the time of his ICU admission. Note Port-a-Cath in right anterior chest wall and diffuse pulmonary infiltrates, most prominently seen in both lower lung fields. B, Chest radiograph of same patient 3 days later; note diffuse alveolar filling process associated with profound hypoxemia. The patient expired despite oseltamivir and ventilatory support, with severe hypotension and acute kidney injury.

Lung pathology in fatally infected patients who underwent autopsy examination revealed a diffuse alveolar filling process, often with early hyaline membrane formation that was sometimes accompanied by focal areas of hemorrhage. The alveolar lining was usually thickened, with evidence of lymphocytic infiltrates and early organization with fibrosis. A typical lung tissue section of a patient with fatal influenza pneumonia is seen in Figure 140-2. Lung tissue in deaths occurring early in the presentation of influenza pneumonia often revealed diffuse immunohistochemical evidence of viral infection and intraalveolar hemorrhage.

Figure 140-2 Lung pathology of fatal case of primary influenza pneumonia in previously healthy 20-year-old woman. Note diffuse alveolar filling, squamous metaplasia, lymphocytic infiltrates, focal hemorrhage, loss of ventilatable lung tissue.

(Figure courtesy David Horn, MD.)

In children, the median age of hospitalized patients was 5.0 years (range 1 month to 17 years); 54.4% were female, and the mean PRISM III score was 9.^14–1624 One or more chronic comorbid illnesses were observed in 70.2% of patients: lung disease (44%), neurologic diseases (19%), immune suppression or immunodeficiency (16%), history of prematurity (9%), and congenital heart disease (7%). Mechanical ventilation was used in 68% of children admitted to ICU, and the median duration of ventilation was 6 days (range 0-67).

Clinical and Laboratory Diagnosis

Significant difficulties with definitive virologic diagnosis existed in the early phase of the 2009 influenza outbreak that were partially rectified as the pandemic unfolded. Fever and upper respiratory symptoms were present in almost all patients who progressed to critical illness. However, shortness of breath, a symptom not typical of uncomplicated influenza virus infection, was likely suggestive of severe disease. Other clinical signs noted in patients with severe disease have included hemoptysis, frothy pink sputum, and purulent sputum with diffuse lung crackles. Percutaneous oximetric assessment of oxygenation or arterial blood gas evaluation of PO₂ should be performed when assessing a patient with suspected severe influenza. Relative hypoxia should trigger further assessment including a chest radiograph. Laboratory findings typically found at presentation with severe disease include normal or low-normal leukocyte counts and elevated creatine kinase^22,24,28 (Figure 140-3).

Figure 140-3 Suggested algorithm in the workup and management of suspected severe influenza pneumonia in the critical care unit.

Early laboratory diagnosis of influenza infection is greatly facilitated by the use of reverse transcriptase–polymerase chain reaction (RT-PCR) methodology. This assay should be employed when available in the evaluation of a patient with suspected severe influenza. Immunofluorescent techniques, enzyme-linked immunoassays, and other rapid diagnostic tests of clinical specimens often lack diagnostic sensitivity.^31,32 Viral cultures require up to 1 week for processing. Whereas RT-PCR is the preferred definitive diagnostic technique and has very high sensitivity, the adequacy of the clinical specimen is essential. Standard nasopharyngeal swab samples are adequate but can be falsely negative. Nasopharyngeal samples should be repeated in 48 to 72 hours if diagnostic suspicion remains. Paired nasopharyngeal and tracheal aspirates are useful for RT-PCR in intubated patients and may increase the diagnostic yield in critically ill patients.

Supportive Care

Almost all patients with severe infection in the ICU setting will have deficits in oxygenation and subsequently require ventilatory support.^22,24,33 Shock and renal failure can occur during therapy as a consequence of efforts to optimize oxygenation though diuresis coupled with high intrathoracic pressures and limited venous return.^24,28 Other important but less frequently seen disorders at presentation may include encephalitis (with or without obtundation or seizure activity), cardiac injury (myocarditis, pericarditis, conduction defects), and rhabdomyolysis.¹⁷

Most critically ill patients with severe influenza will manifest evidence of ARDS; supportive care for severe hypoxemia with diffuse pulmonary disease and supplemental oxygenation and ventilation assistance is required.²⁴ During pandemic periods, patients are often relatively young compared with non-pandemic years, and much greater numbers can be expected to be in need of ventilatory support than during a usual flu season.^{18,22,24,29,30}

Primary influenza pneumonia is unusual in that patients often display a relative insensitivity to usual measures of oxygenation assistance with PEEP. Controlled ventilation with attention to a lung-protective strategy,³⁴ in combination with appropriate sedation and judicious use of neuromuscular blockade, is appropriate. Avoidance of volume overload (and judicious diuresis) may also be associated with reduced duration of ventilation and length of stay in ICU for most patients with ALI and ARDS, and this strategy should be attempted for patients with influenza.^30,35 Other ventilation measures (despite unproven benefit in other forms of ARDS) that might improve oxygenation for individual patients have included prone positioning and inhaled nitric oxide.^36,37 HFOV is currently being evaluated as a rescue therapy for patients with severe ARDS in randomized controlled trials³⁸ and might be an option in patients with influenza-related refractory hypoxemia. ECMO remains a controversial option to manage severe respiratory failure in influenza-associated ALI in adults.^29,30,39 Clinicians in Australia similarly recommend consideration of ECMO for refractory hypoxemia in influenza infection.⁴⁰ HFOV and ECMO might be considered as a salvage therapy in centers familiar with these modalities in desperately ill patients.

Antiviral Therapy

In severely ill patients with suspected influenza, early initiation of antiviral therapy should be based upon clinical presentation and epidemiologic data and not delayed pending laboratory confirmation.^22,24,41 Various influenza strains are circulating throughout the world, and susceptibility to currently available antiviral agents is strain specific. The 2009 H1N1 swine influenza variant was resistant to amantidine but sensitive to neuraminidase inhibitors including oseltamivir and zanamavir.⁴² Oseltamivir-resistant strains were isolated during the 2009 H1N1 influenza A pandemic but fortunately were uncommon.⁴³ In contrast, the seasonal H1N1 influenza A strains circulating in 2008 and onwards are almost uniformly resistant to oseltamivir, yet many remain susceptible to zanamivir.^44–46 At this time, only an oral form of oseltamivir and an inhaled form of zanamivir are available for use.

Initiation of antiviral therapy within 48 hours of onset of symptoms of seasonal influenza is associated with a 1-day or greater reduction in duration of symptoms in ambulatory patients.^44,47 Oseltamivir therapy may reduce the risk of secondary bacterial superinfection.⁴⁸ Early therapy of severe influenza A 2009 H1N1 infections requiring ICU support with neuraminidase inhibitors contributed to improved outcomes.⁴⁹

Little data are available to guide the optimal dose or duration of therapy for antiviral agents. Severe influenza infections, including those caused by the 2009 H1N1 strain, can represent a systemic in addition to a pulmonary infection,⁵⁰ favoring the use of a systemic rather than an inhaled antiviral agent. Despite concerns over inadequate gastrointestinal absorption of oseltamivir among critically ill patients, published studies indicate comparable blood levels in ICU patients as compared with normal volunteers.⁵¹ Available evidence suggests that an oseltamivir dose of 75 mg twice daily is adequate; higher doses might be indicated and are the current subject of ongoing clinical trials.

Viral shedding can be prolonged in hospitalized patients with seasonal or pandemic influenza. In one study, approximately one-third of patients continued to shed live virus at least 1 week after symptom onset.⁵² Neuraminidase-inhibitor therapy for longer than 5 days has been used in outbreak situations and in immunocompromised patients known to shed virus for prolonged periods, but formal recommendations on optimal duration are lacking.³⁰ The intravenous neuraminidase inhibitor, peramivir, is available on a compassionate basis for emergency use in severe influenza pneumonia. The recommended dose is 600 mg of peramivir intravenously once daily for 5 days.⁵³

Adjunctive Pharmacologic Therapy

Several potential adjunctive immunomodulatory or antiviral therapies for treatment of severe influenza exist. Convalescent serum/plasma or hyperimmune globulin derived from patients who have recovered from influenza has been used for many decades. A series of studies were performed using convalescent plasma/serum during the 1918 pandemic and have recently undergone a meta-analysis showing that early, but not late, administration of such products may be associated with a significant survival benefit.⁵⁴ In addition, several case series suggest the possibility that similar therapy may be of use in severe influenza A/H5N1 infection.^55,56

High-dose corticosteroid therapy has been advocated for a variety of infectious and inflammatory conditions.⁵⁷ Corticosteroids have been useful as adjunctive therapy to suppress inflammatory responses in certain serious infections including severe influenza pneumonia. The uncertain benefits and known risks of corticosteroids in the presence of ongoing infection warrant caution before employing this strategy in primary influenza pneumonia. The use of glucocorticoid therapy for influenza is best limited to randomized clinical trial protocols rather than uncontrolled use. Similarly, a wide variety of immunomodulator agents are commercially available and might have salutary effects in selected patients (e.g., statins, peroxisome proliferator activated receptor alpha and gamma [PPARα, PPARγ] agonists, resveratrol). Many of these agents are readily available at low cost in developing countries and should be studied in controlled clinical trials in patients with severe influenza pneumonia.⁵⁸

Secondary Bacterial Pneumonia

Available evidence indicates that the majority of deaths from the 1918 pandemic occurred as a consequence of secondary bacterial infection.^18,19 Similarly, a substantial number of the deaths from the 1957 and 1968 pandemics were caused by bacterial co- or superinfection. The common pathogens in all series have been S. pneumoniae, group A streptococci, S. aureus, and Haemophilus influenzae. Given the frequency of secondary bacterial infection, clinicians should have a low threshold for considering antibiotic coverage against these commonly observed pathogens.

Secondary bacterial pneumonia as a complication of viral pneumonia takes two forms: mixed viral/bacterial pneumonia and postinfluenza pneumonia during the convalescent phase of influenza. Postinfluenza pneumonia is generally attributable to damaged airways and poor mucociliary clearance mechanisms following severe influenza pneumonia.¹⁸ The early mixed form of bacterial pneumonia during ongoing viral replication in the airways is more complex, with possible synergism between the bacterial and viral pathogens. Apoptosis of pneumocytes induced by the viral PB1-F2 protein facilitates pneumococcal growth in lung tissue.⁵⁹ Pneumococci bind to epithelial surfaces more readily if sialic acids have been cleaved by neuraminidase.⁶⁰ Viral neuraminidase from influenza virus has been found to promote pneumococcal adhesion in lung tissues and increase lethality in experimental pneumococcal pneumonia.⁶¹ Early institution of effective antiviral agents with neuraminidase inhibitors might serve to decrease virus replication and decrease the risk for secondary pneumonia.⁴⁸

Infection Control in the ICU

Patients with suspected influenza should be managed using droplet precautions by healthcare professionals, who should wear a standard surgical tie mask. There are different recommendations as to which face mask is optimal and whether N95 masks or similar personal respirators might be preferable to surgical tie masks. A recent study found limited to no additional protection of N95 masks in comparison to surgical masks, yet many still advocate their use during cough-inducing procedures when treating patients with influenza.⁶² Vaccines, when available against circulating strains of influenza, should be mandatory for all healthcare workers unless specific contraindications exist. Healthcare workers should also consider appropriate gloves when likely to have contact with body fluids or to touch contaminated surfaces, and they should wear gowns during procedures and patient care activities where clothing might be contaminated. Protective eyewear is recommended when providing direct care in close proximity to the patient.⁶³ Patients with suspected influenza should be in single patient rooms, if available, during the initial phase of hospital admission. If clinical demand exceeds the availability of such quarters, then cohorting of patients with influenza in common areas may be necessary. Influenza patients who must be transported outside of the room should wear a mask if tolerated, or when necessary, an oxygen delivery system that limits the spread of aerosols.

With respect to infection prevention and control related to mode of ventilatory assistance, there is circumstantial evidence from the SARS (severe acute respiratory syndrome) epidemic that noninvasive ventilation and HFOV may promote excess aerosolization of viral-laden particles and place surrounding patients and staff at risk. Limited evidence suggests that the process of endotracheal intubation, especially in an uncontrolled setting, may be associated with increased risk of acquiring infection; however, this risk is mitigated if adequate personal protective equipment is worn.^64,65 HFOV circuits should be equipped with microbial filters and a scavenger system to the exhalation port to limit aerosol generation.

Global Critical Care Collaboration

A working group composed of members from the international critical care community formed the International Forum for Acute Care Trialists (InFACT) to aid with global collaborative research in critical care.⁶⁶ For the 2009 H1N1 pandemic, the InFACT group focused on developing a case report form as a reference for generating an international “minimal clinical dataset” through collaboration with members of global critical care societies. This web-entry case report form system was available to clinicians around the world to contribute patient-based data in a manner that can be analyzed in real time and help inform decision makers and clinicians during the outbreak. InFACT also supports large, simple, investigator-initiated interventional studies in many countries. The impact of the InFACT initiative will only be determined over time, but this is an important global critical care attempt to more efficiently and more inclusively improve the care of critically ill patients.