Nonsuppurative Poststreptococcal Sequelae: Rheumatic Fever and Glomerulonephritis

History

Guillaume de Baillou (1538–1616), also known as Ballonius, first clearly distinguished acute arthritis from gout. Thomas Sydenham (1624–1689) described chorea but failed to associate this entity with other manifestations of ARF. Raymond Vieussens (1641–1715) published pathologic descriptions of mitral stenosis and aortic insufficiency. It remained, however, for William Charles Wells, in 1812, to emphasize the association of rheumatism and carditis and to provide the first clear description of subcutaneous nodules. Jean-Baptiste Bouillard, in 1836, and Walter B. Cheadle, in 1889, published extensive studies of rheumatic arthritis and carditis that synthesized the syndrome of rheumatic fever and have come to be regarded as classic works in this field and that form the basis for modern clinical concepts of ARF. In 1904, Ludwig Aschoff described the specific rheumatic lesion in the myocardium.

In 1880, J. K. Fowler pointed out the association between sore throat and rheumatic fever from his personal experience and, shortly after the dawn of the 20th century, Bela Schick identified ARF as one of the “nachkrankheiten” (sequelae) of scarlet fever. The introduction of Rebecca Lancefield's grouping system for β-hemolytic streptococci allowed clarification of the epidemiology of the disease by a number of investigators in the United States and the United Kingdom, including Coburn, Collis, Rammelkamp, Denny, Wannamaker, Massell, and Stollerman. Finally, the widespread introduction of antibiotic agents after World War II resulted in the development of strategies for primary and secondary prevention of rheumatic fever.

Etiology and Pathogenesis

ARF is a delayed nonsuppurative sequela of upper respiratory infection caused by group A streptococci (GAS), a conclusion firmly supported by several lines of evidence. There is a close temporal relationship between epidemics of streptococcal sore throat and scarlet fever and epidemics of ARF. About two-thirds of patients with ARF relate a history of preceding pharyngitis; even in the absence of such clear-cut evidence, elevated serum levels of antistreptococcal antibodies almost always document recent streptococcal infection. Prospective studies of primary and recurrent ARF have shown that this disease occurs only after an immunologically significant streptococcal infection. Finally, continuous antimicrobial prophylaxis, when successful in preventing intercurrent group A streptococcal infections, also effectively prevents ARF recurrences in rheumatic persons.

An intriguing and as-yet unexplained aspect of the host-parasite relationship is the fact that, as far as is known, cutaneous streptococcal infections do not initiate ARF. This may indicate a requirement for the pharyngeal site, with its rich endowment of lymphoid tissue, for initiation of the disease process, or it may result from a lack of rheumatogenicity among the so-called pyoderma strains of GAS. However, some suggest that cutaneous infections might play a role in the etiology of ARF, based on epidemiologic observations and circulating GAS strains in hyperendemic areas, but this remains unclear.

Host genetic factors appear to influence the susceptibility to ARF. Observational studies in the 19th century recognized familial tendencies to develop ARF, and in the early 1940s, studies showed familial tendencies to develop the disease, with greater risk occurring in children if both parents had rheumatic heart disease. Considerable efforts examining human leukocyte antigen type predisposition to ARF have been undertaken, and more recently, an allele in an immunoglobulin heavy-chain locus was shown to be associated with rheumatic heart disease in Micronesian populations. Alternatively, clustering by family also might reflect differences in the microbiome.

A substantial body of evidence indicates that GAS do vary in their rheumatogenic potential. Studies of outbreaks of streptococcal pharyngitis show that strains of certain M serotypes or genotypes are strongly and repeatedly associated with ARF ( Table 198.1 ), whereas strains of other equally prevalent types fail to initiate the disease or even to reactivate it in exquisitely susceptible hosts. Investigations of endemic ARF cases in Trinidad and Chile indicated that streptococci causing ARF belong to different serotypes than those causing acute glomerulonephritis (AGN) when the conditions occur simultaneously in the same population. Strains of GAS isolated from ARF patients may, however, differ widely in different geographic locales. Although the association of pyoderma strains of GAS (see Chapter 197 ) with ARF has been postulated by Australian investigators, such strains have never been definitively associated with ARF even when, as sometimes occurs, they colonize the throat. A study comparing group A streptococcal pharyngitis serotypes in Chicago in the 1960s, when ARF was common, to serotypes causing pharyngitis from 2001 to 2007, when ARF was very uncommon, showed marked reduction in circulating rheumatogenic types in the latter period.

Rheumatogenic streptococcal strains exhibit distinct biologic characteristics. Their M protein molecules share a particular surface-exposed antigenic domain against which ARF patients mount a strong immunoglobulin G (IgG) response. These strains fail to elaborate α(l)-lipoproteinase (so-called streptococcal opacity factor), and they are frequently heavily encapsulated. The latter feature is manifested by the formation of mucoid colonies on blood-agar plates. Whether such strains express a unique rheumatogenic antigen, however, remains unknown.

It is probable that not all strains of rheumatogenic serotypes are equally dangerous. The propensity of a given strain to cause ARF likely depends on its degree of virulence, a reflection of quantitative factors such as expression of M protein, hyaluronate, or other less well-defined biologic properties. Virulence is likely to be enhanced in epidemiologic settings that favor rapid person-to-person transmission.

Although GAS is the causative agent of rheumatic fever, the exact mechanism whereby this microorganism induces the disease remains unexplained. Several theories have been advanced. These include the following: (1) toxic effects of streptococcal products, particularly streptolysins S or O, which are known to be capable of inducing tissue injury; (2) serum sickness–like reaction mediated by antigen-antibody complexes, perhaps localized to sites of tissue injury; (3) autoimmune phenomena induced by similarity or identity of certain streptococcal antigens to a wide variety of human tissue antigens ; and (4) binding of streptococcal M protein to the CB3 region of collagen type IV, which may trigger antibody to collagen, leading to ground substance inflammation in the subendothelial matrix of endothelium and perivascular connective tissue.

Although none of these theories has been unequivocally proved or refuted, most attention has been focused on the concept of autoimmunity related to molecular mimicry. Interest in this mechanism has been spurred by the identification of antibodies in the sera of patients with ARF or rheumatic heart disease (RHD) that react with the human heart in a variety of test systems. These so-called heart-reactive antibodies (HRAs) are also present at a much lower titer in sera of patients with uncomplicated streptococcal pharyngitis. The presence of bound immunoglobulin and complement in the myocardia of children dying of rheumatic carditis suggests that circulating HRAs may have pathogenetic significance.

Molecular techniques have been used to study the relationship between specific peptides of the M protein molecule and human tissues. Epitopes of streptococcal M proteins have been identified that share antigenic determinants with cardiac myosin, sarcolemmal membrane proteins, synovium, and articular cartilage.

Goldstein and colleagues described a cross-reaction between group A streptococcal polysaccharide and a structural glycoprotein of human and bovine heart valves. Such a cross-reaction might explain the observation that serum levels of antibody to group A streptococcal carbohydrate appear to remain elevated for many years in patients with rheumatic valvulitis but not in rheumatic patients without valvulitis, and the levels decline remarkably after valve resection.

There is evidence that antibody to cardiac myosin in rheumatic carditis targets the S2 region and is similar among diverse populations worldwide. Many children with Sydenham chorea have circulating antibodies that react both with neurons of the caudate and subthalamic nuclei and with group A streptococcal cell wall carbohydrate. Taken together, these cross-reactive and toxic phenomena could explain most of the individual manifestations of ARF. On the other hand, it should be emphasized that no direct proof exists that these systems play any role in the pathogenesis of rheumatic fever.

Much of the work reviewed in the preceding paragraphs, particularly that related to HRAs and group A streptococcal carbohydrate, focused on humoral immune responses to streptococci. Indeed, serum antibody responses to streptolysin O, non–type-specific M-related antigens, and almost every other streptococcal antigen are, on average, more vigorous in patients with ARF than in those with uncomplicated streptococcal infections. However, it is likely that cellular immune responses to streptococcal antigens also play a critical role in the etiology of ARF. Preparations of streptolysin S contain a nonspecific mitogen that is closely related but separable from the hemolytic activity. In rheumatic persons, lymphocyte reactivity to streptococcal cell walls and membranes is heightened, but the reactivity to membranes is more striking and persists for several years after an acute attack. During active rheumatic carditis, both the number of helper (CD4) lymphocytes and the ratio of CD4 to CD8 cells are increased in heart valves as well as in peripheral blood. Production of interleukin-1 and interleukin-2 is enhanced. That both M protein and streptococcal pyrogenic exotoxins function as superantigens suggests a potential mechanism mediating the unrestrained immunologic assault postulated to cause ARF.

Pathologic Findings

Rheumatic fever is characterized pathologically by the presence of exudative and proliferative inflammatory lesions of connective tissue, most notably in the heart, joints, blood vessels, and subcutaneous tissue. In the early stages of the disease, there is fragmentation of collagen fibers, cellular infiltration that is predominantly lymphocytic, and fibrinoid deposition. This is followed shortly by the appearance of the myocardial Aschoff nodule ( Fig. 198.1 ). The Aschoff nodule is a perivascular focus of inflammation that consists of an area of central necrosis surrounded by a rosette of large mononuclear and giant multinuclear cells. The nuclei of these cells may contain a clear area just within the nuclear membrane (owl-eyed nucleus) or present a serrated (caterpillar) appearance, depending on their orientation in microscopic cross section. Such cells are known as Anichkov myocytes, although immunohistochemical studies demonstrated that they are of macrophage-histiocyte origin. Cardiac findings may include pericarditis, myocarditis, endocarditis, or combinations of these. Endocarditis involves the left side of the heart in almost all instances. A thickened and roughened area is frequently seen in the left atrium above the base of the posterior leaflet of the mitral valve (MacCallum patch). Valvular lesions begin as edema and cellular infiltration of the leaflets and chordae, with small verrucae along the line of closure. As healing progresses, the valves may become thickened and deformed, the chordae shortened, and the valve commissures fused, thereby resulting in valvular insufficiency with stenosis manifesting years later.

The joint lesions are characterized by fibrinous exudate over the synovial membrane and serous effusion without joint destruction. Histologic findings include cellular infiltration and fibrinoid degeneration. Subcutaneous nodules resemble Aschoff bodies in many features. They consist of a central zone of fibrinoid necrosis surrounded by histiocytes and fibroblasts; perivascular accumulations of lymphocytes and polymorphonuclear leukocytes are also apparent. Although scattered areas of arteritis and petechial hemorrhages have been found in the brain, their relationship to Sydenham chorea remains uncertain.

Epidemiology

ARF is most frequent in children ages 5 to 15 years. Only approximately 5% of cases occur in children younger than 5 years. Indeed, its relative rarity in infants and pre–school-aged children has led to the idea that repeated “primary” infections might be a precondition for the development of this disease. Both initial and recurrent episodes also occur in adults uncommonly. There is no clear-cut gender predilection, although a female preponderance exists in certain clinical manifestations, notably mitral stenosis and Sydenham chorea when the latter occurs after puberty. In temperate climates, rheumatic fever tends to occur more during winter or spring and less frequently during the summer.

It is difficult for North American physicians to comprehend the magnitude of the problem of ARF in developing countries. The disease is highly endemic in the Middle East, the Indian subcontinent, and selected areas of Africa and South America. The World Health Organization has estimated that approximately a half-million individuals worldwide acquire rheumatic fever annually, of whom 300,000 develop RHD. More than 15 million persons have been estimated to be living with this disease, and 319,000 die each year from RHD or its complications. Extraordinarily high rates of ARF and RHD are seen in Aboriginal populations, such as those in New Zealand and Australia, and in Pacific Islanders. The annual incidence of ARF in Aboriginal children ages 5 to 14 years in Australia's Northern Territory was 150 to 380 per 100,000, and the point prevalence of RHD among the Aboriginal population has approached 2%. In addition, the recognition that silent mitral regurgitation is common in school-age populations in many parts of the developing world may lead to a large increase in the estimated number of individuals affected by ARF/RHD.

The overall incidence of ARF in the United States cannot be ascertained precisely because of inherent difficulties in diagnosing the disease and because most states no longer maintain rheumatic fever registries. There is general agreement, however, that the incidence of ARF and RHD declined markedly during the 20th century in the United States and Western Europe. The rate of decline appears to have been particularly steep during the 1960s and 1970s. A survey in Memphis, Tennessee, indicated that during 1977 through 1981, the incidence of ARF in white suburban schoolchildren was only 0.5 per 100,000 annually. Similar rates were reported from many geographic areas of the United States. Traditionally, ARF in the United States was largely a disease of lower socioeconomic groups. The incidence was much higher in blacks than whites, a fact that appears to relate to basic environmental conditions rather than to any genetic predisposition of blacks for the development of ARF. The major identified predisposing environmental condition is household crowding. The degree of crowding markedly influences the acquisition rate of GAS (see Chapter 197 ) and hence the risk of development of ARF.

In the mid-1980s, a resurgence of ARF occurred in many communities in the United States. Beginning in early 1985, an epidemic occurred in Salt Lake City, Utah, and the surrounding intermountain area. By 2000, more than 500 cases had been diagnosed at the Primary Children's Medical Center in Salt Lake City. Smaller clusters of ARF, ranging from 15 to 40 cases, were reported during approximately the same time period from the following regions: Columbus and Akron, Ohio; Pittsburgh, Pennsylvania; Nashville and Memphis, Tennessee; Kansas City, Missouri; Morgantown and Charleston, West Virginia; Dallas, Texas; and New York City, New York. Moreover, for the first time in many years, outbreaks occurred in army and navy training camps.

Quite surprisingly, several of the 1980s civilian outbreaks involved children of middle-class families residing in suburban or rural settings. The group A streptococcal strains most strongly associated epidemiologically with these ARF outbreaks belong to the well-recognized rheumatogenic serotypes (e.g., types M-1, M-3, M-5, M-6, and M-18); particularly prominent in this regard were highly mucoid strains of M-18. Since the 1980s ARF has been very uncommon in most areas of the mainland United States, though outbreaks of rheumatic fever have been reported in the United States, such as that which occurred in Utah in the 1980s.

Persons who have suffered an initial attack of ARF have a marked predilection to develop recurrences after subsequent episodes of streptococcal pharyngitis. The risk of recurrence after streptococcal infection is highest within the first few years after the initial attack and then declines. It is unclear whether the reason for this decline is the length of time since the preceding attack or the older age of the patient. Nevertheless, rheumatic patients remain at an increased risk of recurrence well into adult life. Two other factors that positively correlated with a risk of rheumatic recurrences after streptococcal infection are the magnitude of the anti–streptolysin O (ASO) response and the presence of preexisting heart disease. In the classic studies conducted at Irvington House, New York, for example, 56% of streptococcal infections occurring in persons with RHD and accompanied by fourfold or higher ASO titer rises induced an ARF recurrence.

Clinical Manifestations

Rheumatic fever manifests itself as a variety of signs and symptoms that may occur singly or in combination. The most important of these, in terms of diagnosis, have been termed the major manifestations and include carditis, polyarthritis, chorea, subcutaneous nodules, and erythema marginatum. Certain additional findings frequently present in ARF, but those nonspecific in nature constitute the so-called minor manifestations: fever, arthralgia, first-degree heart block, and acute-phase reactants in the blood (C-reactive protein [CRP], leukocytosis, and erythrocyte sedimentation rate [ESR]).

The latent period between the onset of preceding streptococcal sore throat and the onset of ARF averages 19 days. The range is 10 days to 5 weeks. The average latent period is the same for recurrent attacks as for initial episodes.

The mode of onset is variable. If acute polyarthritis is the initial complaint, the disease may have a rather abrupt onset and may be marked by fever and toxicity. On the other hand, when isolated mild carditis is the initial manifestation, the onset of ARF may be insidious or even subclinical.

Most attacks begin with polyarthritis. Carditis, if it appears, usually does so early in the course of the disease. Overall, arthritis occurs in approximately 75% of first attacks of ARF, clinically evident carditis in 40% to 50%, chorea in 15%, and subcutaneous nodules and erythema marginatum in less than 2%. These incidences vary with age; carditis occurs most frequently when ARF strikes younger children, whereas the proportion of cases with arthritis increases with the age of the patient.

Carditis is the only manifestation of ARF that has the potential to cause long-term disability or death. Heart involvement in ARF may be pancarditis involving the endocardium, myocardium, and pericardium. While myocarditis and pericarditis may occur in ARF, the predominant manifestation of carditis is involvement of the endocardium presenting as valvulitis. Thus myocarditis or pericarditis without evidence of valvulitis is inconsistent with ARF. Nevertheless, in the absence of high fever or symptoms of acute pericarditis or congestive heart failure, it may be asymptomatic. Carditis almost always manifests itself within the first 3 weeks of an attack of ARF, if it appears at all. The clinical signs of carditis include the development of organic heart murmur(s) not previously present, cardiac enlargement, congestive heart failure, pericardial friction rubs, or signs of effusion. Because echocardiography is superior to auscultation for detection of valvular pathology, the 2015 Jones criteria were revised to include mitral regurgitation demonstrated using echocardiography. Thus now both pathologic and physiologic criteria that meet World Heart Federation criteria (rather than only pathologic) valvular involvement fulfills the criterion for carditis. Intractable heart failure with pancarditis may cause death in the acute phase of ARF, but this is rare. Echocardiographic studies show that patients with ARF and congestive heart failure have preserved left ventricular systolic function and severe mitral or aortic regurgitation or both. Serum levels of cardiac troponin I are not elevated in ARF patients with congestive heart failure. Thus the cause of heart failure appears to be acute valvular dilatation rather than myocarditis.

Chronic inflammatory changes involving the myocardium and endocardium may lead to the delayed development of chronic RHD ( Fig. 198.2 ). Endocarditis involves the mitral valve more frequently than the aortic valve. There are three characteristic murmurs of acute rheumatic carditis: a high-pitched blowing holosystolic apical murmur of mitral regurgitation, a low-pitched apical mid-diastolic flow murmur (Carey Coombs murmur), and a high-pitched decrescendo diastolic murmur of aortic regurgitation at the secondary and primary aortic areas. Murmurs of mitral and aortic stenosis are associated with chronic but not with acute rheumatic valvular disease. The tricuspid valve is involved much less frequently and the pulmonic valve very rarely. Delayed atrioventricular conduction, as manifested by first-degree or greater degrees of heart block, is a toxic phenomenon associated with ARF but is not in itself diagnostic of rheumatic carditis.

Joint involvement in ARF ranges from arthralgia without objective findings to frank arthritis characterized by heat, swelling, redness, and exquisite tenderness. There is an inverse relationship between the severity of joint involvement and the risk of development of carditis. The most frequently involved joints are the knees, ankles, elbows, and wrists. The small joints of the hands are less frequently affected, and the spine is only rarely involved. When the course of the illness has not been suppressed by antiinflammatory drugs, classically, multiple joints are usually involved; approximately 50% of patients develop arthritis in more than six joints. Arthritis in ARF is typically migratory in nature, that is, the inflammation travels from joint to joint. Once a joint becomes involved, inflammation begins to subside within a few days to a week. The evolution of arthritis in individual joints tends to overlap, so multiple joints may be inflamed at the same time. The typical migratory polyarthritis pattern may not be present, however, if effective antiinflammatory therapy is administered early in the course of the disease. In ARF, arthritis responds dramatically to salicylates or nonsteroidal antiinflammatory drugs (NSAIDs), and the absence of such a response should call the diagnosis into question. Moreover, the classic migratory pattern is not invariable. In some cases, the pattern may initially be additive, persisting in several joints simultaneously or even, rarely, monoarthritic. It should be noted that the 2015 revision of the Jones criteria introduces less stringent arthritis/arthralgia criteria only for those individuals in moderate- and high-risk populations ( Table 198.2 ).

In most cases, the entire bout of polyarthritis subsides within 4 weeks, leaving no residual articular damage or deformity. One possible exception to this is the very rare occurrence of the so-called Jaccoud form of periarticular fibrosis after rheumatic arthritis.

The existence of poststreptococcal reactive arthritis distinct from ARF has been postulated to occur in some patients whose arthritis is atypical in time of onset or duration, is nonmigratory, is unaccompanied by other major manifestations of rheumatic fever, and fails to respond promptly to salicylate therapy. The ultimate prognosis of such cases is unknown, but, in a very few cases, mitral valve disease has apparently ensued. Although the issue remains controversial, it seems prudent to consider all cases of poststreptococcal polyarthritis that fulfill the Jones criteria as representing ARF, provided that other common causes of polyarthritis have been excluded.

Subcutaneous nodules are rare but are almost always associated with severe carditis and tend to occur several weeks after its onset. They are firm and painless and vary in size from a few millimeters to 2.0 cm. Such nodules are usually found over extensor surfaces or prominences and over tendons. Common sites are adjacent to elbows, knees, wrists, or ankles and over Achilles tendons, the occiput, or spinous processes of the vertebrae. Their number varies from one, requiring a thorough physical examination to detect, to a few dozen, which can be quite prominent as described in 1889 by Cheadle ( Fig. 198.3 ). They usually persist for a few weeks. Somewhat similar but more persistent lesions are seen in rheumatoid arthritis. Pathology of the lesions is variable depending on the time since onset, but usually includes fibrous tissue (increasing in older lesions) and polymorphonuclear cells, as described by Jones in 1937 ( Fig. 198.4 ).

Erythema marginatum is a rare nonpruritic, nonpainful erythematous eruption usually seen on the trunk or proximal aspects of the extremities but rarely, if ever, on the face. The individual lesions are evanescent, moving over the skin in serpiginous patterns that can change before the observer's eyes and are often likened to smoke rings, with a tendency to advance at the margins while clearing in the center ( Fig. 198.5 ). The lesions are usually macular with slightly raised margins and appear to be more a vasomotor phenomenon than a manifestation of cutaneous pathologic changes. Individual lesions may come and go in minutes to hours, but the process may go on intermittently for weeks to months and can be exacerbated by a warm cloth. Histopathologic examination may show perivascular infiltration of lymphocytes and neutrophils in the dermis.

Sydenham chorea (St. Vitus dance) is a neurologic disorder characterized by emotional lability, muscular weakness, and rapid, uncoordinated, involuntary purposeless movements. The choreiform movements disappear during sleep and may be partially suppressed by sedation. The nonrhythmic movements are most notable in the face, hands, and feet. Sensation remains intact. Detailed descriptions of the nature of the choreiform movements can be found elsewhere. Individual attacks in hospitalized patients usually last 2 to 4 months.

Chorea may occur in close temporal association with other major rheumatic manifestations or in isolated form (pure chorea). In cases of pure chorea, laboratory evidence of acute inflammation (elevated CRP, elevated ESR) or recent streptococcal infection (elevated levels of antistreptococcal antibodies) may be lacking. This observation, which led investigators in the past to question the relationship of ARF to pure chorea, is now known to result from the fact that Sydenham chorea often occurs with a substantially longer latent period than the other manifestations of ARF. Relapses of pure chorea may occur in some patients despite faithful adherence to prophylaxis with intramuscular benzathine penicillin. Some patients with pure chorea are found on follow-up to have RHD manifested by mitral regurgitation or even mitral stenosis.

Interest has focused on the possibility that certain other neurobehavioral conditions, including tics, obsessive-compulsive disorder, and Tourette syndrome, may be poststreptococcal sequelae. This hypothesized entity has been termed p oststreptococcal a utoimmune n europsychiatric d isorders a ssociated with s treptococci (PANDAS). A prospective blinded cohort study of patients meeting suggested diagnostic criteria for PANDAS found that, after group A streptococcal infection, exacerbations of childhood tics and obsessive-compulsive disorders were more frequent in such patients than in control subjects. However, there has been no consistent relationship demonstrated between exacerbations and GAS infections. Streptococcal infection was not the only or even the most common antecedent of such exacerbations. The relationship of streptococci to this proposed entity is unproven at this time.

Several clinical manifestations of ARF occur with some frequency but are not in themselves specific enough to be considered major criteria. These minor criteria include fever, which accompanies almost all ARF attacks at their onset, and arthralgia. Slight differences in these minor criteria have been incorporated into the revised Jones criteria between low- and moderate- to high-risk populations. The pulmonary parenchyma in ARF rarely may be involved by a variety of pathologic processes, including pulmonary edema, atelectasis, pulmonary embolism, or thromboses. Some believe that, in addition, a specific rheumatic pneumonia may occur in rare cases. Abdominal pain or epistaxis can occur.

The average duration of an attack, in the absence of antiinflammatory therapy, is approximately 3 months. Less than 5% of cases persist for longer than 6 months, justifying the designation of “chronic” rheumatic fever. Stollerman listed the criteria for continuing clinical activity as follows: joint symptoms, new organic murmurs, changing heart size, congestive heart failure in the absence of long-standing valvular disease, subcutaneous nodules, sleeping pulse rate higher than 100 beats/min, erythema marginatum, chorea, elevated CRP levels, and a rectal temperature of 100.4°F or higher for 3 or more consecutive days.