Epidemiology

The overall incidence and severity of ARF have decreased in recent years in developed Western countries and in prosperous countries of Asia. Reliable morbidity data on the occurrence of ARF in total populations are lacking because studies often consider only a segment of a population. The trend seems clear, however. Some of the best long-term data come from Denmark, where rheumatic fever has been a reportable disease for many years. A steady decline has been occurring since 1900, except for a peak during World War II ( Fig. 29.1 ). In the United States, rheumatic fever is not a reportable disease, but mortality rates ( Fig. 29.2 ) and hospital discharge rates ( Fig. 29.3 and Table 29.1 ) have shown a steady decline. Although this decline was already under way, it seems to have been accelerated by the introduction of penicillin.

Reported annual incidence of ARF in Denmark, 1862–1962.

(Modified from Public Health Board of Denmark. Reported rheumatic fever incidence in Denmark, 1862–1962. In: Vendsborg P, Hansen LF, Olesen KH. Decreasing incidence of a history of ARF in chronic RHD. Cardiologia. 1968;53:332–40. Used with permission of S. Karger AG, Basel.)

U.S. national mortality rates from rheumatic fever in individuals 5 to 19 years old, 1921–78. Values are age adjusted to the 1950 U.S. population. Trend lines are fitted for each era (1921–45 and 1946–78) separately and take International Classification of Diseases (ICD) revisions into account. The arrow separates the two eras.

(From Massell BF, Chute CG, Walker AM, Kurland GS. Penicillin and the marked decrease in morbidity and mortality from rheumatic fever in the United States. N Engl J Med. 1988;318:280–6.)

Annual rates (per 100,000) of discharge of patients with rheumatic fever from short-stay nonfederal hospitals in the United States, 1965 to 1983.

(From National Hospital Discharge Survey, NCHS. Modified from Gordis L. The virtual disappearance of rheumatic fever in the United States: lessons in the rise and fall of disease. Circulation. 1985;72:1155–62.)

The dramatic decline in the incidence of rheumatic fever began in the United States in the late 1940s (see Fig. 29.2 ). During the late 1950s, the 1960s, and the early 1970s, studies in the United States showed annual rates of 13.5 to 62.5 first attacks per 100,000 children 5 to 14 years of age; however, these studies are not strictly comparable in design. Some of them were conducted in low-income urban areas, locations in which the incidence of rheumatic fever was thought to be higher. Secondary prevention of rheumatic fever with penicillin prophylaxis to protect against recurrent attacks probably became widespread in the 1960s. Denny and colleagues showed the possibility of preventing initial attacks with injectable penicillin in 1950 in military camps after the original observation by Massell and colleagues. Similar conclusive controlled studies were not repeated in the general or pediatric populations or with oral penicillin.

The most compelling evidence that appropriate medical intervention helps reduce the number of initial attacks of rheumatic fever comes from a study by Gordis of health care availability in an at-risk inner-city Baltimore population. The rate of ARF was reduced by 60% over the course of a decade in only those census tracts receiving a comprehensive care program, with that reduction occurring only in patients with an identifiable preceding clinical respiratory tract infection. A similar trend with small numbers of patients was seen for the Navajo and Papago American Indian populations in school-based intervention programs and in an Alaskan program. Other school-based interventions have reduced streptococcal prevalence rates but have not gone the necessary further step to show a reduction in rheumatic fever morbidity in a controlled, carefully demarcated population. A recent systematic review and meta-analysis of available evidence supports the premise that a community- or school-based initiative can reduce rheumatic fever by an estimated approximately 60% (relative risk, 0.41; 95% confidence interval [CI], 0.23–0.70), acknowledging the eligible studies are of variable quality but are unlikely to be replicated.

Promising evidence is accruing with reduction of first-episode ARF in a school-based program in a well-demarcated population in New Zealand that may eventually provide evidence for primary prevention of ARF using an oral penicillin preparation outside of a military population. Recurrences in this population had already been controlled.

An unexplained high incidence of ARF persists in Hawaii, especially in Polynesian and part-Polynesian children. Similarly, rates are higher in larger populations of Polynesian (including indigenous Maori) children in New Zealand, although increasing evidence from that country points toward low socioeconomic status, including housing and income issues, and associated factors such as a lesser degree of education playing a role. In some areas of the continental United States, the rate of endemic rheumatic fever continues to exceed that of the general population in some children traditionally considered to be at risk (urban African Americans), although not in others (recent Hispanic immigrants). In contrast, in a hospital-based study in another urban area, Hispanic children, reflecting the pediatric population being served, were the most affected.

From the beginning of 1985, the number of patients with ARF in several centers on the U.S. mainland increased. Although these numbers were not large, they represented a definite increase in the outbreak areas ( Table 29.2 ). The resurgence in Utah since 1985 led to more than 600 cases. Nationally, however, the number of diagnoses of ARF, in the absence of robust surveillance data, appeared to continue to decline gradually from 1984 through 1990. The populations (aside from clusters in military populations) generally did not seem to be those considered to have been at risk in the past; most patients were white and middle class and lived in suburban or rural communities with ready access to medical care. In the Utah and Tennessee outbreaks, the families were larger than the state average. The military outbreaks were the first in 2 decades in U.S. military personnel. Factors other than the widespread use of antibiotics and improved availability of health care are likely to be important because group A streptococci are rated moderately infectious agents in close contacts. Overcrowded living circumstances long have been considered a risk factor. This hypothesis was not substantiated as an important risk factor by conditional logistic regression in a modern (Yugoslavian) study but was supported in a recent ecologic study also exploring multivariant relationships where ARF rates are high. The disquieting feature in these more recent U.S. outbreaks was the severity of the illness, especially in the Salt Lake City outbreak: 72% of patients had clinical carditis ( Table 29.3 ), 19% had severe carditis with or without congestive heart failure, and three patients required mitral valve replacement. However, this report was from a cardiology center, which is likely to reflect a referral bias. Earlier U.S. reports documented clinical carditis in initial attacks of rheumatic fever in 40% to 51% of patients (1951 through 1965). Longitudinal observations at the same institution suggest a decreasing frequency of clinical carditis (73% from 1921 to 1930, 51% from 1951 to 1960).

With the use of Doppler echocardiography, carditis rates in the Utah patients increased to 91% (see Table 29.3 ). The duration of secondary penicillin prophylaxis and endocarditis prophylaxis is an important consideration for patients with nonclinical carditis. The place of echocardiography in the diagnosis and management of rheumatic fever continues to evolve (see “ Laboratory Findings ”). Rheumatic fever presenting with arthritis may be misdiagnosed or not be admitted to the hospital, which is likely to affect estimates of rates of carditis in studies that are not population based.

A more recent assessment of hospitalized ARF disease in the United States in children younger than 21 years of age found a low rate of 14.8/100,000 hospitalized children (503 cases). The prevalence of RHD, which represents the result of many years of exposure to the risks for acquiring ARF, seems to have declined in the United States over the course of many years.

Many different racial and ethnic groups have been deemed unusually susceptible to rheumatic fever. They usually have been minority groups within a given area who are of lower socioeconomic status than the general population (e.g., Malays in Singapore, Arabs in Israel, Bantus in South Africa, indigenous Maori and Pacific Polynesians in New Zealand, blacks in the United States, aborigines in Australia). In the United States, when differences in socioeconomic status or degree of crowding were taken into account, the differences in the incidence of rheumatic fever, prevalence of RHD, and mortality from ARF and RHD generally declined or disappeared, at least in the population studied. Rates in African Americans were slow to decline, however. Rates became very low in the 1970s, at least in some areas, but cases persisted in other areas.

Gordis and associates caution that some other socioeconomically determined factor closely paralleling crowding could be the actual determinant of rheumatic fever. Most authorities agree that the reduction in the incidence and severity of rheumatic fever that has been noted in the United States and Western Europe might be due partly to a higher standard of living and less crowding. In a case-control study in the former Yugoslavia, home dampness, change of place of residence during the previous 5 years, low maternal education, body weight below normal, frequent sore throats, and a positive family history of rheumatic fever were found to be significant risk factors.

A relationship between group A streptococcal throat infection and rheumatic fever was recognized in observations of the latter occurring after outbreaks of scarlet fever. The attack rate noted after group A streptococcal throat infection varied widely (from 3% at Warren Air Force Base to 0.39% in Chicago children). Further observations in the latter study revealed that exudative pharyngitis, a positive throat culture with persistence of group A streptococci beyond 21 days, and the development of significant antibody (antistreptolysin O [ASO]) responses were associated with a higher attack rate that approached 3% in the children studied. In the New Zealand study an estimated attack rate of 0.4% was found in the school population (aged 5–17 years) studied, with approximately 60% with exudative pharyngitis from another study in the same population.

Rheumatic fever, similar to streptococcal pharyngeal infection, occurs most commonly in children 5 to 15 years of age. First attacks of rheumatic fever rarely occur in children younger than 3 years of age or in adults older than 40 years because of the relative infrequency of streptococcal infection at these ages and perhaps other factors.

The incidence of ARF is highest in the spring and winter months in temperate zones and coincides with the seasonal variation in streptococcal pharyngitis. This incidence may be related to the greater tendency for spread of streptococcal infection by closer contact during the colder and damper months, at least in some climates. In other climates, a seasonal peak for ARF is less pronounced.

Pathogenesis

Evidence points toward ARF being preceded by a group A streptococcal upper respiratory tract infection ( Fig. 29.4 ). The events that occur after such infection and that culminate in rheumatic fever are poorly defined, suggesting a complex interaction of numerous factors. With the resurgence of interest in this disease after the outbreaks in the 1980s in the United States and more recently with increased global interest, laboratory data collected with the use of modern technologies have led to some new insights. Pathogenesis involves the host, the environment (see “ Epidemiology ”), and group A streptococci, individually and collectively. Group A streptococci possess an enormous number of virulence factors. The M protein is a major surface protein on the surface as coiled-coil fibrils. The most common method of typing currently is by nucleotide sequencing of the variable 5′ end of the M protein gene ( emm ). Considering the full architecture of the M protein there are three main groups represented by the prototypic M-5, M-80, and M-77 molecules. This approach may aid future epidemiology and vaccine development. So-called rheumatogenic strains of group A streptococci have been the subject of much discussion, more recently in relation to the U.S. focal upsurges in rheumatic fever in the 1980s and 1990s. Because no factor has been described or isolated, however, such strains remain a hypothesis. To date, rheumatic fever has been shown to occur only after nasopharyngeal infection, although debate and research continue. Why the site of infection seems to predispose individuals to the development of rheumatic fever remains an enigma, perhaps related to skin lipids. Acute glomerulonephritis develops after skin or throat infections with a nephritogenic type of group A streptococci (e.g., M-49 and M-12).

Suggested pathophysiologic model for rheumatic fever.

(Modified from Azevedo PM, Pereira RR, Guilherme L. Understanding rheumatic fever. Rheumatol Int . 2012;32(5):1113–20.)

Certain streptococcal M protein serotypes are implicated strongly and repetitively in epidemics of ARF. Serotypes M-3, M-5, M-14, M-18, and M-24 have been reported more than once in outbreaks, and M-1, M-6, M-19, M-27, and M-29 have been reported once only. Distinct nucleotide sequences that determine different gene subfamilies encoding the M or M-like protein antigenic domain ( emm gene) may be the cause of these variations in streptococcal rheumatogenic potential, depending on the site of infection. Other equally prevalent M types rarely, if ever, have been associated with epidemics of the disease or have failed to cause recurrences in susceptible patients. The U.S. resurgence in the 1980s lends limited support for this concept. No predominance of a single serotype within a specific geographic zone was identified in any of the published outbreaks (see Table 29.2 ). Specific M types and their production of mucoidal colonies, considered to be related to the amount of M protein and virulence, may be more relevant to epidemic than to endemic rheumatic fever.

There are limited reports of endemic rheumatic fever–associated emm types. In Auckland, New Zealand, where rheumatic fever is endemic, in a large urbanized minority population an average of 50 highly scrutinized new cases of ARF currently occur annually in children aged 5 to 14 years of age (annual age-specific rate of 25/100,000 per year; Maori and Pacific Polynesians 60/100,000). Nine years (1984 to 1992) of surveillance of group A streptococcal isolates from hospitalized pediatric patients (one centralized children’s facility for 8 of the 9 years in question) yielded 2410 isolates. Only three of 38 throat isolates (32 from well-documented cases of rheumatic fever, six from siblings) were strains described as possibly rheumatogenic (one each of M-1, M-3, and M-6). None was described as mucoidal. In that series, M types 6, 53, 55, and 66 (and NZ-1437, now known as M-89, when sibling isolates were included as cases) were statistically more likely to be associated with a case of ARF. Further prospective emm surveillance in ARF cases in Auckland demonstrates a wide diversity of emm types in this population where rheumatic fever is endemic ( Fig. 29.5 ). Isolates associated with ARF and pharyngitis in Hawaii ( Table 29.4 ) are rarely seen in the continental United States. Of importance, the assumed link between rheumatic fever, the disease, and the disease-causing emm types is only temporal. It may itself not be sufficient to build a hypothesis for a possible association of initial infection site and ARF.

emm types in acute rheumatic fever cases with potential coverage by the 30 valent vaccine and cross-reaction, Auckland 2009–12.

Strain selectivity is a further consideration. No documented evidence has shown that all members of an M type may be equally able to elicit ARF. Some streptococci from a particular serotype may be associated with ARF and acute poststreptococcal glomerulonephritis, although the two sequelae rarely occur simultaneously. M types have been shown to be composed of genetically diverse streptococci, not all of which may be established within a community. This has been again demonstrated with modern tools, with analysis of the full-length sequence of 51 M proteins from Brazilian and Belgium strains of group A streptococci. Genetic analysis of serotype M-18 from ARF cases in Utah separated by 12 years, however, showed strains nearly genetically identical. The M type denotes possibly nothing more than a shared type-specific marker, with the property of rheumatogenicity as yet remaining elusive. Streptococcal strains that are opacity factor negative (a lipoprotein lipase) are unlikely to be rheumatogenic, according to some investigators. This finding does not hold up in all geographic areas. Surveillance of group A streptococci in multiple geographic zones will be necessary to guide M protein–based vaccine development. Other vaccine options are being explored (see later discussion).

Although current evidence strongly implicates an immunologic mechanism in the pathophysiology of rheumatic fever, the details of how the disease develops are unclear. Evidence to date strongly suggests an abnormal cell-mediated and humoral immune response to cell membrane streptococcal antigens, which, because of molecular mimicry of human tissues, may result in continued damage to the cardiovascular and nervous systems. The findings of circulating immune complexes in most patients and the deposition of C3 and immunoglobulin in the myocardium of patients dying of ARF support an abnormal immune response in rheumatic fever.

M proteins from some rheumatogenic group A streptococcal types share antigenic determinants with myosin, with the sarcolemma of cardiac muscle, and with antigens of articular cartilage and synovium. The immune response to streptococci may mistake the host antigens as foreign and result in tissue damage. Other streptococcal antigens, such as the group A carbohydrate component, are candidates for mistaken cross-reaction with a glycoprotein in human heart valves. Group A streptococci have components that can amplify or downregulate the immune response. The mechanisms for molecular mimicry leading to central nervous system dysfunction are less defined.

The site of the initial streptococcal infection may be important; lymphatic channels have been shown between the tonsils and the heart. Unusual compartmentalization of rheumatic antigen–positive non-T cells has been shown in patients with ARF, with no positive cells detected in rheumatic tonsils but increased numbers in peripheral blood.

Cell-mediated immunity to streptococcal antigens also is enhanced in patients with rheumatic fever. The lymphocytic infiltrate of heart valves was found to be composed predominantly of CD4 ⁺ helper cells. Increased expression of human leukocyte antigen (HLA)-DR on fibroblasts, which can present antigens to CD4 ⁺ lymphocytes (cytotoxic/suppressor T cells), has been observed on the heart valves of patients with acute carditis. The cytotoxicity induced in normal human helper and suppressor cells in vitro by purified protein from a type M-5 group A streptococcal organism has been shown to destroy several human cell types, including cultured myocardial cells. T-cell subset study results are conflicting, but production of interleukins is reported to be enhanced. The role of M protein and streptococcal pyrogenic exotoxins as superantigens is being explored and perhaps might explain the exaggeration of the streptococcal immune response.

ARF pathogenesis is mediated by autoimmune mechanisms: the production of antibodies and T cells that react with self-antigens. Antibodies have been observed deposited in the myocardium and valves of ARF patients following death, interstitial immune nodules (known as Aschoff bodies) are a hallmark of rheumatic carditis, and CD4 ⁺ and CD8 ⁺ T cells have been shown to be attached to rheumatic heart valve endothelium. Despite this relatively well-documented histopathology, there is still no consensus on which antigens initiate the autoimmune response, nor is there a clear understanding of the immune cell profile in ARF. A better understanding of pathogenesis is crucial to develop interventions, both preventative and therapeutic. The application of modern immune-profiling technologies, which have been used with success in other autoimmune diseases, has the potential to identify triggers of autoimmunity and characterize the immune signatures for ARF.

The genetic background of the human host seems to influence susceptibility to rheumatic fever. There are theories about the roles certain genes might play. Aggregation of rheumatic fever cases in families has been recognized for some time. Low concordance for inheritance has been reported in monozygotic twins, although affected siblings have significant concordance for arthritis, residual RHD, and chorea. A recent systematic review and meta-analysis of twin studies estimated heritability across all studies of 60%, suggesting whole-genome scanning might provide a clinically useful genetic risk prediction tool for rheumatic fever and RHD. Patarroyo and associates found that the B lymphocytes of patients with rheumatic fever have a specific marker (883 alloantigen) associated with host rheumatic susceptibility. It seems to transcend ethnicity, although studies in India and Africa were less supportive, and may be similar to an immune response gene. This work has been extended with the use of monoclonal antibodies to family members of patients with rheumatic fever. Class I HLA molecules have not been associated with ARF. Many studies in different populations have shown an association with HLA-DR but without a single HLA marker for susceptibility. Genetic factors alone seem highly unlikely to be responsible for susceptibility to rheumatic fever. No predictive marker for susceptibility to rheumatic fever has been defined to date. Studies in different populations have yielded conflicting results exploring the possibility that gene polymorphisms influence severity of heart disease in rheumatic fever patients.

Vaccine Development

Immunity to group A streptococci and to rheumatic fever is thought to depend largely on antibodies to M protein, a major virulence determinant; such antibodies can opsonize the bacteria in the presence of neutrophils. Immunity is thought to be strain-specific and to depend on antibodies to the variable serotype-specific regions of the protein, and earliest vaccine development followed this pathway. Two early landmark studies showed that vaccines containing purified M protein evoked protective immune response in humans. Antibodies against the variable amino-terminal end of the M protein opsonize streptococci in a type-specific manner, but the results of experiments in animals suggest that the conserved carboxyl-terminal end also may be an immune target. Some human evidence suggests that this conserved epitope acts as a subunit vaccine. Complexities in this area include the risk for inducing cross-reacting antibodies that could injure rather than protect. Separation of the peptide fragments of M proteins (epitopes) that evoke type-specific and not cross-reacting antibodies was an important step. Most of the cross-reactive epitopes identified have been located in the middle of the M proteins ( Fig. 29.6 ) and can be separated from the N-terminal, hypervariable protective epitopes that evoke antibodies with the greatest bactericidal activity.

Schematic representation of the structural and functional domains of streptococcal M protein.

(From Dale J. Current status of group A streptococcal vaccine development. In: Finn A, Pollard A, eds. Hot Topics in Infection and Immunity in Children IV: Advances in Experimental Medicine and Biology , vol. 609. New York: Springer; 2008:53–63.)

More than 200 different defined serotypes (M types) of group A streptococci have been identified. Vaccine development falls into two broad approaches: vaccines based on common protective antigens of group A streptococci and multivalent vaccines based on type-specific N-terminal regions of the M protein. A 26-valent vaccine with components guided by North American surveillance of ARF, invasive infections, and pharyngitis has been shown to be safe and immunogenic in adults. This vaccine includes 80% to 90% of important serotypes in North America. More recent epidemiologic studies with additional data from North America and Europe, as well as some from developing countries, led to a 30-valent vaccine with broader coverage and promising animal model data against vaccine serotypes. In addition, the researchers reported significant levels of bactericidal antibodies against a further 24 of 40 nonvaccine serotypes of group A streptococci. This vaccine is now in phase I trials in North America.

However, despite 98% coverage of pharyngitis isolates in North America, preliminary data from prospective studies of pharyngitis infections in Mali and South Africa indicate only 40% and 59% coverage. As stated earlier there are few studies with pharyngeal emm typing associated with rheumatic fever cases. Epidemiologic studies of group A streptococci from low-income countries are sparse. In general, a wider diversity of emm types are seen, indicating that vaccine coverage with current tools may be incomplete in other areas (see Fig. 29.6 ), such as in Hawaii, Asia, Cape Town in South Africa, Mali, or New Zealand, where rheumatic fever is endemic in some populations. More comprehensive information for serotype prevalence in such areas of high endemicity is urgently needed. Cluster typing of S. pyogenes throat and skin isolates provides groups more likely to be associated with pharyngitis or skin infection with the latter more often seen from ARF endemic areas. However, there remains a paucity of pharyngeal isolates from these areas.

Rheumatic Fever in Developing Countries

A global profile for ARF and RHD has emerged in the past decade. The burden of RHD, larger than previously thought, is able to be better quantified with echocardiography ( Fig. 29.7 ). RHD is considered one of the few preventable chronic diseases. Despite impressive declines in incidence in developed countries (see Figs. 29.1 and 29.3 ), globally RHD remains the most common form of acquired heart disease. Four-fifths of the world’s population live in developing countries where the prevalence of RHD suggests that the incidence of ARF remains at high levels in areas characterized by crowded living quarters and lower socioeconomic conditions. In Soweto, South Africa, the prevalence of RHD in the 1970s was estimated by cardiac auscultation at 7.1/1000 schoolchildren. A recent estimate using echocardiography in two developing-country settings increased this estimate by nearly 10-fold. Studies have been repeated in other settings ( Fig. 29.7 ) with differing echocardiographic criteria but similar estimates of burden. The true meaning of these findings in asymptomatic subjects in high-prevalence populations requires a better understanding of the natural history of subclinical carditis, standardization of the echocardiographic criteria, and cost-benefit analyses in various settings. Data on natural history is accruing. A recent consensus standardization guideline has been published. Reliable, quality-controlled global supplies of benzathine penicillin for prevention of further attacks may need to be considered.

Rheumatic heart disease prevalence rates in children: echocardiography-based screening versus clinical examination.

(From Marijon E, Mirabel M, Celermajer DS, Jouven X. Rheumatic heart disease. Lancet. 2012;379:953–64.)

An important recent prospective study in patients 14 years of age or older from Soweto highlights a significant disease burden of new-onset RHD presenting in the third and fourth decades of life, suggesting a presumed unrecognized or subclinical earlier ARF episode, most likely in childhood. RHD is an important risk factor in obstetrics in low-resource settings and in other vulnerable groups. The contribution of RHD to the burden of hospitalized disease resulting from strokes, atrial fibrillation, and other sequelae requires more elucidation.

Given the difference in medical care delivery, evaluation of the incidence of ARF must be viewed with caution. Estimates suggest, however, an annual incidence of 200 to 400 cases/100,000 population in Soweto. The incidence of rheumatic fever (as judged by hospital admissions for RHD between 1966 and 1980) has remained stable in India during this period of rapid decline in the United States and the West. Some recent publications have attempted to better define the global burden of ARF (and RHD and other group A streptococcal diseases) with variable rigor. Community-based secondary penicillin prophylaxis programs in developing countries are considered cost-effective and may be more achievable than is primary prevention.

However, the case can be argued, even in resource-poor settings, for a comprehensive primary and secondary prevention program, particularly because benzathine penicillin is listed as a World Health Organization (WHO) essential medicine.

Pathology

The unique pathologic lesion of rheumatic fever is the Aschoff body, which generally is considered to be a granuloma that results from injury to collagen fibers. Classically Aschoff bodies are found in the heart, usually in the left atrial appendage, but similar foci can be found in the synovia of the joints and in and around joint capsules, tendons, and fascia.

The early pathologic response to rheumatic fever may be an exudative reaction with Aschoff-like bodies as an inflammatory focus. They are cardiac or extracardiac, with a central area of fibrinoid necrosis surrounded mostly by polymorphonuclear leukocytes. Clinically, this condition may be manifested as arthritis and spontaneously subside in 2 to 4 weeks. No residual joint damage results. The proliferative phase of classic Aschoff nodules, which consists of central necrosis surrounded by a rosette of large mononuclear cells, giant multinuclear cells, and other cell types, is confined to the heart and usually causes pancarditis with simultaneous involvement of all three layers (i.e., the pericardium, the myocardium, and the endocardium). This event may result in permanent valvular damage in the following order of frequency: the mitral valve, the aortic valve, the tricuspid valve, and, rarely, the pulmonary valve.

The heart disease encountered clinically usually is mitral regurgitation, aortic regurgitation, or both. The scarring that leads to valvular stenosis (mitral or aortic) typically takes decades to develop but may occur much faster in hyperendemic areas. This process is not the full story, however, because although rheumatic mitral valve stenosis occurs more commonly in India and occasionally in other less advantaged populations, it was never a common occurrence in the United States or the United Kingdom at the height of rheumatic fever incidence.

The presence of Aschoff bodies is not evidence of rheumatic activity because these lesions are found in biopsy specimens of the left atrial appendage many years after an acute attack of rheumatic fever. Little is known about the pathology of Sydenham chorea, and the pathologic changes cannot be related to the clinical manifestations. Patients rarely die of this form of rheumatic fever.

Clinical Course

The stage is set for the development of rheumatic fever in a susceptible host after a pharyngeal infection by one of the types of group A β-hemolytic streptococci. Not all serotypes (M or emm types) appear to be equally rheumatogenic (see “ Pathogenesis ”). Rheumatic fever develops in approximately 1% to 3% of children with known epidemic untreated exudative pharyngitis and a culture positive for group A streptococci. The frequency decreases to less than 1%, as shown in the one controlled study involving children, when patients with less severe or less precisely diagnosed streptococcal infections are included. This finding has been replicated more recently in a large community- and school-based randomized, controlled trial with high consent and retention rates in New Zealand in an area with endemic rheumatic fever. The attack rate observed was approximately 0.2%.

The preceding pharyngitis is not recognized as an illness by the patient or parents in approximately 10% to 33% of cases of ARF, although 50% to 60% of patients remember having a sore throat. In the New Zealand trial, episodes of sore throat (with appropriately increased streptococcal serology) in this carefully monitored series preceded development of ARF in 14 of 19 (74%) of the cases enrolled in the program at the time of presentation. In some case series, this figure is lower but the methodology is less rigorous (see Table 29.2 ). The infection is followed by a latent period that averages 19 days in duration, during which time the patient seems well. The range seems to be between 1 and 5 weeks but has been difficult to establish. In the New Zealand trial, the average latent period was 27 days (range, 2 to 49 days) for seven rheumatic cases with proven group A streptococcal pharyngitis and raised streptococcal titers. The average latent period is the same for recurrent attacks as for initial episodes.

ARF then begins. Table 29.3 suggests a clinical profile in epidemics in the United States, although recurrent cases with their increased risk for carditis are included. The clinical profile in all settings is similar. In a prospective study in India, 67% of initial episodes were associated with migrating arthritis involving one or more of the large joints accompanied by a fever of 38°C to 39°C (100.4°F to 102.2°F), malaise, and anorexia. Just as the redness, swelling, and pain in a knee subside, the whole process may start again in the ankle. The elbows and wrists are likely to be involved. Arthritis of the hip also occurs. The joint involved may be exquisitely tender, with the patient exhibiting extreme reluctance to move the joint or mobilize. Typically, multiple joints are involved in tandem, with overlap over the course of time, when symptoms are not suppressed by antiinflammatory therapy. The whole polyarthritic episode usually subsides over the course of 4 weeks, with no residua remaining. The joint inflammation may be of low grade in some individuals, without limitation of motion or outward manifestations of redness and swelling (arthralgia). The literature supporting monoarthritis not developing into migrating polyarthritis in ARF is unconvincing. However, the clinician should act cautiously in an area endemic for rheumatic fever in the early phase of presentation. The development of polyarthritis in a patient with a culture-negative monoarthritis (e.g., a hip joint in a patient without prior antibiotic exposure) may be aborted by nonsteroidal antiinflammatory drugs (NSAIDs). An early echocardiogram may assist in establishing the diagnosis (see “ Laboratory Findings ”).

In a New Zealand series 30% of sequential ARF cases ( n = 119) in a population-based study presented with monoarthritis, 14% (5/34) of the monoarthritis cases had received an NSAID agent, and 74% (25/34) of confirmed ARF patients with monoarthritis at presentation had echocardiographic changes diagnostic of ARF. At the time of the study, and according to the New Zealand diagnostic criteria, monoarthritis was allowed as a major criterion only when the patient had received an NSAID. On the basis of this new information the New Zealand diagnostic criteria now allow monoarthritis without exhibition of an NSAID as a major criterion ( www.heartfoundation.org.nz ). More recently the American Heart Association (AMA) has followed suit.

Carditis generally appears early in the illness (first 2 to 3 weeks) if it is going to occur. It may be silent in the absence of elevated fever or symptoms of pericarditis or cardiac failure. The clinical signs are the development of a new heart murmur, cardiac enlargement, congestive cardiac failure, a pericardial friction rub, or signs of cardiac effusion. An echocardiogram early on may be diagnostic of carditis using international diagnostic criteria even in the absence of a murmur.

However, the mode of onset of ARF can be quite variable, and access to medical care, the health knowledge of the patient or family, or the awareness of health care personnel may be factors determining when in the illness the patient initially seeks medical care. Indolent carditis, as an insidious or even subclinical presentation now recognized by the Jones criteria ( Table 29.5 ), may present without other manifestations to fulfill the Jones criteria because the initial episode is sometime in the past. This was an uncommon but recognized presentation in the New Zealand school trial.

In a classic presentation on examination, the striking findings are the patient’s pallor and discomfort, especially on movement of the affected joints. Carditis usually appears within 3 weeks of the start of the illness, and, in the absence of high fever or symptoms of acute pericarditis or congestive heart failure, it may be asymptomatic. Examination of the heart may reveal in at least half of patients a grade 2/4 apical pansystolic murmur that is transmitted to the axilla (mitral insufficiency) with or without an apical mid-diastolic flow murmur (Carey-Coombs murmur); half of these patients also may experience an early diastolic grade 2/4 murmur at the left sternal edge (aortic insufficiency). The latter can also occur alone. The tricuspid valve is not usually involved and the pulmonary valve very rarely. In addition, less commonly, the child can have congestive heart failure or cardiac enlargement, indicative of active carditis. Carditis is more likely to occur in younger children. Pericarditis may be suspected with muffled heart sounds, a frictional rub, or chest pain. It becomes less common as ARF in a population becomes less severe. Ruptured chordae tendineae in a patient with fulminant carditis may present as unilateral pulmonary edema. Death is a rare but well-described sequela of the acute phase of the disease. Murmurs of mitral and aortic stenosis are associated with chronic but not with acute rheumatic valve disease.

Echocardiography in ARF cases with congestive cardiac failure has shown preserved left ventricular systolic function and/or severe mitral or aortic regurgitation. Serum levels of cardiac troponin I are not elevated in ARF patients with heart failure.

The distinctive rash, erythema marginatum, was observed in up to 12% of patients in these case series ( Fig. 29.6 ; see Table 29.3 ). It is neither pruritic nor painful. The pink, slightly raised macules, usually seen initially on the trunk and proximal ends of the extremities and never on the face, fuse centrally and coalesce to form a serpiginous pattern. The lesions may disappear after a few hours or may reappear intermittently over the course of weeks, especially after a warm shower or bath.

Subcutaneous nodules, usually associated with severe carditis, also occur uncommonly (<10% of patients). They are firm and painless and are found over bony surfaces or prominences and over tendons. ARF is not likely to be diagnosed on the basis of the latter two major criteria without another major criterion.

Sydenham chorea, or St. Vitus dance, may be the only manifestation of rheumatic fever, or it may be associated with other disease manifestations. All laboratory parameters, including streptococcal serology, may be normal. It becomes less common as ARF becomes less severe in a population, presumably as a result of improved awareness and earlier presentation of other manifestations of ARF. Chorea is characterized by purposeless (most often bilateral, uncoordinated, involuntary) movements, mostly of the hands, feet, and face, that develop over the course of weeks and are accentuated by excitement and emotional stress. They disappear during sleep. Sensation remains intact. The speech can be explosive and indistinct, and the handwriting can be clumsy. Handwriting is a useful objective means of monitoring the course of the disease. The child has difficulty counting rapidly and holding the protruded tongue still. The fingers and wrists are hyperextended when the fingers are outstretched, and the palms usually are turned outward when the arms are held above the head. Handgrip generally is weak and may consist of spasmodic contractions followed by rapid relaxation (so-called milkmaid grip). The patient may be easily irritated and quarrelsome. Chorea typically is a delayed manifestation of rheumatic fever and may develop after other signs of the disease have subsided, commonly appearing 2 to 6 months after the streptococcal infection. Historically, observers think residual heart disease occurs less commonly when chorea is the only manifestation of rheumatic fever, but in the echocardiographic era this hypothesis may prove not to be the case. The importance of prophylaxis in those with chorea to prevent recurrent attacks and possible subsequent carditis was reaffirmed in Kuwait. Permanent serious residual neurologic deficits have not been observed. A 25-year review found the duration of chorea to be 1 to 22 weeks, with a median of 12 weeks. Rare cases may last 2 to 3 years. Recurrent attacks are common and may occur despite faithful adherence to prophylaxis with intramuscular benzathine penicillin. The neuropsychiatric sequelae of chorea were reviewed recently.

Some cases of chorea are mild or atypical and may be confused with motor tics or the involuntary jerks of Tourette syndrome. Confusion may ensue between Sydenham chorea and these conditions. The term pediatric autoimmune neuropsychiatric disorder associated with streptococcal infection (PANDAS) refers to a subgroup of children with tic or obsessive-compulsive disorder whose symptoms may develop or worsen after an episode of group A streptococcal infection. The following five criteria have been used to define the PANDAS subgroup :

• 1.
  

  The presence of a tic disorder, obsessive-compulsive disorder, or both

  
  
• 2.
  

  Prepubertal age of onset (usually between 3 and 12 years of age)

  
  
• 3.
  

  Abrupt onset of symptoms, episodic course of symptom severity, or both

  
  
• 4.
  

  Temporal association between exacerbations of symptoms and streptococcal infection (approximately 7 to 14 days)

  
  
• 5.
  

  Presence of neurologic abnormalities during periods of exacerbation of symptoms (typically adventitious movements or motoric hyperactivity)

The evidence supporting PANDAS as a distinct disease entity has been questioned. In any population with a high prevalence of ARF, clinicians should rarely (if ever) make a diagnosis of PANDAS and rather should err on the side of overdiagnosis of ARF and secondary prophylaxis. If ARF is excluded, secondary prophylaxis is not needed, but these patients should be followed carefully to ensure they do not develop carditis in the long term (see Chapter 82 ).

The average duration of an attack of ARF is approximately 3 months when unaltered by antiinflammatory therapy. Fewer than 5% of cases persist longer than 6 months with active symptoms, so-called chronic rheumatic fever.

Laboratory Findings

The degree of inflammation in patients with ARF is measured by nonspecific indicators, such as the erythrocyte sedimentation rate (ESR) and C-reactive protein (CRP). Unless the patient has taken corticosteroids or salicylates, these test results almost always are positive in patients with polyarthritis or acute carditis but are normal in patients with chorea. The magnitude of the ESR is proportional to the intensity of the inflammatory reaction but is not site specific (i.e., it can be high with polyarthritis or carditis). The ESR is typically more than 60 mm/h in ARF. The ESR may be decreased in congestive heart failure, whereas the CRP may be elevated in congestive heart failure attributable to any cause. The ESR may remain elevated for 6 weeks to 3 months in an untreated attack of ARF. Antiinflammatory agents may reduce the ESR, but it rebounds if the drugs are stopped before the rheumatic process has run its course. Although chronic elevation of the ESR (>6 months) is not understood, it is not a sufficient reason on its own to limit a patient’s activities. The CRP may reflect the patient’s rheumatic activity more precisely than may the ESR.

Chest radiographs are useful for detecting cardiomegaly, which may be caused by dilation, preexisting heart disease, or pericardial effusion. The degree of enlargement is helpful in judging severity. The electrocardiogram may show a prolonged atrioventricular conduction time, usually evidenced by a prolonged PR interval or even greater degrees of heart block. Generally, an increase in the PR interval in tracings with comparable rates is considered significant. Upper limits of normal by age are available. Atrioventricular conduction abnormalities per se bear no relation to the ultimate prognosis of patients. Changes of myocarditis and pericarditis also are seen.

Every patient with suspected rheumatic fever should be considered for an echocardiogram. The AMA’s recent revision of the Jones criteria allows echocardiographic changes without a concomitant murmur as a major criterion. The roles of two-dimensional and Doppler echocardiography in establishing the diagnosis and determining the diagnosis and prognosis of ARF are now supported by multiple clinical studies ( www.heartfoundation.org.nz ). In a prospective blinded study using febrile controls and strict color and pulsed Doppler criteria, pathologic left-sided heart regurgitation could be differentiated from physiologic regurgitation. Several centers using similar strict echocardiographic guidelines have observed subclinical carditis in ARF. Echocardiographic changes of ARF with usually mitral and/or aortic valvular regurgitation may be helpful in consolidating poly- or monoarthritis or chorea as true ARF.

In the New Zealand prospective randomized trial of prevention of first episode ARF (see earlier discussion), only three additional cases (3 of 59 [5%]) met the case definition for ARF using the New Zealand Rheumatic Fever diagnostic criteria, which allowed echocardiographic carditis to be a major or minor criterion in the presence of other major or minor criteria and evidence of preceding streptococcal infection. Every case was scrutinized by an independent group of clinicians, including a cardiologist. Monoarthritis at this time was not a major criterion in the New Zealand ARF diagnostic criteria. Australia similarly incorporated echocardiographic criteria into guidelines. Publications included in a systematic review found a prevalence of 17% of subclinical carditis in ARF.

In 2012, an international consortium published guidelines on minimal criteria for a diagnosis of RHD with echocardiography. These criteria do not address the differentiation between acute carditis and chronic RHD. The criteria for pathologic regurgitation are unlikely to be different and have been incorporated into other guidelines. Morphologic changes may be different in ARF and may require further clarification. Assessment of left ventricular size and function is an important part of the evaluation of a patient with rheumatic fever, especially if surgery is being considered.

A positive throat culture for group A β-hemolytic streptococci as evidence of a recent streptococcal infection seldom is found in an ARF case, although 50% of such patients could be carriers of the organism. In the New Zealand randomized trial, of patients with ARF who reported sore throat episodes ( n = 14), 57% (eight of 14) had a positive throat culture and appropriately increased serology. A positive culture may be helpful if it can be related to the time of the acute infection.

Corroboration of a previous streptococcal infection is better determined serologically. Rising titers are preferred to raised titers. Many streptococcal antibody tests are available. Antibody titers may be elevated in the absence of clinical or bacteriologic evidence of streptococcal pharyngitis (five of 19 ARF cases in the New Zealand program). The ASO titer is the most popular antibody test. It measures the inhibition of rabbit red blood cells by specific antibody to streptolysin O, an extracellular product of β-hemolytic streptococci that in its reduced form hemolyzes red blood cells. The “normal” level for an ASO titer usually is defined as the highest titer exceeded by only 20% of a population, but it is influenced significantly by age, geography, season, and other factors. ASO titers of 500 Todd units or greater are rare findings in normal schoolchildren and are good evidence of a recent streptococcal infection. ASO titers of less than 250 Todd units could be considered normal; titers of 250 to 320 Todd units should be considered borderline elevated. Approximately 50% of patients with ARF have ASO titers in this range, and approximately 60% have titers of 500 Todd units or greater. Conversely, ASO titers can be normal in 20% of patients with ARF.

A recent streptococcal infection is more likely to be shown if more than one antibody titer is measured (e.g., antistreptokinase and antihyaluronidase). Anti-deoxyribonuclease B is currently the most favored test because of its better reproducibility.

The onset of clinical ARF usually coincides with the peak of the streptococcal antibody response. It may stay elevated for many weeks. The absence of an elevated antistreptococcal titer, if three different antibodies are measured, means, however, that the clinician can be 95% certain that the patient has not had a streptococcal infection within the recent past. In patients with pure chorea or indolent carditis, antibody levels may have declined to normal because of the length of the latent period between the development of streptococcal infection and the manifestation of this symptom.

The synovial fluid in joints affected by ARF contains 10,000 to 100,000 white blood cells/mm ³ , which are mostly neutrophils. The protein concentration is approximately 4 g/dL, glucose levels are normal, and a good mucin clot is present. A more recent report in consecutive patients with septic arthritis ( n = 111) and ARF ( n = 119) found a significantly lower synovial fluid white cell count (WCC) in the latter group of patients (median 20.2 vs. 102.5/mm ³ ; P = .004); however, there was no significant difference in the percentage of polymorphonucleocytes (median 63.5% vs. 79.0% of synovial fluid WCC; P = .761) between the two groups.