Acute rheumatic fever (ARF) and its sequel, rheumatic heart disease (RHD) are the most important causes of morbidity and mortality resulting from infection with the bacterium group A Streptococcus.1 It has been estimated that 95% of the nearly 20 million cases of RHD in the world, and up to ∼500 000 deaths each year due to ARF and RHD, occur in developing countries.1 Unlike ischaemic heart disease, RHD causes severe disability and premature death, mainly in adolescents and young adults during their years of greatest social and economic productivity.2 3

There are renewed efforts by the World Heart Federation to combat ARF and RHD through health promotion, improved surveillance and the application of effective prevention strategies.2 4 A key requirement for success of this global effort is the availability of high-quality information on the incidence of ARF and prevalence of RHD. A systematic review of RHD prevalence has confirmed the high levels of RHD in the Pacific region, the Indian subcontinent, Asia, sub-Saharan Africa, Mediterranean, Latin American and the Caribbean islands.5 Indeed, the continuing high burden of RHD in Africa and Asia has been highlighted by an echocardiographic screening study which found that 2–3% of school-age children in Cambodia and Mozambique have RHD.6 Data showing that rheumatic valve disease accounts for an overwhelming 89.5% of cardiac disease associated with pregnancy and over a third of maternal deaths in Egypt7 further depict long-term sequelae of ARF and, for women first diagnosed during such presentations, reflect the insufficient detection rates of ARF.

There is, to the best of our knowledge, no systematic summary of prospective population-based studies of the incidence and temporal trends of ARF in the world. A recent analysis of the incidence of ARF included all available retrospective and prospective studies of the subject, in an effort to estimate the annual number of ARF cases in different regions of the world.1 We have conducted a systematic review of prospective population-based incidence studies of ARF to assess the burden and temporal trend of ARF in developed and developing countries.

METHODS

We included prospective, population-based studies involving men and women regardless of age which used Duckett–Jones criteria to establish the diagnosis, and reported the incidence of first attack of ARF. Medline and Old Medline (1944 to 2005), Embase (1944–2005) and 35 other health-related bibliographic databases were searched for studies conducted from 1944 onwards, using predefined search terms. Reference lists of included articles were scanned. No language restrictions were used for conducting the search but only studies published in English and French were considered for the review.

Titles and abstracts of all citations identified were assessed using the predefined eligibility criteria. For studies that initially seemed to meet the inclusion criteria but were subsequently excluded, reasons for this decision were documented (fig 1). All studies selected for inclusion by one author (KBT) were checked independently by the other two authors (JAV and BMM), with disagreements being resolved by consensus. A flow chart of the study selection process was prepared according to the QUOROM statement (fig 1).8

• Download figure
• Open in new tab
• Download powerpoint

Figure 1 Flow chart illustrating search results. Reasons for exclusion: I, inappropriate measure of occurrence and insufficient raw data for population-based incidence to be calculated; II, inappropriate study design; III, insufficient information on use of use of Duckett–Jones criteria; IV, not an original study (includes reviews, editorials and letters); V, article in language other than English or French. Further attempts to obtain all these articles/abstracts failed.

We extracted the following data from each study: location, duration, source population size, target age, interventions used, source of cases and methods of case ascertainment (table 1). Each included study was appraised for quality using a 15-point checklist based on existing guidelines.9^–11 One or zero was allocated depending on whether a particular criterion had been met, giving a maximum total quality score of 15. We computed the overall mean annual incidence rate for each study (which was obtained by dividing the total number of incident cases over the entire study period by the estimated annual study population at risk, then further dividing by the total number of years of study) and the annual specific incidence rate for studies that reported an annual rate of ARF cases in order to assess temporal trends.

RESULTS

Description of included studies

A total of 449 potentially relevant citations were identified, yielding 10 studies that were eligible for inclusion in the review (fig 1, table 1) (These citations are available from the authors on request.) These studies had been conducted in 10 countries: the Americas (USA,12 Chile,13 and Martinique and Gouadaloupe14); Europe (Czechoslovakia,15 and Sweden16); the Middle East (Kuwait,17 18 and Iran19); Asia (India20); and Australasia (New Zealand21), and were published between 1967 and 1996. The study duration ranged from 3 to 12 years, with a mean (SD) of 6 (3.6) years.

Studies used either the modified13^–15 19 or revised versions11 17 18 20 21 of the Duckett–Jones diagnostic criteria. In total, five studies used active case detection for the assessment of the incidence of ARF,12 14 18 20 21 with health education of health professionals being the method most commonly used to promote the identification of ARF cases.12 14 21 The remaining five studies used passive case detection,13 15^–17 19both European studies falling within this group.15 16

Methodological quality of included studies

Table 2 shows the results of the quality appraisal for each study. The mean (SD) score was 8 (2) out of 15 and the score for four studies fell below the group mean.13 15^–17 Thus 40% of the studies were of poor methodological quality, and none of the studies scored the full marks of 15 for high quality. Overall, these studies were of low or moderate quality.

Incidence data

Table 3 shows the incidence rates reported in each study. The overall mean incidence rate of first attack of ARF per annum for each study, calculated over the entire study period for each study, ranged from 5 to 51/100 000 (mean for all studies together being 19 per 100 000; 95% CI 9 to 30/100 000). While the highest overall mean annual incidence emerged from a single Indian study (51/100 000 per year),20 a subgroup analysis of the Maori community in New Zealand showed an incidence of >80/100 000 per year.21

The first of two Kuwaiti studies was one of the few studies demonstrating incident ARF in children under 5 years of age.17 In the French Caribbean Islands study, only the incidence over the past 10 years of the 11-year study was taken into consideration because of the very high initial reporting rate following introduction of the active surveillance programme.14 The second Kuwaiti study yielded by far the highest raw number of incident cases of any other included study (ie, between 79 and 114 new cases of first attack ARF per year of study)18; despite it having a population size similar to three studies conducted in Czechoslovakia,15 the Caribbean14 and the USA,12 and considerably lower than Chile.13 The Indian study showed by far the highest overall mean annual incidence rate of 51/100 000.20

Factors influencing the incidence of ARF

We assessed the impact of surveillance method, geography, age and sex on the reported incidence of ARF. As expected, studies using active surveillance methods exhibited relatively higher incidence rates, which ranged from 8 to 51/100 000 (median 22; interquartile range 10–23/100 000), compared with those employing passive surveillance with lower incidence rates of 5–35/100 000 (median 16; interquartile range 5–18/100 000). Crude incidence rates and age-adjusted rates derived from studies from the USA,12 Chile,13 Martinique and Gouadaloupe,14 and Sweden16 showed a low ARF incidence of ⩽10/100 000 per year; while those from Eastern Europe,15 Middle East,17^–19 Asia,20 and Australasia21 had a higher incidence >10/100 000 per year. Six of the 10 included studies assessed the incidence of ARF in the first three decades of life,16^–18,20,21 and the remainder included all age groups.12^–14 19 In two studies, the incidence per year of age could be obtained,15 16 and in one study the childhood and adult incidence measures could be analysed separately.15 The single study providing data on the incidence of the first attack of ARF for male and female subjects separately, found no differences between the sexes.20 Most other studies (8/10) reported male to female ratios for all episodes of ARF (ie, first and recurrent attacks combined), with no consistent pattern evident. The remaining study had no information on the sex distribution of cases.15

Trends over time

The annual specific incidence rate in three (ie, Martinique and Gouadaloupe,14 New Zealand,21 and Kuwait18) of the five studies documenting yearly incident cases showed a falling incidence of ARF over time (table 4 and fig 2). However, the remaining two studies (ie, USA and Iran) were conducted over 3 years and showed no clear direction in trend.12 19 There is, however, a discernable trend towards falling rates of ARF in studies that have tracked its incidence at a population level.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2 Annual specific incidence rate (temporal trend) of first attack of acute rheumatic fever.

DISCUSSION

To the best of our knowledge, we report the first systematic review of prospective population-based studies of the incidence of first attack of ARF from across the world. The overall mean annual incidence rates were lowest in American and Western European countries (⩽10/100 000),12^–14 16 with relatively high rates reported in Eastern Europe, Asia, Australasia and the Middle East (>10/100 000).15 17^–21 No studies that were available from Africa, a continent which is considered to harbour nearly 50% of the cases of RHD in children under the age of 15 years.1 While the highest overall mean annual incidence emerged from a single Indian study (51/100 000 per year), a subgroup analysis of the Maori community in New Zealand showed an incidence rate >80/100 000 per year. The regional variation in incidence of ARF was associated with a fall in the incidence of ARF over time in the few studies that have monitored the temporal trend of the disease. Declining trends in our studies may have been due to a combination of improved awareness of the disease among the community and healthcare workers, leading to improved diagnosis and treatment of streptococcal pharyngitis and greater use of secondary prevention of ARF.14 20 22

Fifty per cent of studies included in this review used the earlier modified version of the Duckett–Jones diagnostic criteria rather than the more up-to-date revised diagnostic criteria for ARF; the latter criteria require the evidence of recent streptococcal infection for diagnosis, while the former merely considered it a minor manifestation.23 The requirement for evidence of recent streptococcal infection in the revised criteria is considered by some to render the diagnostic criteria less sensitive in cases presenting after evidence of antecedent streptococcal infection has subsided, or where other manifestations are delayed.23 Apart from the USA, most of the countries using the less up-to-date criteria were middle-income countries. The use of different diagnostic criteria over time clearly represents an important limitation in summarising studies in this field.

This systematic review has disclosed a paucity of high-quality, population-based prospective studies of the incidence of ARF in the world. We found population-based studies conducted in only 10 out of over 190 countries of the world; the last of which ended over a decade ago. No prospective population-based incidence studies of ARF were found in Africa, the second hardest hit world wide by disability-adjusted life years lost to RHD. A similar experience was found by Steer and colleagues in their systematic review of the prevalence of RHD.5 There is a pressing need for good-quality information on the incidence of ARF to provide a basis for health policy, resource allocation and clinical management, particularly in Africa.

Previous reports on the incidence of rheumatic fever have reported median incidences as high as 374 per 100 000 of the population among the aboriginal communities of Australia and New Zealand.1 We have restricted this systematic review to population-based studies that report on the first attack of ARF. The exclusion of information on recurrent attacks of ARF provides the minimum estimate of incidence of ARF in affected regions. Our results, which show high rates of first attack of ARF in Eastern Europe, the Middle East and Asia, Australasia, are consistent with the known trends in the prevalence of RHD in these regions.1 6

The World Health Organization has emphasised the collection of epidemiological data as a crucial step in planning and implementing national programmes for the prevention and control of RF and RHD.23 These epidemiological data will allow policymakers and practitioners to identify groups or locations that are most affected by ARF and RHD in order to direct and concentrate control efforts appropriately. The current state of ARF and RHD surveillance programmes in countries most affected by the disease is deficient.24 25 This systematic review of ARF incidence, and the aforementioned systematic review of RHD prevalence studies, highlight the lack of quality data as well as the absence of reports on rheumatic fever outbreaks from developing countries.5 The scarcity of reliable surveillance data is one of many barriers preventing developing countries from mounting an appropriate and effective response to combat ARF and RHD.4

Implications for policy and practice

An immediate priority for the global effort to prevent ARF and RHD is to establish comprehensive and sustainable surveillance systems for the disease.4 The creation and maintenance of registers or databases of ARF and RHD cases is a proven strategy for the secondary prevention of the disease.26 The registers, which are used as tools for case management to track cases and to ensure that patients are receiving appropriate prevention and treatment, may be used as a basis for prospective incidence studies of ARF at a population level. The establishment of registers is a minimum requirement that can be achieved in almost any setting where the will to establish ARF and RHD control programme exists.