The incidence of Perthes disease varies markedly both between countries and within countries down to a local level. The disease is more common in white than in Asian or black African children. The disease is associated with deprivation; with a steep disease gradient across social class groups. This epidemiology alongside the lack of concordance in twins suggests a strong environmental etiology, with little evidence to suggest a genetic predisposition. Children are frequently short, with a growth pattern described as “rostral-sparing”. A propensity to associated congenital anomalies suggests an intrauterine cause.
Perthes disease occurs in around 5 boys for every affected girl. The disease affects children between 2 and 14 years old with the peak age of onset, among white children, being 5 years old. The age of onset has been demonstrated to be later among Indian children. Approximately 15% of cases are bilateral.
There is a marked geographic variation in the frequency of Perthes disease. Studies of incidence are difficult to directly compare, as inappropriate population denominators have sometimes been used. Table 1 shows those studies where the annual incidence rates have been calculated using a denominator of 0- to 14-year-old child years.
|Year||Study||Location||Annual Incidence Figures per 100,000 Population (0–14 y old)|
|1966||Molloy and MacMahon||Massachusetts, USA||—||—||5.7|
|1972||Gray et al||British Columbia, Canada||8.4||1.6||5.1|
|1978||Barker et al||Wessex, UK||8.7||2.0||5.5|
|1978||Barker et al||Merseyside, UK||16.9||5.0||11.1|
|1978||Barker et al||Trent, UK||12.0||3.0||7.6|
|1982||Purry||Eastern Cape, South Africa (Blacks)||—||—||0.5|
|1982||Purry||Eastern Cape, South Africa (Mixed race)||—||—||1.7|
|1982||Purry||Eastern Cape, South Africa (Whites)||—||—||10.8|
|1983||Hall et al||Liverpool, UK||25.8||4.9||15.6|
|1988||Joseph et al||Vellore, Eastern India||—||—||0.4|
|1988||Joseph et al||Udupi, Western India||—||—||4.4|
|1989||Hall and Barker||Yorkshire, UK||—||—||6.1|
|1992||Moberg and Rehnberg||Uppsala, Sweden||8.5||2.1||6.3|
|2000||Kealey et al||Northern Ireland||—||—||11.6|
|2001||Margetts et al||Liverpool, UK||—||—||8.7|
|2005||Rowe et al||Korea||6.1||1.5||3.8|
|2005||Pillai et al||Southwest Scotland||—||—||15.4|
|2006||Kim et al||Japan||1.5||0.2||0.9|
|2006||Wiig et al||Norway||—||—||9.2|
There is a significant geographic variation in disease incidence between countries, within countries, and even between small local areas. Equatorial regions have a low incidence of disease while Northern Europe has the highest documented incidence. This divergence may at least partly be explained by race, because evidence from the Eastern Cape of South Africa shows that black African children living alongside African whites have a substantially lower disease incidence, and people of mixed race an intermediate level (black: 0.45; mixed race: 1.7; white: 10.8 per 100,000 children 0–14 years old). Investigators from predominantly white countries, with significant black populations, have similarly identified few cases of Perthes disease among black children, New York, USA: 14 black children among 358 Perthes disease cases ; Connecticut, USA: 2 black children among 203 cases ; Massachusetts, USA: 1 black child amongst 86 cases. Population studies from Asian countries have found an incidence between that of black and white individuals, with incidence rates of 0.93 per 100,000 children 0 to 14 years old from a population study of Japan, and 3.85 per 100,000 children 0 to 14 years old in Korea (see Table 1 ).
There is data to show regional disease distributions within three countries, namely England, India, and Norway. All show a variation in incidence of at least twofold. In England the southern region of Wessex had half the incidence of Merseyside in the North West, with Trent in the Midlands having an intermediate value (11.1 vs 7.6 vs 5.5 cases per 100,000 children 0–14 years old per annum). In India, incidence in Vellore (south-east) was 0.4 per 100,000 per annum for 0- to 14-year-olds whereas in Manipal (south-west) it was 4.4. In Norway, an incidence of 3.6 in Northern Finmark county contrasted with 16.7 per 100,000 per annum in Western Sogn og Fjordane.
Incidence even varies notably within small areas. This divergence was first demonstrated within Merseyside (United Kingdom), with a very high incidence in inner-city Liverpool prompting investigators to examine the relationship between deprivation and disease. A steep social class gradient of disease from 4 per 100,000 per annum in Social Class 1 (highest social standing) to 31.7 per 100,000 per annum in Social Class 5 (lowest social standing) was found. Similar social class and deprivation trends have now been reported by several other studies, although this association is not universally demonstrated.
Few studies have sought to measure the influence of time on disease incidence. Between 1924 and 1960, Peic reviewed outpatient data from an orthopedic clinic in Dortmund, Germany. He noted that the incidence of cases tended to increase 5 years after a period of economic recession. However, his method of case ascertainment and definition of source population raises doubts as to the robustness of this association. A Scottish study similarly noted an association between “gross value added” (a measure of regional wealth) and the incidence of Perthes disease, though the period of observation was just 10 years and case numbers were small. The only area to record trends in disease incidence over a prolonged period is Merseyside. The disease incidence has been measured from 1976, with the last published data considering the period up to 1995. The incidence within Liverpool was found to have halved between the 1976 and 1995, with rates falling from 16.9 to 8.7 cases per 100,000 per annum. This time period was associated with a progressive increase in the standard of living within the city, which may account for the decline in incidence.
The pattern of variation in incidence by social class, over small geographic areas, and with time all indicate a major environmental influence in the cause of the disease. This influence appears to act early in childhood or in prenatal life. Family studies show that in large case series disease occurs in approximately 0.5% to 0.8% of the parents and 2% to 4% of siblings of affected children. Twin pairs within such case series have identified 10 dizygous pairs and 5 monozygous pairs, each with a single concordant pair. Given this low concordance in twins, it seems likely that familial aggregation is attributable to a shared environment rather than a genetic susceptibility.
Analysis of the age distribution of onset of cases shows that it conforms to a log-normal distribution. This pattern is typical of incubation periods and is consistent with a single exposure acting at a critical period in the hip development, either prenatally or in the first 2 years of life.