We have investigated the annual incidence of Perthes’ disease in Dumfries and Galloway (Southwest Scotland), in relation to the population density and socio-economic status. The incidence of Perthes’ disease in rural Scotland is comparable with that in urban areas (15.4 per 100 000). There was a direct association between the incidence of Perthes’ disease and deprivation scores, with the highest incidence in the most deprived areas. A higher incidence of Perthes’ disease was noted in areas with a lower population density compared with those with a higher density. We found no correlation between population density and deprivation scores for individual electoral wards within the region.
The incidence of Perthes’ disease shows considerable geographical variation both between and within countries. Epidemiological studies have suggested that it is an urban phenomenon associated with overcrowding, poor living standards and social deprivation.1–4
We have assessed the mean annual incidence of Perthes’ disease in rural Southwest Scotland in children aged 0 to 14 years and its relation to population density and socio-economic status.
Patients, Methods and Observations
Dumfries and Galloway in Southwest Scotland is a predominantly rural area. It is the third largest region in Scotland, covering 2380 square miles with a population of 147 765. The largest town is Dumfries with a population of 32 136, followed by Stranraer with 11 348 and Annan with 8930. All other villages and towns have a population of 4000 or less. There are approximately 60 people per square mile, which is significantly lower than the average of 168 people per square mile in Scotland and reflects the sparsely populated nature of the region.5 Dumfries and Galloway has a low wage economy, 13% lower than the Scottish mean and 20% lower than the national mean in the UK.5,6 The population has remained almost static with a change of only 0.03% between the censuses in 1991 and 2001.5,7–9 The orthopaedic services for the region are centralised in Dumfries.
Patients with Perthes’ disease were identified by a retrospective review of case notes over a period of ten years from 1992 to 2002. In each case, the diagnosis was made by a consultant orthopaedic surgeon based on the history, clinical examination and radiographic findings. Dumfries and Galloway is divided into 47 individual electoral wards. The cases were allocated to their respective wards using unit postcode addresses. Patients were also contacted to confirm their place of residence when the diagnosis of Perthes’ disease was made as they might have subsequently moved.
Population data were obtained from the General Register for Scotland census of 2001.7–9 For the purpose of statistical analysis the electoral wards were grouped according to their population density. The geography of Dumfries and Galloway is such that some electoral wards have an extremely low population density, while others are relatively high. Grouping the density figures into 20% bands, the electoral wards were divided into density quintiles with 1 being of the highest density and 5 the lowest (Fig. 1⇓). The actual band widths were selected using MapInfo professional 7.5 as being the most appropriate quintile boundaries.
The socio-economic status of individual electoral wards was assessed using the Scottish Indices for Multiple Deprivation (SIMD 2003).10–13 The SIMD measures deprivation in five separate domains: income; employment; health deprivation and disability; education, skills and training; and geographical access to services. Each domain score is individually weighted to generate a composite score. There are 1222 electoral wards in Scotland. The most deprived ward is given a rank of 1 and the least deprived a rank of 1222. For the purpose of analysis, electoral wards in Dumfries and Galloway were ranked in order for all scores for Scotland. They were then grouped into decile ranks in order of deprivation. Of all 1222 wards, the top 122 highest scoring wards were given a rank of 1. The band width was set to 122 ward intervals to create ten equal bands (Fig. 2⇓).
The mean annual incidence of Perthes’ disease for Dumfries and Galloway as a whole and for individual electoral wards was calculated. The relationship between the incidence of Perthes’ disease, the population density and the SIMD scores was analysed.
Between 1992 and 2002, 40 patients were diagnosed with Perthes’ disease. There were 31 boys and nine girls (male to female ratio of 3.4:1). The mean age at diagnosis was 6.5 years. A history of antecedent trauma was noted in 12.5%. There was bilateral involvement of the hip in 7.7%. There were three patients who had either a parent or a sibling affected with Perthes’ disease. The at risk population (age group 0 to 14 years) was 25 991, representing 17.6% of the total population.
The mean annual incidence of Perthes’ disease for Dumfries and Galloway was 15.39 per 100 000 children aged 0 to 14 years.
Of the 47 electoral wards 22 had no cases of Perthes’ disease. Table I⇓ shows the mean annual incidence for Perthes’ disease for electoral wards grouped according to population density. The incidence of Perthes’ disease for wards grouped according to their deprivation scores is shown in Table II⇓. Even without further analysis, the data indicates a gradient showing a higher incidence of Perthes’ disease in areas of low SIMD scores and lower population density.
The Spearman rank test was first used to examine the relationship between the SIMD scores and the mean annual incidence, and between the population density scores and the mean annual incidence. The Spearman coefficient revealed a correlation between both the SIMD and the annual incidence (r = −0.738), as well as between the population density and the annual incidence (r = −0.900). In both cases, the correlation was found to be statistically significant (p = 0.037).
The chi-squared test of association was applied to identify the nature of statistical significance between the variables. For the purpose of analysis, this involved bracketing the cases into two separate groups. Wards with SIMD scores of 1 to 5 were designated as deprived and those with scores of 6 to 10 as advantaged. The chi-squared analysis revealed that the deprived areas had significantly more cases of Perthes’ compared with the advantaged areas (chi-squared = 7.763; p ≤ 0.01). We then grouped wards with scores of 1 to 3 as high density areas, and wards with scores of 4 to 5 as low density areas. This suggested that areas with a lower population density had significantly more cases of Perthes’ compared with the higher density group (chi-squared = 9.914; p ≤ 0.01).
The Spearman correlation coefficient was also used to examine the relationship between the deprivation scores and population density for all 47 individual wards in Dumfries and Galloway. There was no statistically significant relationship between the two variables (r = 0.027, p = 0.855). Since the level and nature of deprivation may vary within areas of similar population density, a two-way cross-tabulation was generated based on the SIMD score and the population density quintile with the mean annual incidence. This indicated that the incidence increased with deprivation in all of the five groups of population density. The maximum incidence of Perthes’ disease was seen in those electoral wards with relatively higher deprivation and lower population density.
The mean annual incidence of Perthes’ disease in Southwest Scotland (15.39 per 100 000 children aged 0 to 14 years) is comparable with that reported by Hall et al3 in urban Liverpool (15.6 per 100 000). Our results however, are higher than the incidence reported from other regions of Britain such as Wessex (5.5 per 100 000), Yorkshire (6.1 per 100 000), Trent (7.6 per 100 000), Mersey (11.1 per 100 000) and Northern Ireland (11.6 per 100 000).2,14,15 The gender ratio and the age at diagnosis compare well with previous reports on Scottish populations by Wynne-Davies and Gormley16 and Cameron and Izatt.17 Table III⇓ shows our data on the incidence of Perthes’ disease in Southwest Scotland in relation to those reported from other parts of the world.
It is generally accepted that the incidence of Perthes’ disease varies with socio-economic factors.1,3,4,15,16 Deprivation studies in rural areas are more difficult than in urban areas, as there may be both deprived and non-deprived areas side by side in the same unit of analysis such as the postcode. This may have a ‘smoothing effect’, which eliminates the area being designated as very deprived or very affluent, but urban areas tend to have larger neighbourhoods that share their socio-economic status more uniformly. The SIMD 2003 takes into account a wider variety of factors to measure deprivation compared with techniques such as the Carstairs or Townsend index which have been used previously.10,11 It is therefore a more appropriate measure of deprivation in rural settings. Data from this study show an increasing incidence of Perthes’ disease in areas of higher levels of deprivation, as measured by the SIMD scores. In Southwest Scotland, there were significantly more cases of Perthes’ disease in children from lower income families and more deprived households.
Current epidemiological studies from Britain, support a higher urban than rural incidence of Perthes’ disease.2–4 Hall et al3 reported a correlation with population density and overcrowding in their study in Liverpool. They also observed a very low incidence of Perthes’ disease in the small towns and rural areas of Yorkshire.14 Barker et al2 also suggest an increased association with urban areas from their comparative study on Perthes’ disease in Wessex, Trent and Mersey. Purry1 made a similar observation for urban pre-disposition in South Africa but Kealey et al15 found no urban-rural divide in Northern Ireland. Our results indicate that within Southwest Scotland, the incidence of Perthes’ disease is higher in areas of relatively lower population density, compared with those areas with a higher density. This suggests a paradoxically higher rural than urban incidence of the disease in the region.
We agree with Kealey et al,15 that it is a misconception that deprivation and poverty are urban phenomena, concentrated in inner city areas and large public housing estates. This explains the previous suggestions1–4 that the incidence of Perthes’ disease is related to the effects of overcrowding, urbanisation and population density. Although underprivileged neigbourhoods are the most affected in predominantly urban settings such as Liverpool, in regions like rural Scotland, the distribution of deprivation can be quite different. A substantial proportion of low-income households in such areas are based in the more remote rural locations. These sparsely-populated areas may have low-wage economies, higher unemployment and poor access to services. They are as deprived in health terms as their densely-populated inner-city counterparts. Areas of Dumfries and Galloway, with SIMD scores 1 to 2 share the same deprivation category as inner city Glasgow.
We feel that in areas like Southwest Scotland where there are no large cities and people are living in small towns and villages are areas of relative urbanisation, they may have better job opportunities, better access to services and may be relatively better off than those living in the more remote parts of the country. This could account for the paradoxically higher incidence of Perthes’ disease in areas of low population density indicated by this study, as they would be the most deprived. Hence, parts of Scotland and England with a similar population profile to Dumfries and Galloway, may exhibit a higher incidence of Perthes’ disease in sparsely-populated rural areas. Further investigation would be required to validate such a statement. A predominantly rural incidence of Perthes’ disease has been reported from South India by Joseph et al,22 but the study does not elaborate on population or socio-economic data, which makes comparison difficult.
The epidemiology of Perthes’ disease, and in particular it’s wide geographic variations in incidence, suggests major environmental influences in its aetiology. It has been suggested that socio-economic status affects the nutritional status of the child and also influences maternal behavior.3,15,23 Studies on the incidence of Perthes’ disease in Dortmund from 1924 to 1960 indicated wide variations in the incidence, which correlated with fluctuations in the German economy.24 The term ‘gross value added’ is now used to measure the total wealth created in a regional economy.5 Measured at basic prices it is similar to what was formerly known as the gross domestic product at factor cost. Examining the relationship between this measurement over the same period and the annual incidence of Perthes’ disease in Dumfries and Galloway from 1992 to 2002 we noticed an increased incidence of Perthes’ disease corresponding to the period 1996 to 1998 which saw a sharp decline in the local economy.5,25 The maximum number of cases was recorded in 1997 (17.5%). The relatively short study period prevents us from drawing any further inference on this relationship. The mean annual incidence was calculated by identifying patients through a retrospective review of hospital records. It has been argued that patients from rural areas are less likely to seek medical advice compared with those from urban areas. Although this may be true for minor ailments, it is unlikely that a persistent painful hip would be ignored by a patient or by a general practitioner without seeking specialist advice.2 We accept that during the study period it is possible that a few patients could have remained undiagnosed and that certain patients from border areas could have been referred to another hospital. Although the number of these lost patients would be very small, if any, the term minimal average annual incidence as suggested by Gray et al18 may be more accurate.
The minimal mean annual incidence of Perthes’ disease in predominantly rural Southwest Scotland is similar to that reported from regions of higher urbanisation. The incidence increases with deprivation and poor living standards. Areas of lower population density showed a paradoxically higher incidence of Perthes’ disease. Sparsely populated, remote rural areas can be as disadvantaged socio-economically as crowded areas in the inner city.
The authors would like to acknowledge Mr Andrew McAuley, Statistician, Department of Clinical Effectiveness, NHS Lanarkshire for help with data analysis and Ms Ananda Allan, Senior Health Intelligence Analyst, Department of Public Health and Strategic Planning, NHS Dumfries for the population data and cartography.
No benefits in any form have been received or will be received from a commercial party related directly or indirectly to the subject of this article.
- Received May 16, 2005.
- Accepted June 7, 2005.
- © 2005 British Editorial Society of Bone and Joint Surgery