Background Nonmelanoma skin cancer (NMSC) is the most common cancer affecting white-skinned individuals and the incidence is increasing worldwide.
Objectives This systematic review brings together 75 studies conducted over the past half century to look at geographical variations and trends worldwide in NMSC, and specifically incidence data are compared with recent U.K. cancer registry data.
Methods Following the development of a comprehensive search strategy, an assessment tool was adapted to look at the methodological quality of the eligible studies.
Results Most of the studies focused on white populations in Europe, the U.S.A. and Australia; however, limited data were available for other skin types in regions such as Africa. Worldwide the incidence for NMSC varies widely with the highest rates in Australia [> 1000/100 000 person-years for basal cell carcinoma (BCC)] and the lowest rates in parts of Africa (< 1/100 000 person-years for BCC). The average incidence rates in England were 76·21/100 000 person-years and 22·65/100 000 person-years for BCC and squamous cell carcinoma (SCC), respectively, with highest rates in the South-West of England (121·29/100 000 person-years for BCC and 33·02/100 000 person-years for SCC) and lowest rates by far in London (0·24/100 000 person-years for BCC and 14·98/100 000 person-years for SCC). The incidence rates in the U.K. appear to be increasing at a greater rate when compared with the rest of Europe.
Conclusions NMSC is an increasing problem for health care services worldwide. This review highlights a requirement for prevention studies in this area and the issues surrounding incomplete NMSC registration. Registration standards of NMSC should be improved to the level of other invasive disease.
Basal cell carcinoma (BCC) and squamous cell carcinoma (SCC) are the two most common subtypes of nonmelanoma skin cancer (NMSC). Although they share many similarities, they have different incidence rates and important aetiological differences. BCC is the most common cancer in many countries worldwide1 and although the mortality rate is exceptionally low,2 NMSC represents both a significant economic burden to health services and can cause significant morbidity especially as most NMSCs occur on highly visible areas such as the head and neck and face.
In comparison with other malignancies, little is known about the incidence. The rate is increasing in many countries3–6 and although the reason for this is unclear, it may be linked to increased sun-seeking behaviours and improved registration procedures. It is important to realize that NMSC incidence is based on a lifetime of exposure to risk factors and as such the older population currently being diagnosed used very little sun protection and received very little education regarding the effects of ultraviolet radiation (UVR). A primary prevention study in Australia ran for 20 years before beneficial effects were observed.6,7
The main difficulty in measuring the incidence comes from poor registration practice in the majority of countries. Usually only the first case of NMSC in a patient is registered, any subsequent tumours are not included and multiple tumours are not differentiated. The majority of data on NMSC incidence comes from local studies of incidence in a certain geographical location.
The recently updated National Institute for Health and Clinical Excellence (NICE) guidelines state a need to establish the true nature of the epidemiology of BCC.8 This systematic review aims to summarize all the existing literature regarding the incidence of not only BCC but also SCC worldwide and for the U.K. to make comparisons between that literature and incidence data from U.K. cancer registries.
Materials and methods
Inclusion and exclusion criteria
Any epidemiological study that assessed the incidence rate of BCC, SCC or both subtypes combined was included. In order to improve the chance a study was sufficiently rigorous to deal accurately with the complexities of NMSC epidemiology, the studies needed to examine the incidence of NMSC exclusively, as opposed to all malignancies. Incidence rates had to be stated or sufficient data provided to allow the calculation of incidence rates. If two papers reported results in the same population at the same point in time (such as provisional rates) the data were compared and counted only once. Studies which focused solely on genetic syndromes were excluded.
A comprehensive search strategy was developed, and EMBASE (from 1980) and Medline (from 1950) were searched to March 2011 using the following search string to identify relevant papers: carcinoma, basal cell, basal cell nevus syndrom/, carcinoma, squamous cell non melanoma or nonmelanoma or non-melanoma, skin, cutaneous, incidence, epidemiology. No restrictions on language were imposed during the search strategy. A second author independently reviewed titles and abstracts and disagreements were resolved through discussion with a third author.
Two of the authors (A.L. and J.L.-B.) independently extracted the following data for each study: author, country and region, study period, reported rates of NMSC, BCC or SCC, source of the study’s population (e.g. cancer registry, histology laboratory etc.), whether histological confirmation was obtained, unit of analysis (i.e. number of patients or number of tumours) and method of standardization for incidence rates. Disagreements were resolved through discussions with a third author (F.B.-H.). Values for approximate latitude were obtained using Google Earth.9
The South West Cancer Intelligence Service provided standardized data from all cancer registries in England. Northern Ireland and Scotland provided standardized data. No data were available from the Welsh cancer registry. The data for England were entered into ArcGIS for Windows (Esri, Redlands, CA, U.S.A.) and displayed on a map of England.
The included studies were assessed for methodological quality using a modified methodological quality-assessment tool which was developed and based on existing assessment tools.10,11 The criteria listed in Table 1 were assessed, with each point equally weighted.
(a) Did the study directly sample the population, as opposed to using cancer registries?
Description of methods
(b) Is the method of counting tumours stated?
• Number of tumours or number of patients
(c) Is there histological verification?
Reporting of incidence rates
(d) Were separate rates reported for BCC and SCC?
(e) Are gender-specific incidence rates reported?
(f) Are age-specific incidence rates reported?
(g) Were the rates age standardized?
(h) Are confidence intervals used?
(i) Is there information about the ethnicity or skin type of the population?
• Must state actual proportions
(j) Were the data standardized to a major population?
• U.S. population, European population or world population
Incidence rates are presented as rates per 100 000 person-years. The incidence rates from studies which standardized to the same population were compared directly. Graphs were constructed using data from male incidence rates because these maximized the number of comparable figures, as few studies reported overall standardized incidence rates for male and females combined.
The search strategy identified 3083 papers of which 275 were deemed potentially eligible based on abstracts, and 75 were finally included in the review (Fig. 1).3–7,12–81 Thirty-one studies were excluded as they did not focus exclusively on NMSC. These papers were all cancer incidence reviews and in the majority of cases did not report NMSC incidence at all. Fifteen papers were excluded as no translation was available.
The included papers provide data on 38 different countries worldwide over the period 1955–2007. Twelve incidence studies were based on Australian populations, more than any other country. Most of the papers reported age-standardized incidence rates. Thirty-eight were standardized to the worldwide population, eight to the U.S. population and eight to the European population. Fifteen papers did not age-standardize incidence rates to any population.
The methodological quality of the studies varied with a range of scores between 1 and 9. Forty-eight (64%) studies were deemed of high quality (score ≥ 6); however, no study met all the criteria and only 19 provided information about the skin types or ethnicity of the study population. The majority (65%) of studies used histological verification. Fifty-three studies used records from sources such as cancer registries and hospitals with the remaining 22 studies obtaining their own data from sources such as questionnaires. Table 2 includes methodological information about each of the studies included in this review.
Table 2. Study characteristics
Direct or indirect (population)
THIN, The Health Improvement Network; GP, general practitioner.
The data provided by the cancer registries are shown in Table 3. As the incidence of BCC is much higher than SCC in the U.K., the distribution of NMSC appears the same as BCC. The incidence for BCC and SCC is shown diagrammatically in Figure 2.
Table 3. Directly standardized annual incidence rates for nonmelanoma skin cancer (NMSC), basal cell carcinoma (BCC) and squamous cell carcinoma (SCC) in the U.K. per 100 000 person-years: 2000–2006
All rates are directly standardized to the European population.
East of England
Yorkshire & the Humber
Basal cell carcinoma
U.K. registry data
The South-East regions of the U.K., including London, the South-East coast and East Anglia, have lower rates of BCC than more Northern regions of the country where the incidence is similar at around 100/100 000 person-years. There appears to be a divide between the North of England and Scotland with rates dropping from 104·12/100 000 person-years in Yorkshire to 90·4/100 000 person-years in Scotland as a whole. Scotland has a similar incidence to Northern Ireland (86·8/100 000 person-years).
Data from systematic review
U.K. One study conducted in the North Humberside region, between the years 1978 and 1991, showed an increase in BCC incidence from 38·8/100 000 person-years to 115·6/100 000 person-years.42 A study using a primary care database (The Health Improvement Network) found the European age-standardized rate of BCC across the U.K. to be 89/100 000 person-years between 1996 and 2003.15 Only two studies were found that looked at BCC in Wales and reported the incidence to be 75·1/100 000 person-years5 and 83·1/100 000 person-years,64 respectively, in 1988. Ten years later in 1998, the rate had increased to 114·2/100 000 person-years.5 A study in Scotland reported rates increasing from 35·6/100 000 person-years in 1979 to 97·5/100 000 person-years in 2003 (standardized to the European population).81
Europe Figure 3 shows data from males from 14 comparable studies assessing the incidence of BCC in Europe.5,15,17,18,25,34,41,48,50,56,57,64,67,68 All of the studies show that the rates of BCC have increased at a similar rate over the past four decades, on average increasing by 20/100 000 person-years every 15 years, a 5·5% increase per year.34 The two South Wales-based studies report a higher incidence than in other European countries.5,64 A further study conducted in the North Humberside region of the U.K.42 could not be included in this graph due to data being standardized to a different standard population (England and Wales); however, this study showed that the U.K. incidence rates could be increasing much faster than in other European countries: an increase from 38·8/100 000 person-years in 1978 to 115·6/100 000 person-years in 1991 was seen.42 Incidence rates in the Netherlands are also not shown due to differing populations where the male and female combined incidence rate was 87·5/100 000 person-years in 2003 (standardized to the European population).77 Elsewhere in Europe, Switzerland and Italy appear to have the highest incidence rates at around 70/100 000 person-years in 1995;17,48,50 for comparison, the incidence of BCC in Slovakia was only 38/100 000 person-years in 1994,57 even though Slovakia lies at a similar latitude to Switzerland and Italy. In Denmark, the male incidence rate increased from 34·2/100 000 person-years to 91·2/100 000 person-years between 1978 and 2007.79 In the same study, an even greater increase was reported in female incidence rates (27·1/100 000 person-years in 1978 to 96·6/100 000 person-years in 2007). The lowest rates were observed in Croatia at 33·6/100 000 person-years between 2003 and 2005.80
North America Manitoba (Canada) lies at a similar latitude to Europe and hence is exposed to comparable levels of UVR. Despite this, a relatively high incidence of BCC was seen both in Manitoba, where the male incidence was 93·9/100 000 person-years in 2000,4 and in Alberta at 147·0/100 000 person-years in 2006.78
Five comparable studies were found regarding BCC incidence in the U.S.A.13,21,35,40,65 Arizona and New Mexico are two adjacent states in the South-West of the U.S.A., lying at approximately 34o N latitude. Both states appear to have similar incidence rates. A rate of 935·9/100 000 person-years was reported for 1996 in Arizona.35 Two Northern states also share similar incidence rates. New Hampshire in New England lies at a latitude of 43o N. Minnesota lies more centrally but at a very similar latitude of 44o N. The two states have nearly identical BCC incidence rates at approximately 170/100 000 person-years in 1980.21,40,65 All studies were standardized to the same population and appear to show a similar rate of increase of approximately 2% per year.
Australia The search found eight studies which looked at incidence of BCC in Australia.6,7,19,32,33,52,58,63 Four studies estimated the incidence based on a national survey,6,7,32,52 two papers focused on a population in Townsville, Queensland19,58 and a further two papers followed a population in Nambour.33,63 Although all standardized to the world population, the two regional studies are not likely to represent the whole country. As the Nambour study involved screening by experienced dermatologists, these results cannot be compared directly with the national survey.
The incidence of BCC in Australia was higher than anywhere else in the world. The national surveys showed that the incidence of BCC has been increasing gradually since 1985, and appears to be reaching a plateau.6,7,32,52 The rate of 884/100 000 person-years (both sexes combined) in 2002 is 10 times the rate recorded in the U.K.15 The four studies that followed up the smaller populations show a decreasing incidence rate.19,33,58,63 In the case of the Nambour study group the male incidence rate decreased from an average incidence of 2074/100 000 person-years between 1985 and 1992 to 1813/100 000 person-years between 1997 and 2006.33,63
Squamous cell carcinoma
U.K. registry data
Squamous cell carcinoma incidence data are listed in Table 3 and displayed in Figure 2. The incidence rates of SCC vary less across the U.K. than those of BCC. The rate of 23·73/100 000 person-years in Yorkshire is representative of most of the country north of London which, like BCC, has the lowest recorded SCC rate in the U.K. at 14·98/100 000 person-years. The South-West counties again have a notably higher incidence rate at 33·02/100 000 person-years.
Data from systematic review
U.K. An increasing incidence rate of SCC from 15·9/100 000 person-years in 1978 to 28·6/100 000 person-years in 1991 was reported in the North Humberside region.42 The two Welsh studies reported rates of SCC at 19/100 000 person-years64 and 15·1/100 000 person-years5 in 1988, respectively, with only a slight increase in incidence rate being seen 10 years later (15·8/100 000 person-years in 19985). In Scotland, the incidence rate of SCC increased from 16·1/100 000 person-years in 1979 to 36·9/100 000 person-years in 2003.81
Europe For comparison, all European studies reporting male SCC incidence rates standardized to the world population are displayed in Figure 4.
Nineteen studies were identified that measured the incidence of SCC in Europe.5,17,18,25,34,41,48,50,56,57,64,67,68,74–77,79,80 All studies suggested an increasing trend, although the rate of increase varied from country to country. Like BCC, the South Wales studies show the highest incidence rates of approximately 31·7 per 100 000 person-years.64 Switzerland had the highest SCC incidence rate of all of mainland Europe, as well as showing the fastest increase from 14·2/100 000 person-years in 1978 to 28·9/100 000 person-years in 1997.48,50 Northern European countries such as Norway, Finland and Denmark reported very low rates of SCC of less than 10/100 000 person-years, and the increase in incidence over time is also much slower than in the other countries.34,56,75,79 Although Sweden has a similar geographical location to these Northern European countries, the incidence of SCC has shown a sharper increase.74 A second Swedish study between 1990 and 2005 found that the average SCC incidence rate was 34·4/100 000 person-years for males and 15·4/100 000 person-years for females.76 As for BCC, the incidence rate of SCC in Croatia was lower than elsewhere in Europe at 8·9/100 000 person-years.80
North America Five studies standardized to the U.S. population were found.13,22,35,40,65 A divide in incidence rate of SCC was seen according to latitude, although the overall trend is less clear than that seen with BCC rates. A study in Arizona based on a histologically verified cancer registry showed a gradual decline in incidence of SCC from 1985 to 1991 to a steady level of approximately 290/100 000 person-years.35 New Mexico reported an increasing incidence of SCC,13 the rate of increase being approximately the same as in the Northern state of New Hampshire.40 As with BCC, much lower incidence rates were found in the Northern than in the Southern states.13,22,35,40,65 In Alberta, Canada, rates increased from 45·0/100 000 person-years in 1988 to 60·2/100 000 person-years in 2006.78
Australia Seven comparable studies were found that stated the incidence of SCC in Australia.6,7,19,32,33,52,58 The national postal survey found an almost identical pattern in SCC6,7,32,52 as was seen with BCC, relating to a gradual increase since 1985, with the rate of increase slowing over time. In 2002, the rate of combined male and female SCC was 387/100 000 person-years.6 Much higher incidence rates of SCC were observed in the Nambour and Townsville study populations,19,33,58,63 with a world-standardized incidence rate of 1035/100 000 person-years in males living in Nambour from 1985 to 1992.33 However, as with BCC, these much smaller populations are not representative of the country as a whole.
This systematic review brings together 75 papers on the incidence of NMSC and is the largest systematic review to date on the subject.
The NICE guidelines to establish the true nature of NMSC epidemiology are complex. Cancer registries are certainly improving practices following guidance but more should be done to bring NMSC epidemiology in line with other invasive diseases. A particular area of focus should be the encouragement of general practitioners to send all excised material for histological confirmation. Incidence studies such as those compared in this review should, as a minimum, standardize all rates to a common population, use histological confirmation and provide basic information such as the unit of analysis (by tumour or by patient). This will allow much more accurate comparison at both national and international levels.
Variations in ethnicity may partly explain the BCC distribution in the U.K. There is a comparatively large nonwhite population in the South-East of England.82 Darker skin types have a much lower risk of BCC and this may be reflected by the registry data. If the registries could divide the data by county as opposed to Strategic Health Authority then the contribution of ethnicity may be clearer. The South-West of England is associated with the highest UVR exposure and also reports the highest incidence of BCC.
It is possible that the variation seen in the U.K. data is due to variation in registration practice. Poor registration is highly likely to be the cause of the extremely low rates seen in the South-East of England. Although this reduces the utility of the comparisons here, it does highlight well the issues of NMSC registration in the U.K. and gives good evidence that registration procedures must be vastly improved and standardized.
Although the different standard populations prevent direct comparisons of incidence, the rate of increase can be examined. BCC incidence is increasing by approximately 1/100 000 persons per year in mainland Europe;34,48,57 however, the incidence in the U.K. is increasing at a rate of approximately 6/100 000 persons per year.15,42 This difference in the rate of increase is not discussed in the literature and should be examined in the future.
When comparing the incidence rates in mainland Europe, all countries appear to be similar with Switzerland and Italy reporting slightly higher rates.17,50 Switzerland has the highest average altitude in Europe and as such will have higher UVR levels. UVR levels in Italy are also likely to be high due to the low latitudes and high-altitude Northern regions bordering Switzerland. Manitoba lies at a similar latitude to the U.K. and should in theory have similar levels of UVR. In 2000, the world-standardized male BCC incidence rate was 93·9/100 000 person-years4 which, although at the higher end of the European figures, fits the trend observed here.
The best evidence for increasing BCC incidence with decreasing latitude is seen in North America where the divide between the Northern and Southern states is clear. Rawashdeh and Matalka60 conducted a U.S.-standardized BCC incidence study in Jordan from 1991 to 2000 and reported incidence rates of just 8·8/100 000 person-years (both sexes combined). Jordan lies at a similar latitude to North America. This large difference in incidence rate illustrates the effect that skin type has on the geographical distribution of incidence rates worldwide. This theory is further supported by the very low incidence rates observed close to the equator in countries such as Singapore.66
Australia has by far the highest incidence of BCC in the world. Although intermittent UVR exposure is more important than total exposure in BCC aetiology,83 the very high UVR levels in Australia are likely to be raising the incidence rates to the levels observed in this review. Support for this theory can be found in the NMSC subtype proportions. The BCC : SCC ratio in Australia is approximately 5 : 2.7 In the U.K., the BCC : SCC ratio is 4 : 1.42 As Australians are exposed to long-term UVR, they are more likely to develop SCC than in other countries such as the U.K., where intermittent exposure is more common.
The distribution of SCC in the U.K. is similar, but not identical, to that of BCC, highlighting differences in the aetiology of the two tumours. London has the lowest SCC incidence in the U.K., although this is still not as low as the incidence of BCC. According to the registry data, the South-West again appears to report the highest incidence of SCC. Rates across the North of the country are generally similar, with the West Midlands reporting a slightly lower rate. The same registry also reported a lower than expected rate for BCC so it is possible that this area is more affected by under-reporting than other areas.
The division between Northern and Southern states in the U.S.A. persists in the SCC data and again is almost certainly due to differences in UVR levels. According to the authors, the gradual decrease seen in Arizona35 may not be representative of the whole state and incidence may be increasing in other regions.13 As is the case in Europe, SCC rates are increasing but not as rapidly as for BCC.
A potential area for future review is the variation in use of sun protection and sun exposure behaviour worldwide and the correlation with NMSC incidence rates over time. These factors are highly likely to contribute to the geographical variation in rates reported here.
This is the largest systematic review of incidence studies to date regarding NMSC. The size and completeness is a key strength that allows comparisons on a level not previously possible. The inclusion and exclusion criteria also succeeded in providing enough studies to create a narrative without threatening validity. The U.K. registry data provide an interesting comparison that other reviews do not include and help to highlight the rapidly increasing incidence rate in the U.K. compared with Europe.
The main limitation encountered when making comparisons was the different standard populations used. Although these different populations cannot account for large differences observed between countries, they do prevent comparisons between countries lying at similar latitudes. The main source of error in this review is differences in data collection. Cancer registries will be affected by under-reporting to a greater extent than studies performed in the community. Without histological confirmation, the incidence is also likely to be overestimated. Although these issues would present a problem for a detailed statistical analysis, they are less of a problem for a descriptive narrative such as this. Whenever studies were compared, care was taken to ensure that the studies were based on similar methodologies. For example, a study counting number of patients would not be compared with a study counting number of tumours.
The data presented here provide a good representation of NMSC incidence in white populations. There have, however, been very few good epidemiological studies undertaken in lower latitude regions such as Africa. These regions contain a very wide range of ethnicity resulting in a wide range of incidence rates. The incidence studies that have been located all use very different methodologies, rendering comparisons extremely difficult. Any future studies should aim to include the information discussed above in order to allow accurate comparison internationally, particularly as NMSC incidence data from cancer registries is likely to be highly limited.
This review focuses exclusively on nonmelanoma malignancies and does not cover the incidence of melanoma. Melanoma skin cancer receives a much higher level of registration and was not included in the majority of the papers reviewed here. The substantial variation between BCC and SCC incidence rates is caused by the differences in their aetiology. Melanoma has a different aetiology again, rendering any comparisons between the diseases much less relevant.
In conclusion, in white populations, NMSC has the highest incidence of all cancers. Despite this it is rarely included in national cancer statistics and there are very few papers looking at incidence rates in each country. Although NMSC does have a low mortality rate, it still has a significant impact on quality of life and is placing a large financial burden on health care services.
All the registry data were provided with warnings of under-reporting and problems with multiple tumour registration. The fact that Wales does not officially publish any NMSC data indicates that current registry guidelines are not sufficient in obtaining accurate measurements.
A recent update to NICE guidelines8 highlighted the need to establish the true epidemiology of NMSC. This review summarizes past and current knowledge in this area but is also in agreement with NICE recommendations of improving registration and including multiple tumours. In order to conduct accurate prevention studies on the effects of sun protection, it is essential to bring registration in line with other malignancies. This could be done by histologically verifying all cases of NMSC and reporting them to national cancer registries.
As NMSC risk is due to a number of life-long factors, it is important to make these changes now before the incidence rates rise to a much higher level. BCC and SCC have become a major problem in Australia with primary prevention programmes only just starting to have an effect after 20 years. If incidence rates were to continue to rise unchecked in the U.K., NMSC may start to represent a major public health problem and a significant burden to the National Health Service.
What’s already known about this topic?
• The incidence rate of nonmelanoma skin cancer is high compared with other malignancies and is increasing.
• The current best source of incidence data comes from many observational studies of populations in a defined, often small, geographical location.
• Incidence rates vary geographically according to various factors including ultraviolet radiation exposure, sun-seeking behaviours and skin type.
• Recent National Institute for Health and Clinical Excellence guidelines recommend research in basal cell carcinoma epidemiology and improved tumour registration practice.
What does this study add?
• This is the largest and most complete systematic review in this field to date.
• This is the first review that has pulled together all observational studies in a systematic way.
• The review highlights the high and increasing incidence rates in the U.K. compared with other European countries.
• These findings are important as they highlight the need for early prevention.
The authors extend their thanks to Dr Sean McPhail of the South West Cancer Intelligence Service for the provision of NMSC incidence data from every cancer registry in England. We also acknowledge Mr Jeremy Hinds for his invaluable assistance in producing the maps displaying the U.K. incidence data.