eTable 1. Comparison of Demographic Characteristics of Kaiser Permanente of Northern California (2012) to the U.S. Population (Census 2010)
eTable 2. KPNC Incidence Rates of Basal Cell Carcinoma Standardized to the 2010 U.S. Population by Age, Sex and Race
Asgari MM, Moffet HH, Ray GT, Quesenberry CP. Trends in Basal Cell Carcinoma Incidence and Identification of High-Risk Subgroups, 1998-2012. JAMA Dermatol. 2015;151(9):976-981. doi:10.1001/jamadermatol.2015.1188
The incidence of basal cell carcinomas (BCCs) is increasing globally, but incidence rates in the United States are difficult to quantify because BCCs are not reportable tumors.
To estimate annual BCC incidence rates by age, sex, and race/ethnicity to identify demographically distinct high-risk subgroups and to assess changes in rates over time.
Design, Setting, and Participants
In this retrospective cohort study (January 1, 1998, through December 31, 2012), we studied 147 093 patients with BCC from Kaiser Permanente Northern California, a large, integrated health care provision system, identified using a previously validated BCC registry.
Main Outcomes and Measures
We estimated annual BCC incidence rates by age, sex, and race/ethnicity and assessed changes in rates over time. The BCC incidence rates were standardized to the age, sex, and race/ethnicity distribution of the 2010 US Census population.
In models adjusting for age, sex, and race, male patients had higher rates than female patients (incidence rate ratio [IRR], 1.65; 95% CI, 1.60-1.70). Persons 65 through 79 years of age and those 80 years and older had higher rates than persons 40 through 64 years of age (IRR, 2.96; 95% CI, 2.86-3.06; and IRR, 5.14; 95% CI, 4.94-5.35, respectively). Whites had higher rates than multiracial persons (IRR, 1.96; 95% CI, 1.80-2.13), Hispanics (IRR, 8.56; 95% CI, 7.79-9.41), Asians (IRR, 33.13; 95% CI, 27.84-39.42), and blacks (IRR, 72.98; 95% CI, 49.21-108.22).
Conclusions and Relevance
We estimate that BCCs occur in approximately 2 million Americans annually. Our findings provide an updated estimate of the incidence of BCCs, highlight the changing epidemiologic findings, and better identify demographically distinct high-risk subgroups.
Nonmelanoma skin cancers (NMSCs), including basal cell carcinomas (BCCs) and squamous cell carcinomas (SCCs), are the most common malignant tumors in the United States and have been estimated to affect more than 2 million Americans annually.1 Despite the large population affected, NMSCs have been difficult to study in part because of their exclusion from cancer registries and national cancer surveillance programs, such as the Surveillance Epidemiology and End Results program. Epidemiologic estimates of NMSC incidence rates are often based on periodic surveys and are generally outdated, with the last National Cancer Institute–funded survey performed more than 3 decades age.2 Studies that rely on disease codes have historically combined BCCs and SCCs because they share the same International Classification of Diseases identifiers. Such data limitations hinder the study of BCC etiology, prevalence, incidence, and disease burden.
The incidence of NMSC has been reported to be increasing globally,3 yet the incidence of NMSC in the United States is not well characterized. The direct cost of treating NMSC in the United States has been estimated to be more than $1.4 billion annually,4 making NMSC among the top 5 most costly cancers.5 The US government has responded to this epidemic by allocating resources to improve epidemiologic research and preventive efforts, including release of a docket by the Centers for Disease Control and Prevention to obtain information from the public on preventing skin cancer, bipartisan legislative efforts to establish a skin cancer research fund for the National Institutes of Health, and most recently, in 2014, the US Surgeon General’s Call to Action to Prevent Skin Cancer.6 However, the absence of US data on the incidence of NMSC has hampered the development of consistent health care policies to guide screening and preventive measures. There is an urgent need to better define incidence rates of NMSC, especially BCCs, which account for most NMSCs (>80%).
In 1997, Kaiser Permanente Northern California (KPNC) implemented a computerized electronic records system for all pathology specimens received for examination, recording specimen types, anatomical locations, gross and microscopic diagnoses, and assigned Systematized Nomenclature of Medicine (SNOMED) codes. SNOMED codes are systematically organized, computer-searchable medical terms that constitute comprehensive clinical terms for clinical documentation and reporting.7 These codes were used to identify a BCC registry at the KPNC, which has previously been validated.8 In this study, we estimate the incidence of BCC during a 15-year period ending in 2012. We calculate the incidence of BCC in our total population by age, sex, and race/ethnicity and compare our standardized incidence rates to other population-based BCC incidence estimates. In doing so, we estimate the incidence of BCC, highlight the changing epidemiologic findings, and identify high-risk demographic subgroups.
The KPNC is a large, integrated health care delivery system that provides comprehensive health care and pharmaceutical benefits to a large and diverse community-based population of more than 3.2 million members residing in Northern California. The KPNC membership represents approximately 33% of the insured population and 28% of the total population in its service area. The member population reflects the general population in the Northern California region, although, as an insured population, it underrepresents persons with very low levels of education or income.9
This study was approved by the Kaiser Foundation Research Institute Institutional Review Board. The Declaration of Helsinki protocols were followed, and a waiver of informed consent was obtained.
We identified all KPNC members from January 1, 1998, through December 31, 2012, and extracted every electronic pathology report with a SNOMED code of M809xx for specimens collected during that period. We have previously found that pathology reports with SNOMED code M809xx have a positive predictive value of 99.2% for being a true BCC.8 Given that BCC is a disorder that is subject to multiple primary tumors and the disorder can be cured with appropriate management, we reasoned that a person could have a BCC, be treated, and then be at risk for the disorder again. Because of the difficulty in distinguishing new primary tumors from the automated pathology reports, we required a 365-day period with no electronic pathology reports consistent with a SNOMED BCC code (365-day clear period) before counting a BCC as being a new primary case. We considered this to be a very conservative estimate of new primary BCC incidence rates. As a sensitivity analysis, we used a 90-day clear period. We required patients to be active KPNC members at the time the specimen was collected.
Patient age and sex were extracted from the KPNC membership databases. Although race/ethnicity is not systematically captured for all members, the race/ethnicity of most patients (approximately 94%) was identified via a search of electronic medical records, membership databases, and other administrative databases. For patients who did not have race/ethnicity recorded in administrative databases, we imputed race/ethnicity using the Bayesian Improved Surname and Geocoding algorithm, which uses US Census data and surname to impute the probabilities that the patient belonged to 1 of 5 racial/ethnic groups (non-Hispanic white [white], Hispanic, Asian, black, or multiracial).10
The KPNC tracks membership on a monthly basis. Member-months for each calendar year by race/ethnicity, sex, and age (in 1-year increments) were calculated. The BCC cases were summarized for each calendar year, and in the primary analysis using the 365-day clear period, persons could contribute no more than one BCC to the numerator in any given year. Annual incidence rates were calculated by dividing the number of BCC cases in each year by the KPNC membership person-years in that year. For reporting, 1-year age strata were summarized into 4 age categories (<40, 40-64, 65-79, and ≥80 years of age). The 95% CIs for incidence rates were calculated based on a normal approximation to the Poisson distribution.
We used Poisson regression to examine sex, age, and race/ethnicity in relation to BCC incidence in the most recent year of complete available data (2012), providing point and interval estimates of incidence rate ratios (IRRs) for each covariate adjusted for the other covariates. When making inferences about the burden of disease in the US population and trends in incidence rates, we used the direct standardization method to apply the age- (in 1-year increments), sex-, and race/ethnicity-specific KPNC BCC incidence rates in each year to the age, sex, and race distribution of the 2010 US Census population. All incidence rates are presented as cases per 100 000 person-years.
Poisson regression was also used for point and interval estimation of the mean annual percentage change (APC) in BCC incidence rates during the study period (1998- 2012), with calendar year as a continuous variable in the regression models. These analyses were stratified by (1) race, categorical age group, and sex; (2) race and age group (combining male and female patients and adjusting for sex); and (3) race and sex (combining and adjusting for all age categories).
We identified 221 624 BCC cases in the KPNC population between 1998 and 2012, representing 147 093 unique persons. Mean age was 65.25 years, median age was 66 years (range, 2-105 years), and 80 588 patients (54.8%) were male (Table 1). The BCCs primarily occurred in whites (n = 135 538 [92.1%]).
The BCC incidence rate among all KPNC members in year 2012 was 535 cases per 100 000 person-years (Table 2), with rates for male patients being 46% higher than for female patients (640 per 100 000 person-years for male patients vs 439 cases per 100 000 person-years for female patients). Rates increased with increasing age, and the rate increase was higher for whites than nonwhites. In models adjusting for age, sex, and race, male patients had higher rates than female patients (IRR, 1.65; 95% CI, 1.60-1.70), persons 65 through 79 years of age and those 80 years and older had higher rates than persons 40 through 64 years of age (IRR, 2.96; 95% CI, 2.86-3.06; and IRR, 5.14; 95% CI, 4.94-5.35, respectively), and whites had higher rates than multiracial persons (IRR, 1.96; 95% CI, 1.80-2.13), Hispanics (IRR, 8.56; 95% CI, 7.79 -9.41), Asians (IRR, 33.13; 95% CI, 27.84-39.42), and blacks (IRR, 72.98; 95% CI, 49.21-108.22).
The KPNC member population differs from the US population, particularly with respect to distribution by race/ethnicity, with the KPNC member population being approximately 48% white compared with 64% in the US population (eTable 1 in the Supplement). The age-, sex-, and race/ethnicity-adjusted annual KPNC BCC incidence rates ranged from 513 cases per 100 000 person-years in 1998 to 600 cases per 100 000 person-years in 2012, approximately a 17% overall increase in rates during the 15-year period (eTable 2 in the Supplement).
Given that BCCs occur overwhelmingly among whites (92.1%), further analyses were performed limited to that subgroup. Figure 1 illustrates sex-specific KPNC BCC incidence rates among whites, standardized to the 2010 US Census population, and illustrates that although absolute rates are higher among male patients, the rate of increase is actually steeper among female patients. Between 1998 and 2012, the BCC incidence increased in white male patients from 941 cases per 100 000 person-years to 1069 cases per 100 000 person-years and in white female patients from 620 cases per 100 000 person-years to 774 cases per 100 000 person-years. Overall among whites, rates increased approximately 18% during 15 years.
Figure 2 illustrates age-specific KPNC BCC incidence rates among white male patients, standardized to the 2010 US Census population by male patients (Figure 2A) and female patients (Figure 2B). The increase in incidence is greatest among persons 80 years or older for both sexes (APC, 2.72; 95% CI, 2.35-3.10; for female patients and APC, 1.67; 95% CI, 1.35-1.99; for male patients). Rates among middle-aged women increased with time (APC, 0.69; 95% CI, 0.44-0.93; for those aged 40-64 years; APC, 1.28; 95% CI, 1.01-1.54; for those aged 65-80 years). For male patients, rates held steady for the 40- through 64-year-old group (APC, 0.02; 95% CI, −0.20 to 0.24) but increased for those aged 65 through 79 years (APC, 1.16; 95% CI, 0.96-1.37). Rates among white male patients and female patients 40 years or younger appear stable over time (APC, 0.49; 95% CI, −0.29 to 1.29; for female patients and APC, −0.17; 95% CI, −1.08 to 0.74; for male patients).
Poisson models indicate that BCC incidence rates increased during 1998 to 2012 by approximately 0.87% per year for whites (all ages and sex combined), with increases being the largest overall for female patients (1.11%) or for persons 80 years and older (2.23%). Rates decreased significantly for Asians and somewhat for blacks and multiracial persons, and there was an overall trend toward a decrease among Hispanics.
Sensitivity analyses were performed using a 90-day algorithm, which identified 269 119 BCC cases and an incidence rate of 628 cases per 100 000 person-years in 2012. In models adjusting for age, sex, and race, male patients had higher rates than female patients (IRR, 1.79; 95% CI, 1.74-1.84), persons 65 through 79 years of age and 80 years and older had higher rates than persons 40 through 64 years of age (IRR, 3.11; 95% CI, 3.01-3.21; and IRR, 5.68; 95% CI, 5.48-5.89; respectively), and whites had higher rates than multiracial persons (IRR, 1.98; 95% CI, 1.83-2.14), Hispanics (IRR, 9.23; 95% CI, 8.44-10.10), Asians (IRR, 34.33; 95% CI, 29.18-40.39), and blacks (IRR, 116.57; 95% CI, 73.84-184.02). The age-, sex-, and race/ethnicity-adjusted annual KPNC BCC incidence rates ranged from 628 cases per 100 000 person-years in 1998 to 720 cases per 100 000 person-years in 2012, approximately a 15% overall increase in rates during the 15-year period. Our sensitivity analyses reveal similar relative rates among persons of different ages and races and consistent changes in rates over time (15% vs 17%).
Using a validated BCC registry,8 we report incidence rates of BCCs during a 15-year observation period at the KPNC, a large, integrated health care delivery system. After standardizing to the 2010 US Census population, we found that BCC incidence rates at the KPNC increased from 513 cases per 100 000 person-years in 1998 to 600 cases per 100 000 person-years in 2012—an increase of approximately 17% during this 15-year period. This modest increase is in contrast to previous US-based studies,11- 13 which have reported 80% to 200% increases in incidence rates during a 10- to 20-year period. Previously reported US-based BCC incidence rates have varied widely from 61.5 to 1019 per 100 000 person-years for women and 70.7 to 1488 per 100 000 person-years for men.11- 16 The variation in reported rates is partly attributable to geographic locations of the study populations, periods, and methods for capturing BCC. The most recent US-based BCC incidence rate,11 which was based on data from 2 large, prospective cohort studies (Nurses’ Health Study, 1986-2006, N = 121 701, and Health Professionals’ Follow-up Study, 1988-2006, N = 51 529) estimated age-adjusted incidence rates of 1109 cases per 100 000 person-years for women and 1488 cases per 100 000 person-years for men. The cohorts used in that study were predominantly white health care professionals who ranged in age from 40 to 70 years during the follow-up. To permit meaningful direct comparison of our data, we conducted a subanalysis limited to whites aged 40 to 70 years and calculated the standardized incidence rates in 2006 as 846 per 100 000 for women and 1194 per 100 000 for men. Thus, our calculated rates are somewhat lower than those reported by Wu et al11; however, they relied on self-reported BCC (rather than pathologic confirmation), and they stated that previous reports have confirmed validity of the BCC self-report in 90% of women and 80% of men. If we take into account a 10% to 20% margin of error in their reported rates, our calculated rates are quite similar.
In examining demographic subgroups, we found that after adjusting for age and race, the BCC rates for male patients were 1.7 times that of female patients, which is similar to previously reported sex differences in the BCC incidence.14,16 With respect to age, those who were 65 years or older had an increased risk of BCC during the observation period. Whites were at highest risk for BCCs, with adjusted incidence rates 8 to 70 times those of Asians, blacks, or Hispanics. Rates of BCCs for persons other than whites did not increase during the study period, and in fact for some groups rates decreased modestly. The highest risk subgroup were whites 80 years or older, and that subgroup experienced the biggest increase in BCC incidence during the observation period, with an estimated mean APC of 2.23%. Given a recent focus in the dermatology literature on skin cancer screening and behaviors in nonwhites,17- 19 it is important to quantify risk by examining standardized incidence rates of the most common skin cancer among different age, sex, and racial/ethnic subgroups to help guide screening and prevention efforts.
Applying our 2012 incidence rates to the US population, we would expect nearly 2 million incident BCCs annually in the United States. This annual BCC incidence rate estimate appears to be slightly higher than what is reported on a National Institutes of Health website, which estimates the annual incidence of all NMSC (not just BCC) at 2 million.20 A previously published claims-based study1 estimated the total number of persons treated for NMSC in 2006 at 2 152 500 but could not differentiate between BCCs and SCCs because both cancers share the same claims code. Given that the ratio of incident BCCs to SCC is approximately 4:1, our estimate of BCC incidence is higher than previously reported.
Unlike previously published reports, we did not find an increased rate of BCC among individuals younger than 40 years,21 and we did not find an increased incidence among Hispanics. With respect to age, we found a mean APC of 1% to 2% among whites 65 years and older during the observation period, which is lower than the 4.2% mean APC in NMSC cases in the Medicare population reported from 1992 to 2006.1 Direct comparison of these data are difficult because Rogers et al1 did not differentiate SCC from BCC and reported on all NMSCs.
A unique strength of our study was the use of a validated skin cancer registry within a large integrated health care provision system that includes all pathology reports and is based in a diverse population. One limitation of this study is that data were collected from a single health care setting, which is composed of insured health plan members and underrepresents persons with very low levels of education or income, although a previous study22 examining patient and tumor characteristics in incident skin cancers at the KPNC found that our health care system data closely mirror state and national registry data for reportable skin cancers (melanoma). Furthermore, it is possible that BCCs were diagnosed outside the KPNC setting and thus not captured in the pathology database. However, seeking care outside a prepaid health plan and incurring additional out-of-pocket expense is unlikely. Some BCCs may be removed without submitting a pathology specimen for definitive diagnosis. However, the KPNC standard of care is to submit specimens suspected of being malignant for histopathologic diagnosis, so such events also are likely to be rare. Finally, although BCCs often arise repeatedly and may arise concurrently in susceptible individuals, SNOMED codes do not distinguish among primary, subsequent, recurrent, or multiple concurrent BCCs, making it difficult to identify true incident disease. Because of this uncertainty, we performed sensitivity analyses using a less conservative 90-day clear period for defining new cases, in addition to our more conservative base-case 365-day clear period. The latter algorithm allows individuals to contribute at most 1 BCC per year, whereas the former allows up to 4 new BCCs per person to count toward incidence rates in any given year. Although incidence rates were in general approximately 20% higher using the 90-day algorithm, the relative rates among persons of different ages and races were similar, as were the changes in rates over time and the mean APCs in rates.
We found that the overall incidence of BCC slightly increased during a 15-year observation period, but this increase was not remarkable. In extrapolating our data to the United States, we estimate that approximately 2 million individuals develop at least one BCC in the United States in a given year. In addition to providing a current estimate of the BCC incidence rate, we identify highest-risk demographic subgroups, namely, whites older than 65 years, and show that male patients are at increased risk compared with female patients. Our findings can help guide selection of high-risk demographic groups for future screening and prevention efforts.
Accepted for Publication: March 24, 2015.
Corresponding Author: Maryam M. Asgari, MD, MPH, Department of Dermatology, Massachusetts General Hospital, 50 Staniford St, Ste 270, Boston, MA 02114.
Published Online: June 3, 2015. doi:10.1001/jamadermatol.2015.1188.
Author Contributions: Dr Asgari had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Asgari, Ray.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: Asgari, Moffet, Ray.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Quesenberry.
Obtained funding: Asgari.
Administrative, technical, or material support: Moffet.
Study supervision: Asgari.
Conflict of Interest Disclosures: Dr Asgari reported receiving grant funding as a research investigator to her institution (Kaiser Permanente) from Pfizer Inc and 3 Valeant Pharmaceuticals, but these funding organizations are not relevant to the current work. No other disclosures were reported.
Funding/Support: This work was supported by grants U19 CA79689 and R01 CA166672 from the National Cancer Institute at the National Institutes of Health (Dr Asgari).
Role of the Funder/Sponsor: The funding source had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and the decision to submit the manuscript for publication.