Assessment of Rebound in Cancer Diagnoses in California in 2022, by Stage, Sociodemographic Factors, and Rurality.

INTRODUCTION
The coronavirus disease 2019 (COVID-19) pandemic impacted access to healthcare services throughout the United States (U.S.), resulting in reduced cancer screening rates (1–3) and delayed cancer diagnoses in 2020 (4–7). Using data from the National Cancer Institute (NCI) Surveillance Epidemiology and End Results program (RRID:SCR_006902), which represents approximately 48% of the U.S. population, Burus et al. reported that the age-adjusted incidence rate of all cancers combined in 2020 was 9.4% (−8.5%, −10.5%) less than what was observed in 2019 (8). Moreover, there was a reduction in observed incidence for 11 common cancer types, with differences in the age-adjusted incidence in 2020 compared to 2019 ranging from 4.3% less for pancreas to 15.4% less for melanoma (8).
Consequently, it has been anticipated that there will be a rebound in cancer cases (more cancers diagnosed than expected based on past trends) as access to healthcare services returns to pre-pandemic levels and previously undiagnosed cases are diagnosed. Yet, reports using data through 2021 concluded that there was no evidence of a rebound (8–11). The observed incidence rate of all tumors across 22 SEER registries in 2021 was not different than the expected rate (rate ratio: 1.0; 95% CI: 0.97, 1.03) (9). Moreover, certain sites, including lung and bronchus and pancreas, continued to have lower observed incidence rates (2–5%) than expected. In that analysis, the only cancer type with a suggested rebound was distant breast cancer (rate ratio: 1.09; 95% CI: 1.04, 1.13) (9). Kim et al. also reported that there was no rebound overall but did observe small differences by sex and race-ethnicity: the incidence rate in 2021 was 2% higher than expected among females and 0.8% lower than expected among males. By race-ethnicity, the incidence rate in 2021 was 5.4% higher than expected among Non-Hispanic Asian or Pacific Islander individuals and 1.2% lower than expected among Non-Hispanic White individuals; there were no differences between the observed and expected incidence rates in other racial-ethnic groups.
Given that the timeline for resuming healthcare services related to cancer screening and detection following the COVID-19 pandemic varied geographically, it is possible that rebounds in cancer cases in smaller geographic areas or population groups may not be detectable when looking across all SEER registries. California contributes the largest number of cancers diagnosed across all states in the U.S. (approximately 10% of all cases) and includes three SEER cancer registries (12). It also includes a racially, ethnically, and geographically diverse population. Thus, this state provides a unique opportunity to examine variation in potential post-pandemic rebound of cancer cases by cancer type, stage, sociodemographic factors, and regional factors. For this analysis, we quantified the difference in observed versus expected cases, overall and for 12 common cancers, by age, sex, race-ethnicity, stage at diagnosis, and rurality. This information is needed to continue to improve targeted cancer screening, diagnosis, and treatment programs.

MATERIALS AND METHODS
Data from SEER, submitted by the three regional registries in California, were used to identify invasive tumors diagnosed in California in 2001–2022 (n= 3,439,814). For each of the 22 years, we used the November submission file that corresponded to the first complete dataset for a given year (November 2024 submission was used for 2022, November 2023 submission was used for 2021, etc.) to standardize the completeness of case reporting across years and limit differences in observed versus expected cases driven by cancer registry operations.
We focused on total invasive tumors as well as 12 common cancers: bladder, breast, colon and rectal, kidney and renal, leukemia, lung and bronchus, melanoma of the skin (hereafter melanoma), non-Hodgkin lymphoma, pancreas, prostate, thyroid, and uterine. Together, these cancers accounted for 78% of all invasive tumors diagnosed in California in 2022. The International Classification of Diseases for Oncology (ICD-O-3) site and histology codes used to define the cancer types are provided in the Supplementary Materials and Methods. We obtained information on variables of interest from SEER, including age at diagnosis, sex, race and ethnicity, stage at diagnosis, and urban/rural status of county of residence at diagnosis (see Supplementary Materials and Methods for SEER*Stat variables used). Race and ethnicity data were obtained from medical records and the North American Association of Central Cancer Registries (NAACCR) identification algorithms (13,14).
To calculate the ratio of the observed to expected (O/E) number of tumors, the number of cases diagnosed in 2022 from the three regional registries that comprise reporting regions in California (Cancer Registry of Greater California, Greater Bay Area Cancer Registry, Los Angeles Cancer Surveillance Program) were used for the observed values. We used the NCI’s method to calculate expected cases, which is the method SEER uses to estimate registry completeness rates (15). First, the Joinpoint Regression Program (v5.4.0.0) was used to estimate a regression line to the observed count data from 2001 to 2021 (omitting 2020) with up to three joinpoints using Poisson variance, a log transformation, and the weighted Bayesian information criterion model selection method (16). The expected number of cases in 2022 was then projected using the average annual percentage change (AAPC) for 2018–2021 omitting 2020. We calculated 95% confidence intervals (CI) for the O/E ratio using the delta approximation on the variance of the O/E ratio, where the variance of the expected (the Joinpoint projected count) is calculated using the empirical quantile method for confidence interval estimates of the AAPC (17).
A limitation of the NCI’s method for calculating expected number of cases is that the O/E ratio from stratified results can differ from the overall O/E ratio for the combined sample. Therefore, SEER recommends not combining data from registries when using Joinpoint models to calculate the expected number of cases. In addition, combining data for all of California may obscure differences by geographic regions. Therefore, we conducted all analyses separately for the Greater California, Greater Bay Area, and Los Angeles regional registries. The Cancer Registry of Greater California includes 48 counties in California (approximate 2022 population: 22.1 million), Greater Bay Area Cancer Registry includes nine counties (approximate 2022 population: 7.2 million), and the Los Angeles Cancer Surveillance Program includes one county (approximate 2022 population: 9.7 million) (18).
Cases with other or unknown race or ethnicity were not shown in stratified results for race-ethnicity but were included in overall results and results stratified by other factors. Similarly, leukemia was not included in analyses stratified by stage. To minimize the influence of small numbers, we suppressed results when either the observed or expected number of cases in 2022 was less than 100.

Data Availability.
The data used in this study are publicly available from the Surveillance, Epidemiology, and End Results (SEER) Program (www.seer.cancer.gov).

RESULTS
Overall, there were 104,972 invasive tumors diagnosed in Greater California, 42,271 invasive tumors diagnosed in Los Angeles County, and 36,935 invasive tumors diagnosed in the Greater Bay Area in 2022. Characteristics of people diagnosed with invasive tumors in California in 2001–2019, 2020, 2021, and 2022 are presented in Supplementary Table S1 on the tumor level (i.e., an individual may have had more than one tumor diagnosed). Tumors diagnosed in 2022 versus 2001–2019 were more likely to be among people 65 years or older (59% vs. 54%), Asian American individuals (13% vs. 10%), and Hispanic individuals (24% vs. 19%), and less likely to be at the localized stage (46% vs. 49%).
Table 1 shows the difference in observed versus expected total invasive tumors in 2022 by regional registry in California. Figures 1–3 plot the observed and modeled number of invasive tumors by year from 2001–2021, and the observed and expected (projected) number of cases in 2022. There was no difference between the overall observed and expected number of cases in 2022 in any of the regional registries. The only cancer sites with statistically significant differences between the observed and expected number of cases in 2022 were breast cancer in Greater California and uterine cancer in Los Angeles County. In Greater California, breast cancer cases were 5% lower than expected in 2022 (O/E: 0.95; 95% CI: 0.90, 0.98). In Los Angeles County, uterine cancer cases were 9% lower than expected in 2022 (O/E: 0.91; 95% CI: 0.86, 0.96).
We examined the O/E ratios (95% CI) for total invasive tumors diagnosed in 2022 by age, sex, race-ethnicity, stage, and rurality for the three regional registries (Supplementary Tables S2-S4). The only statistically significant differences between observed and expected number of total invasive tumors were for non-Hispanic Native Hawaiian or Pacific Islander individuals in Greater California where 13% fewer tumors were observed than expected (O/E: 0.87; 95% CI: 0.51, 0.94) and for distant stage tumors in the LASCP, where the observed number of distant tumors was 11% lower than expected in 2022 (O/E: 0.89; 95% CI: 0.80, 0.99).
Supplementary Tables S2-S4 also show the O/E ratios and 95% CI for 12 common cancer sites examined by age, sex, race-ethnicity, stage, and rurality. Results differed by registry. Generally, there were few differences in the observed number of tumors in 2022 compared to expected. Where differences occurred, the observed number of cases was more often lower than expected. In Greater California, the sites/population groups with fewer observed than expected cases included: breast (localized and regional stage, non-Hispanic White; O/E: 0.92–0.93), lung (non-Hispanic White, rural; O/E: 0.71–0.95), melanoma (remote and unknown stage, non-Hispanic White; O/E: 0.66–0.90), bladder (<65 years, large metropolitan areas; O/E: 0.90–0.92), kidney and renal (non-Hispanic White, 65+ years, large metropolitan areas; O/E: 0.87–0.89), and leukemia (small metropolitan areas, O/E: 0.77). Colorectal cancer of unknown stage (O/E: 1.54; 95% CI: 1.27, 1.76) and localized melanoma (O/E: 1.24; 1.01, 1.44) were the only two sites with more observed tumors than expected in Greater California. In the Greater Bay Area, the observed number of pancreas tumors was lower than expected (O/E: 0.90; 95% CI: 0.71, 0.99), while the observed number of CRC tumors among males and Asian American individuals was higher than expected (O/E: 1.12–1.16). In Los Angeles county, non-Hodgkin lymphoma among non-Hispanic Black individuals (O/E: 0.79; 95% CI: 0.60, 0.99) and uterine cancer in several sub-groups (<65 years, 65+years, Hispanic, localized stage; O/E: 0.83–0.93) had lower than expected case counts, while CRC of unknown stage (O/E: 1.26; 95% CI: 1.02, 1.52) and prostate cancer among Hispanic males (O/E: 1.24; 95% CI: 1.04, 1.46) had higher observed than expected number of tumors.

DISCUSSION
In this report, we used 2001–2022 data from SEER to examine the difference between observed versus expected number of total invasive tumors in California in 2022, overall and for 12 cancer types. As with 2021 patterns, we observed no overall evidence of a rebound in cancer cases in 2022. In fact, for the cancer sites and population groups where there were statistically significant differences, most continued to have fewer tumors diagnosed in 2022 than expected.
The reduction in incident cancer cases in 2020 has been widely reported (11). Using data from the National Cancer Database (NCDB), Nogueira et al. reported a 12.4% reduction in total cancer cases in 2020 compared to what was expected based on data from 2018–2019 (19). In the NCDB, melanoma and prostate cancer were among the cancers with the largest difference between observed versus expected cases (observed about 20% lower than expected for both). These observations raised the concern that there will be a rebound in cancer cases in subsequent years, where cases that were not diagnosed in 2020 are eventually diagnosed resulting in higher observed cases than expected. Yet, emerging reports concluded there was no rebound in 2021 (8,11). Using the most recent data available, in this analysis, we confirmed that there was also no rebound in California through 2022. While there was some variation, the number of cancer cases diagnosed in 2022 in California was similar to what was projected based on trends from 2001–2021 with few exceptions.
One of the notable exceptions was the markedly lower (29% fewer) number of lung cancer cases observed in rural areas in 2022. Only Greater California includes rural areas, so this was the only registry that contributed data to this estimate. It is not known why lung cancer cases remained lower than expected in rural areas. It is possible that this reflects an emerging trend in reduced lung cancer incidence in rural areas due to reductions in cigarette smoking (20). Another explanation for lower number of tumors in rural areas than expected may be reductions in the population size of rural areas of California (21); but it’s not clear why this would impact lung cancer specifically. Lastly, it is possible that individuals at higher risk of lung cancer due to smoking and those with un-diagnosed lung cancer may have been more vulnerable to respiratory disease and died during the COVID-19 pandemic (22). People living in rural areas of California have higher incidence of lung cancer and are often diagnosed at later stages compared to those in non-rural areas (23,24), potentially resulting in relatively more people with un-diagnosed lung cancer in rural areas compared to urban areas during the pandemic. This competing cause scenario is difficult to verify or test. Further research is needed to determine if this pattern can be replicated in other rural areas of the U.S. and inform whether increased evidence-based lung screening in rural areas is warranted.
Within registries, two cancer sites had statistically significantly lower numbers of observed tumors than expected: female breast in Greater California and uterine cancer in Los Angeles County. The most recent report to the nation concluded that cancer incidence is increasing among females, with the largest increases observed in these specific cancer sites (11). Between 2017–2021, female breast cancer increased 1.6% on average per year (95% CI: 0.9, 2.0) and cancer of the corpus and uterus increased 0.8% on average per year (95% CI: 0.4, 1.2). The AAPC were even higher among Asian American/Pacific Islander and Hispanic females, which make up a larger portion of the California population compared to the U.S. population. Among Asian American/Pacific Islander females, on average between 2017–2021, breast cancer increased 3.3% per year (95% CI: 2.1, 4.3) and uterine cancer increased 2.5% per year (95% CI: 2.2, 2.8). Among Hispanic females, corresponding AAPCs were 1.7 (1.2, 2.7) and 2.7 (2.5, 3.6), respectively. Given the rising incidence of these two cancer sites among females, it is especially concerning to see lower number of cases diagnosed in 2022 compared to expected. This may result in more regional and distant stage breast and uterine cancers being diagnosed in California in coming years.
Our analysis had several strengths and limitations. We used the same method for calculating the expected number of cases that the NCI SEER program uses and used the November submission file for each year to limit the possibility of registry-related administrative processes influencing our results. However, the method developed by the NCI cannot anticipate abrupt changes in the future and can be sensitive to sudden recent changes in the observed number of cases (15,25). Additionally, while California is a large diverse state with complete cancer surveillance data, we were unable to examine differences in the observed versus expected number of cases for specific groups with small population sizes in the state, such as American Indian or Alaska Native individuals. Lastly, we conducted several sub-group analyses and therefore some of our statistically significant findings could be due to chance.
In summary, despite the large reduction in cancers diagnosed in California during 2020, there does not appear to have yet been a rebound in cancer cases in 2022, consistent with observations for 2021. In fact, certain sites remain low, including lung cancer in rural areas, breast cancer among females in Greater California, and uterine cancer in Los Angeles County. It is possible that a rebound of delayed diagnoses may still occur. The longer it takes for the rebound to manifest, the greater likelihood that there will be an increase in late-stage diagnoses. However, our data suggest that the low incidence in 2020 cannot entirely be explained by delayed diagnoses due to healthcare impacts during the pandemic. It is also possible that the approximate 10% reduction in tumors reported in 2020 may have been due, in part, to individuals dying from competing causes. In either case, our data demonstrate that the impact of the COVID-19 pandemic on cancer incidence varied by cancer type and across population groups. Continued close monitoring of cancer diagnoses, particularly in smaller geographic areas and population groups, is needed to continue to improve cancer control strategies.

Supplementary Material
12345