Impact of the COVID‐19 pandemic on international cutaneous squamous cell carcinoma incidence: A systematic review and meta‐analysis

Abstract Background Non‐melanoma skin cancer (NMSC) is the most common cancer globally in white ethinicity populations, and cutaneous squamous cell carcinoma (cSCC) is the second most common subtype. The COVID‐19 pandemic severely impacted public and private healthcare systems. Many studies have reported reduced cancer diagnoses during the pandemic. The impact of the COVID‐19 pandemic on global cSCC and NMSC incidence is poorly reported. Objectives The aim was to conduct a systematic review and meta‐analysis to assess the impact of the COVID‐19 pandemic on global cSCC and NMSC incidence rates, compared with 2019 incidence rates. Two primary outcome measures were used: crude incidence rate ratios (CIRR) and age‐standardised incidence rate ratios (ASIRR). Methods A structured search was undertaken on 23 March 2023 using grey literature and four electronic databases: MEDLINE, CINAHL, EMBASE and Web of Science. Studies published before January 2020 were excluded. A quality assessment was undertaken using A. Lomas quality assessment tool. CIRR outcomes were synthesised in a meta‐analysis, while ASIRR outcomes were narratively synthesised. Results Fourteen cancer registries were included, capturing data from 13 countries across Europe. Variation was observed in NMSC and cSCC incidence across the cancer registries. Pooled cSCC crude incidence rates in 2020 were equal to crude incidence rates in 2019 (cSCC‐CIRR 1.00 (95% confidence interval (CI) 0.94–1.06). In 2021, the pooled result indicated a non‐significant 8% increase in cSCC crude incidence rates, compared with 2019 (cSCC‐CIRR 1.08 (95% CI 0.98–1.19). Significant reductions were reported in NMSC incidence across all meta‐analyses in 2020 and 2021 compared with 2019. Heterogeneity was observed across most pooled estimates (I 2>75%). Conclusion There was a lack of high quality data on cSCC incidence rates recorded during the pandemic outside of Europe. The COVID‐19 pandemic resulted in no significant changes in cSCC incidence across Europe. By contrast, NMSC incidence fell across Europe following the pandemic. Significant reductions in pooled NMSC incidence rates may reflect a delay in basal cell carcinoma presentation, diagnosis and treatment. Although annual incidence rates for cSCC were not affected by the pandemic, delays in treatment may still have occurred, which may result in poorer outcomes yet to be fully understood.

Significant reductions in pooled NMSC incidence rates may reflect a delay in basal cell carcinoma presentation, diagnosis and treatment.Although annual incidence rates for cSCC were not affected by the pandemic, delays in treatment may still have occurred, which may result in poorer outcomes yet to be fully understood.

| BACKGROUND
4][5][6][7] Despite this, NMSC incidence was the most common cancer globally in 2017, with cSCC incidence estimated at 2.4 million in 2019. 8,9The rising incidence of cSCC has a high economic burden. 2,10,11he COVID-19 pandemic severely impacted public and private healthcare systems globally. 12To reduce COVID-19 mortality, countries implemented lockdowns, promoted stay at home campaigns, and advised vulnerable populations such as the elderly or immunocompromised to shield themselves. 13,14Hospital staff were redeployed to frontline services and cancer services were restricted. 15This was associated with a sharp decline in all cancer diagnoses in April 2020, with cancer diagnoses recovering to pre-pandemic levels by June-October 2020 according to one global systematic review. 168][19][20][21][22][23] Factors such as the public fearing to attend medical care and restrictions in accessing healthcare for non-urgent appointments may have influenced this decline. 18,19Moreover, organisations such as the American College of Mohs Surgery advised delaying lower risk surgeries for up to 3 months to balance COVID-19 risks with cancer outcomes. 13][26][27][28][29][30] The World Health Organisation (WHO) NMSC incidence report in 2020 extrapolated pre-pandemic data, therefore current international NMSC incidence rates are unknown. 31This systematic review assessed whether cSCC and NMSC incidence rates in 2020 and 2021 changed due to the impact of the COVID-19 pandemic.

| Registration
The protocol was registered on PROSPERO (ID: CRD42022376497).The review was reported according to the Meta-analyses Of Observational Studies in Epidemiology statement.

| Eligibility criteria
Studies were included if they reported cSCC and/or NMSC incidence rates, or provided adequate data to calculate incidence rates, both before and after 1 January 2020, which aligns with WHO's recognition of SARS-CoV-2 cases. 12No restrictions were applied on

What is already known about this topic?
� Incidence rates of cutaneous squamous cell carcinoma (cSCC) and Non-melanoma skin cancer (NMSC) are high and increasing in most countries worldwide.� NMSC incidence data are often poorly recorded worldwide.� The COVID-19 pandemic impacted healthcare systems globally resulting in reductions in recorded incidence of some cancers.� The impact of the COVID-19 pandemic on NMSC incidence rates is not well reported.

What does this study add?
� cSCC incidence rates showed no significant change in 2020 and 2021 compared with 2019 in 13 countries across Europe.� NMSC incidence rates were lower in 2020 and 2021 compared to 2019 in 13 countries across Europe.This suggests there may have been delayed diagnosis of non melanoma skin cancers other than squamous cell carcinoma in these countries.
study design.Studies on specific populations such as genetic syndromes or renal transplant patients were excluded.Studies only providing estimated incidence rates after 1 January 2020 were excluded.Studies reporting data in the same population occurring at the same time point were assessed, with only one study being included.

| Literature search
A comprehensive search was performed on 23 March 2023.Four electronic databases were searched systematically using date restrictions between 2020 and March 2023 (MEDLINE, CINAHL, EMBASE, Web of science).Date restrictions were applied to meet the eligibility criteria. 32Search strategy focused on the main concepts: cSCC, NMSC, COVID-19, and epidemiology. 33Grey literature was searched on Google Scholar and Google.Thirty-three countries, classified with high-quality epidemiological data by the Global Burden of Disease (GBD) study, were screened using Google for cancer registries or government reports. 34o language restrictions were imposed on search strategy results and translations were obtained.See Table S1 for full search strategy.After removing duplicates, two reviewers (JW, KGD) independently screened the titles and abstracts of all studies to assess eligbility.Both reviewers independently conducted full-text reading on the studies considered potentially eligible.Uncertainties were resolved by a third reviewer (ZV/NJL).

| Data extraction
Full data extraction was conducted independently by two reviewers (JW, KGD).A data extraction template was created for the review and was used to extract: author, year, country, setting, study design, sex, diagnosis, method of confirmation, and crude and/or agestandardised incidence rates.Corresponding authors were emailed for missing information and non-English studies were translated.Discrepancies between reviewers were resolved via a third reviewer (ZV/NJL).

| Quality assessment
After a pilot assessment, a quality assessment was conducted independently by two reviewers (JW, KGD) using a tool previously by published by Lomas et al. 2 The tool consisted of 10 questions, each equally weighted.Discrepancies were resolved with a third reviewer (ZV/NJL).

| Synthesis of results
Two primary outcomes compared cSCC and/or NMSC incidence rates during the 2020-21 COVID-19 pandemic with 2019 incidence rates to determine incidence rate ratios.Incidence rates are expressed per 100 000 person-years.Separate analyses for cSCC and NMSC data were performed, and studies reporting both were included in both analyses.Studies reporting the International Classification of Diseases, 10th Edition (ICD-10) code 'C44 excluding BCC' were included in the cSCC analyses. 35

| Meta-analysis
Meta-analyses were performed on crude incidence rate ratios (CIRR).The inverse variance statistical method was applied, and due to the differential impact of the pandemic between populations, a random effects model was used. 36Four forest plots were generated using Review Manager version 5.4.1. 37Statistical heterogeneity was evaluated using the Higgins I 2 statistic, with considerable heterogeneity indicated if I 2 >75%. 36IRRs were calculated using the formula: The natural logarithm of the CIRR and standard error were calculated and entered into Review Manager. 38

| Narrative synthesis
Age-standardised incidence rate ratio (ASIRR) were narratively synthesised as all studies could not be compared to the same standardised population structure due to inadequate age-group specific incidence data between included studies.ASIRRs were calculated using the formula: � ðcSCC age−standardised incidence rate during COVID−19Þ ðcSCC age−standardised incidence rate in 2019Þ Statistical differences between sex were determined using Poisson method.

| Study selection
The study selection process is outlined using a PRISMA flow diagram (Figure 1). 39A total of 14 studies were included in the review.

| Study characteristics
Table 1 describes the characteristics of studies included.The review encompassed 14 cancer registries across Europe.All studies reported CIRRs, of which 11 provided ASIRRs.The COVID-19 pandemic group included study data recorded between January 2020 and December 2022.The median study duration in the pandemic group was 365 days (IQR = 365 days).cSCC data were reported in six studies, NMSC data in seven studies, and four studies used the ICD-10 C44 classification (NMSC) excluding BCC.Sex-specific data was available in 12 studies.

| Quality assessment
The methodological quality of the studies varied with a range of scores between two and eight.Twelve (86%) were deemed of high quality (score ≥6). 2 However, no study met all the criteria, directly sampled the population, as opposed to using cancer registries, and no study provided information about ethnicity or skin type.All studies included histological verification, as European cancer registries consider the 'most valid basis of diagnosis', such as histology, in counting a cancer case. 7Table S2 reports quality assessment results.

| Crude incidence rate ratios (CIRR)
A summary table of results for CIRRs is provided in Table S3.The meta-analysis results for cSCC-CIRR and NMSC-CIRR are displayed in forest plots in Figures 2 and 3, respectively.All meta-analyses, except one, indicated considerable heterogeneity (I 2 >75%) in each yearly CIRR group.

| cSCC-CIRR
Ten studies compared cSCC incidence in 2020 against 2019.This covered a total population area of approximately 72 million, encompassing eight national studies and two regional studies.Compared with 2019, Catalonia, Spain, reported the greatest decrease cSCC-CIRR of 0.83 (confidence interval (CI) 0.79-0.88),Abbreviations: ASIRR, age-standardised incidence rate ratios; B, both male and female data recorded together; BCC, basal cell carcinoma; CIRR, crude incidence rate ratio; cSCC, cutaneous squamous cell carcinoma; F, female only; M, male only; NMSC, non-melanoma skin cancer.
F I G U R E 2 Forest plot reporting cSCC crude incidence rate ratios: comparing data recorded during 2020 and 2021 against cSCC crude incidence rates in 2019.cSCC, cutaneous squamous cell carcinoma.
WALL ET AL.
The pooled result for cSCC-CIRR was 1.00 (CI 0.94-1.06),suggesting a non-significant change between pooled cSCC crude incidence rates in 2020 compared to 2019.
Six studies compared cSCC incidence in 2021 against 2019.This covered a total sample size of approximately 45 million, encompassing four national and two regional studies.Schleswig-Holstein, Germany, was the only study with a significant decrease in cSCC-CIRR of 0.93 (CI 0.87-0.98).By contrast, Denmark reported 35% higher cSCC crude incidence rates in 2021, compared with 2019 (cSCC-CIRR 1.35, CI 1.29-1.41).The pooled result of cSCC-CIRR was 1.08 (CI 0.98-0.1.19),representing a non-significant 8% increase in cSCC crude incidence rates in 2021 compared with 2019.Scotland and Catalonia, Spain, reported significantly higher cSCC-CIRR in 2021 compared with 2020.

| NMSC-CIRR
Seven studies compared NMSC incidence in 2020 against 2019 which included four national, two regional, and one local study.A total population area of approximately 80 million was captured.Northern Ireland reported the greatest decrease in NMSC-CIRR (0.74, CI 0.71-0.77),whereas Belgium had a smaller but still significant decreases in NMSC-CIRR (0.92, CI 0.91-0.94).The pooled NMSC-CIRR was 0.83 (CI 0.77-0.89).This suggested a significant reduction in NSMC crude incidence rates in 2020 compared to 2019.

| Age-standardised incidence rate ratios (ASIRR)
A summary table of results for ASIRR is provided in Table S4.Eleven cancer registries in Europe were included.All studies reported male and female data separately.
Four studies compared cSCC incidence in 2021 against 2019 which included three national and one regional study.The standard populations reported Figure 4b demonstrates that cSCC-ASIRR scores ranged between 0.90 and 1.10 in males and between 0.96 and 1.09 in females.All studies reported higher cSCC-ASIRRs in 2021 compared to cSCC-ASIRRs in 2020, with the greatest increase observed in Scotland.Two studies reported cSCC-ASIRR for the whole population.Netherlands reported a cSCC-ASIRR of 1.00 while Scotland reported a cSCC-ASIRR of 1.09.

| NMSC-ASIRR
Six studies compared NMSC incidence in 2020 against 2019 which included four national and two regional studies.The standard populations reported were: ESP 2013 (n = 4) and ESP 1976 (n = 2).Figure 4c demonstrates that NMSC-ASIRR decreased comparably in both sexes.ASIRR scores ranged between 0.75 and 0.92 in males and between 0.69 and 0.92 among females.Four studies reported NMSC-ASIRR for the whole population.The NMSC-ASIRR scores ranged between 0.73 and 0.92.The lowest and highest NMSC-ASIRR was reported in Northern Ireland and Belgium, respectively.
Two studies compared NMSC incidence in 2021 against 2019 (Figure 4d).Scotland and Schleswig-Holstein.NMSC-ASIRR scores ranged between 0.90 and 0.93 in males and between 0.86 and 0.90 among females.

| DISCUSSION
This systematic review assessed 14 studies on the impact of the COVID-19 pandemic on cSCC and NMSC incidence rates.
The pooled result for cSCC-CIRR reported a 0% change between 2020 and 2019 crude incidence rates.By contrast most cancers, of any type, had been reported as having lower incidence rates in 2020. 40,41The presentation of cSCC is rapid in onset and often painful therefore diagnosing most cases of cSCC either face-toface or via telemedicine is not challenging for experienced clinicians.Delayed presentation during the pandemic may have been less likely for cSCC than for other cancers with less obvious visual or sensory impact resulting in a lower threshold for seeking urgent medical attention.However, before the pandemic, cSCC incidence increased consistently year on year in most populations globally, and predictions indicated continued upward trends. 2,10,42Given this trend, no change between cSCC incidence in 2019 and 2020 may reflect a reduction in cSCC presentation by patients.Furthermore, our findings suggested lower cSCC-ASIRR in males than females in 2020, compared to 2019 although this was non-significant.This requires further research but may be due to differences in health seeking behaviours between sexes during the pandemic. 43Delays in cSCC could lead to fatal outcomes. 44ooled estimates indicated significant reductions in NMSC incidence rates in 2020 and 2021, compared with 2019.However pooled estimates for changes in cSCC incidence rates lacked significance.Approximately 75% of NMSC cases constitute BCC, suggesting the reductions mainly involved BCC. 45The pandemic may have impacted BCC and cSCC incidence differently due to differences in care pathways. 46,47BCCs may have been managed within more routine care pathways which faced larger delays. 48,49By contrast, cSCC may have prioritised on emergency and cancer pathways during the pandemic. 48,49Furthermore, BCC's slower development and less symptomatic growth than cSCC could have led to fewer individuals seeking medical attention. 47Lastly, BCC is often diagnosed incidentally during whole body skin examinations, which were less frequent during the pandemic. 50,51igher cSCC and NMSC incidence rate ratios were observed in 2021 compared with 2020, suggesting a trajectory towards pre-pandemic levels of healthcare service provision.This trend is supported by WHO data which indicated reduced excess mortality in 2021, a measure used to assess the impact of the COVID-19 pandemic. 52,535][56] cSCC may be more common in frail individuals who had high COVID-19 mortality which may have partially influenced the impact of the pandemic on cSCC incidence rates. 47Despite this, cSCC incidence rates in 2021 were not significantly higher than 2019, although this could be due partially to a lack of reported data.
Several studies assessing the earlier months of 2020 have reported an increased proportion of higherrisk cSCCs, compared to 2019 data. 17,23,57Various factors were associated with a larger tumour diameter in these studies.This included older age, residency in nursing homes, limited exposure to skin cancer campaigns during the lockdowns and and delays in patients seeking initial GP consultations. 17,57Conversely, studies assessing the longer term impact of the pandemic on cSCC tumour characteristics have yielded heterogenous results.A Netherlands study analysing national data revealed no significant changes in cSCC tumour stage distribution, whereas several local studies in Europe have reported an increased number of thicker cSCC tumours, therefore potentially impacting patient outcomes. 25,58,59

| Strengths and limitations
CIRRs do not adjust for differences in age structure between studies.However, comparing closely matched dates in the meta-analyses minimises potential changes in age structures.Additionally, higher mortality rates observed among individuals over 65 during the pandemic may offset the increasing age trend in populations, mitigating potential bias introduced. 60ASIRRs outcomes removed the confounding effect of age on NMSC diagnosis, and scored higher in the quality assessment as most standardised data to a major population, increasing the external validity. 61However, fewer studies were available for analysis, leading to less conclusive and generalisable results.
Considerable statistical heterogeneity (I 2 >75%) was observed in most pooled estimates suggesting variability in the effect estimates rather than random error alone. 36Clinical heterogeneity arised from the inclusion of various countries and settings, each with different population densities and varying lockdown periods. 62ethodological diversity is present due to variation between countries in the quality of recording of skin cancer data.Cancer registries may underestimate incidence by not recognising multiple cSCCs in one individual. 7Furthermore, challenges in data collection during the early stages of the pandemic may further contribute to information bias. 63Exploration of heterogeneity was restricted as most included limited demographic data, some studies were relatively small and some meta-analyses included a limited number of studies. 36he search strategy aimed to have a global representation but all of the publications which met the eligibility criteria were from Europe.The grey literature search focused on 33 countries, selected based on the 5-star mortality data rating from the GBD and mainly included high-income, white ethnicity countries. 34This limits how representative the data may be on a global perspective not only because incidence rates vary significantly globally for cSCC but also the response to the pandemic and healthcare service provision varied.Furthermore, data collection from private healthcare to cancer registries is unclear, possibly underestimating incidence during the pandemic considering increased private healthcare utilisation was reported. 64nother limitation is that a publication bias assessment wasn't performed.Many cancer registries were included, which follow strict publication policies, and the limited number of cohort studies in each primary outcome prevented Egger's Test or funnel plot analysis. 65,66he COVID-19 pandemic led to higher mortality rates in the frail elderly population who are more susceptible to developing cSCC. 59As a result we might have expected cSCC incidence to fall because of this effect but observed data did not confirm this.Therefore due to potential bias the results should be interpreted with caution.
The long-term impact of the pandemic on cSCC incidence remains unclear.Global travel restrictions may reduce ultraviolet exposure and cSCC incidence, while encouragement for outdoor exercise may increase cSCC incidence. 67Other vairables such as changes in sun protection behaviours and climate change may also impact cSCC incidence, complicating attribution to the pandemic alone. 68

| Future research
There was a lack of high quality cSCC data from countries outside Europe.Further research should compare studies using age-standardised incidence rates with the same major population structure enabling more accurate comparisons between countries.Furthermore, future studies should assess the potential impact of delayed cSCC diagnoses on patient outcomes, aligning with James Lind Alliance priorities. 69his should include assessing the effectiveness of teledermatology in managing diagnostic delays, given its limitations such as accessibility. 30][72] Lastly, alignment of NMSC epidemiology with other invasive diseases is required.Governmental support is vital in improving electronic record systems and standardised data collection globally.

| CONCLUSION
In conclusion, evidence indicates that the impact of the COVID-19 pandemic on cSCC incidence varied across Europe but appeared to be reducing with time.The available data showed no difference in cSCC crude incidence rates between 2020 and 2019 and suggested a non-significant increase of 8% in 2021 compared to 2019.Significant reductions were reported in NMSC incidence rates in 2020, suggesting a backlog in BCC cases.The generalisability of these findings is limited due to high statistical heterogeneity, a lack of age-standardised data, and the scarcity of cSCC incidence rates recorded during the pandemic outside of Europe.Further research is needed to explore the longterm impact of the pandemic on cSCC outcomes.

F
I G U R E 1 PRISMA flow diagram of literature search and study selection process.A search was conducted on 23rd March 2023, including grey literature and four bibliographic databases.Three thousand, four hundred forty-seven studies were initially screened with 14 studies meeting the eligibility criteria and included for analysis.

F I G U R E 3
Forest plot reporting NMSC crude incidence rate ratios comparing 2020 and 2021 crude incidence rates against NMSC crude incidence rates in 2019.NMSC, non-melanoma skin cancer.F I G U R E 4 Summary of findings for cSCC and NMSC age-standardised incidence rate ratios.(a) Comparing cSCC age-standardised incidence rates in 2020 against 2019 in eight studies; (b) Comparing cSCC age-standardised incidence rates in 2021 against 2019 in four studies; (c) Comparing NMSC age-standardised incidence rates in 2020 against 2019 in six studies; (d) Comparing NMSC age-standardised rates in 2021 against 2019 in two studies.cSCC, cutaneous squamous cell carcinoma; NMSC, non-melanoma skin cancer.WALL ET AL. were: ESP 2013 (n = 2) and ESP 1976 (n = 2).
Summary of characteristics of all included studies.