Growing Deficit in New Cancer Diagnoses 2 Years Into the COVID-19 Pandemic: A National Multicenter Study

Abstract

Background.

Large decreases in cancer diagnoses were seen early in the COVID-19 pandemic. However, the evolution of these deficits since the end of 2020 and the advent of widespread vaccination is unknown.

Methods.

This study examined data from the Veterans Health Administration (VA) from 1 January 2018 through 28 February 2022 and identified patients with screening or diagnostic procedures or new cancer diagnoses for the four most common cancers in the VA health system: prostate, lung, colorectal, and bladder cancers. Monthly procedures and new diagnoses were calculated, and the pre-COVID era (January 2018 to February 2020) was compared with the COVID era (March 2020 to February 2022).

Results.

The study identified 2.5 million patients who underwent a diagnostic or screening procedure related to the four cancers. A new cancer was diagnosed for 317,833 patients. During the first 2 years of the pandemic, VA medical centers performed 13,022 fewer prostate biopsies, 32,348 fewer cystoscopies, and 200,710 fewer colonoscopies than in 2018–2019. These persistent deficits added a cumulative deficit of nearly 19,000 undiagnosed prostate cancers and 3300 to 3700 undiagnosed cancers each for lung, colon, and bladder. Decreased diagnostic and screening procedures correlated with decreased new diagnoses of cancer, particularly cancer of the prostate (R = 0.44) and bladder (R = 0.27).

Conclusion.

Disruptions in new diagnoses of four common cancers (prostate, lung, bladder, and colorectal) seen early in the COVID-19 pandemic have persisted for 2 years. Although reductions improved from the early pandemic, new reductions during the Delta and Omicron waves demonstrate the continued impact of the COVID-19 pandemic on cancer care.

As soon as the COVID-19 pandemic necessitated restrictions on elective health care in March 2020,^1,2 concerns were expressed about the impact of the pandemic on cancer care.³ Reports have confirmed the anticipated reductions in diagnosis and treatment of patients with cancer, showing decreases in new diagnoses of 25% to more than 50%. These studies have focused primarily on the early months of the pandemic.^3–7 The decreases in cancer diagnosis and care have been observed through the end of 2020.⁸

In a nationwide study of the Veterans’ Affairs (VA) Health Care System, our group demonstrated a 45% decrease in colonoscopies in 2020 compared with previous years and an associated 20% drop in new colon cancer diagnoses.⁹ Other cancers, including prostate, lung, and bladder cancers, had drops of 13% to 23% in new diagnoses.

Since the arrival of COVID-19 and the initial shutdown, widespread waves of COVID-19 variants, including Delta, Omicron, and other localized surges have caused renewed concerns about seeking routine health care. The impact of these waves on health care systems’ and patients’ cancer detection activities, such as cancer screening or diagnostic studies for concerning symptoms, has not been well studied since the early pandemic. Increasing adoption of vaccinations¹⁰ and natural immunity from prior infection, with attendant drops in COVID-related hospitalization rates, could result in the return of patients and health systems to cancer detection rates approaching levels existing before COVID despite periodic increases in COVID-19 caseloads. Conversely, recurrent surges and continued reluctance of patients to enter high-risk hospital environments may perpetuate the deficits observed in the early pandemic months.

Understanding the continuing effect of the pandemic on cancer care will provide insight into the risk for undiagnosed cancer and the need for additional efforts to recruit patients back to normal cancer screening and diagnostic programs, focusing on cancers with the greatest deficits. It will also inform expectations about the potential delayed surge in late-stage cancer presentations.

Prior studies have examined changes to cancer care in the early pandemic, but no study has examined how these trends have changed or persisted 2 years into the pandemic. We examined changes in diagnostic and screening procedures as well as new cancer diagnoses from the beginning of the pandemic through the peak of the COVID-19 Omicron wave in early 2022. We studied the four most common cancers in the VA population: prostate, lung, bladder, and colorectal cancers.¹¹ We examined the temporal trends in cancer-related diagnostic or screening procedures and new cancers, the backlog of unperformed procedures and undiagnosed cancers accumulated from the start of the pandemic through the Omicron wave, and the geographic variation in these deficits across the United States.

METHODS

Data Sources

Data from the VA’s electronic medical record (Computerized Patient Record System [(CPRS]) stored in the Veterans Affairs Informatics and Computing Infrastructure (VINCI) allows for nationwide, real-time, patient-level analysis of more than 9 million veterans treated at 1298 VA medical facilities.¹² Encompassing all in- and outpatient visits in the VA health system, VINCI provides administrative, demographic, and clinical information on a diverse set of patients and a unique opportunity to study cancer trends in the United States. The Institutional Review Board of the University of Maryland School of Medicine and the Baltimore VA Medical Center Research and Development Committee approved this study.

Study Population

The study enrolled all patients with a clinical encounter in a VA facility from 1 January 2018 through 28 February 2022. Encounters were searched for International Classification of Diseases-10 (ICD-10) codes related to the four cancers: prostate, lung, bladder, and colorectal (Table S1). We also identified Current Procedural Terminology (CPT) codes (Table S2) for procedures that could lead to the diagnosis of these cancers, including prostate biopsies, screening or diagnostic chest computed tomography (CT) scans, cystoscopies, colonoscopies, sigmoidoscopies, and fecal occult blood tests (FOBTs).

The CPT codes were able to identify FOBTs for colon cancer screening and computed tomography (CT) scans for lung cancer screening. Because the CPT codes indicating a colonoscopy for colon cancer screening were so infrequent, we did not attempt to distinguish types of colonoscopies, which led to our inclusion of colonoscopies for non-cancer indications. Similarly, we could not confirm a specific concern for cancer in non-screening CT scans, prostate biopsies, or cystoscopies. We ensured that no patient in our cohort had a prior diagnosis for the same cancer before the study period by reviewing encounters from 1 January 2016 to 31 December 2017. The outcomes of interest were the number of cancer-related diagnostic or screening procedures (defined by CPT codes) and new cancer diagnoses (defined by ICD-10 codes).

Statistical Analyses

The outcome measures were grouped by the month and year of the patient encounter. Using pre-COVID data from 1 January 2018, to 31 December 2019, a monthly baseline, accounting for seasonal variation, was calculated. We then calculated normalized monthly encounters from 1 January 2020 to 28 February 2022 as a percentage of the pre-COVID baseline. The annual deficit in the 26 months spanning 2020 to 2022 for each outcome was calculated as a raw number and as a percentage of the baseline annual volume. Annual deficits were calculated for each state to examine geographic variation (including the District of Columbia and Puerto Rico).

A heat map of the United States for each procedure was generated to examine geographic variation by state. Observed-to-expected (O/E) ratios for each 6-month period since the pandemic began were calculated for each procedure in each state. Funnel plots compared procedural O/E ratios in each state, with 95% confidence intervals calculated based the expected number of cases performed. Scatter plots, linear regression, and Pearson’s correlation coefficient were used to evaluate the correlation of the O/E ratios for screening and diagnostic procedures with the O/E ratios for new cancer diagnoses by state and month during the COVID era. All statistical analyses and graphs were generated with R (v. 4.0.2; R Foundation for Statistical Computing, Vienna, Austria).

RESULTS

From 1 January 2018 to 28 February 2022, we identified 2.5 million patients undergoing diagnostic or screening procedures potentially related to our cancers of interest (prostate, lung, colorectal, and bladder cancers), including 1.7 million in the pre-COVID era and nearly 800,000 in the COVID era (Table 1). Compared with the patients undergoing these procedures in the pre-COVID era (1 January 2018 to 29 February 2020), the patients in the COVID era (1 March 2020 to 28 February 2022) tended to be younger (38% vs 27% younger than 60 years). The patients in the COVID era were predominantly male (90%), with 23% black and 7.2% Latino patients. For more than 300,000 patients, a new cancer was diagnosed during the study period (180,870 in the pre-COVID era vs 136,963 in the COVID era), including prostate (57%), lung (20%), bladder (12%), and colorectal (11%) cancers.

TABLE 1.

Demographic characteristics of patients undergoing screening or diagnostic procedures for cancer in 2018–2022

Variable	Pre-COVID n (%)	COVID n (%)	P value
n	1,739,497 (68.8)	790,568 (31.2)
Age: years (SD)	67 (59–72)	64 (53–72)	<0.001
Age category (years)			<0.001
<40	54,095 (3.1)	41,848 (5.3)
40–49	73,250 (4.2)	76,606 (9.7)
50–59	347,131 (20)	179,256 (22.7)
60–69	620,682 (35.7)	213,263 (27)
70+	644,339 (37)	279,594 (35.4)
Sex			<0.001
Female	130,573 (7.5)	81,360 (10.3)
Male	1,608,924 (92.5)	709,208 (89.7)
Race			<0.001
White	1,242,854 (77)	535,358 (73.7)
Black	350,059 (21.7)	168,808 (23.2)
Other	20,730 (1.3)	22,450 (3.1)
Latino	100,724 (6)	51,188 (7.2)	<0.001
Location			<0.001
Inpatient	11,152 (0.6)	5735 (0.7)
Outpatient	1,728,345 (99.4)	784,833 (99.3)
Procedure			<0.001
Prostate biopsy	33,164 (1.9)	12,401 (1.6)
CT chest	530,313 (30.5)	216,405 (27.4)
CT chest screening	93,698 (5.4)	35,561 (4.5)
Cystoscopy	89,140 (5.1)	32,727 (4.1)
Colonoscopy	413,980 (23.8)	161,232 (20.4)
FOBT	579,202 (33.3)	332,242 (42)

CT computed tomography, FOBT fecal occult blood test

Diagnostic and screening procedures declined sharply from March to May of 2020, then rebounded to near baseline levels between June and September of 2020 (Fig. 1A). Colonoscopies were an exception, as they never fully rebounded.

FIG. 1.

Temporal trends in monthly diagnostic and screening procedures in the Veterans’ Affairs (VA) health system from 2018 to 2022 as (A) an absolute number and (B) a percentage of the 2018–2019 baseline. The estimated (C) monthly and (D) cumulative number of unperformed cases based on the 2018–2019 seasonally adjusted baseline is presented. The gray, shaded region indicates the pre-pandemic era, defined as before 11 March 2020

In the COVID era, only two procedures (FOBT and CT chest screening) achieved rates that exceeded those in the pre-COVID era. When normalized to a seasonally adjusted percentage of the 2018–2019 procedure volume (Fig. 1B), we found that colonoscopies, prostate biopsies, and cystoscopies never attained a period of procedure volume exceeding 100% to “make up” for procedures not performed during March to May of 2020. In fact, these procedures never returned to 2018–2019 baseline levels. They accrued monthly deficits of hundreds to tens of thousands (in the case of colonoscopies) during most months after the pandemic began (Fig. 1C). These monthly deficits resulted in large cumulative deficits over the 2 years since the pandemic began (Fig. 1D). During the first 2 years of the pandemic, VA medical centers performed 13,022 fewer prostate biopsies, 32,348 fewer cystoscopies, and 200,710 fewer colonoscopies compared with two pre-COVID years (2018–2019). This equates to a cumulative 2-year deficit of 48%, 33%, and 68% of annual volume, respectively.

When these temporal trends were evaluated by state, we observed substantial variations across states and 6-month periods from the start of COVID. Heat maps of percentage change in colonoscopies performed by state showed uniform decreases of 50% or more from baseline during the first 6 months of the pandemic (Fig. 2A). Although these deficits improved across the United States over the next 18 months (Fig. 2B–D), most states remained below baseline throughout the pandemic. States such as Wisconsin and Colorado were able to achieve near normal colonoscopy volume and even exceed their 2018–2019 baseline numbers later in the COVID era, but still accrued respective 2-year deficits of 26% and 34% of annual colonoscopy volume during 2 years. Mississippi, Louisiana, and Arkansas always had a deficit of more than 20%, and many periods had deficits greater than 50%, leaving these states with a cumulative 2-year COVID deficit of 138%, 91%, 122%, respectively.

FIG. 2.

Heat map of the change in the number of colonoscopies performed for cancer in 2020–2022 compared with the seasonally adjusted 2018–2019 baseline in each state by 6-month period from the start of the pandemic. A March to August 2020. B September 2020 to February 2021. C March to August 2022. D September 2021 to February 2022. Green states represent a relative increase in procedures performed in 2020, and pink indicates a relative decrease in procedures performed. Gray indicates no data for that state

Computed tomography scans of the chest (both screening and diagnostic) showed deficits during the first 6 months of the pandemic, but rebounded to greater than baseline volumes after this period (Fig. S1). Prostate biopsies and cystoscopies showed patterns similar to those of colonoscopies, although the deficits generally were not as severe (Figs. S2 and S3). We could not identify consistent geographic patterns among these four diagnostic and screening procedures.

The findings of wide variation among the states were confirmed in funnel plots of O/E ratios of 6-month procedure numbers. For colonoscopies, only 20% of the states fell within the 95% confidence intervals calculated based on their expected volume, which persisted in all 6-month intervals through February 2022. The funnel plots also demonstrated the dramatic decrease in colonoscopies during the early pandemic (March to August 2020), in which only 33% of expected colonoscopies were performed (Fig. 3A). Although procedures increased during the next 18 months, O/E ratios remained below baseline from September 2020 to February 2021 (O/E ratio, 0.69), from March to August 2021 (O/E ratio, 0.86), and from September 2021 to February 2022 (O/E ratio, 0.81) (Fig. 3B–D). Computed tomography scans of the chest dropped only to 74% of the expected value in the first 6 months of the pandemic, then returned to baseline or higher (Fig. S4). Prostate biopsies and cystoscopies showed trends similar to those for colonoscopies (Figs. S5 and S6).

FIG. 3.

Funnel plots of observed-to-expected ratio of diagnostic and screening procedures performed for cancer in 2020–2022 by expected number of procedures (based on 2018–2019 data) per state for colonoscopies in (A) March to August 2020, (B) September 2020 to February 2021, (C) March to August 2022, and (D) September 2021 to February 2022

New diagnoses of cancer showed a temporal trend similar to that seen for diagnostic and screening procedures. There was a dramatic decrease from March to June 2020, and a rebound toward baseline levels by September 2020 (Fig. 4A). When examined as a percentage of diagnoses normalized by 2018–2019 baseline data, the pattern of decreased diagnoses was consistent across prostate, lung, bladder, and colorectal cancers. Two years into the pandemic, overall cancer diagnoses remained consistently below pre-COVID levels (Fig. 4B). These persistent monthly deficits (Fig. 4C) continued adding to a cumulative deficit of nearly 19,000 undiagnosed prostate cancers and 3300 to 3700 undiagnosed cancers each for the lung, colon, and bladder (Fig. 4D). These deficits are equivalent to respectively 19%, 10%, 15%, and 17% of the annual new diagnoses of prostate, lung, bladder, and colorectal cancer in the VA health system.

FIG. 4.

Temporal trends in new cancer diagnoses in the Veterans’ Affairs (VA) health system from 2018 to 2022 are presented as (A) monthly cases and (B) percentage of seasonally adjusted baseline. The estimated (C) monthly and (D) cumulative number of undiagnosed cancers in 2020–2022 are shown. The gray, shaded region indicates the pre-pandemic era, defined as before 11 March 2020

When correlating the O/E ratios of diagnostic or screening procedures performed for cancer and new diagnoses of cancer by state and month, we found a weak but significant correlation in all four cancer types (Fig. 5). The correlation between prostate biopsy and new prostate cancer was strongest (R = 0.44; p < 0.01), followed by cystoscopy for bladder cancer (R = 0.27; p < 0.01). Colorectal cancer and lung cancer had weaker correlations.

FIG. 5.

Scatterplot demonstrating the correlation between the observed-to-expected (O/E) ratio for procedures performed to diagnose a cancer (i.e., colonoscopy) and the O/E ratio for new diagnoses of that cancer (i.e., colorectal cancer) by state and month. Pearson’s correlation coefficient and the p value for each cancer are listed

DISCUSSION

This first study to examine the impact of COVID-19 after widespread vaccinations and innate immunity found find that 2 years into the pandemic, deficits in cancer diagnostic procedures and diagnoses of new cancers continued to grow. The VA health system performed 200,000 fewer colonoscopies in the 2 years after the pandemic started than in the 2 years before the pandemic, including a deficit of 5872 in February 2022 during the Omicron wave of COVID-19. Similarly, we diagnosed 30,000 fewer cancers in the first 2 years of the pandemic than in the 2 years before. Although subsequent COVID-19 waves did not have the same impact on procedures and diagnoses as the initial moratorium at the start of the pandemic, large deficits in procedures again accrued during the Omicron wave. This finding confirms the continued ability of new COVID-19 variants to have an adverse impact on cancer detection efforts even 2 years into the pandemic. This study also linked decreased use of screening and diagnostic procedures with real decreases in new cancer diagnoses, although the correlation is weak.

Although interest in COVID-19 may be fading from public view and media attention, the continued disruptions to cancer detection efforts, even late into the pandemic, are disturbing. Disruptions to cancer care early in the COVID-19 pandemic were to be expected and have been well documented.^{4,5,8,13–16} Although concerns were raised about the potential long-term consequences of these disruptions,^17,18 attention to this issue has waned as interest in the pandemic decreases and social activities normalize. The current report indicates that cancer detection efforts have not normalized, and there is continued avoidance of more invasive procedures for cancer screening and diagnosis.

Conversely, CT scans of the chest, particularly screening CT scans for lung cancer, increased above baseline levels during the later study period. Screening CT scans had already been increasing in use before the pandemic. Therefore, our findings may reflect a continuation of a pre-existing trend unrelated to the pandemic in response to studies that have confirmed the utility of CT scans in improving outcomes of select populations at risk for lung cancer.¹⁹

In the COVID era, FOBTs also showed increased use. It is intuitive that low-risk procedures with limited resource strain on hospitals and requiring minimal inter-personal contact were more resilient, and these activities bounced back after the early surge in the pandemic. To avoid colonoscopies, FOBTs may have been used preferentially. The need for administration of sedation, increased contact with facilities and medical personnel, and increased strain on medical facilities may have created a low threshold for postponing or canceling invasive colonoscopies. Although FOBTs did increase during the pandemic (13,000 more FOBTs were used during the first 2 years of the pandemic), this increase was not sufficient to offset the large drop in colonoscopies (200,000 fewer colonoscopies during the first 2 years of the pandemic).

Colonoscopies have shown especially large deficits during the pandemic. Although prostate biopsies and cystoscopies are also invasive procedures that require close contact with the medical team and may need sedation, these tests may be performed for cause (i.e., to address some new symptoms or concerning findings on other studies). Colonoscopies are unique in that they are invasive, require sedation, and often are screening tests with no signs or symptoms of disease. Asymptomatic patients may have decided that the risks of COVID-19 provided sufficient reason to avoid potentially unnecessary contact with medical facilities.

Concerns have been raised that state or regional policy differences in health care shutdowns and re-openings, COVID vaccination or infection rates, and the emphasis on “return to normal” in different areas would lead to major differences in the rate of normal health care activities.²⁰ These concerns led our team to a careful examination of state differences in diagnostic and screening procedures. Although we did find dramatic differences in rates of procedures between states after accounting for baseline activity, we did not identify important geographic patterns. These findings indicate that these differences were not important or that categorizing by state may be too blunt to identify patterns. Further analysis of county or medical center catchment area may allow for differences to be detected at a more granular geographic level.

We also found continued decreases in new diagnoses of our four cancers of interest. These findings suggest that the reduction in procedures is translating into identification of fewer cancers. This study is the first to demonstrate a correlation, albeit weak, between decreased use of screening and diagnostic procedures and identification of fewer cancers. Prostate biopsies and cystoscopies showed the strongest correlation, as would be expected given that these studies are performed to address a symptom or finding and would have a stronger link to the identification of cancer. Conversely, screening procedures available for lung and colorectal cancer may have led to a smaller correlation, which may have been detected only in a study with a large sample. Lung cancers showed the smallest decrease in new diagnoses, as would be expected, given that CT scans were performed largely at or above pre-pandemic levels. Given the large number of these procedures that likely were not performed for cancer and the large number of screening studies required to identify a single cancer, it is not surprising that these correlations were weak. Random fluctuations in cancer incidence, migration in and out of states, cancer program availability, and hospital staffing changes, many of which may be COVID pandemic related, could each be larger drivers of new cancer diagnoses than the rate of cancer detection procedures.

The current study provided several potential insights. In contrast to studies from the early pandemic, which assumed a rapid return to baseline and worried about a need for excess capacity to make up for missed procedures,^4,15,21,22 this study showed improved but persistent deficits in procedures to detect cancer. The impact of the Omicron wave is especially disconcerting, nearly 2 years after the start of the pandemic.

Colonoscopy has shown the largest deficits, which may be explained by the more intense contact with the health system (e.g., sedation and recovery room stay) compared with CT scans or FOBT. These data indicate that patients may be more willing to re-engage with less invasive screening (e.g., FOBT) as a mitigation strategy for this and future pandemics. In addition, procedures that usually are performed in response to symptoms or prior testing, such as prostate biopsy or cystoscopy, demonstrated less dramatic decreases during the COVID era. However, these procedures also had a stronger correlation with decreased new cancer diagnoses, making even small decreases in procedures potentially more important.

Further research on the mitigating and aggravating factors associated with cancer care disruptions is required. A study from Washington state found greater reductions in breast cancer screening during the pandemic among black, Hispanic, and other racial and ethnic groups than among white patients.²¹ There also were greater reductions in mammograms among rural patients and patients with Medicaid or no insurance. Determining which patient subgroups are at the greatest risk for missing appropriate cancer care may allow focused outreach to communities most at risk for undiagnosed cancer.

We are focusing future research on the following topics: evolving disparities in cancer care of the VA population, comparison of trends and disparities between the VA and other health systems, and changes in cancer-related morbidity and mortality in the COVID era. Unfortunately, studies investigating the impact of cancer screening on cancer-related mortality often require 10 to 20 years of follow-up evaluation to identify a mortality difference.^23–27 Although the large sample available in the VA or other administrative datasets will shorten this timeline, early studies on cancer outcomes may require interim measures such as metastatic disease at presentation or late-stage cancer presentation to detect significant differences. Given the large population in the VA and a pre-COVID rate of metastatic disease at presentation for colorectal cancer of 9.4%, we have calculated the minimal detectable difference in the COVID era to be 10.8% with a power of 0.8 and a significance of 0.05. This difference could be detected by early 2025.

This study must be understood in the context of its limitations. The VA population is older, making cancer more prevalent, so these findings may not be generalizable to non-VA populations. The predominantly male cohort makes it difficult to study breast or cervical cancers. We chose cancers with the highest incidence in the VA health system, and other cancers may show different trends. Because VINCI provides administrative data linked to rich clinical details, miscoding of procedures or diagnoses likely was low but may exist.²⁸ Administrative data did not allow us to identify patients who presented for procedures due to a specific concern for cancer. However, because all these procedures can result in a new diagnosis of cancer and because many indications for these procedures (e.g., bleeding per rectum, hemoptysis, shortness of breath) may be the presenting symptom of a new cancer, we included all screening and diagnostic procedures that may result in a new diagnosis of cancer. This limitation in our methodology likely diluted any correlation of changes in diagnostic and screening procedures with new cancer diagnoses but also exaggerated the number of cancer procedures being missed.

With the VA’s Care in the Community (CITC) program, veterans may obtain care outside the VA system, although prior studies have shown similar trends between the VA and other health care systems.^4,13 Because the VA system never returned to the baseline levels of many procedures, such as colonoscopies and prostate biopsies, it does not appear that the VA system has a large backlog that would have been referred for outside care. Further research is needed to confirm these findings this far into the COVID-19 pandemic.

CONCLUSION

Major disruptions in new diagnoses of cancer among four common cancer types (prostate, lung, bladder, and colorectal) seen early in the COVID-19 pandemic have persisted for 2 years despite the decreased impact of the pandemic. Although deficits in monthly screening and diagnostic tests and new diagnoses of cancer have decreased from the early months of the pandemic, new reductions in diagnostic efforts during the Delta and Omicron waves demonstrated the continued impact of the COVID-19 pandemic on cancer care. Decreases in the use of screening and diagnostic procedures are correlated with identification of fewer new cancers. Health systems need urgent efforts to identify patients and communities at greatest risk and focus outreach to reengage these communities in their routine cancer screening and medical care.