Age and gender patterns in COVID-19: Insights from a Regional COVID-19 testing center

ABSTRACT

Introduction:

Asymptomatic and targeted testing are critical for mitigating the spread of COVID-19. Understanding testing demand is crucial for efficient personnel and resource management. The study analyzes testing data to uncover factors influencing testing demand.

Methods:

Testing and infection data were consolidated into weekly intervals, grouped into pediatric (0–17.99 years), adult (18–64.99 years), and geriatric (65 + years) cohorts, and plotted longitudinally. Changes in overall testing, infectivity, and age-appropriate vaccination availability were analyzed between groups. Simple paired t-table Chi-square testing calculated significance (P < 0.05).

Results:

Of 69,612 COVID-19 tests, 13,447 (19.3%) were pediatric, 48,084 (69.1%) were adult, and 8,081 (11.6%) were geriatric. A high correlation was seen for pediatric vs. adult (0.73; P < 0.05), pediatric vs. geriatric (0.62; P < 0.05), and adult vs geriatric (0.93; P < 0.05). Pediatric testing differed from adult and geriatric testing during peak overall testing (both P < 0.001) and infectivity (P < 0.001 and 0.002, respectively). Pediatric testing increased 1.7-fold (P < 0.001) following immunization availability. Gender differences were seen between all groups with pediatrics being less female (P < 0.001). Changes in overall testing, infectivity, and immunization availability didn’t change gender ratios significantly within or between groups. Pediatric testing was less female for peak overall testing (P < 0.001) and infectivity (P = 0.032).

Conclusion:

Pediatric testing was more associated with times of high infectivity and did not benefit from vaccination availability. Gender differences occurred between groups, but influences did not change gender ratios within each group. Understanding factors altering COVID-19 testing patterns can inform management of resource and personnel allocation for the next public health crisis.

Introduction

COVID-19, first isolated in the Wuhan province of China in December 2019, is primarily spread through human-to-human transmission.[1] In March 2020, the World Health Organization declared this multi-system disease a pandemic with over 750 million confirmed cases and nearly 7 million deaths as of June 2023.[2] Many efforts in research, therapeutic strategies, and vaccinations have occurred to both limit the spread and manage morbidity.[3] Due to varying viral infectivity rates, testing protocols monitoring infectivity have become increasingly more vital. Multiple studies have examined the efficacy of protocol implementation. The success of drive-thru testing has been illustrated as a means of optimizing safety and efficiency.[4,5,6]

Prior research outlines significant barriers to COVID-19 testing. The most common obstacles cited are the cost of testing, discomfort in testing procedures, low health literacy, low healthcare system trust, lack of availability and accessibility of testing sites, stigma, and consequences of positive testing.[7,8] Testing challenges have also been outlined in rural settings.[9] These include unique disparities of economic limitations for healthcare facilities that serve disproportionately uninsured or under-insured patients and an older population with a high degree of chronic illness.[10,11,12,13] The most plausible strategies to mitigate these obstacles include offering free testing, promoting awareness of testing importance, presenting various testing options including home testing, multiple types of testing centers (i.e., drive-thru, walk-up, home testing), providing transportation, and supporting self-isolation (e.g., salary support or housing).[7,14] Previous research has not drawn similar conclusions regarding the best course of action.[15,16] This makes the investigation of testing in specific populations important.

Understanding drivers for testing is complicated but is of utmost importance as it impacts personnel and resource allocation. At the pandemic onset, the number of tests performed weekly spiked.[17] Testing, positive tests, and percent test positivity have mirrored simultaneous peaks in viral symptoms.[18] Many institutions such as schools, hospitals, and workplaces require repeat routine testing of asymptomatic individuals or those returning from vacations.[19,20,21] Some research has identified variables, like observed holidays and compliance with mask-wearing that influence certain population’s testing,[15,16,22] but no comparison has been evaluated across multiple different age groups.

The pediatric population presents a unique challenge due to their lack of autonomy to seek elective medical care. Asymptomatic or mild symptoms such as lethargy, malaise, myalgia, sore throat, runny nose, sneezing, gastrointestinal symptoms, and fatigue,[11,23,24] make differentiation from other seasonal viruses (RSV and influenza) difficult.[25,26] This results in a low threshold for testing which includes: exposure to SARS-CoV-2 infected individuals, parental concern, and asymptomatic screening for travel and surgery.[27,28,29] State and federal requirements for school attendance and sports participation further complicate pediatric testing rates.[30]

Age is an independent risk factor for SARS-CoV2 infection outcomes[31] with the geriatric population accounting for an estimated 81% of all COVID deaths.[10,32] This, along with longer colonization times, influences testing patterns.[33] Furthermore, the geriatric population has specific exposure opportunities that could impact testing frequency such as child care for grandchildren, church-related activities,[34] and assistance in transportation and shopping.[35] All of these increase exposure for healthy, but vulnerable, community-dwelling seniors. Nursing home environments also lead to increased exposure. Healthcare-associated iatrogenic and visitation-specific influences occur in group home living environments, where the intimate nature of care required by seniors makes physical distancing from staff difficult.[36,37] Cognitive disorders exhibited by many limit their understanding of the importance of physical distancing and utilization of personal protective equipment (PPE).[36,38] Close living arrangements and financial challenges limit PPE.[36,39] These unmodifiable risks result in increased exposure and testing, including routine testing of all residents in nursing homes to prevent spread.[36,37]

Gender must be considered as females have more positive attitudes toward seeking medical care.[40] This may represent a heightened anxiety toward perceived risks to personal health, of which the genders differ.[41] Males have been shown to receive less COVID testing than females for age groups except for extremes in age (0–9 and 70–79 years old).[42] Males, however, tend to have worse morbidity and mortality both in this[43] and the previous two coronaviral epidemics,[42] which includes higher rates of positive results, hospitalizations, ICU admissions, and death in the male population.[42] This is thought to be due to both immunological and gendered differences in habits, including mask-wearing, in which females utilize 1.5x more.[22]

Prior research has largely occurred in metropolitan areas[15] and emphasizes the need for efficient and reliable testing facilities.[4] Detailed trends in COVID-19 testing with a large sample size at a rural health center have yet to be studied and are essential to understanding this pandemic and the next. Vulnerable rural populations have been shown to have lower testing rates[17] from decreased testing availability.[17,44] Multi-sector collaboration, additional funding, and high levels of creativity and flexibility are needed to alleviate the challenge of testing vulnerable populations.[45] This study aims to understand how age and gender influence COVID-19 testing patterns in a regional testing center. We predict that testing rates for separate age groups differ within various levels of testing and are influenced by factors such as infectivity and vaccination availability.

Methods

The Family Medicine department of an academic medical center, primarily servicing a rural population, developed a drive-through testing center in their associated hospital’s adjacent parking area. This regional testing center was staffed during business hours five days a week (excluding holidays) from 3/2020 to 4/2022. Test results and de-identified demographic information of gender and age were collected from tested individuals. Age groups were defined as Pediatric (0-17.99 years), Adult (18-64.99 years), and Geriatric (65+ years) to reflect distinct stages of life and likely differences in COVID-19 exposure and healthcare interactions [Figure 1].

Figure 1.

Methodologic flow diagram

Cohort analyses were conducted retrospectively. Data were collected daily, collated into weekly intervals, and analyzed longitudinally [Figure 2]. Longitudinal plots of weekly testing for each age group were compared against the other two groups by assessing Pearson correlation calculations. These calculations were repeated after shifting the plots in one-week intervals in either direction to see if a Lag relationship existed [Figure 3].

Figure 2.

Time series plot age-tiered groups Age-tiered data from the regional testing center is grouped in weekly intervals and displayed longitudinally over the 110 weeks that testing occurred

Figure 3.

Cross-correlation plot for weekly testing between age tier groups. (a) Pediatric vs. Adult, (b) Pediatric vs. Geriatric, and (c) Adult vs. Geriatric. Vertical bars symbolize magnitude of Pearson correlations (y-axis). A zero lag occurs when both age tier groups are analyzed when temporally synchronized (without shifting one curve either forward or backward). A positive lag represents a shift of the weekly testing curve for the first-listed age tier group forward over the second-listed age tier group by the number of weeks represented on the x-axis. Negative lag represents the Pearson correlation when the first-listed curve is moved backward the number of weeks listed on the x-axis. Findings of 0.20 to 0.39 indicate low levels of correlation, while calculations of 0.40 to 0.59 indicate moderate correlation and 0.60+ indicates a high correlation

Taylor’s change point analysis was used to determine where significant breaks in the total weekly testing occurred based on naturally occurring breaks in the results.[46] This was used to define Overall Testing as ‘Low’ if < 350 tests/week, ‘Intermediate’ if 350-900 tests/week, ‘High’ if 900-1300 tests/week, and ‘Peak’ if > 1300 tests/week. Weekly positive percentage rate (PPR) calculations were made by dividing the positive results for the week by the total tests per week. Testing center Infectivity (PPR for the testing center) was defined as ‘Low’ for a PPR of 0-5%, ‘Intermediate’ for a PPR of 5.1-10%, ‘High’ for a PPR of 10.1-20%, and ‘Peak’ for a PPR over 20%. Immunization Availability was defined as ‘No’ for times when age group-specific immunizations were not available and ‘Yes’ for times when they were. Overall change ratios were created for each age group related to these three variables of overall testing, infectivity, and immunization availability. This was done by assigning a baseline level to one variable rating (Low for overall testing and infectivity, and No for immunization availability) and dividing the rates of the other designations by that rate [Figure 4].

Figure 4.

Ratio of Overall Changes. The calculations are based on per-week ratios. The baseline was set as 1.0 (Low Testing and PPR rates and No Immunization Available) and other variables were based on that baseline rate. The specific testing with the groups reveals similarities for Adult NG and Geriatrics (P = 0.947, 0.143, and 0.547). The Pediatrics differed from both Adult (all three P < 0.001) and Geriatric (P < 0.001, 0.002, 0.004)

The three age categories were further separated into female and male designations. Gender differences between the groups were assessed with a simple t-test and Chi-square analysis. Similarly, weekly testing rates were compared between females and males for changes in testing level, infectivity, and immunization variables [Table 1]. Findings were considered significant for P values <0.05. This project was designated exempt by Marshall University’s Institutional Review Board date approved on 8/18/2021.

Table 1.

Gender difference within and between age groups for changing variables

Level	Category	Pediatric		Adult		Geriatric		Chi Square	P Value
		Female	Male	Female	Male	Female	Male
Testing*	Low	49.1%	50.9%	61.0%	39.0%	57.2%	42.08%	2.102	P=0.350
	Intermediate	50.7%	49.3%	58.3%	41.7%	56.8%	43.2%	1.830	P=0.401
	High	52.1%	47.9%	58.8%	41.2%	57.0%	43.0%	2.632	P=0.368
	Peak	50.0%	50.0%	61.01%	38.6%	58.2%	41.8%	14.699	P=<0.001
		0.2663		1.6467		0.0874
		P=0.966		P=0.649		P=0.993
Infectivity^	Low	49.9%	50.1%	56.9%	43.1%	55.4%	44.6%	0.868	P=0.648
	Intermediate	49.0%	51.0%	58.4%	41.6%	55.7%	44.3%	2.6271	P=0.269
	High	50.8%	49.2%	60.0%	40.0%	58.0%	42.0%	4.4265	P=0.109
	Peak	50.9%	49.1%	61.4%	38.6%	57.4%	42.6%	6.8636	P=0.032
		0.1172		1.957		0.1831
		P=0.990		P=0.581		P=0.980
Immunization	No	50.9%	49.1%	57.7%	42.3%	55.6%	44.4%	1.3694	P=0.504
	Yes	50.2%	49.8%	61.0%	39.0%	58.4%	41.6%	5.3444	P=0.069
		0.0144		1.0723		0.142
		P=0.905		P=0.300		P=0.706

Calculations for the data were conducted on real per-week numbers but expressed as percentages. *Overall Testing gender ratio differences (pediatric vs. adult P<0.001, pediatric vs. geriatric P=0.001, adult vs. geriatric P=0.995). ^Infectivity gender ratio differences (pediatric vs. adult P=0.010, pediatric vs. geriatric P=0.233, adult vs. geriatric P=0.384). “Immunization Availability gender ratio differences (pediatric vs. adult P=0.021, pediatric vs. geriatric P=0.259, adult vs. geriatric P=0.667)

Results

Over the 110 operating weeks, 69,612 tests were conducted, with an average of 632.8 tests/week and a weekly range of 22–2,361 tests. Of these, 13,447 (19.3%) were pediatric, 48,084 (69.1%) were adult, and 8,081 (11.6%) were geriatric. The ages of those tested ranged from 1 day to 101 years. Overall, there were 40,003 (57.5%) females and 29,609 (42.5%) males.

High correlations of testing behaviors [Figure 3] were observed at a 0-week lag between pediatric and adult (0.73; P < 0.05), pediatric and geriatric (0.62; P < 0.05), and adult and geriatric (0.93; P < 0.05). As the longitudinal plots were shifted a week, there was a symmetric correlated decrease in both directions (e.g. shifting the pediatric longitudinal plot one week ahead of the adult plot or one week behind). A high correlation of testing behaviors (>0.60) persisted for adult vs geriatric correlation for two weeks in both directions.

Testing ratios [Figure 4] differed between the three groups (P < 0.001). Sub-analyses reveal the pediatric group differed, especially in Peak groups (9.5-fold vs 6.7- and 7.1-fold), from the other age groups (both P < 0.001) while adult and geriatric groups were similar (P = 0.947). Infectivity rates differed between the three groups (P < 0.001). Again, pediatrics differed from both adult (P < 0.001) and geriatric (P < 0.002) groups, while no difference was seen between adult and geriatric groups (P = 0.143). This was due to no change in testing for adults and geriatrics during Low and Intermediate infectivity. Once Age-Appropriate immunizations were available, testing decreased to 0.8-fold for adults and 0.9 for geriatrics but increased to 1.7-fold for pediatrics (P < 0.001).

Age groups were predominately female, however the proportion [Figure 5] differed for each group (P < 0.001). Pediatric patients were 50.4% female (6,783 female vs 6,664 male), while adults were 59.5% female (28,591 females vs 19,493 males) and geriatrics were 57.3% female (4,629 females vs. 3,452 males). Within each age group, there were no differences between genders for any variable: overall testing, infectivity, and immunization availability [Table 1]. However, an even gender distribution of pediatric testing at Peak levels of testing was seen, where adult and geriatric patients were more decidedly female (P = 0.001). Similarly, during Peak times of infectivity, pediatric testing remained more male than adult or geriatric testing (P = 0.032). No gender differences were seen before or after immunization availability for pediatrics (P = 0.905), adults (P = 0.300), or geriatrics (P = 0.706). No identifiable difference was noted between the groups, except the pediatric group was less female than adults (P = 0.021).

Figure 5.

Gender by Age. Gender ratios for pediatric patients differ from both adult (P < 0.001) and geriatric (P < 0.001) patients. The adult gender ratio also differs from the geriatric gender ratio (P < 0.001)

Discussion

Testing volume during an infectious pandemic is important to understand as it helps predict the utilization of resources and personnel. Factors driving testing volume during the COVID-19 pandemic have not yet been studied adequately.[15,16] It is likely that influences such as overall testing rates, infectivity, and immunization availability impact individuals of differing ages and genders independently. Using weekly testing rates allows easier management of a large data set and minimizes daily testing variability (from holidays, inclement weather, and other disruptions). Analyzing how testing rates differ for these variables has yielded the following conclusions.

Age-specific testing patterns show a higher correlation between adults and geriatrics, indicating similar testing patterns in these groups

Time series plots of weekly testing patterns [Figure 1] show some uniformity in testing patterns across age groups. It is important to contextualize the changes seen by events that occurred during those times. The first two peaks, occurring from April 20 to May 20 and June 20 to August 20, largely corresponded to the Governor’s reopening plan and businesses requiring testing for individuals returning from vacations, respectively. The following three spikes in testing related to Alpha (October 20 to February 21), Delta (August 21 to November 21), and Omicron (December 21 to March 21) variants. Purely visible analysis is blunted by the magnitude of the adult numbers. Correlation analysis is high for each of the age groups at 0-lag and falls off symmetrically as a one-time series plot is lagged in either direction [Figure 2]. The closest correlation is with adults and geriatrics (0.950) and maintains a high level of correlation even lagging two weeks in either direction (0.705 and 0.690). Lack of autonomy and different influencing factors (school vs occupation) likely contribute to decreased pediatric correlation comparatively.[27,28,29,30]

Pediatric testing changes associate more with infection rates than adults or geriatrics

Despite correlation in age-specific testing patterns, differences occur between the fold-increases in testing rates (P < 0.001) [Figure 5]. Larger changes between High overall testing times for pediatrics compared to adults and geriatrics (3.9-fold vs. 5.6-fold and 6.0-fold) and Peak overall testing times (9.5-fold vs. 6.7-fold and 7.1-fold) are seen. These are tightly linked to the Delta and Omicron waves. Differences were also noted for pediatrics for infectivity [Figure 4]. Here, increased testing during Intermediate infectivity times is only seen in pediatrics (2.1-fold vs 1.0-fold and 0.9-fold). This creates a stepwise increase in pediatric testing that is not seen in adults and geriatrics (P < 0.001). These differences point toward a higher proportion of adult and geriatric testing during low infectivity and likely represent the alternate rationale for testing (job-related,[19,20,21] living situation,[36,37,39] and anxiety about own health[33]).

Immunization availability was not impactful for the pediatric population

Immunization availability [Figure 4] did not seem to impact pediatric testing rates as opposed to adults and geriatrics (P < 0.001). Testing rates increased following pediatric vaccine availability (1.7-fold). Likely explanations include both attitudes toward pediatric vaccination limiting its utilization and its availability which only occurred during the Peak infectivity of the Delta-variant when testing was highest. No statistical differences were seen between adults and geriatrics after age-appropriate vaccine availability (P = 0.549).

Gender differences are seen between but not within age groups when analyzed for testing, infectivity, and immunization effect

As expected,[40,41] the overall gender ratio was predominantly female (40,003 females vs 29,609 males; 57.5%). This was driven by the adult population [Figure 5], which comprised 48,084 of the 69,612 tests [Figure 4]. Interestingly, the geriatric population, which historically is thought to be more female, was less female than adults (57.3% vs 59.5%; P < 0.001). Pediatrics, however, had an even gender distribution (50.4% female), which occurred when patient choice to seek testing was largely removed from the equation. Children lack the autonomy to access testing themselves, therefore guardians must initiate testing, and this is likely not based on the child’s gender. No gender differences were seen within any age group when testing rates were analyzed by overall testing, infectivity, and immunization availability [Table 1] meaning that these changing variables do not alter gender rates within each group. What is seen is that the stresses of higher overall testing and infectivity push adult (61.1% and 61.4% female) and geriatric (58% and 57.4% female) further female while pediatrics maintains an even gender distribution (50.0% and 50.9% female), widening the gap. This creates significant differences between the three groups during Peak overall testing (P < 0.001) and infectivity (P = 0.032) [Table 1].

This study’s single-site venue limits its generalizability to other regions with different sizes and governance structures. It was conducted at a rural, tertiary care center and its results may not be generalizable to other populations. Also, as a retrospective study, causation cannot be established; rather our findings represent associations. As gender was only collected in a binary fashion, non-traditional gender-identity results were not studied. Furthermore, other factors impacting testing patterns (employment, school status, and sports) may exist. Nevertheless, this work does represent a methodology by which this and future pandemics can be analyzed. Future research should continue to examine drivers for testing during COVID-19 in multiple varied settings, especially in vulnerable populations.

Conclusion

Prior to the COVID-19 pandemic, no one knew how the American public would react to public health strategies designed to mitigate viral spread, like routine and targeted testing. While overall similarities exist between testing patterns between pediatrics, adults and geriatrics, pediatrics are more closely associated with infectivity. Other factors, including vaccination availability, seem more relevant to adults and geriatrics than pediatrics. Also, although gender differences exist between these age-tiered groups, these studied influences do not alter the gender ratios within each group. Heightened levels in testing and infectivity drive these overall gender differences between pediatric and both adult and geriatric populations, further apart. Likely COVID-19 will not be the last pandemic in our lifetime and these findings suggest that public health strategies targeting testing behaviors should consider the unique needs of different age groups to better prepare us for the next pandemic.

Conflicts of interest

There are no conflicts of interest.

Funding Statement

Nil.