This study employed a retrospective observational design to investigate the impact of the COVID-19 pandemic on primary care visits for respiratory illnesses across 9 countries. The study period spanned from January 1, 2018, to December 31, 2021.

Data for this study were gathered from diverse sources, including electronic medical records and billing claims. Specifically, information was sourced from visits to primary care physicians in Australia, Canada, China, Norway, Singapore, Sweden, and the United States. Moreover, data were obtained from primary care clinics in Argentina and Peru, encompassing visits to various healthcare providers within a primary care setting.

The dataset covers the period from January 1, 2018, to December 31, 2021, with the exception of Peru, where visit data was available only from January 2019. Although the onset of the pandemic varied across countries, for the purpose of this study, we defined the pre-pandemic period as January 2018 to March 2020, and the pandemic period as April 2020 to December 2021 in all countries except for China, where the pandemic was declared by the end of January 2020.

The representativeness of the data regarding primary care physician visits varied by country (Supplementary Table S1). Further detailed information about the INTRePID datasets can be found elsewhere, as described in prior publications (16, 17).

The primary outcome was monthly visits for respiratory conditions across different countries, taking into account both virtual and in-person visits and categorizing them based on the type of respiratory condition, regardless of age, gender or other demographic factors. Virtual visits included video-calls and telephone consultations between patients and primary care physicians. Consultations associated with diagnostic codes for respiratory conditions were identified in each country and divided into eight groups: asthma, emphysema/chronic obstructive pulmonary disease (COPD), respiratory allergies, other respiratory diseases, lower respiratory tract infection (LRTI), upper respiratory tract infection (URTI), influenza and COVID-19. Because of variations in reporting, we use the general term influenza to include all reported influenza-like illnesses. COVID-19 includes both suspected and confirmed cases as some coding systems do not differentiate between the two. These groups were also combined to form two major categories, chronic respiratory diseases (asthma, COPD, respiratory allergies, other respiratory diseases) and infectious respiratory diseases (LRTI, URTI, influenza, COVID-19). Singapore data specific to COVID-19 were unavailable in the first year of the pandemic because they were recorded in a different system. See Supplementary Tables S2–S10 for a full description of diagnostic codes and description of the billing coding systems used for categorization in each of the countries.

We conducted an analysis to compare pre-pandemic and pandemic time periods to determine the impact of COVID-19 on primary care for each country. We calculated the difference between the average volume of respiratory monthly visits before and after the onset of the COVID-19 pandemic in each country, along with 95% confidence intervals (CI) and associated p-values using Welch’s t-test, with a significance level of 0.05. This statistical approach was chosen to account for potential variations in sample sizes and variances between the two time periods. Additionally, we computed the percent change in average monthly visits from pre-pandemic to pandemic, along with 95% CI, for all countries and conditions.

Accounting for the variation within the different studied groups, we calculated the standardized difference in means using Cohen’s d test with Hedges correction. This adjustment accounts for biases in small sample sizes. Cohen’s d is defined as the difference in means divided by an estimate of the pooled standard deviation and incorporates a correction factor based on the size of the samples being compared (18, 19).

Furthermore, we visually presented the proportion of respiratory visits as a percentage of total visits by respiratory condition and by modality (in-person vs. virtual). These visit rates were calculated using total monthly visits (of any reason) as denominators except in Sweden and Peru. For these countries, we used total coded visits as the denominator, as uncoded visits were more likely to be with other health-care providers and potentially occurred concomitantly with a visit to a primary care physician. We generated a plot illustrating the trends in respiratory visits throughout the studied timeframe, alongside a trendline representing the incidence of new COVID-19 cases per 100,000 population. The data points for this plot were extracted from https://ourworldindata.org/coronavirus (20).

We supplemented our data analysis by conducting a survey among INTRePID collaborators, who served as points of contact in each country and were actively engaged in their local, regional, and national COVID response efforts. The survey aimed to describe the accessibility of care for patients with COVID-19 during the first and second years of the pandemic. Based on the framework developed by Huston et al. (6), which delineates the roles of various healthcare sectors in COVID-19 assessment, our survey employed a 5-point Likert scale offering respondents a range of options to indicate the frequency of occurrence for specific scenarios or activities related to COVID-19 care accessibility. Respondents could select from the following response options: “always,” “common,” “sometimes,” “occasionally,” and “never..” Refer to Supplementary Table S11 for a comprehensive description of the questionnaire employed.

Patients and the public were not involved in the study design phase due to its highly technical nature; however, members of the public in INTRePID countries read the manuscript to ensure acceptable methods and interpretation. Specifically, the Patient and Clinician Engagement (PaCE) group, a well-established international patient advisory committee within the North American Primary Care Research Group (NAPCRG) (21), confirmed that our study was of public interest and offered important feedback on our results and discussion.

Our study found a notable decline in the rate and average monthly volume of primary care visits for respiratory concerns following the onset of the COVID-19 pandemic. Acute respiratory infection visits experienced a more pronounced decrease compared to non-infectious chronic respiratory illness visits. This trend suggests that COVID-19 mitigation measures likely impacted infections such as influenza and URTIs, while chronic respiratory illnesses such as COPD and asthma were less amenable to these efforts. Predictably, COVID-19 had a lesser effect on seasonal respiratory illnesses such as allergies. Our findings align with existing research demonstrating a decrease in respiratory virus activity during the COVID-19 pandemic. This decline has been linked to reductions in cases of acute respiratory illnesses and influenza-like illnesses. Furthermore, studies have shown a positive impact on chronic respiratory diseases, with fewer hospital admissions for asthma and COPD exacerbations during the pandemic’s early stages, which coincides with the implementation of national lockdowns and non-pharmaceutical interventions (23).

The changes observed in non-infectious respiratory illnesses might stem from reduced access to in-person primary care or insufficient reporting of virtual visits not accounted for in the available data. Before COVID-19, only a few primary care practices conducted virtual visits, and there may have been delays in recording, coding, and reporting virtual visits during the early phase of the pandemic.

While non-COVID-19 acute infection visits generally decreased, this change varied across countries participating in INTRePID. For example, respiratory illness accounted for 21.8% of visits in Singapore before the onset of the COVID-19 pandemic. This rate dropped below 10% of visits in the first few months of the pandemic and slowly levelled out at just over 10% of visits, with very few attributed to COVID-19. Prior to COVID-19, there was seasonal variation in Canada, with peaks near 15% of primary care visits attributed to respiratory illness. While there was an overall decrease in the rate of respiratory illness, it still hovered around 5%, with a significant number of COVID-19 patients. Our results were similar to those found in the United Kingdom, showing a marked decline in acute respiratory illness with flattening of seasonal variation while maintaining the usual incidence and seasonal variation associated with allergic rhinitis (24).

The variation in respiratory illness visits across countries also highlights differences in the location of COVID-19 infection care within our study population. Reflecting the diverse healthcare landscapes and pandemic responses globally, the dominant sites of COVID-19 care varied significantly. These variations encompass the dominant sites of COVID-19 care, including primary care versus other facilities, and align with the fluctuating waves of COVID-19 infections across the globe. Huston et al. studied the primary care and public health response to COVID-19 in 6 countries in early 2020 at the start of the pandemic (6). They found that COVID-19 assessment centers were the dominant location for triage of potential COVID-19 cases. In accordance with the Houston et al. study, neither their analysis nor our study identified primary care practices as the predominant COVID-19 testing and assessment locations in Australia, Canada, or the United States. Devi and colleagues reported that the majority of patients in their multi-country study had seen a health professional during the pandemic (63%) (25). This was even higher in Argentina, the only country that overlaps our research, which may explain the smaller drop in overall visit volume seen in this country.

There is evidence that many infectious respiratory illnesses were much less common during the first 1–2 years of the COVID-19 pandemic (26, 27). Stephenson reported that while overall ambulatory visits dropped by just 5% between 2019 and 2020, the number of “common cold” visits dropped by 51% (3). Rodgers et al. reported that during the first few months of the pandemic, respiratory visits to the emergency department (ED) were twice the pre-pandemic rate; however, by the end of 2020, ED respiratory infections were below pre-pandemic rates (28). Liu et al. found lower rates of most respiratory pathogens among hospitalized children with lower respiratory tract infections (29). Lockdowns, social distancing, and mask mandates may have contributed to protection from COVID-19 (30, 31) and many other endemic and seasonal infections such as respiratory syncytial virus (RSV), influenza and other common rhino and adenoviruses (32, 33). The lower rate of respiratory infections seen among INTRePID participants may indicate the response to robust public health measures aimed at minimizing the spread of contagious illness. With the reduction in COVID-19 mitigation efforts in 2021, there was a resurgence in common respiratory infections (34, 35). Most recent evidence from December 2022 in the United States and Norway revealed major increases in RSV and influenza (36). Renati and Linder reported that a majority of acute respiratory infections may not require a clinical consultation (37). The additional fear of transmission and the restrictions in place in many healthcare settings may have been enough to keep patients with mild to moderate COVID-19 infection away from the clinic. Bullen et al., in a 9-country survey, found that 60% of physicians and pharmacists reported patient “reluctance to visit a healthcare setting.” (38).

Insights gained from previous pandemics provide crucial context for understanding the dynamics of viral interactions during outbreaks. For instance, the emergence of influenza A (H1N1) pandemic in 2009 had a significant impact on the circulation of other respiratory viruses between 2009 and 2011 (39). Studies observed unusual patterns in virus activity following the influenza A (H1N1) pandemic peak. Research conducted in France suggested a delay in the circulation of respiratory syncytial virus (RSV) during the 2009–2010 season, compared to previous years (40). Similarly, a study conducted in the United Kingdom found that some cases initially diagnosed as influenza during the summer outbreak were actually caused by other respiratory viruses (41). These findings underscore the interplay between dominant strains like the influenza A (H1N1) during the 2009 pandemic and other respiratory viruses. While our study focused on acute respiratory visits without examining the specific viruses involved, similar dynamics were observed during the COVID-19 pandemic (39).

Goodyear-Smith and colleagues found that the perceived strength of the primary care system was not associated with a lower COVID-19 mortality rate (42). However, they also found that the perceived strength of a pandemic plan with robust implementation was associated with lower COVID-19 mortality. Local, regional, and national planning for COVID-19 recovery should also include planning for the management and resurgence of other respiratory infections. Primary care plays a crucial role in vaccination and may need to be part of post-pandemic immunization catchup and annual management (43, 44). Virtual visits may also play an important role in primary care and require further research to maximize their impact.

This research comes with some limitations. First, we collected data from numerous sources with large variations in availability. While some INTRePID countries provided comprehensive national-level data, others provided limited data from a few clinics or regions. For instance, data from Peru (45) include over 8000 primary care practices, representing nearly 70% of the population. Data from smaller samples, such as those from China and the US, may not reflect the regional variation nor the national experience in primary care. Also, the usage of country names to define the regions is primarily for clarity and comprehension purposes rather than a direct comparison between the entire country populations. However, we have at least in part achieved a global footprint in relation to the sampling frame.

Our aim was not necessarily to compare countries to each other but rather to compare pre-pandemic with COVID-19 pandemic time periods among the participating INTRePID countries. We present unadjusted analyses as social and demographic variables were not available among all participant data. As many countries managed COVID-19 outside of the typical primary care setting, the COVID-19-related visits presented here reflect the impact on and role of primary care in the typical primary care settings rather than the full impact of COVID-19 in the community.

Moreover, primary care physicians staffed COVID-19 assessment clinics established outside the conventional primary care settings in numerous INTRePID countries. However, the visit rates for these patients were not accounted for in our data, except in Norway.

Another limitation arises from the reliance on coding systems themselves. For instance, due to the emergence of COVID-19 as a new diagnosis, some coding systems did not differentiate between suspected and confirmed cases. Unfortunately, this limitation in data availability prevented us from making a stratified analysis of suspected and confirmed cases. However, given the widespread epidemic nature of COVID-19, in the midst of a local wave, it is likely that most suspected cases were true COVID-19 infections.

We acknowledge that COVID-19-related visits to primary care do not reflect COVID-19 cases or death rates (20). The results presented in this study do not represent the impact of the global COVID-19 pandemic on primary care. We show the impact of the pandemic within each of the participant countries and particular regions involved.

Telehealth virtual visits are a safe and effective alternative to in-person clinic visits (14). While it has limitations, such as the inability to perform physical exams (46), it allows for efficient triage, effective symptom assessment, and the provision of timely medical advice, especially during times when in-person visits were restricted. Many primary care practices increased their use of virtual telehealth visits throughout the COVID-19 pandemic, particularly in the first year of the pandemic. While the INTRePID data included in-person as well as virtual visits, the rapid shift to virtual visits may not have generated a full encounter in the medical record or billing number in some countries, resulting in a loss of primary care practice visit data. Virtual visits, particularly early in the pandemic, may have been audio only and did not generate a full encounter in some countries’ medical or billing records.

Future research should explore the long-term impact of the pandemic on primary care utilization patterns and healthcare delivery. Further investigation into the effectiveness of virtual visits and strategies to address the underreporting of encounters is warranted. Moreover, understanding the interplay between pandemic response measures and the resurgence of respiratory infections will inform future public health interventions and pandemic preparedness efforts. This study also identifies the urgent need to consider methods to harmonize and curate data from various sources as a method to conduct robust international primary care research.