Visits and Hospital Admissions Decreased During COVID-19 Pandemic: A Pattern Common to Several Highly Developed Countries, Despite Huge Differences in Other Aspects of Disease and Care

Abstract

The publication of ‘Impact of COVID-19 on Pediatric Care in Japan: Analysis of National Health Insurance Claims Data’ [1] in Acta Paediatrica comes 5 years after the start of the COVID-19 pandemic and over 7 million deaths worldwide.

In addition to the impact on everyone's life, both personal and professional, the COVID-19 pandemic had a huge impact on published research: Over 460 000 papers in PubMed—as a rough comparison, a search for ‘preterm birth/infant’ yields about 170 000 results.

Amidst the abundance (or overdose) of articles, what can we learn from the paper of Ohnishi et al.?

The paper describes a very focused consequence of the pandemic, namely its impact on healthcare across Japan. Using the insurance claims database, which covers 98% of all healthcare episodes in the country, it reports the volume of outpatient visits and hospital admissions before (fiscal years 2016–2019) and during (2020–2021) the COVID-19 pandemic. The data are further broken down by age groups, including both paediatric and adult care.

The information is limited to the number of outpatient visits and number of patient-days hospitalised, and the results are therefore purely descriptive, showing a decrease in visits/admissions during the first 2 years of the COVID-19 pandemic. The decrease was greatest in 2020 and partially reversed in 2021. The study neither provide data on emergency visits/admissions and Paediatric Intensive Care Units admissions, nor on diagnoses.

The comparison between age groups shows that the decrease in hospital admissions and outpatient visits was most pronounced in the 0–4 and 5–9 years age groups, while the impact was minimal for older children and adults, both for inpatients and outpatients.

Remembering that COVID-19 spared children [2], none of the data presented are surprising. The reasons for hospital admissions and outpatient visits differ between infants/young children and older adults, with the latter more likely to have non-communicable ‘chronic’ conditions leading to hospital admission, whereas children are often admitted for infectious diseases.

This is supported by the fact that, in Japan, university and tertiary hospitals experienced a smaller decline in activity than community hospitals, suggesting that the decline was mainly in acute care, while more complex children with chronic conditions were spared.

The results presented in the paper [1] do not allow us to go any further. The scarce, almost absent granularity of data is the main limitation of this report.

However, its importance lies in the fact that to assess the impact of a pandemic, population-based data are needed to avoid biased conclusions. As a matter of fact, this is especially true for COVID-19, where the rush to publish partial results from individual institutions was most evident, and where much of the enormous literature suffers from this and now seems hopelessly outdated.

Descriptive studies like this one take time and resources to collect data and transform ‘big data’ into easy-to-understand results. In this case, the results are available almost 3 years after the study period—but this is not uncommon when analysing national data.

The data presented cover only the first part of the pandemic, characterised by the sudden appearance of a new disease, home confinement and lockdowns, and a complete overhaul of hospital organisation, but before the full emergence of the omicron variant, with increased hospitalisations and PICU admissions, and before the widespread use of vaccines in paediatrics.

In deciding whether this report is relevant to us, the main question is whether the Japanese experience can be generalised to other countries.

My country (Italy) was one of the first COVID hotspots in the ‘western’ world and was severely affected by the wild-type virus. By March 2022, Italy—a country of 60 million inhabitants—had 14.6 million cases (about 24% of the population) and 159 000 deaths (mortality rate 2.65 per 1000; lethality of 11 per 1000 cases). Japan (about 125 million inhabitants) had 6.4 million cases (about 5% of the population) and 28 000 deaths (mortality rate 0.22 per 1000, lethality 4 per 1000 cases) [5]. These data suggest huge differences in both the incidence and severity of the disease.

I am not aware of data on the volume of healthcare activity for Italy broken down by age, but, similar to Japan, the total number of hospital admissions fell abruptly by 16% in 2020 compared with 2019, with a slight recovery in 2021 (−12%). However, in the face of this similar pattern, the absolute number of admission patient-days in Italy was 58.8 million (about 1 patient-day/person), compared to 296.9 million in Japan (about 2.4 patient-days/person) [6].

It is therefore difficult to judge whether the Japanese experience can be ‘generalised’ to Italy: yes and no. What is the way forward?

Trivial as it may seem, one lesson to be learnt from aggregate data is that while biology is the same for all humans (and this is also true for the SARS-COV2 virus), other aspects of healthcare are very different between countries, even at similar levels of development.

Evidence-based medicine (currently the main paradigm of care) focuses heavily on the ‘biological’, individual side of medicine. Its methodology is primarily concerned with reducing the risk of bias due to confounding or other ‘errors’ in the design/analysis of studies and maintains a ‘hierarchy of evidence’ (e.g., that randomised clinical trials provide better quality evidence than observational studies). However, randomised clinical trials do not, or only minimally, address the generalisability or transferability of results which is taken for granted, and do not require representativeness of samples [7].

Observational descriptive studies such as this one, on the other hand, focus on public health aspects of care and to be valid require that the data be a census covering a defined population, as in the present study, to avoid spurious estimates. This requires painstaking data collection over a period of time and produces results that are much less timely, especially when using standard administrative data.

This difference brings us back to the tension between EBM (by its nature universal) and the variability of procedures and outcomes that we see in the real world between different providers (countries, regions, hospitals, etc.). Surely, if we had the opportunity to take a close-up of investigations right down to the individual level, we would always see variation in any system. As I have argued elsewhere, ‘context matters’ [8]; and it is important to conduct macro- or systems-level analyses to relate changes at the individual or patient level to overall systematic changes [9].

The case of COVID shows that ‘non-strictly biological’ aspects of the disease (the timing of the spread of infection in different countries, which in turn affected which viral variants were at work, the preparedness of the health system, and finally the availability of vaccinations, to name a few) played a much more important role in the variation in outcomes than strictly biological knowledge (the use of EBM interventions to curb the pandemic and care for the sick). Why did Japan, where the first case of COVID was documented on 15 January 2020, fare so much better than Italy and other European countries? Especially considering that all these countries share the same knowledge base?

To return to the initial question of ‘what can we learn?’, there is nothing better than measuring a phenomenon to know what is happening. To quote one of the fathers of quality improvement, W.E. Deming, ‘Without data, you are just another person with an opinion’. This report provides a bird‘s eye view of a pattern of healthcare provision and use that is common to countries with similar levels of development but very different COVID experiences and (presumably) healthcare organisations. Despite uncertain generalisability, it provides a rock-hard nugget of data—COVID has reduced hospital admissions and visits—that is indisputable. What the report does not measure is probably much more important than what it does measure [10]—looking beyond counts at ‘quality’ (details, diagnoses, secondary effects) rather than ‘quantity’—and reinforces the idea that what we know (the ‘evidence’) is less important than how we use it—even when caring for individual patients.

Luigi Gagliardi: conceptualization, writing – original draft, writing – review and editing.

The author declares no conflicts of interest.