Trends in non-COVID-19 hospitalizations before and during the COVID-19 pandemic period, United States, 2017-2021
The COVID-19 pandemic has profoundly disrupted the dynamics of health care and disease in the United States. The wealth of data provided by this health insurance billing database has enabled wide-scale examination of these disruptions across a wide range of disease categories. The majority of categories saw a decline in the relative incidence of hospitalizations in April and May 2020, while the timing of the incidence returning to pre-pandemic levels varied by diagnostic chapter, and some chapters did not. reached pre-pandemic levels of incidence of hospitalizations by the end of the study period.
Hospitalizations with a primary diagnosis code for respiratory infectious diseases (eg, influenza, acute lower respiratory tract infections) and otitis media (eg, middle ear diagnosis subchapter) have were grouped together in our analysis given their common dramatic decrease in IRR in April and May 2020 and their relative low incidence during the fall/winter 2020/2021. These results support previous studies of influenza, RSV and other upper and lower respiratory tract infections during the 2020/2021 winter season.10.22. While the decline in influenza could be partially attributed to a decrease in the number of samples sent for breath testing, as evidenced by a 61% decrease in US samples sent for testing in 2020, studies have shown that the rate of positivity among the samples sent had also decreased, by around 98%, during the winter of 2020/202111.
Unlike the subchapters with large declines in IRR, the group of subchapters with increased hospitalization rates (group B), included diagnoses of behavioral syndromes associated with physiological disorders, hemolytic anemias, and intellectual disabilities. Interestingly, these diseases did not appear to have a clear common mechanism to explain the trends of the pandemic period. Potential drivers of sporadic increases in any disease incidence during the pandemic period include: (1) changes in the underlying population covered by the insurance payers included in our analysis or changes in hospitals (or types of hospitals that) are part of the insurance service areas, (2) diagnoses that are a direct result of prior SARS-CoV-2 infection, and (3) mechanisms that cannot be elucidated at the using this dataset, such as diagnostic screening, ordering tests, and prioritizing hospitalization versus outpatient facilities.
Similar to other non-COVID-19 trend studies, we see an immediate relative decline in hospitalizations for neoplasms that rebounded later in 2020, followed by a shorter period of relative decline in winter 2020/202118. The largest decline was recorded among benign tumors in April 2020 (−77%, PI: −76%, −78%). Neoplasms of the skin, thyroid and male genital tract also saw a decrease in hospitalizations of 30% or more in April 2020. The most likely explanations for fluctuations in neoplasm diagnoses are disruptions in cancer screenings and subsequent diagnoses resulting from limitations in access to health care during the early pandemic period followed by an influx of patients needing delayed cancer screenings and referrals23,24,25,26.
Primary injury hospitalizations related to hospitalizations remained below expected rates in 2020, apparently in contrast to the results of a similar study in Denmark, in which injury hospitalizations appeared to be returning to expected incidence levels.18. Because our analysis relies on hospitalized patient data, we’re likely only seeing a small portion (and serious enough to require hospitalization) of the total injuries that presented during this time. Because countries had different approaches to pandemic-related shutdowns and patient triage, the timing of rebounds for non-infectious disease hospitalizations may not align perfectly across countries. .
The ICD-10 chapter on pregnancy, childbirth and the puerperium saw only an estimated decrease in relative incidence of 11% (95% PI: 9%, 13%) in April 2020. As expected, admissions for maternal delivery did not fluctuate during the pandemic period; however, birth outcomes could be influenced by the level of prenatal care and pregnancy-related stress as the pandemic continued through late 2020 and early 2021. Throughout the period of study, interesting signals were noted among hospitalizations for newborn gestational disorders and pregnancies with an abortion outcome. The incidence was outlying and did not appear to be clearly altered from the changes of the COVID-19 pandemic in healthcare. The increase in incidence in 2020 and 2021 for these diagnostic codes could be the product of an upward trend in data reported to the database or an accurate reflection of related pregnancy/newborn outcomes to the COVID-19 pandemic.
Group C, as a whole, could be characterized as showing declines in relative incidence in correspondence with the waves of COVID-19 occurring in the United States, mainly in March and April 2020 and January 2021. This supports the hypothesis according to which these hospitalizations are primarily those that may be delayed during peak COVID-19 transmission cycles. A Spanish study that included “factors influencing contact with health services” as a diagnostic chapter of interest, showed a relative increase in September and October 2020, which may reflect delayed care-seeking behavior20.
Due to the passive nature of data reporting to the healthcare billing center and potential differential reporting by the insurance payer, different payers may contribute to relative increases or decreases in disease rates in a way which may not represent generalizable trends across all regions of the United States during the study period. Other explanations for the relative increase in disease diagnoses include payer catchment redefinitions (and sudden inclusions of specialty facilities), reimbursement policies, and primary and secondary diagnosis coding practices. However, sensitivity analysis, where 20% of the top 100 payers were excluded from the analysis, revealed similar results, suggesting that high payers were not erroneously driving specific diagnostic category results. For most diagnostic chapters and sub-chapters, there was an increase in hospitalizations in January and February 2020, indicating fluctuation in billing data reported to the database used for this analysis. This complicates the interpretation of high incidence rates among some diagnostic sub-chapters during the pandemic period, as it may reflect the overall influx of data to this clearinghouse that began early in the year. calendar year 2020.
Our analysis was limited to primary diagnostic codes and it is possible that IRR trends could vary if all secondary codes were included for analysis. This study was also limited to inpatient data; therefore, we are unable to discern whether the relative increases or decreases in hospitalizations may be related to a change in location of treatment (i.e. cases treated as outpatients who would otherwise be hospitalized or vice versa) . Likewise, we are unable to make inferences about the prevalence or incidence of the disease in the general population, as the threshold for hospitalization may have changed during the pandemic period and the rates of hospitalization for disease correspond rarely to disease incidence rates in the general population.27. This phenomenon was noted among appendicitis cases, resulting in more cases being treated with non-surgical management prior to delayed surgical treatment at the start of the pandemic period.28.29. Unfortunately, the database used for the analysis did not include information on gender or race/ethnicity, thus limiting generalizability interpretations. Because the database was not updated after September 2021, analyzes of later waves of the pandemic were not possible and later trends may differ from early trends in non-COVID-related hospitalizations. 19.
This study is bolstered by the large population size (including U.S. patients with public and private insurance) and the ability to investigate rare disorders that might otherwise not be captured in a trend analysis. disease using localized data sources. Our ability to access multiple years of pre-pandemic data also strengthens our confidence in estimates of relative incidence during the pandemic period. Examination of billing data allowed this study to overcome limitations in disease reporting due to COVID-19-related disruptions, as demonstrated for data from the National Notifiable Disease Surveillance System (NNDSS )8.
There were substantial changes in hospitalization patterns across many different disease categories early in the COVID-19 pandemic, some of which have persisted for at least a year. The broad trends identified here suggest a number of hypotheses about the mechanisms behind these trends. The decline in infectious respiratory disease may likely be the result of NPI use, social distancing, and changes in daily life that have altered contact patterns. Meanwhile, this research has highlighted potential areas for further investigation by experts in the field, such as trends in hospitalizations for intellectual disabilities, newborn gestational disorders, and mental health disorders. . We hope that specialists and general practitioners, as well as epidemiologists, may find this study useful when analyzing their own data or dissecting trends in rare diseases.