This retrospective descriptive study analyzed US healthcare claims from 2014 to 2023 to characterize trends in hospitalizations for all-cause pneumonia and influenza (ACPI) among adults (>18 years). The authors compared incidence trends when ACPI was identified only in the principal diagnosis position versus when an expanded approach included selected principal diagnoses with a secondary ACPI code. The effort aimed to assess whether reliance on principal diagnostic codes alone might underestimate ACPI hospitalization rates or distort temporal trends, particularly given evolving coding practices during the study period.
The investigation used de-identified administrative claims data to identify inpatient hospitalizations for ACPI between 2014 and 2023. Two principal definitions were applied:
Principal ACPI hospitalizations: admissions with an ACPI code listed in the principal diagnostic field only.
Expanded principal ACPI hospitalizations: admissions with either a principal ACPI code or a principal code for sepsis, meningitis, COPD, empyema, or acute respiratory failure when ACPI appeared as a secondary code.
The expanded principal definition was intended to capture hospitalizations in which a severe, related principal diagnosis was coded first while ACPI was recorded elsewhere in the diagnosis list. The authors also compared these definitions against ACPI codes present in any diagnostic field.
Incidence derived from the strict principal ACPI definition demonstrated the largest overall decline across 2014–2023. That decline was more pronounced in younger adult age groups compared with older adults, indicating differential temporal patterns by age. In contrast, the expanded principal definition showed less temporal variation, suggesting that changes in how principal diagnoses were selected influenced observed trends when using principal-only case definitions.
The expanded principal ACPI definition captured between 48.5% and 63.4% of ACPI codes that appeared in any diagnostic field, indicating that a substantial proportion of ACPI-coded hospitalizations would be missed by a principal-only approach. The authors report that using only principal diagnostic codes would exclude approximately 30% of hospitalizations that met the expanded principal ACPI definition. This discrepancy implies that principal-only definitions can exaggerate the magnitude of decline in ACPI hospitalizations and underestimate the absolute burden recorded in claims data.
The study also documents a shift toward greater use of sepsis and COPD codes in the principal diagnostic position over the study period. The authors suggest that such coding shifts may partially explain the larger apparent decline observed when using the principal-only definition, because cases that would previously have been recorded with principal ACPI codes may have been reclassified under other principal diagnoses while retaining ACPI in secondary fields.
Prevalence of risk factors indicative for pneumococcal pneumonia, influenza, or respiratory syncytial virus (RSV) increased during the study period, with a notable rise between 2020 and 2023. Importantly, in 2023 more than 96% of adults aged 65 years and older hospitalized with ACPI had at least one risk factor of interest recorded in the three years prior to hospitalization. The authors assessed these risk factors as markers of susceptibility and potential indications for vaccination against pneumococcus, influenza, or RSV.
The analysis demonstrates that case definition selection strongly influences trends derived from administrative claims. Relying solely on principal diagnostic codes may disproportionately inflate the observed decline in ACPI hospitalizations between 2014 and 2023 because it omits hospitalizations where ACPI appears in secondary positions alongside principal codes for sepsis, COPD, or acute respiratory failure. The expanded principal definition reduced temporal variability and captured a larger share of hospitalizations with ACPI present in any field.
The authors attribute part of the changing diagnostic-position patterns to coding practice changes over the study period. They caution that administrative-data estimates of disease burden must account for diagnostic coding behavior when used for surveillance or to infer temporal changes.
Ethical approval was not required for this study because it used secondary, non-identifiable healthcare claims data; consent for publication was likewise unnecessary. The authors disclose competing interests: several contributors are employees of Merck & Co., Inc., Rahway, NJ, USA, and may hold stock or stock options. The report does not provide additional methodological details beyond the summary reported in the abstract, and any specifics not stated there were not reported in the source.
Analysts and clinicians using claims data should be aware that principal diagnostic coding alone can undercount ACPI hospitalizations and may exaggerate declines over time.
Expanded definitions that include related principal diagnoses with secondary ACPI codes capture substantially more hospitalizations and show less temporal fluctuation.
Rising prevalence of risk factors for pneumococcal, influenza, and RSV disease, particularly among older adults, underscores the relevance of vaccination strategies; in 2023 nearly all adults ≥65 hospitalized with ACPI had at least one recorded risk factor in the prior three years.
Changes in coding practice, including greater use of sepsis and COPD as principal diagnoses, should be considered when interpreting longitudinal trends from administrative claims.
This claims-based study from 2014–2023 indicates that how diagnostic codes are used substantially affects measured trends in ACPI hospitalizations. Using only principal diagnosis fields may miss a meaningful fraction of hospitalizations and exaggerate declines, while an expanded principal approach captures more events and yields more stable estimates over time. These findings support careful case-definition selection and awareness of coding shifts when using administrative data to monitor respiratory infection hospitalizations.