Abstract Introduction Robust estimates of adult hospitalised community-acquired pneumonia (CAP) incidence are critical for evidence-based public health decisions. Historically, the Hospital Episodes Statistics (HES) database which uses ICD-10 diagnosis codes, has been criticised as unreliable for estimating hospitalised CAP incidence. A recently published study reported 2019/20 CAP incidence from one large Bristol hospital based on ICD-10 data and an annualised 21-day prospective survey. Both approaches had similar results which were substantially higher than previous UK estimates. To further validate this past ICD-code-based analysis, we compared pneumonia incidence from a new ongoing prospective hospital study in Bristol with HES data over one year.

Methods Pneumonia event data (01Aug2020–31Jul2021) were extracted from HES for adults (≥18 years) in the same manner as the prior Bristol incidence study (i.e., ICD-10 codes J12–J18 in positions 1–5). Results were compared to incidence estimates from AvonCAP, an ongoing prospective study in two large Bristol hospitals designed to determine acute lower respiratory tract disease incidence. AvonCAP excludes hospital-acquired infections, and final clinical diagnoses, including pneumonia, are abstracted via chart review. Annual incidence (per 100,000) stratified by age was calculated using pneumonia admission numbers and previously established population denominators.

Results 1-year incidence estimates from AvonCAP and HES, stratified by age group, were: 103 vs 102 (18–34 y), 304 vs 317 (35–49 y), 698 vs 738 (50–64 y), 1245 vs 1448 (65–74 y), 2627 vs 2979 (75–84 y), 5502 vs 6657 (≥85 y) and 714 vs 802 (≥18 y) (table 1). The Pearson Correlation Coefficient between the incidence ranges was high (R=0.9996; P=0.0004). The HES versus prospective study incidence ratio increased with age (Range: 0.99 [18–34 years] to 1.21 [≥85 years]).

Conclusion This analysis showed a high correlation between CAP incidence estimates derived from HES and AvonCAP, validating the accuracy of HES coding for hospitalised CAP in Bristol. The trend towards higher incidence in older ages in the HES versus prospective estimates may be due to incomplete exclusion of hospital-acquired pneumonia from HES estimates. The conduct of audits, such as the National Adult BTS CAP audit, may result in coding accuracy improvements, thereby improving the reliability of HES in supporting adult hospitalised CAP incidence estimates.