Using hospital discharge registries that are merged into a research database hosted by the Department of Clinical Epidemiology, Aarhus University Hospital, Denmark, we identified all patients with a first-time discharge diagnosis of perforated or bleeding peptic ulcer during the study period. The discharge registries, established in 1977, contain information on patients' civil registry numbers, dates of admission and discharge and up to 20 discharge diagnoses coded by physicians according to the International Classification of Diseases (ICD) (8^th edition until the end of 1993, 10^th edition thereafter). The discharge registries do not contain data on whether perforated or bleeding peptic ulcer was the reason for hospitalisation or whether it occurred as a complication during a hospitalisation with other diseases. ICD-8 codes utilized for identification of patients with perforated peptic ulcer were 53100, 53101, 53108, 53109, 53209, 53309, and 53409. The ICD-10 codes were K251, K252, K255, K256, K261, K262, K265, K266, K271, K272, K275, K276, K281, K282, K285, and K286. ICD-8 codes utilized for identification of patients with bleeding peptic ulcer were 53190, 53192, 53195, 53290, 53390, 53490 and ICD-10 codes were K250, K254, K260, K264, K270, K274, K280, and K284.

All Danish citizens are assigned a unique civil registry number encoding age, gender, and date of birth and death or migration. This number is included in all Danish administrative registries and enables unambiguous linkage between these 12.

To obtain information on death or migration, we linked the study cohort to the Danish Civil Registration System, which has kept records for the entire Danish population since 1968 on vital status [dead, alive], date of death, residence, and migration.

We used the Charlson comorbidity index as a summary measure for burden of comorbidity 13. The Charlson index includes 19 major disease categories, including cardiovascular diseases, diabetes, chronic pulmonary, liver, renal and peptic ulcer disease, and solid and haematological malignancies. For computing the Charlson index score, a weight is assigned to each disease category, and the score is the sum of these weights. The index has been adapted and validated for use with hospital discharge registry data in ICD-based databases for the prediction of short- and long-term mortality 14. For each patient we identified, in the hospital discharge registries, all primary and secondary discharge diagnoses for all hospitalisations preceding the date of the first hospitalisation for complicated peptic ulcer; from these data we computed the Charlson index score 15. We defined 3 levels of comorbidity: low (index score of 0); moderate (index score of 1–2); and high (index score of more than 2). For this study, diagnoses of perforated and bleeding peptic ulcer were excluded from the Charlson index, since they defined our study population. Diagnoses of previous uncomplicated peptic ulcer disease (excluding peptic ulcer bleeding and perforation) were also removed from the index and included as a separate variable in the analyses.

Information on use of drugs potentially associated with both advanced age and a poor outcome of bleeding and perforated peptic ulcer was obtained through the prescription databases of Viborg, North Jutland, and Aarhus counties. The prescription databases contain information on all prescriptions redeemed in any pharmacy in the county, including patient's civil registry number, the type and amount of drug prescribed and the prescription date 16. By computerized data linkage we ascertained whether patients hospitalized with perforated peptic ulcer had redeemed prescriptions for oral glucocorticoids, low- or high-dose aspirin and/or NSAIDs within 60 days of admission. For patients hospitalized with bleeding peptic ulcer we retrieved information on redeemed prescriptions for oral glucocorticoids, aspirin, NSAIDs, vitamin K antagonists, calcium channel blockers, and antidepressants within 60 days of admission.

With the exception of aspirin and low-dose (200 mg) ibuprofen (which accounts for only 14% of ibuprofen use in Denmark), all included drugs are available in Denmark on the prescription-only basis 17.

Based on the date of first admission with perforated or bleeding peptic ulcer we constructed Kaplan-Meier survival curves and life-table estimates of 30-day mortality for the main study variables: age group (15–64, 65–79, 80+ years), level of comorbidity (according to the Charlson score categories), gender, use of ulcer-related drugs, and previous hospitalisation with uncomplicated peptic ulcer disease. To compare 30-day mortality in the three different age groups we first computed standardized 30-day mortality rates, using direct standardization in the oldest age groups (65–79 years and 80+ years) to the distribution of comorbidity, and gender in the youngest age group (15–64 years). Second, we used Cox's regression analysis to estimate 30-day mortality rate ratios (MRRs), while adjusting for comorbidity level, gender, use of ulcer-related drugs, and previous hospitalisation with uncomplicated peptic ulcer disease. To examine the effect of comorbidity on the association between advanced age and 30-day mortality, we then stratified the analyses by the three levels of comorbidity. The Cox's regression analyses were repeated after including comorbidity diagnoses made during the index hospitalisation with complicated peptic ulcer. To examine whether age-related mortality changed during the study period we also stratified the analyses on calendar year band (1991–1997 vs. 1998–2004).

The assumption of proportional hazards in the Cox's regression model was assessed graphically and found appropriate.

All analyses were performed using SAS version 8.2 (SAS Institute Inc, Cary, NC).

The study was approved by the Aarhus University Hospital Registry Board and by the Danish Data Protection Agency.

Descriptive data are provided in table 1. We identified 2,061 patients with a first-time discharge diagnosis of perforated peptic ulcer; of these 1,356 patients (61.4%) were more than 65 years old and 513 patients (25.7%) were aged 80+ years. A total of 38.5% of patients had at least one previous discharge diagnosis included in the Charlson comorbidity index and the burden of comorbidity increased with age. The majority of elderly patients (74.6% and 81.3% of the 65–79 and 80+ year old, respectively) were current users of NSAIDs, low- or high-dose aspirin or oral glucocorticoids.

Overall 30-day mortality from perforated peptic ulcer was 25.3%, but it increased from 8.9% among patients younger than 65 years to respectively 28.5% and 46.0% among patients aged 65–79 years and 80+ years (table 2). The standardized 30-day mortality was 24.2% among patients aged 65–79 years and 44.6% among those aged 80+ years. As seen in figure 1, the higher mortality among elderly patients was most pronounced during the first 2 weeks of hospitalisation with perforated peptic ulcer. The adjusted 30-day MRRs were 2.8 (95% CI: 2.1–3.6) among patients aged 65–79 years and 5.3 (95% CI: 4.0–6.9) among patients aged 80+ years, each compared with patients younger than 65 years (table 2). Including comorbidity diagnoses made during the index hospitalisation with perforated peptic ulcer did not change any of the MRRs. Further, we found no difference in age-related MRR between calendar year bands, 1991–1997 and 1998–2004 (data not shown). Regardless of comorbidity level, MRRs were consistently higher among elderly than among younger patients (table 3). As seen from table 3 the MRRs associated with increased age were highest among patients with low comorbidity levels. By contrast, the absolute differences in 30-day mortality between the three age groups were similar across the comorbidity index groups.

We identified 7,232 patients with a first-time discharge diagnosis of bleeding peptic ulcer, of whom 5,146 (71.2%) were older than 65 years of age and 2,372 (32.7%) were aged 80+ years. The prevalence of patients with comorbidities increased from 31.5% among patients younger than 65 years to 51.6% among patients aged 80+ years (table 1).

Overall 30-day mortality of bleeding peptic ulcer was 10.8%, but rose from 4.3% among patients younger than 65 years to 10.2% and 17.0% among patients aged 65–79 years and 80+ years, respectively (table 2, figure 1). After standardization, the 30-day mortality decreased to 8.7% among the 65–79 year old and to 16.9% among those aged 80+ years. Adjusted MRRs were 2.2 (95% CI: 1.7–2.8) and 3.7 (95% CI: 2.9–4.7) among patients aged 65–79 years and 80+ years compared with patients younger than 65 years. Including comorbidity diagnosis made during the index hospitalisation with bleeding peptic ulcer did not change any of the MRRs and we found no difference in age-related MRRs between calendar year bands, 1991–1997 and 1998–2004 (data not shown). The stratified analyses consistently showed higher MRRs among elderly patients compared with younger patients regardless of level of comorbidity (table 3). Again the MRRs associated with increased age were highest among patients with low comorbidity level whereas the absolute differences in 30-day mortality among the three age groups were similar across the three strata of comorbidity.