Original Contribution

The American Journal of Gastroenterology (2005) 100, 1685–1693; doi:10.1111/j.1572-0241.2005.41833.x

A Nationwide Study of Mortality Associated with Hospital Admission Due to Severe Gastrointestinal Events and Those Associated with Nonsteroidal Antiinflammatory Drug Use

For the list of the members of Asociación Española de Gastroenterología, see Appendix.

Angel Lanas MD1, Maria Angeles Perez-Aisa MD1, Faust Feu MD2, Julio Ponce MD3, Esteban Saperas MD4, Santos Santolaria MD5, Luis Rodrigo MD6, Joaquim Balanzo MD7, Eduardo Bajador MD8, Pedro Almela MD9, Jose M Navarro MD10, Fernando Carballo MD11, Manuel Castro MD12 and Enrique Quintero MD13 on behalf of the Investigators of the Asociación Española de Gastroenterología (AEG)

  1. 1Servicio de Aparato Digestivo, Hospital Clínico Zaragoza, Zaragoza;
  2. 2Servicio de Aparato Digestivo, Hospital Clinico de Barcelona, Barcelona;
  3. 3Servicio de Aparato Digestivo Hospital La Fe, Valencia;
  4. 4Servicio de Aparato Digestivo, Hospital Val'de Hebron, Barcelona;
  5. 5Servicio de Aparato Digestivo, Hospital San Jorge, Huesca;
  6. 6Servicio de Aparato Digestivo, Hospital de Asturias, Oviedo;
  7. 7Servicio de Aparato Digestivo, Hospital de San Pau, Barcelona;
  8. 8Servicio de Aparato Digestivo, Hospital Miguel Server, Zaragoza;
  9. 9Servicio de Aparato Digestivo, Hospital Clínico, Valencia;
  10. 10Servicio de Aparato Digestivo, Hospital Costa del Sol, Marbella;
  11. 11Servicio de Aparato Digestivo, Hospital Universitario de Guadalajara;
  12. 12Servicio de Aparato Digestivo, Hospital del Valme, Sevilla;
  13. 13Servicio de Aparato Digestivo, Hospital Universitario de la Laguna, Tenerife, Spain

Correspondence: Angel Lanas, MD, Service of Gastroenterology, University Hospital, Lozano Blesa, 50009 Zaragoza, Spain

Received 19 November 2004; Revised  0000; Accepted 1 March 2005.





The worst outcome of gastrointestinal complications is death. Data regarding those associated with nonsteroidal antiinflammatory drug (NSAID) or aspirin use are scarce.



To determine mortality associated with hospital admission due to major gastrointestinal (GI) events and NSAID/aspirin use.



The study was based on actual count of deaths from two different data sets from 2001. Study 1 was carried out in 26 general hospitals serving 7,901,198 people. Study 2 used a database from 197 general hospitals, representative of the 269 hospitals in the Spanish National Health System. Information regarding gastrointestinal complications and deaths was obtained throughout the Minimum Basic Data Set (CIE-9-MC) provided by participating hospitals. Deaths attributed to NSAID/aspirin use were estimated on the basis of prospectively collected data from hospitals of study 1.



The incidence of hospital admission due to major GI events of the entire (upper and lower) gastrointestinal tract was 121.9 events/100,000 persons/year, but those related to the upper GI tract were six times more frequent. Mortality rate was 5.57% (95% CI = 4.9–6.7), and 5.62% (95% CI = 4.8–6.8) in study 1 and study 2, respectively. Death rate attributed to NSAID/aspirin use was between 21.0 and 24.8 cases/million people, respectively, or 15.3 deaths/100,000 NSAID/aspirin users. Up to one-third of all NSAID/aspirin deaths can be attributed to low-dose aspirin use.



Mortality rates associated with either major upper or lower GI events are similar but upper GI events were more frequent. Deaths attributed to NSAID/ASA use were high but previous reports may have provided an overestimate and one-third of them can be due to low-dose aspirin use.



Studies suggest that, while the incidence of peptic ulcer and its complications is declining, hospitalization rates and ulcer complications are on the increase in the elderly population (1,2). The rise in ulcer complications in the elderly is most likely due to longer life expectancy in Western countries, which is associated with higher incidences of both rheumatic and cardiovascular diseases, and consequently an increased use of nonspecific nonsteroidal antiinflammatory drugs (NSAIDs) and aspirin in this population (1).

In the worst cases, gastrointestinal (GI) complications can result in the death of a patient, but data on most severe outcomes are scarce. Randomized, controlled trials alone are unlikely to provide reliable estimates of the incidence of rare events, such as death from GI complications associated with nonspecific NSAID use, due to the limited population they study. Cohort and case–control studies involve larger sample sizes but are vulnerable to bias of a variety of kinds. To overcome the limitations of different designs, a U.K. study developed a model to quantify the frequency of rare adverse events associated with chronic NSAID use by combining data from published, randomized, controlled trials, and observational studies (3). The study concluded that, on average, one in 1,200 patients taking chronic oral NSAIDs for at least 2 months will die from gastroduodenal complications as a direct result of NSAID use (3). This figure extrapolates to about 2,000–2,500 deaths each year in the United Kingdom (3,4) although other estimates from the same country have lowered that figure to 400 deaths in people older than 60 (5), based on a widely quoted population study (6). In the United States, another widely quoted report had put that figure to 16,500 cases per year (7,8).

Fewer data are available on the incidence of major lower GI events resulting from NSAID exposure, although epidemiological studies do suggest that nonspecific NSAIDs increase the risk of lower GI bleeding, perforation, obstruction, and diverticular disease (9,10,11). It is now recognized that hospitalizations due to lower GI complications are more common than previously suspected. Recently, a post hoc analysis from a large, randomized GI outcomes trial reported that serious lower GI events (including bleeding and hospitalization for perforation, obstruction, ulceration, or diverticulitis) accounted for a large proportion of all serious GI events in patients receiving naproxen (12). As previous estimates of complications for the whole GI tract have focused on upper GI events, it is possible that, when lower GI events are also included, mortality figures could be even higher among NSAID users.

In order to provide the best estimate of mortality associated with major GI events, and those associated with NSAID use, the present study uses data based on the actual number of deaths nationwide, including data from 197 Spanish general hospitals that account for 80% of the total hospitalizations within the Spanish National Health System (NHS).



This was an observational study conducted in the Spanish NHS, which utilized two different large data samples (study 1 and study 2) reporting the frequency of hospital deaths associated with severe and major GI events arising from any part of the GI tract (upper or lower). Events were coded as the primary diagnosis for hospital admission, and any events occurring during, or as a consequence of, hospitalization were not included. Therefore, this study accounts only for complications or serious events occurring in the community that lead to hospital admission, which represent the vast majority and more genuine gastrointestinal complications or severe events associated or not with NSAID/aspirin use. It does not take into account serious GI events occurring during hospitalization, which represent a different and probably more severe subset of patients with different pathogenic mechanisms to those occurring in the community (e.g., stress ulcers in intensive-care unit). In the same way, we did not include deaths occurring after hospital discharge, which was not possible to detect with the methodological approach used in the study.

The majority of the Spanish population uses the Spanish NHS, which provides open access to all health-care services including hospitals, drugs, and diagnostic and therapeutic procedures. Although private practice coexists with the NHS, it represents less than 20% of all hospitalizations within the country. Therefore, data provided from general hospitals account for more than 80% of all potential events. In this way, the data obtained have been extrapolated and considered representative of the total population.

Study 1

Study 1 was carried out in 26 Spanish general hospitals that serve the needs of approximately 8 million people distributed across the entire country. All serious GI events resulting in admittance at each participating hospital were registered in the Minimum Basic Data Set (MBDS) from January 1 to December 31, 2001. The MBDS is the most accurate source of computerized data provided by these hospitals, and comprises both administrative and clinical data for each hospitalization. Data are coded according to the International Classification of Diseases (9th revision, Clinical Modification [CIE-9-MC]), which has been widely used in Spain and across Europe for both administrative and research purposes (13,14,15,16). The most important and relevant clinical information provided by this system is the primary diagnosis, which is considered to be the cause of hospitalization based on the clinical judgment of the medical doctor who admits the patient to hospital. In order to provide a systematic tool for the management of the information, a patient-classification system based on Diagnosis Related Groups (GRD), version 18.0, was used, as adopted by the Health Care Financing System. In this way, each hospitalization event is unequivocally classified into one of the GRDs according to the main diagnosis and other variables, including: previous surgical procedures; patient gender and age; complications; co-morbidities; patient status at discharge; and use of health-care resources. Thus, the system identified all patients who died during hospitalization. However, since the system does not provide information on mortality associated with events after discharge from hospital, we can only provide a "minimum-mortality guarantee."

In order to obtain the total number of hospitalizations due to events affecting either the upper or lower GI tract, all GRDs of the gastroduodenal tract or small bowel and colon, which were related to peptic-ulcer disease, or complications such as bleeding or perforation, were included in the analysis (numbers 531–535, 578, 562, and 569 and all subcategories which included peptic ulcer bleeding, gastric, duodenal ulcer or acute mucosal bleeding, gastrojejunal ulcer, gastrointestinal bleeding, gastrointestinal perforation, rectal bleeding, small bowel diverticuli or diverticulitis with bleeding, colonic diveticuli, or diverticulitis with bleeding). Of the potential GI complications due to ulcer diseases, GI obstruction is the least common, and it has become increasingly rare over the last decade (1,2). Furthermore, most hospitalized cases of obstruction are unrelated to ulcer diseases. Therefore, in order to avoid an overestimation of this diagnosis, it was not included in the study.

Finally, in order to provide specific information regarding clinical data and drug use during the month before hospitalization in patients who died, charts of patients who died were manually reviewed. Patient identification was not possible since data were coded anonymously as digit numbers.

Study 2

Study 2 used data provided by IASIST, a company offering the largest hospital database in Spain, which includes clinical and economic information. IASIST belongs to the Solucient Group, Chicago, USA, an external organization with access to the PDHP01 database, which contains data from the MBDS of 197 general hospitals in Spain that are considered to be the representative of the 269 hospitals in the Spanish NHS. This data sample represents 77% of the total hospitalizations occurring during 2001. Information and results obtained from this database have been extrapolated to the entire Spanish population. For consistency, study 2 employed the same GRDs identified for study 1, but also provided data on the following parameters:

  1. Severity of events, based on data from refined GRDs (version HCFA 18) (17)
  2. Mean length of hospital stay due to admissions for GI events, calculated from the day of admission to the day of discharge. Length of hospital stay is also used to estimate hospitalization costs as, in a hospital context, the fixed number of resources used by each patient is high.
  3. Cost of the event, estimated by the NHS from the relative weight of each GRD multiplied by the mean cost of the Production Unit (equivalent to the cost of producing a hospitalization event of unitary relative weight). This variable differs according to the hospital complexity (level).

Deaths Attributed to NSAID and Aspirin Use

The number of complications and mortality events attributed to either NSAID or aspirin use (P - NSAIDs) was estimated from the following formula:

Unfortunately we are unable to provide accessible alternative text for this. If you require assistance to access this image, please contact help@nature.com or the author

where P-NSAIDs is a function that depends on the prevalence of NSAID and aspirin exposure in the study population (P), and the incidence of GI complications in patients exposed and not exposed to NSAIDs and aspirin; R is the risk ratio or the quotient of the incidence rate of GI complications in those exposed and not exposed to NSAID and aspirin (18,19).

Data regarding drug use obtained from retrospective hospital chart reviews are not accurate at all to provide reliable estimates of the proportion of events attributed to NSAID/aspirin use. It would also have needed the review of thousands of charts from potential controls where the transcription of drug and NSAID use is even less reliable. Therefore, the prevalence of drug use in patients with or without GI complications was obtained from a parallel study involving prospective collection of data, which was carried out at the same time and in the same hospitals as those used in study 1. Preliminary data from this study have been recently reported (20,21). The estimated rates of GI complications in those exposed and not exposed to NSAIDs have been calculated on the basis of matched control population by age and on logistic regression models where sex and other confounders were introduced in the models (20,21).


According to local epidemiology data, it was expected that about 0.067% of patients treated with any NSAID would die due to a severe GI complication (22). Using these data, we would need to study 416 in detail in order to extrapolate nationally with the precision level of plusminus3% and alpha value of 0.01. Both quantitative and qualitative variables were analyzed using only descriptive statistics (mean and frequency). We used the approximate normal distribution-based method to obtain the confidence intervals. All statistical tests were performed using SPSS Software Products, Version 10.0 (SPSS, Inc., Chicago, IL).



Study 1

A total of 8,010 serious GI events (upper and lower GI tract) were reported. Extrapolating study 1 data of the whole Spanish population (40,850,540) gave a total annual estimate of GI complications of 41,409 patients (Table 1). The mortality rate from GI complication events calculated from study 1 was 5.7% (95% confidence interval [CI]= 4.9–6.7). More patients were hospitalized or died as a result of upper GI events (mostly upper GI bleeding) than serious lower GI events: n = 5,814 versus n = 1,031 hospitalized, respectively, and n = 332 versus n = 55 deaths, respectively. However, the percentage of deaths as a result of serious upper or lower GI events was similar (5.7%vs 5.3%, respectively). Hospitalizations due to GI perforation represented just 2.4% of all reported events. However, 30.1% of all GI perforation events died, which was far in excess of that calculated for either upper or lower GI bleeding events.

The mean age of patients who died as a result of their complications was 76plusminus13.5 yr, and 89.7% of patients were aged >60 yr. Among those who died, there were slightly more men than women (57.2%vs 42.8%). Approximately half of the patients who died had some GI disease history, including history of peptic ulcer (21.6%), GI bleeding (15.3%), or dyspepsia (13.5%). Among those who died, the most prevalent comorbidity was a history of cardiac disease (65.1%) or hypertension (40%). Twenty-seven percent of all patients who died were receiving proton pump inhibitors, and an additional 10% were taking other antisecretory drugs prior to hospitalization. All deaths occurred in hospital, and all of them were either directly associated with the GI event (43.1%) or were a consequence of complications developed during hospitalization prompted by the GI event.

Study 2

Hospitalization for a serious GI event represented 1.4% of hospital admissions for any cause in 2001. A total of 50,114 GI events were recorded in study 2 (Table 2), and 47,591 patients were hospitalized for those GI events (1.05 events/patient). This represents 121.9 events/100,000 people and 1.262 NHS hospital beds permanently occupied. Mean length of hospital stay was 7.81 days per event and the mean cost was 2,311 euros (euro dollar) per event, representing additional costs to the NHS of 116 million euro dollar. Ten percent of all serious GI events required surgical intervention. Sixty-two percent of patients hospitalized with serious GI events were male, and the mean age of all hospitalized patients in study 2 was 64 plusminus 20 yr. As in study 1, the percentage of patients experiencing a GI complication event rose with increasing age (Fig. 1). The mortality rate calculated from study 2 data was similar to that obtained in study 1 (Table 1). A total of 2,816 patients died due to serious GI events in hospital resulting in a mortality rate of 5.62% (95% CI: 4.8–6.8), which was higher than the overall 3.7% rate associated with hospitalizations for any reason.

Figure 1.
Figure 1. - Unfortunately we are unable to provide accessible alternative text for this. If you require assistance to access this image, please contact help@nature.com or the author

Distribution of patients with gastrointestinal complications by age and sex (study 2).

Full figure and legend (59K)

The most common coded GI events resulting in hospitalization were gastric or duodenal ulcer bleeding (21.3% and 23.1% of all GI events, respectively) and GI bleeding (27.5%) (Table 2). GI perforations resulted in the longest hospital stay of all complication events (Table 2), and, in agreement with the findings of study 1, were associated with a mortality rate in excess of that for the more commonly recorded GI bleeding events (30.3%vs 9.9%). As expected, the length of hospital stay and mortality rates increased with the severity of the complication event. The mean hospital stay for GI bleeding increased from 5.6 days for mild events to 13.1 days for severe events with a corresponding rise in mortality rates from 0.9% to 26.5%.

Complications and Deaths Attributed to NSAID Use

Estimates of complications and deaths attributed to NSAID or aspirin use were based on prospective data obtained from a case–control study carried out in the majority of hospitals included in study 1 (20,21). According to these data, and those obtained from other similar studies in Spain (9), the prevalence of NSAID/aspirin use (including OTC use) in the study population was estimated to be approximately 19% (17.7% to 19.8%). The estimated rate of GI complications in patients not previously exposed to NSAIDs/aspirin was over 120 per 100,000 patients/year, while in patients previously exposed to NSAIDs/aspirin rates were substantially higher at 480/100,000 patients/year. Therefore, the proportion of complications and deaths attributed to NSAID/aspirin use was 36.3%.

Using the data given above, the estimated number of GI complication events and deaths attributed to NSAID/aspirin use for the entire country was 15,031 and 860, respectively, for study 1, and 18,191 and 1,022, respectively, for study 2 (Table 2). When the number of complications and deaths were calculated by the type of NSAID used, low-dose aspirin was responsible for no less than 8.2% and no more than 12.2% of all complications and deaths; therefore, between 4,109 and 6,113 complications, and 231 and 343 deaths, approximately, were due to low-dose aspirin use. Coxib use rate was low (around 1.3% of the adult population during that period of time). Therefore, GI events and mortality due to this type of drugs were negligible and the estimations based on that proportion would be probably unreliable.

The data reported translate into a mortality rate as a consequence of NSAID/aspirin-related GI complications (upper and lower GI tract) of 21.0–24.8 cases per million inhabitants. In 2001, 35.4 million prescriptions of NSAIDs/aspirin were filled in Spain (23). Based on these data, the mortality rate due to NSAID/aspirin-associated GI complications for study 1 and study 2 was 24.4–28.8 cases per million prescriptions.

Further, it has been estimated that 20.6% of the Spanish adult population use an NSAID for at least 1 month during any given 12-month period (22), thus the mortality rate is calculated as a maximum of 15.3 cases per 100,000 NSAID users. The estimated mortality rates for the entire Spanish population and for NSAID/aspirin users are lower than previous values reported for the United States and some other reported in the United Kingdom (Fig. 2). Other report from the United Kingdom, however, suggest a much lower incidence, which is even lower to those reported in this study, but it refers to patients older than 60 yr (5).

Figure 2.
Figure 2. - Unfortunately we are unable to provide accessible alternative text for this. If you require assistance to access this image, please contact help@nature.com or the author

Mortality rate attributed to NSAID/aspirin-associated GI complications in Spain and frequently reported estimates published in the United States, and United Kingdom (3, 4, 13, 15). *Data estimated from the population registered in 1998. Estimations for mortality in NSAID/aspirin users have been calculated from the adult population with a 20.6% of NSAID use in Spain (17) and a 16.8% in the United States and United Kingdom (7,8).

Full figure and legend (15K)



Previous studies reporting the incidence of GI complications and resulting mortality have been restricted to the analysis of complications occurring in the upper GI tract only. The present investigation reports deaths due to complications of the entire GI tract and the attributed proportions of those deaths due to NSAID use and low-dose aspirin.

Estimates of mortality associated with complications throughout the whole GI tract are important as it is now recognized that NSAID-related lower GI events are relatively common (9,11,12). The proportion of deaths from either upper or lower GI complications reported here was similar (5.7%vs 5.3%, study 1), although the actual number of deaths associated with GI complications was 6.2 times higher in the upper GI tract. It must be noted, however, that the frequency of lower GI complications could have been underestimated due to the definition of some variables (such as "GI bleeding") in the data-classification system used. To provide estimates of GI mortality due to low-dose aspirin use is also important since aspirin is now widely used in the population where the benefits and risks of prevention strategies must be carefully balanced.

While several outcome studies have previously provided estimates of GI complications associated with NSAID use, randomized controlled trials alone are unlikely to provide reliable estimates of the incidence of rare events, such as deaths. Estimations would have to be made based on very few cases and the probability of error is high under these conditions. In addition, patients in these trials (24,25,26) were very different from those included in this ecological study, which represents "real life" in clinical practice. Patients in those outcome GI studies (24,25,26) were in most cases rheumatoid arthritis patients, which are a minority in the overall subset of patients admitted to hospital because of a GI complication. Furthermore, in these trials, patients had been taking relatively high dose of NSAIDs for a long periods of time (6–12 months), which are conditions that are not commonly seen in clinical practice, where patients take lower doses and for shorter periods of time (often intermittent treatment). Outcome trials did not provide a control group not taking NSAIDs, and therefore it is difficult to estimate how much of that proportion of complications could be attributed to NSAID treatment. However, even assuming a 100% attributable risk, the rate of deaths would not be higher than 44.7 deaths per 100,000 NSAID users in our population.

The present nationwide study has tried to overcome some of these limitations by using two different data sets that are representative of common clinical practice: One based on actual counts of GI complications and deaths in 26 Spanish general hospitals, and the other based on data obtained from 197 hospitals which are considered to be the representative of the entire Spanish National Health System, since the activity of most remaining private hospitals is focused on different and less severe pathology. Thus, we believe that the data and extrapolations reported here represent a rigorous and current estimate of death due to GI complications and of those attributed to NSAID use. Results and estimates obtained from the two data sets were similar, reinforcing the validity of the methodology.

The overall rate of deaths due to GI complications and the rate of deaths associated with NSAID/aspirin use reported here are lower than some frequently quoted estimates from previous studies, despite the fact that our figures include both upper and lower GI complications and also refer to low-dose aspirin use (3,4,7,8). Our data, however, agree with some more conservative estimates from the United Kingdom (5) and more recent data regarding current rates of GI complications reported in the United States (19). There are a number of reasons that may account for some of the discrepancies observed in the studies: variation in prescribing practice by country; differences in the extent of OTC NSAID use and in the co-prescription of gastroprotective drugs; decreasing GI complication rates; and differences in study methodologies. Our data would imply a lower NSAID consumption in Spain compared with other countries. However, the annual NSAID prescription rates in Spain are relatively high (35.4 million) (22,23) and are proportionally greater than rates reported in the United Kingdom and 50% of those reported in the United States (70 million), despite Spain's smaller population. In addition, the rate of NSAID use among adults in Spain (20.6%, 95% CI 15.8–25.4) is similar than that determined in the United States (7,22).

It has been estimated that the use of OTC NSAIDs including aspirin could exceed that of prescription drugs by five-to-seven-fold (27). Since the majority of studies investigating NSAID-related GI complications do not take into account the use of OTC medications, it is likely that mortality rates could be underestimated in these instances. This is not the case with the present study as all hospitalizations for GI complications, regardless of the cause, were included in the analysis, and the data used to estimate the proportion of those complications attributed to NSAIDs included OTC use. It is also important to outline that by counting all GI complications, those attributed to low-dose aspirin use emerge as an important part of those complications and deaths. Rates of NSAID/gastroprotective agent co-therapy vary substantially from country to country and this could, in part, account for differences in mortality rates observed between the studies. The proportion of NSAID users receiving gastroprotective agents in the United Kingdom was estimated to be 20–22%, while in the United States it was 29% (28), although these proportions may be higher today. In Spain, the proportion is still higher with greater than or equal to50% of patients taking concomitant NSAIDs and gastroprotective agents (23). In agreement with these data, we observed that a relatively high percentage of patients who died from their GI complications in study 1 were receiving antisecretory drugs (27% PPI, 10% H2-blockers). Estimations of deaths attributed to NSAIDs/aspirin use are sensitive to two variables which are mortality rate in patients with GI complications and the proportion of those complications attributed to NSAID/aspirin use. The retrospective approach in the measurement of the number of complications and deaths used in this study, and the indirect approach used to determine the number of those events attributed to NSAID/aspirin use are two limitations of this study. However, we must outline that the mortality rate of GI complications in our study is within the expected range and in agreement with most recent reports (29,30,31). The rate is lower than that of other reports that have included deaths due to complications developed during hospitalization, which have a higher mortality rate (6). However, we believe that estimations of GI side effect outcomes due to NSAID use should not be based on data including GI complications developed during hospitalization, since they are not representative of NSAID use in clinical practice and may bias the results due to the much higher proportion of deaths. On the other hand, the attributed proportion of GI complications due to NSAID/aspirin use found in this study is very similar to other reports and within the expected range (5,32), which suggest that the approach, as explained in methods, was correct and probably better than using data from charts in a retrospective way.

One strength of our study is that the sample size of the two studies is large, nationwide, closer to common clinical practice, and based on actual account of complications and deaths. Despite this, we assume that the number of deaths may have been underestimated since death due to GI complications may occur at home in patients that do not seek hospital attention or after hospital discharge, but this is also a problem of other studies as well (1,2,3,4,5,6).

The lower mortality rates determined from this study could simply be the result of the decline in GI complication rates, which has been detected in the last 5–10 yr in different studies (1,2,19,33). This may be a consequence of persuasive medical education campaigns aiming to increase awareness in physicians of the unrecognized risks associated with NSAIDs, a concomitant decrease in the rate of Helicobacter pylori infection, or increases in PPI prescriptions. Alternatively, it could be due to the development of less toxic NSAIDs for the treatment of rheumatic diseases, such as the cyclooxygenase (COX)-2 specific inhibitors. Finally, when compared to those located in the upper GI tract, the proportion of lower GI events attributed to NSAID/aspirin use (1:6) in this study was lower than those reported in the VIGOR trial, where the rate of serious lower GI events accounted for 39.4% of all serious GI events among patients taking naproxen (12). The reasons for this apparent discrepancy are unclear but this study has accounted only patients with hospital admission and in VIGOR one-third of those described with lower GI events had no hospital admission. Furthermore, in VIGOR some of the patients (6 of 27) diagnosed with lower GI bleeding had positive tests for fecal occult blood rather than gross rectal bleeding, which can have been missed in this study.

While the results of our study suggest mortality rates from NSAID-related GI complications are lower than previously believed, they are still of great clinical and economic significance. The estimated number of individuals who died as a result of their complications is similar to that for deaths associated with occupational activities according to the figures from the Spanish Ministry of Labour and Social Affairs, and is equivalent to more than 50% of AIDS deaths in 2001, as estimated by the Spanish Institute of Statistics. The management of GI complications places a large economic burden on the Spanish NHS. Figures from 2000 demonstrated that the cost to the Spanish NHS from NSAID-induced gastroenteropathy approached euro dollar600 million, with euro dollar60 million spent on the actual treatment of complications (34). Results from the current study suggest this figure could be even higher for 2001, as the mean cost per complication event in study 2 was estimated to be 2,311 euro dollar, resulting in additional costs to the NHS of 116 million euro dollar. It is therefore of paramount importance that severe GI outcomes are accurately quantified to allow adequate distribution of economic resources in health prevention strategies.

The analysis of the population who died in this study as a result of GI complications shows that the elderly and patients with concomitant diseases represent a vast majority of deaths, suggesting that a bad basal condition was an important factor associated with death related to GI complications.

This highlights the importance of taking ever-greater steps to research new and better alternatives to treat pain and inflammation in the elderly, to heighten physician and public awareness of the associated problems of NSAID therapy, and to educate them on the use of appropriate prevention strategies. It is still very clear that the GI adverse events associated with NSAID and aspirin use pose an important problem to the national health of any country with respect to loss of life and utilization of health-care resources.



  1. Higham, J, Kang, JY, Majeed, A. Recent trends in admissions and mortality due to peptic ulcer in England: Increasing frequency of haemorrhage among older subjects. Gut 2002;50: 460–464. | Article | PubMed | ISI | ChemPort |
  2. Perez-Aisa, MA, Del Pino, L, Siles, M, et al. Clinical trends in ulcer diagnosis in a population with high prevalence of Helicobacter pylori infection. Aliment Pharmacol Ther 2005;21(1):65–72. | Article | PubMed | ISI | ChemPort |
  3. Tramer, MR, Moore, RA, Reynolds, DJ, et al. Quantitative estimation of rare adverse events which follow a biological progression: A new model applied to chronic NSAID use. Pain 2000;85: 169–182. | Article | PubMed | ISI | ChemPort |
  4. Blower, AL, Brooks, A, Fenn, GC, et al. Emergency admissions for upper gastrointestinal disease and their relation to NSAID use. Aliment Pharmacol Ther 1997;11: 283–291.
  5. Hawkey, CJ, Langman, MJ.Non-steroidal anti-inflammatory drugs: Overall risks and management. Complementary roles for COX-2 inhibitors and proton pump inhibitors. Gut 2003;52(4):600–608. | Article | PubMed | ChemPort |
  6. Rockall, TA, Logan, RF, Devlin, HB, et al. Incidence of and mortality from acute upper gastrointestinal haemorrhage in the United Kingdom. Steering Committee and members of the National Audit of Acute Upper Gastrointestinal Haemorrhage. Br Med J 1995;311(6999):222–226.
  7. Singh, G, Triadafilopoulos, G. Epidemiology of NSAID induced gastrointestinal complications. J Rheumatol 1999;26 (suppl 56):18–24.
  8. Wolfe, MM, Lichtenstein, DR, Singh, G. Gastrointestinal toxicity of nonsteroidal antiinflammatory drugs. N Engl J Med 1999;340: 1888–1899. | Article | PubMed | ISI | ChemPort |
  9. Lanas, A, Serrano, P, Bajador, E, et al. Evidence of aspirin use in both upper and lower gastrointestinal perforation. Gastroenterology 1997;112: 683–689.
  10. Langman, MJ, Morgan, L, Worrall, A. Use of anti-inflammatory drugs by patients admitted with small or large bowel perforations and haemorrhage. Br Med J 1985;290: 347–349.
  11. Wilcox, CM, Alexander, LN, Cotsonis, GA, et al. Nonsteroidal antiinflammatory drugs are associated with both upper and lower gastrointestinal bleeding. Dig Dis Sci 1997;42: 990–997. | Article | PubMed | ChemPort |
  12. Laine, L, Connors, LG, Reicin, A, et al Serious lower gastrointestinal clinical events with nonselective NSAID or coxib use. Gastroenterology 2003;124: 288–292. | Article | PubMed | ChemPort |
  13. Ruiz, M, Cirera Suarez, L, Perez, G, et al. Grupo CAMPARA. [Comparability between the ninth and tenth revisions of the International Classification of Diseases applied to coding causes of death in Spain]. Gac Sanit 2002;16(6):526–532.
  14. Ruiz, U, Acedo, K, Buenaventura, R, et al. Implementing total quality management in the Spanish health care system. Qual Assur Health Care 1992;4(1):43–59.
  15. Paz Rodriguez-Perez, M, de la Rosa Rodriguez, G, Lopez-Madurga, ET, et al. Variations on in-hospital mortality rates based on an administrative database aortocoronary bypass mortality rate. Med Clin (Barc) 2000;114(suppl 3):112–116.
  16. Magee, HF.The Hospital Data Project: comparing hospital activity within Europe. Eur J Publ Health 2003;13(suppl 3):73–79.
  17. Steven, H. Medicare severity-refined DRGs: An improved system. Available at http://www.irp.com.
  18. Lilienfeld, AM, Lilienfeld, DE. Foundations of epidemiology. 2nd ed. Oxford University Press: New York; 1980.
  19. Kurata, JH, Nogawa, AN. Meta-analysis of risk factors for peptic ulcer. Nonsteroidal antiinflammatory drugs, Helicobacter pylori, and smoking. J Clin Gastroenterol 1997;24: 2–17. | Article | PubMed | ChemPort |
  20. Lanas, A, Garcia Rodriguez, LA. The use of coxibs does not increase the risk of upper gastrointestinal bleeding in common clinical practice. Gastroenterology 2003;124(4 suppl 1):A93.
  21. Lanas, A, Garcia Rodriguez, LA. Grupo de Estudio del Tracto Digestivo Superior AEG. El uso de coxibs no se asocial a aumento del riesgo de hemorragia digestive alta. Gastroenterol Hepatol 2003;26: 183.
  22. Carmona, L. Estudio EPISER. Madrid: Spanish Society of Rheumatology; 2000: 141–154.
  23. Kimmey, MB, Lanas, A. Appropriate use of proton pump inhibitors with traditional nonsteroidal anti-inflammatory drugs and COX-2 selective inhibitors. Aliment Pharmacol Ther 2004;19(suppl 1):60–65.
  24. Bombardier, C, Laine, L, Reicin, A, et al. Comparison of upper gastrointestinal toxicity of rofecoxib and naproxen in patients with rheumatoid arthritis. N Engl J Med 2000;343: 1520–1528. | Article | PubMed | ISI | ChemPort |
  25. Silverstein, FE, Faich, G, Goldstein, JL, et al. Gastrointestinal toxicity with celecoxib vs nonsteroidal anti-inflammatory drugs for osteoarthritis and rheumatoid arthritis: The CLASS study: A randomized controlled trial. Celecoxib Long-term Arthritis Safety Study. JAMA 2000;284: 1247–1255. | Article | PubMed | ISI | ChemPort |
  26. Silverstein, FE, Graham, DY, Senior, JR, et al. Misoprostol reduces serious gastrointestinal complications in patients with rheumatoid arthritis receiving nonsteroidal anti-inflammatory drugs. A randomized, double-blind, placebo-controlled trial. Ann Intern Med 1995;123(4):241–249. | PubMed | ISI | ChemPort |
  27. Barrier, CH, Hirschowitz, BI. Controversies in the detection and management of nonsteroidal antiinflammatory drug-induced side effects of the upper gastrointestinal tract. Arthritis Rheum 1989;32: 926–932.
  28. Moore, RA. The hidden costs of arthritis treatment and the cost of new therapy—The burden of non-steroidal anti-inflammatory drug gastropathy. Rheumatology (Oxford) 2002;41(suppl 1):7–15.
  29. Singh, G, Mithal, A. Triadafilopoulos G age-adjusted hospitalization rates for complicated gastric and duodenal ulcers in the us: Have cox-2 specific inhibitors and PPIs made any difference? (Abstract). Gastroenterology 2004;126(4 suppl 2):1766.
  30. Gisbert, JP, Gonzalez, L, Calvet, X, et al. Proton pump inhibitors versus H2-antagonists: A meta-analysis of their efficacy in treating bleeding peptic ulcer. Aliment Pharmacol Ther 2001;15(7):917–926.
  31. Lewis, JD, Bilker, WB, Brensinger, C, et al. Hospitalization and mortality rates from peptic ulcer disease and GI bleeding in the 1990s: Relationship to sales of nonsteroidal anti-inflammatory drugs and acid suppression medications. Am J Gastroenterol 2002;97(10):2540–2549.
  32. Straus, WL, Ofman, JJ. Gastrointestinal toxicity associated with nonsteroidal anti-inflammatory drugs. Gastroenterol Clin North 2001;30: 895–920.
  33. van Leerdam, ME, Vreeburg, EM, Rauws, EA, et al. Acute upper GI bleeding: Did anything change? Time trend analysis of incidence and outcome of acute upper GI bleeding between 1993/1994 and 2000. Am J Gastroenterol 2003;98(7):1494–1499. | Article | PubMed | ChemPort |
  34. Lanas, A. Cost stratification of nonsteroidal anti-inflammatory drug-associated gastrointestinal side effects. Med Clin 2000;114(suppl 3):46–53.



List of Investigators

See Table



This study was made possible by funds from grant C03/02 from the National Institute of Research Carlos III and by an unrestricted grant provided by Pfizer Spain to the Spanish Association of Gastroenterology (AEG). Pfizer Spain had no role either in the conduction of the study, in the interpretation of data, or in the writing of the manuscript. Dr. Lanas designed and coordinated the study. Dr. Lanas also contributed to the analysis and interpretation of data and drafted the manuscript. Dr. Perez Aísa coordinated the collection of data from different hospitals and contributed to the analysis of data. The rest of the authors made major contributions to the collection of data, revised data interpretation, and had the opportunity to revise the manuscript: Dr. Lanas has given talk and received research or travel support from AstraZeneca, Merck Sharp & Dhome, Pfizer, Inc., Almirall Prodesfarma, TAP, and Ferrer. Dr. Ponce and Dr. Quintero have given talks and received research and/or teaching grants from AstraZeneca, Merck Sharp & Dhome, Pfizer, Inc., Novartis & Almirall Prodesfarma, Dr. Balanzo receives research, and/or teaching grants from AstraZeneca and Altana; Fernando Carballo receives research and/or teaching grants from Altana. The other authors have not reported conflict of interest.