Decreasing incidence of peptic ulcer complications after the introduction of the proton pump inhibitors, a study of the Swedish population from 1974–2002

In this study all cases of gastric and duodenal ulcer complications diagnosed in Sweden from 1974 to 2002 were identified using the National Hospital Discharge Register (National Board of Health and Welfare, S-10630, Stockholm, Sweden). Since 1987 it is mandatory for hospitals to report to this register. In practise, however, it has been in existence since the early seventies though not in the entire country. We compensated for this when the incidence figures from 1974 to 1986 were calculated, as described below.

The ICD codes used for the search are listed in table 2. In ICD 8 and ICD 9 there are codes named ".99" or ".9" in the last positions which represent "unspecified location without knowledge of bleeding or perforation". These codes does not exist in ICD 10 which means that the patients that were earlier coded as totally unspecified are now included in other codes. This fact could mean that incidences of ulcer complications are underestimated before 1997 when ICD 10 was introduced. During the period 1974–86 between 30–37% of all reported codes for peptic ulcer disease were reported as "totally unspecified". From 1987 to1996 this number was 16–20% according to the National Department of Health. Codes representing ulcer with bleeding as well as perforation are included in both the bleeding and the perforation groups.

The data from the National Department of Health are reliable. In a study from 1994 regarding validity, they found that in 11–14% of the cases the diagnosis reported was not correct if they applied a very strict judgement, i.e. all five figures correct [16]. If they were slightly less stringent, i.e. three figures correct, the inaccurate diagnoses was about 6–8%.

Information on sales of NSAID was obtained from the National Prescription Survey (The National Corporation of Swedish Pharmacies – Apoteket AB). Data on drugs dispensed on prescription during the period 1975 – 1986 used a sample of 1/288 of all dispensed prescriptions. From 1987 – 2002, the sample was 1/25 (4%) of all dispensed prescriptions. However, medication to patients in Nursing Homes, Hospices etc. is only included if their medication had been dispensed on prescription by the pharmacies.

The composition of the combination-product Treo Comp^® was changed in 1985, and the ATC-code was changed accordingly from N02BA71 (ASA, combinations with psycholeptics) to N02AA59 (codeine combinations). The DDD for Treo Comp^® is 3 tablets, corresponding to 1,5 g ASA, but the DDD for ASA is 3 g. One DDD of Treo Comp^® thus corresponds to 0,5 DDD ASA. The figures for Treo Comp^® have been recalculated to DDD's of ASA.

Age-standardised incidence rates were calculated using the Swedish population in 1995 as an external reference. Yearly site, age and gender specific incidence rates were calculated using data from the national population register. Time trends were investigated using linear regression analysis with log(incidence) rates as the dependent variable and the calendar year as an independent variable. An exploratory spline regression analysis of log incidence with 6 uniformly distributed knots was also performed using a piecewise linear function for time. STATA 10.1 was used for the statistical calculations. All p-values are two-sided and p-values below 5% were considered statistically significant.

The ethical committee of Lund University Hospital approved the study.

The sales of prescribed NSAID and ASA in Sweden for the years 1975 through 2002 are presented in figures 5, 6, 7, 8 and 9. In males the sales of NSAID and ASA described as DDD per1000 inhabitants per day rose from 2.3 to13.1 in the age group 0–50 years, from 13.7 to 52.1 in the age group 51–70 years and from 20.3–59.1 in the age group 71–99 years.

The corresponding figures for females were from 3.2 to 20.9 in the age group 0–50 years, from 21.0 to 82.5 in the age group 51–70 years and from 30.6 to 80.8 in the age group 71–99 years.

The sales to females compared to males were higher in all age groups and the difference increased towards the end of the study period.

The total sales of prescribed NSAID/ASA (DDD per1000 inhabitants per day) increased from 13.3 in 1978 to 46.5 in 2002.

The total volume of NSAID and ASA (DDD per1000 inhabitants per day) sold both "over the counter" and prescription sales, increased from 45.0 in 1978 to 57.6 in 2002 according to the statistics of Apoteksbolaget AB, Sweden. This means that sales increased 28% during this period.

The sales of low-dose ASA (DDD per1000 inhabitants per day) in males increased from 2.75 to 3.55 in the age group 0–50 years, from 79.2 to 110.9 in the age group 51–70 years and from 219.0–356.2 in the age group 71–99 years (fig 7).

The sales of low-dose ASA (DDD per1000 inhabitants per day) in females increased from 1.5 to 2.0 in the age group 0–50 years, from 42.3 to 64.2 in the age group 51–70 years and from 156.4–288.9 in the age group 71–99 years (fig 8).

H2 receptor blockers were introduced in Sweden in 1978. In 2002 total sales were 5.7 DDD per1000 inhabitants per day (figure 9).

The first PPI were introduced in Sweden in 1988. In 2002 total sales were 30.5 DDD per1000 inhabitants per day (figure 9).

In the whole material mean age rose from 59.0 years to 67.9 years from 1974 to 2002. The mean age in males with gastric complications rose from 59 years in 1974 to 66.5 years in 2002 The corresponding figures for females were 63.2 and 70.2 years.

The mean age in males with duodenal complications rose from 52.8 years in 1974 to 62.1 years in 2002. In females the corresponding figures were 61.1 and 72.6.

According to the literature, the incidence of peptic ulcer complications appears to be relatively stable over the last three decades with approximately 5 to10 episodes per 100 000 inhabitants/year. Some relevant references are listed in table 1. In elderly women, several authors even report an increasing incidence [11‐14, 27]. At least three studies have compared the incidences of peptic ulcer perforation before and after the introduction of the H2-receptor antagonists but found the incidence to be unchanged [1, 22, 23]. In contrast to these studies, we report a significant decrease in peptic ulcer complications in the Swedish population over the last two decades. Recently, a Danish group report similar results from 1996 to 2004 regarding perforated peptic ulcer but not for bleeding ulcers [15]. Like us they had a relatively large sample size, the Danish study involved a population of 1, 2 million inhabitants. A plausible explanation for the fact that several studies fail to detect any differences might be that the studied populations have often been too small for a disease that has a relatively low incidence to begin with. Differences in the prevalence of the different risk factors for peptic ulcer complications between populations, which will be discussed below, are other possible explanations.

The significant drop in incidence of peptic ulcer complications reported in this study, can possibly be explained by a generational (cohort) influence, for example less exposure to Helicobacter pylori infections among younger generations compared with older birth cohorts. Another possibility is a secular (periodical) phenomenon, for example the influence of extensive exposure of acid-suppressing drugs or changing trends in the use of factors known to cause ulcer or counteract ulcer healing, such as tobacco, alcohol or ulcerogenic drugs. Further, the use of gastroscopy has probably increased during the study period in Sweden enabling the detection and treatment of symptomatic peptic ulcers before complications occur. In our opinion, the explanation for this uniform decline in complications after 1988 is most likely multifactorial.

In the sections below we will discuss some important risk factors for complicated peptic ulcer disease and their possible influence on our results. When interpreting the results one must be aware that this is an ecological study with risk of "ecological fallacy" [28]. The connections discussed below can, of course, be caused by other factors and should therefore be adopted with caution.

We found an increase in the sales of prescribed NSAID during the study period. The number of prescriptions increased with age and was highest among middle-aged and elderly women. As mentioned earlier, elderly women are also the group in which the incidences of peptic ulcer complications are reported to be increasing relative to other groups [1, 11‐14]. However, in conjunction with an increase of prescribed sales there was a fall in non-prescribed sales, resulting in a total increase in NSAID sales of 28% during this period.

Several studies support the notion that NSAID is a risk factor not only in uncomplicated peptic ulcer disease, but also in regard to perforated and bleeding ulcers [2, 3, 5‐7, 10, 29, 30]. A systematic review of 18 case-control studies from 1990 to 1999 reported a pooled RR of 3.6 [2]. The higher risk was maintained during treatment and disappeared after treatment termination. The added risk is dose-dependent and also includes low-dose ASA. Studies report higher risk with all doses, with a RR = 2 for bleeding ulcer with daily intake of 75 mg ASA [8, 9]. According to a study by Sorensen et al from 2000 the risk increase further when low-dose ASA is combined with NSAID [10].

Bearing this in mind one would expect a rise in incidence of peptic ulcer complications when the sales of NSAID increases, which is quit contrary to our findings. Since ours is a descriptive study one can only speculate as to the relationship between peptic ulcer complications and NSAID use. In a recent German paper from 2005 Ohmann et. al. found no difference in the incidence of peptic ulcer bleeding when comparing the years 1989 and 1999 despite significantly higher NSAID consumption in the latter year [26]. Hawkey conducted a prospective, controlled study of patients with bleeding ulcers and found that H pylori-positive NSAID users had twice the risk compared to H pylori negative patients [31]. The prevalence of H pylori is declining and thus could have reduced the effect of increased NSAID use [32‐34].

In the end of the 1990-ies the less ulcerogenic selective cyclo-oxygenase-2 (COX-2) inhibitors were introduced in Sweden. These drugs were not included in this study since they were only available for use in Sweden during the last years of the study period. Christensen et al report a 44% rise in prescriptions of all NSAID after 1997, mainly due to COX-2 inhibitors [15]. The hospitalization rates of peptic ulcer complications after the introduction of COX-2 inhibitors were, however, stable for bleeding ulcers and decreased from 17 to 12/100000 person-years from 1996 to 2004 for perforations. This indicates that the COX-2 selective drugs might have decreased the risk of peptic ulcer complications for the individual NSAID user.

Proton pump inhibitors (PPI) have become one of the most sold drugs in the world and are nowadays also available over the counter in Sweden. It is a well known fact that PPI protect against peptic ulcer complications in NSAID users [3, 35‐38]. Interestingly the introduction of PPI was almost simultaneous with the beginning of a falling incidence in peptic ulcer complications (fig 1, 2, 3, 4). We believe that the extensive use of PPI in the population of Sweden has reduced the incidence of peptic ulcer complications. This conclusion is supported by Vonkeman et al (2007) who performed a nested case-control study from a cohort of 52000 NSAID users [39]. They found a significant risk reduction regarding peptic ulcer complications in NSAID users who used PPI concomitantly. Selective COX-2 inhibitors did not reduce the risk in their study.

In the present study we found a significant decrease in peptic ulcer complications in the mid-eighties that co-inside with the discovery of H pylori as a cause for peptic ulcer disease. H pylori is an important pathogenic factor in uncomplicated peptic ulcer disease, although studies that investigate the connection between H pylori and peptic ulcer complications are somewhat divergent. For example, Gisbert et. al. (2003) calculated a mean prevalence of 68% from 19 studies with a range from 0–100% [40].

We have found five studies comparing patients with peptic ulcer complication patients to patients with non-GI related diseases in the control group. In three of these reports there was no difference in the prevalence of H pylori [5, 41‐44]. Three studies have compared patients with peptic ulcer complications patients to patients with uncomplicated ulcer disease [43, 45, 46]. Surprisingly, the prevalence of H pylori was slightly higher in the latter group.

Helicobacter Pylori is, without a doubt, connected to peptic ulcer disease. In the pathogenesis of perforation and bleeding, however, other factors such as NSAID and smoking are of great importance as well. In the time period covered by this study, the prevalence of H pylori infections has decreased in the western world [47] as has the incidence of peptic ulcer complications in our study (fig 1, 2, 3, 4). Since this is a descriptive study one can only speculate as to whether these two phenomena are correlated or not. There is however no doubt in the literature that eradication of H pylori is mandatory after a peptic ulcer complication, if the patient is H pylori positive. Otherwise the risk of recurrent ulcer disease is high [48‐50].

Smoking is a risk factor for peptic ulcer perforation [5, 18, 51]. A prospective cohort study from Denmark found that smoking > 15 cigarettes daily increased the risk of peptic ulcer perforation 3,5 times [51]. A retrospective study from Norway report that a cohort pattern in prevalence of smoking partly explained the cohort pattern in perforation risk for both sexes [18]. The prevalence of smoking has declined during the last twenty years in Sweden, especially in men, which could perhaps account for fewer ulcer complications [52].