SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Aliment Pharmacol Ther 2010; 32: 901–907

Summary

Background  The risk of dying from gastric cancer appears to have increased among consecutive generations born during the 19th century.

Aim  To follow the time trends of hospitalization for gastric cancer and test whether they confirm such increase.

Methods  Inpatient records of the last two centuries from four hospitals in Scotland and three US hospitals were analysed. Proportional rates of hospitalization for gastric cancer, gastric ulcer and duodenal ulcer were calculated during consecutive 5-year periods.

Results  The data from all seven cities revealed strikingly similar patterns. No hospital admissions for gastric cancer or peptic ulcer were recorded prior to 1800. Hospital admissions for gastric cancer increased in an exponential fashion throughout the 19th and the beginning of the 20th century. In a majority of cities, the rise in hospitalization for gastric cancer preceded a similar rise in hospitalization for gastric ulcer. Hospitalization for these two latter diagnoses clearly preceded hospitalization for duodenal ulcer by 20–40 years.

Conclusions  The occurrence of gastric cancer, gastric ulcer and duodenal ulcer markedly increased during the 19th century. Improvements in hygiene may have resulted in the decline of infections by other gastrointestinal organisms that had previously kept concomitant infection by Helicobacter pylori suppressed.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Gastric cancer, gastric ulcer and duodenal ulcer are linked to upper gastrointestinal infection with Helicobacter pylori. The infection is found ubiquitously in human populations from all continents.1Helicobacter pylori is thought to have emerged out of Africa and followed similar migration routes as taken by the first humans thousands of years ago.2 However, the risk of dying from gastric cancer or peptic ulcer increased only among generations born during the 19th century.3, 4 After reaching a peak shortly before the turn of the century, the risk started to fall again among people born during the 20th century. Whereas the recent fall is explained by a declining infection of the general population with H. pylori,5 the initial rise has remained mysterious. The present study followed the time trends of hospitalizations for gastric cancer, gastric and duodenal ulcer during the 19th century to test whether the initial rise observed in the mortality data can also be observed in other morbidity statistics. Parts of the hospitalization data of gastric and duodenal ulcer have been published in two previous publications.6, 7 These data were included in the present study primarily for comparison with the time trends of gastric cancer, which are new and have not been analysed before.

Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Data sources

The surviving annual inpatient records of the last two to three centuries from the Scottish Royal Infirmaries of Aberdeen, Dundee, Edinburgh, and Glasgow were analysed. These hospitals were chosen because the quality of record keeping in these hospitals was exceptionally meticulous and because their records contained detailed descriptions of all inpatients diagnoses. (i) The Aberdeen Infirmary opened in 1742 and there are diagnostic data from that date, at first compiled from case books and the minutes of the hospital’s managers, and then from 1752 in tabular form in registers of admissions and discharges (except for 1823–1837 when outcomes were not recorded). Annual reports were published from the late 1830s, and the statistical tables were printed until 1917 and exist in manuscript from 1922 to 1934. (ii) The Dundee Royal Infirmary opened in 1798, but diagnostic data for inpatients are available only from 1839 to 1916. (iii) The Edinburgh Royal Infirmary opened in 1729, but its early registers for 1730 to 1759 have not survived. Further admission data are available from 1800–1804, 1839–1849, 1854–1864, 1870 and 1890–1914. (iv) The Glasgow Royal Infirmary has data from its opening in 1794 up to 1930 and was the only centre for acute admissions until the opening in 1874 of the Western Infirmary and of the Victoria Infirmary in 1890.

For comparison and confirmation of the data from Scotland, US hospitals with matching quality of historical records were sought. Eventually, the epidemiological analysis of gastric cancer, gastric and duodenal peptic ulcer in the United States during the 18th and 19th century came to rely on the hospital statistics from three cities, that is, Boston, New York and Philadelphia. (i) The Boston City Hospital opened in 1864 as the first municipal hospital in the United States. The hospital discharges stratified by diagnosis were available through its annual reports from 1864 until 1915. (ii) The New York Hospital was founded in 1782 and has diagnostic records of admissions from 1792 to 1918. (iii) The Pennsylvania Hospital has in its archives manuscript ledgers of admissions from its foundation in 1752 until 1918. Cumulative annual reports were published first as broadsheets and later as booklets. Between 1803 and 1823, only monthly lists were produced and their data have been converted to annual figures. Missing volumes or their microforms were sought in the libraries of the College of Physicians, the American Philosophical Society and the Library Company, but data from 1849 to 1852 and for 1901 could not be found, and have been estimated by linear extrapolation between the preceding and subsequent data.

The resident populations of the four cities from Scotland are available through the decennial census of the British population. The resident populations of the three US cities are available through decennial census tabulations of the US Census Bureau, together with local censuses.

Statistical analyses

Hospitalization for gastric cancer, gastric ulcer, and duodenal ulcer were analysed separately. For each hospital, the numbers of patients diagnosed with any of the three diseases were accumulated over consecutive 5-year periods (quinquennia). The cumulative number of cases in each disease category from five consecutive years was divided by the total number of hospitalizations during the same period and expressed as annual proportional rate per 1000 hospitalizations. The cumulative number of cases was also divided by the resident population of the same period and expressed as population-based rate per million population. In few instances, missing data were estimated by linear extrapolation between data preceding and following a period of incomplete data. The hospital rates were plotted against the time of diagnosis and smoothed by calculating a moving average of three consecutive time periods. The numbers of patients per 5-year periods were treated as Poisson variables with corresponding 95% confidence intervals. Two rates were considered statistically different, if their 95% confidence intervals did not overlap. For most time periods, the hospitalization data were not stratified by age groups and no age adjustment was possible in the comparison of the time trends.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Table 1 contains the seven hospitals and their respective time periods, for which data were available. For comparison with hospitalizations secondary to gastric cancer, gastric ulcer and duodenal ulcer, the Table also lists the total number of hospital admissions during the latest decennium, for which data were available from all hospitals. In each city, the various hospitals served different fractions of the population, which is responsible for some of the variation in the population based rates among the different cities. Hospitalizations for gastric cancer expressed as proportional fraction of total hospitalizations show a smaller variation across different cities than the population-based rates. The same pattern applies to the proportional rates of gastric or duodenal ulcer as opposed to their population-based rates. As the primary focus of the present analysis has been the time trends of gastric cancer during the 19th century, more attention should be paid to the temporal behaviour of the rates from individual cities than to their absolute magnitude. In the following sections, the analysis of time trends is focused on the proportional hospital rates. Almost identical patterns are observed in the time trends of population-based hospital rates.

Table 1.   Number of hospitalizations associated with gastric cancer and peptic ulcer
CityHospital namePeriodPopulation in 1900Admissions in 1900–1909
TotalGCAGUDU
  1. GCA, gastric cancer; GU, gastric ulcer; DU, duodenal ulcer.

Scotland
 AberdeenAberdeen Royal Infirmary1742–1934144 00026 63814628226
 DundeeDundee Royal Infirmary1835–1919159 00036 80712519416
 EdinburghEdinburgh Royal Infirmary1729–1912317 000103 0771300702221
 GlasgowGlasgow Royal Infirmary1795–1930762 00075 02621656732
United States
 BostonBoston City Hospital1865–1915561 00086 12428430749
 New YorkNew York Hospital1792–19061 970 000108 81716917623
 PhiladelphiaPennsylvania Hospital1752–19181 421 00044 602391137

No patients with gastric cancer or gastric ulcer were seen at the Aberdeen Royal Infirmary between 1742 and 1784. The first patient with gastric cancer was admitted to the hospital in 1787; however, admissions for gastric cancer started to become a common feature only after 1841. Hospital admissions for gastric ulcer and duodenal ulcer started only in 1874 and 1898 respectively. During the 19th century, the rate of hospital admission for all three diagnoses increased exponentially, with the rates of gastric ulcer and duodenal ulcer eventually overtaking those of gastric cancer (Figure 1). The rise was statistically significant for all three diagnoses.

image

Figure 1.  Time trends of hospitalizations for gastric cancer (GCA), gastric ulcer (GU), and duodenal ulcer (DU) from four cities in Scotland.

Download figure to PowerPoint

When hospital records in Dundee became available in 1839, gastric cancer was still a rare yet regular diagnosis among inpatients. The first cases of gastric ulcer and duodenal ulcer were treated among inpatients in 1853 and 1899, respectively. Subsequently, the hospital rates for the three diagnoses increased, the rise being statistically significant in all instances. Hospital admissions for gastric ulcer became more common than those for gastric cancer around the turn of the century, whereas the data from Dundee captured only the initial rise in the occurrence of duodenal ulcer (Figure 1).

During the initial period between 1729 and 1844, no patients with gastric cancer or gastric ulcer were hospitalized at the Edinburgh Royal infirmary. The first patients with gastric cancer and gastric ulcer were admitted in 1848 and 1845 respectively, although the subsequent rise was more pronounced in gastric cancer than gastric ulcer. Duodenal ulcer first occurred among inpatients only after 1891. Again, the subsequent rise was statistically significant for all three diagnoses.

No inpatients with gastric cancer were treated in Glasgow Royal Infirmary from 1794 until 1820, the first patient with gastric cancer being admitted in 1821. Three patients with gastric ulcer were admitted to the hospital as early as 1817, but gastric ulcer became a regular feature among inpatients only after 1855. Patients with duodenal ulcer became admitted to the hospital only in 1904. After their first appearance, the rates of the three diagnoses increased exponentially, the increase being statistically significant in all three instances. Inpatient diagnosis of gastric and duodenal ulcer became more frequent than gastric cancer after 1880 and 1920 respectively.

Figure 2 shows the time trends of hospitalization for gastric cancer, gastric ulcer and duodenal ulcer among the three American cities. When data from the Boston City Hospital became available, gastric cancer and gastric ulcer already featured regularly among inpatient diagnoses from 1866 onward. Duodenal ulcer did not start as inpatient diagnosis until 1895. The inpatient data from the New York Hospital between 1792 and 1859 do not contain a single case of gastric cancer, which started to occur among inpatients only in 1862. Gastric ulcer started only in 1877 and duodenal ulcer only in 1895. No cases with gastric cancer were admitted to the Pennsylvania Hospital from 1752 until 1881. Its first inpatient with gastric cancer was admitted in 1882. The first inpatients with gastric ulcer and duodenal ulcer were admitted in 1874 and 1903 respectively. In all three US cities alike, the rates of the three diagnoses increased significantly during the second half of the 19th century and the beginning of the 20th century. In Philadelphia and New York, gastric ulcer became more frequent than gastric cancer shortly before or after the turn of the century, respectively. Compared with the trends from Scotland, the onset of all three diagnoses in the US appeared somewhat delayed and remarkably fewer.

image

Figure 2.  Time trends of hospitalizations for gastric cancer (GCA), gastric ulcer (GU), and duodenal ulcer (DU) from three cities in the United States.

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

The present study used hospital data to follow the time trends of gastric cancer during the 19th century and compare them with the corresponding trends of gastric and duodenal ulcer. The aim of the study was to corroborate previous trends of mortality data, which suggested rising trends of gastric cancer among subjects born during the 19th century. The present data show for the first time an unequivocal rise in the occurrence of gastric cancer throughout the 19th century. The various hospital data from four hospitals in Scotland and three hospitals in the United States revealed a striking conformity. In all data sets alike, hospitalization secondary to gastric cancer hardly existed prior to the beginning of the 19th century. It increased in an exponential fashion throughout the 19th and the beginning of the 20th century. In most of the cities, the rise in hospitalization for gastric cancer seemed to precede a similar rise in hospitalization for gastric ulcer. Hospitalization for both diagnoses clearly preceded hospitalization for duodenal ulcer by 20–40 years.

One potential limitation of the data relates to the validity of medical diagnoses established one hundred to two hundred years ago when physicians rarely unclothed patients. Without the availability of endoscopy and X-ray imaging, obviously, medical descriptions from the 19th century did not meet the present day standards of diagnosis. However, one should not summarily disregard the clinical acumen of previous generations of physicians, who saw a large clientele of patients with common digestive diagnoses and were exposed to frequent necropsy in instances of death.8 Gastric cancer, gastric ulcer and duodenal ulcer were mostly diagnosed based on their acute clinical course accompanied by severe abdominal pain and subsequent complications, such as gastrointestinal haemorrhage, anaemia, gastric outlet obstruction, intestinal perforation, and death. As a majority of cases were diagnosed through their associated complications or through postmortem examinations, it is likely that many instances of less severe ulcer disease went undiagnosed. Other factors that may potentially confound the analysis of historical trends include incomplete record keeping and unreliable population estimates. It is not always possible to assess with certainty the catchment population for individual hospitals. For these reasons, the analysis was focused on proportional hospital rates, although the population-based rates showed almost identical time trends. In spite of their obvious limitations, the historic statistics can still provide a valuable window into the early epidemiology of these common upper gastrointestinal disorders, especially, if assessed in conjunction with other supportive evidence.

The time trends of gastric and duodenal ulcer have been shown to be shaped by an underlying birth-cohort pattern.3 The risk for peptic ulcer disease increased among generations born throughout the 19th century. It peaked among generations born shortly before the turn of the 19th to 20th century and then decreased among all subsequent generations. This temporal behaviour applied similarly to gastric and duodenal ulcer, although the rise and fall of gastric ulcer preceded those of duodenal ulcer by about 10–20 years. It was recently shown that the time trends of gastric cancer also followed a birth-cohort pattern.4 The initial rise and the subsequent peak of gastric cancer among consecutive birth-cohorts preceded similar time trends of gastric ulcer by about 20–30 years. The birth-cohort patterns of gastric cancer and peptic ulcer are seemingly ubiquitous. It can be confirmed in the death rates of peptic ulcer and gastric cancer from most European countries, the United States, Canada, Australia, and Japan.3, 4, 9 The fall in the incidence of gastric cancer and both ulcer types during the past decades is confirmed by a large variety of different health statistics.10–12 It has been ascribed to the declining infection of the general population with H. pylori.5 The initial rise in the occurrence of these three diseases has remained difficult to explain. The present study was undertaken to confirm the rise of gastric cancer in a dataset that is different from the vital statistics. The analysis of cancer and ulcer mortality relied on health statistics generated during the 20th century. Plotting mortality data against the period of birth and their analysis by birth-cohorts opens a window into the time trends of 19th century. By contradistinction, the data of the present analysis is based on data truly generated and assembled during the 19th century itself. The data dealing with gastric and duodenal ulcer have been shown in parts before and are shown here primarily for comparison with the data of gastric cancer.6, 7 The data of gastric cancer confirm the birth-cohort pattern described by the mortality data.4 They also extend the time frame of the previous analysis and clearly indicate that gastric cancer was increasing in the population of Scotland, as well the United States.

Gastric cancer, gastric ulcer and duodenal ulcer are all related to infection of the upper gastrointestinal mucosa with H. pylori. Infection with H. pylori has been confirmed to be ubiquitous among all human populations spread throughout the globe.1Helicobacter pylori is currently thought to have emerged out of Africa and follow similar migration paths as taken by early humans thousands of years ago.2 It appears to have been endemic in human populations since the dawn of mankind. If this is the case, why would gastric cancer, gastric ulcer and duodenal ulcer suddenly start to rise during the 19th century? As data from Japan suggest, this phenomenon has not been restricted to the Western populations of Europe and North America.4 It is also amazing to find similar and seemingly synchronized epidemics of three different gastro-duodenal diseases affecting people in Scotland, as well as the east coast of the United States. The rise in the occurrence of the three diagnoses could stem from varying exposure to environmental risk factors or temporal changes associated with H. pylori infection. Other bacteria besides H. pylori colonize the upper gastrointestinal tract.13 Such species probably compete with H. pylori for its particular ecologic niche. Lactobacillus species, for instance, have been shown to be able to suppress growth of H. pylori in vitro.14 Due to technological discoveries and social advances during the 19th century, industrialized nations witnessed marked improvements in the standards of hygiene affecting all strata of the population.15 These changes in hygiene may have resulted in the decline of infections by other gastrointestinal organisms that had previously kept any concomitant infection by H. pylori suppressed. Improving hygiene could, thus, have caused initially a rise in the occurrence of H. pylori-related diagnoses, before further improvements ultimately led to a decline of the H. pylori infection as well.

In conclusion, hospitalizations for gastric cancer, gastric ulcer and duodenal ulcer increased during the 19th century. Hospital statistics support the time trends revealed initially by the birth-cohort analysis of mortality data. The reasons for the initial rise in the occurrence of gastric cancer, as well as gastric and duodenal ulcer, are presently unknown. They may be linked to the emergence of H. pylori or changes in its mode of acquisition.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

We thank many archivists and librarians for their gracious help: Aberdeen – (Fiona Watson); Dundee Dundee University Archives (Michael Bolik); Edinburgh Lothian Health Services Archives (Michael Barfoot and Alison Gardiner) and The Royal Colleges of Physicians (Iain Milne) and Surgeons (Steve Kerr) of Edinburgh; Glasgow The Royal College of Physicians and Surgeons of Glasgow (Charlotte Crooks, Valerie McClure) and NHS Greater Glasgow Archives (Alistair Tough); Boston Francis A Countway Library (Jack Eckart) and Massachusetts Historical Society (Kimberley Nusco); New York New York Hospital (Adèle Lerner); Philadelphia – American Philosophical Society (Robert Cox, Valerie-Anne Lunz), City Archives, College of Physicians (Evin McLeary, Edward Morman, Christopher Stanwood, Gretchen Worden), Library Company (Phil Lapsansky), and Pennsylvania Hospital (Stacey Peeples). Declaration of personal and funding interests: None.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References