American Samoa

The maritime quarantine of American Samoa completely isolated the island during the 1918 pandemic period; there were no reports of influenza-related deaths during the pandemic period. The success of the quarantine of American Samoa reflected tight control of both the naval base at Pago Pago (by the U.S. Navy) and the extensive coastal areas (by island residents). Of note with regard to the latter, residents of Western Samoa used ocean-going canoes to seek shelter from the influenza epidemic on their home islands on nearby American Samoa. On American Samoa, entry restrictions – including a several-day wait prior to disembarking – were maintained at least until 1920. Influenza eventually arrived in American Samoa in 1926; during the ensuing epidemic, at least one-fourth of the island residents were clinically affected by influenza-related illnesses but few died. Overall mortality during the 1926 epidemic was estimated as 1/1000 island residents which was about 200 times lower than the overall mortality on Western Samoa during the 1918 pandemic-related epidemic [Reference Ravenholt and Foege4, Reference Larsen5].

Table 1.

Six Pacific islands or groups of islands (American Samoa, New Caledonia) and their influenza epidemic mortality experience following being missed by the main pandemic wave of the 1918–1919 influenza pandemic

New Caledonia

Prior to 1918, Australia was a major supplier of goods to New Caledonia. However, during the 1918–1919 influenza pandemic period, restrictions instituted by the Australian Director of Quarantine ended nearly all shipping to the islands for several months. During the quarantine period, influenza was not reported on New Caledonia other than in individuals who had been isolated at the port. In July 1921, an influenza epidemic affected New Caledonia [Reference Peltier6]. During the epidemic, an estimated 80% of the local and 42% of the European-origin population became ill; however, few influenza-related deaths were reported (deaths: local inhabitants, n = 30; Europeans, n = 4). The island's population at the time was estimated as 33 000; thus, overall influenza-related mortality during the epidemic was about 1/1000 [about 160 times lower than the overall mortality (16%) in residents of French Polynesia during a 1918 pandemic-related epidemic] [Reference Peltier6].

Rotuma

Rotuma was affected by a ‘virgin soil’ measles epidemic in 1911; during the epidemic, about 16% of the island's population died. To the colonial administrators of the island, the experience highlighted the exceptional vulnerability of very isolated populations to first-time exposures to clinically virulent infectious agents [Reference Shanks7]. From November 1918 to February 1919, all shipping to Rotuma was stopped to prevent influenza from reaching the island; the preventive measures were apparently successful. A review of mortality records on Rotuma from 1903 to 1960 documented sharp increases of influenza-related mortality in 1927–1928 and 1947–1948. During the respective epidemic periods, 3·8% and 1·9% of the Rotuman population died of pneumonia/influenza (according to causes of death listed in mortality registers). Of note, pneumonia/influenza was a leading cause of death on Rotuma during much of the 20th century [Reference Lambert8].

Yule Island (Mekeo)

Catholic mission records from Yule Island (Mekeo) on the southern coast of Papua documented mortality on the island throughout the first half of the 20th century [Reference Scragg9]. A review of the records revealed sharp increases in baseline annual mortality during periods of widespread influenza-like illness on Yule Island; such ‘influenza epidemic periods’ occurred in 1920, 1931–1932, 1940–1942 and 1948. Of note, during the 1931, 1941, and 1948 epidemic periods, influenza-related mortality was relatively high in young adults, e.g. 35–43% of all deaths were in the 20–49 years age group. The relatively high influenza-related mortality in young adults during influenza epidemics on Yule Island reflected the increased mortality risk in young adults in general (W-shaped age-mortality curve) during the 1918–1919 pandemic period [Reference Scragg9].

Niue

Prior to the Second World War, Niue was an extremely isolated island; it had no airfield and was about 3000 km from its colonial administrators in New Zealand. As a result, during this period, there were very few arrivals to the island, and all were by sea. During the Second World War, an airfield was constructed on the island; contacts between previously isolated islanders and outsiders sharply increased. Influenza epidemics occurred in 1942, 1944, 1949 and 1957 [Reference Taylor10]. During the 1944 epidemic, there were 2351 cases and 129 deaths; case fatality was about 5·5% and overall mortality in island residents was about 4%. During influenza epidemics in 1976, 1977, 1979, 1981 and 1983, there were high illness attack rates (>30%) but very few deaths (e.g. two in 1983).

Jaluit

Jaluit is a small atoll in the southern Marshall Islands. An influenza epidemic occurred in 1910 which killed an estimated 100 island residents – about 10% of the island population. To our knowledge, the 1918–1919 influenza pandemic did not affect Jaluit. An influenza epidemic in 1931 killed 140 persons – about 8·6% of the population [Reference Spennemann11].

American Samoa and New Caledonia were established colonial outposts that escaped the 1918–1919 pandemic because of rigid restrictions on maritime traffic [Reference Cumpston1]. Within a few years after lifting these restrictions, influenza caused high morbidity but relatively low mortality during epidemics on the islands. Of particular note, the influenza-related mortality on these islands during post-pandemic epidemics was much lower than the mortality during epidemics on similar islands during the 1918–1919 pandemic period [Reference Tomkins2, Reference Peltier6]. In contrast, on the very isolated islands of Rotuma, Yule Island, Niue, and Jaluit, post-pandemic influenza epidemics – some as long as several decades after the 1918–1919 pandemic period – caused high morbidity and high mortality [Reference Lambert8–Reference Spennemann11]. Of particular note, influenza-related mortality on these islands during post-pandemic epidemics was comparable to that on similar islands during 1918 pandemic-related epidemics. Together, the findings of this report suggest that extreme social isolation is a significant determinant of mortality risk during influenza epidemics.

There are limitations to this report that should be considered when interpreting the findings. During the first half of the 20th century, the documentation of morbidity and mortality on Pacific islands – particularly, on small isolated islands – was variably complete and accurate. Moreover, in the absence of laboratory tests for influenza viruses, the reliability of reports or imputations of ‘influenza’ as the primary causes of epidemics or epidemic-related deaths is questionable. Some islands (particularly, small isolated islands) considered ‘missed’ for this report may have been affected by pandemic influenza; and some epidemics and deaths attributed to ‘influenza’ for this report may have been caused at least in part by other agents.

Genomic studies with mathematical modelling suggest that three lineages of influenza A(H1N1) viruses were circulating prior to 1918 [Reference Smith12]. As such, it is likely that influenza viruses, other than the 1918 pandemic strain, were circulating in the Pacific region during the post-pandemic period. The reader is referred to our recent hypothesis publication for our justification for thinking that multiple viruses were involved [Reference Shanks and Brundage13]. The huge difference in mortality between the highly lethal epidemic on Western Samoa in 1918 and the much less deadly epidemic on American Samoa in 1926 suggests that the epidemics were caused by different influenza viruses [Reference Tomkins2].

In the early 20th century, residents of isolated Pacific islands compared to those from more socially interconnected communities were probably exposed to fewer and less diverse respiratory infectious agents. Such differences were probably reflected in the characteristics of their immune systems and the natures and degrees of their immunological responses to novel infectious agents including but not limited to influenza virus [Reference Shanks, Hussell and Brundage14]. During post-pandemic influenza epidemics on the isolated islands, both adults and children died at high rates. The finding suggests that the immune systems of human hosts may require multiple exposures to diverse respiratory infectious agents in order to effectively – but not excessively – respond to antigenically novel and potentially virulent respiratory infectious agents.