Study says nondrug measures helped in 1918 flu pandemic

Aug 13, 2007 (CIDRAP News) – An analysis of historical records in 43 US cities indicates that the early use of nonpharmaceutical measures, such as isolating the sick and banning public gatherings, saved lives in the influenza pandemic of 1918-19.

Cities that reacted quickly to the pandemic with combinations of nonpharmaceutical interventions (NPIs) generally had lower peak death rates and total deaths than cities that acted more slowly and used fewer precautions, according to the report by researchers from the University of Michigan and the Centers for Disease Control and Prevention (CDC).

The findings "demonstrate a strong association between early, sustained, and layered application of nonpharmaceutical interventions and mitigating the consequences of the 1918-1919 influenza pandemic in the United States," says the report, published last week in the Journal of the American Medical Association.

The authors, led by Dr. Howard Markel of the University of Michigan, say the findings suggest that nonpharmaceutical measures can play a "critical" role in battling future pandemics, along with vaccines and drugs. The study's senior author is Dr. Martin Cetron, director of the CDC's Division of Global Migration and Quarantine.

However, the researchers acknowledge that they don't know how well the interventions were carried out or complied with, and a few cities that were slow to implement NPIs had relatively low death rates. Also, one expert not involved in the study said questions have been raised about the methods the researchers used.

Despite any limitations, the findings influenced the CDC recommendations released back in February on the use of community mitigation measures in a pandemic, according to Markel. The CDC called for using "early, targeted, layered" measures, including isolation of sick patients, voluntary home quarantine of potentially exposed people, school closings, and social distancing for adults (changing work schedules, banning large gatherings).

Markel told CIDRAP News the NPI analysis was completed in December 2006 and was discussed at a stakeholders meeting the CDC held in preparation for writing its recommendations.

Markel and colleagues focused on three main categories of interventions: isolation and quarantine (assessed as one measure), school closings, and bans on public gatherings. They also considered a fourth category of miscellaneous other interventions, such as work-schedule changes, transportation restrictions, and face-mask ordinances. They gathered data on the use of these measures in major US cities between Sep 8, 1918, and Feb 22, 1919, which covered the pandemic's second wave and most of the third wave.

Data on total deaths and death rates in 43 cities were gathered from the US Census Bureau's Weekly Health Index for 1918-19. In addition, the researchers say they examined all available public health documents on NPIs in the pandemic, along with published historical and medical literature, two daily newspapers from each of the 43 cities, military records, and various other sources. The 43 cities included all major cities with complete archival and mortality records.

To estimate how many deaths were due to the pandemic, the researchers used previous estimates of weekly pneumonia and influenza deaths in each city from 1910 through 1916, and counted excess deaths for the period studied. Information on the NPIs was gathered from the newspapers.

The authors also assessed the "public health response time"—defined as the interval between the increase in pneumonia and flu deaths and the initiation of NPIs—and the total days of NPIs. They estimated that it would take 10 days for an NPI to have an effect on death rates.

The researchers report that all the cities used at least one of the three main categories of NPIs, and 15 used all three at the same time. The most common combination was school closures and public gathering bans, used by 34 cities for a median of 4 weeks. All the cities except New York, Chicago, and New Haven (Conn.) closed their schools for some period of time; the median was 6 weeks.

Excess pneumonia and influenza deaths for the 43 cities totaled 115,340, the report says. Deaths per 100,000 population ranged from 210 for Grand Rapids, Mich., to 807 for Pittsburgh. A statistical analysis showed that the NPIs were a significant factor in the variation of excess death rates both within and between cities. Combinations of NPIs generally appeared to have a more significant impact than single interventions, and combinations that included school closures and public gathering bans seemed most effective.

The speed of cities' initiation of NPIs was a key factor. "Overall, cities that implemented nonpharmaceutical interventions earlier experienced associated delays in the time to peak mortality, reductions in the magnitude of the peak mortality, and decreases in the total mortality burden," the report states.

Some scholars have suggested that the varied effects of the successive waves of the pandemic may account for differences in how different cities were affected. The suggestion is that cities hit relatively hard by the first wave in the spring fared better than other cities during the second wave, because many people had immunity after exposure to the virus in the first wave. But the authors of the new study say they found no association between the severity of each wave and the following or preceding one.

Similarly, they found no links between pandemic mortality and total population, population density, sex distribution, or age distribution.

Many of the cities experienced two peaks in weekly mortality, with the second peak following the cancellation of NPIs prompted by the preceding wave of cases, the article says. None of the cities had a second peak while the first set of NPIs was still in effect.

The report acknowledges that the observed associations between early NPIs and reduced deaths were not entirely consistent. For example, Grand Rapids had the lowest cumulative death rate (210 per 100,000) despite having a fairly slow public health response time of 17 days.

Markel also acknowledged in an interview that it was not possible to determine how fully the various NPIs were enforced. "You can never discern how fully an NPI was carried out," he said. For example, "There were bans on public gatherings, but there also were public gatherings, sometimes in the same cities." He also said cities varied in how they enforced isolation and quarantine at the time.

Nonetheless, it was easy to determine when the various measures were started and stopped, he said, adding, "I feel very comfortable that the main nonpharmaceutical interventions that we measured . . . were conducted, performed and responded to by large numbers of people."

"The bottom line is that you can't let the perfect be the enemy of the good," Markel said. "None of these [NPIs] are absolute in the history of epidemics, but the question is whether they have an impact, and that's what we tried to assess, by looking at associations."

Kathleen M. Neuzil, MD, MPH, chair of the Pandemic Influenza Task Force of the Infectious Diseases Society of America, said Markel and colleagues deserve praise for producing a difficult piece of research with "fascinating and informative" results. But she noted some limitations.

In addition to the lack of data on compliance with NPIs, Neuzeil told CIDRAP News, there was no true "control city," meaning a city that did not adopt any of the major interventions. In addition, "It is difficult to rule out other differences between the cities that could affect mortality rates—eg, do cities that implement public health interventions sooner and more thoroughly also have better medical care or sanitation practices in general that might contribute to decreased mortality?"

Another expert, Michael T. Osterholm, PhD, MPH, said methodologic questions have been raised about the study.

Osterholm, director of the University of Minnesota Center for Infectious Disease Research and Policy, publisher of CIDRAP News, said, "We all want nonpharmaceutical interventions to have the most impact possible in a world where vaccines are not likely to be available and antivirals in short supply. However, I'm aware that there have been methodologic questions raised about this effort by others who have also studied the historical relationship between nonpharmaceutical interventions and pandemics, and who disagree with both the facts and the conclusions of this study. Therefore I think the final chapter has not yet been written on what we might expect from such interventions in the next pandemic."

Neuzil, a senior clinical advisor at PATH, a Seattle-based nonprofit group that works on global health problems, said one major message of the study is that even in cities that seemed to use NPIs effectively, the pandemic still had a substantial impact on illness and death rates. Therefore, vaccines should be the primary focus of efforts to mitigate a future pandemic, with antiviral drugs as an important supportive measure, she said.

Further, she said, "Nonpharmaceutical interventions may be appropriate first measures to delay the devastating effects of the pandemic while vaccine and antiviral resources are mobilized. However, anything but short-term use of these measures could have major effects on societal functioning and the economy that should be considered." For example, she said, many day care centers could suffer financially from school closings.

Markel HM, Lipman HB, Navarro JA, et al. Nonpharmaceutical interventions implemented by US cities during the 1918-1919 influenza pandemic. JAMA 2007 Aug 8;298(6) [Abstract]

See also:

Feb 1 CIDRAP News story "HHS ties pandemic mitigation advice to severity"

Sort Results


Newsletter Sign-up

Get CIDRAP news and other free newsletters.

Sign up now»


Unrestricted financial support provided by

Bentson Foundation 3M Gilead 
Grant support for ASP provided by

  Become an underwriter»