ORIGINAL CONTRIBUTION Nonpharmaceutical Interventions Implemented by US Cities During the 1918-1919 Influenza Pandemic Howard Markel, MD, PhD Harvey B. Lipman, PhD J. Alexander Navarro, PhD Alexandra Sloan, AB Joseph R. Michalsen, BS Alexandra Minna Stern, PhD Martin S. Cetron, MD T HE INFLUENZA PANDEMIC OF 1918-1919 was the most deadly contagious calamity in human history. Approximately 40 million individuals died worldwide, including 550 000 individuals in the United States.1-4 The historical record demonstrates that when faced with a devastating pandemic, many nations, communities, and individuals adopt what they perceive to be effective social distancing measures or nonpharmaceutical interventions including isolation of those who are ill, quarantine of those suspected of having contact with those who are ill, school and selected business closure, and public gathering cancellations.5,6 One compelling question emerges: can lessons from the 1918-1919 pandemic be applied to contemporary pandemic planning efforts to maximize public health benefit while minimizing the disruptive social consequences of the pandemic as well as those accompanying public health response measures?7-10 Most pandemic influenza policy makers agree that even the most rigorous nonpharmaceutical interventions are unlikely either to prevent a pandemic or change a population’s underlying biological susceptibility to the pandemic virus. However, a growing 644 Context A critical question in pandemic influenza planning is the role nonpharmaceutical interventions might play in delaying the temporal effects of a pandemic, reducing the overall and peak attack rate, and reducing the number of cumulative deaths. Such measures could potentially provide valuable time for pandemic-strain vaccine and antiviral medication production and distribution. Optimally, appropriate implementation of nonpharmaceutical interventions would decrease the burden on health care services and critical infrastructure. Objectives To examine the implementation of nonpharmaceutical interventions for epidemic mitigation in 43 cities in the continental United States from September 8, 1918, through February 22, 1919, and to determine whether city-to-city variation in mortality was associated with the timing, duration, and combination of nonpharmaceutical interventions; altered population susceptibility associated with prior pandemic waves; age and sex distribution; and population size and density. Design and Setting Historical archival research, and statistical and epidemiological analyses. Nonpharmaceutical interventions were grouped into 3 major categories: school closure; cancellation of public gatherings; and isolation and quarantine. Main Outcome Measures Weekly excess death rate (EDR); time from the activation of nonpharmaceutical interventions to the first peak EDR; the first peak weekly EDR; and cumulative EDR during the entire 24-week study period. Results There were 115 340 excess pneumonia and influenza deaths (EDR, 500/ 100 000 population) in the 43 cities during the 24 weeks analyzed. Every city adopted at least 1 of the 3 major categories of nonpharmaceutical interventions. School closure and public gathering bans activated concurrently represented the most common combination implemented in 34 cities (79%); this combination had a median duration of 4 weeks (range, 1-10 weeks) and was significantly associated with reductions in weekly EDR. The cities that implemented nonpharmaceutical interventions earlier had greater delays in reaching peak mortality (Spearman r=−0.74, P⬍.001), lower peak mortality rates (Spearman r=0.31, P=.02), and lower total mortality (Spearman r=0.37, P=.008). There was a statistically significant association between increased duration of nonpharmaceutical interventions and a reduced total mortality burden (Spearman r=−0.39, P=.005). Conclusions These 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. In planning for future severe influenza pandemics, nonpharmaceutical interventions should be considered for inclusion as companion measures to developing effective vaccines and medications for prophylaxis and treatment. Author Affiliations: Center for the History of Medicine, University of Michigan Medical School, Ann Arbor (Drs Markel, Navarro, and Stern, and Ms Sloan and Mr Michalsen); and Division of Global Migration and Quarantine, Centers for Disease Control and Pre- JAMA, August 8, 2007—Vol 298, No. 6 (Reprinted) Downloaded From: https://jamanetwork.com/ on 04/13/2020 www.jama.com JAMA. 2007;298(6):644-654 vention, Atlanta, Georgia (Drs Lipman and Cetron). Corresponding Author: Martin S. Cetron, MD, Division of Global Migration and Quarantine, Centers for Disease Control and Prevention, 1600 Clifton Rd, Mailstop E-03, Atlanta, GA 30333 (mcetron@cdc.gov). ©2007 American Medical Association. All rights reserved. INTERVENTIONS DURING 1918-1919 INFLUENZA PANDEMIC body of theoretical modeling research suggests that nonpharmaceutical interventions might play a salubrious role in delaying the temporal effect of a pandemic; reducing the overall and peak attack rate; and reducing the number of cumulative deaths.11-15 Such measures could potentially provide valuable time for production and distribution of pandemic-strain vaccine and antiviral medication. Optimally, appropriate implementation of nonpharmaceutical interventions would decrease the burden on health care services and critical infrastructure. The historical record of the 19181919 influenza pandemic in the United States constitutes one of the largest recorded experiences with the use of nonpharmaceutical interventions to mitigate an easily spread, high mortality and morbidity influenza virus strain (ie, a category 4-5 pandemic using the Centers for Disease Control and Prevention February 2007 Interim Pre-Pandemic Planning Guidance).16 Our study focused on this data set by assessing the nonpharmaceutical interventions implemented in 43 cities in the continental United States from September 8, 1918, through February 22, 1919, a period that encompasses all of the second pandemic wave (SeptemberDecember 1918) and the first 2 months of the third wave (January-April 1919) and represents the principal time span of activation and deactivation of nonpharmaceutical interventions. The purpose was to determine whether city-tocity variation in mortality was associated with the timing, duration, and combination (or layering) of nonpharmaceutical interventions; altered population susceptibility associated with prior pandemic waves; age and sex distribution; and population size and density. METHODS Data Collection We combined systematic historical data collection and contemporary epidemiological and statistical analytic tools. MortalitydatawereobtainedfromtheUSCensus Bureau’s Weekly Health Index17 for 1918-1919, a series of reports listing total deaths and death rates for 43 large US cit- ies. These 43 cities were among the 66 most populous urban centers according to the 1920 census, and all had a population greater than 100 000. Of the 66 most populous cities, the remaining 23 hadincompletearchivalandmortalityrecords. No city with a comprehensive archival record of nonpharmaceutical interventions was excluded. The Weekly Health Index is the most complete extant compilation of weekly pneumonia and influenzamortalitydatainUSurbanareas during the 1918-1919 pandemic. In addition, we captured all of the available public health documents on nonpharmaceutical interventions implementedbythese43citiesduringthe19181919 pandemic, including municipal public health department annual and monthly reports and weekly bulletins; everystateandfederalreportonthe19181919 influenza pandemic published between 1917 and 1922; US Census pneumonia and influenza mortality data from 1910-1920; the corpus of published historical, medical, and public health literature on the 1918-1919 pandemic; 86 different newspapers from the 43 different cities; records of US military installations between1917-1920;andadditionalholdings housed in several major libraries and archival repositories (the complete bibliography of the 1144 primary and secondary sources is available as an online supplement at http://www.cdc.gov /ncidod/dq/index.htm).17-23 Data Analysis From the census reports, we extracted the weekly pneumonia and influenza mortality data covering the 24 weeks from September 8, 1918, through February 22, 1919, for the 43 US cities. In 1920, these 43 cities had a combined population of approximately 23 million (22% of the total US population). A small number of missing values (846 [0.6%] of 136 563 deaths) was imputed. Using estimated weekly baseline pneumonia and influenza death rates generated from the 1910-1916 median monthly values found by Collins et al,18 weekly excess death rates (EDR) were computed. Based on available mortality data and epidemiological reports from ©2007 American Medical Association. All rights reserved. Downloaded From: https://jamanetwork.com/ on 04/13/2020 that era, as well as a recent retrospective statistical analysis, we estimated that those who succumbed to influenza contracted it 10 days earlier.3,24-27 The onset of the epidemic in a particular city was estimated as either the day of the first reported pneumonia and influenza case, or the calendar day of the first recorded pneumonia and influenza death minus 10 days, whichever was earlier. Information on nonpharmaceutical interventions was captured by reviewing at least 2 daily, high-circulation newspapers for each city and available municipal or state health reports. Nonpharmaceutical interventions were grouped into 3 major categories: school closure; public gathering bans; and isolation and quarantine. We also considered an additional general category of ancillary nonpharmaceutical interventions (eg, altering work schedules, limited closure or regulations of businesses, transportation restrictions, public risk communications, face mask ordinances). Nonpharmaceutical interventions were considered either activated (“on”) or deactivated (“off”), according to data culled from the historical record and daily newspaper accounts. Specifically, these nonpharmaceutical interventions were legally enforced and affected large segments of the city’s population. Isolation of ill persons and quarantine of those suspected of having contact with ill persons refers only to mandatory orders as opposed to voluntary quarantines being discussed in our present era. School closure was considered activated when the city officials closed public schools (grade school through high school); in most, but not all cases, private and parochial schools followed suit. Public gathering bans typically meant the closure of saloons, public entertainment venues, sporting events, and indoor gatherings were banned or moved outdoors; outdoor gatherings were not always canceled during this period (eg, Liberty bond parades); there were no recorded bans on shopping in grocery and drug stores. Based on an estimated 10-day time frame between disease onset and death, (Reprinted) JAMA, August 8, 2007—Vol 298, No. 6 645 INTERVENTIONS DURING 1918-1919 INFLUENZA PANDEMIC we estimated that the association of nonpharmaceutical interventions with reductions in EDR occurred 10 days after their actual date of implementation.3,24-27 To test the association of the layering and timing of nonpharmaceutical interventions with mortality, an analysis of variance (ANOVA) model was constructed with weekly EDR as the dependent variable and epidemiological week, city, and the status (on/off) of every combination of nonpharmaceutical interventions as the independent variables. In the ANOVA model, each possible combination of nonpharmaceutical interventions was treated as an independent variable to test for layering effects. Any factor with a P value of less than .10 was included in the model. Because there is ambiguity over the rigor with which the category of ancillary nonpharmaceutical interventions was applied, enforced, and deactivated, we focused primarily on the 3 major categories of nonpharmaceutical interventions discussed above and we included the ancillary nonpharmaceutical interventions in the multivariate model for purposes of completeness. We defined additional outcome (dependent) variables: (1) the time to first peak as the time in days from the activation of the first major category of nonpharmaceutical interventions to the date of the first peak EDR; (2) the magnitude of the first peak as the first peak weekly EDR; and (3) the mortality burden as the cumulative EDR during the entire 24-week study period. We also defined the following independent variables. The first was the public health response time (PHRT) as the time in days (either positive or negative) between the date when weekly EDR first exceeds twice the baseline pneumonia and influenza death rate (2⫻baseline; ie, when the mortality rate begins to accelerate) and the activation of the first major nonpharmaceutical interventions. Interventions that occurred prior to this reference point are recorded as negative PHRT values, indicating that public health officials responded to events prior to the acceleration of weekly death rates. How646 ever, most cities had positive PHRT in that they reacted after the 2⫻ baseline mortality threshold, indicating that the epidemic had already entered its acceleration phase. The second independent variable was total days of nonpharmaceutical interventions, which was defined as the total cumulative number of days that nonpharmaceutical interventions from the 3 major categories were activated during the entire 24-week study period. The ANOVA models were based on the study design of a 43 (city) ⫻ 24 (week) factorial design without replication. Because there is no replication, the city⫻week interaction term was treated as the error term in the multivariate analysis. We considered 4 different nonpharmaceutical interventions. Hence, there are 15 different combinations of these interventions (excluding the no intervention combination). Each of these 15 combinations was either implemented (on) or not implemented (off) in each city for each week. Thus, the effects of each of these combinations of nonpharmaceutical interventions are included in the city⫻week interaction term. Each of these terms (along with their ⫻city and ⫻week interaction terms) were extracted from the original city⫻week interaction term. The remaining unexplained variation was used as the error term in the ANOVA model. The remaining error term is likely to be larger than a true error term generated through replication so the analysis of any effects using this error term can be expected to be conservative. Such a factorial model without replication can be used to test hypotheses but the lack of natural error in the model makes estimates or predictions from the model such as effect size measures and confidence intervals nonestimable. We also generated scatterplots and Spearman rank correlation coefficients to explore the associations between PHRT and each of the 2 additional dependent variables and associations between total days of nonpharmaceutical interventions and mortality burden. We further investigated these associations by using box plots and Wilcoxon rank sum tests to compare the outcomes for the cit- JAMA, August 8, 2007—Vol 298, No. 6 (Reprinted) Downloaded From: https://jamanetwork.com/ on 04/13/2020 ies above and below the median of each independent variable. We also generated scatterplots and Spearman rank correlation coefficients to explore other potential or confounding associations (as independent determinants): (1) EDR in the 4 successive waves of the pandemic; (2) city-specific population size vs EDR; (3) cityspecific population density vs EDR; (4) city-specific population age distribution vs EDR; and (5) city-specific sex distribution vs EDR. Analyses were performed using SAS statistical software version 9.1 (SAS Institute Inc, Cary, NC). RESULTS There were 115 340 excess pneumonia and influenza deaths (EDR, 500/ 100 000 population) in the 43 cities during the 24 weeks analyzed. TABLE 1 shows considerable city-to-city variation in mortality profiles and intervention characteristics; lists the earliest reported dates of the first pneumonia and influenza cases by city, mortality acceleration (2⫻baseline EDR), first implementation of nonpharmaceutical interventions, and first peak EDR; and lists the values for each of the independent and outcome variables described above. TABLE 2 shows the categories of nonpharmaceutical intervention combinations, the number of cities implementing those combinations, and the median and range of duration of implementation by each of the 43 cities during the study period. Every city adopted at least 1 of the 3 major categories of nonpharmaceutical interventions; 15 applied all 3 categories of nonpharmaceutical interventions concurrently. School closure concurrently combined with public gathering bans represented the most common combination, implemented in 34 cities (79%) for a median duration of 4 weeks (range, 1-10 weeks). School closure was ultimately used in some combination with the other categories of nonpharmaceutical interventions by 40 cities (93%). Three cities never officially closed their schools (New York City, New York, New Haven, Connecticut, and Chicago, Illinois, although the latter reported a student absenteeism ©2007 American Medical Association. All rights reserved. INTERVENTIONS DURING 1918-1919 INFLUENZA PANDEMIC rate of ⱖ45% at the peak of its epidemic); 25 cities closed their schools once, 14 closed them twice, and 1 (Kan- sas City, Missouri) closed its schools 3 times. Schools were officially closed a median of 6 weeks (range, 0-15 weeks). The ANOVA multivariate model had an r2 of 86.7% (P⬍.001). Nonpharmaceutical interventions were a significant Table 1. Characteristics of Influenza Pandemic for 43 US Cities Between September 8, 1918, and February 22, 1919 City Albany, NY Baltimore, MD Birmingham, AL Boston, MA Buffalo, NY Cambridge, MA Chicago, IL Cincinnati, OH Cleveland, OH Columbus, OH Dayton, OH Denver, CO Fall River, MA Grand Rapids, MI Indianapolis, IN Kansas City, MO Los Angeles, CA Louisville, KY Lowell, MA Milwaukee, WI Minneapolis, MN Nashville, TN New Haven, CT New Orleans, LA New York City, NY Newark, NJ Oakland, CA Omaha, NE Philadelphia, PA Pittsburgh, PA Portland, OR Providence, RI Richmond, VA Rochester, NY St Louis, MO St Paul, MN San Francisco, CA Seattle, WA Spokane, WA Syracuse, NY Toledo, OH Washington, DC Worcester, MA Public Date of First First Health Total No. of Days of Mortality Case Acceleration Nonpharmaceutical Response Nonpharmaceutical Intervention Interventions Date Time, d b Date a 9/27 10/6 10/9 3 47 9/18 9/29 10/9 10 43 9/24 9/30 10/9 9 48 9/4 9/12 9/25 13 50 9/24 9/28 10/10 12 49 9/4 9/11 9/25 14 49 9/17 9/28 9/26 −2 68 9/24 10/4 10/6 2 123 9/20 10/7 10/5 −2 99 9/20 10/6 10/11 5 147 9/20 10/5 9/30 −5 156 9/17 9/27 10/6 9 151 9/9 9/16 9/26 10 60 9/23 10/2 10/19 17 62 9/22 9/30 10/7 7 82 9/20 9/26 9/26 0 170 9/27 10/6 10/11 5 154 9/13 10/1 10/7 6 145 9/9 9/16 9/27 11 59 9/14 10/6 10/11 5 132 9/21 10/6 10/12 6 116 9/21 10/6 10/7 1 55 9/14 9/23 10/15 22 39 9/10 10/1 10/8 7 78 9/5 9/29 9/18 −11 73 9/6 9/30 10/10 10 33 10/1 10/8 10/12 4 127 9/18 10/4 10/5 1 140 8/27 9/25 10/3 8 51 9/4 9/27 10/4 7 53 10/2 10/7 10/11 4 162 9/8 9/17 10/6 19 42 9/21 9/29 10/6 7 60 9/22 10/6 10/9 3 54 9/23 10/7 10/8 1 143 9/21 10/2 11/6 35 28 9/24 10/7 10/18 11 67 9/24 10/1 10/6 5 168 9/28 10/9 10/10 1 164 9/12 9/18 10/7 19 39 9/21 10/13 10/15 2 102 9/11 9/23 10/3 10 64 9/9 9/12 9/27 15 44 Date of Peak Excess Death Rate 10/24 10/18 10/22 10/3 10/22 10/3 10/21 10/24 10/31 10/24 10/20 10/20 10/12 10/25 10/18 10/27 10/30 10/20 10/10 10/23 10/24 10/16 10/24 10/20 10/23 10/22 10/30 10/18 10/16 11/5 11/2 10/17 10/16 10/26 10/29 11/12 10/29 10/23 11/5 10/14 10/25 10/15 10/7 Excess Pneumonia and Magnitude of Influenza First Peak, Mortality, Excess Deaths/ Deaths/ Time to 100 000 100 000 c Peak, d Population Population d 15 161.8 553.2 9 182.1 559.3 13 70.9 591.8 8 159.9 710.0 12 140.9 529.5 8 125.5 541.0 25 84.8 373.2 18 67.6 451.2 26 83.6 474.0 13 47.3 311.7 20 87.8 410.0 14 55.0 630.9 16 165.2 621.3 6 15.0 210.5 11 38.8 290.0 31 58.1 579.8 19 64.2 493.8 13 74.8 406.4 13 123.1 522.9 12 36.4 291.5 18 37.6 267.1 9 160.1 610.4 9 109.5 586.5 12 172.9 734.0 35 90.1 452.3 12 101.5 533.0 18 107.0 506.2 13 81.7 554.0 13 249.7 748.4 32 130.7 806.8 22 59.4 505.2 11 105.2 574.2 10 112.2 508.3 17 70.2 359.1 21 30.0 358.0 6 55.6 413.2 11 143.0 672.7 17 49.5 414.1 26 66.0 481.8 7 145.4 541.4 10 54.8 294.5 12 140.1 607.6 10 126.1 654.7 a Defined as 2 ⫻ baseline death rate. b Defined as days between 2 ⫻ baseline death rate and first nonpharmaceutical intervention. c Weekly excess death rate. d Total excess death rate through 24 weeks. ©2007 American Medical Association. All rights reserved. Downloaded From: https://jamanetwork.com/ on 04/13/2020 (Reprinted) JAMA, August 8, 2007—Vol 298, No. 6 647 INTERVENTIONS DURING 1918-1919 INFLUENZA PANDEMIC Table 2. Nonpharmaceutical Interventions Implemented in 43 US Cities Between September 8, 1918, and February 22, 1919 Type of Nonpharmaceutical Intervention No. (%) of Cities Implementing Nonpharmaceutical Intervention for ⱖ1 wk (N = 43) a Isolation or quarantine only School closure only Public gathering ban only Isolation and quarantine and school closure Isolation and quarantine and public gathering ban School closure and public gathering ban Isolation and quarantine, school closure, and public gathering ban 15 (35) 22 (51) Median (Range) Duration of Nonpharmaceutical Intervention, wk 1 (1-10) 1 (1-7) 6 (14) 2 (5) 4 (9) 1.5 (1-5) 5.5 (4-7) 4 (2-5) 34 (79) 15 (35) 4 (1-10) 4 (2-6) a Cities often implemented more than 1 nonpharmaceutical intervention combination during the outbreak period, so the total adds to more than 100%. The number of categories of nonpharmaceutical interventions implemented during some part of the outbreak was 1 in 1 city, 2 in 23 cities, and 3 in 19 cities. The total number of weeks that at least 1 nonpharmaceutical intervention was implemented was 4 in 6 cities, 5 in 6 cities, 6 in 8 cities, 7 in 3 cities, 8 in 6 cities, 10 in 5 cities, 11 in 4 cities, 13 in 1 city, 14 in 2 cities, 15 in 1 city, and 16 in 1 city. No cities had at least 1 nonpharmaceutical intervention implemented for durations of 9 and 12 weeks. source of the variation in the weekly EDRs within and between the cities. The ANOVA results are shown in TABLE 3. The multivariate model demonstrates that layered nonpharmaceutical interventions generally had a more significant association with weekly EDR than individual nonpharmaceutical interventions. Specifically, combinations of nonpharmaceutical interventions including school closure and public gathering bans appeared to have the most significant association with weekly EDR (ie, the lowest P values, most were P⬍.001). The large number of significant nonpharmaceutical interventions⫻week interactions in the model confirms that the timing of the implementation of a given combination of nonpharmaceutical interventions was a significant factor in reducing mortality. One caveat is persistent nonpharmaceutical interventions⫻city interactions, meaning that the success of a strategy of nonpharmaceutical interventions in a particular city does not uniformly translate to all other cities. The 2 outlier cities in our study, Grand Rapids, Michigan, and St Paul, Minnesota, exemplify this point. The scatterplots in F IGURE 1A, Figure 1B, and Figure 1C display the associations between the PHRT and each of the 3 dependent variables. Figure 1A displays the association between PHRT in days and time to first peak EDR; cities that implemented non648 pharmaceutical interventions earlier had greater delays in reaching peak mortality (Spearman r = −0.74, P ⬍ .001). Figure 1B shows the association between PHRT and the magnitude of the first peak EDR; cities that implemented nonpharmaceutical interventions earlier had lower peak mortality rates (Spearman r=0.31, P=.02). Figure 1C depicts the association between PHRT and total mortality burden; cities that implemented nonpharmaceutical interventions earlier experienced a lower total mortality (Spearman r=0.37, P=.008). In summary, when comparing the 21 cities with earlier (less than the median) PHRT with the 21 cities with the later (greater than the median) PHRT, there are statistically significant differences for each of the outcome measures (P ⱕ.001; TABLE 4). Figures 1C and 1D show the associationbetweenearly,sustained,andlayered application of nonpharmaceutical interventions and total excess pneumonia and influenza mortality burden in 43 cities. Figure 1C shows the statistically significant association between PHRT and total mortality burden. Figure 1D shows a statistically significant association between increased duration of nonpharmaceutical interventions and a reduced total mortality burden (Spearman r=−0.39, P=.005). In summary, the 21 cities that hadearlierPHRT(ie,lessthanthemedian) JAMA, August 8, 2007—Vol 298, No. 6 (Reprinted) Downloaded From: https://jamanetwork.com/ on 04/13/2020 and the most sustained and most days of nonpharmaceutical interventions had a statisticallysignificantreductioninexcess pneumonia and influenza mortality rates compared with the 21 cities that had later PHRT and fewer days of nonpharmaceutical interventions (Table 4). FIGURE 2 shows the aggregate city mortality curves by region (East, Midwest and South, and West). FIGURE 3 displays 4 city-specific mortality curves, including weekly EDR and the nonpharmaceutical interventions implemented as well as their activation and deactivation dates for St Louis, Missouri, New York City, Denver, Colorado, and Pittsburgh, Pennsylvania. These 4 cities were chosen because they indicate the broad spectrum of outcomes seen in the 43 cities studied as well as for their geographical and social diversity. (The mortality curves for all 43 cities are available at http://www .cdc.gov/ncidod/dq/index.htm.) 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. In exploring alternative and potentially confounding explanations for variation in city-specific EDR, we used a scatterplot to compare the cumulative EDR of the 43 cities during pandemic waves 1 (February-May 1918), 2 (SeptemberDecember 1918), 3 (January-April 1919), and 4 (January-April 1920).2,3 We found no statistically significant association of the EDR across the 43 cities when comparing successive waves. Specifically, the severity or occurrence of wave 1 is not associated, either positively or negatively, with the severity of wave 2; the severity of wave 2 is not associated with the severity of wave 3; and the severity of wave 3 is not associated with the severity of wave 4 (figures appear in the online supplement at http://www.cdc.gov /ncidod/dq/index.htm).28,29 Publishedvirologicalevidenceforstrain variation during wave 2 is limited to a singlegenotypicvariantwithoutevidence for significant phenotypic change in virulence.30-33 Whileplausible,novirologi- ©2007 American Medical Association. All rights reserved. INTERVENTIONS DURING 1918-1919 INFLUENZA PANDEMIC cal evidence yet exists to explain the perplexing mortality difference between the spring 1918 wave, which was reportedly milder, and the subsequent fall and winter waves. Additional studies may clarify the understanding of the 1918 pandemic’s wave phenomena. Similarly, scatterplots comparing the cumulativeEDRtothecity-specificpopulation size and density; sex distribution; and proportion of ages of younger than 1 month to 5 years, 15 to 40 years, and older than 65 years, which corresponded to high reported specific mortality rates in 1918 demonstrated no association. Among the 43 cities we investigated, neitherthecity’spopulationsize,density,sex distribution, nor age distribution accounted for the differences in mortality (figures appear in supplement at http: //www.cdc.gov/ncidod/dq/index.htm). Table 3. Multivariate Model Showing Effect of Combinations of Nonpharmaceutical Interventions on Weekly Excess Death Rates for 43 US Cities Between September 8, 1918, and February 22, 1919 a Source of Variation Type of confounders Week City 1 Nonpharmaceutical intervention School closure ⫻ Week Banning public gatherings Isolation and quarantine ⫻ City Ancillary nonpharmaceutical interventions ⫻ Week ⫻ City 2 Nonpharmaceutical interventions School closure and banning public gatherings ⫻ Week ⫻ City School closure and isolation and quarantine ⫻ Week Banning public gatherings and isolation and quarantine ⫻ Week Banning public gatherings and ancillary nonpharmaceutical interventions ⫻ Week ⫻ City Isolation and quarantine and ancillary nonpharmaceutical interventions ⫻ Week ⫻ City 3 Nonpharmaceutical interventions School closure, banning public gatherings, and isolation and quarantine ⫻ Week School closure, banning public gatherings, and ancillary nonpharmaceutical interventions ⫻ Week ⫻ City School closure, isolation and quarantine, and ancillary nonpharmaceutical interventions ⫻ Week Banning public gatherings, isolation and quarantine, and ancillary nonpharmaceutical interventions ⫻ Week 4 Nonpharmaceutical interventions School closure, banning public gatherings, isolation and quarantine, and ancillary nonpharmaceutical interventions ⫻ Week ⫻ City Error df Sum of Squares Mean Square F Score P Value 29 42 75 677.0 65 557.9 2609.6 1560.9 16.24 9.72 ⬍.001 ⬍.001 1 8 1 1 10 1 13 12 1288.7 4551.8 1342.0 911.1 3976.5 897.3 6122.4 10 257.6 1288.7 569.0 1342.0 911.1 397.7 897.3 471.0 854.8 8.02 3.54 8.35 5.67 2.48 5.59 2.93 5.32 .005 ⬍.001 .004 .02 .006 .02 ⬍.001 ⬍.001 1 9 13 1 4 1 1 1 4 4 1 2 2 681.3 6497.0 6229.9 2335.3 2434.2 292.3 563.9 272.6 7444.6 5547.6 48.1 1507.6 824.7 681.3 721.9 479.2 2335.3 608.6 292.3 563.9 272.6 1861.1 1386.9 48.1 753.8 412.4 4.24 4.49 2.98 14.54 3.79 1.82 3.51 1.70 11.59 8.63 0.30 4.69 2.57 .04 ⬍.001 ⬍.001 ⬍.001 .005 .18 .06 .19 ⬍.001 ⬍.001 .58 .009 .08 1 2 1 762.4 2239.3 691.6 762.4 1119.7 691.6 4.75 6.97 4.41 .03 .001 .04 10 26 1 12 260.5 51 423.8 3451.1 1226.0 1977.8 3451.1 7.63 12.31 21.48 ⬍.001 ⬍.001 ⬍.001 4 2493.5 623.4 3.88 .004 1 51.9 51.9 0.32 .57 8 4535.2 566.9 3.53 ⬍.001 1 503.7 503.7 3.14 .08 9 13 770 6068.3 23 509.7 123 691.2 674.3 1808.4 160.6 4.20 11.26 ⬍.001 ⬍.001 a r2 = 86.7% ©2007 American Medical Association. All rights reserved. Downloaded From: https://jamanetwork.com/ on 04/13/2020 (Reprinted) JAMA, August 8, 2007—Vol 298, No. 6 649 INTERVENTIONS DURING 1918-1919 INFLUENZA PANDEMIC COMMENT During the 1918-1919 influenza pandemic, all 43 cities eventually implementednonpharmaceuticalinterventions but the time of activation, duration, and choice or combination of these nonpharmaceutical interventions appear to have beenkeyfactorsintheirsuccessorfailure. In 1918, decisions to activate nonpharmaceutical interventions were typically triggered by excess morbidity, mortality, or both, as well as situational awareness ofothercommunitiesnearandfar.Because discerning precisely the first arrival of pandemic virus in a community was difficult, we chose to measure public health response time in reference to excess pneumonia and influenza mortality (ie, whenweeklyEDRfirstcrossedthethreshold of 2⫻the baseline and the mortality rates entered an acceleration phase). Figure 1. Scatterplot of Public Health Response Time for 43 US Cities From September 8, 1918, Through February 22, 1919 A Time to first mortality peak by public health response time r = –0.74 P<.001 Pittsburgh, PA 30 Excess Deaths/100 000 Population New York City, NY 35 Time to First Peak Excess Death Rate, d B Magnitude of first mortality peak by public health response time 25 St Louis, MO 20 Denver, CO 15 10 5 St Paul, MN Grand Rapids, MI 0 –15 –10 –5 0 5 10 15 20 25 30 35 250 r = 0.31 P = .02 200 150 Pittsburgh, PA New York City, NY 50 0 –15 Grand Rapids, MI –10 –5 Public Health Response Time, d Excess Deaths/100 000 Population Excess Deaths/100 000 Population 700 Denver, CO 600 500 New York City, NY St Paul, MN 400 St Louis, MO 300 200 –15 Grand Rapids, MI –10 –5 0 5 10 15 20 5 10 15 20 25 30 35 D Total excess pneumonia and influenza mortality by total No. of days of nonpharmaceutical interventions r = 0.37 P = .008 Pittsburgh, PA 0 Public Health Response Time, d C Total excess pneumonia and influenza mortality by public health response time 800 St Paul, MN Denver, CO St Louis, MO 25 30 35 Public Health Response Time, d Pittsburgh, PA r = –0.39 P = .005 800 700 Denver, CO 600 500 New York City, NY St Paul, MN 400 St Louis, MO 300 Grand Rapids, MI 200 20 40 60 80 100 120 140 160 Total No. of Days of Nonpharmaceutical Interventions The 4 cities represented by black circles are discussed further in the text. The 2 cities represented by blue circles are outliers chosen to demonstrate that the associations shown are not perfect. The Spearman rank correlation coefficient was used. Table 4. Implementation Strategy of Nonpharmaceutical Interventions for 21 Cities Between September 8, 1918, and February 22, 1919 Public Health Response Time, d Late (ⱖ7 d) Early (⬍7 d) Outcome Variable Time to peak, d Magnitude of first peak (weekly EDR) Excess pneumonia and influenza mortality rate (total EDR) 25th Percentile 13 54.7 359.1 50th Percentile 18 67.6 451.2 75th Percentile 22 84.8 505.2 25th Percentile 9 101.5 529.5 50th Percentile 11 125.8 580.3 75th Percentile 13 145.4 654.7 P Value ⬍.001 .001 ⬍.001 610.4 ⬍.001 Total Days of Nonpharmaceutical Interventions Most (ⱖ65 d) Excess pneumonia and influenza mortality rate (total EDR) 358.0 451.2 Least (⬍65 d) 505.2 529.5 559.3 Abbreviation: EDR, excess death rate. 650 JAMA, August 8, 2007—Vol 298, No. 6 (Reprinted) Downloaded From: https://jamanetwork.com/ on 04/13/2020 ©2007 American Medical Association. All rights reserved. INTERVENTIONS DURING 1918-1919 INFLUENZA PANDEMIC Figure 2. Aggregate Weekly Excess Death Rates for 43 US Cities by Region From September 8, 1918, Through February 22, 1919 120 West East 80 60 40 20 0 Sep 15 22 29 Oct 13 20 27 Nov 10 17 24 Dec 8 8, 6 3 1 1918 15 22 29 Jan 12 19 26 Feb 9 5, 2 1919 16 23 The total excess death rate is 555 for the East region; 413 for the Midwest and South region; and 529 for the West region. with an enforced staggered business hour ordinance from October 5 through November 3, 1918.34 During this era, New York City’s health department was renowned internationally for its innovative policies of mandatory case reporting and rigidly enforced isolation and quarantine procedures.35 Typically, individuals diagnosedwithinfluenzawereisolatedinhospitals or makeshift facilities, while those suspected to have contact with an ill person (but who were not yet ill themselves) were quarantined in their homes with an official placard declaring that location to be under quarantine. New York City mounted an early and sustained response to the epidemic and experienced the lowest death rate on the Eastern seaboard but it did not layer its response. New York City’s cumulative mortality burden, 452/ 100 000, ranked 15 out of the 43 cities studied. In contrast, Pittsburgh, under orders from the Pennsylvania health department, executed a public gathering ban on October 4, 1918, but city officials delayed until October 24 before implementing school closure. A week later, on November 2, the state rescinded public gathering bans. The city applied its non- ©2007 American Medical Association. All rights reserved. Downloaded From: https://jamanetwork.com/ on 04/13/2020 Midwest and South 100 Weekly Excess Death Rate, No. of Deaths per 100 000 Population Hence, the difference in time between the firstnonpharmaceuticalinterventionsand this excess mortality threshold may be a positive or negative value. For example, in Philadelphia, Pennsylvania, which was affected early and was unprepared to respond, the PHRT was 8 and the EDR was approximately 37/100 000 population at the point of implementing nonpharmaceutical interventions; in contrast, New York City’s PHRT was −11 days and the EDR was 0/100 000 population at the point of implementing nonpharmaceutical interventions. New York City responded to its first influenza cases and the perceived severity of the epidemic in nearby cities without waiting for excess deaths to accumulate. TheUSCentersforDiseaseControland Prevention’s newly released interim communitymitigationguidancerecommends activating nonpharmaceutical interventions when outbreaks due to a pandemic virus strain first are confirmed in a state or metropolitan service region.16 Several theoretical models suggest that the effect oftargeted,layeredstrategiesfornonpharmaceutical interventions may be optimized when community influenza attack rates are 1% or lower.11-15 Given the exponential growth of an unmitigated influenza pandemic, it is reasonable to expect that the timing of interventions will be among the most critical factors. Such expectations and biological realities are consistent with our observations of the 1918pandemic,whenrapidpublichealth response time was a critical factor in the successful application of nonpharmaceutical interventions. Late interventions, regardless of their duration or permutation of use, almost always were associated with worse outcomes. However, timing alone was not consistently associated with success. The combination and choice of nonpharmaceutical interventions also appeared to be critical as confirmed by the multivariate model. For example, New York City reacted earliest to the gathering influenza crisis, primarily with the sustained (⬎10 weeks beginning September 19, 1918) and rigidly enforced application of compulsory isolationandquarantineprocedures,along pharmaceutical interventions late and individually rather than combined. Pittsburgh’s cumulative excess mortality burden (EDR=807/100 000) ranked 43 out of 43 cities during the study period. However,thebenefitsoftheseinterventions were not equally distributed. Those cities acting in a timely and comprehensivemannerappeartohavebenefitedmost in terms of reductions in total EDR. For example, St Louis, which implemented a relatively early, layered strategy (school closure and cancellation of public gatherings), and sustained these nonpharmaceutical interventions for about 10 weeks each, did not experience nearly as deleterious an outbreak as 36 other communities in the study (cumulative EDR=358/ 100 000 population). The 1918 experience suggests that sustained nonpharmaceutical interventions are beneficial and need to be “on” throughout the particular peak of a local experience. Many of the 43 cities in the study, particularly in the Midwest and South and West, experienced 2 peaks of excess pneumonia and influenza mortality (eg, Birmingham, Alabama, Cincinnati, Ohio, Columbus, Ohio, Denver, Indianapolis, Indiana, Kansas City, (Reprinted) JAMA, August 8, 2007—Vol 298, No. 6 651 INTERVENTIONS DURING 1918-1919 INFLUENZA PANDEMIC Louisville, Kentucky, Milwaukee, Wisconsin, Minneapolis, Minnesota, Oakland, California, Omaha, Nebraska, Port- land, Oregon, Rochester, New York, St Louis, San Francisco, California, Seattle, Washington, Spokane, Washing- ton, Toledo, Ohio; see figures in online supplement at http://www.cdc.gov /ncidod/dq/index.htm). These second Figure 3. Weekly Excess Death Rates From September 8, 1918, Through February 22, 1919 A St Louis, MO Total excess death rate: 358/100 000 population Public health response time: +1 d Total No. of days of nonpharmaceutical interventions: 143 140 130 B New York City, NY Total excess death rate: 452/100 000 population Public health response time: –11 d Total No. of days of nonpharmaceutical interventions: 73 Weekly Excess Death Rate, No. of Deaths per 100 000 Population 120 110 Weekly excess death rate 100 2 × baseline mortality First pneumonia and influenza case 90 80 70 60 50 40 30 20 10 0 Sep 8, 1918 Oct 6 Nov 3 Dec 1 Jan 5, 1919 Feb 2 23 School closure Public gathering ban Other a Sep 8, 1918 Oct 6 Nov 3 Dec 1 Jan 5, 1919 Feb 2 23 Isolation, quarantine Otherb C Denver, CO Total excess death rate: 631/100 000 population Public health response time: +9 d Total No. of days of nonpharmaceutical interventions: 151 140 130 D Pittsburgh, PA Total excess death rate: 807/100 000 population Public health response time: +7 d Total No. of days of nonpharmaceutical interventions: 53 Weekly Excess Death Rate, No. of Deaths per 100 000 Population 120 110 100 90 80 70 60 50 40 30 20 10 0 Sep 8, 1918 Oct 6 Nov 3 Dec 1 Jan 5, 1919 School closure Public gathering ban Isolation, quarantine Other c Feb 2 23 Sep 8, 1918 Oct 6 Nov 3 Dec 1 Jan 5, 1919 Feb 2 23 School closure Public gathering ban Other d Type and duration of nonpharmaceutical interventions are indicated under each plot. For the specific nonpharmaceutical interventions, black bars indicate activation. a Business hours restricted, streetcars’ capacity limited. b Staggered business hours, signs with “cover coughs.” c Staggered business hours, warning signs posted in theaters. d Schoolchildren given information to take home, warned not to gather in groups. 652 JAMA, August 8, 2007—Vol 298, No. 6 (Reprinted) Downloaded From: https://jamanetwork.com/ on 04/13/2020 ©2007 American Medical Association. All rights reserved. INTERVENTIONS DURING 1918-1919 INFLUENZA PANDEMIC peaks frequently followed the sequential activation, deactivation, and reactivation of nonpharmaceutical interventions, highlighting the transient protective nature of nonpharmaceutical interventions and the need for a sustained response. For example, Denver (cumulative EDR=631/100 000 population) responded twice with an extensive menu of nonpharmaceutical interventions that included public gathering bans, school closure, isolation and quarantine, and several ancillary nonpharmaceutical interventions and these actions are reflected temporally in its 2-peak mortality curve. Such dual-peaked cities are of particular interest because of the specificity and temporal associations between excess mortality and the triggers of activation and deactivation of nonpharmaceutical interventions and the implications for a causal relationship. Among the 43 cities, we found no example of a city that had a second peak of influenza while the first set of nonpharmaceutical interventions were still in effect, suggesting that each city with a bimodal pattern served as its own control. In dual-peaked cities, activation of nonpharmaceutical interventions was followed by a diminution of deaths and, typically, when nonpharmaceutical interventions were deactivated, death rates increased. History is not a predictive science. There exist numerous well-documented and vast differences between US society and public health during the 1918 pandemic compared with the present. We acknowledge the inherent difficulties of interpreting data recorded nearly 90 years ago and contending with the gaps, omissions, and errors that may be included in the extant historical record. The associations observed are not perfect; for example, 2 outlier cities (Grand Rapids and StPaul)experiencedbetteroutcomeswith less than perfect public health responses. Future work by our research team will explore social, political, and ecological determinants, which may further help to explain some of this variation. The United States of 1918 had many similar features to the present era: rapid transportation in the form of trains and automobiles; rapid means of communication in the form of the telegraph and telephone; large, heterogeneous populations with substantial urban concentrations (although a much higher percentage of the US population lived in rural areas compared with the present); a news system that was able to circulate information widely during the epidemic, including many daily newspapers and broadsheets distributed in communities; and a wide spectrum of public health agencies at various levels of government. When examining the 1918 pandemic, however, it is important to recognize the numerous social, cultural, and scientific differences that do exist between that period and the present. For example, the legal understanding of privacy, civil, and constitutional rights as they relate to public health and governmentally directed measures (such as mass vaccination programs) has changed markedly over the past 9 decades. In addition, public support of and trust in these measures, along with trust in the medical profession as a whole, has shifted over time. Finally, other features of the modern era that need to be considered when applying lessons from history to the present era include the increased speed and mode of travel, above all highvolume commercial aviation; instantaneous access to information via the Internet and personal computers; a baseline understanding among the general population that the etiologic agents of infectious diseases are microbial; and advances in medical technology and therapeutics that have expanded considerably the options available for dealing with a pandemic. Historical contextual issues and statistical limitations aside, the US urban experience with nonpharmaceutical interventions during the 1918-1919 pandemic constitutes one of the largest data sets of its kind ever assembled in the modern, postgerm theory era. Our findings conform to 8 of A. Bradford Hill’s 9 tenets on causal associations in the consideration of disease and the environment.36 Specifically, during the 1918-1919 pandemic, the rela- ©2007 American Medical Association. All rights reserved. Downloaded From: https://jamanetwork.com/ on 04/13/2020 tion of early, sustained, and layered nonpharmaceutical interventions to EDR in 43 US cities demonstrate satisfaction of the criteria of strength (the magnitude and statistical significance of our findings, which also argue against an association by chance alone), consistency (early and combined nonpharmaceutical interventions were consistently associated with reductions in mortality, and our analysis is consistent with 2 recent smaller, preliminary historical epidemiological reports, although these studies look at only 16 US urban centers, do not include actual activation and deactivation time points, duration, or layering of nonpharmaceutical interventions, and rely extensively on secondary historical sources.37,38 Further, our retrospective study is consistent with the results from recent theoretical models of the spread of a contemporary pandemic, which highlight the value of early, combined, and sustained nonpharmaceutical interventions to mitigate a pandemic11-15), specificity (best demonstrated in cities with bimodal mortality peaks when the triggers were activated, deactivated, and reactivated), temporality (interventions always preceded the reduction of EDR), dose response (layering and increased duration of the nonpharmaceutical interventions were associated with better outcomes), biological plausibility (these interventions reduce person-to-person interactions and biologically would be expected to reduce the spread of a communicable agent such as influenza), coherence (our data align with the established body of knowledge on the epidemiology of influenza), and analogy (isolation and social distancing have been demonstrated as effective means of preventing person-to-person spread of other respiratory tract diseases, such as rhinovirus, severe acute respiratory syndrome, respiratory syncytial virus, varicella, and seasonal influenza). The ninth tenet, experiment, could not be demonstrated directly because of the paucity of influenza pandemics in the past century, the trend away from such traditional public health mea- (Reprinted) JAMA, August 8, 2007—Vol 298, No. 6 653 INTERVENTIONS DURING 1918-1919 INFLUENZA PANDEMIC sures for disease control during the past 50 years, and ethical limitations of using population-wide nonpharmaceutical interventions in the absence of a serious threat. These findings contrast with the conventional wisdom that the 1918 pandemic rapidly spread through each community killing everyone in its path. Although these urban communities had neither effective vaccines nor antivirals, cities that were able to organize and execute a suite of classic public health interventions before the pandemic swept fully through the city appeared to have an associated mitigated epidemic experience. Our study suggests that nonpharmaceutical interventions can play a critical role in mitigating the consequences of future severe influenza pandemics (category 4 and 5) and should be considered for inclusion in contemporary planning efforts as companion measures to developing effective vaccines and medications for prophylaxis and treatment. The history of US epidemics also cautions that the public’s acceptance of these health measures is enhanced when guided by ethical and humane principles.39-41 Author Contributions: Drs Markel and Cetron had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Markel, Lipman, Navarro, Stern, Cetron. Acquisition of data: Markel, Navarro, Sloan, Michalsen, Stern. Analysis and interpretation of data: Markel, Lipman, Navarro, Sloan, Michalsen, Stern, Cetron. Drafting of the manuscript: Markel, Lipman, Navarro, Sloan, Michalsen, Stern, Cetron. Critical revision of the manuscript for important intellectual content: Markel, Lipman, Navarro, Stern, Cetron. Statistical analysis: Markel, Lipman, Cetron. Obtained funding: Markel, Cetron. Administrative, technical, or material support: Markel, Navarro, Sloan, Michalsen, Cetron. Study supervision: Markel, Cetron. Financial Disclosures: None reported. Funding/Support: This study was funded by a contract from the US Centers for Disease Control and Prevention (Sol No. 2006-N-08562, Non-Pharmaceutical Interventions Study/contract 200-2006-16894). Role of the Sponsor: The US Centers for Disease Control and Prevention provided funding as part of pandemic preparedness research. Drs Cetron and Lipman from the Division of Global Migration and Quarantine at the Centers for Disease Control and Prevention participated as full scientific collaborators in the investigation. Acknowledgment: Matthew Cartter, MD, MPH, Cleto DiGiovanni Jr, Jeffrey Duchin, MD, Bruce Gel654 lin, MD, Richard Goodman, MD, JD, Daniel Jernigan, MD, MPH, Lisa Koonin, MN, MPH, Anthony Marfin, MD, Martin Meltzer, PhD, William Thompson, PhD, David Shay, MD, and Mary Wilson, MD, provided constructive review of this manuscript. No one metioned in this section was compensated for contributing. REFERENCES 1. Murray CJL, Lopez AD, Chin B, Feehan D, Hill KH. Estimation of potential global pandemic influenza mortality on the basis of vital registry data from the 1918-20 pandemic: a quantitative analysis. Lancet. 2006; 368(9554):2211-2218. 2. Crosby A. America’s Forgotten Pandemic: The Influenza of 1918. New York, NY: Cambridge University Press; 2003. 3. Jordan EO. Epidemic Influenza: A Survey. Chicago, IL: American Medical Association; 1927. 4. Byerly CR. Fever of War: The Influenza Epidemic in the US Army During World War I. New York, NY: New York University; 2005. 5. Markel H. When Germs Travel: Six Major Epidemics That Invaded America Since 1900 and the Fears They Unleashed. New York, NY: Pantheon; 2004. 6. Rosen G. A History of Public Health. Baltimore, MD: Johns Hopkins University Press; 1993. 7. Bell DM; World Health Organization Writing Group. Nonpharmaceutical interventions for pandemic influenza, international measures. Emerg Infect Dis. 2006; 12(1):81-87. 8. Bell DM; World Health Organization Writing Group. Nonpharmaceutical interventions for pandemic influenza, national and community measures. Emerg Infect Dis. 2006;12(1):88-94. 9. Inglesby TV, Nuzzo JB, O’Toole T, Henderson DA. Disease mitigation measures in the control of pandemic influenza. Biosecur Bioterror. 2006; 4(4):366-375. 10. Aledort JE, Bozzette S, Lurie N, Ricci K, Adamson D, Stern S. Working Paper Non-Pharmacological Public Health Interventions for Pandemic Influenza: Proceedings of an Expert Panel Meeting—Final Draft. Washington, DC: RAND Health Center for Domestic and International Health Security; 2006. 11. Ferguson NM, Cummings DAT, Fraser C, Cajka JC, Cooley PC, Burke DS. Strategies for mitigating an influenza pandemic. Nature. 2006;442(7101):448-452. 12. Ferguson NM, Cummings DAT, Cauchemez S, et al. Strategies for containing an emerging influenza pandemic in Southeast Asia. Nature. 2005; 437(7056):209-214. 13. Longini IM Jr, Nizam A, Xu S, et al. Containing pandemic influenza at the source. Science. 2005; 309(5737):1083-1087. 14. Germann TC, Kadau C, Longini IM Jr, Macken CA. Mitigation strategies for pandemic influenza in the United States. Proc Natl Acad Sci U S A. 2006; 103(15):5935-5940. 15. Glass RJ, Glass LM, Beyeler WE, Min HJ. Targeted social distancing design for pandemic influenza. Emerg Infect Dis. 2006;12(11):1671-1681. 16. US Centers for Disease Control and Prevention. Interim Pre-Pandemic Planning Guidance: Community Strategy for Pandemic Influenza Mitigation in the United States: Early, Targeted, Layered Use of Nonpharmaceutical Interventions. Atlanta, GA: Centers for Disease Control and Prevention; 2007. 17. US Bureau of the Census. Weekly Health Index [October 6, 1917 to June 26, 1920]. Located at: New York Public Library, New York, NY. 18. Collins SD, Frost WH, Gover M, Sydenstricker E. Mortality from influenza and pneumonia in 50 large cities in the United States, 1910-1929. Public Health Rep. 1930;45:2277-2329. 19. Ministry of Health. Reports on Public Health and Medical Subjects No. 4: Report on the Pandemic of JAMA, August 8, 2007—Vol 298, No. 6 (Reprinted) Downloaded From: https://jamanetwork.com/ on 04/13/2020 Influenza, 1918-19. London, England: His Majesty’s Stationery Office; 1920. 20. McLaughlin AJ. The Shattuck lecture: epidemiology and etiology of influenza. Boston Med Surg J. 1920;183:1-23. 21. Sydenstricker E. Variations in case fatality during the influenza epidemic of 1918. Public Health Rep. 1921;36:2201-2211. 22. Sydenstricker E. Preliminary statistics of the influenza epidemic. Public Health Rep. 1918;33:23052321. 23. Frost WH. The epidemiology of influenza. Public Health Rep. 1919;34:1823-1837. 24. Eichel OR. A Special Report on the Mortality From Influenza in New York State During the Epidemic of 1918-19. New York: New York State Department of Health; 1923. 25. Klotz O. The pathology of epidemic influenza. In: Studies in Epidemic Influenza. Pittsburgh, PA: University of Pittsburgh School of Medicine; 1919:207. 26. Mills CE, Robins JM, Lipsitch M. Transmissibility of 1918 pandemic influenza. Nature. 2004;432 (7019):904-906. 27. Lyon MW. Gross pathology of epidemic influenza at Walter Reed General Hospital. JAMA. 1919; 72(13):924-929. 28. Vaughan WT. American Journal of Hygiene, Monograph Series. Influenza: an epidemiologic study. In: Baltimore, MD: American Journal of Hygiene; 1921. 29. Morens DM, Fauci AS. The 1918 influenza pandemic: insights for the 21st century. J Infect Dis. 2007; 195(7):1018-1028. 30. Reid AH, Janczewski TA, Lourens RM, et al. 1918 influenza pandemic caused by highly conserved viruses with two receptor-binding variants. Emerg Infect Dis. 2003;9(10):1249. 31. Taubenberger JK, Morens DM. 1918 influenza: the mother of all pandemics. Emerg Infect Dis. 2006; 12(1):15-22. 32. Taubenberger JK, Reid AH, Krafft AE, Bijwaard KE, Fanning TG. Initial genetic characterization of the 1918 Spanish influenza virus. Science. 1997;275 (5307):1793-1796. 33. Taubenberger JK, Reid AH, Lourens RM, Wang R, Jin G, Fanning TG. Characterization of the 1918 influenza virus polymerase genes. Nature. 2005; 437(7060):889-893. 34. Department of Health of the City of New York. Minutes of the Department of Health meetings from August 10 thru December 31, 1918. 1918;Book No. 31. Located at: Municipal Archives of New York, New York, NY. 35. Markel H. Quarantine! East European Jewish Immigrants and the New York City Epidemics of 1892. Baltimore, MD: Johns Hopkins University Press; 1999. 36. Hill AB. The environment and disease: association or causation? Proc R Soc Med. 1965;58:295-300. 37. Hatchett RJ, Mecher CE, Lipsitch M. Public health interventions and epidemic intensity during the 1918 influenza pandemic. Proc Natl Acad Sci U S A. 2007; 104(18):7582-7587. 38. Bootsma MCJ, Ferguson NM. The effect of public health measures on the 1918 influenza pandemic in US cities. Proc Natl Acad Sci U S A. 2007;104 (18):7588-7593. 39. Berman Institute of Bioethics; Johns Hopkins University. Bellagio meeting on social justice and influenza. http://www.johnshopkinsmedicine .org/bioethics/bellagio. Accessibility verified July 9, 2007. 40. Gostin L. Public health strategies for pandemic influenza. JAMA. 2006;295(14):1700-1704. 41. Institute of Medicine; Board on Global Health, Forum on Microbial Threats. Ethical and Legal Considerations in Mitigating a Pandemic Disease Event: Workshop Summary. Washington, DC: National Academies Press; 2007. ©2007 American Medical Association. All rights reserved. LETTERS Saturated fatty acids increase levels of low-density lipoprotein cholesterol (LDL-C) and HDL-C,2 whereas transfatty acids increase LDL-C level but decrease HDL-C level.2,3 This distinction is important, because trans-fatty acids are more strongly associated with the risk of cardiovascular disease than saturated fatty acids due to their undesirable effects on LDL-C and HDL-C levels, endothelial cell function, adipocytes, and inflammatory response.3,4 Dae Hyun Kim, MD, MPH dae-hyun.kim@mail.tju.edu Department of Medicine Jefferson Medical College Philadelphia, Pennsylvania Financial Disclosures: None reported. 1. Singh IM, Shishehbor MH, Ansell BJ. High-density lipoprotein as a therapeutic target: a systematic review. JAMA. 2007;298(7):786-798. 2. Kris-Etherton PM, Yu S. Individual fatty acid effects on plasma lipids and lipoproteins: human studies. Am J Clin Nutr. 1997;65(5)(suppl):1628S-1644S. 3. Mozaffarian D, Katan MB, Ascherio A, Stampfer MJ, Willett WC. Trans fatty acids and cardiovascular disease. N Engl J Med. 2006;354(15):1601-1613. 4. Hu FB, Stampfer MJ, Manson JE, et al. Dietary fat intake and the risk of coronary heart disease in women. N Engl J Med. 1997;337(21):1491-1499. In Reply: Dr Kim distinguishes the effects of saturated fatty acids and trans-fatty acids on lipoproteins. However, reports of the effect of trans-fatty acids on HDL-C have been variable, with a large meta-analysis finding a statistically nonsignificant effect on HDL-C.1 In addition to increasing levels of LDL-C, trans-fatty acids promote vascular inflammation and endothelial dysfunction and reduce paraoxonase activity.2 These lipid and biochemical effects act synergistically to increase cardiovascular disease risk.2 The issue is more complex than indicated by HDL-C. Saturated fat rapidly promotes proinflammatory changes in HDL without changing HDL-C level.3 Thus, as mentioned in our review, dietary intake of both saturated fatty acids and transfatty acids should be avoided and substituted with intake of monounsaturated and polyunsaturated fatty acids. Inder M. Singh, MD, MS Department of Cardiovascular Medicine Krannert Institute of Cardiology Indiana University Medical Center Indianapolis Mehdi H. Shishehbor, DO, MPH shishem@gmail.com Department of Cardiovascular Medicine Cleveland Clinic Cleveland, Ohio 2264 JAMA, November 21, 2007—Vol 298, No. 19 (Reprinted) Downloaded From: https://jamanetwork.com/ on 04/13/2020 Benjamin J. Ansell, MD Department of Medicine David Geffen School of Medicine at UCLA Los Angeles, California Financial Disclosures: Dr Ansell reported receiving speaking honoraria from AstraZeneca, Kos Pharmaceuticals, Merck, and Pfizer; receiving research medication from Merck and Pfizer in the past; and having equity interest in Bruin Pharma. No other disclosures were reported. 1. Mensink RP, Zock PL, Kester AD, Katan MB. Effects of dietary fatty acids and carbohydrates on the ratio of serum total to HDL cholesterol and on serum lipids and apolipoproteins: a meta-analysis of 60 controlled trials. Am J Clin Nutr. 2003; 77(5):1146-1155. 2. Mozaffarian D, Katan MB, Ascherio A, Stampfer MJ, Willett WC. Trans fatty acids and cardiovascular disease. N Engl J Med. 2006;354(15):1601-1613. 3. Nicholls SJ, Lundman P, Harmer JA, et al. Consumption of saturated fat impairs the anti-inflammatory properties of high-density lipoproteins and endothelial function. J Am Coll Cardiol. 2006;48(4):715-720. CORRECTIONS Data Error: In the Review article entitled “Data Extraction Errors in Meta-analyses That Use Standardized Mean Differences” published in the July 25, 2007, issue of JAMA (2007;298:430-437), Figure 4 included incorrect data. The reported point estimate and its 95% confidence interval for the meta-analysis standardized mean differences “−0.74 (−0.98 to −0.49)” for the Edmonds et al article should have read “−0.77 (−1.26 to −0.28).” The error was caused by a wrong label in the Cochrane Library at the time of the study. A meta-analysis was stated to have been done with a random-effects model; however, it was done with a fixed-effect model. The Cochrane Library no longer contains this error. Typographical Errors in Tables: In the Research Letter entitled “Patterns of Prevalent Major Chronic Disease Among Older Adults in the United States” published in the September 12, 2007, issue of JAMA (2007;298[10]:1160-1162), both tables contained typographical errors. In both tables, the column headings of “Estimated Frequency (⫻1000)” and “CVA” were erroneously transposed and the brace under the column head “Disease Pattern, No. of Diseases” should have extended to include the CVA column. Online versions of this article on the JAMA Web site were corrected on October 4, 2007. Unreported Financial Disclosures: In the Original Contribution entitled “Nonpharmaceutical Interventions Implemented by US Cities During the 1918-1919 Influenza Pandemic” published in the August 8, 2007, issue of JAMA (2007; 298[6]:644-654), financial disclosures were inadvertently not reported. On page 654, under “Financial Disclosures,” the following should appear: “Dr Markel reported having received honoraria for delivering lectures on the social and cultural history of the 1918-1919 influenza pandemic at academic conferences or colloquia presented by Yale University, the US Department of Defense, the RAND Corporation, Columbia University, the US Department of Health and Human Services, the Michigan Society for Infection Control and Prevention, University of Michigan, and Emory University. None of the other authors reported financial disclosures.” Incorrect Affiliation: In the Research Letter entitled “Cardiovascular Response to a Modern Roller Coaster Ride” published in the August 15, 2007, issue of JAMA (2007;298[7]:739-741), the affiliations were reversed for 2 authors and 1 author’s name was listed out of order. Joachim Brade, MSc, is affiliated with the Department of Medical Statistics and Dariusch Haghi, MD, is affiliated with the 1st Department of Medicine-Cardiology, University Hospital of Mannheim, Mannheim, Germany. The name for Christian Wolpert, MD, should have been placed last among the list of author names. ©2007 American Medical Association. All rights reserved.