In 1851, the German physician Carl Reinhold August Wunderlich obtained millions of axillary temperatures from 25,000 patients in Leipzig, thereby establishing the standard for normal human body temperature of 37°C or 98.6 °F (range: 36.2–37.5°C [97.2- 99.5 °F]) (Mackowiak, 1997; Wunderlich and Sequin, 1871). A compilation of 27 modern studies, however (Sund-Levander et al., 2002), reported mean temperature to be uniformly lower than Wunderlich’s estimate. Recently, an analysis of more than 35,000 British patients with almost 250,000 temperature measurements, found mean oral temperature to be 36.6°C, confirming this lower value (Obermeyer et al., 2017). Remaining unanswered is whether the observed difference between Wunderlich’s and modern averages represents true change or bias from either the method of obtaining temperature (axillary by Wunderlich vs. oral today) or the quality of thermometers and their calibration (Mackowiak, 1997). Wunderlich obtained his measurements in an era when life expectancy was 38 years and untreated chronic infections such as tuberculosis, syphilis, and periodontitis afflicted large proportions of the population (Murray et al., 2015; Tampa et al., 2014; Richmond, 2014). These infectious diseases and other causes of chronic inflammation may well have influenced the ‘normal’ body temperature of that era.

The question of whether mean body temperature is changing over time is not merely a matter of idle curiosity. Human body temperature is a crude surrogate for basal metabolic rate which, in turn, has been linked to both longevity (higher metabolic rate, shorter life span) and body size (lower metabolism, greater body mass). We speculated that the differences observed in temperature between the 19^th century and today are real and that the change over time provides important physiologic clues to alterations in human health and longevity since the Industrial Revolution.

In men, we analyzed: a) 83,900 measurements from the Union Army Veterans of the Civil War cohort (UAVCW) obtained between 1862 and 1930, b) 5998 measurements from the National Health and Nutrition Examination Survey I cohort (NHANES) obtained between 1971 and 1975, and c) 230,261 measurements from the Stanford Translational Research Integrated Database Environment cohort (STRIDE) obtained between 2007 and 2017 (Table 1). We also compared temperature measurements in women within the two later time periods (NHANES, 9303 measurements; and STRIDE, 348,006 measurements).

Overall, temperature measurements were significantly higher in the UAVCW cohort than in NHANES, and higher in NHANES than in STRIDE (Figure 1; Figure 1—figure supplement 1). In each of the three cohorts, and for both men and women, we observed that temperature decreased with age with a similar magnitude of effect (between −0.003°C and −0.0043°C per year of age, Figure 1). As has been previously reported (Eriksson et al., 1985), temperature was directly related to weight and inversely related to height, although these associations were not statistically significant in the UAVCW cohort. Analysis using body mass index (BMI) and BMI adjusted for height produced similar results (Figure 1—figure supplement 2) and analyses including only white and black subjects (Figure 1—figure supplement 3) showed similar results to those including subjects of all ethnicities.

Figure 1 with 6 supplements see all

Download asset Open asset

Figure 1—source code 1

R code for Figure 1.

https://cdn.elifesciences.org/articles/49555/elife-49555-fig1-code1-v2.rmd

Download elife-49555-fig1-code1-v2.rmd

In both STRIDE and a one-third subsample of NHANES, we confirmed the known relationship between later hour of the day and higher temperature: temperature increased 0.02°C per hour of the day in STRIDE compared to 0.01°C in NHANES (Figure 1, Figure 1—figure supplement 2, Figure 1—figure supplement 4). The month of the year had a relatively small, though statistically significant, effect on temperature in all three cohorts, but no consistent pattern emerged (Figure 1—figure supplement 5). Using approximated ambient temperature for the date and geographic location of the examination in UAVCW and STRIDE, a rise in ambient temperature of one degree Celsius correlated with 0.001 degree (p<0.001) and 0.0004 degree (p=0.013) increases in body temperature in UAVCW and STRIDE, respectively. Because the seasonal and climatic effects were small and the independent variables were unavailable for many measurements, we omitted month and estimated ambient temperature from further models.

We explored whether chronic infectious diseases—even in the absence of a diagnosis of fever—might raise temperature in the UAVCW cohort, by assessing the temperatures of men reporting a history of malaria (N = 2,203), syphilis (N = 465), or hepatitis (N = 24), or with active tuberculosis (N = 738), pneumonia (N = 277) or cystitis (N = 1,301). Only those currently diagnosed with tuberculosis or pneumonia had elevated temperatures compared to the remainder of the UAVCW population [37.22°C (95% CI: 37.20–37.24°C) and 37.06°C (95% CI: 37.03–37.09°C), respectively compared to 37.02 (95% CI: 36.52–37.53)] (Supplementary file 1).

One possible reason for the lower temperature estimates today than in the past is the difference in thermometers or methods of obtaining temperature. To minimize these biases, we examined changes in body temperature by birth decade within each cohort under the assumption that the method of thermometry would not be biased on birth year. Within the UAVCW, we observed a significant birth cohort effect, with temperatures in earlier birth decades consistently higher than those in later cohorts (Figure 2). With each birth decade, temperature decreased by −0.02°C. We then assessed change in temperature over the 197 birth-year span covered by the three cohorts. We observed a steady decrease in body temperature by birth cohort for both men (−0.59°C between birth decades from 1800 to 1997; −0.030°C per decade) and women (−0.32°C between 1890 and 1997; −0.029°C per decade). Black and white men and women demonstrated similar trends over time (Figure 3).

Figure 2

Download asset Open asset

Figure 2—source code 1

R code for Figure 2.

https://cdn.elifesciences.org/articles/49555/elife-49555-fig2-code1-v2.rmd

Download elife-49555-fig2-code1-v2.rmd

Figure 3

Download asset Open asset

Figure 3—source code 1

R code for Figure 3.

https://cdn.elifesciences.org/articles/49555/elife-49555-fig3-code1-v2.rmd

Download elife-49555-fig3-code1-v2.rmd

In this study, we analyzed 677,423 human body temperature measurements from three different cohort populations spanning 157 years of measurement and 197 birth years. We found that men born in the early 19^th century had temperatures 0.59°C higher than men today, with a monotonic decrease of −0.03°C per birth decade. Temperature has also decreased in women by −0.32°C since the 1890s with a similar rate of decline (−0.029°C per birth decade). Although one might posit that the differences among cohorts reflect systematic measurement bias due to the varied thermometers and methods used to obtain temperatures, we believe this explanation to be unlikely. We observed similar temporal change within the UAVCW cohort—in which measurement were presumably obtained irrespective of the subject's birth decade—as we did between cohorts. Additionally, we saw a comparable magnitude of difference in temperature between two modern cohorts using thermometers that would be expected to be similarly calibrated. Moreover, biases introduced by the method of thermometry (axillary presumed in a subset of UAVCW vs. oral for other cohorts) would tend to underestimate change over time since axillary values typically average one degree Celsius lower than oral temperatures (Sund-Levander et al., 2002; Niven et al., 2015). Thus, we believe the observed drop in temperature reflects physiologic differences rather than measurement bias. Other findings in our study—for example increased temperature at younger ages, in women, with increased body mass and with later time of day—support a wealth of other studies dating back to the time of Wunderlich (Wunderlich and Sequin, 1871; Waalen and Buxbaum, 2011).

Resting metabolic rate is the largest component of a typical modern human’s energy expenditure, comprising around 65% of daily energy expenditure for a sedentary individual (Heymsfield et al., 2006). Heat is a byproduct of metabolic processes, the reason nearly all warm-blooded animals have temperatures within a narrow range despite drastic differences in environmental conditions. Over several decades, studies examining whether metabolism is related to body surface area or body weight (Du Bois, 1936; Kleiber, 1972), ultimately, converged on weight-dependent models (Mifflin et al., 1990; Schofield, 1985; Nelson et al., 1992). Since US residents have increased in mass since the mid-19^th century, we should have correspondingly expected increased body temperature. Thus, we interpret our finding of a decrease in body temperature as indicative of a decrease in metabolic rate independent of changes in anthropometrics. A decline in metabolic rate in recent years is supported in the literature when comparing modern experimental data to those from 1919 (Frankenfield et al., 2005).

Although there are many factors that influence resting metabolic rate, change in the population-level of inflammation seems the most plausible explanation for the observed decrease in temperature over time. Economic development, improved standards of living and sanitation, decreased chronic infections from war injuries, improved dental hygiene, the waning of tuberculosis and malaria infections, and the dawn of the antibiotic age together are likely to have decreased chronic inflammation since the 19^th century. For example, in the mid-19^th century, 2–3% of the population would have been living with active tuberculosis (Tiemersma et al., 2011). This figure is consistent with the UAVCW Surgeons' Certificates that reported 737 cases of active tuberculosis among 23,757 subjects (3.1%). That UAVCW veterans who reported either current tuberculosis or pneumonia had a higher temperature (0.19°C and 0.03°C respectively) than those without infectious conditions supports this theory (Supplementary file 1). Although we would have liked to have compared our modern results to those from a location with a continued high risk of chronic infection, we could identify no such database that included temperature measurements. However, a small study of healthy volunteers from Pakistan—a country with a continued high incidence of tuberculosis and other chronic infections—confirms temperatures more closely approximating the values reported by Wunderlich (mean, median and mode, respectively, of 36.89°C, 36.94°C, and 37°C) (Adhi et al., 2008).

Reduction in inflammation may also explain the continued drop in temperature observed between the two more modern cohorts: NHANES and STRIDE. Although many chronic infections had been conquered before the NHANES study, some—periodontitis as one example (Capilouto and Douglass, 1988)— continued to decrease over this short period. Moreover, the use of anti-inflammatory drugs including aspirin (Luepker et al., 2015), statins (Salami et al., 2017) and non-steroidal anti-inflammatory drugs (NSAIDs) (Lamont and Dias, 2008) increased over this interval, potentially reducing inflammation. NSAIDs have been specifically linked to blunting of body temperature, even in normal volunteers (Murphy et al., 1996). In support of declining inflammation in the modern era, a study of NHANES participants demonstrated a 5% decrease in abnormal C-reactive protein levels between 1999 and 2010 (Ong et al., 2013).

Changes in ambient temperature may also explain some of the observed change in body temperature over time. Maintaining constant body temperature despite fluctuations in ambient temperature consumes up to 50–70% of daily energy intake (Levine, 2007). Resting metabolic rate (RMR), for which body temperature is a crude proxy, increases when the ambient temperature decreases below or rises above the thermoneutral zone, that is the temperature of the environment at which humans can maintain normal temperature with minimum energy expenditure (Erikson et al., 1956). In the 19^th century, homes in the US were irregularly and inconsistently heated and never cooled. By the 1920s, however, heating systems reached a broad segment of the population with mean night-time temperature continuing to increase even in the modern era (Mavrogianni et al., 2013). Air conditioning is now found in more than 85% of US homes (US Energy Information Administration, 2011). Thus, the amount of time the population has spent at thermoneutral zones has markedly increased, potentially causing a decrease in RMR, and, by analogy, body temperature.

Some factors known to influence body temperature were not included in our final model due to missing data (ambient temperature and time of day) or complete lack of information (dew point)(Obermeyer et al., 2017). Adjusting for ambient temperature, however, would likely have amplified the changes over time due to lack of heating and cooling in the earlier cohorts. Time of day at which measurement was conducted had a more significant effect on temperature (Figure 1—figure supplement 4). Based on the distribution of times of day for temperature measurement available to us in STRIDE and NHANES, we estimate that even in the worst case scenario, that is the UAVCW measurements were all were obtained late in the afternoon, adjustment for time of day would have only a small influence (<0.05°C) on the −0.59°C change over time.

In summary, normal body temperature is assumed by many, including a great preponderance of physicians, to be 37°C. Those who have shown this value to be too high have concluded that Wunderlich’s 19^th century measurements were simply flawed (Mackowiak, 1997; Sund-Levander et al., 2002). Our investigation indicates that humans in high-income countries have changed physiologically over the last 200 birth years with a mean body temperature 1.6% lower than in the pre-industrial era. The role that this physiologic ‘evolution’ plays in human anthropometrics and longevity is unknown.

All data generated or analysed during this study are included in the manuscript and supporting files.

1. 1. Fogel RW
   2. Costa DL

   (2018) National Archive of Computerized Data on Aging

   Aging of Veterans of the Union Army: Surgeons' Certificates, United States, 1862-1940 (ICPSR 2877).

   https://doi.org/10.3886/ICPSR02877.v2
2. 1. United States Department of Health and Human Services
   2. National Center for Health Statistics

   (1992) National Archive of Computerized Data on Aging

   National Health and Nutrition Examination Survey I, 1971-1975: Medical Examination (ICPSR 8055).

   https://doi.org/10.3886/ICPSR08055.v2

1. 1. Adhi M
   2. Hasan R
   3. Noman F
   4. Mahmood SF
   5. Naqvi A
   6. Rizvi AU

   (2008)

   Range for normal body temperature in the general population of Pakistan

   Journal of the Pakistan Medical Association 58:580–584.

   • PubMed
   • Google Scholar
2. 1. Capilouto ML
   2. Douglass CW

   (1988) Trends in the prevalence and severity of periodontal diseases in the US: a public health problem?

   Journal of Public Health Dentistry 48:245–251.

   https://doi.org/10.1111/j.1752-7325.1988.tb03206.x

   • PubMed
   • Google Scholar
3. Report

   1. Centers for Disease Control, National Center for Health Statistics

   (1975)

   National Health and Nutrition Examination Survey Data, 1971-1974

   US Department of Health and Human Services, Centers for Disease Control and Prevention.

   • Google Scholar
4. Website

   1. Costa DL

   (2019) Union army data: early indicators of later work levels, disease and death

   Accessed February 23, 2018.

   http://uadata.org/cite/
5. Book

   1. Du Bois EF

   (1936)

   Basal Metabolism in Health and Disease

   Lea & Febiger.

   • Google Scholar
6. 1. Erikson H
   2. Krog J
   3. Andersen KL
   4. Scholander PF

   (1956) The critical temperature in naked man

   Acta Physiologica Scandinavica 37:35–39.

   https://doi.org/10.1111/j.1748-1716.1956.tb01339.x

   • PubMed
   • Google Scholar
7. 1. Eriksson H
   2. Svärdsudd K
   3. Larsson B
   4. Welin L
   5. Ohlson LO
   6. Wilhelmsen L

   (1985)

   Body temperature in general population samples. The study of men born in 1913 and 1923

   Acta Medica Scandinavica 217:347–352.

   • PubMed
   • Google Scholar
8. Report

   1. Fogel RW
   2. Wimmer LT

   (1992) Early Indicators of Later Work Levels, Diseases and Death

   National Bureau of Economic Research.

   https://doi.org/10.3386/h0038

   • Google Scholar
9. Report

   1. Fogel RW
   2. Costa DL
   3. Haines M
   4. Lee C
   5. Nguyen L
   6. Pope C
   7. Rosenberg I
   8. Scrimshaw N
   9. Trussell J
   10. Wilson S
   11. Wimmer LT
   12. Kim J
   13. Bassett J
   14. Burton J
   15. Yetter N

   (2000)

   Aging of Veterans of the Union Army: Version M-5

   Chicago: Center for Population Economics, University of Chicago Graduate School of Business, Department of Economics, Brigham Young University and the National Bureau of Economic Research.

   • Google Scholar
10. 1. Frankenfield D
    2. Roth-Yousey L
    3. Compher C

    (2005) Comparison of predictive equations for resting metabolic rate in healthy nonobese and obese adults: a systematic review

    Journal of the American Dietetic Association 105:775–789.

    https://doi.org/10.1016/j.jada.2005.02.005

    • PubMed
    • Google Scholar
11. 1. Heymsfield SB
    2. Harp JB
    3. Rowell PN
    4. Nguyen AM
    5. Pietrobelli A

    (2006) How much may I eat? calorie estimates based upon energy expenditure prediction equations

    Obesity Reviews 7:361–370.

    https://doi.org/10.1111/j.1467-789X.2006.00249.x

    • PubMed
    • Google Scholar
12. 1. Kleiber M

    (1972) A new Newton's law of cooling?

    Science 178:1283–1285.

    https://doi.org/10.1126/science.178.4067.1283

    • PubMed
    • Google Scholar
13. 1. Lamont EB
    2. Dias LE

    (2008) Secular changes in NSAID use and invasive colorectal Cancer incidence: an ecological study

    The Cancer Journal 14:276–277.

    https://doi.org/10.1097/PPO.0b013e318182ee67

    • PubMed
    • Google Scholar
14. 1. Levine JA

    (2007) Nonexercise activity thermogenesis--liberating the life-force

    Journal of Internal Medicine 262:273–287.

    https://doi.org/10.1111/j.1365-2796.2007.01842.x

    • PubMed
    • Google Scholar
15. 1. Luepker RV
    2. Steffen LM
    3. Duval S
    4. Zantek ND
    5. Zhou X
    6. Hirsch AT

    (2015) Population trends in aspirin use for cardiovascular disease prevention 1980-2009: the Minnesota heart survey

    Journal of the American Heart Association 4:e002320.

    https://doi.org/10.1161/JAHA.115.002320

    • PubMed
    • Google Scholar
16. 1. Mackowiak PA

    (1997)

    Fever: Basic Mechanisms and Management

    History of clinical thermometry, Fever: Basic Mechanisms and Management, Philadelphia, Lippincott Williams & Wilkins.

    • Google Scholar
17. 1. Mavrogianni A
    2. Johnson F
    3. Ucci M
    4. Marmot A
    5. Wardle J
    6. Oreszczyn T
    7. Summerfield A

    (2013) Historic variations in winter indoor domestic temperatures and potential implications for body weight gain

    Indoor and Built Environment 22:360–375.

    https://doi.org/10.1177/1420326X11425966

    • PubMed
    • Google Scholar
18. 1. Mifflin MD
    2. St Jeor ST
    3. Hill LA
    4. Scott BJ
    5. Daugherty SA
    6. Koh YO

    (1990) A new predictive equation for resting energy expenditure in healthy individuals

    The American Journal of Clinical Nutrition 51:241–247.

    https://doi.org/10.1093/ajcn/51.2.241

    • PubMed
    • Google Scholar
19. 1. Murphy PJ
    2. Myers BL
    3. Badia P

    (1996) Nonsteroidal anti-inflammatory drugs alter body temperature and suppress melatonin in humans

    Physiology & Behavior 59:133–139.

    https://doi.org/10.1016/0031-9384(95)02036-5

    • PubMed
    • Google Scholar
20. 1. Murray JF
    2. Schraufnagel DE
    3. Hopewell PC

    (2015) Treatment of tuberculosis. A historical perspective

    Annals of the American Thoracic Society 12:1749–1759.

    https://doi.org/10.1513/AnnalsATS.201509-632PS

    • PubMed
    • Google Scholar
21. 1. Nelson KM
    2. Weinsier RL
    3. Long CL
    4. Schutz Y

    (1992) Prediction of resting energy expenditure from fat-free mass and fat mass

    The American Journal of Clinical Nutrition 56:848–856.

    https://doi.org/10.1093/ajcn/56.5.848

    • PubMed
    • Google Scholar
22. 1. Niven DJ
    2. Gaudet JE
    3. Laupland KB
    4. Mrklas KJ
    5. Roberts DJ
    6. Stelfox HT

    (2015) Accuracy of peripheral thermometers for estimating temperature: a systematic review and meta-analysis

    Annals of Internal Medicine 163:768–777.

    https://doi.org/10.7326/M15-1150

    • PubMed
    • Google Scholar
23. Website

    1. NOAA National Centers for Environmental Information

    (2018) Climate at a glance: statewide times series

    Accessed June 3, 2018.

    https://www.ncdc.noaa.gov/cag/
24. 1. Obermeyer Z
    2. Samra JK
    3. Mullainathan S

    (2017) Individual differences in normal body temperature: longitudinal big data analysis of patient records

    BMJ 359:j5468.

    https://doi.org/10.1136/bmj.j5468

    • PubMed
    • Google Scholar
25. 1. Ong KL
    2. Allison MA
    3. Cheung BM
    4. Wu BJ
    5. Barter PJ
    6. Rye KA

    (2013) Trends in C-reactive protein levels in US adults from 1999 to 2010

    American Journal of Epidemiology 177:1430–1442.

    https://doi.org/10.1093/aje/kws443

    • PubMed
    • Google Scholar
26. Website

    1. Richmond D

    (2014) No teeth, no man: dentistry during the civil war

    Accessed July 11, 2018.

    http://civilwarrx.blogspot.com/2014/08/no-teeth-no-man-dentistry-during-civil.html
27. 1. Salami JA
    2. Warraich H
    3. Valero-Elizondo J
    4. Spatz ES
    5. Desai NR
    6. Rana JS
    7. Virani SS
    8. Blankstein R
    9. Khera A
    10. Blaha MJ
    11. Blumenthal RS
    12. Lloyd-Jones D
    13. Nasir K

    (2017) National trends in statin use and expenditures in the US adult population from 2002 to 2013: insights from the medical expenditure panel survey

    JAMA Cardiology 2:56–65.

    https://doi.org/10.1001/jamacardio.2016.4700

    • PubMed
    • Google Scholar
28. Book

    1. Salinger JL
    2. Kalteyer FJ

    (1900)

    Symptomatology and Semeiology. Modern Medicine

    Saunders and Co.

    • Google Scholar
29. 1. Schofield WN

    (1985)

    Predicting basal metabolic rate, new standards and review of previous work

    Human Nutrition. Clinical Nutrition 39:5–41.

    • PubMed
    • Google Scholar
30. 1. Sund-Levander M
    2. Forsberg C
    3. Wahren LK

    (2002) Normal oral, rectal, tympanic and axillary body temperature in adult men and women: a systematic literature review

    Scandinavian Journal of Caring Sciences 16:122–128.

    https://doi.org/10.1046/j.1471-6712.2002.00069.x

    • PubMed
    • Google Scholar
31. 1. Tampa M
    2. Sarbu I
    3. Matei C
    4. Benea V
    5. Georgescu SR

    (2014)

    Brief history of syphilis

    Journal of Medicine and Life 7:4–10.

    • PubMed
    • Google Scholar
32. 1. Tiemersma EW
    2. van der Werf MJ
    3. Borgdorff MW
    4. Williams BG
    5. Nagelkerke NJ

    (2011) Natural history of tuberculosis: duration and fatality of untreated pulmonary tuberculosis in HIV negative patients: a systematic review

    PLOS ONE 6:e17601.

    https://doi.org/10.1371/journal.pone.0017601

    • PubMed
    • Google Scholar
33. Website

    1. US Energy Information Administration

    (2011) Residential energy consumption survey (RECS)

    US Energy Information Administration. Accessed June 3, 2018.

    https://www.eia.gov/consumption/residential/reports/2009/air-conditioning.php
34. 1. Waalen J
    2. Buxbaum JN

    (2011) Is older colder or colder older? The association of age with body temperature in 18,630 individuals

    The Journals of Gerontology Series A: Biological Sciences and Medical Sciences 66A:487–492.

    https://doi.org/10.1093/gerona/glr001

    • Google Scholar
35. Book

    1. Wunderlich CA
    2. Sequin E

    (1871)

    Medical Thermometry, and Human Temperature

    New York: William Wood & Company.

    • Google Scholar

Author details

1. Myroslava Protsiv

   Division of Infectious Diseases and Geographic Medicine, Department of Medicine, Stanford University, School of Medicine, Stanford, United States

   Contribution

   Data curation, Software, Formal analysis, Validation, Visualization, Methodology, Writing—original draft

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0002-7787-5898

2. Catherine Ley

   Division of Infectious Diseases and Geographic Medicine, Department of Medicine, Stanford University, School of Medicine, Stanford, United States

   Contribution

   Formal analysis, Methodology, Writing—review and editing

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0002-8424-7873

3. Joanna Lankester

   Division of Cardiovascular Medicine, Stanford University, School of Medicine, Stanford, United States

   Contribution

   Methodology, Writing—review and editing

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0003-0709-6722

4. Trevor Hastie

   1. Department of Statistics, Stanford University, Stanford, United States
   2. Department of Biomedical Data Science, Stanford University, School of Medicine, Stanford, United States

   Contribution

   Formal analysis, Visualization, Methodology, Writing—review and editing

   Competing interests

   No competing interests declared

5. Julie Parsonnet

   1. Division of Infectious Diseases and Geographic Medicine, Department of Medicine, Stanford University, School of Medicine, Stanford, United States
   2. Division of Epidemiology, Department of Health Research and Policy, Stanford University, School of Medicine, Stanford, United States

   Contribution

   Conceptualization, Resources, Supervision, Funding acquisition, Investigation, Visualization, Methodology, Project administration, Writing—review and editing

   For correspondence

   parsonnt@stanford.edu

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0001-7342-5366

• 174,957

  views

• 4,387

  downloads

• 107

  citations

Views, downloads and citations are aggregated across all versions of this paper published by eLife.