Attack Rate

An attack rate (AR) is a core biostatistical metric in field epidemiology and public health outbreak investigation that measures the cumulative incidence proportion (risk) of a defined, initially disease-free susceptible population that develops a specified illness over a bounded epidemic or exposure window. Despite the historical term "rate," an attack rate is dimensionally a proportion (bounded between 0% and 100%, or 0.0 and 1.0) rather than an instantaneous incidence rate per unit of person-time, because the observation window is implicitly defined as the entire duration of a finite outbreak (such as a 48-hour foodborne gastroenteritis cluster, a 14-day household quarantine period, or a seasonal influenza wave). By comparing attack rates across exposure groups, household contacts, and immunized cohorts, epidemiologists identify contaminated point sources, quantify person-to-person contagiousness via the secondary attack rate (SAR), and calculate vaccine efficacy (VE).

1. Crude (Overall) Outbreak Attack Rate and Infection Attack Rate

In a closed population of N initially susceptible individuals at risk at the beginning of an outbreak (t = 0), if C_new individuals develop clinical cases of the disease before the outbreak concludes (t = T), the Overall (Crude) Clinical Attack Rate is:

AR = ( C_new / N_{susceptible at risk} ) × 100%

Strict definition of the denominator N_{susceptible at risk} requires excluding individuals who cannot biologically develop the outcome—for example, excluding individuals with documented prior sterilizing immunity (such as adults with verified two-dose MMR immunity during a school measles outbreak) or individuals lacking the target organ. When seroprevalence surveys or universal PCR screening capture all asymptomatic and subclinical infections (I_total) in addition to symptomatic cases, the corresponding proportion is termed the Infection Attack Rate (IAR = I_total / N), which represents the final size of an epidemic in compartmental SIR models.

2. Food-Specific (Exposure-Specific) Attack Rates in Point-Source Outbreaks

One of the classic applications of the attack rate—pioneered by Wade Hampton Frost and refined in the U.S. Centers for Disease Control and Prevention (CDC) Epidemic Intelligence Service (EIS) field manual—is retrospective cohort analysis of acute foodborne or waterborne point-source outbreaks (such as Salmonella enterica, enterotoxigenic Escherichia coli, Staphylococcus aureus superantigen enterotoxin, or norovirus following a banquet or conference).

Investigators interview all attendees and construct a two-by-two contingency table for each menu item j, partitioning attendees into those who ate (exposed) the item (a_j ill, b_j well; total exposed N_1,j = a_j + b_j) and those who did not eat (unexposed) the item (c_j ill, d_j well; total unexposed N_0,j = c_j + d_j). Two exposure-specific attack rates and their Attack Rate Ratio (Relative Risk, RR) and Attributable Risk (Risk Difference, RD) are computed:

AR_{exposed, j} = a_j / ( a_j + b_j )

AR_{unexposed, j} = c_j / ( c_j + d_j )

RR_j = AR_{exposed, j} / AR_{unexposed, j}

The true contaminated vehicle is identified by the combination of a high attack rate among the exposed (AR_exposed), a low attack rate among the unexposed (AR_unexposed) (accounting for cross-contamination or recall error), a large, statistically significant Relative Risk (RR ≫ 1.0), and high population attributable fraction explaining the majority of outbreak cases:

Whereas the primary attack rate measures cases arising from a shared initial exposure, the Secondary Attack Rate (SAR)—first introduced by Charles V. Chapin (1903)—measures the probability of person-to-person transmission within a close-contact mixing unit (such as a household, hospital ward, classroom, or military barracks) following the introduction of one or more primary (index) cases.

Mathematical Formula and Denominator Exclusions

Suppose K households each experience C_primary primary index cases introduced from the community. During the subsequent window spanning one minimum to one maximum incubation period (or serial interval) of the pathogen after index symptom onset, C_secondary new cases develop among the remaining household members. To measure person-to-person contagiousness accurately, the primary index cases (and any co-primary cases whose symptoms began within LESS than one minimum incubation period of the index case) are subtracted from both the numerator and the denominator, alongside any household contacts known to be previously immune:

SAR = [ C_secondary / ( N_{total contacts} − C_{primary/co-primary} − N_{previously immune} ) ] × 100%

For example, across 100 households of 4 susceptible people each (N_total = 400), if 1 child per household brings home chickenpox (Varicella-zoster virus, C_primary = 100) and subsequently 255 of the remaining 300 susceptible household members develop chickenpox within 10 to 21 days, the household Secondary Attack Rate is 255 / 300 = 85.0% (whereas the naive overall attack rate would be 355 / 400 = 88.75%).

Relationship Between Household SAR, R₀, and Epidemiological Control

Across infectious pathogens, the household SAR provides a direct empirical gauge of close-contact transmissibility:

• High Household SAR (70% to 90%): Characteristic of airborne pathogens with high basic reproduction numbers (R₀), including measles (SAR ≈ 85%–90%), varicella (SAR ≈ 80%–85%), and pertussis (Bordetella pertussis) (SAR ≈ 80%).
• Intermediate Household SAR (15% to 45%): Characteristic of droplet/aerosol respiratory viruses such as SARS-CoV-2 (ancestral SAR ≈ 16%–19%; Delta SAR ≈ 25%–30%; Omicron SAR ≈ 35%–45%), pandemic/seasonal Influenza A (SAR ≈ 10%–25%), and Ebola virus in unisolated households (SAR ≈ 15%–25%).
• Low Household SAR (< 1% to 8%): Characteristic of zoonotic spillover pathogens poorly adapted to human-to-human transmission, such as Avian Influenza A (H5N1) (SAR < 1%) or MERS-CoV (SAR ≈ 4%–5%), as well as strictly vector-borne diseases (malaria, dengue fever) in the absence of competent mosquitoes inside the dwelling.
• Differentiating Point-Source vs. Contagious Outbreaks: When an epidemic curve displays an abrupt peak followed by a tail of later cases, computing the household SAR distinguishes whether the trailing cases represent delayed incubation from the original point source (low household clustering) or secondary person-to-person propagation (elevated household SAR).

Kermack-McKendrick Final Attack Rate Equation

In a homogeneously mixing population governed by the compartmental SIR model with Basic Reproduction Number R₀ > 1 and no external interventions, the fraction of the population eventually infected by the end of the epidemic—the Final Infection Attack Rate (AR_∞ = Z)—satisfies the implicit transcendental Kermack-McKendrick (1927) Final Size Equation:

Z = 1 − exp( −R₀ · Z )

Even without vaccination, an epidemic never infects 100% of a population for finite R₀, because transmission decelerates once the susceptible fraction drops below the herd immunity threshold (p_c = 1 − 1/R₀), although momentum "overshoots" that threshold:

• For R₀ = 1.5 (herd immunity threshold = 33.3%), the final attack rate Z ≈ 58.3%.
• For R₀ = 2.5 (herd immunity threshold = 60.0%), the final attack rate Z ≈ 89.3%.
• For R₀ = 5.0 (herd immunity threshold = 80.0%), the final attack rate Z ≈ 99.3%.

Calculating Vaccine Efficacy (VE) from Attack Rates

Finally, as formalized by British statisticians Major Greenwood and G. Udny Yule (1915), the clinical efficacy of a vaccine in randomized controlled trials (and vaccine effectiveness in observational field studies) is defined directly as the proportional reduction in the attack rate between the unvaccinated control group (AR_u) and the vaccinated group (AR_v):

VE = [ ( AR_u − AR_v ) / AR_u ] × 100% = ( 1 − RR ) × 100%

For example, in a Phase III randomized trial of 40,000 participants (20,000 vaccinated, 20,000 placebo), if 160 symptomatic cases occur in the unvaccinated placebo group (AR_u = 160 / 20,000 = 0.80%) and 8 cases occur in the vaccinated group (AR_v = 8 / 20,000 = 0.04%), the Relative Risk is RR = 0.04% / 0.80% = 0.05, and the Vaccine Efficacy is (1 − 0.05) × 100% = 95.0%. Furthermore, by tracking the household Secondary Attack Rate from vaccinated breakthrough index cases vs. unvaccinated index cases (VE_{infectiousness} = 1 − SAR_{vaccinated index} / SAR_{unvaccinated index}), epidemiologists separately quantify how strongly immunization blocks onward transmission to contacts in addition to protecting the vaccinated recipient (VE_{susceptibility}).

1. Greenwood, M., & Yule, G. U. (1915). "The statistics of anti-typhoid and anti-cholera inoculations, and the interpretation of such statistics in general." Proceedings of the Royal Society of Medicine, 8(Sect Epidemiol State Med), 113–194. https://doi.org/10.1177/003591571500801433
2. Kermack, W. O., & McKendrick, A. G. (1927). "A contribution to the mathematical theory of epidemics." Proceedings of the Royal Society of London. Series A, 115(772), 700–721. https://doi.org/10.1098/rspa.1927.0118
3. Halloran, M. E., Préziosi, M.-P., & Chu, H. (2003). "Estimating vaccine efficacy from secondary attack rates." Journal of the American Statistical Association, 98(461), 38–46. https://doi.org/10.1198/016214503388619076
4. Madewell, Z. J., Yang, Y., Longini, I. M., Halloran, M. E., & Dean, N. E. (2020). "Household transmission of SARS-CoV-2: a systematic review and meta-analysis." JAMA Network Open, 3(12), e2031756. https://doi.org/10.1001/jamanetworkopen.2020.31756
5. Celentano, D. D., & Szklo, M. (2019). Gordis Epidemiology (6th ed.). Elsevier. ISBN 978-0-323-55229-5