Data on self-reported symptom and rash onset dates for primary and secondary case pairs, including the type of contact that occurred between pairs, were compiled by 12 state and local health departments. We examined serial interval for both earliest symptom onset and rash onset because the latter may be more specific to mpox than the other signs. Earliest symptom onset included any mpox symptom as defined by CDC (2), including rash. We only included cases if there was a high degree of certainty that the secondary case-patient was infected by the primary case-patient (Appendix).

For each case pair, we calculated days between onset of any mpox symptoms and days between rash onset in the primary and secondary case-patients. We used the EpiEstim package version 4.1.2 in R software (The R Foundation for Statistical Computing, https://www.r-project.org) to estimate the distribution of the serial interval for known primary and secondary case pairs using Bayesian methods for symptom and rash onset (Appendix) (7). We did not adjust for right-truncation of the data because we included cases during the stable or declining phase of the outbreak (8).

Figure 1

Figure 1. Empirical and fitted distributions of the serial intervals of rash onset (A) (n = 40 cases) and symptom onset (B) (n = 57 cases) for monkeypox virus, 12 jurisdictions, United...

We received data for 120 case pairs from 13 jurisdictions (Appendix Figures 1, 2). Fifty-seven case pairs met the inclusion criteria (Figure 1; Appendix). Dates of symptom onset among primary case-patients ranged from May 11 to August 13, 2022 (Appendix Figure 3). Forty of the 57 pairs included rash onset dates for primary and secondary case-patients. We also considered in our analysis the type of contact for the case pairs (Appendix Table 1). The gamma distribution provided the best fit to the serial interval data. The overall mean estimated serial interval for symptom onset was 8.5 (95% CrI 7.3–9.9) days (SD 5.0 [95% CrI 4.0–6.4] days) and for rash onset was 7.0 (95% CrI 5.8–8.4) days (SD 4.2 [95% CrI 3.2–5.6] days) (Table; Appendix Tables 2, 3).

Figure 2

Figure 2. Updated estimated cumulative density functions according to a log-normal distribution of monkeypox virus incubation periods, by symptom onset (A) (n = 36 cases) and rash onset (B) (n = 35...

We estimated the incubation period using 22 US mpox cases reported in an earlier study (K. Charniga et al., unpub. data, https://doi.org/10.1101/2022.06.22.22276713) plus 14 cases from our dataset (Appendix). Of the new case-patients, 10 were exposed during a single day. The mean incubation period from exposure to symptom onset for 36 case-patients was 5.6 (95% CrI 4.3–7.8) days (SD 4.4 [95% CrI 2.8–8.7] days), whereas the mean incubation period from exposure to rash onset for 35 case-patients was 7.5 (95% CrI 6.0–9.8) days (SD 4.9 [95% CrI 3.2–8.8] days) (Figure 2; Appendix Figure 4).

Determining the serial interval of a pathogen can inform our understanding of the timing of transmission relative to symptom onset. The serial interval of 8.5 days for symptom onset was similar to an estimate from the United Kingdom of 9.8 days after correcting for right-truncation (5) but shorter than the generation time of 12.5 days reported from Italy (6). The serial interval correlates with human behavior and can decrease with increasing awareness among men who have sex with men or interventions, a pattern similar to that observed among the general population during the COVID-19 pandemic (9). The Italy estimates were from the initial period of the epidemic (May–June 2022), whereas our study was for July–August 2022. Serial interval changes can be very rapid (10). We also found a serial interval of 7.0 days for rash onset. The estimated serial interval for symptom onset was longer than that for the incubation period (5.6 days), suggesting most transmission occurred after the onset of symptoms in the primary case-patient. Conversely, the serial interval for rash onset (7.0 days) was slightly shorter than that for the rash incubation period (7.5 days), which may suggest some prerash transmission; indeed, there were instances in the observed data where secondary case-patients were exposed before onset of reported rash in the primary case-patient. However, the credible intervals for the estimates overlap. The serial interval for symptom onset ranged from 2 to 25 days. This wide range may be attributable in part to variability in the nature and intensity of contact.

The first limitation of this study is that precise ascertainment of symptom and rash onset dates is critical for serial interval estimation, but initial mpox symptoms are often nonspecific and may be unrelated to MPXV infection. Second, despite careful selection of linked primary and secondary case pairs, exposure from additional unknown sources may have occurred. Third, social desirability bias may have factored into the self-reported exposures before infection. Fourth, serial interval may vary by age, underlying conditions, vaccination status, or contact type (route of exposure); we did not stratify our analysis by these factors because of limited data. Fifth, we excluded secondary case-patients who had symptom onset on the same day as or before the primary case-patient to ensure a high degree of confidence linking case pairs; however, the serial interval could be negative (11). Sixth, the serial interval for rash onset could be biased if rash is more quickly identified in the secondary case-patient because of case finding and investigation of the primary case-patient.

Notwithstanding those limitations, our estimate of the serial interval for MPXV infection includes more case pairs than have been reported previously from the United Kingdom (5) and Italy (6). We also provide estimates for rash onset, which may be more reliable than initial symptom onset for determining serial interval.