Source: Kulagin, Fedorova & Ketiladze (1962).
Outbreak period: 18 Oct – 29 Nov 1961. Total cases: 113.
PDF page references use paper page numbers (= PDF viewer page − 4).
PDF source (group identity): Paper p. 6 (PDF p. 10), paragraph 2: > “…a group of persons who remained in the department for 3–4 hours only > on the 24th of October. Out of 24 persons in this group, 18 became ill.”
PDF source (individual incubation data): Paper p. 8 (PDF p. 12), paragraph 2: > “The duration of the incubation period in their group was as follows: > 3 of 17 days, 1 of 19 days, 2 of 21 days, 8 of 22 days, 1 of 23 days, > 2 of 25 days, and 1 of 27 days (an average of 21.5 days).”
| Case_no | n | E_L | E_R | Onset | k | T_lower | T_upper | Interval |
|---|---|---|---|---|---|---|---|---|
| A-1 | 3 | Oct 24 | Oct 24 | Nov 10 | 17 | 16 | 18 | 2 days |
| A-2 | 1 | Oct 24 | Oct 24 | Nov 12 | 19 | 18 | 20 | 2 days |
| A-3 | 2 | Oct 24 | Oct 24 | Nov 14 | 21 | 20 | 22 | 2 days |
| A-4 | 8 | Oct 24 | Oct 24 | Nov 15 | 22 | 21 | 23 | 2 days |
| A-5 | 1 | Oct 24 | Oct 24 | Nov 16 | 23 | 22 | 24 | 2 days |
| A-6 | 2 | Oct 24 | Oct 24 | Nov 18 | 25 | 24 | 26 | 2 days |
| A-7 | 1 | Oct 24 | Oct 24 | Nov 20 | 27 | 26 | 28 | 2 days |
IC likelihood: ✅ Fully applicable. Formula:
T_lower = k − 1, T_upper = k + 1 (within-day
calendar uncertainty only). All T_lower ≥ 16 > 0; interval width is a
constant 2 days across all cases. This is the tightest possible window
consistent with calendar-day recording.
Current status: ✅ INCLUDED (v3b, v4)
PDF source (overview): Paper p. 5 (PDF p. 9), paragraphs 3–4: > “During October… out of 63 permanent personnel, 5 had had hemorrhagic fever > previously and apparently were immune, consequently 58 men were at risk, > of these 52 became ill (89.6%).”
Exposure window left bound (E_L = Oct 7): Paper p. 4 (PDF p. 8), last paragraph: > “the rodents, on 7 Oct, were transferred over to permanent accommodations > for animals (rooms #2 and 12)”
Exposure window right bound (E_R ≈ Oct 25): Animal removal announced Nov 2 (paper p. 9); earliest B-1 onset was Oct 18, suggesting effective exposure ended ~Oct 25.
PDF source: Paper p. 5 (PDF p. 9), paragraph 3: > “out of 23 persons working in the 1st section (rooms #3, 9, 10, 12, and 13), > 20 had become ill”
| Group | n | Cases | AR | E_L | E_R | Onset range |
|---|---|---|---|---|---|---|
| B-1 | 23 | 20 | 87% | Oct 7 | Oct 25 | Oct 18 – unknown |
IC likelihood: ⚠️ Applicable in principle;
informativeness is onset-dependent. Formula:
T_lower = O_day − Oct 25 − 1,
T_upper = O_day − Oct 7 + 1 (window width = 18 days, fixed
by E window).
| Onset example | T_lower | T_upper | Width | Info type |
|---|---|---|---|---|
| Oct 18 (earliest) | −8 → clamped to ~0 | 12 | ~12 | Upper bound only |
| Oct 28 | 2 | 22 | 20 | Interval (weak) |
| Nov 10 | 15 | 35 | 20 | Interval (moderate) |
| Nov 20 | 25 | 45 | 20 | Interval (broader) |
Early-onset cases (onset ≤ Oct 25, i.e., within the exposure window) yield T_lower ≤ 0 — effectively only an upper bound. Later-onset cases provide genuine two-sided intervals of ~20-day width.
Current status: ❌ NOT INCLUDED — individual onset dates unavailable
PDF source: Paper p. 5 (PDF p. 9), paragraph 3: > “only 10 of the 33 workers in the 2nd section became ill”
| Group | n | Cases | AR | E_L | E_R |
|---|---|---|---|---|---|
| B-2 | 33 | 10 | 30% | Oct 7 | Oct 25 |
IC likelihood: ⚠️ Same formula as B-1; onsets likely later (Zone 2 lagged Zone 1 per paper p. 5), so a larger fraction may have T_lower > 0. Window width = 20 days if individual onset data were obtained.
Current status: ❌ NOT INCLUDED — individual onset dates unavailable
PDF source: Paper p. 5 (PDF p. 9), paragraph 3: > “out of 63 permanent personnel, 5 had had hemorrhagic fever previously > and apparently were immune”
Conclusion 4, paper p. 10 (PDF p. 14): > “5 workers of those working constantly in the infected area had this > disease previously (1–3 years ago) and didn’t become ill.”
| Group | n | Cases | Notes |
|---|---|---|---|
| B-3 | 5 | 0 | Prior immunity |
IC likelihood: N/A — no cases; not relevant to incubation estimation.
Current status: N/A (no cases)
PDF source: Paper p. 5 (PDF p. 9), paragraph 4: > “out of 34 workers from other buildings, 14 visited the department’s > premises frequently, and 11 of them (78.5%) became ill with hemorrhagic > fever. The remaining 20 workers visited the department from 1-3 times > during this period and 6 of them (30%) became ill.”
| Group | n | Cases | AR | E_L | E_R | Onset range |
|---|---|---|---|---|---|---|
| C-1 | 14 | 11 | 78.5% | Oct 7 | Nov 1 | Oct 25 – Nov 1 |
IC likelihood: ⚠️ Applicable in principle, but
structurally weak. Formula: T_lower = O_day − Nov 1 − 1,
T_upper = O_day − Oct 7 + 1
Since E_R = Nov 1 and the reported onset range is also Oct 25 – Nov 1, T_lower = O_day − Nov 1 − 1 ≤ −1 for all known onsets → T_lower ≤ 0 for every case in this group. All C-1 cases provide only an upper bound on incubation (P(T < T_upper)), carrying minimal information about the lower tail of the distribution.
| Onset example | T_lower | T_upper | Info type |
|---|---|---|---|
| Oct 25 | −8 → ~0 | 19 | Upper bound only |
| Nov 1 | −1 → ~0 | 26 | Upper bound only |
Current status: ❌ NOT INCLUDED — individual onset dates unavailable; additionally, all cases yield T_lower ≤ 0 (upper bounds only)
| Group | n | Cases | AR | E_L | E_R |
|---|---|---|---|---|---|
| C-2 | 20 | 6 | 30% | Oct 7 | Nov 1 |
IC likelihood: ❌ Doubly limited. (1) Same upper-bound-only problem as C-1 (onset ≤ E_R for most cases). (2) “1–3 visits” means E is not a continuous window but a set of discrete unknown visit dates — the shared [Oct 7, Nov 1] window overestimates E_R per person; true individual E_L/E_R require knowing specific visit dates.
Current status: ❌ NOT INCLUDED — individual onset + visit dates unavailable; IC formula is further confounded by multiple unknown visit dates
PDF source (D-1 and D-2): Paper p. 6 (PDF p. 10), paragraph 1: > “Out of 94 persons who visited the department by chance, 44 became ill. > Out of 13 persons who visited the department often, 10 became ill, and > out of 81 persons who visited the department from 1-3 times, 34 became ill.”
| Group | n | Cases | AR | E_R |
|---|---|---|---|---|
| D-1 | 13 | 10 | 76.9% | Nov 1 |
IC likelihood: ❌ E_L unknown per person; E_R ≈ Nov 1 but individual visit patterns not reported. Cannot compute meaningful T_lower/T_upper without knowing when each person first visited.
Current status: ❌ NOT INCLUDED — individual visit dates unavailable; E_L undefined per person
| Group | n | Cases | AR | E_R |
|---|---|---|---|---|
| D-2 | 81 | 34 | 42% | Nov 1 |
IC likelihood: ❌ Same problems as D-1, compounded by larger n with diverse unknown visit patterns. Some of these individuals may overlap with the “precisely known” 6–14 cases referenced in the paper text, but they cannot be identified without the original records.
Current status: ❌ NOT INCLUDED — individual visit dates unavailable
PDF source: Paper p. 6 (PDF p. 10), paragraph 2: > “a group of persons who remained in the department for 3-4 hours only > on the 24th of October. Out of 24 persons in this group, 18 became ill.”
| Group | n | Cases | AR | E_L | E_R | k range |
|---|---|---|---|---|---|---|
| D-3 | 24 | 18 | 75% | Oct 24 | Oct 24 | 17–27 days |
IC likelihood: ✅ Fully applicable — E_L = E_R = Oct 24, same structure as Group A. Group A IS these 18 cases; the 6 non-cases contribute nothing to incubation estimation.
Current status: ✅ INCLUDED via Group A (D-3 cases = Group A cases)
PDF source: Paper p. 6 (PDF p. 10), paragraph 1: > “cases have been established of single momentary (not more than 5 minutes) > visits to the department after which the disease followed”
| Group | Cases | Notes |
|---|---|---|
| D-4 | unknown | Count not reported |
IC likelihood: ⚠️ Single-visit structure is ideal (E_L = E_R = visit day), but count of cases, visit dates, and onset dates are entirely unreported.
Current status: ❌ NOT INCLUDED — count and individual dates unknown
PDF source: Paper p. 7 (PDF p. 11), paragraph 3: > “Out of 113 persons taken ill, only 5 had first hand and direct contact > with these particular rodents, and out of them 3 became ill. Of the 2 > who didn’t become ill, one was a zoologist who, for many years, worked > with wild rodents in localities with natural foci of infection, and the > other had hemorrhagic fever in 1960.”
| Group | n | Cases | AR | E_L | E_R | Notes |
|---|---|---|---|---|---|---|
| E-1 | 5 | 3 | 60% | Sep 24 | Oct 7 | 2 non-cases: presumed immune |
Rodents arrived Sep 24 (expedition return); transferred to animal rooms Oct 7. E_L = Sep 24 (quarantine contact begins), E_R = Oct 7 (rodents moved).
IC likelihood: ⚠️ Applicable in principle with a
13-day exposure window. Formula:
T_lower = O_day − Oct 7 − 1,
T_upper = O_day − Sep 24 + 1
| Onset example | T_lower | T_upper | Width | Info type |
|---|---|---|---|---|
| Oct 25 (plausible early) | 17 | 32 | 15 | Interval (moderate) |
| Nov 5 | 28 | 43 | 15 | Interval (moderate) |
| Nov 15 | 38 | 53 | 15 | Interval (broader) |
All plausible onset dates yield T_lower > 0. The 15-day interval width is wider than Group A but narrower than Groups B/C. With 3 cases, the contribution would be small but methodologically legitimate.
Current status: ❌ NOT INCLUDED — individual onset dates unreported
PDF source: Paper p. 9 (PDF p. 13), paragraph 4: > “Epidemiological investigation showed that there was no hemorrhagic fever > among those who had contact with patients of the disease. This ruled out > the contagiosity of the infection.”
| Group | Cases | Conclusion |
|---|---|---|
| F-1 | 0 | Human-to-human transmission fully excluded |
IC likelihood: N/A — no cases.
Current status: N/A (no cases)
Paper p. 8 (PDF p. 12), paragraph 2: > “with accurately established dates for the incubation period of 24 men, > the shortest duration was 14 days and the longest 32 days > (an average of 20.4 days)”
Conclusion 3, paper p. 10 (PDF p. 14): > “In 32 cases it was possible to establish accurately the duration of the > incubation period, which consisted on the average of 20.4 days with a > variation of 14-32 days.”
Note: text body says 24 men; Conclusion 3 says 32 cases — inconsistency within the original paper. Group A accounts for 18 of these “precisely known” cases. The remaining 6–14 are other single-visit persons whose individual onset data are not in the current dataset.
| Group | n cases | IC applicable? | T_lower formula | T_upper formula | Interval width | T_lower > 0? | Info quality | Currently included? | Blocking factor |
|---|---|---|---|---|---|---|---|---|---|
| A | 18 | ✅ Yes | k − 1 | k + 1 | 2 days (fixed) | Always (min 16) | High | ✅ Yes (v3b, v4) | — |
| Other precise (~6–14) | ~6–14 | ✅ Yes | k − 1 | k + 1 | 2 days | Yes | High | ❌ No | Individual onset dates not in dataset |
| B-1 | 20 | ⚠️ Partial | O − Oct 25 − 1 | O − Oct 7 + 1 | 20 days | Only if onset > Oct 25 | Low–moderate | ❌ No | Individual onset dates unavailable |
| B-2 | 10 | ⚠️ Partial | O − Oct 25 − 1 | O − Oct 7 + 1 | 20 days | Likely for most | Low–moderate | ❌ No | Individual onset dates unavailable |
| B-3 | 0 | N/A | — | — | — | — | N/A | N/A | No cases |
| C-1 | 11 | ⚠️ Structurally weak | O − Nov 1 − 1 | O − Oct 7 + 1 | 26 days | Never (onset ≤ E_R) | Low (upper bound only) | ❌ No | T_lower ≤ 0 for all cases; onset dates unavailable |
| C-2 | 6 | ❌ Confounded | Unknown per person | Unknown per person | — | Unknown | Very low | ❌ No | Multiple unknown visit dates; onset unavailable |
| D-1 | 10 | ❌ Infeasible | Undefined | Undefined | — | — | None | ❌ No | E_L undefined per person |
| D-2 | 34 | ❌ Infeasible | Undefined | Undefined | — | — | None | ❌ No | E_L undefined per person |
| D-3 | 18 | ✅ Yes | k − 1 | k + 1 | 2 days | Always | High | ✅ Yes (= Group A) | — |
| D-4 | unknown | ⚠️ In principle | O − visit − 1 | O − visit + 1 | 2 days | Yes | High | ❌ No | Count and dates unknown |
| E-1 | 3 | ⚠️ Moderate | O − Oct 7 − 1 | O − Sep 24 + 1 | 15 days | Always (onset >> Oct 7) | Moderate | ❌ No | Individual onset dates unreported |
| F-1 | 0 | N/A | — | — | — | — | N/A | N/A | No cases |
Other precise single-visit cases (~6–14): same model as Group A, highest information density — direct drop-in to v4 with no code changes.
Group E-1 (n=3): 15-day window, all T_lower > 0, moderate information. Small but legitimate contribution; exposure dates well-defined.
Group B-1 late-onset cases: 20-day window; only those with onset after Oct 25 contribute two-sided intervals. Worth stratifying.
Group B-2: similar to B-1 but weaker signal due to lower attack rate and likely later onsets with wider windows.
Groups C-1, C-2, D-1, D-2: not recommended — upper-bound-only contribution or undefined individual exposure windows.