Panel A shows a flow diagram for the generation of the analysis cohort. Panel B shows a kernel-density plot of the distribution of log10 antibody concentrations among the Ministry of Health–reported survivors and the survivor-reported close contacts. The black vertical line indicates the antibody concentration cutoff for determining Ebola virus seropositivity or seronegativity. EU denotes enzyme-linked immunosorbent assay units.
From June 2015 through June 2017, three sites in Liberia enrolled 1145 of the approximately 1500 EVD survivors listed in the Ministry of Health registry and 2785 close contacts with no history of EVD. Among the survivors, the median time from the onset of EVD to the baseline protocol visit was 358 days (interquartile range, 313 to 405) (Table 1). Of the 1108 Ministry of Health–reported survivors who underwent serologic testing, 966 (87.2%) had EBOV-specific antibodies detected at levels equal to or above 548 EU per milliliter; 2350 (88.7%) of the survivor-reported close contacts with serologic test results had antibody levels below 548 EU per milliliter (Figure 1, and Section 2.1 in the Supplementary Appendix). Antibody-positive EVD survivors and antibody-negative close contacts make up the analysis cohort for this study. These groups are referred to as survivors and controls, respectively, when comparisons are made between these two groups.
Antibody-Negative Survivors and Antibody-Positive Contacts
There were 142 survivors reported by the Ministry of Health who had Ebola antibody levels below 548 EU per milliliter (median antibody level, 98 EU per milliliter) (Figure 1). Antibody-negative survivors were less likely than antibody-positive survivors to have a positive EBOV RT-PCR result reported during their acute illness (31% vs. 76%, P<0.001).
Among the 300 close contacts with EBOV antibody levels of at least 548 EU per milliliter, the median antibody level was significantly lower than that among the antibody-positive survivors (3979 EU per milliliter vs. 19,242 EU per milliliter, P<0.001). Among the antibody-positive close contacts, 47% reported having symptoms consistent with EVD within 3 weeks after their linked survivor’s acute illness, and most symptoms were reported with at least twice the frequency among the antibody-positive close contacts than among the antibody-negative close contacts (Table S1 in the Supplementary Appendix). In the antibody-negative close-contact group, 31% of participants recalled symptoms suggestive of EVD within a 3-week period after their linked survivor’s acute illness (P<0.001 for the difference) (Table S1 in the Supplementary Appendix).
Antibody-Positive Survivors and Antibody-Negative Contacts (Controls)
Both survivors and controls reported a wide range of symptoms. A complete list of the symptoms reported by survivors and controls is provided in Tables S4 and S5 in the Supplementary Appendix. Table 2 lists the prevalence of the targeted symptoms and findings. Six symptoms met the criteria for significance (i.e., P<0.0001): urinary frequency (14.7% vs. 3.4%), headache (47.6% vs. 35.6%), fatigue (18.4% vs. 6.3%), muscle pain (23.1% vs. 10.1%), memory loss (29.2% vs. 4.8%), and joint pain (47.5% vs. 17.5%) were reported significantly more often by survivors than by controls at enrollment. Child and adolescent participants in both the survivor group and the control group reported fewer symptoms than adults (Fig. S2 in the Supplementary Appendix). Among antibody-positive survivors, there was no association of antibody levels with the symptoms shown in Table 2 (see Section 2.5 in the Supplementary Appendix).
Objective abnormalities on physical examination were less common than reported symptoms. Survivors had significantly (P<0.0001) higher percentages of abnormal findings on abdominal, chest, neurologic, muscle, and joint examination than did controls (Table 2, and Table S8 in the Supplementary Appendix).
Only a small percentage of survivors had physical findings suggestive of inflammatory arthritis or myositis, with the most common abnormal musculoskeletal findings being muscle tenderness (4.1% of survivors and 0.9% of controls), decreased joint range of motion (2.7% and 1.4%), and joint swelling (0.8% and 0.5%). The prevalence of muscle tenderness on examination was significantly higher among survivors than among controls (odds ratio, 4.35; 95% confidence interval [CI], 2.58 to 7.34; P<0.001), whereas joint swelling (odds ratio, 1.46; 95% CI, 0.58 to 3.65; P=0.42) and decreased range of motion (odds ratio, 1.75; 95% CI, 1.04 to 2.93; P=0.04) did not meet the criteria for significance imposed in this study. The most common abnormalities found on neurologic examination among survivors and controls were abnormal reflexes (1.4% and 0.7%, respectively), tremor (0.9% and 0.2%), gait or balance abnormalities (0.7% and 0.9%), speech abnormalities (0.7% and 0.2%), and cranial nerve abnormalities (0.7% and 0.1%). On abdominal examination, the most common abnormal findings noted among survivors and controls were tenderness (5.6% and 3.7%, respectively), mass (2.4% and 2.0%), and distension (1.9% and 1.4%). The most common abnormal chest findings on examination among survivors and controls were irregular pulse (1.0% and 0.5%, respectively), decreased breath sounds (0.9% and 0.4%), and heart murmur (0.9% and 0.3%). A complete list of the frequencies of individual abnormal findings is provided in Table S8 in the Supplementary Appendix. Evaluations of renal, hepatic, and hematopoietic laboratory measurements did not reveal any clinically relevant differences between survivors and controls at baseline (Table S9 in the Supplementary Appendix).
Clinical and Laboratory Findings at Follow-up Visits
The prevalence of all targeted symptoms decreased at the 6- and 12-month follow-up visits among both survivors and controls (Table 2); however, the odds ratio (survivor vs. control) increased over time for headache and decreased for muscle pain. To assess the rate at which targeted symptoms and findings developed after the baseline visit, data from the 6- and 12-month visits were used to calculate the incidence of targeted symptoms and findings during the first year of follow-up. The incidence of new urinary frequency symptoms and abnormal chest and abdominal findings did not differ significantly between survivors and controls. The incidence of the other targeted symptoms and findings (other than joint and muscle) was higher among survivors than among controls (Table 3), although the overall prevalence decreased after baseline. Follow-up laboratory evaluations did not reveal any emerging differences between survivors and controls (Table S9 in the Supplementary Appendix).
For participants in the longitudinal eye cohort, no significant differences in corrected visual acuity or age-adjusted prevalence of cataracts were noted between survivors and controls either at baseline or at the 12-month visit (Table 2). The median corrected visual acuity was 20/20 (interquartile range, 20/20 to 20/25) for both survivors and controls. On the basis of the World Health Organization criteria for visual impairment,19 the prevalence of moderate-to-severe visual impairment did not differ significantly between survivors and controls at baseline or at the 12-month follow-up visit (Table 2).
At baseline, 149 survivors (26.4%) had evidence of uveitis in at least one eye on ophthalmologic examination, as compared with 77 controls (12.1%) (P<0.0001). Among these participants, 30 survivors (5.3%) and 13 controls (2.0%) had active uveitis (P=0.003). The prevalence of uveitis increased among both survivors and controls between baseline and the 12-month follow-up visit (Table 2). The incidence of new uveitis was significantly higher among survivors than among controls (Table 3).
Analysis of Semen for EBOV RNA
Data points represent model-based estimates of the probability of testing positive for individual samples; vertical bars represent 95% confidence intervals for the probability of a sample being positive, based on samples grouped into 25-day bins; and the piecewise linear black curve shows the sample proportions in these bins. The red curve represents a model-based population trend for the probability of a sample testing positive for Ebola virus RNA.
A total of 267 antibody-positive male survivors (median age, 33 years; interquartile range, 26 to 41) provided 2416 semen samples with which an EBOV RNA determination was successfully conducted (see Section 1.4 in the Supplementary Appendix). Viral RNA was detected in one or more semen samples in 81 men (30%) (Table S3 in the Supplementary Appendix). Among the 252 men who provided more than one semen sample, detection of viral RNA in semen was intermittent in 78 (31%), with 36 men having two negative tests followed by a positive test at least once. The time from acute EVD illness to first semen sampling ranged from 233 to 1178 days (median, 551 days [19 months]). A significant (P<0.0001) decline in the frequency of positive results over time was observed (Figure 2). To date, the longest time from acute EVD illness to the detection of viral RNA in a semen sample is 40 months.
We found no correlation between the persistence of viral RNA in semen and the symptoms shown in Table 2 (see Section 2.6 in the Supplementary Appendix). There was a positive association between the presence of uveitis and the detection of viral RNA in at least one semen sample (odds ratio, 2.62; 95% CI, 1.32 to 5.22).