Volume 12, Issue 4 , Pages 374-379, July 2008
Clinical and laboratory features of Crimean-Congo hemorrhagic fever: predictors of fatality
Article Outline
Summary
Objective
To determine the predictors of fatality among patients with Crimean-Congo hemorrhagic fever (CCHF) based on epidemiological, clinical, and laboratory findings.
Methods
Among the patients with possible CCHF who were referred to Ankara Numune Education and Research Hospital (ANERH) from the surrounding hospitals between 2003 and 2006, those with IgM antibodies and/or reverse transcriptase-polymerase chain reaction (RT-PCR) results positive for CCHF virus in their blood, and who had received only supportive treatment, were included in the study.
Results
Sixty-nine patients with CCHF were admitted to ANERH from various cities of the northeastern part of the central region and southern parts of the Black Sea region of Turkey. Eleven (15.9%) patients died. Age, gender, days from the appearance of symptoms to admission, and initial complaints except bleeding were similar between fatal and non-fatal cases (p
>0.05). Among the clinical findings, ecchymosis (p=0.007), hematemesis (p=0.030), melena (p<0.001), somnolence (p<0.001), and gingival bleeding (p=0.044) were more common among fatal cases. The mean platelet count was 47.569×109/l in non-fatal cases and 12.636×109/l in fatal cases (p=0.003). Among the fatal cases, the mean prothrombin time (PT; 18.4s vs. 13.4s; p<0.001) and the mean activated partial thromboplastin time (aPTT; 69.4s vs. 42.7s; p=0.001) were longer, and the mean alanine aminotransferase (ALT; 1688 vs. 293; p<0.001), mean aspartate aminotransferase (AST; 3028 vs. 634; p<0.001), mean lactate dehydrogenase (LDH; 4245 vs. 1141; p<0.001), mean creatine phosphokinase (CPK; 3016 vs. 851; p=0.004) levels and the mean international normalized ratio (INR; 1.38 vs. 1.1; p<0.001) were higher. In a Cox proportional hazards model, thrombocytopenia of ≤20×109/l (hazard rate (HR) 9.67; 95% confidence interval (CI) 1.16–80.68; p=0.036), a prolonged aPTT ≥60s (HR 11.62; 95% CI 2.40–56.27; p=0.002), existence of melena (HR 6.39; 95% CI 1.64–24.93; p=0.008), and somnolence (HR 6.30; 95% CI 1.80–22.09; p=0.004) were independently associated with mortality.
Conclusions
Thrombocytopenia of ≤20×109/l, a prolonged aPTT ≥60s, the existence of melena, and somnolence were independent predictors of fatality.
Keywords: Crimean-Congo hemorrhagic fever, Fatality
Introduction
Crimean-Congo hemorrhagic fever (CCHF) is caused by a Nairovirus belonging to the genus Bunyavirus of the family Bunyaviridae, and transmitted to humans by Hyalomma ticks or by direct contact with the blood of infected humans or domestic animals.1, 2 The most common clinical signs of CCHF are fever, nausea, headache, diarrhea, myalgia, petechial rash, and bleeding.3
At present, CCHF is a public health problem in many regions of the world including Asia, Eastern Europe, Africa, and Russia.4 CCHF is a severe disease in humans with a fatality rate of up to 80%, most deaths occurring 5 to 14 days after the onset of illness.3, 4, 5 Since 2002, a rapid emergence of CCHF has occurred in the central, northern, and eastern regions of Turkey.6, 7, 8 By the end of 2006, there had been 1103 confirmed CCHF cases and 59 (5.4%) deaths in Turkey.9
This study was undertaken to determine the predictors of fatality among patients with CCHF based on epidemiological, clinical, and laboratory findings.
Methods
This prospective study was carried out at Ankara Numune Education and Research Hospital (ANERH), a referral and tertiary-care hospital in Turkey. Several patients with possible CCHF were referred to ANERH from the surrounding hospitals during the spring and summer seasons of the years 2003 to 2006. Only patients with a definitive diagnosis of CCHF by means of clinical presentation plus the presence of specific IgM antibody and/or detection of viral RNA by reverse transcriptase-polymerase chain reaction (RT-PCR), and patients who had received only supportive treatment, were included in the study. Biochemical and hematological laboratory parameters were measured on a daily basis after admission to the hospital. Data were recorded prospectively on individual forms.
Patients were given preparations of erythrocytes, platelets, and fresh frozen plasma depending on their homeostatic state. Patients who received antiviral therapy were excluded from the study as antiviral therapy might have changed the course of the disease.
Statistical analysis
The Student's t-test was used for independent and paired continuous variables, and proportion comparisons for categorical variables were done using Chi-square tests, although Fisher's exact test was used when data were sparse. A p-value of <0.05 was considered statistically significant. Survival curves with a 95% confidence interval were computed using the Kaplan–Meier method. Cox regression was used to model outcomes. The time start-point was the onset of complaints and the time end-point was either death or discharge from the hospital. For multivariate analysis, only variables with a p value <0.05 were entered into a Cox proportional hazards model and selected using a stepwise selection procedure. Hazard ratios (HR) and 95% confidence intervals (95% CI) were computed from estimated parameters of the final regression model. Software package Stata 9.0 (College station, Texas, USA) was used for the analysis.
Results
Sixty-nine patients were enrolled in the study. These patients were from various cities of the northeastern part of the central region and southern parts of the Black Sea region of Turkey. Their mean age was 50 years, and 45 patients (65.2%) were male. Fifty-four (78.3%) patients were involved in farming/handling livestock. A history of tick bite was present for 37 (53.6%) patients.
The patients showed hemorrhagic signs including epistaxis (26.1%), petechiae (20.3%), ecchymosis (17.4%), melena (17.4%), gingival bleeding (15.9%), hematemesis (13.0%), hematuria (5.8%), and hematoma (2.9%). Other prominent clinical signs were fever, anorexia, myalgia, and headache. Almost all of the patients had leukopenia, thrombocytopenia, and elevated aspartate aminotransferase (AST), alanine aminotransferase (ALT), lactate dehydrogenase (LDH), and creatine phosphokinase (CPK) levels at the time of admission.
Forty patients were IgM positive and 50 patients were RT-PCR positive. Seven (63.6%) out of the 11 fatal cases were only RT-PCR positive.
The case–fatality rate was 15.9%. Fatal cases were hospitalized for a mean of 3.8 days (range 1–9) before death. The duration of hospitalization was approximately 8 days (range 2–19) in non-fatal cases. The disease started a mean of 6.3 days (range 2–10) before hospitalization in fatal cases and 5.8 days (range 1–15) in non-fatal cases.
Age, gender, number of days from the appearance of symptoms to admission, and initial complaints except bleeding were similar between fatal and non-fatal cases (p>0.05). Among the clinical findings, ecchymosis (p=0.007), hematemesis (p=0.030), melena (p<0.001), somnolence (p<0.001), and gingival bleeding (p=0.044) were more common among fatal cases (Table 1).
Table 1. Univariate analysis of demographic and clinical variables for non-fatal and fatal cases with Crimean-Congo hemorrhagic fever
| Characteristic | Non-fatal cases (N=58) | Fatal cases (N=11) | p Value |
|---|---|---|---|
| Age, years | 49.1±16.7 | 53.5±17.4 | 0.435 |
| Female sex | 19 (32.8%) | 5 (45.5%) | 0.418 |
| Duration of complaints until hospitalization, days | 5.8±2.7 | 6.3±2.6 | 0.574 |
| Most common complaints | |||
| Myalgia | 50 (86.2%) | 10 (90.9%) | 0.560 |
| Fever | 49 (84.5%) | 10 (90.9%) | 0.315 |
| Lack of appetite | 47 (81.0%) | 10 (90.9%) | 0.385 |
| Headache | 39 (67.2%) | 6 (54.5%) | 0.418 |
| Nausea and/or vomiting | 29 (50%) | 6 (54.5%) | 0.782 |
| Bleeding (any kind) | 16 (27.6%) | 9 (81.8%) | 0.001 |
| Diarrhea | 11 (19.0%) | 3 (27.3%) | 0.393 |
| Cough | 9 (15.5%) | 1 (9.1%) | 0.498 |
| Clinical findings | |||
| Fever, temperature >38°C | 29 (50%) | 6 (54.5%) | 0.782 |
| Hepatomegaly | 10 (17.2%) | 3 (27.3%) | 0.341 |
| Splenomegaly | 6 (10.3%) | 1 (9.1%) | 0.691 |
| Jaundice | 3 (5.2%) | 1 (9.1%) | 0.509 |
| Rash | |||
| Maculopapular | 3 (5.2%) | 2 (18.2%) | 0.177 |
| Petechiae | 10 (17.2%) | 4 (36.4%) | 0.150 |
| Bleeding | |||
| Ecchymosis | 7 (12.1%) | 5 (45.5%) | 0.007 |
| Hematoma | 0 | 2 (18.2%) | – |
| Epistaxis | 14 (24.1%) | 4 (36.4%) | 0.308 |
| Hematemesis | 5 (8.6%) | 4 (36.4%) | 0.030 |
| Melena | 4 (6.9%) | 8 (72.7%) | <0.001 |
| Hematuria | 2 (3.4%) | 2 (18.2%) | 0.117 |
| Gingival | 7 (12.1%) | 4 (36.4%) | 0.044 |
| Vaginal | 2 (3.4%) | 0 | – |
| Somnolence | 2 (3.4%) | 6 (54.5%) | <0.001 |
In fatal cases, the mean heart rate was 89 bpm (range 80–112) and in non-fatal cases 86 bpm (range 62–120) (p=0.483). Blood pressure was normal in all patients at the time of admission.
The mean platelet count was 47.569×109/l in non-fatal cases and 12.636×109/l in fatal cases (p=0.003). Among the fatal cases the mean prothrombin time (PT; 18.4s vs. 13.4s; p<0.001) and mean activated partial thromboplastin time (aPTT; 69.4s vs. 42.7s; p=0.001) were longer, and the mean ALT (1688 vs. 293; p<0.001), mean AST (3028 vs. 634; p<0.001), mean LDH (4245 vs. 1141; p<0.001), mean CPK (3016 vs. 851; p=0.004) levels and the mean international normalized ratio (INR; 1.38 vs. 1.1; p<0.001) were higher (Table 2).
Table 2. Univariate analysis of laboratory findings for non-fatal and fatal cases with Crimean-Congo hemorrhagic fever
| Laboratory finding | Non-fatal cases Mean value (range) | Fatal cases Mean value (range) | p Value |
|---|---|---|---|
| Lowest platelet count (platelets×109/l) | 47.569 (4.00–183.00) | 12.636 (4.00–34.00) | 0.003 |
| Lowest WBC count (WBC×109/l) | 2.084 (0.70–6.00) | 2.494 (1.10–6.52) | 0.328 |
| Highest AST level, U/l | 634 (17–4202) | 3028 (278–11870) | <0.001 |
| Highest ALT level, U/l | 293 (18–1278) | 1688 (193–7345) | <0.001 |
| Highest LDH level, U/l | 1141 (257–9237) | 4245 (473–7866) | <0.001 |
| Highest CPK level, U/l | 851 (150–7150) | 3016 (216–16372) | 0.004 |
| Longest PT, s | 13.4 (10.2–18.1) | 18.4 (13.1–39.8) | <0.001 |
| Longest aPTT, s | 42.7 (25.8–154) | 69.4 (42.8–85.9) | 0.001 |
| Highest INR | 1.1 (0.8–1.6) | 1.38 (1–1.9) | <0.001 |
| Lowest fibrinogen, mg/dl | 240 (115–389) | 198 (108–424) | 0.121 |
All patients received intensive clinical support, including platelets, fresh frozen plasma, and packed erythrocyte infusions when indicated. None of the patients received antiviral therapy. Fatal cases received significantly more fresh frozen plasma (14 vs. 2 packs, p<0.001) and platelet suspensions (26 vs. 8 packs, p=0.017) than non-fatal cases.
The Cox proportional hazards model was used for fatality analysis. The time start-point was the onset of complaints and time end-point was either death or discharge from the hospital. Thrombocytopenia of ≤20×109/l (hazard rate (HR) 9.67; 95% confidence interval (CI) 1.16–80.68; p=0.036), a prolonged aPTT ≥60s (HR 11.62; 95% CI 2.40–56.27; p=0.002), the existence of melena (HR 6.39; 95% CI 1.64–24.93; p=0.008), and somnolence (HR 6.30; 95% CI 1.80–22.09; p=0.004) were independently associated with mortality (Table 3, Figure 1, Figure 2, Figure 3, Figure 4).
Table 3. Estimated regression coefficient, standard error, p value, and hazard ratio as a function of the risks of the variables according to the Cox proportional hazards model
| Variables | Coefficient | Standard error | p Value | Hazard ratio (95% CI) |
|---|---|---|---|---|
| Activated partial thromboplastin time ≥60s | 3.05 | 9.35 | 0.002 | 11.62 (2.40–56.27) |
| Platelet count ≤20×109/l | 2.10 | 10.46 | 0.036 | 9.67 (1.16–80.68) |
| Melena | 2.67 | 4.44 | 0.008 | 6.39 (1.64–24.93) |
| Somnolence | 2.87 | 4.03 | 0.004 | 6.30 (1.80–22.09) |
Figure 1.
Kaplan–Meier survival estimates by somnolence, time from onset of complaints to death or discharge.
Figure 2.
Kaplan–Meier survival estimates by melena, time from onset of complaints to death or discharge.
Figure 3.
Kaplan–Meier survival estimates by platelet count, time from onset of complaints to death or discharge.
Figure 4.
Kaplan–Meier survival estimates by activated partial thromboplastin time, time from onset of complaints to death or discharge.
Discussion
CCHF virus causes a fatal hemorrhagic syndrome, which has been endemic in the central, northern, and eastern regions of Turkey in recent years. CCHF is an acute and generally self-limiting disease. Clinically, CCHF results in the most severe bleeding and ecchymoses among the hemorrhagic fevers. Hemorrhagic manifestations include gingival bleeding and epistaxis followed by gastrointestinal hemorrhage. Ecchymoses are often large and pressure-linked.10 The clinical presentation of patients with CCHF in this study was similar to that reported previously.10, 11 On admission, the most frequently detected symptoms were fever, myalgia, lack of appetite, headache, nausea/vomiting, and bleeding. In this study, 58% of patients had bleeding symptoms. Epistaxis (26.1%) was the most frequently observed hemorrhagic manifestation followed by petechiae (20.3%), ecchymosis (17.4%), melena (17.4%), gingival bleeding (15.9%), hematemesis (13.0%), hematuria (5.8%), and hematoma (2.9%). Although it has been reported that cardiovascular changes including bradycardia and low blood pressure can be seen,12 none of our patients had signs of cardiovascular abnormalities.
Various fatality rates for CCHF, of up to 80%, have been reported in the literature.1, 3, 11, 13 In this study, a fatality rate of 15.9% was obtained, which is higher than the average for Turkey (5.4%) and previously reported fatality rates.8, 13, 14 The fatality rate reported by the Ministry of Health of Turkey includes all CCHF cases including the ones treated in primary- and secondary-care services. As ANERH is a referral and tertiary-care hospital, more complicated and severe cases are transferred here from the surrounding hospitals; hence, the high fatality rate obtained in this study is probably due to the admission of severe cases. On the other hand, high death rates reported in some series in contrast to the lower fatality rates in our patients can be explained by the wide availability of blood products at our institution compared to other facilities, and also to better supportive care of our patients.
Deaths are reported to occur on days 5–14 of CCHF.15 In this study, the mean duration from the onset of complaints until death was 10.2 days (range 3–19 days).
In fatal cases there was little evidence of an antibody response.16 In seven out of 11 fatal cases no serum IgM or IgG was detected although the virus was identified by PCR. The absence of a significant antibody response in fatalities has been reported previously.17
Any of the following clinical pathologic values during the first 5 days of illness were found to be >90% predictive of fatal outcome in a series of South African CCHF patients: leukocyte counts<10×109/l, platelet counts <20×109/l, AST >200U/l, ALT >150U/l, aPTT >60s, and fibrinogen <110mg/dl.17 Other case series have confirmed that levels of AST and ALT are significantly higher among severe cases (p < 0.05).8, 13, 14 In a study from Turkey, higher AST and ALT levels (>700 and >900IU/l, respectively) were found to have higher sensitivity for severe cases.14
Previously defined severity criteria did not completely fit the patients in the present study (Table 4).14, 17 Leukocytosis was one of the fatality criteria described by Swanepoel et al.,17 but this was only observed in one fatal case and two non-fatal cases in our study. Decreased fibrinogen of ≤110mg/dl was another previously described fatality criterion that was observed only in one patient in our study. In contrast to Swanepoel et al.,17 we observed no renal failure. An aPTT of >60s (HR 11.62; 95% CI 2.40–56.27; p=0.002) was found to be an independent predictor of fatality in the present study. The mean platelet count was 47.569×109/l in non-fatal cases and 12.636×109/l in fatal cases (p=0.003). We found thrombocytopenia of ≤20×109/l (HR 9.67; 95% CI 1.16–80.68; p=0.036) to be a strong independent predictor of fatality. Among the fatal cases, the mean ALT (1688 vs. 293; p<0.001) and the mean AST (3028 vs. 634; p<0.001) levels were higher. Most of our patients had at least one or more risk factors as described above; however, the overall death rate was 15.9%. Further observations with larger numbers of patients are necessary to clarify these inconsistencies.
Table 4. Comparison of predictors of fatality with previously reported criteria (data extracted from univariate analysis)
| Criteria | Swanepoel et al.17 | Ergonul et al.14 | Present study | p Value |
|---|---|---|---|---|
| Increased WBC count, ≥10×109/l | Yes | No | No | 0.400 |
| Decreased platelet count, ≤20×109/l | Yes | Yes | Yes | 0.035 |
| Elevated AST level | Yes | Yes | Yes | <0.001 |
| Elevated ALT level | Yes | Yes | Yes | <0.001 |
| Elevated LDH level | No | No | Yes | <0.001 |
| Elevated CPK level | No | No | Yes | 0.004 |
| Prolonged aPTT | Yes | Yes | Yes | 0.001 |
| Prolonged PT | No | Yes | Yes | <0.001 |
| Decreased fibrinogen, ≤110mg/dl | Yes | Yes | No | 0.121 |
| Ecchymosis | No | No | Yes | 0.007 |
| Melena | No | Yes | Yes | <0.001 |
| Hematemesis | No | Yes | Yes | 0.030 |
| Somnolence | No | Yes | Yes | <0.001 |
Bakir et al. reported that the INR, AST, LDH, and CPK levels were higher in patients with a fatal outcome.8 In this study fatal cases had higher AST, LDH and CPK levels, and a higher INR.
Among the clinical findings, melena and hematemesis were more commonly reported among fatal cases.14 In this study all of the fatal cases had hematemesis and/or melena. Additionally, the risk of fatality was higher in patients with melena (HR 6.39; 95% CI 1.64–24.93; p=0.008).
An altered mental status (such as unconsciousness, confusion, and somnolence) has previously been reported as an indicator of poor prognosis.8, 13, 14 In the present study, the existence of somnolence (HR 6.30; 95% CI 1.80–22.09; p=0.004) showed a strong relationship with the severity of illness and subsequent probability of death.
Treatment options for CCHF are limited. Supportive therapy is the most essential part of case management and includes intensive monitoring to guide volume and administration of platelets, fresh frozen plasma, and erythrocyte preparations.12, 13 Currently, there is no specific antiviral therapy for CCHF approved for use in humans. However, some reports have been published that suggest that oral or intravenous ribavirin is effective for treating CCHF infections.6, 18, 19 In another study from Turkey, it was highlighted that oral ribavirin treatment did not decrease the need for blood and its products and did not have any effect on reducing the fatality rates.13 It is important to mention that there have been no randomized or controlled studies to confirm the effectiveness of ribavirin against CCHF to date. Therefore, the efficacy of ribavirin in the treatment of CCHF is not clear. Immunotherapy has been attempted via passive transfer of CCHF survivor convalescent plasma,20 but its value is also not clear. There is an urgent need to establish effective treatment(s) for CCHF.
To conclude, from the results of our study we report that the most important risk factors for fatality in CCHF are the existence of somnolence and melena, a prolonged aPTT (≥60s), and a decreased platelet count (≤20×109/l). We suggest that such patients should be observed more carefully for intensive supportive therapy.
Conflict of interest: No conflict of interest to declare.
References
- . Measures undertaken in the German Armed Forces Field Hospital deployed in Kosovo to contain a potential outbreak of Crimean-Congo hemorrhagic fever. Mil Med. 2005;170:366–369
- . Crimean-Congo hemorrhagic fever in Kosovo. J Clin Microbiol. 2002;40:1122–1123
- Tick-borne virus diseases of human interest in Europe. Clin Microbiol Infect. 2004;10:1040–1055
- Human MxA protein inhibits the replication of Crimean-Congo hemorrhagic fever virus. J Virol. 2004;78:4323–4329
- . The use of a reverse transcription-polymerase chain reaction for the detection of viral nucleic acid in the diagnosis of Crimean-Congo hemorrhagic fever. J Virol Methods. 1998;70:129–137
- . The characteristics of Crimean-Congo hemorrhagic fever in a recent outbreak in Turkey and the impact of oral ribavirin therapy. Clin Infect Dis. 2004;39:285–289
- Crimean-Congo hemorrhagic fever in Turkey. Emerg Infect Dis. 2004;19:1379–1384
- . Crimean-Congo hemorrhagic fever outbreak in Middle Anatolia: a multicenter study of clinical features and outcome measures. J Med Microbiol. 2005;54:1–5
- . Reports of the Communicable Diseases Department. Ankara, Turkey: Ministry of Health; 2006;
- . Viral disease in haematology. Hematology (Am Soc Hematol Educ Program). 2000;409–423
- . Clinical features of Crimean-Congo haemorrhagic fever in the United Arab Emirates. Infection. 1997;25:364–367
- . Crimean-Congo hemorrhagic fever. Antivir Res. 2004;64:144–160
- Crimean-Congo hemorrhagic fever in Eastern Turkey: clinical features, risk factors and efficacy of ribavirin therapy. J Infect. 2006;52:207–215
- . Analysis of risk-factors among patients with Crimean-Congo haemorrhagic fever virus infection: severity criteria revisited. Clin Microbiol Infect. 2006;12:551–554
- Short report: Crimean-Congo hemorrhagic fever outbreak in Rawalpindi, Pakistan, February 2002. Am J Trop Med Hyg. 2003;69:284–287
- . Antibody response in Crimean–Congo hemorrhagic fever. Rev Infect Dis. 1988;11(Suppl 4):S801–S806
- . The clinical pathology of Crimean-Congo hemorrhagic fever. Rev Infect Dis. 1989;11(Suppl 4):S794–S800
- A patient with Crimean–Congo hemorrhagic fever serologically diagnosed by recombinant nucleoprotein-based antibody detection systems. Clin Diagn Lab Immunol. 2003;10:489–491
- . The efficacy of oral ribavirin in the treatment of Crimean–Congo hemorrhagic fever in Iran. Clin Infect Dis. 2003;36:1613–1618
- . Specific intravenous immunoglobulin for Crimean–Congo haemorrhagic fever. Lancet. 1990;335:791–792
PII: S1201-9712(07)00199-3
doi:10.1016/j.ijid.2007.09.010
© 2007 International Society for Infectious Diseases. Published by Elsevier Inc. All rights reserved.
Volume 12, Issue 4 , Pages 374-379, July 2008
