Severe Scrub Typhus: Prognostic Indicators, Clinical Prediction Rule and Validation

Abstract

Scrub typhus, widely spread throughout the Asia-Pacific region is an acute febrile illness, produced by Orientia tsutsugamushi. In endemic areas the approximate prevalence of the disease is 1 million cases yearly. The reported mortality is as high as 35%. In Thailand, 7,310 cases in 2011 and as many as 9,000 cases in 2012 have been reported, with a mortality rate of 2.6 to 30%, with the highest prevalence in the north. In this region, as much as 15% of patients die albeit timely antibiotic treatment greatly decreases mortality. Patients can develop severe symptoms, experience complications and finally die. Organ involvement, e.g., the central nervous, cardiovascular, respiratory, hepatic and renal systems or multiple organ involvement occurs in 2 to 36% or mortality cases. These complications along with late diagnosis and delayed treatment frequently end in death. Predictive signs of the severity of the disease could aid in identifying and prompting initial therapy, and decrease mortality. This study aimed to investigate clinical indicators to predict the severity of the disease in Chiang Rai Prachanukroh Hospital, Chiang Rai and Nakornping Hospital, Chiang Mai, Thailand, between 2004 and 2010. The clinical indicators of severity included rising pulse rate, urine albumin, aspartate aminotransferase (AST) and creatinine levels. Related indicators also included lowered body temperature, percentage of lymphocytes and serum albumin level and crepitation. The results can be applied to improve clinical assessment, create greater awareness, and prevent complications, leading to a reduced mortality rate. Even though these indicators of scrub typhus have been reported from several areas, stud¬ies have yet to create clinical predictors of scrub typhus severity. This study planned a simple clinical algorithm to score risk established on regular clinical factors to forecast scrub typhus severity when suspected. A total of 6 significant severity indicators were found. These included crepitation, a pulse rate >100/min, an AST level >160 IU/L, a serum creatinine level >1.4 mg/dl, a serum albumin level ≤3.0 g/dl and age >15 years. The item scores varied from the minimum of 0 to the maximum of 4. Each item scores were combined as total scrub typhus severity score which may vary from 0 to 16. The scores accorded these indicators can be applied to classify patients into 3 risk levels; fatal (scores ≥10) which experienced the highest risk of death and should be admitted to an intensive care unit and fully investigated for system abnormalities or other life threatening clinical risks, severe (scores 6-9) which may experience higher risk for complications and should be admitted to hospital for close observation, and non severe (scores ≤5) which may be managed as out-patients. The scores classified patients correctly 68.3%. Discrimination performance of these scores among patients diagnosed with fatal and severe scrub typhus, from patients diagnosed as non severe held a high level of validity presenting a receiver operating characteristic curve (AuROC) equal to 91.48%. The associated value to discriminate among these three groups was inferior, i.e., AuROC equal to 91.22%. These scores can be applied in routine clinical practice to guide and advance patient treatment. These scores must be validated before being applied in routine clinical practice. This risk-scoring system was validated by an independent database of scrub typhus cases at Chiang Rai Prachanukroh Hospital between 2011 and 2012. The general validity of the three levels of severity was less than the development data, i.e., 61.0% versus 68.3%. The general predictive performance of this risk scoring system was inferior, both in differentiating fatal versus non severe and severe, i.e., AuROC=88.66% versus 91.22% as well as non severe from severe and fatal, i.e., AuROC=88.74% versus 91.48%) On the other hand, this difference was not significant, neither clinically or statistically. The performance of validation data would be inferior to development data. By dividing patients into three risk levels; non-severe, severe, and dead, final management for those who were in severe or dead were hospitalization whereas non-severe patients can be treated as out-patient department. Therefore, additional data analysis was carried by combining cases with severe scrub typhus and dead into the severe group and the remaining of the cases into non-severe group. Then, we developed a clinical prediction rule to simplify for adoption as a screening guide and facilitate decision for patients’ admission or refer them to a provincial hospital. Five predictors in the final model were age >60, crepitation, AST>160 IU/L, albumin ≤3.0 g/dl, creatinine>1.6 mg/dl. The item scores varied from 0-3. Each item scores were summed up to total scrub typhus severity score which may vary from 0-9. For application, patients who were scored 0-2.5 points have low risk of severe scrub typhus and can be treated as outpatient department, dispense with antiricketsial, and appointed for follow up. On the other hand, patients with 3-9 pointes were classified in a severe group; they should be hospitalized for close monitoring to prevent severe symptoms and mortality or transfer to higher potential hospital. Accuracy to predict severity was 84.2%, AuROC 84.8%. When apply developed score scheme in validation data, the ability to predict severity was good; 78.5 % from the AuROC curve, but lower than the development data. This score system has an accuracy of 78.4% which is less than development dataset (84.2%). The benefit of these clinical prediction guidelines includes being based on routine, available patient data, namely, demographics, laboratory tests and physical examinations. These clinical guidelines performed reliably well when classifying scrub typhus patients in the three levels of severity in a successive group of subjects. The guidelines were acceptable clinically. They were practical in routine clinical treatment to categorize subjects by severity level. In addition, to the guidelines could discriminate more intensive treatments and investigations and particularly in endemic regions with limited health resources reduce morbidity and mortality rates. Nevertheless, this thesis employed 3 studies that revealed significant constraints when applying the generalized diagnosis of scrub typhus consistent with the WHO definition. These guidelines are the most realistic classification and more related to the real world employed in less economically developed countries, where a definitive diagnosis, e.g., indirect immunofluorescence antibodies, would be completely impractical. In this research, validated information on patients suspected of having scrub typhus was collected retrospectively. Thus, validating these guidelines and confirming their feasibility in the future among patients could establish a less theoretical definition and serve to improve routine practices and treatments.