Multivariable prediction model for predicting deaths in severe dengue cases

doi:10.21203/rs.2.9960/v2

Download PDF

Research article

Multivariable prediction model for predicting deaths in severe dengue cases

https://doi.org/10.21203/rs.2.9960/v2

This work is licensed under a CC BY 4.0 License

Version 2

posted

You are reading this latest preprint version

Background. Many predictive models have been developed to predict an outbreak, identify and stratify dengue but none has predicted death in severe dengue cases. To build a predictive model for deaths in severe dengue, a multicentre retrospective cohort study was conducted.

Methods. Patients with severe dengue based on the World Health Organisation (WHO) 2009 classification were studied. Demographic, clinical and laboratory data were collected at diagnosis of severe dengue. Penalised regression was used for variable selection and model-building. Ten-fold cross-validation with 1000 repeats were performed for internal validation.

Results. A cohort of 786 severe dengue cases, including 35 deaths, was analysed. Our model that predicts death in severe dengue cases comprises eight independent predictors: persistent diarrhoea, body mass index, respiratory rate, platelet count, aspartate transaminase, serum bicarbonate, serum lactate and serum albumin. The area under the receiver operating characteristic curve is 89·6% with a sensitivity of 99·6%, specificity of 23·6%, positive predictive value of 96·6%, negative predictive value of 71·1%, positive likelihood ratio 1·45 and negative likelihood ratio 0·01. We also found that the proportion of patients that were in the febrile phase at diagnosis of severe dengue for the overall cohort, decompensated and compensated shock were 74·3%, 73% and 75·4%, respectively.

Conclusions. We developed a high-performance dengue death prediction model comprising clinical and laboratory data, and deployed an open-access web-based tool (www.saifulsafuan.com/REPROSED2017E2) for any centre to utilise for local validation. We additionally found that a large majority of patients developed severe dengue during the febrile phase.

Keywords: Dengue; severe; model; predict; deaths; phase.

Infectious Diseases

Dengue

severe

model

predict

deaths

phase

The world is burdened with an estimated 96 million dengue infection every year.¹ The World Health Organisation estimated that there are 500,000 severe dengue cases per year.² While the feared Ebola has killed 12,962 people from 1976 to 2015³, dengue is estimated to kill approximately 12,500 people each year.²

In a recent estimate of 12 countries in South East Asia, between the years 2001–2010, annual dengue cases amounted to 2·9 million cases with 5906 deaths per year.⁴ Dengue is a common and endemic in Malaysia. The number of recorded dengue cases in Malaysia has risen from 6543 to 120,836 cases between 1995 to 2015.⁵ The incidence rate had risen from 31·6/100,000 population in the year 2000 to 396·4/100,000 in 2015.⁵ Correspondingly, the number of deaths has increased from 28 in 1995 to 336 in 2015.⁵ The case fatality rate, however, remained at around 0·2% per year, for the exception of years 2011–2012.⁵ Dengue in Malaysia predominantly involves ages 15 and above.⁶

In South-East Asia, the estimated economic burden amounted to USD950 million [95% CI: USD610–1384 million] or USD1·65 [95%CI: USD1·06–2·41] per capita annually.⁴ In 2010, USD73·5 million was spent in Malaysia for its National Dengue Vector Control Program which is equivalent to USD2·68 per capita. This represents 0·03% of Malaysia’s gross domestic product.⁷ Direct medical costs and costs related to productivity loss and premature death, was USD102·2 million in 2009.^7–9

In order to prevent death due to dengue, strategies employing prediction models at various aspects of the infection have been made. Most predictive modelling studies have been conducted for the prediction of dengue outbreak.^10–16 Several studies have looked at two clinical aspects of dengue management: predicting the identification of dengue in unselected cases of febrile illnesses,^17,18 and predicting the severity of the disease.^17,19,20,21 Early recognition of dengue infection and severity stratification would improve clinical management amnd lead to a better outcome.

Predicting and estimating the probability of death is one aspect of clinical dengue management that has never been addressed. The ability to estimate the probability of death in illness will assist in therapeutic decision-making. It also allows assessment of an interventional treatment against a comparator in clinical research. An example is the TIMI score, which has been used in clinical research and decision-making in the management of acute coronary syndromes.²² Currently, there have only been three studies on predicting death in dengue.^23–25

If the prediction of an explosion in the number of dengue cases due to impending climate changes materialises,²⁶ the global community of clinicians caring for patients with severe dengue must be intelligently equipped. Identification of risk factors of death should culminate in the building of prediction models that can be deployed as software applications for clinical use at the bedside. Thus, we conducted a multicentre retrospective cohort study to build a predictive model that will predict death and estimate the probability of death in severe dengue cases. This model is an open-access, web-based ‘bedside’ prediction tool.

We conducted a multicentre retrospective cohort study with a total population sampling of patients with severe dengue who were admitted during the period 1st January 2017 until 31st December 2017. Our report is based on the Transparent Reporting of a multivariable prediction model for Individual Prognosis Or Diagnosis (TRIPOD) 2015 guideline.²⁷

The five participating centres were Hospital Raja Perempuan Zainab II, Hospital Sungai Buloh, Hospital Kuala Lumpur, Hospital Tuanku Ampuan Rahimah, and Hospital Sultanah Aminah. Patients were selected for inclusion into the cohort if, 1) they were ≥15 years old⁶, 2) the case fulfilled the study definition of severe dengue, and 3) the presence of dengue infection was confirmed via laboratory confirmation by the presence of at least one of the following: positive non-structural protein 1 (NS1) antigen, or dengue RNA by reverse transcription-polymerase chain reaction (RT-PCR), or high-titre level of immunoglobulin G (IgG), or positive immunoglobulin M (IgM) from an admission serum sample. Patients with positive IgM-only immunology were also excluded if their clinical and laboratory blood results patterns were incongruous with dengue disease course or had an alternative final diagnosis documented in their case notes. High-titre IgG detects acute dengue infection at titres >1:2560, which therefore is highly specific for dengue infection and also identifies such infection as acute secondary dengue infection. Excluded patients were those who were pregnant and those who were admitted for transfer of care from non-participating centres.

The definition of severe dengue in our study was based on World Health Organisation (WHO) 2009 definition but with adaptation for this study.²⁸ We defined severe dengue by presence of any one of the following: 1) decompensated shock due to severe plasma leakage, 2) compensated shock due to severe plasma leakage, 3) respiratory compromise due to plasma leakage, 4) severe bleeding that required intervention, or 5) severe organ involvement such as acute kidney injury defined by elevated serum creatinine above upper limit normal (according to gender-specific levels), severe hepatitis, myocarditis, or encephalopathy. Decompensated shock was defined by the presence of systolic blood pressure (SBP) less than 90 mmHg, mean arterial pressure (MAP) of less than 65 mmHg, or a drop in systolic blood pressure of more than 40 mmHg from a patient’s known usual baseline readings. Compensated shock required signs of impaired peripheral perfusion, occurring in combination rather than singly, in the presence of systolic blood pressure of ≥90 mmHg. Severe hepatitis was defined as aspartate transaminase (AST) level >1,000 IU/L or alanine transaminase (ALT) level >1,000 IU/L.

Data extracted from records were demographic, co-morbidity, clinical parameters at the time of diagnosis of severe dengue, serial laboratory investigations at the time of diagnosis of severe dengue and at 24 hours later (with their timings), nadir and peak values of selected laboratory investigations, serial treatment, time data and outcome data. A data collection proforma was used to ensure the integrity of data. Time data, in the format of dates and times, were: fever onset, admission, the time of diagnosis of severe dengue, the time of defervescence (exact start of temperature persistently <38°C), the time of occurrence of the outcome and times of all laboratory investigations. Fever onset was determined by careful history-taking, specified to the best estimated time to the nearest hour. If missing it was imputed as midday (12:00 pm). As such, the day of illness may be determined for all variables. The time of onset of severe dengue was taken as the time of diagnosis of severe dengue. The temperature was monitored 4-hourly, and the temperature at diagnosis of severe dengue was determined from the temperature graph, extrapolated by the time of diagnosis of severe dengue. Clinical parameters included warning signs, which are clinical and/or laboratory features that predict the development of severe dengue.²⁸

We calculated the required sample size on the basis of comparison of the area under a receiver-operating characteristic (AUROC) curve with a null hypothesis value of 0·5, i.e. no discriminatory power. A sample size of 21 deaths was needed for this study for an AUROC of 80%, at the confidence level of 95% and a power of 95%.

Statistical Methods

Our analyses were two-pronged, the first involved analyses to describe our cohort, and the other, predictive model building. In order to describe the characteristics of our cohort clearly, we included variables which are related to laboratory investigations at 24 hours after the diagnosis of severe dengue, nadir and peak values of selected laboratory investigations, and variables related to management at 24-hourly time intervals. We described our cohort with descriptive and inferential analyses. Continuous variables were tested for normality with the Shapiro-Wilk test. As our data mostly had non-gaussian distribution, we used non-parametric analyses for data interrogation. Categorical variables were expressed as frequencies and percentages. Continuous variables were summarised as median and inter-quartile range (IQR).

Before proceeding to model-building, we performed univariable and standard multivariable logistic regression on all variables. The outcome or event of interest was any death during the hospital stay. As the study outcome uncommonly occurs with dengue and was expected to be similarly uncommon in our cohort, in the standard multivariable logistic regression, each variable was adjusted for age, gender, centre and time of onset of severe dengue. We then report the adjusted odds ratios in comparison to corresponding unadjusted values. This analysis was made on complete case basis.

Wherever applicable, we reported the statistical parameter with its 95% confidence interval. All tests of significance were 2-sided, and we took p-value <0·05 indicating statistical significance. All analyses were made using R, R Core Team (2016), R Foundation for Statistical Computing, Vienna, Austria.²⁹

Predictive model building

Variables used for predictive model building are called candidate predictors. As our study was focused on building a predictive model that would predict death in a severe dengue cohort, candidate predictors were collected at the onset of severe dengue. There were 28 candidate predictors which can be grouped into variables related to patient characteristics, warning signs in dengue, vital signs and laboratory (see Supplementary Table 1. Candidate Predictors for Model-building Grouped into Types). These candidate predictors were: age, body mass index (BMI), systolic blood pressure, diastolic blood pressure, pulse rate, respiratory rate, white blood cell count, platelet count, haematocrit, aspartate transaminase, alanine transaminase, serum albumin, serum bicarbonate, serum lactate, serum creatinine, gender, presence of any comorbidity, presence of non-structural protein 1 antigen, presence of immunoglobulin M, presence of immunoglobulin G, persistent vomiting, persistent diarrhoea, presence of abdominal pain, presence of abdominal tenderness, presence of clinical fluid accumulation, occurrence of non-severe bleeding, day of onset of severe dengue, severe dengue presentation (presented as severe dengue or not). Persistent vomiting and persistent diarrhoea were quantified as three or more episodes within the 24-hour interval before admission. Clinical fluid accumulation refers to the presence of any ascites or pleural effusion by clinical or radiological detection. In predictive model-building, the outcome of interest was also any death during the hospital stay.

In building our predictive model, we performed variable selection, model search, and internal validation. In order to select truly independent predictors of death, all 28 candidate predictors were simultaneously analysed in a single model using a technique called penalised regression.³⁰ The uncommon outcome (death) led to low events per variable, which prevented all 28 candidate predictors from being simultaneously analysed in a model using standard multivariable logistic regression approach - the number of events restricts the number of variables to be analysed within a model.

Our prediction model was built using multivariable penalised logistic regression using the glmnet and caret packages in R.^31,32 Penalised or regularised regression was chosen as the method is known to be able to perform variable selection in the condition of low events per variable very well.^30,33 In penalised logistic regression, a penalty term is introduced such that non-significant regression coefficients undergo shrinkage towards zero while coefficients of significant or influential variables would be retained as non-zero.³⁰ This is in contrast to the usual method of stepwise logistic regression to select independent predictors whereby variables are chosen based on the magnitude of the p-value. Penalised regression also reduces overfitting, which is usual with logistic regression.^30,33 Overfitting leads to the over-optimistic performance of a model. Another advantage of penalised regression is that it is one method that can be used to address multicollinearity.^30,34 It encourages a grouping effect, where strongly correlated predictors tend to be in or out of the model together.

We performed a model search for the best performing model by tuning different values of model hyperparameters α and λ of penalised regression. The best performing model was chosen by the highest AUROC. We then extracted the corresponding values of α and λ to build the final model. Model search and the final model underwent repeated k-fold cross-validation, using 1000 repeats and ten-folds that generated 10,000 sample models. Repeated k-fold cross-validation is one form of internal validation method of a predictive model. In this technique, resampling with replacement was made: 100,000 datasets were generated. Then, a total of 10,000 models were trained or built on 90,000 datasets and validated on the remaining 10,000 datasets. Cross-validation further minimised over-optimistic performance and served as our internal validation method. Performance measures we reported were the area under the curve of the receiver-operating curve, sensitivity, specificity, accuracy, positive predictive value, negative predictive value, positive likelihood ratio, and negative likelihood ratio. We also reported the calibration curve of the model. We illustrated the overall study pipeline in Figure 1 and the internal validation step in Supplementary Figure 1.

Figure 1. Overall study pipeline.

In prediction model building, missing data within the candidate predictors were imputed using k-nearest neighbours imputation during model-building via the caret package in R.³² Apart from serum lactate and serum bicarbonate, which had 16% and 11·6% missing data respectively, all other candidate predictors had less than 10% missing data. Since the proportion of missing data were less than 20%, we kept all the candidate predictors. Continuous variables among candidate predictors were retained in their original scales.

The study cohort was composed of 786 laboratory-confirmed severe dengue cases of whom 35 (4·5%) died. NS1 antigen was positive in 643 (81·8%), and a small proportion, 54 patients (6·9%) had only positive IgM. Distribution of cases according to centres were: 366 cases (46·6%) from Hospital Kuala Lumpur, 217 cases (27·6%) from Hospital Tuanku Ampuan Rahimah, 103 cases (13·1%) from Hospital Sungai Buloh, 61 cases (7·8%) from Hospital Sultanah Aminah, and 39 cases (5·0%) from Hospital Raja Perempuan Zainab II. Compensated shock occurred in 44·9%, acute kidney injury in 27·1%, decompensated shock in 24%, carditis in 21·1%, severe hepatitis in 16·7%, respiratory compromise in 12·7%, severe bleeding in 7·8%, and encephalopathy in 3·2% of patients.

The median age was 30·3 years old, 57·6% were males, 31·3% had co-morbidity, and 13·4% had no warning signs. Among those without warning signs, 28·6% presented as severe dengue. Severe dengue was diagnosed upon presentation in 23·8% of patients, and patients who were in the febrile phase at admission made up 82·7% of the cohort. The proportion who were in the febrile phase at diagnosis of severe dengue were 74·3%; by subset according to type of severe dengue the proportions were: for decompensated shock 73%, compensated shock 75·4%, respiratory compromise due to fluid accumulation 55%, severe bleeding 45·9%, severe hepatitis 58·8%, acute kidney impairment 77·5%, encephalopathy 68% and carditis 62·7%. The median duration of febrile phase was 4·8 days, the median onset of severe dengue was day 4·0, and median day of admission was on day 3·6. The median length of stay was 4 days, and the median length of illness was 7·6 days. Inotropes were administered in 16·2%, invasive and assisted ventilation in 17·7%, renal replacement therapy in 2·7%, and blood product transfusion in 15·3%. Demographic and clinical characteristics are given in Table 1, and laboratory variables are given in Table 2. Univariate and multivariable analysis of selected variables are given in Table 3 (full univariate and multivariable analysis of all study variables, including management-related variables at different time intervals are provided in Supplementary Table 2). The number of missing observations for each variable are provided in Supplementary Table 3.

Table 1. Demographic and clinical variables.

Table 2. Laboratory variables.

Table 3. Univariate and multivariable analysis of selected variables for risk of death in severe dengue.

Our final model was a penalised logistic regression with elastic net penalty model or an elastic net regression model, defined by model hyperparameters α = 0·1 and λ = 0·01613622. The final model comprised eight independent predictors of death among severe dengue cases: the presence of persistent diarrhoea, BMI, respiratory rate, levels of platelet count, serum bicarbonate, serum lactate, serum albumin, and aspartate transaminase. Since the predictors were timed at diagnosis of severe dengue, the model, therefore, is to be used from this time point onwards. The model regression coefficients and intercept are given in Figure 2. The model coefficients shown in Figure 2 are the regression coefficients of statistically significant variables that make up the final predictive model, i.e. their coefficients remained non-zero after shrinkage.

We reported the performance of the model in terms of discrimination measures (Table 4) and calibration. The final model has a high discrimination performance as represented by its AUROC of 89·6% [95% CI: 89·5 - 89·7]. In comparison, the raw standard non-cross-validated logistic regression model has an over-optimistic AUROC of 91·9% (Figure 3). The final penalised regression model has a very high sensitivity of 99·62% [95% CI: 99·61 - 99·63] but a modest specificity of 23·6% [95% CI: 23·2 - 24·0]. Nonetheless, the model has a very high accuracy of 96·23% [95% CI: 96·21–96·26]. Calibration refers to how closely the predicted probabilities of death agrees with the observed probabilities. Calibration of the probabilities was assessed by plotting observed probabilities versus predicted probabilities: a calibration curve. The calibration curve of the model showed excellent agreement between prediction and actual deaths when the probability of death is 60% and beyond, whilst below 60%, the agreement was less strong (Figure 4). During variable selection, centre and day of onset of severe dengue were found to be non-significant (coefficients underwent shrinkage to zero). The prediction model may be accessed at the following website: www.saifulsafuan.com/REPROSED2017E2.

Table 4. Performance measures of the final model, penalised logistic regression with elastic net penalty model.

Figure 2. Regression coefficients of the final model, penalised logistic regression with elastic net penalty model.

Figure 3. Receiver-operating characteristic curves of the raw model (standard non-cross-validated logistic regression model) vs the final model (elastic net regression model with repeated K-fold cross-validation).

Figure 4. Calibration curve of the final model, penalised logistic regression with elastic net penalty model.

We have developed a predictive model whichis available as an open-access web-based tool that has performed well with high certainty with an AUROC 89·6% [95% CI: 89·5–89·7] and a high accuracy of 96.23% [95% CI: 96.21–96.26]. The model comprises 8 independent predictors, incorporating demographic, clinical, and laboratory variables. The model was developed through regression search for influential variables of the disease. We believe the model reflects all important underlying pathophysiological aspects of the disease represented by these variables and therefore led to its high performance. The model has a very high sensitivity of 99·6% but a modest specificity of 23·6%. In view that such a model is intended to capture and prevent potential deaths, and the overall high AUROC, the model would be very useful in clinical practice. Misidentification of a non-fatal case is an acceptable compromise compared to missing a fatal case altogether. A possible explanation for the modest specificity is the fact that the real underlying outcome-determining pathophysiological mechanisms in dengue has not been clearly elucidated and only biomarkers involved in these mechanisms could improve the specificity of any model.

We developed this model to be employed at the time of severe dengue diagnosis, which may reasonably be assumed as the time of onset of severe dengue. This is likely the time when outcome-determining pathophysiological processes become critical. This is the earliest and most appropriate choice of time to prognosticate a patient in terms of death in severe dengue. We believe the selection of the point of time of prediction is crucial in the development of a dengue death prediction model. An earlier time point would be too early where pathophysiological processes may not have reached outcome-determining significance. Selection of a later time point would be too late for the prediction model to be beneficial. With that in mind and the fact that the model performed well, we postulate that investigating processes represented by these variables could elucidate the pathophysiology of dengue with better clarity. We also believe that the model may be used to track the progression of a patient with severe dengue through the course of illness, assists in guiding prognostication and decision-making.

Making predictions in dengue research has gained momentum. Modelling studies in outbreak prediction utilised several predictive analytics which includes ensemble methods, time series regression, and support vector machine.^10–16 Modelling studies involving earlier clinical aspects of management of dengue - identifying and stratifying dengue - used decision trees, logistic regression, and structural equation models.^17–21 However, only three studies have modelled prediction of death.^23–25

Huang et al. studied all patients with dengue that included 34 deaths and identified five independent predictors of death: age >64 years old, diabetes mellitus, systolic blood pressure <90 mmHg, chronically bedridden, and haemoptysis.²³ However, this study used a scoring method. Risk scoring is not without weaknesses and has been deemed to have serious problems.³⁵ Risk scoring involves converting continuous predictors into categorical predictors. The conversion results in loss of granularity of information contained within continuous predictors.³⁶ Moreover, the presence of any of the first three predictors only occurred in 162 (20·6%) cases of our cohort and the combination of all three in only 2 cases. It is imperative in clinical prediction that a predictor is sufficiently prevalent for the achievement of reasonable accuracy.³⁷

Md-Sani et al. examined severe dengue cases and predicted death at the onset of severe dengue similar to the current study but at a single centre.²⁴ That study however, had only 20 events (deaths).²⁴ Building a predictive model to predict death among severe dengue cases is challenging as dengue death, the event or outcome of interest, is actually uncommon - in Malaysia it is just above 0.2% of all dengue cases.⁶ This poses a huge problem in that, in the usual approach of statistical modelling using multivariable logistic regression, 5–10 events per candidate predictor variable (EPV) are required.³⁸ This ratio dictates the number of candidate predictors that may be simultaneously analysed in the multivariable model. When the EPV is less than this, the number of candidate predictors that may be simultaneously analysed to identify which among them are truly independent predictors is limited. Thus, Md-Sani et al. employed the approach of using adjustment or controlling variables instead of assessing all candidate predictors simultaneously in one model. The present study is larger, multicentred and used a different analysis technique which addressed overfitting, multicollinearity and the low number of events per variable.

Pinto et al. which had a large cohort with 61 severe dengue deaths, built a simple predictive model comprising only four categorical predictors: age (binary, cutoff age 55), haematuria, gastrointestinal bleeding and thrombocytopaenia (binary, cutoff platelet count 20,000 cells/mm3).²⁵ In our study, we did not specifically identify the source of bleeding. However, assuming the variable warning sign of spontaneous bleeding tendencies and the variable severe bleeding represent haematuria and gastrointestinal bleeding, respectively, the presence of any of Pinto et al.’s predictors occurred only in 259 (33%) cases in our cohort. The presence of all four predictors occurred only in 3 cases. Therefore, these models may not be adequate to prognosticate death in severe dengue. Our model kept variables in their original continuous attribute, and persistent diarrhoea was the only categorical variable. Persistent diarrhoea occurred in 38·5% of our cohort, which is more prevalent than any of the predictors of Huang et al. and Pinto et al. if applied to our cohort. Thus, because of this and the predominantly continuous attribute of predictors in our model, we obtained a higher performance accuracy. Additionally, Pinto et al. used the WHO 1997 dengue classification instead of the WHO 2009 schema which Malaysia has adopted in clinical practice.^6,28

While there are other studies on death in dengue, they examined the association with death in unselected dengue cases and were not prediction modelling studies.^39–46 Our study is the first modelling study, based on the latest and widely adopted WHO 2009 classification scheme, to model prediction of death in severe dengue cases. As mentioned above, there have been many studies that built models for identification of dengue and stratification of severity in dengue. Our model complements this and completes the prediction aspect of clinical management.

The study additionally documented another interesting finding. We found that almost three-quarters of the cases were still in the febrile phase (study definition temperature >38°C) at diagnosis of severe dengue. This finding supports similar documentation in a previous study.²⁴ A similar proportion was found for those who had shock. This finding is different from what majority of guidelines have stated where shock only occurs during the critical phase, i.e. upon defervescence and later.^6,28 Current clinical practice only recognises that severe dengue would occur only during the critical phase, thereby missing severe dengue which could occur earlier in the febrile phase. Our study provided quantified evidence that severe dengue could occur early in the febrile phase and clinicians should be vigilant of this fact in order to prevent deaths.

The limitation of our study is its retrospective design in which missing data was inevitable. However, only variables of utmost importance were included as candidate predictors for variable selection. We included serum bicarbonate and serum lactate though these had higher missing proportions (11·6 and 16%, respectively) as we believe they play essential roles in determining not only progression to severe disease but also death.^24,47,48 Even though external validation of the model was not performed, we believe that the repeated k-fold cross-validation algorithm we employed had ensured the robustness of the model for unseen data. A final limitation is that though our model may save costs due to its accessibility, it will require additional laboratory-related resources. Nevertheless, any severe dengue cases should be treated in a setting with adequate resources to implement evidence-based clinical practice, and the laboratory predictors in our model are commercially available. Zakaria et al. demonstrated that the WHO 2009 dengue severity stratification scheme classifies more patients (4·6%) into the most severe form as compared to the previous WHO 1997 scheme (0·7%).⁴⁹ They highlighted that this might pose a significant impact on hospital resources. Our model can potentially prioritise patients to local resources based on their probability of death.

In conclusion, we have developed a dengue death prediction model comprising clinical and laboratory data and deployed an open-access web-based tool for any centre to utilise for local validation. The findings from this study would be valuable to the global community of clinicians who treat dengue, hopefully paving better and tailored clinical decision-making and resource management. In terms of research, the tool may be a useful yardstick, similar to how TIMI and APACHE are useful to cardiovascular and critical care medicine, respectively.^22,50,51

Ethics approval

The study was approved by the Medical Research and Ethics Committee (MREC), Ministry of Health of Malaysia (NMRR–16–2074–33141).

Consent for publication

Not applicable.

Availability of data

Datasets used and source codes to our analyses are given in supplementary files accompanying this article.

Competing interests

The authors declare that they have no potential conflicts of interest.

Funding

This work was supported by Medical Research Grant (NMRR–16–2074–33141), National Institutes of Health, Ministry of Health Malaysia.

Authors’ contributors

SSM, MZ, RH, AMJ, PYO, SK, AS, and MM collected and contributed to data. SSM, MZ, AMJ, MA, SK, and MM designed the study. SSM analysed the data. SSM, KNL, MZ, and RH wrote the first draft of the manuscript. All authors performed data interpretation and critically reviewed the manuscript.

Acknowledgements

The authors thank Dr Siti Zubaidah A. Subki from Medical Development Division, Ministry of Health Malaysia, Dr Lee Jen Ven and Dr Duratul’ain M. Nazri from the Clinical Research Centre, Hospital Kuala Lumpur, for their commitment and diligence in their assistance running this study; and the Director-General of Health Malaysia for his permission to publish this article.

Bhatt S, Gething PW, Brady OJ, Messina JP, Farlow AW, Moyes CL, Drake JM, Brownstein JS, Hoen AG, Sankoh O, Myers MF. The global distribution and burden of dengue. Nature. 2013;496(7446):504.
World Health Organisation. https://www.who.int/news-room/fact-sheets/detail/dengue-and-severe-dengue. Accessed July 6, 2019.
World Health Organisation. https://www.who.int/news-room/fact-sheets/detail/ebola-virus-disease. Accessed July 6, 2019.
Shepard DS, Undurraga EA, Halasa YA. Economic and disease burden of dengue in Southeast Asia. PLoS Negl Trop Dis. 2013;7(2):e2055.
Malaysian Remote Sensing Agency and National Disease Control Division, Ministry of Health Malaysia. Statistik. In: iDengue untuk komuniti. http://idengue.remotesensing.gov.my/idengue/content/statistik.pdf. Accessed 17 November 2018.
Ministry of Health Malaysia, Medical Development Division, Malaysia Health Technology Assessment Section (MaHTAS). Clinical Practice Guidelines: Management of Dengue Infection in Adults. 3rd Edition. Putrajaya: Ministry of Health Malaysia; 2015.
Packierisamy PR, Ng CW, Dahlui M, et al. Cost of dengue vector control activities in Malaysia. Am J Trop Med Hyg. 2015;93(5):1020–7.
Shepard DS, Undurraga EA, Lees RS, Halasa Y, Lum LC, Ng CW. Use of multiple data sources to estimate the economic cost of dengue illness in Malaysia. Am J Trop Med Hyg. 2012;87(5):796–805.
Shepard DS, Undurraga E, Lees R, Halasa YA, Lum LCS, Ng C. Errata. Am J Trop Med Hyg. 2013;88:606.
Ho CC, Ting CY, Raja DB. Using Public Open Data to Predict Dengue Epidemic: Assessment of Weather Variability, Population Density, and Land use as Predictor Variables for Dengue Outbreak Prediction using Support Vector Machine. Indian J Sci Technol. 2018;11(4).
Ahmad R, Suzilah I, Najdah WM, Topek O, Mustafakamal I, Lee HL. Factors determining dengue outbreak in Malaysia. PLoS One. 2018;13(2):e0193326..
Loshini T, Asirvadam VS, Dass SC, Gill BS. Predicting localised dengue incidences using ensemble system identification. In: 2015 International Conference on Computer, Control, Informatics and its Applications (IC3INA), 5-7 Oct 2015;Bandung, Indonesia. New York: IEEE; 2015. 6–11.
Buczak AL, Baugher B, Moniz LJ, Bagley T, Babin SM, Guven E. Ensemble method for dengue prediction. PLoS One. 2018;13(1):e0189988.
Morin CW, Monaghan AJ, Hayden MH, Barrera R, Ernst K. Meteorologically driven simulations of dengue epidemics in San Juan, PR. PLoS Negl Trop Dis. 2015;9(8):e0004002..
Yamana TK, Kandula S, Shaman J. Superensemble forecasts of dengue outbreaks. J R Soc Interface. 2016;13(123):20160410.
Ramadona AL, Lazuardi L, Hii YL, Holmner Å, Kusnanto H, Rocklöv J. Prediction of dengue outbreaks based on disease surveillance and meteorological data. PLoS One. 2016;11(3):e0152688.
Tanner L, Schreiber M, Low JG, et al. Decision tree algorithms predict the diagnosis and outcome of dengue fever in the early phase of illness. PLoS Negl Trop Dis. 2008;2(3):e196.
Cucunawangsih DB, Sungono V. Scoring Model to Predict Dengue Infection in the Early Phase of Illness in Primary Health Care Centre. Arch Clin Microbiol. 2015;6: 2.
Lee VJ, Lye DC, Sun Y, Leo YS. Decision tree algorithm in deciding hospitalisation for adult patients with dengue haemorrhagic fever in Singapore. Trop Med Int Health. 2009;14(9):1154–9.
Potts JA, Gibbons RV, Rothman AL, et al. Prediction of dengue disease severity among pediatric Thai patients using early clinical laboratory indicators. PLoS Negl Trop Dis. 2010;4(8):e769.
Carrasco LR, Leo YS, Cook AR, et al. Predictive tools for severe dengue conforming to World Health Organization 2009 criteria. PLoS Negl Trop Dis. 2014;8(7):e2972.
Antman EM, Cohen M, Bernink PJ, et al. The TIMI risk score for unstable angina/non–ST elevation MI: a method for prognostication and therapeutic decision making. JAMA. 2000;284(7):835-42.
Huang CC, Hsu CC, Guo HR, Su SB, Lin HJ. Dengue fever mortality score: A novel decision rule to predict death from Dengue fever. J Infect. 2017;75(6):532–40.
Md-Sani SS, Md-Noor J, Han WH, et al. Prediction of mortality in severe dengue cases. BMC Infect Dis. 2018;18(1):232.
Pinto RC, de Castro DB, de Albuquerque BC, et al. Mortality predictors in patients with severe dengue in the State of Amazonas, Brazil. PLoS One. 2016;11(8):e0161884.
Hales S, De Wet N, Maindonald J, Woodward A. Potential effect of population and climate changes on global distribution of dengue fever: an empirical model. Lancet. 2002;360(9336):830–4.
Moons KG, Altman DG, Reitsma JB, et al. Transparent Reporting of a multivariable prediction model for Individual Prognosis or Diagnosis (TRIPOD): explanation and elaboration. Ann Intern Med. 2015;162(1):W1–73.
Department of Control of Neglected Tropical Diseases, World Health Organization. Epidemic, & Pandemic Alert. Dengue: guidelines for diagnosis, treatment, prevention and control. Geneva: World Health Organization; 2009.
R Core Team. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. 2016. URL https://www.R-project.org
James G, Witten D, Hastie T, et al. An introduction to statistical learning. New York: Springer; 2013; Chapter 6 Linear Model Selection and Regularization (Section 6.2 Shrinkage Methods): 214-227.
Friedman J, Hastie T, Tibshirani R. Regularization Paths for Generalized Linear Models via Coordinate Descent. Journal of Statistical Software. 2010;33(1):1-22.
Kuhn, M. Building Predictive Models in R Using the caret Package. Journal of Statistical Software. 2008;28(5):1-26.
Pavlou M, Ambler G, Seaman SR, et al. How to develop a more accurate risk prediction model when there are few events. BMJ. 2015;351:h3868.
Zou H, Hastie T. Regularization and variable selection via the elastic net. Journal of the Royal Statistical Society: Series B (Statistical Methodology). 2005;67(2):301-20.
Hubbard D, Evans D. Problems with scoring methods and ordinal scales in risk assessment. IBM J Res Dev. 2010;54(3):2–1.
Altman DG, Royston P. The cost of dichotomising continuous variables. BMJ. 2006;332(7549):1080.
Schummers L, Himes KP, Bodnar LM, Hutcheon JA. Predictor characteristics necessary for building a clinically useful risk prediction model: a simulation study. BMC Med Res Methodol. 2016;16(1):123.
Vittinghoff E, McCulloch CE. Relaxing the rule of ten events per variable in logistic and Cox regression. American Journal of Epidemiology. 2007;165(6):710-8.
Amâncio FF, Heringer TP, Fassy LB, et al. Clinical profiles and factors associated with death in adults with dengue admitted to intensive care units, Minas Gerais, Brazil. PLoS One. 2015;10(6):e0129046.
Werneck GL, Macias AE, Mascarenas C, et al. Comorbidities increase in-hospital mortality in dengue patients in Brazil. Mem Inst Oswaldo Cruz. 2018;113(8).
Almas A, Parkash O, Akhter J. Clinical factors associated with mortality in dengue infection at a tertiary care centre. Southeast Asian J Trop Med Public Health. 2010;41(2):333–40.
Khalil MA, Tan J, Khalil MA, Awan S, Rangasami M. Predictors of hospital stay and mortality in dengue virus infection-experience from Aga Khan University Hospital Pakistan. BMC Res Notes. 2014;7(1):473.
Karunakaran A, Ilyas WM, Sheen SF, Jose NK, Nujum ZT. Risk factors of mortality among dengue patients admitted to a tertiary care setting in Kerala, India. J Infect Public Health. 2014;7(2):114–20.
Chowdhury R, Pan K, Sarkar A, Chakrabarti S, Samanta T. Predictors of dengue mortality in a tertiary care hospital at kolkata: a cross-sectional study. Int J Med Sci Public Health. 2013;2(2):254–8.
Acharya V, Khan MF, Kosuru S, Mallya S. Predictors of mortality in adult patients with dengue: a study from South India. International Journal of Research in Medical Sciences. 2018;6(5):1605–10.
Singh J, Zeya MT, Dhir G, et al. Mortality and Severity Predictors of Dengue Fever. Scholars Journal of Applied Medical Sciences. 2014;2(6A):1958–61.
Thanachartwet V, Oer-Areemitr N, Chamnanchanunt S, et al. Identification of clinical factors associated with severe dengue among Thai adults: a prospective study. BMC Infect Dis 2015;15(1):420.
Shankar-Hari M, Phillips GS, Levy ML, et al. Developing a new definition and assessing new clinical criteria for septic shock: for the Third International Consensus Definitions for Sepsis and Septic Shock (Sepsis-3). JAMA. 2016;315(8):775–787.
Zakaria Z, Zainordin NA, Sim BL, et al. An evaluation of the World Health Organization’s 1997 and 2009 dengue classifications in hospitalized dengue patients in Malaysia. The Journal of Infection in Developing Countries. 2014;8(07):869-75.
Bouch DC, Thompson JP. Severity scoring systems in the critically ill. Continuing Education in Anaesthesia, Critical Care & Pain. 2008;8(5):181-5.
Vincent JL, Moreno R. Clinical review: scoring systems in the critically ill. Critical care. 2010;14(2):207.

Table 1. Demographic and clinical variables^a.
	Died (N=35)	Survived (751)
	Median (IQR) or n(%)	Median (IQR) or n(%)	p^b value
Age (years)	42 (22·6)	30 (20)	·001
Gender (male)	20 (57·1)	433 (57·7)	NS
BMI (kg/m²)	28·1 (7·1)	24·7 (7·7)	·001
Systolic BP (mmHg)	105 (3·5)	111 (24)	NS
Diastolic BP (mmHg)	67 (25)	68 (18)	NS
Mean arterial pressure (mmHg)	81·3 (22·3)	83 (18·7)	NS
Pulse rate (bpm)	106 (41)	94 (27)	NS
Respiratory rate (breaths/min)	23 (8)	20 (2)	<·0001
Presence of any co-morbidities	19 (54·3)	227 (3·2)	·004
Hypertension	11 (31·4)	97 (12·9)	·003
Diabetes mellitus	7 (2·0)	70 (9·3)	·04
COPD/Asthma	5 (14·3)	46 (6·1)	NS
Heart disease	2 (5·7)	28 (3·7)	NS
Chronic kidney disease	1 (2·9)	7 (0·9)	NS
Chronic liver disease	1 (2·9)	1 (0·1)	·03
Associated bloodstream infection	7 (20)	28 (3·7)	·0001

Laboratory diagnosis of dengue fever
NS1	31 (88·6)	612 (81·5)	NS
IgM	19 (54·3)	348 (46·3)	NS
IgG	13 (37·1)	236 (31·4)	NS

Types of severe dengue
Compensated shock	14 (4·0)	339 (45·1)	NS
Decompensated shock	15 (42·9)	174 (23·2)	·01
Severe bleeding	15 (42·9)	46 (6·1)	<·0001
Severe hepatitis	25 (71·4)	106 (14·1)	<·0001
Acute kidney impairment	28 (8·0)	185 (24·6)	<·0001
Encephalitis/encephalopathy	5 (14·3)	20 (2·7)	·0007
Cardiac complications	19 (54·3)	147 (19·6)	<·0001
Respiratory compromise	21 (6·0)	79 (1·5)	<·0001

Warning signs at diagnosis of severe dengue
Persistent vomiting	23 (65·7)	369 (49·1)	NS
Persistent diarrhoea	23 (65·7)	280 (37·3)	·001
Abdominal pain	22 (62·9)	307 (4·9)	·01
Abdominal tenderness	12 (34·3)	141 (18·8)	·03
Third space fluid accumulation	16 (45·7)	125 (16·6)	<·0001
Spontaneous bleeding tendencies	13 (37·1)	98 (13·0)	·0002
Raised hematocrit with rapid drop of platelet	19 (57·6)	325 (44·2)	NS

Timing of events
Length of illness	6·9 (3·4)	7·6 (2·6)	NS
Length of stay	3·1(3·6)	4·0 (2·8)	NS
Day of onset of severe dengue	3·7 (1·7)	4·0 (2·0)	NS
Length of febrile phase	4·3 (2·3)	4·8 (2·1)	NS
Febrile severe dengue	14 (51·9)	561 (75·1)	·009
Febrile phase at admission	18 (66·7)	621 (83·2)	·03
Presentation as severe dengue	3 (8·6)	184 (24·5)	·04
Abbreviations: IQR, interquartile range; BMI, body mass index; BP, blood pressure; COPD, chronic obstructive pulmonary disease; NS1, non-structural protein 1 antigen; IgM, immunoglobulin M; IgG, immunoglobulin G; NS, not significant.
^aCategorical variables are summarised as n(%). Continuous variables are represented as median (IQR).
^bUnivariate logistic regression was used to compare differences between groups. Analysis was made on complete case basis.

Table 2. Laboratory variables^a.
	Died	Survived
	Median (IQR) or n(%)	Median (IQR) or n(%)	p^b value
At diagnosis of severe dengue
WBC (×10³/μL)	7·5 (5·7)	4·0 (2·9)	<·0001
Hb (g/dL)	14·8 (4·4)	14·6 (3·2)	NS
Platelet (×10³/μL)	23 (35)	67 (99)	·0001
Hct (%)	43·8 (13)	43·5(8·6)	NS
AST (U/L)	813 (1565)	105 (202)	<·0001
ALT (U/L)	322 (480)	65·5 (123·5)	·0001
Albumin (g/dL)	32 (10)	36 (8)	<·0001
INR	1·41 (·46)	1·1 (1·43)	NS
Serum Bicarbonate (mmol/L)	17·9 (5·925)	23 (3·9)	<·0001
Serum Lactate (mmol/L)	3·92 (5·55)	1·6 (1·06)	<·0001
Serum Creatinine (μmol/L)	115 (46·4)	85 (41)	·003

At 24 hours after diagnosis of severe dengue
WBC (×10³/μL)	8·05 (9·9)	4·1 (3·95)	<·0001
Hb (g/dL)	12·55 (4·3)	13·8 (3)	·008
Platelet (×10³/μL)	24·5 (33)	49 (73)	·007
Hct (%)	38·7 (12·2)	4·55 (7·8)	·02
AST (U/L)	3628 (9560)	153 (282)	<·0001
ALT (U/L)	1384 (1746·5)	89 (159)	<·0001
Albumin (g/dL)	30 (12)	31 (8)	NS
INR	1·67 (·76)	1·08 (·19)	<·0001
Serum Bicarbonate (mmol/L)	15·4 (1·4)	23 (3·2)	<·0001
Serum Lactate (mmol/L)	6·1 (12·3)	1·3 (·8)	<·0001
Serum Creatinine (μmol/L)	147·5 (77·5)	69 (34·9)	<·0001

Nadir and Peak Values
Baseline Hct (%)	41 (7·05)	41·2 (7·6)	NS
Peak Hct (%)	49·9 (7·05)	46 (8·4)	·005
Day of Peak Hct	4·5 (2·3)	4·4 (2·2)	NS
Nadir Platelet (×10³/μL)	8 (11)	25 (48)	·0007
Day of Nadir Platelet	4·1 (1·9)	5·4 (1·8)	NS
Peak AST (U/L)	4202 (12307·5)	165·5 (288)	<·0001
Day of Peak AST	5 (2·2)	5·0 (2·2)	NS
Peak ALT (U/L)	1510 (2550)	99 (166)	<·0001
Day of Peak ALT	5·2 (2·3)	5·2 (2·4)	NS
Peak Creatinine (μmol/L)	203·5 (13·5)	89 (41)	<·0001
Day of Peak Creatinine	5·3 (3·0)	4·1 (2·2)	NS
Abbreviations: IQR, interquartile range; WBC, white blood cell count; Hb, haemoglobin; Hct, haematocrit; AST, aspartate transaminase; ALT, alanine transaminase; INR, international normalised ratio; NS, not significant.
^aCategorical variables are summarised as n(%). Continuous variables are represented as median (IQR).
^bUnivariate logistic regression was used to compare differences between groups. Analysis was made on complete case basis.

Table 3. Univariate and multivariable analysis of selected variables for risk of death in severe dengue.
	Unadjusted OR^a (95% CI)	p^a value	Adjusted OR^b (95% CI)	p^b value
Clinical parameters at diagnosis of SD
Age (years)	1·03 (1·01-1·06)	·001	-	-
Gender (male)	0·98 (·50-1·97)	NS	-	-
BMI (kg/m²)	1·07 (1·03-1·12)	·001	1·07 (1·02-1·12)	·006
Systolic BP (mmHg)	0·98 (·97-1·00)	NS	0·98 (·96-·99)	·008
Diastolic BP (mmHg)	0·99 (·97-1·01)	NS	0·98 (·96-1·00)	NS
Mean arterial pressure (mmHg)	0·98 (·96-1·01)	NS	0·98 (·95-1·00)	·03
Pulse rate (bpm)	1·02 (1·00-1·04)	NS	1·03 (1·01-1·05)	·01
Respiratory rate (breaths/min)	1·13 (1·07-1·19)	<·0001	1·12 (1·05-1·20)	·0003
Presence of any co-morbidities	2·74 (1·39-5·49)	·004	1·60 (·71-3·59)	NS
Associated bloodstream infection	6·46 (2·43-15·39)	·0001	3·58 (1·21-9·54)	·01

Laboratory confirmation of dengue fever
NS1	1·76 (·68-5·99)	NS	1·61 (·59-5·68)	NS
IgM	1·38 (·70-2·75)	NS	1·45 (·71-2·99)	NS
IgG	1·29 (·62-2·57)	NS	1·37 (·63-2·90)	NS

Warning signs at diagnosis of severe dengue
Persistent vomiting	1·98 (·99-4·18)	NS	1·42 (·63-3·26)	NS
Persistent diarrhoea	3·22 (1·61-6·79)	·001	2·54 (1·23-5·49)	·01
Abdominal pain	2·45 (1·23-5·06)	·01	2·44 (1·18-5·26)	·02
Abdominal tenderness	2·26 (1·06-4·57)	·03	2·25 (1·01-4·79)	·04
Third space fluid accumulation	4·22 (2·09-8·43)	<·0001	3·15 (1·43-6·85)	·004
Spontaneous bleeding tendencies	3·94 (1·88-7·98)	·0002	2·7 (1·21-5·82)	·01
Raised hematocrit with rapid drop of platelet	1·72 (·85-3·54)	NS	1·27 (·60-2·73)	NS

Timing of events
Length of illness	1·01 (·93-1·07)	NS	0·96 (·87-1·03)	NS
Length of stay	1·00 (·90-1·06)	NS	0·93 (·83-1·01)	NS
Febrile severe dengue	0·36 (·16 - ·78)	·009	0·58 (·24-1·42)	NS
Febrile phase at admission	0·40 (·18 - ·96)	·03	0·46 (·18-1·20)	NS
Presentation as severe dengue	0·29 (·07 - ·82)	·04	0·42 (·10-1·25)	NS

At diagnosis of severe dengue
WBC (×10³/μL)	1·16 (1·08 – 1·25)	<·0001	1·13 (1·04-1·22)	·004
Hb (g/dL)	1·08 (·94 -1·25)	NS	1·05 (·90-1·23)	NS
Platelet (×10³/μL)	0·98 (·97-·99)	·0001	0·98 (·97-·99)	·004
Hct (%)	1·00 (·96 – 1·05)	NS	0·98 (·94-1·04)	NS
AST (U/L)	1·00 (1·00 -1·00)	<·0001	1·00 (1·00-1·00)	·0001
ALT (U/L)	1·00 (1·00 -1·00)	·0001	1·00 (1·00-1·00)	·01
Albumin (g/dL)	0·85 (·80 – ·91)	<·0001	0·86 (·80-·93)	·0001
INR	1·43 (1·00 -2·78)	NS	1·15 (·79-1·95)	NS
Serum Bicarbonate (mmol/L)	0·70 (·63 – ·77)	<·0001	0·72 (·63-·80)	<·0001
Serum Lactate (mmol/L)	1·50 (1·31 – 1·75)	<·0001	1·39 (1·23-1·61)	<·0001
Serum Creatinine (μmol/L)	1·01 (1·00 -1·01)	·003	1·00 (1·00-1·01)	NS
Abbreviations: CI, confidence interval; WBC, white blood cell count; Hb, haemoglobin; Hct, haematocrit; AST, aspartate transaminase; ALT, alanine transaminase; NS, not significant. Age, gender, centre and day of onset of severe dengue were used as controlling variables.
^aUnivariate logistic regression was used to compare differences between groups. Analysis was made on complete case basis.
^bMultivariable logistic regression was used to compare differences between groups; age, gender, centre and day of onset of severe dengue were used as controlling variables. Analysis was made on complete case basis.

Table 4. Performance measures of the final model, penalised logistic regression with elastic net penalty model.
Performance	Mean (95% CI)
AUROC	89·6 (89·5-89·7)
Sensitivity	99·62 (99·61-99·63)
Specificity	23·59 (23·15-24·02)
Accuracy	96·23 (96·21-96·26)
Positive predictive value	96·56 (96·54-96·58)
Negative predictive value	71·13 (70·24-72·02)
Positive likelihood ratio	1·45 (1·44-1·47)
Negative likelihood ratio	0·011 (·011-·012)
Abbreviations: CI, confidence interval; AUROC, area under receiver operating characteristic curve.

Download PDF

Version 2

posted

You are reading this latest preprint version

Multivariable prediction model for predicting deaths in severe dengue cases

Status:

Version 2

Abstract

Figures

INTRODUCTION

METHODS

RESULTS

DISCUSSION

Declarations

References

Tables

Supplementary Files

Status:

Version 2