Utility of Clinical Improvement and Platelet Count Recovery Time in Counseling Children Hospitalized With Suspected Dengue in A Resource-Poor Setting

Abstract

Objective:We aimed to evaluate the utility of platelet count recovery time in children with suspected dengue, altered by interventions such as platelet transfusions, diagnostic groups and lowest platelet counts, in counseling and discharge planning. We also evaluated the usefulness of improvement in clinical signs and symptoms for discharge planning.
Methodology: Baseline data was collected from 41 consecutive hospitalized children who were suspected to be suffering from dengue, during the recent confirmed dengue epidemic in North India (Sep-Nov 2006). All children were managed with standardized guidelines suggested by the World Health Organization (WHO), and time taken for the recovery of platelet counts (>100,000/cumm) was recorded.
Results: The median recovery time(that is the time at which 50% of children had recovered) was noted to be 3 and 4 days respectively, for groups Dengue and Dengue haemorrhagic fever (all grades), using Kaplan-Meier survival analysis (p=0.003). One way analysis of variance showed the two groups to be different, with regards to platelet count recovery time (p=0.004). Duration of hospitalization correlated strongly with the platelet count recovery time (r=0.82, p<0.0001). The lowest levels of platelet count during the course of hospitalization predicted platelet count recovery time (p=0.003). There was fair to moderate agreement (kappa = 0.31-0.5) for improvement in clinical symptoms and signs at the time of discharge, except for a convalescent rash (kappa=0.7).
Conclusion: The information derived from this study is useful for counseling and discharge planning of hospitalized children suspected to be suffering from dengue, during epidemics in the countries of South Asia where people have to make significant out-of-pocket payments.

Dengue, Platelet recovery time, counseling

Thrombocytopaenia is an essential diagnostic criterion of dengue and dengue haemorrhagic fever (1). The major threat from dengue is bleeding and dengue shock syndrome. Even though there is a poor correlation of bleeding with platelet counts, it is a major scare for both the public and the physicians during a confirmed epidemic, and is often the reason for hospitalization, generally at the nadir of platelet counts (4-7 days after onset of fever)(2),(3). This leads to inappropriate platelet transfusions in a large number of patients, resulting in increased morbidity and hospitalizations (4). Most patients of dengue have recovery of platelet counts, and even the utility of preventive transfusions in Dengue Shock Syndrome has been questioned (5),(6). However, in resource poor settings and developing countries, an increasing number of patients are susceptible to secondary severe dengue infections, resulting in the hospitalization of an overwhelming number of patients during major dengue epidemics. Unfortunately, the health infrastructure in these countries predominantly consists of private health care providers, and people have to make significant and catastrophic out-of-pocket payments for availing health services (7). Thus, it is not only the clinical indications, but also the ability-to-pay (ATP) which plays an important role in the decision for hospitalization. Patients are thus keen to know about the duration of hospitalization for them make an informed decision. Timing of platelet count recovery is generally associated closely with clinical improvement in dengue infections, and is a good objective marker. However, recovery of platelet count after hospitalization is confounded by multiple factors such as diagnostic groups, co-infections, platelet/plasma transfusions, lowest platelet counts and duration of fever at the time of hospitalization (5). We retrospectively evaluated the platelet count recovery time and examined it’s correlation with duration of hospitalization in a cohort of children hospitalized with suspected dengue during the dengue epidemic of 2006 in northern India (8). Along with platelet count recovery time, we also assessed other clinical signs and symptoms for counseling and discharge planning.

Material and Methods

We collected data on demographic variables, diagnostic groups, information on platelet and plasma transfusions, lowest platelet counts and platelet recovery time in days (platelet count >100,000), on a structured performa from 41 consecutive children hospitalized with suspected dengue between September and November 2006, during a confirmed dengue epidemic at a general hospital in north India. The diagnosis of dengue was made as per the World Health Organization (WHO) guidelines and other criteria relevant to our settings (9),(10),(11). All children underwent serological confirmation using rapid ELISA tests for IgM and IgG antibodies. Each child was managed using standardized guidelines suggested by the World Health Organization (WHO) during the epidemics for smaller hospitals (9).

We aimed to document the platelet count recovery time altered by platelet/plasma transfusions, diagnostic groups and lowest platelet counts during the course of hospitalization. All children had platelet counts daily, and were discharged when platelet counts were greater than 50,000/cumm, and showed a rising trend along with improvement in clinical signs and symptoms. Two physicians (blinded to each other) recorded the improvement in symptoms and signs in a graded manner, at the time of discharge. The children were followed up every 24 hours for two to four days, to assess the progress as regards to their health. Those children who were discharged with platelet counts between 50,000-100,000/cumm, had repeat platelet counts daily or until more than 100,000/cumm to study the trend in platelet counts, and time for recovery was recorded in this visit. Five children were excluded from the analysis, as they had always had platelet counts of more than 100,000/cumm during the course of hospitalization.

Indications of Platelet Transfusions
Platelet concentrates were transfused, when the platelet count was less than 25,000/cumm, with evidence of significant bleeding from mucosal surfaces, haematuria, epistaxis, haematemesis or haematochizia. Plasma transfusions were given only when there was a deranged coagulation profile. In exceptional cases with platelet counts less than 50,000/cumm with severe bleeding, patients were transfused when they requested transfusion and could not be convinced about waiting patiently (patient autonomy was respected in these cases). Even though the futility of platelet transfusion at 10,000-20,000/cumm (5) has been debated about, we used a cut-off criteria of 25,000/cumm along with bleeding, as there were three deaths in adults during the same time period, with severe bleeding followed by shock, even in patients with platelet counts more than 20,000/cumm.

Sample Size Calculation
Dengue haemorrhagic fever (all grades) has a longer platelet count recovery time than dengue fever, which is universally masked with platelet and plasma transfusions in severe cases, in clinical practice. Thus, assuming a mean difference of 1.5 days with a wide standard deviation of two days, a sample size of 30 was calculated.

Data Entry and Statistical Analysis
Data was initially entered in Microsoft (Excel) spread sheet, and was transferred to the MEDCALC statistical software (Belgium) file. Cases were dichotomized into the diagnostic group’s Dengue fever and Dengue haemorrhagic fever (all grades I-IV). Data on lowest platelet counts, maximum haematocrit, platelet count recovery time, duration of fever at the time of hospitalization, and duration of hospitalization (days), were entered as continuous variables. Need for transfusion and all other signs and symptoms, were coded as dichotomous variables. One way Analysis of variance (ANOVA) was done to analyze the difference between the two diagnostic groups. Kaplan-Meier survival curves were constructed to know the median platelet recovery time (defined as platelet count>100,000). Pearson correlation coefficient was calculated to see the association between platelet count recovery time and duration of hospitalization, as well as with symptoms (abdominal pain, vomiting, fatigability, fever, anorexia), and signs (bleeding manifestations, hepatomegaly, conjunctival suffusion/facial flushing, convalescent rash) at the time of discharge. An inter-rater agreement using kappa statistics was made for the assessment of the utility of improvement in clinical symptoms and signs at the time of discharge.
We also attempted to study the independent predictors of platelet recovery time using a multiple regression model (backward selection), with five variables of duration of fever at the time of hospitalization, lowest platelet count, haematocrit levels, diagnostic groups (dengue haemorrhagic fever all grades vs. dengue), and need for transfusions (plasma and/or platelet).

The MEDCALC statistical software licensed version 9.3.7.0 (Belgium) was used for the analysis.

Results

Baseline characteristics of children are noted in (Table/Fig 1). Major clinical manifestations and laboratory parameters are shown in (Table/Fig 2). The lowest median (range) of platelet counts studied according to the diagnostic groups were: Group 1 82,000(20,000-1,65000) and Group 2 40,000(10,000-1,21000). The lowest platelet counts for individual patients are highlighted in (Table/Fig 3). Blood cultures suggested co-infections in five children [Enterobacter (2), mixed growth of Enterobacter and Salmonella typhi (1), Salmonella typhi (1), Escherichia coli (1)]. Twelve children (29%) received platelet and/or plasma transfusions. Platelet count recovery time was 3 and 4 days for the two groups, respectively (p=0.003) (Table/Fig 4). The groups were significantly different among themselves (F-ratio 9.65; p=0.004, Analysis of Variance). Multiple regression analysis revealed the lowest platelet count during course of hospitalization to be a significant independent predictor of platelet recovery time in days (Table/Fig 5). There was a trend for lower platelet counts to cause bleeding (petechiae, haematuria, melena, epistaxis or gum bleeds) [r= -0.41, p= 0.01, 95%CI –0.7 to -0.1]. A platelet count less than 55,000 had a sensitivity and specificity of 72.7% and 72.2%, respectively, for bleeding manifestations (Table/Fig 4).Duration of hospitalization correlated strongly with platelet recovery time(r= 0.82, p<0.0001, 95%CI 0.67- 0.9). Platelet count recovery time also correlated well with improvement in clinical symptoms and signs (r=0.5 to 0.7). However, there was a fair to moderate agreement for clinical symptoms and signs at the time of discharge (k=0.3 to 0.5), except for convalescent rash (k=0.7).

Discussion

We evaluated the platelet count recovery time and utility of clinical signs and symptoms in a cohort of hospitalized children with suspected dengue, managed using WHO protocol, where the natural history of the disease will inevitably be altered by management strategies. We showed that the lowest platelet count at the time or during the course of hospitalization, independently predicted platelet count recovery time (days). Twenty-three children (56%) were noted to have lowest platelet counts at hospitalization, while the rest had a nadir of 1-3 days after hospitalization. A diagnosis of dengue haemorrhagic fever would on an average, result in an added day for recovery of platelet counts (Table/Fig 3) and (Table/Fig 5).This information is especially important for resource-poor settings and countries of South Asia, where out-of-pocket payments are a major form of expenditure on health for hospitalization during acute illnesses (7). In our series of patients, two-thirds (28/41) of patients made an average expenditure of 187 United States Dollars (USD) for hospitalization in the form of out-of-pocket payments (detailed data not shown in the manuscript).

Epidemics of dengue might cause a significant impoverishing effect on the poor in developing countries, and hospitalization in resource-poor settings may be governed not only by the acuity of sickness, but also by the ability-to-pay. In our management of patients during the epidemic of 2006 in north India, the patients were noted to be keen to know the anticipated duration of hospitalization, as it directly affected the total incurred costs. Accurate counseling about the duration of hospitalization in this situation, was thus thought of to be of paramount importance, to help patients make an informed choice and to guide discharge planning during dengue epidemics. From the observations made in the present study, the specific question of patients at the time of hospitalization on the time for platelet count recovery can be answered with reasonable certainty. A similar duration of recovery time of platelet count has been noted by Lum et al (5). They noted the median duration of thrombocytopenia to be 4 and 5 days for patients with Dengue Shock Syndrome during two distinct time periods.

It is important to realize that patients also need to be counseled about the importance of the overall improvement in the clinical picture, as has been suggested by the World Health Organization (WHO). WHO suggests clinical improvement in the form of improved appetite, decrease in fever, absence of bleeding manifestations and platelet counts >50,000/cumm as the discharge criteria for hospitalized dengue patients (9),(12).

An important factor which motivated the present analysis and focus on platelet count recovery time is the subjective nature of clinical symptoms and signs which may lack of reliability. This was highlighted in the present study by only fair to moderate agreement in cimprovement in clinical signs between physicians. The only sign that reached good agreement between physicians was the presence of convalescent rash at the time of discharge. Platelet count recovery correlated well with the improvement in clinical signs and symptoms in our patients. Also, there were no falling trends in platelet counts, seen after discharge in any child. Platelet counts more than 50,000 appeared to be a reliable cut-off criterion, consistent with WHO guidelines for discharge, as bleeding manifestations were seen almost always with platelets less than 55,000 in the present study (Table/Fig 6).

It is important to realize that even though the platelet count recovery was useful in this series, unnecessary chasing of platelet counts may result in inappropriate transfusions of platelet concentrates (4). We were very cautious and conservative in the use of our transfusion practices, and followed a strict guideline as mentioned in the methods section. We also audited our transfusion practices. Out of 12 patients who received transfusions in our series of patients, 11 have evidence of significant bleeding. The mean platelet count for this group was 26,500/cumm (only four patients had platelet counts more than 25,000 but less than 50,000/cumm; even these patients had major bleeding). Seven patients with a platelet count between 28,000-51,000/cumm, with no evidence of bleeding, did not receive any transfusions, reassuring about our adherence to guidelines.

We also did a subgroup analysis to evaluate transfusions in patients with dengue haemorrhagic fever. Significantly, more patients with DHF grade III/IV received transfusions, highlighting the increasing severity of thrombocytopaenia in children with Dengue Shock Syndrome [5/17 (DHF I/II) vs. 5/6(DHF III/IV); 95% CI -0.91- -0.17].

Conclusion

The duration of hospitalization of children with suspected dengue in our cohort, correlated well with platelet recovery time, as well as with improvement in clinical signs and symptoms. Convalescent rash is a reliable sign for discharge planning. Median platelet count recovery time was noted to be 3 and 4 days for dengue and dengue haemorrhagic fever, respectively in children with suspected dengue who are managed with WHO guidelines. This information will be useful for patients with suspected dengue, who are likely to be hospitalized in a resource poor setting.

Acknowledgement

The author acknowledges the help of Dr Mamta Waikar, MD, during data collection and management of children.