Estimating utility scores in young children with acute rotavirus gastroenteritis in the UK

Abstract

Objective: To estimate utility scores for different severities of acute rotavirus gastroenteritis in children aged <5 years in the UK.

Methods: UK general practitioners (n=25) and paediatricians (n=25) rated four different health state descriptions of acute rotavirus gastroenteritis using the EuroQol (EQ-5D) questionnaire for children aged <18 months and 18 months to 5 years. EQ-5D scores were modified to account for limited self-care and mobility, and converted into utility values using the standard algorithm using UK data.

Results: General practitioners rated the mean utility for primary care cases at 0.781 (standard deviation (sd) 0.263) and 0.688 (sd 0.345) for the younger and older age groups, respectively. For hospitalised cases the corresponding scores were 0.425 (sd 0.243) and 0.200 (sd 0.386). Paediatricians rated the mean utility for hospitalised severe cases at 0.595 (sd 0.171) and 0.634 (sd 0.217) in the younger and older groups, respectively, and for hospitalised very severe cases at 0.256 (sd 0.251) and 0.077 (sd 0.340), respectively. In all cases, the utility differences between the health states were statistically significant (p<0.0001).

Conclusions: Acute rotavirus gastroenteritis substantially impairs quality of life in children aged <5 years as rated by health professionals. This study provides useful quantitative utility estimates for economic evaluations.

Keywords: :

• paediatric vaccination
• rotavirus
• utility

Introduction

Acute rotavirus gastroenteritis is a highly contagious viral disease that is most common during the winter and mainly affects infants and young children less than 5 years old. The main symptoms of acute rotavirus gastroenteritis are vomiting, fever and profuse watery diarrhoea, which may result in serious dehydration1,2. It has been estimated that almost every child will be infected with the virus before the age of 5 years3. Over 600,000 children will die annually from rotavirus-related illness worldwide4. Most of the deaths (over 80%) occur in the developing world3.

In industrialised countries death from acute rotavirus gastroenteritis is rare, but the disease burden is substantial. For example, rotavirus is responsible for 50% of hospital admissions for acute gastroenteritis in children aged <5 years in Australia5. In the UK, the number of children aged <5 years hospitalised for acute rotavirus gastroenteritis is estimated to be as high as 17,000 a year, or 5.2 per 1,0006. Other developed countries report similar rates; 7.5 per 1,000 in Australia5, and 3 per 1,000 in the European Union7.

Many rotavirus infections are hospital acquired. A study in a UK paediatric hospital estimated that rotavirus was responsible for 19% of healthcare-associated acute gastroenteritis8, and across the European Union countries 21% of inpatient cases of rotavirus gastroenteritis were hospital acquired7. Hospitalised cases of acute rotavirus gastroenteritis have been estimated to cost approximately € 900– €1,800 per case in four European Union countries (Belgium, France, the Netherlands and the UK)9. Acute rotavirus gastroenteritis is also a substantial burden on primary care, accounting for up to 29% of the visits to general practitioners (GPs) for infectious intestinal disease in children aged <5 years, or over 150,000 GP visits per year in the UK10.

There is no specific treatment for rotavirus infection4, and the aim of clinical management in most cases is the prevention of dehydration1. However, oral rehydration therapy can be difficult to administer successfully in children with severe vomiting, which is common in acute rotavirus gastroenteritis4, and rotavirus gastroenteritis occurs mainly in the winter, when health services are already under pressure. This makes vaccination an attractive option to prevent rotavirus-related illness and hospitalisations6,8,11, with the potential to considerably reduce the associated morbidity and healthcare costs in both primary and secondary care.

A recent study in infants in their first 2 years of life in six European countries showed that the vaccine RIX4414 (Rotarix^™*) was highly effective, reducing acute rotavirus gastroenteritis episodes of any severity by 87%12. RIX4414 is a monovalent vaccine derived from the most common human rotavirus strain, G1P13. It provides cross-protection against most other serotypes and is given in two oral doses13. A second rotavirus vaccine, RotaTeq^™†, a pentavalent vaccine based on a bovine strain (WC3), has also demonstrated efficacy and is administered in three oral doses 13,14. The research presented in the current paper is applicable to both vaccines. When deciding whether to introduce and fund mass rotavirus vaccination programmes, healthcare providers will require data on the cost effectiveness of vaccination as well as on safety and efficacy. Cost-utility analysis, in which health benefits are expressed in quality-adjusted life years (QALYs), is a widely accepted approach for assessing the cost effectiveness of healthcare technologies15, and is applied by bodies such as the UK National Institute for Health and Clinical Excellence16. Calculation of the potential gain in QALYs from implementing a vaccination programme requires a health-related quality of life (HRQoL) weighting or utility value for acute rotavirus gastroenteritis health states, on a scale between 0 (death) and 1 (full health).

Such utility values are presently lacking, because of the difficulty of obtaining those values from young children and/or their direct environment. A study in Canada presented as an abstract has estimated utility values for children with acute rotavirus gastroenteritis by proxy assessment using the Health Utilities Index Mark 2 (HUI2), but did not distinguish between different severities of the illness17. A recently published study in Germany has estimated HRQoL in young children with diarrhoea by proxy assessment using a visual analogue scale18. To the authors’ knowledge, no previous study has estimated utility scores for different severities of acute rotavirus gastroenteritis using a recognised HRQoL instrument with a validated method for converting the scores into utility values.

The objective of the present study was to estimate utility values for various severities of acute rotavirus gastroenteritis in children aged <5 years in the UK. Results from the study have been presented at the International Society for Pharmacoeconomics and Outcomes Research 9th Annual European Congress in 200619.

Patients and methods

Utility was rated by 25 GPs and 25 paediatricians as proxy respondents, as infants and young children would be unable to complete a HRQoL questionnaire. Physicians were selected instead of parents as they were considered more likely to be able to distinguish between different severities of acute rotavirus gastroenteritis. All respondents were working within the UK National Health Service and had a minimum of 5 years and maximum of 25 years of experience. They were drawn from five geographical regions of the UK (Scotland, Wales, Southeast England, Midlands and North/Northeast England).

Currently there is no disease-specific HRQoL questionnaire developed for diarrhoea in children and few specific instruments are designed to assess the generic HRQoL in children aged <5 years20. HRQoL in young children can be assessed using parents or physicians as proxy respondents20,21. Several tools are available for measuring HRQoL and utility22,23. For the present study the authors decided to use the EuroQol (EQ-5D) questionnaire24, completed by healthcare professionals acting as patient proxies with the necessary clinical experience to rate the health state of children with acute gastroenteritis. The EQ-5D was chosen because it is widely used, was designed to be applicable in multiple countries, and the rating scores derived from it can be converted to utility values using a standard country-specific algorithm25. The EQ-5D rates quality of life in five dimensions or domains: mobility, self-care, usual activities, pain/discomfort and anxiety/depression24. Respondents were presented with health state descriptions, representing the clinical presentation of different severities of acute rotavirus gastroenteritis, and asked to rate each health state for infants aged <18 months and children aged 18 months to 5 years. Two age bands were used because most children aged 18 months or over are able to walk and so the mobility domain of the EQ-5D is more relevant in this age group. Health state descriptions were checked by a GlaxoSmithKline physician (Dr Norman Begg) with both clinical and public health experience.

GPs do not routinely test for rotavirus infection and so were presented with two health state descriptions of acute infectious gastroenteritis, one describing a case severe enough to be referred to hospital, and the other describing a case that could be managed in primary care. The GPs were not asked to decide whether they would refer each case, the criterion of referral was explicit in the health state descriptions presented (Table 1). The health state descriptions were based on the major symptoms of acute rotavirus gastroenteritis described in the clinical literature, including frequency of liquid diarrhoea, presence of fever and/or vomiting, the risk of dehydration and the ability of parents to manage rehydration at home1,2,26,27 (Table 1). All GPs were presented with the same two health state descriptions.

Table 1. Health state descriptions presented to respondents.

General practitioners
Primary care only	Referred to hospital
>Acute gastroenteritis NOT severe enough to warrant referral for hospital admission	Acute gastroenteritis severe enough to warrant referral for hospital admission
Liquid diarrhoea (loose stools, non-bloody), 1–3 episodes/day	Liquid diarrhoea present (more than usual loose stools, non-bloody), 4 or more episodes/day
Vomiting either not present or infrequent	Vomiting may or may not be present
Fever may or may not be present	Fever may or may not be present
Dehydration may be present BUT parents/carers are able to manage rehydration at homeORNo signs of dehydration present and risk of dehydration is LOW	Signs of dehydration are present AND parents/carers are unable to manage rehydration at homeORThere is a high risk of dehydration (>4 vomits/day; >8 liquid stools/day; child aged <6 months)
Paediatricians
Hospitalised, severe	Hospitalised, very severe
Acute rotavirus infection severe enough to warrant admission or at least short-term observation	Acute rotavirus infection severe enough to warrant admission or at least short-term observation
Vesikari score ≥10	Vesikari score ≥10
Signs of 3–8% dehydration present OR high risk of dehydration	Signs of ≥9% dehydration present
Parents unable to manage rehydration

Vesikari score described in reference 2.

Paediatricians were presented with two health state descriptions of acute rotavirus infection with a Vesikari score2 of at least 10, which indicates a clinical severity sufficient to be considered for admission to hospital. The two health states were differentiated by the severity of dehydration present (Table 1). All paediatricians were presented with the same two health state descriptions.

Respondents were asked to rate the health status of a child in each of the two age bands and each of the described health states in relation to the child's normal capability in full health. If they felt a domain was not applicable for a child of the given age band, they could mark the domain ‘not applicable’. Where a respondent marked a domain ‘not applicable’, this was assigned a default rating of 1, meaning ‘no impairment’. In the main analysis, the self-care and mobility domains were assigned a rating score of 1 for children aged <18 months, and the domain of self-care a rating score of 1 in children aged 18 months to 5 years, regardless of whether respondents rated these domains. This modification was applied because children aged <5 years would normally have limited capacity for self-care and children aged <18 months would normally have limited mobility, so it may not be valid to attempt to rate the impact of acute rotavirus gastroenteritis on these domains. A secondary analysis considered the raw data scores, without these modifications, and both sets of data are presented here.

The raw and modified EQ-5D scores were converted to weighted utility values using a published algorithm25. This assigns each EQ-5D score a utility value based on a survey of a representative sample of the UK population using the time trade-off method25.

Descriptive statistics were compiled, including mean, standard deviation, median and 95% confidence intervals (CI). Non-normality was tested using kurtosis and skewness, and confirmed by the Kolmogorov-Smirnov test. Comparisons between different age groups were performed using the Wilcoxon Signed Rank test, as the data were not normally distributed. However, reporting focuses on mean values, as these are of more interest for economic evaluation than medians. p-values ≤ 0.05 determined statistically significant differences. Descriptive summary statistics were analysed using Microsoft Excel and the comparative statistics were analysed using Stata (StataCorp, Texas, US) version 9.

Results

A total of 25 GPs and 25 paediatricians participated in the survey. Their distribution across the five geographic regions of the UK is shown in Table 2.

Table 2. Respondent characteristics.

	GPs (n=25)	Paediatricians (n=25)
Number (%) in each region
Scotland	5 (20)	4 (16)
Wales	5 (20)	6 (24)
Southeast England	5 (20)	6 (24)
Midlands	5 (20)	5 (20)
North/Northeast England	5 (20)	4 (16)
Years of experience
Mean	15.8	9.3
SEM	0.95	1.07
Median	18.0	8.0
Number of cases of acute gastroenteritis seen each month
Mean	18.1	20.6
SEM	4.00	4.60
Median	12.0	12.0

GP, general practitioner; SEM, standard error of the mean.

The main (modified) utility scores derived from the EQ-5D ratings provided by GPs are shown in Table 3, with the raw scores for comparison. The mean modified disutility score (1 – the utility score) for a child with acute gastroenteritis managed by the GP was –0.219 in the younger age group (aged <18 months), and –0.312 for children aged 18 months to 5 years, indicating substantial impairment of HRQoL. In both age groups, the more clinically severe health state requiring referral to hospital was associated with a significantly (p<0.0001) lower utility score than less severe gastroenteritis that could be managed in primary care. The raw utility scores showed the same pattern as the modified scores in the main analysis.

Table 3. General practitioner utility scores.

	Age <18 months		Age 18 months to 5 years
	Primary care only	Referred to hospital	Primary care only	Referred to hospital
EQ-5D utility scores (modified)
Mean	0.781	0.425	0.688	0.200
95% CI*	0.678, 0.884	0.330, 0.520	0.553, 0.824	0.049, 0.352
sd	0.263	0.243	0.345	0.386
Median	0.796	0.362	0.760	0.128
Lower quartile	0.760	0.197	0.620	–0.117
Upper quartile	1.000	0.433	0.883	0.620
p-value for difference between health states	p<0.0001		p<0.0001
EQ-5D utility scores (raw)
Mean	0.601	0.102	0.634	0.032
95% CI*	0.434, 0.768	–0.042, 0.246	0.473, 0.794	–0.131, 0.195
sd	0.426	0.367	0.409	0.417
Median	0.760	0.082	0.746	0.079
Lower quartile	0.436	–0.166	0.587	–0.331
Upper quartile	0.883	0.242	0.796	0.336

CI, confidence interval; sd, standard deviation.

Modified = EQ-5D scores adjusted to show no impact in mobility and self-care domains in age group <18 months, and no impact in self-care domain in age group 18 months to 5 years. Raw = actual EQ-5D score from respondents.

* Normal distribution assumption.

Table 4 presents the modified and raw utility scores from the paediatricians’ EQ-5D ratings of acute rotavirus gastroenteritis admitted to hospital. The mean modified disutility scores were –0.405 for children aged <18 months and –0.366 for children aged 18 months to 5 years with severe rotavirus gastroenteritis needing hospitalisation. In the older age group hospitalised with very severe rotavirus gastroenteritis, the lower bound of the 95% CI was below zero, indicating that this health state could have a utility value lower than death. Although the EQ-5D questionnaire cannot be scored at less than zero, the process used to convert the EQ-5D scores into utility values can produce negative values. As with the ratings provided by GPs, the difference in utility scores between the two severities of illness was statistically significant (p<0.0001). The raw data scores displayed the same pattern, with lower utility scores in the more clinically severe health states in both age groups (Table 4).

Table 4. Paediatrician utility scores.

	Age <18 months		Age 18 months to 5 years
	Hospitalised severe	Hospitalised very severe	Hospitalised severe	Hospitalised very severe
EQ-5D utility scores (modified)
Mean	0.595	0.256	0.634	0.077
95% CI*	0.528, 0.662	0.157, 0.354	0.549, 0.718	–0.057, 0.210
sd	0.171	0.251	0.217	0.340
Median	0.689	0.197	0.620	0.048
Lower quartile	0.433	0.099	0.620	–0.117
Upper quartile	0.689	0.433	0.991	0.293
p-value for difference between health states	p<0.0001		p<0.0001
EQ-5D utility scores (raw)
Mean	0.208	–0.208	0.542	–0.093
95% CI*	0.100, 0.315	–0.335, –0.080	0.445, 0.638	–0.242, 0.055
sd	0.275	0.325	0.246	0.379
Median	0.137	–0.331	0.516	–0.163
Lower quartile	0.079	–0.380	0.516	–0.331
Upper quartile	0.208	–0.095	0.621	0.079

CI, confidence interval; sd, standard deviation.

Modified = EQ-5D scores adjusted to show no impact in mobility and self-care domains in age group <18 months, and no impact in self-care domain in age group 18 months to 5 years. Raw = actual EQ-5D score from respondents.

* Normal distribution assumption.

Figure 1 compares the raw and modified utility scores in the younger age group (aged <18 months) for all four health states. The utility score was highest (least impaired) in the health state requiring primary care only, and lowest (most impaired) in the health state representing hospitalisation for very severe rotavirus illness. The more severe of the health states rated by GPs (presenting in primary care and referred to hospital) and the less severe of the hospitalised health states rated by paediatricians both had intermediate utility scores, consistent with their intermediate clinical severity.

Figure 1. EQ-5D utility scores (mean and 95% confidence interval) in age group <18 months.

Figure 2 presents the raw and modified utility scores for the older age group (18 months to 5 years). Consistent with the results for the younger age group, these data also show the highest utility score in the health state requiring primary care only and the lowest score in the hospitalised very severe health state.

Figure 2. EQ5D utility scores (mean and 95% confidence interval) in age group 18 months to 5 years.

The differences between the modified and raw scores were consistently larger in the younger age group (Figure 1) than in the older age group (Figure 2). This is to be expected, as two domains were modified in the younger age group and only one was modified in the older age group. Figures 3 and 4 present the utility scores for the two age groups as box and whisker plots, showing median and interquartile ranges.

Figure 3. Box and whisker plot showing utility scores in age group >18 months.

Figure 4. Box and whisker plot showing utility scores in age group 18 months to 5 years.

Discussion

This study estimated utility scores for different severities of acute rotavirus gastroenteritis in children aged <5 years in the UK, using GPs and paediatricians as proxy respondents. Each group of health professionals rated two health states of differing clinical severity for children in each of two age groups using the EQ-5D questionnaire. The EQ-5D scores were then converted into utility values using the standard algorithm as described by the EuroQoL group25.

As expected, all four health states in both age groups were associated with impaired HRQoL, as indicated by mean and median utility scores. This finding suggests that acute rotavirus gastroenteritis imposes a burden of decreased HRQoL on patients, and consequently that prevention of infection could provide valuable gains in utility. There was some variation between respondents’ scores, as would be expected (Tables 3 and 4). Exploration of the factors underlying such variations, such as potential differences in perception between regions or between rural and urban settings, could be an interesting subject for future study, but is beyond the scope of the present paper. Both GPs and paediatricians rated the more clinically severe health states as having significantly (p<0.0001) lower utility scores than the less severe health states, and this result held true across both age groups. Thus, it appears that the utility score derived from the EQ-5D used in this study is able to differentiate between different severities of acute rotavirus gastroenteritis, and should therefore be able to provide useful estimates of utility for use in economic evaluations.

The results of the main analysis are further supported by the raw scores, which followed a similar pattern of lower utility in the more clinically severe health states. As expected, the numerical difference between the raw and modified scores was greater in the younger age group (aged <18 months), reflecting the fact that two of the five EQ-5D domains (mobility and self-care) were modified in this age group, while in the older age group only the self-care domain was modified. It should be noted that the modifications to the score provide a conservative estimate of the utility impact, as no impairment was assumed in the modified domains, even if the respondents indicated that there was impairment. As a result, the raw utility scores were always lower than the modified scores across all age groups and all health states (Tables 3 and 4, Figures 1 and 2). Thus, this study may have underestimated the true impact of acute rotavirus gastroenteritis on HRQoL. A further potential limitation of the study may be that it only considered the impact of acute rotavirus gastroenteritis on the HRQoL of the patients, without attempting to capture the impact on parents or carers. This would also tend to underestimate the loss of utility attributable to rotavirus infection, as a recent cost-effectiveness study using utility estimates from a Canadian study17 reported that the number of QALYs lost by carers was one of the three parameters with the greatest effect on the modelled results28.

The EQ-5D was developed for use in adults24 and has not been validated in children aged <5 years. However, as reported in a recent review29, no satisfactory generic HRQoL instrument has yet been developed for use in this age group. The ability of the EQ-5D-derived utility scores to discriminate between the different severities of health states presented to the respondents in the present study suggests that it provides a reasonably reliable measure of utility. The study was conducted using proxy respondents because of the obvious difficulties in obtaining responses in this age group. This is consistent with current theory and practice, as the lower age limit for self-reported instruments in children is generally 5–6 years30, and a proxy respondent is the recommended approach21. Healthcare professionals were chosen for this study rather than parents because it was considered that they would be more likely to be able to distinguish between different clinical severities of acute rotavirus gastroenteritis.

There is a clear need for more data on the effect of rotavirus infection on utility scores for use in conducting economic evaluations of rotavirus vaccines15. A recent cost-effectiveness study of rotavirus vaccination in the UK suggested that health service funding of a rotavirus vaccine programme may be considered appropriate if there were a sufficient gain in quality of life for the parents and children involved, but was unable to estimate the potential gain because of the absence of utility data31. A separate analysis in the UK28 estimated QALYs for both patients and parents by applying utility data derived from a Canadian study, although there was no differentiation according to disease severity.

The present study provides the first estimate of the impact of acute rotavirus gastroenteritis on HRQoL in infants and young children that distinguishes between different severities of illness and utilises a recognised HRQoL instrument (the EQ-5D) with a validated method25 for converting to utility scores. As such, it offers a valuable contribution to research on the potential cost effectiveness of rotavirus vaccination programmes in Europe. Economic modelling studies applying these estimates are underway in several European countries, including the Netherlands32, Belgium33, Italy34 and the UK35. These should provide new information to add to previous cost-effectiveness studies of rotavirus vaccination that have been based on the Canadian utility estimates that did not distinguish between different severities of rotavirus illness28,36, or that have considered other outcome measures such as disability-adjusted life years37.

In conclusion, the present study provides quantitative utility estimates for young children with varying severities of acute rotavirus gastroenteritis, which should be useful in economic evaluations of rotavirus vaccines.

Acknowledgements

Declaration of interest: This study was conducted on behalf of GlaxoSmithKline Ltd, Middlesex, UK. The authors thank Carole Nadin who provided medical writing services and Michael Aristides for his advice and input to the design of the study.

Notes

^*Rotarix is a trade mark of the GlaxoSmithKline group of companies.

^†RotaTeq™ is a trade mark of Merck & Co.