Observed costs and health care use of children in a prospective cohort study on community-acquired pneumonia in Geneva , Switzerland

QUESTIONS UNDER STUDY: Despite various efforts to estimate cost-effectiveness of pneumococcal conjugate vaccines, only scarce information on the cost burden of paediatric community acquired pneumonia (CAP) exists. The objective of this study was to prospectively calculate direct and indirect costs associated with treatment of CAP from a society perspective in children between 2 months and 16 years of age seeking care at a tertiary hospital in Geneva, Switzerland between December 2008 and May 2010. METHODS: This cost of illness study population comprised children aged from 2 months to 16 years of age seeking care for CAP at the University Children’s Hospital Geneva from January 2008 through May 2010 (a subset of patients taken from a larger multicentre prospective cohort). Hospital-associated costs for episodes of pneumonia were computed according to the REKOLE® system. Nonhospital costs were estimated by parental interviews at baseline and follow-up on day 14. RESULTS: The overall cost for one episode of CAP was 11’258 CHF; 23’872 CHF for inpatient treatment and 1009 CHF for outpatient treatment. Severe pneumonia cases per World Health Organisation (WHO) definition used significantly more hospital resources than non-severe cases: 21’842 CHF versus 3479 CHF (p <0.0001). CONCLUSION: Childhood CAP results in a significant medical cost burden that may have been underestimated in previous cost-effectiveness analyses of pneumococcal vaccine strategies.


Introduction
With an estimated 156 million new clinical cases per year, childhood community acquired pneumonia (CAP) remains a leading aetiology of child morbidity and mortality worldwide and creates a significant cost burden [1].Incidence rates in western countries, based on prospective population-based studies from the 1970s-1980s, are assumed to be around 20-40/1000 in children below 10 years of age [2].The hospitalisation rate has been reported to be between 0.15% and 20.6% of all-cause pneumonia depending on case definition and region [2][3][4].Streptococcus pneumoniae is the most important causative agent of paediatric pneumonia in children and is thought to account for around 40% of cases [5][6][7][8].Routine infant immunisation with 7-valent pneumococcal conjugate vaccine (PCV 7) has decreased rates of pneumonia admissions in young children both in the US and Europe [5][6][7].However, the effect has been partially offset by serotype replacement [9].To expand on the success of vaccine intervention, new vaccines such as a 13-valent pneumococcal conjugate vaccine (PCV 13) have been developed.Increasingly, national recommendations about pneumococcal vaccines have included assessments of cost effectiveness, And Immunisation strategies with PCV 7 have been studied in a large number of countries.Interestingly, especially in Europe, very few reports on the cost burden of childhood pneumonia exist [10,11].Consequently the majority of cost-effectiveness analyses have been based merely on approximate cost estimation and expert panel reviews and hence are lacking in accuracy.The objective of this study was to calculate direct and indirect costs associated with treatment of CAP in children between 2 months and 16 years of age seeking care at a tertiary hospital in Geneva, Switzerland.

Study population / study design
The study population comprised of children aged from 2 months to 16 years of age seeking care for CAP at the University Children's Hospital Geneva from January 2008 through to May 2010.This was a subset of patients taken from a larger multicentre prospective cohort study investigating aetiologies of CAP.The study was conducted at paediatric emergency departments of 3 major hospitals in Switzerland (Geneva, Lausanne and Sion).Cases from Geneva alone were considered for this analysis as hospital-associated costs were only available for this subgroup.For pneumonia cases, inclusion criteria were: ≥2 months and ≤16 years of age, fever (>38 °C) and cough, and increased respiratory rate for age or respiratory distress, and radiographic pneumonia based on a paediatric radiologist's chest X-ray (CXR) read.Exclusion criteria were: chronic lung or heart diseases, immunodeficiency syndrome and hospitalacquired pneumonia.A child hospitalised more than once was counted as having a new case, provided that the child was symptom-free for at least 30 days between presentations.Written informed consent was obtained from the participants before enrolment.Ethical approval was obtained from the Research Ethics Committees of the Hospitals of Geneva, Lausanne and Sion.Case investigation included demographic data (age, sex, and vaccine status), clinical data, chest X-ray, and laboratory tests.At enrolment, parents were interviewed about direct and indirect expenditures caused by the child's illness before hospitalisation.Two weeks after initial enrolment (or after hospital discharge in case of prolonged hospitalisation) parents were again asked about expenditures and indirect costs accrued during the follow-up period.For cases in Geneva, about 17% of patients did not follow-up at 15 days.All hospitalassociated costs were acquired for the entire hospitalisation period.Severe CAP episodes were identified per World Health Organisation (WHO) classification, that is, abnormal respiratory rate for age associated with the presence of ≥1 of the following: signs of respiratory distress (retractions, abdominal breathing, nasal flaring), moderate to severe dehydration, or oxygen requirement.

Cost estimation
This cost of illness analysis was directed from a societal perspective.As such it included both direct and indirect costs.Direct costs were defined as the cost of resources used for treating a CAP, whereas indirect costs were defined as the value of resources lost due to a CAP.Direct medical costs, defined as the costs incurred for treating CAP, were calculated from hospital-associated costs as well as parental interviews.Provider costs, which meant patient-specific hospital-associated costs for both ambulatory treatment in the Paediatric Emergency Department as well as inpatient treatment, were computed according to the REKOLE ® system [12].REKOLE ® is a comprehensive accounting algorithm used by Swiss hospitals to internally compute provider costs associated with outpatient and inpatient visits.Costs were attributed to categories as per REKOLE ® guidelines.Categories included imaging, laboratory, medical and treatment services (including physician care and paramedical care; inpatient drugs; materials such as chest tubes; transport, overhead), nursing care, anaesthesia, intensive care unit, surgical procedures, as well as operating room.The parental report was used to estimate costs associated with medical visits outside of our institution, outpatient medications, childcare, as well loss of productivity.Loss of productivity was computed by multiplying the mean number of work absenteeism in hours by the average daily wage in Geneva in 2008 [13].Outpatient medication prices were calculated using a 2010 Geneva drug price list (including VAT) provided by our pharmacy department.Outpatient medications were recorded in categories (e.g., beta-lactam antibiotic).The smallest package size of the most common prescribed drug from each category was used for calculation.One package was assumed for every child that took the same drugs before and after consultation.Prices of generic formulation were used when available.Patients lost to follow-up were completed as if no further expenses had occurred.

Data analysis
Data were managed and analysed using Excel (Microsoft, version 2007) as well as SAS (version 9.3).For inpatient provider costs, mean and median were calculated.As the distribution of cost data was heavily skewed, non-parametric testing as described by Hahn and Meeker was used to compute 95% confidence intervals for medians [14].Wilcoxon rank test and Kruskal-Wallis test were used to assess for difference across two and three categories, respectively.

Results
From January 2008 to May 2010, 191 treatment episodes of CAP were enrolled into the study in Geneva.Only episodes (n = 176) for whom hospital-associated (outpatient and inpatient) cost data was available were included into the analysis.For all (n = 174) patients one treatment episode corresponded to one case except for one patient who was hospitalised twice and treated one time as an outpatient.For this patient only the first treatment episode was included.Baseline characteristics are displayed in table 1.The admission rate among the study population was 47% (n = 78).The mean hospital stay was 7.9 days (range 1-30, median 5).A total of 12 children (15% of admissions) were admitted to the intensive care unit (ICU).When using a definition for severe pneumonia as per WHO (see method section), 70 (40%) cases could be classified as severe.

Hospital-associated costs
Cost estimates associated with inpatient and outpatient treatment at the Children's Hospital Geneva are displayed ously reported annual incidence rates of 6.6/ 1000 for patients less than two years and 5.0/1000 for children less than 5 years in Switzerland [17]), one would expect 328 CAP cases per year in children aged 2 month to 16 years living in the canton of Geneva.In 2009, 74 cases were included into our study.The remaining pneumonia cases were either treated by primary care providers or not enrolled into the study (provider failed to enrol patient or parents refused participation).Estimating that around 70% of all cases presenting to the Paediatric Emergency Department were included in the study, one could assume that around 106 cases presented to the Paediatric Emergency Department.Consequently, 222 (67.7%) would have been treated at a general paediatrician's office.Setting the cost for a routine consultation at a general paediatrician's office at 250 CHF (including prescribed medications), and based on the results from our study, yearly paediatric CAP-associated inpatient and outpatient treatment costs in Canton of Geneva would be 1'113'390 CHF and 115'394 CHF, respectively.

Discussion
To our knowledge this study is the first detailed report on the cost burden of childhood CAP in Europe.This study is limited by the fact that it is a single-centre study as well as inherent issues of the REKOLE ® system: the system assigns costs based on hospital criteria that may vary across care-centres.Though overall, when comparing our cost data with assumptions used for cost-effectiveness calculations related to introduction of PCV-7 in Switzerland, it appears that costs were probably underestimated.In the costeffectiveness study, cost data was calculated on the basis of resource utilisation by a subset of 114 patients treated at the University Hospital in Geneva between 1991 and 2000 (the patient population is hence very similar to our study).Data was then completed by expert panel consultation [17].Cost assumptions for uncomplicated and complicated pneumococcal pneumonia were 1950 CHF and 7200 CHF, respectively which is below our estimates.One could argue that cost assumptions for pneumococcal pneumonia should be set even higher given that the latter are associated with higher complication rates compared to all causes of pneumonia.Yet, the Swiss cost-effectiveness model by Ess and al. was quite robust to health care cost variations but more sensitive to variations in incidence and case fatality rate [17].The calculated costs in our study were also higher than reports from other western regions.For example, Black et al. allocated USD 1464 (1997 value, 1380 CHF) to one pneumonia episode with radiographic consolidation treated at Kaiser Permanente in California, USA [18].The analyses included medical and non-medical costs.In Germany, outpatient, office-based treatment for pneumonia was estimated at € 77 (in 2000 values); costs of hospital admission for pneumonia at € 2424 based on an average length of stay of 7.9 days [19].Our analysis may overestimate costs for outpatient treatment for pneumonia as it was conducted in a tertiary referral centre which is certainly a limitation of our study.This is why the admission rate was 42% which is significantly higher than that in previous reports (2)(3)(4).Based on our incidence estimates around 63% of cases of uncomplicated pneumonia would have been treated at outpatient paediatric offices-a population that was not included in our study.Based on the design, our cohort only included cases with radiographic consolidation whereas most outpatient pneumonias may be diagnosed by clinical evaluation alone.Furthermore, one could argue that there is a tendency to ordering a greater number of diagnostic tests at a tertiary level university hospital.On the other hand our study excluded children with chronic medical conditions who are at great risk for developing complicated pneumonia.In a study in Germany, Kalies et al. reported that up to 21% cases of invasive pneumococcal disease occurred in patients with chronic medical conditions [20].In the present study we also estimated indirect costs and family expenditure both of which have hardly been reported for childhood CAP.The majority of costs were reflected by direct medical costs.Given the small contribution of indirect medical costs we assume that treating patients lost to follow-up at 15 days as if no further costs occurred introduced no significant bias in our analysis.Only 37% of patients reported leave of absence from work which may appear an underestimation.However, in 2009, one partner was working 49% part-time or less among around 62.3% of couples.Loss of parental income may indeed only be present in around 40% of cases [21].This may be much higher on other countries.In the initial US cost-benefit analyses by Lieu et al, more than half of the projected savings were from reduced work-loss by parents who care for ill children or averted productivity loss due to disability or death caused by pneumococcal disease [22].The overwhelming majority of direct medical costs were determined by hospitalisation.This expected difference in resource utilisation underscores the necessity of reducing admission rates and length of stay whenever possible.Clearly, not all patients with pneumonia require referral to secondary care as evidenced by a recent study from Norway that reported a secondary care consultation of 147 / 100'000 per year for children with CAP [4].Indications for inpatient treatment, besides complications such as hypoxemia and dehydration, have traditionally included parental antibiotic treatment.Several studies, including two in the developed world, have compared parenteral and enteral antibiotics.There is good evidence that oral and parenteral antibiotic treatment may be equivalent in uncomplicated CAP [23][24][25].The picture may increasingly be complicated by the observation that whilst studies show reduction in overall incidence of pneumonia after introduction of PCV-7, rates of severe pneumonia may be rising [26].Complicated, and hence more expensive, pneumonia cases have been attributed to serotype 1, 3, and 19A, among others [27,28].Serotype 1, 3, and 19A are not included in PCV7 but in PCV13.Consequently, there may be a cost-benefit in replacing PCV7 by PCV13.Resistant strains of bacteria, new serotypes of pneumococcus, the increase in empyema and necrotising pneumonia, the interaction between viruses and bacteria, and highly virulent PVLpos S. aureus have proven to be increasingly challenging (9,28).

Conclusion
Childhood CAP continues to be a significant cause of morbidity and results in significant medical cost burden.Cost analysis should play a significant role in evaluation of preventive immunisation strategies and case management.To our knowledge this study is the first detailed report on the cost burden of childhood CAP in Europe and suggests that CAP-associated costs may have been underestimated in previous cost-effectiveness analyses of pneumococcal vaccine strategies.More effort should be made to treat uncomplicated pneumonia on an outpatient basis as this may lead to substantial savings.As a matter of course treatment decisions should mainly be taken on clinical and not on cost considerations, especially in an era challenged by the changing epidemiology of CAP after the introduction of PCV.

Table 1 :
Baseline characteristics of study population and CAP-related outcomes.
Swiss Medical Weekly • PDF of the online version • www.smw.ch

Table 2 :
Hospital costs associated with CAP.
(15)hospital costsEstimated prescription medication costs are outlined in table3.Only four patients were not prescribed antibiotics either before or after treatment at the Children's Hospital Geneva.Interestingly, 20% (35) of patients were prescribed cough syrup and/or nasal decongestants; treatments that were shown to be non-effective or even harmful in patients with respiratory symptoms[15].A total of 15 fathers (Taking all charges together the estimated mean overall cost for one episode of CAP in our study cohort was 11'258 CHF; 23872 CHF for inpatient and 1009 CHF for outpatient treatment.Assuming a population at risk of about 82000(15)and an annual incidence rate of 4.0 cases/1000 for children aged two months to 16 years (based on previ-
childrenSwiss Medical Weekly • PDF of the online version • www.smw.ch