Journal of the New Zealand Medical Association, 18-June-2004, Vol 117 No 1196
Economic cost of community-acquired pneumonia in New Zealand adults
Guy Scott, Helen Scott, Maria Turley, Michael Baker
The prime objective of economic evaluations in healthcare is to provide policy analysts and decision-makers with information on the costs and effects of medical interventions. Both incremental costs and effectiveness data are required for these evaluations.
Community-acquired pneumonia is an important cause of morbidity and mortality from infectious disease in developed countries. Both pneumonia incidence and mortality rates increase with age. The elderly and people with co-morbid illnesses have the highest risk of illness or death from pneumonia.1
There are few published studies attempting to quantify national costs of treating this disease. ‘Without such data, it is difficult to assess whether new therapies and treatment strategies are needed to improve patient outcomes’.2 Accordingly, the aims of this study were to estimate the incidence and economic cost of pneumonia in New Zealand adults.
New Zealand age-specific hospital admission rates (average of 3 years 2000–2002),3 2003 population data,4 and international rates of hospitalisation were used to estimate resource utilisations for pneumonia. The target population was adults aged 15 years and older.
Resource utilisations related to the 2003 year. Costs were measured incrementally from a societal perspective and reported in 2003 NZ dollars (NZ$1=US$0.5809 mid-rate June 2003).5
Market prices (exclusive of goods and services tax [GST]) and wages were used as proxies for unit costs. All identifiable transfer payments were excluded from the analysis. Discounting was unnecessary as costs related to a single year.
We have chosen to classify health costs as: (a) direct medical, (b) direct non-medical and (c) indirect.6 [The literature provides a number of different cost classifications. A US authority7 classifies costs as: (a) direct costs (direct healthcare costs and direct non-healthcare costs and the value of patient time for treatment), and (b) productivity costs (changes in production or output). Another system described by British and Canadian authors8 delineates costs by sector or perspective as: (a) health sector, (b) patient and family, and (c) other sectors.]
Hospitalisation data were only sourced from the New Zealand Health Information Service, National Minimum Dataset, and cover public hospital discharges. (These data are adjusted for transfers, patients whose usual residence is outside New Zealand, and inconsistent stays.) Records were extracted for adults with a first (primary) diagnosis of pneumonia (ICD-10-AM 2nd edition J10.0, J12-18).
Age-specific hospitalisation rates were calculated using the average number of public hospital discharges for 2000–2002 (three years) and the mean resident estimated population at June 30 for 2000–2002. To estimate 2003 volumes, these age-specific rates (2000–2002) were applied to the population as at 30 June 2003.
Hospitalisation rates for community-acquired pneumonia suggest that the actual number of cases of pneumonia in the population is between 3 and 10 times the number hospitalised. In developed countries, the proportion of patients with community-acquired pneumonia admitted to hospital ranges from just under 10% to just over 70%, 9–16 (Table 1).
Table 1. Hospitalisation rates for community-acquired pneumonia (CAP)
The proportion of patients admitted to hospital is likely to increase with age. Accordingly, we estimated the proportion of pneumonia cases hospitalised as follows: 20% in those aged 15–64 years, 40% in those aged 65–74 years, 60% in those aged 75–84 years, and 70% in those aged over 85 years (31% over all ages).
To establish the total number of pneumonia episodes, the number hospitalised in each age group was divided by the proportion hospitalised. Episodes of pneumonia managed in the community were calculated by subtracting the number of hospitalised cases from the total number of episodes of pneumonia in the population.
Two general practitioner (GP) consultations were allocated per patient treated in the community, and one consultation per hospitalised admission was evaluated in the base case (we assumed that all hospitalised cases were first seen by a GP). The cost per GP consultation was obtained from Statistics New Zealand.17
A course of medicine for community treatment of pneumonia was defined as an antibiotic and an analgesic for 10 days. A frequently prescribed antibiotic (amoxycillin clavulate) and an analgesic (paracetamol) formed a course of drug treatment. The medicine unit costs were the pharmacy selling price (including dispensing fee) (Hataitai Pharmacy, Wellington. Personal communication, 4 June 2004).
An informal telephone survey of GPs indicated that one sputum test and one chest X-ray could be allowed for each episode of pneumonia but variations in practice suggested a range of values should be investigated in the sensitivity analysis. The laboratory test used for the analysis was ‘sputum (excluding tuberculosis)’.18 Chest X-rays were valued using the price charged by a private provider (Standard chest X-ray, Wakefield Hospital, Wellington. Personal communication 17 April 2003).
Hospital unit costs were derived from diagnosis-related group (DRG) costings. This is a case-mix classification system in which cases with similar costs are categorised within broader groupings relating to the same or similar organ or system of the body.
New Zealand uses the Australian National Diagnosis Related Groups (AN-DRGs). The study used AN-DRG (version 3.1), which provided the average base contract price paid to public hospitals in 2001.19 A weighted average AN-DRG cost was derived by using hospital discharge volume data and the average contract prices paid for AN-DRGs in 2001.
The following AN-DRGs accounted for most of the volume of pneumonia cases: AN-DRG 170 respiratory infections/ inflammations age >54 with complications and/or co-morbidities [w cc]), 171 respiratory infections/inflammations (age >54 without complications and/or co-morbidities [w/o cc] or (age <55 w cc), and 172 respiratory infections/inflammations age <55 w/o cc. This weighted average price was inflated to 2003 prices using the Producers Price Index (Inputs for Health and Community Services).20
Transport costs for each patient visit to a GP (to obtain an X-ray, or to return from hospital) were based on a single trip of 2.5 kilometres by private motor vehicle at $0.62 per kilometre.21 Base case private motor vehicle trips were: 2 trips for each GP consultation, 2 trips per X-ray in the community, and 1 trip home for those hospitalised. We used an ambulance call-out charge of $61.3322 for the cost of transport to hospital. For ambulance call-outs, the base case was 1 trip per hospitalised case.
Production loss and leisure time foregone for all pneumonia patients was assessed as follows: 2 weeks for those treated in the community, and 3 weeks for hospitalised cases (1 week in hospital, 2 weeks at home). Patients’ time for consultations and X-rays was assumed to occur within the period off work. Productivity and leisure time foregone was valued using average weekly earnings for males and females combined for June 2003 ($539 per week).23
Univariate sensitivity analysis (to determine the main cost drivers) involved increasing variables of interest by 10% (holding all other factors constant) and recording the change in total costs induced. Multivariate sensitivity analysis using Monte Carlo sampling24 was conducted to assess the impact of simultaneous changes in key assumptions.
We estimated that there were 26,826 episodes of community-acquired pneumonia in New Zealand adults in 2003; a rate of 859 per 100,000 population (Table 2). In those persons aged 65 years and over, the incidence of pneumonia was 1,882 per 100,000. Pneumonia caused 8,278 hospitalisations (265 per 100,000) in New Zealand adults.
It was found that pneumonia in New Zealand incurs direct medical costs of $29 million ($1,095 per episode), direct non-medical costs of $1 million ($26 per episode), and productivity loss of $33 million ($1,244 per episode). Total costs amounted to $63 million or $2,366 per episode (Table 3).
The cost per case treated in the community was $1,280 compared with the total cost of a hospital treated case of $4,800 (Figure 1).
Table 2. Incidence of community-acquired pneumonia in New Zealand adults in 2003
Univariate sensitivity analysis demonstrated that the two major cost determinants were productivity and hospital costs. When productivity loss rose by 10% (holding all other factors constant) total costs increased by 5% and when hospital admissions rose by 10% total costs rose by 4% over the base case result. If these cost drivers were altered by 20% instead of 10% the resultant impact on total costs doubled.
For example, when productivity loss rose by 20% over the base case, total costs increased from 5% to 10%. When the productivity loss of those aged 65 years and over was excluded, total costs fell by 15% (Table 4).
Figure 1. Treatment pathways for community-acquired pneumonia in New Zealand adults
Table 3. Economic cost of community-acquired pneumonia in New Zealand adults (base case)
Figure 2. Multivariate sensitivity analysis
The multivariate sensitivity analysis using Monte Carlo methods found that when the simulation model was run through 10,000 iterations total costs ranged between $56 million at the 5th percentile to $72 million at the 95th percentile (in other words, there is a probability of 90% that an estimate would fall between the 5th and 95th percentiles). (Table 4 and Figure 2).
Table 4. Sensitivity analysis of the cost of community-acquired pneumonia in New Zealand adults
There are no actual recorded data on the total incidence of pneumonia (and the number of cases treated in the community in New Zealand or internationally).25 Accordingly, all estimates of incidence must be based upon recorded hospital discharge data.
The overall incidence of pneumonia in the population is likely to be much higher than indicated by hospitalisation data (as the majority of patients diagnosed with pneumonia are managed at home).26 Similarly, there are no accurate data on the community management and treatment (diagnostic tests and drugs used).27
We employed Monte Carlo analysis to investigate uncertainty in our estimates of incidence of pneumonia and cost of treatment. As we did not have sufficient information to specify the distributions relating to the inputs under investigation, the multivariate sensitivity analysis used triangular distributions. A triangular distribution assumes that there is an equal likelihood of sampling any value between the ‘low’ and ‘most likely’, and an equal probability of selecting any number between the ‘high’ and ‘most likely’ values.
As leisure time and paid (and unpaid) productive activity have value for all people, we estimated these time costs in the same manner for all ages. In many instances, an elderly patient may require care provided by another family member who cannot be otherwise employed. Thus, the inclusion of productivity and leisure time loss can be justified on both willingness-to-pay and opportunity cost criteria. If analysed from a societal perspective, these indirect costs are as important as the hospital costs.
If indirect costs were confined to those patients in paid employment, productivity costs would be substantially reduced. However, this would imply that unpaid production and leisure time activities are not valued by society. The frictional cost method28–30 was not used to estimate productivity loss—as we wished to evaluate the full potential cost of production and leisure time foregone (and did not wish to deduct an allowance for the degree of scarcity of labour in the economy). Our cost estimates are conservative, as we did not estimate intangible costs relating to quality of life or death.
The univariate sensitivity analysis showed that hospital costs were one of the two major cost determinants. The major generator of direct medical costs for pneumonia is the number of hospitalisations, which will become increasingly important as the population ages.
The results of this cost of illness study indicate that prevention and community treatment programmes should focus on the 65 years and older age groups. Potential savings in hospital costs alone would justify this approach.
Author information: Guy Scott, Economist, Massey University and ScottEconomics Ltd, Wellington; Helen Scott, Economist, ScottEconomics Ltd, Wellington, Maria Turley, Epidemiologist, Ministry of Health, Wellington; Michael Baker, Public Health Physician, Wellington School of Medicine, University of Otago, Wellington
Acknowledgements: We thank Craig Wright, Statistician Ministry of Health (who extracted the hospitalisation data from the National Minimum Dataset for us) and the unknown referee (who provided valuable information and suggestions that enabled us to improve our paper).
Correspondence: Dr Guy Scott, Department of Applied and International Economics, Massey University (Wellington Campus), Private Bag 756, Wellington. Fax: (04) 801 2794; email: G.Scott@massey.ac.nz
issue | Search journal |
Archived issues | Classifieds
| Hotline (free ads)
Subscribe | Contribute | Advertise | Contact Us | Copyright | Other Journals