zyxw zyxwvutsrqpo zyxwvutsrq DrlS COMMENT The Cost of Insulin-dependent Diabetes Mellitus (IDDM) in England and Wales zyxwvutsrq A. Grayarb, P.FennaTc and A. McCuireard Oxford Centre for Health Economics Research”, Centre for SocioLegal Studies, Wolfson College, Oxfordb, School of Management and Finance, University of Nottingham‘, and Department of Economics, City University, London, U P This study estimates the direct health and social care costs of insulin-dependent diabetes mellitus (IDDM) in England and Wales in 1992 to be €96 million, or €1021 per person in a population with IDDM estimated at 94000 individuals. These costs include insulin maintenance, hospitalization, CP and out-patient consultations, renal replacement therapy, and payments to informal carers. Expenditure i s concentrated on younger age groups, with one-third of the total expended on those aged 0-24. Around one-half of the total costs can be directly attributed to IDDM, with the remainder associated with a range of complications of the disease. The single largest area of service expenditure is renal replacement therapy. The cost estimates are most sensitive to incidence rates of IDDM, numbers on dialysis and average duration of dialysis. A further €113 million pounds may be lost each year due to premature deaths resulting in lost productive contributions to the economy. The direct and indirect costs of IDDM are therefore significant. The cost of illness framework presented here should facilitate the economic evaluation of new and existing treatment regimens, which may improve value for money by reducing costs and/or increasing the quality or quantity of life for people with IDDM. KEY WORDS Insulin-dependent diabetes mellitus Cost of illness Health economics Introduction further research i s required. The methodology of the study differs from previous studies of diabetes in adopting an incidence-based approach to estimate the population with IDDM, and its mortality. zyxwvutsr zyxwvuts Diabetes mellitus i s one of the most common chronic diseases, and has major economic as well as health implications for patients, health services, and society in general. Laing’ used primary diagnosis as a basis for costing with the finding that direct costs were f83.4 million and disability costs f60.9 million in 1979. Gerard et estimated the total cost of diabetes in 1984 to be at least f259 million. Finally, Laing and Williams3 found that the total costs of care of people with diabetes in England and Wales were f484 million in 1986/7, or almost 5 % of total expenditure on the National Health Service (NHS). However, none of these studies has attempted to distinguish between insulin-dependent (Type 1 ) diabetes mellitus (IDDM) and non-insulin-dependent (Type 2) diabetes mellitus (NIDDM),although each has a different epidemiology and therefore different economic and policy implications. The aim of this study i s to provide economic data relevant to decisions about the prevention and treatment of IDDM, by estimating the direct and indirect socio-economic costs of IDDM in England and Wales in 1992 using a cost-of-illness approach. Comparative information for other diseases is also provided. Finally, the study illustrates some of the difficulties involved in establishing the actual costs associated with IDDM, and highlights areas in which Correspondenceto: Dr A.M. Gray, Oxford Centre for Health Economics Research, Wolfson College, Oxford, OX2 6UD, UK. 1068 CCC 0742-3071/95/121068-09 0 1995 by John Wiley & Sons, Ltd. Methods and Data Cost of illness (or ’burden of illness’) analysis is a method of calculating the resources used to prevent, detect, and treat a disease, in absolute terms and in relation to other diseases. The approach has been proposed as a way of informing priority setting in biomedical r e ~ e a r c h ,and ~ can help to inform or monitor programmes of prevention and treatment. It has been applied to a range of different diseases, including Alzheimer’s di~ease,~ epilepsyt6 mig~ a i n e ,multiple ~ sclerosis,B ~ t r o k e , diabetes ~ mellitu~,~ and benign prostatic hyperplasia.’O Such studies invariably include a range of direct costs to the health and personal social services. Some studies have also attempted to estimate indirect costs such as the loss to society of productive output because of sickness absence from work, early retirement, reduced productivity or mortality. This raises a number of methodological problems, for example concerning the treatment of those not in formal employment. In this study, indirect costs have been estimated, but are presented separately from the direct costs.This ensures comparability with the majority of other studies, and focuses on the data most relevant to decision-makers. In recent years increasing attention has also focused on the role of informal carers and the burden of illness DIABETIC MEDICINE, 1995; 12: 1068-1076 zyxw DTT7 on relatives and friends. We recognize this as a potentially important burden to measure, and have attempted to quantify the informal care burden to the extent that it is recompensed by the Social Security system. There are two main ways of conducting cost-of-illness studies. The first is a ’top-down‘ approach, based on aggregate data on mortality, hospital admissions, GP consultations etc. This approach takes advantage of available national data, but may give a misleading picture if the incidence of the disease in question i s changing or has changed in the past. It also relies on aggregate information being classified such that the disease of interest is readily identifiable. The second main method is a ’bottom-up’ approach, in which data on disease incidence and prevalence are combined with information on disease and treatment probabilities to construct an estimate of the annual incidence of a range of treatments with associated costs. This approach can make full use of the international epidemiological data on incidence, and is more flexible, but also more complex to set up. This study makes use of a variant of the bottom-up approach. The prevalence of I D D M has been estimated using available information on incidence and survival probabilities. The likelihood of hospitalization or of other formal health care associated with I D D M or its complications has then been applied to the estimated total population of patients with I D D M to derive an estimate of the annual volume of treatment. From this, costs of treatment have been attached, giving a total annual cost of illness associated with IDDM. COMMENT Pittsburgh,’ a seven-fold excess mortality was applied to the England and Wales all-population age-specific mortality rates to obtain an estimate of the survival of patients with IDDM. This generated an estimate of the total population with IDDM alive at a point in time, and their age distribution. Life years lost and working years lost were calculated by assuming that each individual dying from a cause attributable to IDDM would otherwise have lived for the remaining population life expectancy. Morbidity and Health Service Utilization At least four major categories of complications accompany diabetes mellitus: cardiovascular and peripheral vascular complications, ophthalmic complications, neurologic complications, and renal complications. Studies of the incidence and prevalence of these complications frequently do not differentiate between patients with I D D M and NIDDM. Retinopathy i s thought to be more common among people with IDDM than NIDDM; diabetic neuropathy may be more commonly associated with I D D M than NIDDM; renal complications are relatively common among all patients with diabetes, but the population with IDDM has been estimated to be up to 15 times more likely to experience end stage renal disease than those with NIDDM.’4-’8 There i s very little information on the link between diabetic morbidity and health service utilization. However, an American study has reported age-specific hospitalization rates and lengths of stay for diabetes and its complications in comparison with the non-diabetic p o p ~ l a t i o n . ’The ~ study does not distinguish between hospitalizations for IDDM and NIDDM, but does nevertheless represent the most detailed source of information on the link between morbidity and hospital utilization, and has been used in the baseline calculations of the present study. We have assumed that hospitalization rates for diabetes and its complications in England and Wales are the same as in the USA. This assumption seems reasonable given that acute hospital sector hospitalization rates for all conditions in England and Wales are almost identical to those in the USA: 1.29 and 1.34 per thousand, respectively, in 1986.20 For hospitalizations directly attributable to diabetes we have made use of actual age-specific English lengths of stay derived from Health Service Indicators. For complications of diabetes, we have assumed that UK lengths of stay were the same as in the USA: again, this assumption i s supported by evidence that, for all acute admissions, the average length of stay in England and Wales (7.8 days in 1986) is virtually the same as in the USA (7.6 days). O n the basis of the review of diabetic complications above, we have assumed that the relative risk of hospitafization is the same for patients with IDDM as for the entire population with diabetes for vascular disease and neurological complications, and double the relative risk of the entire population with diabetes for zyxwvutsrqpon zyxwvut Epidemiology of IDDM Reliable information on the incidence of IDDM i s comparatively rare, especially for adult age groups. A recent study used prospective registration of new diagnoses of diabetes and hospital discharge and death certificates to estimate the incidence of I D D M in the the Oxford health region for the population under 21 during 1985/6.” These incidence figures have been used as the baseline figures in this study, and applied to the total population of England and Wales aged under 21, thus generating a cohort of individuals with IDDM. It is therefore assumed that the incidence of I D D M falls to zero from the age of 21 onwards. Sensitivity analysis was performed using the 95 % confidence intervals from this study, and using data from a study in Rochester, MinnesotaT2which reports incidence rates of I D D M for all adult as well as juvenile age groups. This latter study was not considered appropriate for the baseline analysis because of the comparatively small size of the sample (75 cases compared to 237 in the Oxford study), the country of origin, and the age of the data (1945-69). The subsequent survival of this cohort was then calculated using available information on the excess mortality risk of individuals with I D D M compared to the population as whole. In line with the results of a follow-up study of almost 2000 patients with IDDM in COSTS OF IDDM IN ENGLAND AND WALES zyxwv 1069 COMMENT zyxw ophthalmic and renal complications. These assumptions have all been tested using sensitivity analyses. In addition to in-patient utilization, people with IDDM are much more likely to experience end-stage renal disease, and therefore to be in need of renal replacement treatment, which may or may not be hospital based. There is substantial evidence that many patients with diabetes in the UK who are suitable for renal replacement treatment do not receive it;2' nevertheless, treatment rates have been increasing and the European Dialysis and Transplantation Association (EDTA) report that in 1989 10.4 Yo of patients in the UK accepted onto renal replacement treatment had insulin-dependent diabetes.22, 2 3 In 1990 the rate at which new patients were accepted for renal replacement therapy in the United Kingdom . ~ ~ would indicate was 60.7 per million p ~ p u l a t i o n This 3053 new patients in England and Wales, of whom 305 would have IDDM. This is the baseline figure used in this study. Once accepted for renal replacement therapy, it has been assumed for simplicity that all patients receive continuous ambulatory peritoneal dialysis (CAPD), the most common treatment modality for people with diabetes. The duration of dialysis is taken to be 7 years, which is the mean life expectancy of high risk patients receiving independent dialysis according to survival analyses of UK patients on the EDTA registry.24 Finally, a major item of health care is the routine maintenance of insulin. In this study it is assumed that all people with IDDM follow a standard regimen involving routine blood glucose testing, subcutaneous injection of 42 units of insulin a day with a fresh hypodermic needle daily, and routine out-patient consultations at 4-monthly intervals. We therefore assume that insulin dependency is synonymous with insulin treatment amongst people with IDDM, and therefore that any 'honeymoon' period i s so short that it has no major effect on insulin maintenance costs. DT17 was up-rated to €42.50 in 1992 prices.25 The average cost of a primary care consultation consists of two components: the time and services of the GP and the costs of any pharmaceuticals prescribed. In 1990-91 the cost per consultation averaged €9.46, and the average cost of a prescription was f6.72. This gives a total of €16.18, or f17.81 in 1992 prices.'" The annual cost of renal replacement therapy using CAPD was estimated at €13000 in 1990; 27 this was updated to f 15 558 in 1992 prices. Assuming an average duration of 7 years therapy, and discounted at 6 Yo per annum so that all costs are given in present values, this gives a dialysis cost averaging f86851 per individual. Finally, the costs of routine insulin maintenance were obtained from the 1992 tariffs for appliances and from September 1992 MIMS. These gave annual costs of €77.70 for blood glucose testing strips, f24.09 for hypodermic needles, €1 45.50 for insulin, and f 127.50 for out-patient attendances, a total of €374.79 per person with IDDM per year. Surveys have confirmed that informal caring is a major contributor to health care and may have substantial financial consequences for carers.28,29 However, there is no easy way of placing a value on services of such carers, who mainly go unpaid. Here, we include only informal care which is recompensed through the social security system via Attendance Allowance or Invalid Care Allowance, plus Mobility Allowance paid directly to those with IDDM. These benefits are paid for specific adaptations and aids, or for services rendered, and are not strictly transfer payments which are normally excluded from cost of illness studies as they comprise money passed from one part of the population to another as part of a general redistribution, with no net resource consequences. It has been assumed that these payments are three times as likely to be made to those with IDDM as to the population as a whole. Age-specific payment rates to the England and Wales population were therefore trebled and applied to the age structure of the population with IDDM. This estimate is varied in the sensitivity analysis. To estimate the indirect costs associated with IDDM, the average annual earnings of all male and female manual and non-manual workers, of €12925 in 1990, were multiplied by the average economic activity rate (that is, the proportion of the population in or seeking employment) for the relevant age group, and then by the age-specific unemployment rate,3o to obtain an estimate of the likely annual earnings that an individual who died would otherwise have received from paid employment. This was then multiplied by remaining working years at death, and discounted at the Treasuryapproved public sector rate of 6 'YO per annum, to obtain an estimate of the lifetime earnings loss from each death. Finally, this was multiplied by the total number of deaths from IDDM to obtain an estimate of the total annual productive loss from mortality attributable to IDDM. It was assumed that those who died before reaching zyxwvuts zyxwvutsrqp zyxwvuts zyxwvutsrqp Resource Use and Costs Information was collected from various sources on the use of health care resources attributable to each associated morbidity, to establish the direct costs of IDDM. The average costs attributable to treatment episodes were then used to calculate total direct costs. All costs are expressed in 1992 prices, the most recent year for which all price data are available. Cost data for particular diagnostic categories are not routinely available in the NHS at present. Consequently we use the national average cost per in-patient day in acute or mainly acute This was €145 in 1989-90 (€913 per case with an average length of stay of 6.3 days), and was updated to f 177 in 1992 prices using the Hospital and Community Health Services Pay and Price Inflation Index, as supplied by the Department of Health. We estimated the cost of an average out-patient attendance using the Department of Health's figure of f27.85 in 1986/7 for acute non-teaching hospitals; this 1070 zyxw zyx A. GRAY ET AL. zyx Dm zyxwvutsrqp zyxwvutsrq COMMENT working age would otherwise have worked an average working life with average wage rates, activity rates, and unemployment rates. Total Expenditure Related to IDDM Tables 2 and 3 show the total health and social care expenditure related to IDDM, by disease category (that is, expenditure directly attributable to diabetes, or expenditure related to various complications) and by agegroup, respectively. Table 2 shows that the total direct costs of IDDM in 1992 came to f95.6 million. The major costs of IDDM are those directly attributable to the disease; these account for €47.3 million, or 49 o/o of the total. Almost half of these directly attributable costs are associated with routine insulin maintenance (€23.1 million, made up of €13.6 million on insulin, and €9.5 million on blood glucose testing strips and syringe/needles). In addition most of the f12.5 million spent on out-patient consultations directly attributable to diabetes are routine consultations associated with insulin maintenance. The second most important disease category in expenditure terms is renal complications, which account for a further €30.2 million. Vascular complications account for just under 5 %, of total costs, and ophthalmic and neurologic complications each contribute around 1 O/O of the total. The single most expensive area of service provision is renal replacement therapy, which annually accounts for f26.5 million or almost 28 Yo of the total costs of IDDM. Routine insulin maintenance and hospital in-patient costs are other major areas of service provision. Annually we estimate a total of 19612 in-patient admissions, an admission rate of 209 per 1000 population with IDDM. Throughout the analysis we make a baseline assumption that each hospital admission is associated with 1 outpatient attendance and 3 GP consultations. In addition, it is assumed that routine insulin maintenance involves 3 out-patient attendances per annum. Thus the total out patient consultations comes to 300354, at an annual cost of €12.8 million, or 13.3 Yo of total costs. Total GP consultations come to 37690 annually, costing approximately €1 million, or 1.1 o/o of the total costs associated with IDDM. zyxwvutsrqponm zyxw zyxw Results Incidence and Prevalence Table 1 reports the total population of individuals with IDDM estimated in this study at 93 581. Using previously published estimates of the prevalence of IDDM and NIDDM combined,” it can be calculated that our estimated population with I D D M is approximately 19 Yo of the total number of people with diabetes in England and Wales, accounting for the entire population with diabetes below the age of 20, falling to 18 Yo in the 5059 age group and less than 0.1 YO in the 70-79 age group. life Years and Potential Working Years 1ost The baseline analysis gives a total of 2014 deaths per annum from IDDM, resulting in an annual loss of 50993 life years, and 23691 potential working years. These figures are equivalent to 0.8 Yo of all life years lost, and to 1.8 YO of all potential working years lost. IDDM is therefore responsible for approximately the same proportion of total life years lost as are all diseases of the genitourinary system (0.9 Yo), and for a larger proportion than are all infectious/parasitic diseases (0.7 YO) or musculoskeletal diseases (0.7 %). The larger impact on potential working years lost reflects the severity of IDDM and its early onset. The annual deaths from I D D M estimated in this study result in a life expectancy at birth of 51 years compared to 76 in the population as a whole, implying an average of 25 years of life lost by each individual with IDDM: this result is very similar to the estimates given by P a n ~ r a m . ~ ’ Table 1 . IDDM in England and Wales: baseline estimated population and prevalence zy Age group Population of England and Wales (000s) Diabetes prevalence per 1000 Total population with diabetes IDDM population IDDM prevalence per 1000 IDDM as YO of total population with diabetes All 0-9 10-19 2 0-2 9 30-3 9 40-49 50-59 60-69 70-79 50 955 6 646 6315 8143 7 070 6814 5 292 5 040 3 701 1932 9.8 0.5 2 4 4.7 7.5 13.6 20.8 35.6 31.3 499 359 3 323 12 630 32 572 33 229 51 105 71 971 104 832 131 755 60 471 93 581 4 344 17191 19177 18 297 16 726 12615 5 048 182 1.84 0.65 2.72 2.36 2.59 2.45 2.38 1 .oo 0.05 0.00 19 100 100 59 55 33 18 5 0 0 >80 COSTS OF IDDM IN ENGLAND AND WALES 0 1071 COMMENT zyxw zyxwvu Dm Table 2. Total health and social security expenditure related to IDDM by disease category (f millions 1992) Disease category Routine insulin maintenance Hospital costs Directly attributable to diabetes 93 581 23.1 11 .o 0.7 12.5 47.3 Vascular compl ications 93 581 4.5 0.1 0.1 4.8 Ophthalmic complications 93 581 1 .o 0.06 0.04 1.11 Neurological complications 93 581 1.2 0.03 0.03 1.2 Renal complications 93 581 Total 93 581 Percent General practitioner consulation costs Out-patient consultation costs Cost of social security non-transfer payment costs Population with IDDM Renal replacement therapy costs Total costs 3.5 0.1 0.09 26.5 30.2 23.1 21.1 1 .o 12.8 11 .o 26.5 95.6 24.2 22.1 1.1 13.3 11.5 27.7 100 zyxwvutsrqp Finally, Table 2 summarizes the costs of social security payments attributable to IDDM. In total, we estimate that f l l million annually arises under this heading, of which 45 % consists of mobility allowance payments (f4.9 million), 39 70 consists of attendance allowance (f4.3 million), and the remaining 16 % consists of invalid care payments ( f l . 7 million). Table 3 shows a similar breakdown of the total costs of IDDM, but by age group. The largest share of expenditure is incurred on younger age groups, with the 0-24 age group responsible for f32 million or 33 % of the total, compared with just f6.2 million attributable to the 65+ age group, or 6 Yo of the total. However, on a cost per person basis, the highest costs are incurred in the oldest age group (f3686 per person per year, compared to f853 per person per year in the 25-44 age group). In total, the €96 million annually expended on GP, hospital, out-patient and social security related services for people with IDDM works out at €1021 per person with IDDM per year. of IDDM. The table indicates that the annual total of just over 2000 deaths among the population of IDDM results in potential earnings losses equivalent to f 11 3 million in present values. This loss peaks in the 50-54 age group. Sensitivity Analysis Table 4 shows the estimated indirect costs-that is, costs arising from lost production due to premature mortality- Each of the main assumptions used in estimating the cost of illness of IDDM has been varied in the model to assess the sensitivity of the baseline results to these assumptions. Table 5 reports the results of this sensitivity analysis with respect to the total cost of IDDM. Reducing the excess mortality risk of IDDM increases the size of the population with IDDM and its annual hospitalizations, and therefore increases the total burden associated with IDDM to approximately f102 million annually. Increasing the excess mortality risk has the opposite effect, reducing the burden to f85 million annually. Using the 95 % confidence intervals around the baseline incidence rates,” the total annual cost of IDDM varies from f77 million to €113 million. Incidence is therefore the key parameter within the model. Making use of the American incidence data for all Table 3. Total health and social care expenditure related to IDDM, by age-group (f millions 1992) zyxwvutsrq Indirect Costs Age group 0-24 2544 45-64 65+ Total 1072 Population with IDDM 1227 36443 24215 1695 93 581 Total costs of insulin maintenance, excluding out-patient attendances Hospitalization costs 7.7 9.0 6.0 0.4 23.1 8.9 6.3 5.4 0.5 21.1 General practitioner consultation costs Out-patient consultation costs 0.6 0.2 0.2 0.1 1.1 4.4 4.9 3.2 0.2 12.8 Cost of social security non-transfer payments Cost of renal replacement therapy 1.7 3 .0 5.4 0.9 1 1 .o 8.5 7.5 6.2 4.2 26.5 Total cost Total cost per patient with IDDM (fs) 31.8 31.1 26.4 6.2 95.6 1019 853 1091 3686 1021 zyxw A. GRAY ET AL. Drn zyxw COMMENT Table 4. Indirect costs associated with IDDM zyxw zyxwv zy Age group (yr) Annual deaths from IDDM Remaining working years Average economic activity rate (1990) Average unemployment rate (1 990) Average annual earnings, adjusted for activity and unemployment rates Discounted annual lost earnings (f) Discounted total lost earnings (f millions 1992) 0-4 45-49 50-54 55-59 60-64 65-69 70-74 75-79 19 5 12 39 41 42 51 72 119 198 307 401 405 266 36 0 47 47 47 47 42 37 32 27 22 17 12 7 2 0 0 0 0.635 0.635 0.635 0.635 0.79 0.785 0.785 0.785 0.785 0.732 0.732 0.668 0.546 6.8 6.8 6.8 6.8 6.2 6.2 5.5 5.5 5.3 5.3 6.0 6.0 6.0 7650 7650 7650 7650 9577 951 7 9588 9588 9608 8959 8893 81 16 6633 49 755 66 584 89 104 119241 145811 140 247 135 035 126 660 115697 93 870 74 559 45 305 12162 0.9 0.3 1.1 4.6 6.0 5.9 6.8 9.1 13.8 18.6 22.9 18.2 4.9 Total 201 3 5-9 10-14 15-1 9 20-24 25-29 30-34 35-39 40-44 113.3 Table 5. Sensitivity analyses of main assumptions used in study on direct costs Assumption Baseline assumption Population with IDDM Annual hospitalizations Total direct costs (f millions 1992) 93 581 19612 95.6 102.4 85.4 Baseline - 5-fold excess IDDM mortality risk 7-fold 100934 20 764 12-fold excess IDDM mortality risk 7-fold 81 535 17 694 IDDM incidence at lower 95% CI IDDM incidence at higher 95% CI Rochester all-ages IDDM incidence data m idpoin t 68 956 11 7861 92 853 14 382 24 761 17 086 113.5 98.4 midpoint Oxford incidence data 77.4 IDDM hospitalization risk no different to all diabetic population 2-fold excess for ophthalmic and renal complications 93 581 17 848 93.1 English average length of stay 20 YO above US figure no difference 93 581 19612 98.9 4 general practitioner consultations per hospitalization 3 93 581 19612 95.9 3 Out-patient consultations per hospitalization IDDM = 5 YO of new dialysis cases 1 10 Yo 10 Yo 7 years 7 years 3 x average 93 581 19612 97.3 93 581 19612 82.3 93 581 93 581 93 581 19612 19612 108.9 89.1 19612 101.4 93 581 19612 88.3 93 581 19612 102.9 zyxwvutsrqp IDDM = 15 YO of new dialysis cases Average duration of dialysis = 5 years Average duration of dialysis = 9 years Likelihood of taking up benefits same as population average Likelihood of taking up benefits 5 times population average COSTS OF IDDM IN ENGLAND AND WALES 3 x average 1073 COMMENT zyxw zyxwvutsrq zyxwvutsrqp age groups reported in Melton etal.,I2 the total population with IDDM is very similar to that estimated using the Oxford incidence data, at 92 853 compared to 93 581 : this is because the reported incidence rates among children and adolescents in the American survey are substantially lower than in the Oxford study, offsetting almost exactly the additional cases arising from adult incidence rates. The total annual cost of I D D M using the American incidence data rises by 3 % to €98 million, the increase being due mainly to the older age distribution and hence higher health care costs of the population with IDDM which the American incidence data generates. Assuming no differential between the IDDM and NIDDM populations in hospitalization risk for any complication of diabetes, the total cost falls to €93 million per annum. Altering the assumptions concerning length of stay, GP consultations and out-patient consultations produces relatively small changes in the total cost, which then varies between f96 million and €99 million per annum. However, the total cost is much more sensitive to the values used for the numbers on dialysis and the average duration of dialysis: altering the baseline assumption that 10 YO of all new dialysis patients have IDDM to 5 YO and 15 Yo (in effect, halving or increasing by 50 Yo the number of patients with I D D M receiving dialysis) results in total costs respectively falling to €82 million or rising to €109 million. Similarly, reducing or increasing the average duration of dialysis by 2 years alters the total cost to €89 million and €101 million, respectively. Finally, altering the assumption that people with I D D M are three times as likely as the population as a whole to receive the three social security benefits included in the analysis also alters the costs substantially: from €88 million if there i s no difference between people with IDDM and the population as a whole, to €103 million if the likelihood increases to five times that of the general population. Discussion and Conclusions This study has estimated that the total cost of illness associated with I D D M is f96 million annually. Using sensitivity analysis, the maximum range around this estimate is from €77 million to €1 13 million. The results are most sensitive to the assumptions concerning the incidence of IDDM. Table 6 places these results within the context of a number of previous cost of illness studies. These studies refer to different base years, but the reported costs are all recalculated in the table in 1992 prices to improve comparability. The greatly differing financial burden associated with different disease categories is clear. As Table 6 also shows, previous cost of illness studies have focused on NHS hospital and primary care. It has been relatively uncommon to incorporate community care, formal care from non-NHS sources such as local authorities, private and voluntary sectors, or informal care provided by friends and relatives. The table shows that 1074 DTT7 some diseases such as Alzheimer's disease and stroke are major users of health and social resources. The resource implications of IDDM do not compare with these major diseases, but are nevertheless significant, exceeding a number of other diseases and health problems such as multiple sclerosis, migraine, and benign prostatic hyperplasia. In addition to these direct costs of IDDM, the premature mortality associated with the disease gives rise to indirect costs, that is, productive losses measured in terms of likely earnings, of approximately €1 13 million per annum. Few British estimates exist of indirect costs associated with other diseases: as Table 6 shows, comparable losses from benign prostatic hyperplasia have been estimated at between €3 million and €14 million per annum. The baseline estimate of the direct burden associated with IDDM i s equivalent to around 15 YO of the total burden which at least one study has estimated for all d i a b e t e ~ .Thus ~ the financial burden of I D D M appears to be slightly lower than the crude prevalence of the disease might indicate: our baseline estimate suggests that I D D M accounts for approximately 19 OO/ of the total population with diabetes. This discrepancy arises primarily because the population with NIDDM is on average much older, and it is in these older age groups that the heaviest health care costs tend to be incurred. This younger age composition of the population with I D D M compared with those with NIDDM may also explain why in our study the single most expensive area of service provision was found to be renal replacement therapy (accounting for almost 28 YOof total costs) rather than cardiovascular complications, which in American studies of the total population with diabetes have been by far the most costly c ~ m p l i c a t i o n . 'In ~ our study, all vascular complications account for only 5 O/O of total costs. There are a number of ways in which the analysis could be refined, for example, by measuring G P consultations independently of hospitalizations, using age-specific excess risk ratios, and estimating less crudely the takeup of social security benefits. We would also like to explore ways of extending the analysis into the nursing home sector. Despite these possible refinements, the methodology used in this study produces results which are reasonably robust with respect to the assumptions made, and also allows the effects of alternative assumptions to be considered. It should be noted that the direct costs to the health sector identified in this study refer only to patients with I D D M who have complications of diabetes which are being treated. There is substantial evidence that people with diabetes who have end-stage renal disease have in the past been less likely than people who do not have diabetes to receive renal replacement therapy, and various estimates have been made of the need for and cost of additional treatment f a ~ i l i t i e s . ~It~should , ~ ~ also be stressed that questions about appropriate levels of expenditure cannot be answered by cost of illness studies, which indicate only what the cost is, not whether it is zyx zyxw zyx A. GRAY ET AL. z zyxwvu COMMENT Table 6. The estimated total cost of a range of diseases, England and Wales, in 1992 f s NHS hospital care Disease group -~ General practitioner care Community and non-NHS care Total direct costs lndirect costs Total direct and indirect costs 81 9 n.i." 143 n.1. 34.1 n.1. 11 n.i. 1101 838 660 138 96 70-92 36 27 n.i. 1101 838 660 138 209 74-1 09 36 27 ~ Alzheimer's disease Stroke Diabetes Epilepsy IDDM Benign prostatic hyperplasia Multiple sclerosis Migraine 2 78 81 8 446 106 60.4 61-80 20 3 4 20 71 32 1.1 9-1 2 5 24 n.i. n.i. n.i. 113 4-1 7 n.i. n.i. 'n.i., not included in study. too low or too high. Answers to such questions demand information on the costs and effectiveness of particular treatment^.^^ Such cost-effectiveness analyses, however, may be facilitated and encouraged by cost of illness studies. The present study also provides pertinent background information for those interested in IDDM, which remains an important health problem touching on many different aspects of the health and social service sectors, as well as on the lives of individuals and of society as a whole. 11 12 13 McPherson CK. Economic burden of treated benign prostatic hyperplasia in the United Kingdom. Br j Urol 1993; 71: 290--296. Bingley P, Gale E. Incidence of insulin dependent diabetes in England: a study in the Oxford region. Br M e d J 1989; 298: 558-560. Melton LJ, Palumbo PJ, Chu C-P. Incidence of diabetes mellitus by clinical type. Diabetes Care 1983; 6: 75-86. Dorman, IS, Laporte RE, Kuller LH, Cruikshanks KJ, Orchard TJ, Wagener DK, et a/.. The Pittsburgh InsulinDependent Diabetes Mellitus (IDDM)Morbidity and Mortality Study: mortality results. Diabetes 1984; 33: 271-276. American Diabetes Association. Diabetes: 1991 Vital Statistics. Alexandria, Virginia: ADA, 1991. Lloyd C, Orchard T. The prevelance of complications in IDDM in the U S . Practicaf Diabetes 1991; 9: 171-173. Alberti K, DeFronzo R, Keen H, Zimmet P, eds. lnfernational Textbook o f Diabetes Mellitus. Chichester: John Wiley, zyxwvutsrqp 14 15 Acknowledgements Earlier versions of this paper were presented to a seminar at the University of Oxford Nuffield Department of Clinical Medicine, and to an IBC Diabetic Medicine conference. We are grateful to participants at both events, and to two anonymous referees for a number of constructive comments. Amylin Europe Ltd provided material support for the study. References 16 1992. 17 Neil HAW, Mather HM, Thompson AV, Thorogood GH, Hill RD, Mann JI. The Oxford Community Diabetes Study: 18 19 Evidence for an increase in the prevalence of known diabetes in Great Britain. Diabetic M e d 1987; 4: 539-543. Neil HAW, Gatling W, Mather HM, Thompson AV, Thorogood M, Fowler GH, eta/. Diabetes in the elderly: the Oxford community diabetes study. Diabetic M e d 1987; 4: 539-543. American Diabetic Association. Direct and lndirect Costs of Diabetes in the United States in 1987. Alexandria, Virginia: ADA, 1989. Organization for Economic Co-operation and Development (OECD). Health Data File. Paris: OECD. Joint Working Party on Diabetic Renal Failure of the British Diabetic Association, Renal Association, and the Research Unit of the Royal College of Physicians. Treatment of and mortality from diabetic renal failure in patients identified in the 7985 United Kingdom survey. Br M e d J 1989; 299: 1 135-1 136. European Dialysis and Transplantation Association. Combined Report OR Regular Dialysis and Transplantation in Europe, Vol. XX. London: EDTA, 1989. Beech R, Gulliford M, Mays N, Melia J. Epidemiologically Based Needs Assessment: Renal Disease. London: Department of Health, 1992. Wood IT, Mallick NP, Wing AJ. Prediction of resources needed to achieve the national target for treatment of renal failure. Br M e d J 1987; 294: 1467-1470. Department of Health. Health and Personal Social Service Statistics 1989. London: HMSO, 1989. zyxwvutsrqp zyxwvutsrq zyxwvutsrqpon 1 Laing W. The cost of diet-related disease. In: Turner M, ed. Preventative Nutrition and Society. London: Academic Press, 1981. 2 Gerard K, Donaldson C, Maynard A. The cost of diabetes. Diabetic M e d 1989; 6: 164-1 67. 3 Laing W, Williams R. Diabetes: A Model for Health Care Management. London: Office of Health Economics, 1989. 4 Black D, Pole JD. Priorities in biomedical research: indices of burden. Br J Prev Soc M e d 1975; 29: 222-227 5 Gray AM, Fenn PF. Alzheimer's disease: the burden of the illness in England. Health Trends 1993; 25: 31-37. 6 Griffin J, Wyles, M. Epilepsy: Towards Tomorrow. London: Office of Health Economics, 1991. 7 Blau IN, Drummond MF. Migraine. London: Office of Health Economics, 1991. 8 O'Brien B. Multiple Sclerosis. London: Office of Health Economics, 1987. 9 Dale S. Stroke. London: Office of Health Economics, 1989. 10 Drumrnond MF, McGuire A], Black NA, Pettigrew M, COSTS OF IDDM IN ENGLAND AND WALES 20 21 22 23 24 25 1075 COMMENT zyxw zyxwvutsrqp zyxwv zyxwvutsrq Department of Health. The Government’s Expenditure Plans 1992-93 to 1994-95. Cmd 1913. London: HMSO, 1992. 27 West R. Organ Transplantation. London: Office of Health Economics, 1991. 28 Green H. Informal Carers. General Household Survey, 1985. London: HMSO, 1988. 29 Glendenning C. The Costs of Informal Care: Looking Inside the Household. London: HMSO, 1992. 30 Office of Population Censuses and Surveys. Social Trends 1993. London: HMSO, 1993. 31 Panzram G. Epidemiological data on excess mortality and 26 1076 32 33 life expectancy in IDDM-a critical review. Exp C/in Endocrinol 1984; 83: 93-1 00. Feest TG, Mistry CD, Grimes DS, Mallick NP. Incidence of advanced chronic renal failure and the need for end stage renal replacement treatment. Br Med 1 1990; 301 : 897-900. Diabetes Control and Complications Research Group (DCCT).The effects of intensive treatment of diabetes on the development and progression of long-term complications in insulin-dependent diabetes mellitus. New Engl 1 Med 1993: 329: 977-986. zyx A. GRAY E l AL.