What is known about this topic
- • Cost of diabetes from provider perspective in Thailand
- • Cost of diabetes from patient perspective in Thailand
- • Cost of different medical services availed by the diabetic patients in Thailand
Diabetes is a common metabolic disorder with increasing burden in Thailand. The chronic nature along with associated complications makes the disease very costly. In Thailand, there exist some studies on cost of diabetes; however, those studies estimated the cost either from provider or from patient perspective. In order to capture the complete picture of economic burden caused by diabetes, using prevalence-based approach; the present study estimated the cost of illness of diabetes from societal perspective, the broadest viewpoint covering all costs irrespective of who incur them. Data were collected from 475 randomly selected diabetic patients who received treatment from Waritchaphum hospital in Sakhon Nakhon province of Thailand during 2007–2008 with a response rate of 98%. A micro-costing approach was used to calculate the cost. The direct medical cost was calculated by multiplying the quantity of medical services consumed by their unit costs while indirect cost was calculated by using human capital approach. The total cost of illness of diabetes for 475 study participants was estimated as USD 418,696 for the financial year 2008 (1 USD = 32 THB). Of this, 23% was direct medical cost, 40% was direct non-medical cost and 37% was indirect cost. The average cost of illness per diabetic patient was USD 881.47 in 2008 which was 21% of per capita gross domestic product of Thailand. Existence of complications increased the cost substantially. Cost of informal care contributed 28% of total cost of illness of diabetes. Therefore, the disease not only affected the individual but also the family members, friends and neighbours. The economic and social burden of the disease therefore emphasises the need for initiatives to prevent the disease prevalence and counselling to the diabetic patients to prevent the progression of the disease and its devastating complications.
Diabetes Mellitus is one of the important public health concerns of Thailand in recent years (Rawdaree et al. 2006, Nitiyanant et al. 2007). Recent economic changes of the Kingdom, reflected by rapid industrialisation, urbanisation and increased wealth at both national and household levels, has led to an increasing proportion of Thai population living with diabetes. The 2003–2004 health examination survey revealed that there was a threefold increase in diabetes prevalence among the Thai population during 1991–2004 (from 2% to about 7%) (Ministry of Public Health Thailand 2009). While measuring the health status of Thai people using Disability Adjusted Life Years as an indicator, it was found that diabetes ranked eighth and third for males and females respectively in 2004. The hospitalisation rate for diabetes in Thailand had shown a rising trend over the years, from 33.3 per 100 000 population in 1985 to 91.0 in 1994 to 380.7 in 2003 and 586.8 in 2006 (Ministry of Public Health Thailand 2009). Hence, Thailand is inevitably moving towards the burden of such a public health problem.
People with diabetes are prone to consequences in both short-term and long-term complications. The chronic nature of diabetes and its devastating complications make it a very costly disease. In the United States, total estimated cost of diabetes in 2007 was USD 174 billion while in Latin America and the Caribbean, total annual cost associated with diabetes was estimated as USD 65 216 million in 2000 (Barcelo et al. 2003, American Diabetes Association 2008). In Thailand, studies on the cost of diabetes estimated the cost either from provider or from patient perspective (Putsook 1999, Moolasarn et al. 2005, Pongcharoensuk et al. 2006, Riewpaiboon et al. 2007). As per the knowledge of the researchers, no study had yet been conducted in Thailand to explore the cost of illness of diabetes from societal perspective, the broadest viewpoint covering all costs irrespective of who incurs them. The present study was an attempt towards this end.
Based on epidemiological data used, cost of illness studies can be either prevalence based or incidence based. Prevalence-based cost of illness studies measure the economic burden of a disease for a given period of time, generally a year, whereas the incidence-based approach measures the economic burden from the onset of disease until cure or death (Pagano et al. 1999). The present study was a prevalence-based cost of illness study and it presented a societal perspective of cost of diabetes (Kobelt 2002). In societal perspective, all costs, regardless of who incurred them are included. Thus, costs to the healthcare service, to social services, to patients and also to the rest of the society in the form of production losses are included, but transfer payments (payments made without any corresponding production such as pension) are ignored.
In Thailand, district hospitals were the majority and covered about 92% of all districts in 2007. The proportion of 30-bed hospitals in all district hospitals was 56% (Ministry of Public Health Thailand 2009). The present study was conducted in one such 30-bed district hospital, namely Waritchaphum Hospital at Waritchaphum district of Sakhon Nakhon Province in north eastern Thailand. The study participants were the diabetic patients who received treatment from the study hospital and the health centres under the study hospital during the financial year 2008. In Thailand health system, health centres are the public healthcare facilities at the subdistrict level that provide primary health-care, health promotion and prevention (no inpatient services). The patients were identified according to International Classification of Diseases, tenth revision (ICD-10 codes = E10–E14).
As the present study was the first study which estimated cost of diabetes from a societal perspective in Thailand, there was no sufficient background information to calculate a formal sample size for this study. Based on available budget and time constraints (Kelley et al. 2003) and sample size used in other cost of illness studies in Thailand (Riewpaiboon et al. 2007, 2008a,b), a sample size of 475 (one-third of all diabetic patients who received treatment from the hospital and health centres during the financial year 2008) was fixed for the present study. A total of 1415 diabetic patients (inpatients and outpatients taken together) received treatment from the study hospital and health centres in 2008. Assuming non-availability and refusal rate of 15%, 546 patients were randomly selected from the electronic diabetic patient database of the study hospital and were targeted to be sequentially contacted at their homes for consent and their time for interview. The study team stopped contacting patients as soon as complete data were available from 475 diabetic patients.
In this study the cost components consisted of both direct and indirect costs. The direct costs reflected the resources used in treating or coping with the disease, including expenditures for medical care and the treatment of illness. Direct cost was divided into two subcategories – (1) direct medical costs which included costs of hospitalisation, outpatient visits, drugs, laboratory tests, materials, emergency services (such as dressings for diabetic patients), dental services and traditional medicine services (e.g. foot massage for diabetic patients who had absence of foot pulse) and (2) direct non-medical costs which included cost of transportation to the healthcare providers, time loss of the patient and the accompanied person for visiting healthcare providers, costs of meals and accommodation during these visits, costs of personal facilities needed (e.g. home modifications, personal devices) and cost of informal care (the care provided by the family members, friends, relatives and neighbours without financial compensation). Indirect cost included the societal cost of morbidity, permanent disability and premature mortality. The direct medical costs are mostly provider costs while direct non-medical and indirect costs are costs to the patients.
Unit cost of medical services was measured directly from the study hospital records. Medical service utilisation data which included inpatient stay, outpatient visit, emergency, laboratory, traditional medicine, dental services, drug use, materials use etc were collected by the study team members by reviewing medical records of every study participant for the study period (October 2007 to September 2008), while information on direct non-medical cost and indirect cost was collected from all study participants through direct personal interview method. A structured questionnaire was administered to the study participants to obtain their information. The final version of the questionnaire was divided into four main sections. The first section collected demographic information of the study participants while the second section collected data on ability of the study participants to perform routine jobs. Information on direct non-medical and indirect cost was collected in the third section. Direct medical cost information at other health facilities was also collected in this section. The final section dealt with informal care. Content validity of the questionnaire was assessed by the experts and pretested before administration. The field data collection was done by a team of trained health centre and hospital staff of Waritchaphum district during January 2009 to March 2009. At the end of each interview, the researchers checked the completeness of data collection and the incomplete questionnaires were sent back to the field.
The direct medical cost was calculated by multiplying the quantity of medical services consumed by their unit costs. The standard costing method was used to calculate the unit cost of medical services at Waritchaphum hospital (Brouwer et al. 2001). At first, all departments of the hospital were classified into 12 patient care cost centres and 13 supporting cost centres. The direct cost of each cost centre was calculated by summing up its labour, capital and material costs. After deriving the direct cost of all cost centres, the direct cost of the supporting cost centres were allocated to patient care cost centres by employing the simultaneous equation method (Drummond et al. 2005). To calculate the unit cost of services at the patient care cost centres, averages were taken for those cost centres that produced homogeneous products such as outpatient department, inpatient department and pharmacy. However, for the cost centres producing heterogeneous products (e.g. laboratory, radiology, emergency room), a micro-costing approach was used (Brouwer et al. 2001). Micro costing is a method of allocating the cost of production of the cost centre to each unit of service. The first step was to value the resources directly consumed by each unit of service. Then the shared cost of the cost centre was allocated to the services in proportion to the direct cost of the services (Riewpaiboon et al. 2007).
The estimated unit costs of services at health centres in five Thai provinces from a study of Kongsawatt (1997) were used for calculating the cost per visit at health centres. The outpatient visit and inpatient day cost at the provincial hospital was calculated on the basis of the approximate average of four study results which were conducted in different provincial hospitals in Thailand (Tisayaticom 2000, Jawrakate 2001, Pattanaphesaj 2008, Koopitakkajorn 2009). All costs were converted into 2008 prices by using the consumer price index for medical care in Thailand. For the drug and laboratory cost per visit at the provincial hospital, the same costs incurred at Waritchaphum hospital were used on the assumption that the drug and laboratory cost per visit will almost be the same at district and provincial hospital. Unit costs of inpatient stay, outpatient visits and dispensing cost at Waritchaphum hospital, provincial hospital and health centres are presented in Table 1.
|Healthcare providers||Unit cost (USD)|
|Inpatient stay (per day)||74.33|
|Outpatient visit (per visit)||3.58|
|Dispensing (excluding drugs) (per prescription)||1.69|
|Inpatient stay (per day)||117.64|
|Outpatient visit (per visit)||6.60|
|Outpatient visit (per visit)||1.65|
In this study the patients themselves identified their caregivers. The data on time spent for informal care was collected by using the recall method (Van den Berg & Spauwen 2006a). The recall period was 1 month prior to the date of interview. It was hypothesised that the informal care was needed for (1) activities of daily living, e.g. personal care, moving around the house, going to the toilet, dressing, eating etc, (2) household activities of daily living, e.g. preparation of food and drinks, cleaning the house, washing, ironing etc (3) instrumental activities of daily living which include visiting family or friends, healthcare contacts, going to the bank etc and (4) healthcare activities including managing medications, glucose monitoring, foot care etc. Respondents were asked how much time the informal caregivers spent on the above activities in minutes or hours per day or hours per month whichever was appropriate for the specified activities. Informal care was valued with the opportunity cost method (Van den Berg et al. 2006b) which included cost of foregone paid, unpaid and leisure work.
In this study indirect costs associated with diabetes included health related days absent from work and/or normal activities, leisure time loss, lost earning capacity from permanent disability and lost productivity from premature mortality. A human capital approach was used for indirect cost calculation (Pritchard & Sculpher 2000). For calculating work absence and normal activity lost days, the number of such lost days mentioned by the study participants during last 3 months from the date of interview was taken into account (following maximum allowable recall period of 3 months) and then extrapolated for the whole year (Kobelt 2002). For estimating mortality cost, the number of deaths occurring among the study participants during the study period was considered and for permanent disability, the patients who reported during interview that they were out of the labour force because of disability were considered. Their Barthel Index (BI) score also confirmed their severity of disability (Mahoney & Barthel 1965). For calculating loss of productive life for both deceased and persons with permanent disability, the age of 60 years was considered (the official retirement age in Thailand).
The official minimum wage rate of Sakhon Nakhon province (148 baht per day) was used as the base case in the calculations of indirect cost and cost of time loss for the study participants, accompanied persons and informal caregivers (Ministry of Labour 2008). As most of the study participants worked in agriculture (55%) and 31% were out of the workforce because of age, they did not have a regular income. Some may have earned more while others will have earned less than the minimum wage, and setting their income to be equal to the minimum wage was considered appropriate. When a person in the active labour force dies or is out of the labour force because of permanent disability, his contribution to the country’s gross domestic product (GDP) is lost. Hence, a sensitivity analysis was conducted by using GDP per capita in mortality and permanent disability cost calculation in order to capture how the assumption of using minimum wage rate affected total cost of illness of diabetes.
A constant 5% growth rate in the minimum wage rate was used to calculate income in the future years. That was the average percentage increase in minimum wage in Sakhon Nakhon province for last 5 years (Ministry of Labour 2008). The projected GDP growth rate was used from the International Monetary Fund’s (IMF) World Economic Outlook, 2008 report for Thailand. A 3% discount rate was used to convert future earnings to current value. However, this discount rate was varied to 0% and 6% to see the effect on cost (Edejer et al. 2003).
All costs will be presented in this paper in US dollars using an exchange rate applicable at the time of data collection, 1 USD = 32 Thai Baht.
Statistical analyses were performed using SPSS Release 14.0. Descriptive statistics were used to summarise data on demographic characteristics, clinical status and costs. Cost comparisons between groups were analysed by the Kruskal-Wallis (Altman 1991) or the Jonckheere-Terpstra test (Field 2005). When three or more subgroups were ordered and the ordering was expected beforehand to be reflected in the median costs (such as patient with longer duration of disease will have higher median cost), the Jonckheere-Terpstra test was applied as it takes into account the order of subgroups. Statistical significance was considered when P value was less than 0.05. For either the Kruskal-Wallis or the Jonckheere-Terpstra test, if the P value was found less than the chosen significant level, the Bonferroni correction method (Altman 1991) was applied to the P values of all pairwise comparisons (using the Mann–Whitney test) of the subgroups to guard against Type I errors while making multiple comparisons.
The research protocol was approved by the Institutional Review Board of Mahidol University and it conforms to the provisions of the Declaration of Helsinki. Written informed consent was obtained from all respondents who participated in this study.
A total of 483 diabetic patients from Waritchaphum hospital were contacted for interview. Five patients were not available and three refused to be interviewed, with 475 participating (participation rate 98%). Patients’ demographic and clinical characteristics are presented together with the subgroup analysis in Tables 3 and 4. Out of a total of 475 study participants, 354 (74%) were females and the mean age of all participants was 59.3 (SD 11.4) years. Most had primary education (90%) and were agriculturists (55%). As regards the type of diabetes, only four patients were found with type 1 diabetes (1%). The mean duration of the disease was 7.2 (SD 6.4) years and the mean fasting blood sugar level was 156.1 (SD 37.5) mg dL−1. Diabetic complications were found among 31% of the study participants while 65% had co-morbidity. Thirteen (3%) study participants died during the study period and 5 (1%) were found to be permanently disabled as measured using BI score. The mean age of the deceased and persons with permanent disability were 62.5 (SD 11.4) and 50.8 (SD 5.2) years respectively and the mean duration of disease was 9.8 (SD 6.2) and 5.4 (SD 2.9) years respectively.
|Categories||Number (%)||Cost of illness||P value*|
|Less than 35||7 (1.5)||329.47||66.91–1036.46||<0.001|
|75 and above||45 (9.5)||543.45||125.40–899.99|
|Duration of the disease|
|1–5 years||264 (55.6)||110.40||71.52–335.64||<0.001|
|6–10 years||100 (21.1)||148.73||95.28–395.51|
|11–15 years||61 (12.8)||287.25||118.13–984.23|
|16–20 years||30 (6.3)||521.81||114.04–1402.84|
|21 and above||20 (4.2)||718.58||222.69–1837.95|
|Type of diabetes|
|Type 1||4 (1.0)||747.87||465.92–1483.02||0.167|
|Type 2||471 (99.0)||139.51||82.01–513.42|
|Fasting blood sugar level‡ (mg dL−1)|
|180 and above||100 (21.5)||281.06||104.44–843.30|
|With co-morbidity||308 (64.8)||148.70||87.87–548.78||0.072|
|Without co-morbidity||167 (35.2)||125.40||73.64–494.85|
|Number of co-morbidities|
|One co-morbidity||203 (65.9)||138.56||82.74–436.01||0.013|
|Two co-morbidities||85 (27.6)||170.89||96.95–666.42|
|Three or more co-morbidities||20 (6.5)||425.14||129.14–1841.19|
|Number (%)||Cost of illness||P value*|
|With complications||148 (31.2)||479.93||120.43–1552.22||<0.001|
|Without complications||327 (68.8)||115.12||73.54–286.08|
|Number of complications|
|One complication||96 (20.2)||261.32||100.45–826.29||<0.001|
|Two complications||44 (9.3)||758.09||240.28–2749.14|
|Three and more complications||8 (1.7)||2131.34||1091.34–3795.43|
|Types of complications†|
|Micro and macrovascular||11 (2.3)||666.42||201.03–2706.59|
|Microvascular and cataract||23 (4.8)||745.44||376.41–1357.58|
Total cost of illness of diabetes was estimated to be USD 418 696 in 2008 (1 USD = 32 THB) (Table 2). Of this, 23% was direct medical cost, 40% was direct non-medical cost and 37% was indirect cost. While looking at the components of different costs, the major component in direct medical cost was the hospital care or the inpatient service cost which accounted for 11% of total cost of illness. Next came drug and outpatient visit cost (about 3% of total cost of illness). In direct non-medical cost, the major component was the cost of informal care which was 28% of total cost of illness and this was the highest individual cost component. In indirect cost, cost of permanent disability contributed about 19% of total cost of illness while the contribution of mortality cost was about 17%.
|Cost components||Total||% of total COI||Mean||Median||Inter-quartile range|
|Direct medical cost||94878.88||22.6||199.75||62.46||46.90–102.57|
|Other service utilisation cost||5951.03||1.4||12.53||0.00||0.00–2.25|
|Expenditure at other health facilities||3391.72||0.8||7.14||0.00||0.00–0.00|
|Direct non-medical cost||166932.53||39.7||351.44||54.84||24.54–317.22|
|Cost of personal facilities||3118.44||0.7||6.57||0.00||0.00–1.25|
|Time loss of patient||10536.22||2.5||22.18||17.34||11.56–24.28|
|Time loss of accompanied person||3394.98||0.8||7.15||0.00||0.00–2.60|
|Payment to paid caregivers||956.25||0.2||2.01||0.00||0.00–0.00|
|Cost of informal care||116549.49||27.8||245.37||0.00||0.00–164.96|
|Total direct cost||261811.40||62.3||551.18||138.56||81.90–510.48|
|Total indirect cost||156885.04||37.5||330.28||0.00||0.00–0.00|
|Work absence/leisure time lost||5332.63||1.3||11.23||0.00||0.00–0.00|
|Cost of permanent disability||78266.78||18.7||164.77||0.00||0.00–0.00|
|Total cost of illness||418696.45||100.0||881.47||139.79||82.01–522.50|
The cost of illness due to diabetes was also calculated for subgroups of patients identified on the basis of economic and clinical criteria. These results are presented in Table 3. While comparing cost of illness among different age groups (P < 0.001), the highest median cost was found for the age group 75 years and above. After adjusting by the Bonferroni method, it was found that the costs were significantly different for the following pairs of age groups (35–44 years and 55–64 years, adjusted P = 0.002; 35–44 years and 65–74 years, adjusted P < 0.001; 35–44 years and 75 years and above, adjusted P < 0.001; 45–54 years and 75 years and above, adjusted P = 0.001; 55–64 years and 75 years and above, adjusted P < 0.001). There were no significant differences in median cost of illness for male and female patients (P = 0.754). The median cost of illness was found to increase with the duration of the disease (P < 0.001). As expected, median cost was the highest for those with disease duration more than 20 years. After the Bonferroni adjustment, costs were significantly different for those who suffered from the disease for 1–5 years than those who suffered for 11–15 years (adjusted P < 0.001), 16–20 years (adjusted P < 0.001) and 21 years and above (adjusted P < 0.001). Costs were also different for the pair of disease duration 6–10 years and 21 years and above (adjusted P = 0.002). Median cost of illness did not differ significantly for type 1 and type 2 diabetic patients (P = 0.167). Cost of illness was also significantly different for different fasting blood sugar levels (P = 0.002). The highest median cost was found for those whose average fasting blood sugar level was 180 mg dL−1 and above during the study period. After being adjusted, costs were different between the pair of fasting blood sugar level 131–179 mg dL−1 and 180 mg dL−1 and above (adjusted P < 0.001). Co-morbidities are the diseases or conditions that co-exist with the primary disease (diabetes) but they also stand on their own as specific diseases. Among 308 study participants who suffered from co-morbidities, 83 (27%) suffered from hypertension, 54 (17%) from disorders of lipoprotein metabolism and other lipidaemias and 37 (12%) from both hypertension and disorders of lipoprotein metabolism and other lipidaemias. The median cost was found to increase with the number of co-morbidities (P = 0.013; adjusted P = 0.001 between the pair one co-morbidity and three or more co-morbidities).
For identifying the complications status, the clinicians thoroughly reviewed the medical records of all study participants following the definitions of complications used in the Thailand Diabetes Registry Project (Chetthakul et al. 2006, Krittiyawong et al. 2006, Leelawattana et al. 2006, Ngarmukos et al. 2006). Cataracts were observed in 15% of the diabetic patients. The second highest prevalence was diabetic nephropathy (12%) followed by diabetic foot complications (9%), diabetic retinopathy (3%), coronary artery diseases (2%) and stroke (2%).
The costs were compared for different types of complications as well as with and without complications. While comparing the costs among the patients with and without complications, it was found that the median cost for those with complications was significantly higher than those without complications (P < 0.001) (Table 4). Further, the median costs were found to increase progressively with the increase in number of complications (P < 0.001). The Bonferroni adjustment showed significant associations of costs between one complication and two complications (adjusted P = 0.001) and between one complication and three and more complications (adjusted P = 0.001). The costs also varied significantly for different types of complications (P = 0.001). Costs of microvascular complications, microvascular complication plus cataract and micro and macrovascular complications were significantly different from the cost of cataract alone (adjusted P < 0.001 0.002 and 0.008 respectively).
This prevalence-based cost of illness study presented a societal perspective of cost of illness of diabetes in Thailand during the financial year 2008. While comparing mean age and percentage of female patients in all diabetic patients at the study hospital and the study sample, it was found that the mean age for all diabetic patients was 59.7 years while the same for the sample was 59.3 years, percentage of female patients among all were 73% while the same in the sample was 74%. These characteristics of the study participants were also quite similar to those found in other diabetes studies in Thailand. For example, the studies by Nitiyanant et al. (2001, 2007) and Chaikledkaew et al. (2008) found the mean age of diabetic patients to be 59.2 years, 58.2 years and 59.0 years respectively and percentages of female patients in total sample 70%, 72% and 66% respectively. As regards disease characteristics, the mean duration of the disease was 7.2 years, most of the patients had type 2 diabetes (99%) and the mean fasting blood sugar level was 156.0 mg dL−1. All these results are also comparable with other diabetic studies in Thailand as well as in other countries. For example, Nitiyanant et al. (2001, 2007), Riewpaiboon et al. (2007) and Chaikledkaew et al. (2008) found the percentage of type 2 diabetic patients to be 95%, 94%, 96% and 99% respectively in their studies in Thailand. Barcelo et al. (2003) also reported a percentage of 97% for type 2 diabetic patients in a study in Latin America and the Caribbean. Nitiyanant et al. (2001, 2007) noted the average disease duration 6.2 years and 8.7 years respectively and fasting blood sugar level 167.2 mg dL−1 and 150.9 mg dL−1 respectively among the diabetic patients in Thailand.
In this study, the contribution of direct cost in total cost of illness was higher than that of indirect cost. In general, indirect cost contributes more in total cost of illness as compared to direct cost because of the cost components included in the study (Ettaro et al. 2004). In the present study, time loss of the patients and accompanied persons during visits to the healthcare providers, transportation cost and cost of informal care were incorporated which were generally neglected in cost of illness studies of diabetes following societal perspective even in recent times (Barcelo et al. 2003, American Diabetes Association 2008). This study showed that informal care alone contributed about 28% of total cost of illness, hence, exclusion of this cost obviously underestimated the direct cost of illness of diabetes in other studies. Further, in this study, the calculation of mortality and permanent disability cost was done by using official minimum wage rate of Sakhon Nakhon province to average out the income level of the study participants. This probably underestimated the indirect cost in this study. To check how the assumption of using minimum wage rate in mortality and permanent disability calculation affected total cost, a recalculation of these two costs was done by using GDP per capita. It was found that the contribution of indirect cost in total cost of illness increased from 37% to 61% and the total cost of illness increased by 60%. Hence, it is clear that the indirect cost calculation based on minimum wage rate provided the lowest cost of illness estimate and this clarified the reason of dominance of direct cost in total cost of illness. Other cost of illness studies in Thailand also used GDP per capita (Youngkong et al. 2002) and both minimum wage and GDP per capita (Riewpaiboon et al. 2008b) to calculate indirect cost. Riewpaiboon et al. (2008b) noted that the minimum wage provided the lowest estimate of indirect cost. Another important assumption was to use a 3% discount rate to convert future earnings to current value. This rate was varied to 0% and 6% and for that total cost of illness changed by 13% (increase) and 9% (decrease) respectively.
While comparing the prevalence of diabetic complications among the study participants with another primary care based study of Thailand by Nitiyanant et al. (2007), it was found that the prevalence of diabetic retinopathy, cataract and nephropathy was lower in the present study (3% vs. 14%, 15% vs. 22%, 12% vs. 17% respectively) while the prevalence of coronary artery diseases, foot ulcers was higher (2% vs. 1%, 7% vs. 1% respectively). There may have been a lower prevalence of nephropathy and retinopathy in the present study because no tests were done to confirm the existence of the complications; the complications were detected by reviewing medical records of the study participants following the definitions used in other diabetic studies in Thailand. Hence, it is possible that the present study approach failed to capture all types of complications.
The costs for patients with complications were substantially higher than those without complications and the costs were found to increase progressively with the increase in number of complications. These are consistent with other studies which noted that diabetic complications and co-morbidities had a significant positive impact on healthcare costs (Bhattacharyya 1998, Rosenzweig et al. 2002). In Thailand, Riewpaiboon et al. (2007) and Chaikledkaew et al. (2008) noted that total healthcare costs were significantly associated with age, gender, type of diabetes and complications. The present study also found significant association of costs of diabetes with age and complications. It should be noted that the other two studies in Thailand used multivariate analysis to examine those associations; whereas, the present study used univariate analysis which did not adjust the effects of confounders. This could be a reason for not finding association of cost with gender and type of diabetes in this study.
There was only one missing value in the dataset as one respondent could not report the transport cost from home to the healthcare provider. That value was replaced by taking median of all other values of the variable (Heyse et al. 2001). Hence, it can be assumed that the results of the study were not affected by missing values.
The following limitations of the study should be noted. First, the study covered diabetic patients who received treatment at the study hospital and the health centres. However, those who did not require treatment were omitted from the cost estimation and the resulting cost per case may have been inflated. Second, the cost of outpatient visit and inpatient days at the health centres and provincial hospital were calculated on the basis of approximate results of some previous studies. Hence, the direct medical cost did not present the exact cost incurred at these two healthcare facilities. Third, indirect cost calculation did not take into account reduced earnings or productive capacity due to disability – it only considered the early retirement caused from disability. Fourth, the present cost of illness estimate ignored the intangible cost such as pain and suffering from the disease. Finally, it should be noted that the indirect cost may be overestimated if human capital approach is used to calculate such cost and hence, the frictional cost approach is preferred (Koopmanschap 1996, Drummond et al. 2005). However, the information needed to apply the frictional cost approach was not available at the time of the study hence the traditional human capital approach was used for indirect cost calculation.
The present study found that the average cost of illness per diabetic patient was USD 881.47 in 2008 which was 21% of per capita GDP of Thailand. It should be noted that the cost of informal care contributed 28% of total cost of illness of diabetes. Therefore, the disease not only affected the individual but also the family members, friends and neighbours. Hence, the hidden cost associated with informal caregiving should be considered for future analysis of public health consequences of diabetes. As diabetes related morbidity and mortality generate high cost to the society, this fact should also be considered when priorities are set in the healthcare system. The present study also showed that the patients with diabetic complications had to incur substantially higher cost. However, much of this cost associated with the disease is preventable through improved diet and exercise, prevention initiatives to reduce the prevalence of diabetes and its co-morbidities and improved care. Therefore, in Thailand, the healthcare providers may set up interventions such as diabetic patient counselling, pharmaceutical care or disease management to delay the progression of co-morbidities or complications that diabetic patients may possibly have in the future.
The present study was a part of the research project under the fellowship program “Asia Fellows Awards 2008–2009 (Cohort X)” of the first author. The fellowship program was administered by the Asian Scholarship Foundation, Bangkok, Thailand and funded by the Ford Foundation. We express our gratitude to the Asian Scholarship Foundation for providing this opportunity. We are grateful to the National Research Council of Thailand for giving permission of conducting the present study. We thank Faculty of Pharmacy, Mahidol University, Bangkok, for being the host institute of the research project. Special thanks go to the professors and students of Division of Social and Administrative Pharmacy, Department of Pharmacy, Mahidol University for their help and support at various stages of this research. Finally, we are indebted to all staff of Waritchaphum hospital and health centres for their support and co-operation during data collection.
The authors have no conflict of interest to declare.