Costing of diabetes mellitus type II in Cambodia
© Flessa and Zembok; licensee Springer. 2014
Received: 10 June 2014
Accepted: 16 September 2014
Published: 1 November 2014
Diabetes Mellitus Type II (T2DM) is a major and growing medical, social and economic burden in the East-Asian country of Cambodia. However, no economic modelling has been done to predict the number of cases and the budget impact.
This paper forecasts the epidemiological and economic consequences of T2DM in Cambodia. The Ministry of Health and related donor agencies are supported to select the most cost-effective interventions against the disease. At the same time this paper demonstrates the relevance and potential of health economic modelling for least developed countries.
We developed a Markov-Model for the specific situation of Cambodia. Data was taken from the scientific literature, grey literature in Cambodia and key-informant interviews.
The number of people living with T2DM is steadily increasing from 145,000 in the year 2008 to 264,000 in the year 2028 (+82 %). In the year 2008 the diagnosed T2DM patients would incur costs of some 2 million US$ to cover all of diabetes treatment. 57 % of this amount would have to be spent for OAD-therapy, the rest for insulin therapy. In the year 2028 this amount will have grown to some 4 million US$. If all patients (incl. non-diagnosed) had to be paid-for the respective figure would be 5.5 million and 11 million US$.
Screening for T2DM is only cost-effective if the sensitivity of the test is high while the unit price is low. The results of this simulation call for targeting the high-risk groups. However, an increased availability of Oral Anti-Diabetic and Insulin Therapy is highly cost-effective.
Type 2 Diabetes Mellitus is a major public health challenge in Cambodia. The simulations clearly indicate that prevention and treatment of this disease is highly cost-effective. However, not everything that is cost-effective might be affordable in Cambodia. This country will require external support to ease the growing burden of T2DM.
KeywordsCambodia Costing Diabetes Markov-Model Prediction South-East Asia T2DM
Diabetes Mellitus Type II (T2DM) is a non-communicable disease (NCD) with a high prevalence and an increasing incidence worldwide. The International Diabetes Federation (IDF) estimates that about 350 million people are affected worldwide and some 90% of diabetes patients live in low- or middle-income countries . IDF also expects a strong increase of diabetes cases which will result in tremendous human suffering, death cases and enormous costs of treatment . Thus, T2DM is a disease of high public health importance and a good representative to analyse the economic impact of NCDs on the health care system of Cambodia and other low-income countries.
In the year 2011 diabetes was top-8 cause of death worldwide with 1,392,000 death cases (2.6% of all death cases worldwide) and a rate of 20 per 100,000 inhabitants. For South-East Asia (and the Western Pacific) WHO Region the respective figures are rank 8 (8), 388,000 (284,000) death cases, 2.8% (2.4%) of all death cases) and a rate of 21 (16) per 100,000. In the year 2000 diabetes had been only on rank 10 worldwide and in the two target regions of this paper . Thus, diabetes is of high public health relevance. It is one of the most prominent NCDs and a major cause of morbidity and mortality worldwide as well as in South-East Asia and the Western Pacific Region.
The South-East Asian country of Cambodia has a specific historical pathway that led to a late start of the epidemiological transition and a specific age structure. The population pyramid of 1990 shows the dramatic impact of the Khmer Rouge period (1975–79). However, the population of Cambodia is rapidly aging. As chronic-degenerative diseases are mainly age-determined , it is extremely likely that the absolute and relative importance of NCDs such as T2DM is going to increase in Cambodia.
Currently the population is still rather young so that the share of morbidity and mortality caused by NCDs such as diabetes is slightly lower than in other countries of the region. The STEPS-Survey 2010  records a prevalence rate of T2DM of the adult Cambodians (25–64 yrs.) of 2.9% (170,000 people), but there are strong diversions between age groups, regions and gender. Younger adults (23–34 yrs.) have a lower prevalence rate (1.1%); older groups (55–64 yrs.) a higher rate (6%). The prevalence rate of the urban population (5.6%) is much higher than of the rural population (2.3%). At the same time, women (3.3%) are more affected than men (2.5%). Other surveys  come to slightly different figures but confirm the general finding that T2DM is a major public health problem in Cambodia.
According to WHO, Diabetes causes 3,122 death cases (2008) and a loss of 39,000 disability adjusted life years (DALY) in Cambodia per year (2004, latest up-date of WHO burden of disease database) . IDF records 5,540 death cases in 2013 . However, to our knowledge no estimate of the economic cost of T2DM from a society point of view was attempted. However, this disease is the fourth important NCD cause of death and the sixth importance NCD cause of loss of quality of life in this country.
Prevention and treatment of T2DM in Cambodia is limited as the public health services do hardly provide any diabetes care, T2DM is not included in the "basic package of health care" and oral anti-diabetic drugs (OAD) and insulin are not available in the whole country. The Royal Government of Cambodia wants to address this rising disease and take care of its population, but for this purpose it urgently needs a prediction of the expected number of T2DM cases in Cambodia, determine the most cost-effective intervention strategies, and estimate the budget impact of prevention and treatment of this disease.
Unfortunately, no economic modelling of T2DM was ever done in Cambodia. There are some cost estimates from other South-East Asian countries, such as Indonesia -, Malaysia ,, Philippines -, Thailand  and Vietnam  but these cannot easily be transferred to Cambodia with its unique age structure, economic situation and age structure. A very rudimentary costing comes from the World Diabetes Foundation  which run a project of diabetes care which cost 15.70 US$ per patient annually or 6.00 US$ per consultation with all treatment measures. Raguenaud et al.  estimated 48 US$ per year (p. a.) for a glibenclamide mono-therapy and 192 US$ p. a. of bi-therapy with glibenclamide and metformin. However, these figures are based on a pilot project of limited scope and geographical extent. They cannot be taken as a standard for a public health intervention in the entire country under the wings of the Ministry of Health.
Thus, our knowledge of the economic burden of T2DM in Cambodia is as limited as our insights into the cost-effectiveness of certain interventions in this country. There is a great need to base decisions of the Ministry of Health of the Royal Government of Cambodia as well as of the international health partners on evidence. Consequently, we have to make a micro-costing, a budget-impact and a cost-effectiveness analysis of interventions against T2DM. For this purpose we developed a Markov model of T2DM in Cambodia which will be presented in the next section. The results (to follow in section three) build the foundation of the Annual Operational Plan of the Ministry of Health in Phnom Penh. Consequently, our analysis takes the perspective of the Ministry of Health of Cambodia, but the insights are of relevance to other least developed countries in particular in the South-East Asian region.
Cambodia is a least developed country with a gross domestic product of about 1,000 US$ (gross domestic product in current US$ per capita in 2013) . Although the situation has improved within the last few years, still some 20% of the population is still below the official poverty level permitting hardly to survive daily life.
The Ministry of Health finances first-line (health centres) and second-line (hospitals) health facilities. However, patients have to pay modest fees. The services of these facilities are defined by the Ministry of Health as a so-called "basic package of health care services" . Currently, this package includes only prevention and treatment of communicable diseases as well as diagnosis related to maternity. Due to financial constraints, NCDs are not included in this package although it is generally accepted that they have a high priority . Cambodia spends about 50 US$ per annum per capita which is not enough to include chronic-degenerative diseases in the basic package.
The population under the poverty-line is exempted from paying the fees by a system called "Health Equity Fund" (HEF). The health care provider receives the equivalent amount from the HEF which is partly financed by the international donor community and the Royal Government of Cambodia. However, if a patient suffers from a disease which is not covered by the basic package (such as diabetes) he has to seek services from private providers who usually charge fees unaffordable for the majority of the population. Thus, very limited treatment is available for the majority of diabetic patients.
Most diabetic patients are undiagnosed until they have severe complications as no blood-sugar testing is usually performed in health centres. As anti-diabetic drugs are frequently only available in private pharmacies for high prices, even those who are diagnosed will receive proper treatment only if they can afford it. The same is true for insulin. Consequently, self-help groups and "peer-educator" groups  have been started. MoPoTsyo (Patient Information Centre), for instance, supports diabetes patients with drugs and insulin. At the same time it teaches them to become peer educators for their villages. This includes the search for potential diabetic cases. Peer Educator Networks have been accepted by the Government as an important contribution to the fight against NCDs , but until now no funds are allocated by the Ministry so that most Cambodians remain without any support if they develop diabetes.
The analysis follows the standards of health economic evaluation stipulated by Schulenburg et al.  and takes the perspective of the Ministry of Health of Cambodia. The calculation and prognosis of the costs of different interventions against T2DM is of high importance for this Ministry. In addition, the Ministry of Health is obliged to include only new interventions into the basic package of services if they are cost-effective in comparison to other possible interventions. Thus, we also calculate the incremental cost effectiveness ratio (ICER) as this is a possible statistic that allows comparing the cost-effectiveness of possible candidates to be included into the basic package.
As diabetes is an NCD the likelihoods of developing diabetes or transitioning to another stage of the disease are constant. Consequently, the epidemiology of T2DM is frequently simulated with Markov Models . However, only few models were designed for resource-poor countries . None represents the specific structure and parameters of the Cambodian demography and the health care system. Thus, it was necessary to design a specific model and search for the relevant parameters of Cambodia.
An undiagnosed diabetic case without complications can either stay in this condition or develop complications. For each complication (diabetic nephropathy, retinopathy, neuropathy, angina pectoris, peripheral vascular disease (PVD), myocardial infarction, stroke, heart failure) the respective transition probabilities (e.g. p_retino; p_neuro; p_stroke…) are applied. The patient can also die from diabetes (e.g. p_stroke_death) or non-diabetes (p_death_rest).
age-specific incidence of T2DM
probability of clinical diagnosis with persisting complications
probability of clinical diagnosis without complications or in the first year after development of complications
probability of change of therapy to OAD-therapy
probability of change of therapy to OAD/insulin or mono-insulin
share of patients taking insulin in those requiring it
share of patients taking OAD medication in those requiring it
myocardial infarction-related mortality
incidence diabetic nephropathy
incidence diabetic retinopathy
incidence distal symmetric neuropathy
incidence angina pectoris
incidence myocardial infarction
incidence heart failure
unit cost FPG-test
Annual cost outpatient treatment
Annual cost laboratory exams
Annual average cost for OAD medication
Annual average cost for insulin therapy (weighted mean for monotherapy and insulin-OAD-combined therapy)
Annual average cost for inpatient treatment
Basic Parameters of Markov Model
The mathematical prognosis follows a standard approach. The vector w t represents the number of individuals in the specific health states (e.g. "healthy", "undiagnosed, no complications", "undiagnosed, complications", "diet", "OAD, no complications", "insulin", "OAD, complication", "insulin, complication" "non T2DM death", "T2DM death") in time t. The respective vector w t+1 for the next period (t + 1) is calculated by multiplying the vector with the matrix of transition probabilities A , i.e.
Consequently, the number of individuals in the specific health states in time t is calculated as where w 0 denotes the vector of the numbers of individuals in the conditions in the beginning of the simulation.
Table 2 exhibits the most important parameters and the sources. Where no country-specific data was available, we used international data and adjusted it to Cambodia. For this purpose we confronted Cambodian and international experts with the international data and asked them for their assessment of the relevance for Cambodia. With these estimates we firstly simulated the demographic and epidemiological development from 1988 to 2007. The results of these simulations were compared with the real figures from Cambodian statistics . Finally, we adapted the parameters until the simulation fitted the reality to a best possible extent. This "calibration process" (e.g. ) was also used to adjust the transition of stages. As it is estimated that diabetes manifests about 10 years earlier in Asian countries than in Europe  the respective statistics had to be fitted. This standard approach of disease modelling under uncertainty calls for a set of sensitivity analyses which are presented in section Sensitivity analyses. The basic parameters are also expressed in the (Table 1).
The treatment costs are crucial so that detailed analysis was done based on the data of MoPoTsyo , data from WHO-CHOICE , the international drug price indicator guide  and additional research on market prices in Cambodia . The costs of Metformin and of Glibenclamide were estimates as 25 US$ and 3 US$ p.a. p.c. With the assumption that 12% use mono-therapy of Metformin, 10% mono-therapy of Glibenclamide and 78% the bi-therapy the average cost per OAD-patient are about 25 US$. The cost per ml of insulin were estimated with 0.72 US$ . At an average consumption of 10 ml we have annual cost of 86 US$ for insulin and 17 US$ for syringes. The combined therapy was estimated to cost 129 US$ p.a. p.c. Assuming that international standards  apply to Cambodia we reckoned that 57% of insulin requiring T2DM patients receive a monotherapy the average cost of insulin therapy is 114 US$ p.c. p.a. As far as not stated differently, all values are expressed in 2013 US$.
Based on these parameters we made a forecast of the epidemiology and economic impact of T2DM in Cambodia until 2028.
The basic simulation projects the epidemiological and economic consequences if the Government does not change its current policy, i.e., if the prevention and treatment of T2DM is not included in the basic health care package. In this case, only 12.5% of people requiring OAD and insulin can afford them by own funds or has access to them by charitable organisations. The basic simulation assumes that this rate remains stable.
The majority of T2DM cases remain undiagnosed. The rate is almost stable (63–64%) as the basic simulation assumes that no additional interventions are implemented. The number of diagnosed patients with complication, however, is increasing to a higher extent than the number of diagnosed patients without complications. In 2008 59% of diagnosed patients had complications, in the year 2028 this rate will have increased to 68%. This is due to the fact that the aging population and the longer duration of diabetes will result in more cases with complications.
In the year 2008 the 54,000 diagnosed T2DM patients would incur costs of 2 million US$ to cover all of diabetes treatment. 57% of this amount would have to be spent for OAD-therapy, the rest for insulin therapy. In the year 2028 this amount will have grown to 4 million US$ to cover 97,000 patients. If all patients (incl. non-diagnosed) had to be paid-for the respective figure would be 5.5 million US$ and 11 million US$. These calculations assume stable prices and no discounting. Thus, the costs will double due to a shift of patients towards the expensive patients with complications and requiring insulin. The expected average cost per diagnosed diabetic patient p. a. was 38 US$ in the year 2008 and will be 42 US$ in the year 2028.
The budget impact depends highly on the precision of the parameter estimates. Most critical is the estimate of the hospital cost. An increase of the cost by 50% leads to an increase of the cumulated cost by 40%, whereas the respective increase of the OAD-medication will result only in an increase of the total cost of 6%. Another critical parameter is the general and T2DM-related mortality. The earlier people die, the lower will be the cost. If the T2DM-related mortality increases by 20%, the cumulative cost will decrease by 16%. If the general mortality increases by the same rate, the respective decrease will be 17%. Even if we change all parameters towards the "best" possible case (according to estimates from the literature and from Cambodian experts), the budget requirement will increase until 2028.
In summary, Cambodia is facing an increasing number of diabetic patients requiring OAD- and insulin therapy. If the Government includes prevention and treatment of T2DM in the basic package, these costs will have to be refinanced by the Government even if the share of patients under treatment remains stable. Compared with the actual budget of the Ministry of Health these amounts seem not too high – but we have to keep in mind that they are additional to the existing expenditure for health care! Thus, there is a need to analyse the cost-effectiveness of interventions and select the best use of scarce resources.
The basic simulation assumes that a patient is on average undiagnosed for seven years . Consequently, treatment can only start late. At the same time, complications, early transition to higher states and diabetes-related death are higher than necessary. The purpose of screening programs is to determine T2DM cases as early as possible, ideally at the on-set of the disease to avoid these negative consequences. The impact of screening on the epidemiology and cost depends on the share of the population screened (e.g. only high-risk groups or total population) as well as on the sensitivity of the test. Without doubt, detecting every diabetic patient within one year will be technically impossible. However, in a first step we would like to present this utopic scenario of perfect screening, in the second step we will analyse the consequences of lower screening sensitivity and reduced coverage.
As first step we calculated the model for the case of the ideal screening program that detects every patient within one year after the on-set of T2DM. The consequences are an increase of the number of people living with T2DM and a strong decline of the number of T2DM-related death cases. In the year 2028, for instance, 275,000 people will live with T2DM instead of 264,000. The cumulative number of T2DM-related death cases from 2008 to 2028 will be 462,000 instead of 476,000. The share of diagnosed diabetics with complications sinks from 68% to 48%, and more diabetic cases need no medication (12% versus 8%). The total cumulative treatment costs increase from 61 to 68 million US$. However, the average cost per diagnosed diabetic patient of the year 2028 declines from 42 US$ to 33 US$.
The incremental cost-effectiveness ratio (ICER) compares the cost of an intervention with the effect (e.g. years of life saved, lives saved). An intervention is usually assessed as cost-effective if the cost per life-year saved (LYS) is less than the annual gross national product per capita (p.c.). In the case of Cambodia, an intervention is cost-effective if the ICER is less than 1,000 US$. If we assume the cost of urine sugar screening done by MoPoTsyo (i.e. 0.07 US$ p. c. plus 1 US$ for the confirmation test) the corresponding ICER is 464 US$/YLS, i.e., screening the entire population > 35 years annually for T2DM is cost-effective if the sensitivity is 100%.
In the second step we vary the coverage (share of population covered) and the sensitivity of the test. If we assume that 100% of the population are screened the cost per screening could be 1.05 US$ if the screening is 100% sensitive, 0.64 US$ if the sensitivity is 75% and 0.23 US$ if it were 50% sensitive. If the sensitivity goes below 36% screening will not be cost-effective at all.
Oral anti-diabetic therapy
OAD is not included in the basic package of health care services in Cambodia. Consequently, these drugs are only available for the very few protected by private health insurances or for those who can afford to pay for these drugs out-of-pocket. Consequently, it is estimated that only 12.5% of those diagnosed diabetics requiring OAD-therapy comply ,. In this section we will analyse the consequences of improved access to OAD-therapy. This could, for instance, be possible if the Royal Government of Cambodia provides free or highly subsidized access to these drugs at rural health centres, i.e., if OAD medication becomes an element of the basic package of health care in Cambodia.
An ICER of 100 US$ per YLS is highly cost-effective, and this result does hardly depend on the interest rate. Even at a rate of 7.5% the ICER merely increases to 105 US$/YLS. However, cost-effectiveness does not imply feasibility as the budget constraints have to be respected. In the year 2028 Cambodia would have to spend 6.7 million US$ instead of 4 million US$ if all patients requiring OAD-therapy had access to this medication. A difference of 2.7 million US$ p. a. is significant for the National Budget.
Insulin therapy in Cambodia is quite expensive and difficult to obtain. Consequently, it is estimated that only 12.5% of those requiring insulin therapy have access to it ,. In the following we will analyse the consequences of improved access to insulin therapy.
Consequently, improved access to insulin is cost-effective in Cambodia, but not as effective as improved access to OAD-therapy. From a purely economic perspective, improving OAD-therapy would have a higher priority than insulin therapy. However, patients requiring insulin are most likely suffering more than patients who can still deal with OAD-therapy. Thus, insulin therapy must have a high priority. However, the cost of improved access to insulin is considerable. If all patients requiring insulin are to receive full insulin coverage the additional cost of the year 2008 will be 2,473,319 US$, i.e., 6.5 million instead of 4.1 million US$. Cambodia will have to seek funds for this additional demand.
In most countries neither screening nor OAD- and insulin therapy are stand-alone interventions. A combined approach is seen best to reduce the burden of T2DM in Cambodia. Consequently, the last scenario analyses the consequences of introducing a "perfect" screening program and fully covering all patients requiring OAD- or insulin therapy.
The consequence of this optimal prevention and treatment approach is impressive: 582,000 years of life are saved and 78,000 death cases (–27%) are averted within the simulation period. The number of diagnosed diabetic cases increases from 37% to 90%. Among the diagnosed patients, the rate of those with complications declines from 68% to 51%. More patients do not require any medication (9% instead of 8%), but also more patients require insulin (21% instead of 17%).
However, substantial resources are needed to achieve these results. The annual budget for professional diabetes prevention and treatment in the year 2008 would be 8 million US$, i.e. 3.9 million US$ more than the current budget. This does not include the cost of mass screening. The model calculates that 6.1 million Cambodians of the respective age group (≥35 yrs.) will be living in Cambodia by then so that the costs of mass screening would be substantial. Assuming that the unit cost calculated for MoPoTsyo are representative for the entire country (0.07 US$ per screening unit) we could add 429,000 US$ just for screening. However, the ICER is still quite favorable. Without cost of mass screening it is 203 US$ per YLS (r = 5%), with mass screening 215 US$/YLS. Consequently, from a health economic perspective diabetes prevention and treatment is highly recommendable.
The results of the basic simulation and the intervention analyses strongly depend on the reliability of the parameters. Until today the most serious problem of any modelling of (health) economic interventions in developing countries is the "the almost complete unreliability of what data is available and the absence of any recent data" . Therefore, we conducted sensitivity analyses to assess the reliability of the predications. Parameters which were certain (e.g. fertility) were not included in the sensitivity analysis.
We realize that changes of the mortality and the incidence have a strong impact on the cumulative number of patient years. The incidence expresses the likelihood that a patient in the status "healthy" develops T2DM within a period. A change of this parameter will have a stronger impact on the epidemiology. Another important factor is the non-T2DM-related mortality ("rest mortality"). The higher the likelihood to die from any other cause of death, the lower is the number of patient years. In addition, patients who would require a certain treatment but cannot receive it will have a higher mortality (increase mortality non-compliance). The higher this increase, the lower is the cumulative number of patient years. All other variables have very limited impact on the patient years. It should also be noted, that all changes of parameters have under-proportional impact on the patient years.
Best- and worst-case scenario
Present value of cost [US$]
Probabilities and unit cost
Probabilities and unit cost
Consequently, there is a great need for further research on basic epidemiological parameters and unit costs. For the time being our estimates of the basic model and the interventions are better than anything else in the literature – but they have to be taken with some caution. However, it is very unlikely that all parameters will be wrong so that this data can be taken as a starting-point of discussion.
Type 2 Diabetes Mellitus is a major public health challenge in Cambodia. Currently 170,000 inhabitants are estimated to have the disease , but health planners have to expect higher rates in future. The Markov model calculates that the number of cases of T2DM was 172,000 in the year 2013 and will grow to 264,000 in the year 2028. Based on the population ≥ 35 yrs. this will correspond to a prevalence rate of 4%. There is no doubt that the Ministry of Health, the Civil Society and health partners will have to invest effort to prepare for this "epidemic" of T2DM in Cambodia.
IDF  projects 364,000 diabetic cases in the year 2030 for Cambodia, but their model is rather simplistic in comparison to our Markov model. However, we have to accept that our forecast might underestimate the reality to come as we assume that transition probabilities remain constant. This means, for instance, that the likelihood of developing T2DM at a certain age remains stable through the simulation period. This assumption might be wrong if nutritional habits and physical exercise change. With the availability of processed food even in rural areas and the reduction of the share of population engaged in physical work it might be that our estimates turn out to be too conservative.
However, even at given transition probabilities the economic burden of T2DM for Cambodia is tremendous. Chronic complications are the main cost driver, and their number is increasing with time. The model calculates that 65,711 cases would require additional treatment for chronic complications in the year 2028. This results in a strong increase of cost from 2008 to 2028, i.e. from 2 million US$ to 4 US$. This refers to average annual cost per diabetic case of 42 US$.
Prevention and treatment of T2DM is not only a financial problem. Currently, the majority of health care workers in rural health centres are not well trained in the detection of cases and treatment of patients. Cambodia will have to invest effort to train their doctors, nurses and other health professionals in the prevention and treatment of NCDs, such as diabetes. As the numbers increase, there is an urgent need to intensify the existing training programs and tailor them even more towards NCDs.
At the same time Cambodia has to safeguard a stable supply of OAD-medication and insulin (including syringes etc.). This as well is not only a financial challenge. Cambodia is a tropical country where temperatures easily go up to 40 degrees Celsius. Thus, insulin requires an uninterrupted cooling chain. For most people in rural areas refrigerators are an unaffordable luxury. Providing community cooling systems for people living with diabetes (e.g. one per village) could be discussed. However, logistics will remain an issue.
The cost estimates for diabetes treatment differ widely within studies from South East Asia. Andayani & Imaningsih estimated a figure of 239.64 US$ p. a. for one Indonesian hospital, Andayani  calculated annual cost of 252.00 US$ for another hospital of this country. The direct cost which Ibrahim et al.  calculated for Malaysia (572.70 US$ p. a.) and Chatterjee et al.  for Thailand (555.33 US$) are even higher. However, the differences between these figures and the results of the simulation presented in this paper are the result of the very low cost of the Cambodian health system. Martin  found in her cost study of Cambodian public health services that the average cost of a health centre outpatient visit were 1.00 US$, the average inpatient case in the medical department of a district hospital (CPA-1) had costs of 20 US$. This is only a fraction of what it would cost in Thailand, Malaysia or Indonesia.
Economic data of South-East Asian countries 2012
GNP [Int. US$]
Cost estimates [US$ p. a.]
As demonstrated in section Results the results of this simulation are highly sensitive to changes of the cost per service unit in hospitals. The model used the average cost per patient in medical departments of Cambodian hospitals as presented by Martin . To our knowledge no cost analysis exists analysing the exact cost per diabetic inpatient case in Cambodia. This calls for further research.
In addition, the cost estimates of Martin reflect the situation of public health care providers. The costs of private providers are much higher. If we assume that private providers are included in diabetes care in Cambodia the cost will strongly increase and the cost-effectiveness of interventions will decrease. The Royal Government of Cambodia might make public-private partnership agreements to safeguard country-wide professional diabetes care while containing the costs.
The economic development of the country until 2028 will have two consequences. On one hand side, it is most likely that prices and wages will increase so that some of the cost estimates of this paper will be obsolete. However, the public budget will also increase. It is not yet determined whether the prices and wages of the health care sector will grow stronger or less than the public budget. If the prices and wages increase with the same rate as the gross national income but the national budget grows at a lower rate it is possible that the ICER of interventions remains constant but the feasibility to finance these interventions declines.
Our simulations show that improved access to OAD-medication and to insulin is cost-effective for Cambodia. The Royal Government of Cambodia should be advised to include professional treatment of T2DM patients in its basic package of health care services. Even a slight increase of coverage (from currently 12.5% to 25%) has already a tremendous impact on mortality and life years.
The cost-effectiveness of screening for T2DM depends on some parameters. Firstly, screening is cost-effective only if the sensitivity of the test is high enough. However, urine sugar tests might not be sufficiently precise as they might have a high number of false-negative cases in particular if made only once. They are inexpensive – but if the sensitivity is not at least 37% they will not be cost-effective even if we could limit the screening population to 1% of the adults. The estimates of the sensitivity of urine glucoses tests in the literature do not give a clear answer. Davies et al. estimated a sensitivity of 43% , Hanson et al. of 64% , Friderichsen and Maunsbach of 21% . Other authors showed that the sensitivity differs strongly whether the test is done before or after glucoses intake . Under the conditions of a Cambodian rural health centre, a urine blood sugar test is frequently quite unreliable. However, as the WHO rightly puts it: "Despite its low sensitivity, urine glucose testing may have a place in low resource settings where no other procedure is possible. This is particularly so, of course, when the prevalence of undiagnosed diabetes is likely to be high" .
The simulations show that prevention and treatment of T2DM can be cost-effective even under the conditions of a least developed country like Cambodia. However, we must be aware that providing preventive and curative health care services at the point of provider is only one component of managing T2DM. Frequently undiagnosed diabetic cases will not show-up in health centres because they live too far away, have no transport or no time, or do not perceive any symptoms which would lead to a visit for diagnosis or therapy. At the same time, diagnosed patients might not comply with the instructions of the professional health care workers. Consequently, community-based approaches will be required that find potential patients, help them obtain professional help and improve compliance. The peer-educator network MoPoTsyo is such a community-based system that has shown the capability of strongly improving the situation of people living with T2DM. The Strategic Plan  foresees a nation-wide role of such networks in the fight against T2DM. The Ministry of Health could learn from the experiences of MoPoTsyo.
Diabetes is a great challenge to all stakeholders of the Cambodian health sector. It has financial, medical, demographic and social implications. However, the main challenge might be to develop another mind-set while addressing this chronic-degenerative disease. Until now, the entire health care system is based on the infection paradigm. One agent (e.g. plasmodium) causes one disease (malaria) and the connection between both is almost linear and deterministic. Consequently, seeking for the one cause, preventing or eliminating it is sufficient to provide proper health care. Chronic-degenerative diseases such as T2DM, instead, follow a multi-cause-multi-effect model, i.e., many risk factors and conditions contribute to their development (incl. a genetic risk). At the same time, the same set of risk factors might cause none or very different NCDs as no simple deterministic relationship between causes and effects exists. Consequently, a new focus on NCDs will require a new thinking about medicine, health care and the engagement of the health partners. Type 2 Diabetes Mellitus might be a door-opener to this different paradigm of health care in Cambodia.
AZ prepared the model and did the simulations. SF developed the research concept, supervised the work of AZ and drafted this paper. All authors read and approved the final manuscript.
The authors would like to thank their colleagues from the Ministry of Health of Cambodia, of the Deutsche Gesellschaft für Internationale Zusammenarbeit (GIZ) and MoPoTsyo for their support and input. This research has been partly financed by the German Federal Ministry for Economic Cooperation and Development (BMZ) through the Social Health Protection (SHP) Project implemented by the Deutsche Gesellschaft für Internationale Zusammenarbeit (GIZ) GmbH.
- IDF Diabetes Atlas. , [http://www.idf.org/diabetesatlas/FAQs]
- Diabetes. , [http://www.who.int/mediacentre/factsheets/fs312/en/]
- Disease and injury regional mortality estimates, 2000–2011. , [http://www.who.int/healthinfo/global_burden_disease/estimates_regional/en/index.html]
- Meade M, Emch M: Medical Geography. The Guilford Press, New York, London; 2005.Google Scholar
- Prevalence of non-communicable disease risk factors in Cambodia. STEPS Survey Country Report 2010. Royal Government of Cambodia, Phnom Penh; 2010.Google Scholar
- King H, Keuky L, Seng S, Khun T, Roglic G, Pinget M: Diabetes and associated disorders in Cambodia: two epidemiological surveys. Lancet 2005, 366: 1633–1639. 10.1016/S0140-6736(05)67662-3View ArticlePubMedGoogle Scholar
- Burden of Disease Database. , [http://www.who.int/healthinfo/global_burden_disease/estimates_country/en/index.html]
- IDF DIABETES ATLAS. IDF, Brussels; 2014.Google Scholar
- Soewondo P, Ferrario A, Tahapary DL: Challenges in diabetes management in Indonesia: a literature review. Globalization and health 2013, 9: 63. 10.1186/1744-8603-9-63PubMed CentralView ArticlePubMedGoogle Scholar
- Andayani T, Imaningsih I: Cost analysis of antidiabetic drugs for diabetes mellitus outpatient in Kodya Yogyakarta Hospital. Malay J Phar Sci 2007, 5: 19–23.Google Scholar
- Andayani T: Cost analysis of diabetes mellitus therapy in Dr. Sardjito hospital Yogyakarta. Maj Farm Indones 2006, 17: 130–135.Google Scholar
- WP SE, Azimatun N, Amrizal M, Rohaizan J, Saperi B: Economic burden of diabetic care in government health facilities in Selangor. Jurnal Kesihatan Masyarakat 2009, 15: 17–26.Google Scholar
- Ibrahim SS, Sulaiman SAS, Gillani SW, Hassali MA, Abougalambou ASI, Mohamed M: Evaluation of direct medical cost among outpatients of type 2 diabetes mellitus: cost analysis in teaching hospital of Malaysia. Folia Medica Facultatis Medicinae Universitatis Saraeviensis 2010, 45:.Google Scholar
- Beran D, Higuchi M: Delivering Diabetes Care in the Philippines and Vietnam Policy and Practice Issues. Asia Pac J Public Health 2013, 25: 92–101. 10.1177/1010539511412177View ArticlePubMedGoogle Scholar
- Higuchi M: Costs, availability and affordability of diabetes care in the Philippines. 2009.Google Scholar
- Higuchi M: Access to diabetes care and medicines in the Philippines. Asia Pac J Public Health 2010, 22: 96S-102S. 10.1177/1010539510373005View ArticlePubMedGoogle Scholar
- Chatterjee S, Riewpaiboon A, Piyauthakit P, Riewpaiboon W, Boupaijit K, Panpuwong N, Archavanuntagul V: Cost of diabetes and its complications in Thailand: a complete picture of economic burden. Health Soc Care Community 2011, 19: 289–298. 10.1111/j.1365-2524.2010.00981.xView ArticlePubMedGoogle Scholar
- World Diabetes Foundation: Program Evaluation - Creating a model for diabetes health care services in provincial level in Cambodia. In Book Program Evaluation - Creating a model for diabetes health care services in provincial level in Cambodia. World Diabetes Foundation.Google Scholar
- Raguenaud M-E, Isaakidis P, Reid T, Chy S, Keuky L, Arellano G, van Damme W: Treating 4,000 diabetic patients in Cambodia, a high-prevalence but resource-limited setting: a 5-year study. BMC Medicine 2009, 7:.Google Scholar
- World Development Indicators 2014. The World Bank Press, Washington D.C; 2014.Google Scholar
- Cambodian Demographic Health Survey and Cambodia Socio-Economic Surveys Analysis. Kingdom of Cambodia, Phnom Penh; 2011.Google Scholar
- National Strategic Plan for the Prevention and Control of Noncommunicable Diseases. Ministry of Health, Phnom Penh; 2014.Google Scholar
- van Pelt M, Lucas H, Men C: Yes, they can. In Transforming Health Markets in Asia and Africa: Improving Quality and Access for the Poor. Edited by Bloom G, Kanjilal B, Lucas H, Peters DH. Routledge; 2012:115.Google Scholar
- von der Schulenburg J-MG, Greiner W, Jost F, Klusen N, Kubin M, Leidl R, Mittendorf T, Rebscher H, Schöffski O, Vauth C: Deutsche Empfehlungen zur gesundheitsökonomischen Evaluation. Prӓsidium ab 1 Januar 2008 2007, 23.Google Scholar
- Tarride J-E, Hopkins R, Blackhouse G, Bowen JM, Bischof M, Von Keyserlingk C, O'Reilly D, Xie F, Goeree R: A review of methods used in long-term cost-effectiveness models of diabetes mellitus treatment. Pharmacoeconomics 2010, 28: 255–277. 10.2165/11531590-000000000-00000View ArticlePubMedGoogle Scholar
- Liu X, Li C, Gong H, Cui Z, Fan L, Yu W, Zhang C, Ma J: An economic evaluation for prevention of diabetes mellitus in a developing country: a modelling study. BMC Public Health 2013, 13: 1–11. 10.1186/1471-2458-13-1View ArticleGoogle Scholar
- Life tables Cambodia: Excel-Tabelle. , [http://apps.who.int/gho/data/view.main.60270]
- Wilke T, Ahrendt P, Schwartz D, Linder R, Ahrens S, Verheyen F: Inzidenz und Prӓvalenz von Diabetes mellitus Typ 2 in Deutschland. Dtsch Med Wochenschr 2013, 138: 69–75. 10.1055/s-0032-1327394View ArticlePubMedGoogle Scholar
- Ringborg A, Lindgren P, Yin DD, Martinell M, Stålhammar J: Time to insulin treatment and factors associated with insulin prescription in Swedish patients with type 2 diabetes. Diabetes Metab 2010, 36: 198–203. 10.1016/j.diabet.2009.11.006View ArticlePubMedGoogle Scholar
- MoPoTsyo patient information centre: Annual report 2011 In Book MoPoTsyo patient information centre: Annual report 2011. MoPoTsyo Patient Information Centre, City; 2012.Google Scholar
- Intensive blood-glucose control with sulphonylureas or insulin compared with conventional treatment and risk of complications in patients with type 2 diabetes (UKPDS 33) Lancet 1998, 352: 837–853. 10.1016/S0140-6736(98)07019-6Google Scholar
- Ballard DJ, Humphrey LL, Melton LJ, Frohnert PP, Chu PC, O'Fallon WM, Palumbo PJ: Epidemiology of persistent proteinuria in type II diabetes mellitus Population-based study in Rochester, Minnesota. Diabetes 1988, 37: 405–412. 10.2337/diab.37.4.405View ArticlePubMedGoogle Scholar
- Ahmed KR: Incidence of diabetic retinopathy: A 15 year follow-up in a hospital population (Bangladesh). In Masterarbeit. Universitӓt von Oslo; 2009.Google Scholar
- Sands ML, Shetterly SM, Franklin GM, Hamman RF: Incidence of distal symmetric (sensory) neuropathy in NIDDM. The San Luis Valley Diabetes Study Diabetes care 1997, 20: 322–329.Google Scholar
- Mata-Cases M, Prado-Lacueva C, Salido-Valencia V, Fernández-Bertolín E, Casermeiro-Cortés J, García-Durán M, Jabalera-López S, Fernández-Sanmartín MI: Incidence of complications and mortality in a type 2 diabetes patient cohort study followed up from diagnosis in a primary healthcare centre. Int J Clin Pract 2011, 65: 299–307. 10.1111/j.1742-1241.2010.02503.xView ArticlePubMedGoogle Scholar
- van Pelt M: Diabetes situation in Cambodia: personal communication. In Book Diabetes situation in Cambodia: personal communication. MoPoTsyo Patient Information Centre, City; 2013.Google Scholar
- WHO-CHOICE unit cost estimates for service delivery. , [http://www.who.int/choice/country/country_specific/en/index.html]
- Case study: Diabetes, Cambodia's silent killer. , [http://www.capacity.org/capacity/opencms/en/topics/health-systems/accessible-care-for-diabetes-in-cambodia.html]
- Frye JE: International drug price indicator guide: 2012 edition. In Book International drug price indicator guide. 2012th edition.Google Scholar
- Martin A: Cambodia hospital costing and financial management study: Final Report. In Book Cambodia hospital costing and financial management study: Final Report. Ministry of Health, City; 2013.Google Scholar
- Flessa S, Greiner W: Gesundheitsökonomik Eine Einföhrung in das wirtschaftliche Denken für Mediziner. Springer, Berlin; 2013.Google Scholar
- Cambodian Demographic and Health Survey 2010. Royal Governement of Cambodia, Phnom Penh; 2010.Google Scholar
- Loyo HK, Batcher C, Wile K, Huang P, Orenstein D, Milstein B: From model to action using a system dynamics model of chronic disease risks to align Community Action. Health Promot Pract 2013, 14: 53–61. 10.1177/1524839910390305View ArticlePubMedGoogle Scholar
- Khuwaja AK, Khowaja LA, Cosgrove P: The economic costs of diabetes in developing countries: some concerns and recommendations. Diabetologia 2010, 53: 389–390. 10.1007/s00125-009-1581-7View ArticlePubMedGoogle Scholar
- HPV Information Centre - Data Query. , [http://www.hpvcentre.net/dataquery.php]
- WHO: Comprehensive cervical cancer control: a guide to essential practice. World Health Organization; 2006.Google Scholar
- Harris MI, Klein R, Welborn TA, Knuiman MW: Onset of NIDDM occurs at least 4–7 yr before clinical diagnosis. Diabetes Care 1992, 15: 815–819. 10.2337/diacare.15.7.815View ArticlePubMedGoogle Scholar
- Cheng S-H, Chen C-C, Tseng C-H: Does medication adherence lead to lower healthcare expenses for patients with diabetes? Am J Manag Care 2013, 19: 662–670.PubMedGoogle Scholar
- Cambodia M: Clinical Practice Guidelines Type 2 Diabetes. In Book Clinical Practice Guidelines Type 2 Diabetes. City: 2013.Google Scholar
- Lane PA: The state of the Tanzanian economy 1984. In Book The state of the Tanzanian economy 1984. City: 1984.Google Scholar
- IDF Diabetes Atlas 5th Edition: Country Estimates Table 2030. , [https://www.idf.org/sites/default/files/IDFAtlas5E_Detailed_Estimates_0.xls]
- Martin A: Cambodia Hospital Costing and Financial Management Study. National Social Security Fund, Phnom Penh; 2012.Google Scholar
- Atwood B: Development Co-operation Report 2012: Lessons in Linking Sustainability and Development. OECD Publishing, Paris; 2014.Google Scholar
- World Data Bank. [http://www.databank.worldbank.org]
- Davies MJ, Williams D, Metcalf J, Day DL: Community screening for non-insulin-dependent diabetes mellitus: self testing for post-prandial glycosuria. Q J Med 1993, 86: 677–684. 10.1093/qjmed/86.10.677View ArticlePubMedGoogle Scholar
- Hanson RL, Nelson RG, McCance DR, Beart JA, Charles MAN, Pettitt DJ: Community screening for non-insulin-dependent diabetes mellitus: self testing for post-prandial glycosuria. QJ Medicine 1993, 86: 2133–2140.Google Scholar
- Friderichsen B, Maunsbach M: Glycosuric tests should not be employed in population screening. J Public Health Med 1997, 19: 55–60. 10.1093/oxfordjournals.pubmed.a024588View ArticlePubMedGoogle Scholar
- Shinozaki T, Yamaoka K, Yano E: [Validity of urinary glucose test for diabetes screening in workplace regular medical checkups]. [Nihon koshu eisei zasshi] Japanese journal of public health 1999, 46: 790–798.PubMedGoogle Scholar
- WHO: Screening for type 2 diabetes: report of a World Health Organization and International Diabetes Federation meeting. Geneva: 2003.Google Scholar
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly credited.