Discussion

In Hong Kong, the healthcare budget is under constant stress. According to the most updated statistics released by the Hong Kong Food and Health Bureau, the total health expenditure amounted to HK$75 billion (US$9.6 billion), contributing to about 5% of the local gross domestic product in 2007.9 Although the proportion of healthcare expenditure in terms of total gross domestic product seems to be minimal, with the rising trend of healthcare expenses at 6.3% per year and the emerging problem of an aging population, planning for allocation of efficient and fair medical resources is indispensable. This cost-of-illness analysis of stable CAD gives an estimation of the first-year direct medical cost involved in the management of this condition, which may in turn provide some implications for the resource consumption pattern and possible preventive measures that may lighten the financial burden.

Some features were noted with regard to the demographic characteristics of the stable CAD patients included in our study. The majority of the patients recruited were men (70%). A similar pattern was found in other studies in which over 60–80% of patients were men.4 ,5 ,9 However, a meta-analysis found that women in fact had a similar or slightly higher incidence of SA symptoms than men, which seemed to contradict our findings. A possible explanation is that the diagnosis of SA was more difficult in women than in men, as atypical symptoms like dyspnoea and weakness were more prevalent in women and the perception of the chest pain was not the same in women as in men, resulting in higher incidence of SA diagnosis and admission in the male population.10 ,11 The rate of revascularisation was as high as 62% in our study, which is significantly higher when compared with other countries like the UK, which had a rate of only 6%. A reason for this can be our subject retrieval method. In this study, patients were mostly identified from coronary catheterisation (CC) records. In local practice, after CC had shown vessel obstruction, subsequent PCI would usually be immediately arranged and thus, patients receiving CC had a relatively higher potential of subsequent PCI.

An obvious variation in LOS was observed, with the longest average LOS in the CABG group, having over a threefold difference from that of the PCI group and the no procedure group. In general, patients with long hospital stay of over 10 days were seldom observed in patients not receiving any surgical or cardiac catheterisation procedures. Only 3% of patients not undergoing procedures, contrasting with 70% of CABG and 11% of PCI patients, had stayed for over 10 days. Therefore, the longer the hospital stays, the higher the cost. The hospital stay was related to the complexity of the procedures performed.

When compared with cost analysis studies in other countries, the cost contribution pattern demonstrated a few differences. In the UK cost of angina pectoris study, the top two cost contributors were also revascularisation and hospitalisation costs, but the former was 35%, which was slightly higher than the latter one, 32%.4 The difference in cost patterns may be attributed to the different cost breakdown components included. In this UK study, a coronary angiogram was counted as a part of the procedural cost under revascularisation but in our study, it belonged to the class of diagnostic procedures. Although the rate of PCI and CABG was only 6% in the UK study, it was 62% in ours; the inclusion of the high-cost coronary angiogram compensated for the low procedural rate. Thus it still had a significant contribution to the total cost. Besides, the average LOS was slightly longer in Hong Kong (5.4 days) than in the UK (3.8 days), resulting in the difference in hospitalisation costs. In addition, the outpatient clinic visits showed the greatest contribution to the cost of newly diagnosed stable CAD in the USA, which accounted for a total cost of US$722 million (38% of the total cost) in 2000.7 Ranking second was hospitalisation cost, which was largely different from outpatient cost, contributing to 16% of the total cost (US$307 million). The difference seen may be attributed to the different study design and various managements in different geographical locations. In this US study, medical costs under consideration excluded diagnostic exams, radiological tests and coronary revascularisation when compared with our study. Also, the reasons for the less dominant role of outpatient visit costs in our study may partly be caused by the local resources constraint in our healthcare system. In Hong Kong, the public healthcare system is under great stress and most citizens are fully dependent on public services. Over 90% of the total secondary and tertiary care including specialist outpatient clinic follow-up were provided by the public healthcare system.12 In 2007, 5.5 million consultations were provided by the specialist outpatient clinics and yet, public hospitals were facing a manpower crisis with insufficient numbers of doctors. As a result, outpatient services in the local community are always in a state of saturation and thus follow-up visits are inevitably arranged infrequently for less acute and non-fatal diseases like stable CAD. On the other hand, American citizens are used to private clinic visits and the outpatient visit costs did include private clinic fees, contrary to the inclusion of public spending only in our study. Thus, the different inclusion criteria and different public-private service reliance may account for the difference observed.

Another major finding from this study was that among the three comorbidities concerned, the total annual cost was the highest in patients with hyperlipidaemia, followed by DM and HTN. With regard to the correlation between the cost and the disease state, diverse patterns were observed among the three groups of comorbidities. Hyperlipidaemia was associated with significantly higher total medical costs, as well as in the costs of most individual items like hospitalisation, diagnostic examination, laboratory investigations, medications and procedures. The higher cost in hyperlipidaemia cases can be explained by the increased risk of atherosclerosis in the presence of poor LDL control, which likely increases the chance of revascularisation. Such an increase in the need for invasive therapy directly caused a rise in revascularisation costs, which was confirmed by the high statistically significant difference found in the hyperlipidaemia group, and indirectly increased hospital admissions and inpatient costs. Considering the fact that revascularisation was the second leading contributor to the first-year cost of stable CAD, the higher cost in hyperlipidaemia was understandable. It is reasonable to deduce that aggressive control and management of serum lipid level can be an effective means of minimising the cost by lowering the need for procedures. One piece of supporting evidence from a meta-analysis involving over 90 000 hyperlipidaemia subjects conducted in 2005 showed that for hyperlipidaemia patients, the need for first coronary revascularisation was reduced by 23% with every 1 mmol/l of LDL cholesterol reduction. Lowering LDL cholesterol by 1.5 mmol/l with the use of sustained statin therapy showed a decrease in the incidence of major vascular events by 21%.13 Hence, the use of lipid-lowering agents like statins in stable CAD patients is warranted. However, the cost-effectiveness of lipid lowering therapy in terms of lowering total medical costs needs further evaluation.

In the case of DM, the statistical significance found in the cost of laboratory investigations between the DM group and non-DM group was expected. As more routine monitoring such as HbA1C, whole blood glucose, random plasma glucose level and fasting blood glucose level were required for assessing the blood glucose control in DM patients, more laboratory tests were needed, thus causing increased expenses in the DM group. Although it has been shown that DM increased the risk of cardiovascular complications,14 the total cost and other cost components in this study did not reveal a significant difference. This may be due to the fact that medical costs involved in conditions other than those related to CAD were not included in the cost calculation. Therefore, even if complications like AMI arise, the subsequent resources consumption was not taken into account. And, the baseline blood glucose of our study subjects was good in general, the mean HbA1c was 6.7%. Another point is that the costs of DM medications were relatively low when compared with the total cost of CAD, and consequently minimal difference in total cost was reported.

Cost analysis in HTN patients showed a different scenario with a higher total cost in the non-hypertensive group. Such difference may be caused by the uncertainty during the recruitment of the HTN patient group. Contrasting with the cases of DM and hyperlipidaemia, the diagnosis of HTN could not be adequately identified from the medications prescribed or the baseline blood pressure level. Thus, the recruitment was solely dependent on the physicians’ diagnosis noted in the patients’ medical record. Yet, in practice, chronic illnesses may not be recorded in every admission record and therefore, some of the patients with no marked diagnosis of HTN may be inappropriately classified as non-HTN due to the limitation in accessing actual disease record.

Stable CAD is an early stage disease which may progress to more advanced stages like unstable angina and MI if it is not properly managed. Apart from the more severe symptoms experienced by the individuals, the cost incurred due to AMI was also higher. When compared with our previous finding in AMI patients,8 a few significant differences were observed. Patients with unstable angina had a longer average hospital stay of 16.4 days, when compared with 5.4 days in stable CAD. Majority of AMI patients required over 8 days of hospitalisation while stable CAD patients mostly required only 1–6 days, thus explaining the much higher inpatient cost in AMI subjects. In addition, the cost of bed stay in the coronary care unit for most MI patients is significantly higher than the cost of a general ward bed for most stable CAD patients. Furthermore, drugs used in the two conditions showed a diverse pattern. The major difference was the use of fibrinolytic agents, which was the usual practice in AMI but not in stable CAD. Thus, the use of high unit-cost fibrinolytics accounted for the one-third higher medication cost in AMI. Third, the cost of emergency services for AMI and stable CAD was 30-fold. As regards the resource consumption pattern characteristics in AMI and SA patients with comorbidities of HTN, DM and hyperlipidaemia, both studies found that patients with hyperlipidaemia had a higher management cost than those with HTN or DM.

In a meta-analysis involving 2950 non-acute CAD patients with half of them received PCI and half of them managed with conservative therapy, those with an absence of recent MI showed no significant difference from PCI in term of mortality, MI and the need for subsequent revascularisation over the use of drugs alone. It concluded that PCI currently performed in these patients were not reasonably cost-effective.15 Also, the guidelines on the management of stable CAD published by the European Society of Cardiology 200612 and the Scottish intercollegiate Guidelines Network 2007,16 mentioned certain circumstances when interventional approaches would not offer prognostic gain and the procedures were considered as highly elective without true urgency. Referring to the findings in our study, up to 62% of subjects received either PCI or CABG; there is a possibility that the number of procedures carried out were not justified for their cost-effectiveness. Despite the self-financed nature of PCI consumables, operations would certainly bring about their own costs, additional hospitalisation days and laboratory tests, making up substantial healthcare spending. Given the current budget constraint, risk stratification and cost-effectiveness of invasive versus conservative therapy need to be carefully considered in decision making in the management of stable CAD. Optimal use of medical therapy is likely to be a much more cost-effective initial treatment strategy for most low-risk stable CAD patients without prognostic disadvantage compared with an early invasive strategy.

Since HTN, DM, hyperlipidaemia and progression to AMI have been shown to burden the healthcare system, proper management of general health and care involved in the early stage of disease can definitely help ease the stress. Preventive strategies rather than therapeutic approaches should be of higher priority in the management of most of the chronic diseases. Family doctors, community nurse services, public health education and pharmacists’ efforts on drug review, monitoring the progress and control of these diseases are crucial for the overall health management in a society, especially when facing an aging population. A multidisciplinary approach involving various healthcare professionals for disease prevention and management is therefore another measure suggested for cost saving, which can also help share the workload from doctors and nurses in hospital settings.

This study has several limitations. First of all, our study included a small sample pool and was adopting a retrospective way of data retrieval. Therefore, some information may not be accessible and the actual duration of therapy was difficult to assess. Second, this study calculated the direct medical costs incurred in the diagnosis and management of stable CAD in the subsequent year following initial diagnosis. The unit cost data were obtained from the Hong Kong Government Gazette 2003, which lists the costs of various procedures, laboratory investigations, inpatient care, clinic services and emergency admissions at that time. However, as inflation has taken place between 2003 and 2008, the price in 2003 may not truly represent the costs in 2008, making the cost calculation unable to reflect the actual cost involved in the management of CAD in the study period. Third, this study only involved patients from the Prince of Wales Hospital and therefore the usual practice and treatment choice in our cluster may not be representative, making the generalisation of the cost less applicable. Another drawback of this study is that in retrieving data concerning hypertensive patients, the presence of HTN may not have been properly recorded in the patient records and also may not be easily identified from the medications prescribed. Therefore, some hypertensive patients who had no confirmed diagnosis of HTN may be neglected and counted as non-hypertensive, affecting the generation of a reliable correlation between cost and HTN. In addition, this study had taken into account only the direct medical costs involved from the perspective of a public health provider. Intangible costs and other costs incurred by patients like disease-related loss of productivity, impact on patients and family members’ quality of life, as well as the tremendous cost of PCI consumables were not addressed, making the study not comprehensive enough to represent the whole cost-of-illness to the society. Therefore, future research may look further into this aspect and evaluate the direct and intangible costs from the societal perspective to complement this current study. Furthermore, investigations on the cost-effectiveness between conservative treatment and interventional therapy in stable CAD in the local setting may be done to provide more guidance on the choice of management.