Data on the population-based epidemiology and healthcare burden of ethnic Chinese or Asian TSC patient are lacking. This study is a territory-based study describing the epidemiology of TSC patients in Hong Kong over a period of 23.7 years (from 1995 to 2018).
The local disease burden of TSC patients is reflected by the disease prevalence, survival profile and direct costs. The utilization of CDARS allows complete capture of all TSC cases under the care of our public health system, which handles over 80% of all admission in Hong Kong. In particular, most patients with chronic illness are followed up by the public health system. Moreover, CDARS is an electronic system of all medical records under the Hong Kong Hospital Authority (25). Previous local studies have demonstrated accurate ICD-9 coding and completeness of territory-based data retrieved from this system (29, 30). The population-based nature of this study in the calculation of prevalence and healthcare utilization minimized the selection bias which could occur in studies conducted in a specific referral centre.
Our study presents a number of novel findings which fill the knowledge gap on the prevalence and survival profile of local TSC patients. In particular, the age-specific survival rate of TSC patients is highlighted. Survival profile of our local TSC patients was favourable, with nearly all of them survived beyond 20 years old. The median mortality age and the IQR of mortality age were corroborated by the 2017 study in Bath TSC Clinic, United Kingdom (median mortality age: 33 years, IQR: 26–46 years) (12). Their study reported no death before the age of 16. This difference could be attributed to the availability of a better management and care for TSC patients provided by a specialized TSC clinic in the UK, which is not available in Hong Kong. However, both studies suggested that TSC patients could potentially enjoy a long life expectancy.
Another important parameter is the availability of mTOR inhibitors. Our data showed that it was not widely used in our TSC patients (47 patients, 16.5%). mTOR inhibitors (i.e. Sirolimus and Everolimus) are found to be effective in suppressing the overall disease progression of TSC, through inhibiting the mTOR pathway and anti-tumour effect (13). The Hong Kong government has recently approved Everolimus for TSC treatment. However, there are certain prescription restrictions. The criteria for subsidized prescription are: [1] presence of astrocytoma or SEGA, [2] renal angiomyolipoma greater than 3cm. Access to Everolimus by other TSC patients were restricted, limiting the choices to the cheaper mTOR inhibitor, Sirolimus alone. The mTOR inhibitor prescription, together with prevalence data shown in this study, would be informative to our government in terms of financial estimation of the cost required for mTOR inhibitors subsidies for TSC patients.
This study further delineates the direct cost attributed to TSC patients in the public hospital system, over a study period of 10 years. The results reveal a high healthcare burden disproportionate to its prevalence. Although disproportionately high inpatient cost compared to prevalence have been discussed in rare disease population in general (31, 32), there are several considerations in healthcare planning worth highlighting in the context of TSC. First, TSC patients have higher mean inpatient length of stay (9.36 days) when compared with the whole rare disease population in Hong Kong (6.1 days) (23). Second, TSC patients have favourable survival, and adult TSC patients were nearly three times that of paediatric. Hence, that means that healthcare resources were required in both paediatric units and specialities in adult medicine units. Third, after comprehensively reviewing the healthcare service domains other than inpatient cost, an astonishing number of 9.23 OP visits per patient-year, being 4.91 times more than the number of general population, was noted. The number of A&E attendance was also noticeably higher than the general population. Consequently, these suggest that on top of the considerable inpatient requirement, A&E and OP clinic are particularly resource-intensive service domains in the management of TSC. This is compatible with our knowledge that TSC patients develop a number of life-threatening complications which require hospital admissions and regular follow-ups (24, 33). Locally speaking, TSC patients were referred to multiple frequent subspecialties when they exceed paediatric age. According to overseas experience (34), measures taken to centralize the OP service to TSC patients, such as training doctors in a multi-disciplinary can optimize the healthcare resources provided to TSC patients.
We are aware of two other Taiwan studies in the published literature that estimated the epidemiology on a population level with the use of electronic healthcare system instead of cross-sectional respondents (23). Comparing other epidemiology and direct costs studies of TSC, the prevalence found in our study (3.87 in 100,000) is higher than Taiwan’s studies (1.05 and 1.58 in 100,000 respectively) (17, 18). The accuracies of their studies, however, could be limited by the shorter study periods (3 years and 13 years respectively), or cases that are outside the national health insurance database. With a long study period (23.75 years), we are confident our study depicts a less biased prevalent rate, suggesting that our results are of relevance to clinical and policy decision makers.
Comparing service utilization, we noted that the inpatient admission of our TSC patients (9.36 day per patient-year) is higher than overseas population-based studies (3.25 days in Sweden (20), 6.2 days in UK (33)). OP visit in Hong Kong (9.23 visits per patient per year) was also higher than that reported in Sweden (4.7) but lower than reported in UK (13.7). In Kingswood et al’s study, higher outpatient visit, GP visit, inpatient admission were noted on TSC patients in comparison with an age-, sex- and last record date-matched Comparator Cohort. Our study did not use a matched comparator cohort, because we aimed at finding out the actual service utilization and cost of TSC patients incurred in the healthcare system over 10 years, taken into account their age and sex. On the other hand, we took one more step forward in delineating A&E utilization of TSC patients compared with the general population. The results revealed that TSC patients required more A&E visits than the general population, which provided another prospective in viewing the disproportionate economic burden TSC patients posed on a healthcare system. We are aware that Rentz et al. and Reaven et al. (23) also reported high physician visits and A&E visits. Nevertheless, Rentz et al. was carried through recruiting questionnaire respondents who had internet access and received alert on a few patient advocacy websites; hence, selection bias towards TSC patients with higher cognitive function and less severe forms of the phenotype was highly probable. Besides, their study focused more on the comparison of direct cost in a claim-based manner. As with our study, we delineated the healthcare utilization attributed to TSC patients with the use of an electronic system capturing all medical records under the care of our public health system, over a study period of 10 years. Consequently, our data captured is likely to be of relevance to clinical and policy decision makers.
There are 3 broad caveats to the study findings. First, despite the use of an electronic system which captures all medical records under the care of our public health system, there might be a minimal chance of missing out TSC patients who had never attended local public hospitals. However, TSC patients often require referral to multiple specialists due to its multisystem manifestation, thus are less likely handled in private hospitals alone. In order to reduce the likelihood of missing cases, in the study design, we adopted a long retrospective study period (23.7 years). Therefore, we are confident that the potential uncaptured cases are minimal and did not systemically bias our results. Second, under-diagnosis of TSC clinically could be a major source of error that renders the prevalence in this study underestimated. Since the diagnosis of TSC in local centres was mainly by clinical findings, failure of diagnosis might be possible. Given the broad spectrum of disease manifestation and severity, the identification of TSC in mildly affected individuals could be challenging (35). To further improve the accuracy of prevalence estimation, we also expanded the search criteria. A search using the combination of the diagnostic codes of the major and minor features listed in Updated Diagnostic Criteria for Tuberous Sclerosis Complex 2012(26) was performed; however, there was no additional cases identified from the CDARS system. With no previously undiagnosed cased could be found even with the expansion of search criteria, there are unlikely to be systematic biases in underdiagnosis of TSC locally. Third, data on genetic diagnosis was only obtained from one hospital, as not all patients had genetic diagnosis available. Whether such information is representative for all local patients requires further verification.