Health and economic effects on patients with type 2 diabetes mellitus in the long run: predictions for the Chilean population

Diabetes is a major public health problem that causes a high economic and disease burden, mainly due to its related complications. In Chile, like other jurisdictions across the globe, statistics in the last few years indicate that DM2, and its consequences, are increasing over time [19,20,21]. This study allowed us to estimate the effects of DM2 in the long run in the population of recently diagnosed patients with this condition and characterise the variability across different individuals based on their baseline risk covariates. Our results showed that the expected costs and QALYs of an adult with DM2 were USD 8660 and 12.44 QALYs. We also found that costs are explained by two independent elements: (i) the baseline risk which is associated with complications that demand healthcare resources; (ii) and the length of life from the diagnosis. The former relates to the higher expenditure on healthcare due to complications and the latter to the fact that because more time alive also consumes more resources. In terms of QALYs, our findings showed that either baseline risk or length of life from diagnosis showed a significant inverse association with QALYs. While the length of life contributes directly to the life years, baseline risk contributes to both, a decrease in HRQoL and length of life.

To our knowledge, this is the first modelling study in a low-and-middle-income country that reports the long-term outcomes and costs. One of the strengths of our study is that not only provides a characterization of heterogeneity in costs and outcomes and their magnitudes, but also a quantification of the potential impact of controlling parameters such as HBA1c, SBP, BMI and LDL on the cost and QALYs. For example, a reduction of 1% of the HbA1c is associated with a 3.8% reduction in costs. Likewise, this impact is also observed in HRQoL, which should be interpreted as the amount of disutility averted because reducing one unit of the baseline risk covariate. Although these magnitudes are informative, we are aware that cannot be claimed a hierarchy among those covariates. The fact that HBA1c showed the highest coefficient in the model, does not mean it has the highest impact, because it depends on the equivalence of the units of each covariate.

Although short-term cost estimates are useful because they inform the amount of money one health system must allocate to it in one year, we argue that long-term estimates complement the information for health policy decisions. These estimates can offer projections that may illustrate more clearly the relevance of the disease over time. For example, assuming an incidence rate of 6.9 per 1000 patient-years [23], we can calculate that an annual cohort of new DM2 patients in a population older than 15 years (96,880 patients) [24] determines a long-term expenditure of USD 839 million for the Chilean health system. This calculus assumes that patients will have access to the same services offered today, or -if new services are covered- they do not carry additional healthcare costs. Putting this estimate in context, this means that new DM2 patients in one year represent a long-term expenditure equivalent to 6.6% of the total annual public health expenditure of the country.

Our model also predicted that, among all complications attributable to DM2, MI was the most frequent event and the most frequent cause of death. This finding is consistent with previous literature that indicates that cardiovascular disease, and particularly, ischemic cardiac diseases are the most frequent complications [19, 25,26,27,28]. Furthermore, our results are also coherent with results of the UKPDS model [12] and mathematical models that have used the UKPDS as an input (25). Up to our knowledge, there is no study reporting an incidence estimate throughout the life’s patient. It is worth noting that our estimates are function of the baseline characteristics of the Chilean population, which may explain some differences with other studies where the baseline risk pool of the population differs.

One limitation of our model is that over 1% of the patients remain alive at the age of 95 years old, in circumstances that national statistics indicate that this proportion is only 0.15%. This lack of accuracy of our model at this stage may be explained by the general estimates of mortality provided in the lifetables which assume the same probability of death for all people above 85 years old. In addition, as described above, the risk equations we used to inform our model were obtained from a British cohort, which might have contributed to this lack of precision. Notwithstanding, we argue that this small difference in absolute terms does not determine significant changes in our results.

Another limitation of these results relates to the generalizability to other jurisdictions. We are aware that our estimates are based on the baseline risk distribution of Chilean patients and direct costs of the Chilean healthcare system, which may limit its application overseas. However, we argue that some relevant findings are valid across jurisdictions. For example, the independent relationship between risk, life expectancy and cost are both likely to hold beyond Chile. Although the higher risk is associated with lower life expectancy (negative relationship), both explain greater cost (positive relationship). Likewise, life expectancy and age are also significantly associated with cost, but in opposite direction. While the positive association between life expectancy and cost is expected, the negative association between age and cost seemed counterintuitive, since there was no clear reason to support the idea that between two patients whose time alive from diagnosis is the same, the younger is associated with fewer costs. To examine this finding, we explored the interaction between age and time alive. This analysis showed an estimate statistically significant (0.002; p-value < 0.0001) meaning that cost increases by 2% for every additional unit (year) of this interaction term. Thus, increases in age and time alive from diagnostic contribute to a higher cost.

Finally, we acknowledge that DM2 is a multifactorial disease, which interacts with many other conditions. Future research on modelling should link DM2 to other pre-diabetes disorders, such as insulin resistance or obesity, for a better characterization of the disease process. Furthermore, the precision of costs and outcomes estimates can also be improved by linking the diabetes model to specific models for each complication. Last, including other attributable complications, not considered in the UKPDS study, is also desirable for a more comprehensive representation. Although we could adapt an independent hypoglycaemia model to our DES model, other complications such as non-alcoholic fatty liver disease should be incorporated in future modelling exercises. In conclusion, DM2 is a highly prevalent condition that determines a significant expenditure on the health system and substantial health losses either in quantity or HRQoL. Although the control of cardiovascular risk factors and the metabolic control of the disease; both have an important impact on costs and outcomes, the main impact is achieved by postponing the age of onset of the disease.

留言 (0)

沒有登入
gif