Validation of overall survival extrapolations made by TLV in the assessment of cost-effectiveness of oncology drugs in Sweden - A pilot study comparing extrapolated and observed life-years gained.

AIM: The aim of the study is to assess the accuracy of overall survival (OS) extrapolations in cost-effectiveness analysis made by the Dental and Pharmaceutical Benefits Agency (TLV) to decide on the reimbursement and use of oncology drugs in Sweden. MATERIAL AND METHODS: TLV appraisals for oncology drugs were identified during a 5-year period (2013-2017). To be included each appraisal and health economic model must include a TLV base case extrapolation of OS. Further, Kaplan-Meier (KM) estimates on OS from the original and follow-up clinical trials must be available. Potential follow-up trials on OS were identified in ClinicalTrials.gov and the Lund University Libraries databases, and in the Swedish Medical Products Agency (MPA) and the European Medicines Agency (EMA) databases. In cases where the KM estimates were not available, data points from figures published in TLV's appraisals were extracted using the semi-automated tools Digitizelt and WebPlotDigitizer. The accuracy of survival extrapolations was assessed by comparing extrapolated and observed life-years (LYs), using three different measures: 1) difference in LYs between the treatment and control group; 2) LYs in the treatment group, 3) LYs in the control group. RESULTS: We study TLV's preferred OS extrapolations and show that on average they overestimate the observed mean gain in LYs by 24%, and underestimate observed LYs by 3% and 11% in the treatment and control group, respectively. CONCLUSIONS: We conclude that it is feasible to validate OS extrapolations by comparing extrapolated and observed life-years. Even if survival extrapolations are reasonably accurate for the treatment group, this may still imply that extrapolations of LYs gained deviates to a larger extent. Follow-up studies on OS should be carried out to an increased extent to be able to validate, update and improve OS extrapolations in cost-effectiveness analysis of oncology drugs.

A median voter model of health insurance with ex post moral hazard.

One of the main features of health insurance is moral hazard, as defined by Pauly [Pauly, M.V., 1968. The economics of moral hazard: comment. American Economic Review 58, 531-537), people face incentives for excess utilization of medical care since they do not pay the full marginal cost for provision. To mitigate the moral hazard problem, a coinsurance can be included in the insurance contract. But health insurance is often publicly provided. Having a uniform coinsurance rate determined in a political process is quite different from having different rates varying in accordance with one's preferences, as is possible with private insurance. We construct a political economy model in order to characterize the political equilibrium and answer questions like: "Under what conditions is there a conflict in society on what coinsurance rate should be set?" and "Which groups of individuals will vote for a higher and lower than equilibrium coinsurance rate, respectively?". We also extend our basic model and allow people to supplement the coverage provided by the government with private insurance. Then, we answer two questions: "Who will buy the additional coverage?" and "How do the coinsurance rates people are now faced with compare with the rates chosen with pure private provision?".

Antidepressants and the suicide rate: is there really a connection?

Recent research claims that the major part of the observed reduction in suicide rates during the 1990s can be explained by the increase in the prescription of antidepressants. However, this conclusion is based on research that only looks at raw correlations; confounding effects from other variables are not controlled for. Using a rich Swedish data set, we reinvestigate the issue. After controlling for other covariates, observed as well as unobserved, that might affect the suicide rate, we find, overall, no statistically significant effects from antidepressants on the suicide rate; when we do get significant effects, they are positive for young persons. Regarding the latter result, more research is needed before any firm policy conclusion can be made.

Examining the quality of health economic analyses submitted to the Pharmaceutical Benefits Board in Sweden. The first year.

This study assessed the quality of health economic documentation submitted to the Pharmaceutical Benefits Board (PBB) in Sweden. Two different instruments were used in the evaluation: the PBB checklist, which was constructed by the authors from the PBB guidelines for health economic evaluations, and the QHES, a validated quality assessment instrument. Some areas that seem especially problematic, or where the quality was particularly low are identified and discussed. Also, we present the cost per quality-adjusted life-year that the companies have presented and how this related to the PBBs decisions.

Why did drug spending increase during the 1990s? A decomposition based on Swedish data.

OBJECTIVE: To decompose drug spending in Sweden between the years 1990 and 2000. This paper updates a previous study, which looked at the period 1990-1995, by providing an additional 5 years of data (1995-2000) and extending the previous analysis in a number of ways. METHODS: The paper builds on the earlier work that showed that changes in drug spending could be decomposed into three components: price, quantity and a residual. The size of the residual is a measure of the impact of changes in drug treatment patterns on drug spending. The data set used in this paper was collected from Apoteket AB (The National Corporation of Swedish Pharmacies) and was based on comprehensive information (inpatients as well as outpatients) on drug deliveries from wholesalers to pharmacies. Data were obtained for aggregate drug spending (from 1990-2000) and for spending according to anatomical therapeutic chemical (ATC) classification system group. RESULTS: Real drug spending increased by 119% during the study period. The residual rose by 67% indicating the switch from cheaper to more innovative and expensive drug therapies was a major cost driver. Real drug spending would have increased by about 31% if there had been no changes in treatment patterns. The second driver of drug spending was the quantity of drugs consumed, which increased by 41%. The main reason for the larger quantity sold appears to be increases in the intensity of medication in terms of defined daily doses per patient, rather than a larger number of patients starting drug treatment. Real prices decreased during the 10-year study period. We found large differences between ATC groups in terms of spending growth. The ATC groups that have contributed the most to the increase in spending are: drugs that affect the CNS (N), the alimentary tract and metabolism (A) and the cardiovascular system (C), which are also the three largest groups in terms of sales. For all three groups, it was the residual that mainly drove costs. CONCLUSION: This study indicates very clearly that the main driving force behind the increase in drug costs in Sweden between 1990 and 2000 was the change in drug therapy from old to new and more innovative and expensive drug therapies. This shows the importance of carrying out economic evaluations of new more costly drugs in order to make an assessment of the social benefits of a switch from a cheaper to a more expensive drug.

Changes in drug spending for different age groups during the 1990s? Evidence from Sweden.

The objective of this review is to examine how drug spending in different age groups changed during the 1990s. Time series analysis of registered data on prescription drug spending were performed, along with two decompositions, one of which was spending in three components: price, quantity (defined daily dose) and residual. The size of the residual is a measure of the impact of changes in drug treatment patterns on drug spending. The other decomposition was of the quantity component in three subcomponents: defined daily doses per person on medication, population share on medication and population size. Both decompositions are made separately for different age groups. How spending for different age groups has developed in different therapeutic areas was also studied. The main outcome measures were prescription drug spending over time within different anatomical therapeutic chemical groups and across different age groups of the population. It was found that the older the age group, the more drug spending had increased, both in absolute and in relative terms, during the 1990s. However, for some anatomical therapeutic chemical groups, younger age groups have experienced faster spending growth. The most notable example being anatomical therapeutic chemical group N CNS, where spending grew fastest by 350%, for those aged between 20 and 39 years. Furthermore, changed treatment patterns, such as a switch to more expensive drugs, is the main explanation for higher spending in all age groups. Higher spending is also due to a larger number of defined daily doses sold, which is almost totally due to the fact that each person on medication in the year 2000 utilized more defined daily doses than in 1990. Changing age structure explains a negligible share of the increase in drug spending, but elderly patients did have a key role in the spending surge since they increased their per capita spending the most.