Art Modeling Cherish Model 1 97
Abdul Soumphonphakdy
At the risk of alienating what few old newspaper pals of mine still have jobs, the industry they (and I!) so cherish, which has suffered mind-blowing valuation losses and several dozen rounds of downsizing both in personnel and column inches, is still bloated after all these years, with costs that no publisher would dream of incurring if he was starting a newspaper from scratch in 2009.
The importance of credible, trustworthy numerical simulations is obvious especially when using the results for making high-consequence decisions. Determining the credibility of such numerical predictions is much more difficult and requires a systematic approach to assessing predictive capability, associated uncertainties and overall confidence in the computational simulation process for the intended use of the model. This process begins with an evaluation of the computational modeling of the identified, important physics of the simulation for its intended use. This is commonly done through a Phenomena Identification Ranking Table (PIRT). Then an assessment of the evidence basis supporting the ability to computationally simulate these physics can be performed using various frameworks such as the Predictive Capability Maturity Model (PCMM). Several critical activities follow in the areas of code and solution verification, validation and uncertainty quantification, which will be described in detail in the following sections. The subject matter is introduced for general applications but specifics are given for the failure prediction project.
The manufacturer submitted separate economic models for each spinal muscular atrophy (SMA) type: type I, type II, and type III.6 The models allowed estimation of health care costs, life-years (LYs), and quality-adjusted life-years (QALYs). The models had initial cycles that reflected the timing of outcome assessment in the relevant clinical studies. For time points beyond the time horizon of the clinical studies, cycles corresponded with the timing of the administration of nusinersen (every four months). Time horizon varied by SMA type: 25 years for type I, 50 years for type II, and eight years for type III. The analyses were conducted from the Canadian public health care system perspective. Costs and outcomes were discounted at an annual rate of 1.5%, and expected values of costs, QALYs, and LYs were obtained through probabilistic analysis.
In the SMA type I model, the cohort entered the model at their baseline clinical status. Each cycle patients could transition to other health states that included maintenance of baseline clinical status; whether this improved, worsened, or had no improvement; milestones consistent with SMA type II (e.g., sits without support, stands with assistance, walks with assistance and stand/walks unaided); and death. The analysis was run over a time horizon of 25 years. Cycle length varied at the onset of the model. Patients could transition between health states at 2, 6, 10, 13, and 14 months. The first four transition points related to the timing of clinical assessment in the ENDEAR study7 and the latter cycle corresponded to a dosage of nusinersen. Subsequent cycles were every four months, which conformed to the timing of dosages of nusinersen.
In the SMA type II model, the cohort entered the model at their baseline clinical status. Each cycle patients could transition to health states reflecting worsening, no improvement, mild improvement, and moderate improvement from baseline clinical status; states relating to whether the patient could stand or walk with assistance; milestones consistent with SMA type III (e.g., stand unaided and walks unaided); and death. The analysis was run over a time horizon of 50 years. For the first 15 months of the model, the cycle length was three months conforming to the timing of clinical assessment in the CHERISH study.8 Subsequent cycles were every four months, which conformed to dosages of nusinersen.
In the SMA type III model, health states included non-ambulatory, ambulatory, and death. Patients could enter the model at either the ambulatory or non-ambulatory health states. The analysis was run over a time horizon of 80 years. For the first 27 months of the model, the cycle length was three months, which conformed to the timing of clinical assessment in the CS2 and CS12 clinical studies.5 Subsequent cycles were very four months, which conformed to dosages of nusinersen.
For both SMA types I and III, utility values were derived from a vignette study where five experts in SMA rated derived health state descriptions relating to the health states within the models. For SMA type II, utility values were obtained from a mapping study of quality of life values observed in the CHERISH trial and EuroQol 5-Dimensions questionnaire values. Both studies used to estimate utility values were unpublished.10,11
The reporting of the cost estimates used within the model lacked transparency and health care costs appear to be derived from a German study.15 The methods for interpolating the costs of care into the Canadian context are limited; however, given the high cost of nusinersen, the impact of additional health care cost would be limited.
Although utility values for SMA type II were available form this study, the manufacturer used a different set of utility values for the SMA type II model. This unpublished study used data from the CHERISH study relating to responses to the Pediatric Quality of Life Inventory instrument at each assessment point that were then mapped to the EQ-5D utility scores based on a published mapping algorithm to derive utility values for each state. The manufacturer chose to not use the actual values for specific states when it was felt the ordering of states by utility value was incorrect. The recent CADTH guidelines for economic evaluation suggest that direct measurement should be used to elicit utility values and mapping should be discouraged.9
One patient submission was received, which was prepared jointly by the Canadian Organization for Rare Disorders and Cure SMA Canada. The submission was based on the results of one focus group, four interviews, and a survey. Most of the respondents were caregivers and family members. The submission cited issues for patients with SMA, which included physical functioning, the ability to breathe unassisted, difficulties swallowing, and the ability to conduct activities of daily living. The manufacturer accounted these aspects within their economic model by considering aspects of SMA in the model health states. Impacts on families and caregivers were raised as an aspect of the condition, as well. This was not considered by the manufacturer in its pharmacoeconomic submission.
582128177f