We have extensive experience conducting Systematic Reviews (SRs) using methods equivalent to Cochrane Reviews. This involves forming groups of experienced individuals who independently conduct tasks such as paper selection, data extraction, and Risk of Bias assessment, resolving discrepancies through group discussions. Additionally, we adhere to PRISMA guidelines, ensuring clear criteria for paper selection, as evidenced by works such as [Miwa2019], [Nishimura2021], and [Sugano2018].

In analyzing SR results, we utilize various methods including the Fixed Effect Model, Mixed Effect Model considering differences in effects between papers, and meta-regression analysis focusing on patient differences across studies. Techniques like Bayesian statistics and hierarchical Bayesian models are advantageous in Mixed Effect Models and meta-regression, enabling statistical analysis using Markov Chain Monte Carlo method.

Moreover, even without direct comparisons from randomized controlled trials (RCTs), indirect comparisons can be made by combining RCTs with different comparators including placebos, forming what is known as Network Meta-analysis. For instance, [Higuchi2017], [Miwa2017], and [Miwa2019] conducted a Network Meta-analysis comparing the healing effects of reflux esophagitis between Vonoprazan and proton pump inhibitors (PPIs). By including all PPIs connected via RCTs in a network, indirect comparisons between Vonoprazan and other PPIs became possible, with statistical evaluation facilitated through Bayesian methods.

Furthermore, [Iwasaki2016a] performed a Network Meta-analysis of angiotensin II receptor blockers (ARBs) for hypertension. Since ARB RCTs targeted patients with varying degrees of hypertension, meta-analysis evaluated the outcome of "excess systolic blood pressure," defined as systolic blood pressure exceeding 130mmHg. This choice was made as Real World Data analysis suggested that the difference in systolic blood pressure depends on pretreatment values, while the ratio of excess systolic blood pressure does not, thus potentially aligning patient backgrounds. Similarly, [Nishimura2021] conducted a Network Meta-analysis of oral antidiabetic drugs (OADs).

Additionally, though not documented in papers, we conducted projects using meta-regression analysis. For instance, in evaluating surgical outcomes, various studies, including case reports, are available, each linked to a specific preoperative condition value. By conducting regression analysis with these values as explanatory variables and outcomes as dependent variables, we estimated the improvement in outcomes by reducing these values, predicting the effect (or "hard outcome").

Ultimately, with well-adjusted confounding factors in Real World Data, it should be possible to compare various treatments without resorting to Network Meta-analysis. [Nakamura2023] attempted to replicate the results of [Nishimura2021] using Real World Data from Medical Data Vision (MDV) to assess the feasibility of this approach.

Real World Evidence (RWE) refers to evidence generated from Real World Data (RWD) rather than from clinical trials. While Milliman owns RWD through consulting for health plans/insurance companies and providing data warehousing services in the United States, it doesn't possess proprietary RWD in Japan. Instead, it recommends and analyzes optimal RWD tailored to clients' needs and publishes findings in peer-reviewed academic journals. Below, evidence generated from different RWD sources is presented:

National Database (NDB):

The NDB stores nearly all data from Japan's health insurance systems, making it an unparalleled database with high comprehensiveness. [Mizumaki2023] compared the incidence rates of various diseases before and after Helicobacter pylori eradication using NDB data. Unlike commercial RWD, which may lack long-term patient follow-up, the NDB provides data from 2009 onward, making it optimal for long-term patient tracking. [Igarashi2023] studied the treatment trends of valve disease, among other planned COVID-19 treatment evolution studies.

Japan Medical Data Center (JMDC):

JMDC's health insurance union data comprises enrollment, prescription, and health examination data, pioneering Japan's RWD. [Sruamsiri2018] compared treatment continuation rates and total medical expenses, including non-drug costs, among psoriasis patients using biological drugs. [Nishimura2022] compared the hazard ratios of cardiovascular events among type 2 diabetes patients by medication type. [Iwasaki2023a] analyzed the difference in medical expenses between health insurance and national insurance unions by combining JMDC and DeSC Healthcare data.

Medical Data Vision (MDV):

MDV's clinical data from DPC hospitals includes inpatient/outpatient accounting, laboratory data, and discharge summaries. [Izumi2023] conducted a cross-sectional study on valve disease types and treatment methods. [Uno2020] illustrated treatment patterns and per-patient medical expenses for multiple myeloma. [Eguchi2022] surveyed the reality of recurrent hospitalizations for heart failure. [Kondoh2022] estimated the prevalence of idiopathic pulmonary fibrosis (IPF) using MDV and NDB open data.

DeSC Healthcare:

DeSC Healthcare's data includes health insurance, national insurance, and late-stage elderly population data, featuring a high proportion of elderly individuals for most diseases, mortality data, and data on quality of life and productivity via kencom. [Takeshima2023a] analyzed medical expenses, quality of life, and productivity for anemia patients, providing a multifaceted assessment of the burden of anemia.

Real World Data Co., Ltd.:

As a pioneer in providing electronic medical record (EMR) data in Japan, Real World Data Co., Ltd. offers EMR data considered more suitable for medical research than claims data. [Ono2020] conducted a validation study defining diseases/events using EMR-recorded diagnoses as the gold standard and utilizing receipt information only.

Moreover, projects like [Takeshima2023b] reproduced the control group of a COVID-19 drug trial using NTT Data Corporation's Millennia EMR data, demonstrating the potential of RWD for trial control group alternatives. While ensuring internal validity remains a challenge due to lack of randomization, Milliman has successfully replicated RCT control groups using RWD, contributing to various publications. Additionally, Milliman conducts research using diverse RWD sources such as Wellbee's Personal Health Record (PHR) [Takeshima2017] and PREVENT's data [Yamamoto2022] for evaluating interventions to prevent diabetes complications.

Milliman is an actuarial consulting company specializing in promoting the sound development of insurance businesses. Actuaries play a crucial role in the health insurance sector, which is considered a type of insurance. They are important not only in the United States, where private health insurance functions, but also in countries like Germany, where public insurers compete.

In the United States, Milliman conducts simulations of healthcare insurance system reforms, particularly focusing on pharmaceutical companies as clients. [Fitch2008] estimated the optimal level of copayments for diabetes treatment drugs in the context of the U.S. healthcare insurance system. Setting the copayment at the right level affects medication adherence and complication rates, which in turn impact healthcare costs.

In social simulation, one of the most challenging aspects is incorporating interactions. Human societies consist of complex interactions, making their simulation a non-trivial task. Agent-Based Model (ABM) is one of the most common models that accounts for interactions.

ABM involves setting up "agents" on a computer, defining interactions between them, allowing agents to autonomously make decisions and take actions, and observing the outcomes. [Itsumura2023] observed how the prescription rate of a COVID-19 treatment drug is influenced by the number of infected individuals and physicians' decisions. The simulation utilized NetLogo, a programming language specifically designed for ABM.

Additionally, ABM is used to simulate the impacts of market strategies by insurance companies and policy changes on markets and regional health. ABM serves as a valuable tool for understanding the complexities of social systems and predicting their responses to various interventions.

Embedded Value (EV) in pharmaceutical companies represents the total present value of future probabilistic cash flows for each license or product owned, discounted at a hurdle rate. This theoretically aligns with the company's value as perceived by investors. Therefore, changes in EV can explain fluctuations in stock prices. However, EV differs from the capital amount stated in statutory accounting, particularly for businesses dealing with long-term cash flows like pharmaceuticals, where conventional accounting principles often fail to appropriately account for period matching. For instance, when purchasing licenses, not all expenses can be immediately recognized as assets, leading to temporary decreases in capital despite predictions of significant future revenue contributions. Thus, statutory accounting capital does not accurately reflect a company's true value.

EV is already utilized by many life insurance companies, which, like pharmaceuticals, deal with long-term cash flows. The necessity for internal management accounting systems to accurately grasp insurance performance and period profits is emphasized.

The core of calculating EV for pharmaceutical companies lies in predicting future cash flows, considering factors such as anticipated patient numbers, treatment ratios, drug shares in treatments, drug prices, production costs, phase three trial success probabilities, and external factors. These factors, including their interactions, are probabilistically predicted to forecast future cash flows, from which EV is computed.

Using EV, one can estimate how much a company's value would increase or decrease under certain circumstances, such as downward revisions of anticipated phase three trial success rates, unexpected high competitor drug prices, or the smooth implementation of related governmental policies. This serves as a compass when navigating uncertain business waters.