Cost-effectiveness analysis of mHealth applications for depression in Germany using a Markov cohort simulation

Abstract

Regulated mobile health applications are called digital health applications ("DiGA") in Germany. To qualify for reimbursement by statutory health insurance companies, DiGA have to prove positive care effects in scientific studies. Since the empirical exploration of DiGA cost-effectiveness remains largely uncharted, this study pioneers the methodology of cohort-based state-transition Markov models to evaluate DiGA for depression. As health states, we define mild, moderate, severe depression, remission and death. Comparing a future scenario where 50% of patients receive supplementary DiGA access with the current standard of care reveals a gain of 0.02 quality-adjusted life years (QALYs) per patient, which comes at additional direct costs of ~1536 EUR per patient over a five-year timeframe. Influencing factors determining DiGA cost-effectiveness are the DiGA cost structure and individual DiGA effectiveness. Under Germany's existing cost structure, DiGA for depression are yet to demonstrate the ability to generate overall savings in healthcare expenditures.

Fig. 1. Markov probability analysis.

Treatment 2 is excluded from the visualization since the value differences to treatment 1 are too marginal to be recognized in the graph. Cycle = 0 represents the starting proportions, cycle = 20 shows the proportions at the end of the simulation horizon. One cycle length is equal to 3 months. a Shows the mild depression health state, b Shows the moderate depression health state, c shows the severe depression health state, d shows the remission health state and e shows the death health state. Blue line = Treatment 1 without DiGA; red line = Treatment 3 with DiGA future scenario. The Markov probability analysis graphs show the proportions of the cohort in treatment 1 and treatment 3 belonging to the defined health states over the simulation horizon.

Fig. 2. Cost-effectiveness plane.

The graph shows the effectiveness and total direct costs of each treatment scenario over the simulation horizon and per patient. Blue dot = Treatment 1 without DiGA; yellow square = Treatment 2 with DiGA standard of care; red cross = Treatment 3 with DiGA future scenario.

Fig. 3. Tornado diagram of one-way sensitivity analysis assessing the effect of selected parameters on the ICER.

ICER = Incremental cost effectiveness ratio, WTP = willingness-to-pay ratio referred to quality-adjusted life years; the horizontal bars represent the range of ICER due to changes in the model’s input parameters for one average patient in the cohort. Codification in the legend according to the following scheme: Input variable name (base case value: upper/lower range to lower/upper range), red = upper range of the input variable variation; blue = lower range of the input variable variation. p_DiGA_severe_moderate = transition probability to move from severe to moderate depression health state with use of DiGA; QALY_moderate = quarterly quality adjusted life years of an average patient in moderate depression health state; c_DiGA = quarterly cost of using a DiGA; QALY_severe = quarterly quality adjusted life years of an average patient in severe depression health state; QALY_mild = quarterly quality adjusted life years of an average patient in mild depression health state; QALY_remis = quarterly quality adjusted life years of an average patient in remission health state; pDiGA_moderate_mild = transition probability to move from moderate to mild depression health state with use of DiGA; pDiGA_mild_remis = transition probability to move from mild depression to remission health state with use of DiGA; c_severe = quarterly total direct costs of an average patient in severe depression health state; c_moderate = quarterly total direct costs of an average patient in moderate depression health state; c_mild = quarterly total direct costs of an average patient in mild depression health state; c_remis = quarterly total direct costs of an average patient in remission health state.

Fig. 4. Extended univariate sensitivity analysis of input variable quarterly cost of using a DiGA.

The graph shows the comparison of total direct costs of treatment scenario 1 and treatment scenario 3 depending on the quarterly cost of using a DiGA. The analysis shows that once the quarterly cost of using a DiGA fall below the threshold of ~20 EUR, the total direct costs of scenario 3 are lower compared to scenario 1. Blue line = Treatment 1 without DiGA; red line = Treatment 3 with DiGA future scenario.

Fig. 5. Probabilistic sensitivity analysis (PSA).

Each data point marks the result of one of the 10,000 simulation runs, where input parameters were randomly drawn from defined distributions according to Table 2 to assess parameter uncertainty. DiGA treatment gained average QALYs of 2.19 (95% CI; 2.18–2.20 QALY) with mean total direct costs of  ~ 9476 EUR (95% CI; 9466–9486 EUR) per patient in the cohort; compared to the treatment without DiGA with average QALYs of 2.17 (95% CI; 2.16–2.18 QALY) with mean total direct costs of ~7936 EUR (95% CI; 7926–7946 EUR) per patient in the cohort. Blue dots = Treatment 1 without DiGA; red dots = Treatment 3 with DiGA future scenario.

Fig. 6. State transition diagram of the CMM model.

The arrows show the possible transition paths of the patient cohort. The red numbers show the numbering of the health states. This graph shows the five modeled health states in the CMM model (mild depression, moderate depression, severe depression, remission, death).