Evaluation of Da2 Pilot Projects and Recommended Actions

...

Evaluation of the Project

Strengths of pilot:

- The model is scalable and can be universally applied to every patient admitted.

- Provides the good accuracy levels of about 79%.

- Create segmentation which helps healthcare facilities to allocate resources effectively.

Concerns about execution of the pilot:

- Accuracy not in the high levels, thus cost implications for some inaccurate predictions can be high.

- Changing trends like identification of a new variable would lead to change in the algorithm, which can be time consuming and costly.

Factors affecting the difficulty and cost effectiveness of implementing the pilot:

- Creating a readmission risk score for every patient would require huge amount of data storage and management that can be costly.

- Easy to implement once the algorithm is programmed.

- Changes in the trends can cause algorithm modification and can increase the cost significantly, given the universality of the pilot.

Pilot 3: Advanced Illness Management

Summary of the Project and Results

The primary aim of the pilot is to reduce unnecessary ED visits and hospitalization of the patients. It predefines the criteria for a patient to be a part of AIM, and once the patient included in the model, it suggests that the patient has high chance of hospitalization or revisiting the facility. The patients who are part of this Management are offered priority health services with regular care and are imparted with self-awareness about their condition and required medication and precautions to be taken. The first Cohort included 25 people that visited the facility 96 times in 6 months. Following the start of the AIM program the same patients visited only 33 times in the next six months, showing a big success of the program.

Evaluation of the Project

Strengths of pilot:

- Highly efficient model, as the result were very effective.

- Initial results showed it lowered the cost to healthcare system.

- Provides larger scope of improving healthcare by reducing the number of revisits and increasing the quality and patient satisfaction.

Concerns about execution of the pilot:

- Low sample size, which may show some biased results.

- High level of integration required of every discipline.

- Multidisciplinary effort from the beginning.

Factors affecting the difficulty and cost effectiveness of implementing the pilot:

- High involvement of every discipline makes it difficult to implement.

- Lower patient qualification into this model would mean lower cost of data storage

- The target segment is lower in proportion, which may lead to cost ineffectiveness in the longer run given the amount of involvement required to run the process.

Conclusions from Our Evaluations

For P1: Mapping Underserved Communities: Although the pilot is inexpensive and easy to implement, the scope of the pilot that is limited to providing the details of the area that is fairly inaccessible to health care and the use of biased and older data casts doubt over the credibility of the results and thus a new set of data organization is needed. The process doesn’t require high level of integration and uses fairly simple methodology which makes it a cost effective pilot. But since it doesn’t use the data of the patients that don’t put address in their records it can show some biased results and also the use of a legacy census data casts doubts whether the results are reflection of the latest trends.

For P2: Reducing Readmissions: The pilot has a higher scope that all the pilots, but its accuracy needs to be improved further to make it an ideal model to implement. As we can see the pilot is applicable to each of the patient that predicts if the patient will be needed to re-visit the facility, but has an accuracy rate of 79% in predicting a patient’s risk of readmission within 30 days of discharge. Since the model has uses in providing quality health care by identifying the problems associated with the patient at an early stage and even provide the segmentation of every patient in terms of risk evaluation, we need to make sure that the model is as efficient as possible in order to provide a quality healthcare and increase our overall Return on Investment.

For P3: Advanced Illness Management: The pilot is built on highly efficient model with small sample testing and targets only a small segment, thus requires high level of involvement of multidiscipline, so we need to increase its scope and credibility to make it more universal. As we can see from the results that it provided some positive results in improving the health care, but it requires clinical social workers, nurse practitioners, licensed practical nurses, and registered nurses. The team then assessed all the needs of the patient to create a plant that aligned with the physician’s medical plan. In the whole process the team had to stay in touch with the patient.

Recommended Next Steps

For P1: Mapping Underserved Communities, CHS should improve the data input by:

- Using more dynamic data than just census data, even if it is obtained at a cost.

- The model should involve more diverse set of public and private data, including health surveys, administration enrollment and billing records and medical records used by several entities such as hospitals, physicians and health plans.

- We should find a way to make people fill in the area to which they belong if not complete address to get a diverse set of data.

For P2: Reducing Readmission, CHS should try to increase the accuracy by:

- We need to do some more regression, add