Real-World Impact on Clinical Diagnostic Decisions and Hospital Finance Performance | DxPrime Case Study

This case study analyzes the real-world impact of DxPrime on diagnostic decision-making and the Case Mix Index (CMI).​

The findings demonstrate that DxPrime improved upstream clinical workflows in diagnostic decision-making and documentation, implemented with a soft-touch interface design and without hard alerts or workflow interruptions.

These upstream improvements further translated into measurable downstream impact: the hospital-wide Case Mix Index (CMI) increased by 3.4%, more accurately reflecting patient severity.

The study also found that, even within information-dense EHR environments with numerous interface elements, physicians still noticed and responded to visual changes in the DxPrime interface.

After opening DxPrime, physicians adopted its recommendations. This indicates that DxPrime functions as a practical, decision-support tool embedded within the diagnostic workflow, gaining physician trust and consistent use in daily clinical practice, unlike conventional alert-driven notification systems that are often ignored.

Diagnostic Impact

36%

Primary Diagnosis Adoption Rate

When DxPrime suggested a primary diagnosis,
approximately one-third of physicians adopted the recommendation

47%

CC/MCC Adoption Rate

When DxPrime suggested CC/MCC,
approximately half of physicians adopted the recommendation

CMI Impact

+3.4%

Case Mix Index (CMI)

More accurately reflects patient severity

Why It Matters

Reflects Real-World Clinical Usage Behavior

• Whether physicians notice and activate DxPrime is entirely voluntary, without interruptive alerts or forced workflow stops
• Behavioral impact is driven by real-time clinical decision support, not retrospective coding requirements or administrative interventions
• Consistent results observed across hospitals of different sizes and specialties

Study Scope & Design

Hospital-Wide Deployment with Soft BPA

• Three hospitals selected by scale: Small (200–500 beds), Medium (500–1,000 beds), and Large (1,000–2,000 beds)
• All hospitals are in Taiwan, with full hospital-wide deployment of DxPrime
• Soft BPA* mechanism: Feedback is delivered only through subtle color changes of the DxPrime button, without interruptive alerts, pop-ups, or forced workflow constraints. Physician response is entirely voluntary
• Study period: October to December 2024 (3 months)
• Coverage across 33 specialties: 23,935 inpatient cases and 1,319 physicians using DxPrime

* BPA: Best Practice Advisory

Primary Diagnosis Adoption Analysis

• In cases where DxPrime suggested a primary diagnosis, physicians accepted and revised or added diagnoses in 36.12% of cases
• The highest adoption rate was observed in medium-sized hospitals, reaching 49.81%

Hospital	Suggested Primary Diagnosis	Adopted by Physicians	Adoption Rate (%)
Total	2,766	999	36.12
S	522	151	28.93
M	530	264	49.81
L	1,714	584	34.07

CC/MCC Adoption Analysis

• In cases where DxPrime suggested CC/MCC, physicians accepted the recommendation and added CC/MCC in 47.1% of cases
• The highest adoption rate was observed in large hospitals, reaching 66.4%

Hospital	Suggested CC/MCC	Adopted by Physicians	Adoption Rate (%)
Total	6,029	2,842	47.10
S	1,557	272	17.50
M	1,305	468	35.90
L	3,167	2,102	66.40

* This table comprises only cases that meet all of the following criteria:

1. Completed a full inpatient episode
2. DxPrime was activated during hospitalization
3. No CC/MCC documented before analysis
4. DxPrime recommended adding CC/MCC

CMI Analysis

• Weighted average CMI improvement:
  ((3.24% × 4,821) + (2.95% × 5,283) + (3.67% × 13,831)) ÷ 23,935 = 3.42%

Hospital	Inpatient Cases	CMI Before DxPrime	CMI After DxPrime	CMI Impact (%)
Average				+3.42
S	4,821	0.7819	0.8073	+3.24
M	5,283	1.2664	1.3037	+2.95
L	13,831	1.3004	1.3482	+3.67

* This table comprises only inpatient cases during the study period. CMI is calculated at the hospital level.

---