Digital decision support tools showed parallel improvements with no between-group differences in kidney outcomes among adults with chronic kidney disease in Chinese primary care centers

This cluster randomized clinical trial evaluated the impact of a digital decision support system (CDSS) embedded within electronic health records against government-supported, nephrologist-delivered training for managing chronic kidney disease (CKD). The study involved 3,390 adults aged 18 years or older across various Chinese primary care centers. The primary objective was to assess whether the digital intervention could enhance the composite outcome of kidney-related and cardiovascular hospitalizations over a 36-month period, though initial analysis focused on a six-month follow-up window.

The intervention group utilized the CDSS tool designed to guide clinical decisions, while the control group received standard care supplemented by specialized training from nephrologists. This comparative approach aimed to isolate the specific value added by the digital technology versus traditional educational interventions. The setting of primary care centers is critical, as it represents the frontline where most CKD management occurs, yet resources and specialist availability can vary significantly.

Regarding the primary outcome of CKD diagnosis rates, the data revealed a notable increase in both study arms. The intervention group saw rates rise by 21.4 percentage points, whereas the control group experienced an increase of 27.9 percentage points. Despite these substantial absolute changes, the statistical analysis indicated a nonsignificant between-group difference, with an adjusted odds ratio of 0.91 and a 95% confidence interval ranging from 0.72 to 1.14. This suggests that the digital tool did not prevent the natural progression of diagnosis rates compared to the training program.

Medication utilization patterns also demonstrated similar trends across both cohorts. The use of renin-angiotensin-aldosterone system inhibitors improved in parallel for both groups, with no significant between-group differences observed. Similarly, the adoption of sodium-dependent glucose transporter 2 inhibitors followed a comparable trajectory, showing parallel improvements without statistical separation between the digital and training arms. These findings imply that the underlying clinical behaviors regarding guideline-directed medical therapy were influenced by factors beyond the specific intervention method.

Lipid management outcomes mirrored these results, with low-density lipoprotein cholesterol control showing parallel improvements in both the intervention and control groups. However, blood pressure control and glycated hemoglobin control did not show any improvement attributable to the intervention. The lack of distinct advantage for the CDSS in these metabolic parameters highlights the complexity of modifying clinical practice in primary care settings. The absence of reported adverse events, discontinuations, or tolerability issues suggests the digital tool was safe, but its efficacy in altering hard clinical outcomes remained unproven in this initial phase.

Several limitations constrain the interpretation of these findings. As a phase 1 analysis relying solely on the initial six-month follow-up data, the study may not capture long-term effects or secondary outcomes relevant to the 36-month primary endpoint. Furthermore, absolute numbers for medication use were not reported, and p-values were not explicitly provided, with only confidence intervals available for statistical inference. These data gaps prevent a full assessment of the intervention's robustness.

Ultimately, this preprint study indicates that the digital decision support system did not demonstrate an independent effect over nephrologist-delivered training. The practice relevance is tempered by the lack of a detected advantage for the technology in this specific context. While the CDSS was well-tolerated and integrated into the workflow, the comparative effectiveness against expert-led training was not established in this initial analysis. Future research with longer follow-up periods and larger sample sizes may be necessary to determine if sustained use of such tools yields different results in chronic disease management.