Clinical Decision Support System, Antihypertensive Treatment Intensification, and Blood Pressure Control Authors: Jiali Song, Yanchen Liu, Qinggang Shang, Jiamin Liu, Haibo Zhang, Zeming Zhou, Wei Wang, Jiapeng Lu, Xin Zheng Published: 2026 ABSTRACT Key PointsQuestionIs the implementation of a clinical decision support system (CDSS) associated with increased antihypertensive treatment intensification and improved blood pressure control in primary care practices?FindingsIn this post hoc secondary analysis of a cluster randomized clinical trial comprising 4612 patients, CDSS implementation was associated with higher antihypertensive treatment intensification rates but not with improved blood pressure control compared with usual care. Greater treatment intensity was associated with larger reductions in systolic blood pressure.MeaningThese findings suggest that while CDSS may be associated with increased antihypertensive treatment intensification, additional implementation strategies may be needed to improve blood pressure control. Introduction Hypertension is the leading modifiable contributor to cardiovascular diseases,1 accounting for approximately 1.28 billion adults worldwide in 2019.2 Despite the availability of antihypertensive medications and substantial efforts to improve hypertension management, blood pressure (BP) control remains suboptimal, particularly in primary care settings, where most patients with hypertension are managed.3,4,5 In China, 23% of adults are affected by hypertension, yet only 15% achieve adequate BP control.6 Guideline-concordant care in clinical practice is insufficient, and another important factor associated with poor BP control is therapeutic inertia, defined as the failure to initiate or intensify antihypertensive therapy when BP remains above recommended targets.7,8 Clinical decision support systems (CDSSs) have been proposed as a strategy to improve hypertension care by providing physicians with real-time, guideline-based treatment recommendations during patient encounters.9,10 Previous studies have demonstrated the clinical benefits of multicomponent interventions that incorporate CDSSs, often targeting multiple aspects of care, including treatment intensification, medication adherence, and care coordination.11,12,13,14,15 In the Learning Implementation of Guideline-Based Decision Support System for Hypertension Treatment (LIGHT) trial, the use of a guideline-based CDSS was shown to improve guideline-concordant antihypertensive therapy in Chinese primary care settings.16,17 However, it remains unclear whether CDSS implementation alone increases treatment intensification when BP remains uncontrolled and whether such changes translate into improved BP control. Accordingly, we conducted a post hoc analysis based on data from the LIGHT trial, a pragmatic cluster randomized clinical trial designed to evaluate the effectiveness of a CDSS in improving hypertension management.16 In the present analysis, we further assessed whether CDSS implementation was associated with improved treatment intensification and BP control. We also explored between-practice heterogeneity in treatment intensification and BP control. Methods Study Population The design and results of the LIGHT trial have been described in detail elsewhere.16,17 In brief, the LIGHT trial was a cluster randomized clinical trial conducted from August 1, 2019, to July 31, 2022, in primary care practices in China. After a 3-month baseline period during which all practices delivered usual care, practices were randomized to either a CDSS intervention or usual care. Patients with hypertension who were taking up to 2 classes of antihypertensive medications and had systolic BP lower than 180 mm Hg and diastolic BP lower than 110 mm Hg were eligible for enrollment. In practices assigned to the intervention group, the CDSS provided guideline-based recommendations to physicians at the point of care, typically suggesting treatment escalation when BP remained uncontrolled. Patients were followed up during routine clinic visits, which were recommended at least every 3 months. At each visit, physicians recorded BP measurements, medical history, patient-reported medication use, antihypertensive prescriptions, and other relevant clinical information in the electronic system. The study protocol was approved by the Central Ethics Committee of Fuwai Hospital and the ethics committees of the participating sites (trial protocol and statistical analysis plan in Supplement 1). Individual informed consent for CDSS use was waived because the intervention was considered minimal risk to participants. We followed the Consolidated Standards of Reporting Trials (CONSORT) reporting guideline for this analysis. In this post hoc analysis, we restricted the analyses to patients with uncontrolled BP at baseline and excluded patients from low-volume practices (<10 patients with uncontrolled BP at baseline). The final cohort consisted of 4612 patients from 93 practices (eFigure 1 in Supplement 2). Outcomes and Covariates The outcomes of this study included antihypertensive treatment intensification and BP control. We defined treatment intensification as increasing the classes or doses of antihypertensive medication during visits with uncontrolled BP, consistent with prior studies.18 BP control was primarily defined as BP less than 140/90 mm Hg.19,20,21 We also used BP of 160/100 mm Hg22 or higher and BP of 130/80 mm Hg19 or higher as alternative BP thresholds for intensifying treatment. Other patient characteristics were collected in the face-to-face interview, including age, sex, educational level, smoking status, history of diabetes, history of hyperlipidemia, history of stroke, and classes of antihypertensive medication used. Information on practice and physicians was obtained by questionnaires, which included the location, number of patients with hypertension, and age, sex, and educational attainment of physicians. Statistical Analysis Statistical analysis was conducted from August 1, 2024, to September 1, 2025. Baseline characteristics of practices and patients were summarized using frequencies and percentages for categorical variables and mean (SD) values or median (IQR) values for continuous variables, as appropriate. We summarized the median frequency and IQR of treatment intensification during visits with uncontrolled BP and BP control at the last follow-up. We estimated the adjusted treatment intensification and BP control rates using hierarchical models. To characterize heterogeneity across practices, we estimated the median odds ratio (MOR).23,24 Three hierarchical models (unadjusted, patient adjusted, and fully adjusted) were built to examine whether rates and MORs attenuated after accounting for patient and practice characteristics. These models accounted for the clustering of visits, with visits nested within patients and patients within practices. We further stratified practices by practice characteristics, including geographic location, workforce composition, and patient volume, and explored the association of treatment intensification with these practice characteristics. The treatment intensification score was defined as (observed − expected medication intensifications) divided by the number of visits for each patient. Expected intensifications were determined according to guideline recommendations for visits with uncontrolled BP, such that higher scores indicated more frequent intensification relative to guideline expectations.25 The score was included as a continuous variable in the model to examine its association with change in systolic BP from baseline to the last visit, adjusting for baseline systolic BP, intervention group, and total number of visits. Restricted cubic splines were also used to detect potential nonlinear associations.26 In sensitivity analyses restricted to the intervention group, models were additionally adjusted for treatment fidelity to assess whether adherence to CDSS recommendations was associated with treatment intensification and BP control. All the analyses were performed using SAS software, version 9.4 (SAS Institute Inc). All P values were from 2-sided tests, and results were deemed statistically significant at P < .05. Results We included 4612 patients from 93 practices with uncontrolled BP at baseline (median age, 63 years [IQR, 52-74 years]; 2648 men [57.4%] and 1964 women [42.6%]) (Table 1). Of these, 2134 were in the CDSS group and 2478 were in the usual care group (eFigure 1 in Supplement 2). Baseline BP distributions were similar between groups, although several baseline characteristics differed modestly, including age, educational level, smoking status, diabetes, hyperlipidemia, and the number of antihypertensive medication classes used. Across participating practices, the median number of patients with hypertension managed per practice was 919 (IQR, 591-1240) (Table 1). Participating physicians had a median age of 43 years (IQR, 33-56 years), and 90.2% (313 of 347) held a bachelor’s degree or higher. Baseline characteristics of participating practices, physicians, and patients are summarized in Table 1. Table 1. Baseline Practice, Physician, and Patient Characteristics by Intervention GroupCharacteristicOverallCDSSUsual care Primary care practices No. of primary care practices934548No. of patients with hypertension managed per site, median (IQR)919 (591-1240)838 (622-1171)925 (588-1461)No. of in-service physicians per site, median (IQR)8 (4-10)7 (3-11)8 (5-10) Participating physicians No. of participating physicians347176171Age, median (IQR), y43 (33-56)43 (33-55)43 (34-60)Sex, No. (%) Male139 (40.1)68 (38.6)71 (41.5) Female208 (59.9)108 (61.4)100 (58.5)Educational level, No. (%) Bachelor’s degree or above313 (90.2)161 (91.5)152 (88.9) Junior college or below34 (9.8)15 (8.5)19 (11.1) Participants No. of participants461221342478Age, median (IQR), y63 (52-74)62 (51-74)64 (53-75)Sex, No. (%) Male2648 (57.4)1234 (57.8)1414 (57.1) Female1964 (42.6)900 (42.2)1064 (42.9)Educational level, No. (%) Primary or low928 (20.1)412 (19.3)516 (20.8) Middle school1553 (33.7)788 (36.9)765 (30.9) High school or above2131 (46.2)934 (43.8)1197 (48.3)Current smoker, No. (%)720 (15.6)368 (17.2)352 (14.2)Disease history, No. (%) Diabetes 917 (19.9)397 (18.6)520 (21.0) Hyperlipidemia676 (14.7)268 (12.6)408 (16.5) Stroke247 (5.4)109 (5.1)138 (5.6)Systolic BP, mean (SD), mm Hg148.5 (10.4)148.6 (10.5)148.3 (10.4)Systolic BP category, No. (%) <140 mm Hg620 (13.4)283 (13.3)337 (13.6) ≥140 and <160 mm Hg3307 (71.7)1528 (71.6)1779 (71.8) ≥160 mm Hg685 (14.6)323 (15.1)362 (14.9)Diastolic BP, mean (SD), mm Hg84.8 (10.7)85.0 (10.6)84.7 (10.7)Diastolic BP category, No. (%) <90 mm Hg2738 (59.4)1273 (59.7)1465 (59.1) ≥90 and <100 mm Hg1581 (34.3)712 (33.4)869 (35.1) ≥100 mm Hg293 (6.4)149 (7.0)144 (5.8)Classes of antihypertensive medications, No. (%) 0491 (10.6)268 (12.6)223 (9.0) 12296 (49.8)1026 (48.1)1270 (51.3) ≥21825 (39.6)840 (39.4)985 (39.7)Abbreviations: BP, blood pressure; CDSS, clinical decision support system. Baseline Practice, Physician, and Patient Characteristics by Intervention Group Abbreviations: BP, blood pressure; CDSS, clinical decision support system. Treatment Intensification In the unadjusted model, the treatment intensification rate was significantly higher in the CDSS group than in the usual care group (38.6% [95% CI, 33.1%-44.5%] vs 9.9% [95% CI, 7.9%-12.2%]; odds ratio [OR], 5.76 [95% CI, 4.10-8.10]; P < .001) (Table 2). After adjustment for patient and practice characteristics, the difference remained statistically significant (47.3% [95% CI, 40.1%-54.6%] vs 11.6% [95% CI, 8.9%-14.9%]), with an adjusted OR of 6.87 (95% CI, 4.90-9.64; P < .001) favoring the CDSS group. The corresponding MORs were 1.22 (95% CI, 1.18-1.29) in the CDSS group and 1.08 (95% CI, 1.07-1.10) in the usual care group (eTable 1 in Supplement 2). After further adjustment for adherence to CDSS recommendations in the CDSS group, the MOR decreased to 1.09 (95% CI, 1.07-1.11). Across BP thresholds, treatment intensification increased with a higher threshold in both groups, and the MOR varied across thresholds within each group (eTable 2 in Supplement 2). Table 2. Treatment Intensification by the Intervention GroupModelAntihypertensive treatment intensification rate (95% CI), %Odds ratio (95% CI)P valueCDSS (n = 4940)Usual care (n = 6809)Model 0: unadjusteda38.6 (33.1-44.5)9.9 (7.9-12.2)5.76 (4.10-8.10)<.001Model 1: adjusted for patient characteristicsb42.3 (35.4-49.5)9.6 (7.4-12.5)6.87 (4.75-9.93)<.001Model 2: adjusted for patient and practice characteristicsc47.3 (40.1-54.6)11.6 (8.9-14.9)6.87 (4.90-9.64)<.001Model 3: adjusted for patient and practice characteristics, and treatment fidelityd56.3 (49.1-63.2)NANANAAbbreviations: CDSS, clinical decision support system; NA, not applicable. a Model 0: no covariates were included; practices and patients were modeled as random effects. b Model 1: based on model 0; individual patient-level covariates were included as fixed effects, including sex, age, systolic blood pressure, classes of antihypertensive medication used, educational level, current smoking status, and medical history (diabetes, dyslipidemia, and stroke). c Model 2: based on model 1; additional adjustment was made for practice-level characteristics, including geographic location, number of in-service physicians, educational level of physicians, age of physicians, and patient volume. d Model 3: based on model 2; additional adjustment was made for treatment fidelity. Owing to missing data on treatment fidelity for 77 cases, the total number of cases included in the analysis was 4863. Treatment Intensification by the Intervention Group Abbreviations: CDSS, clinical decision support system; NA, not applicable. Model 0: no covariates were included; practices and patients were modeled as random effects. Model 1: based on model 0; individual patient-level covariates were included as fixed effects, including sex, age, systolic blood pressure, classes of antihypertensive medication used, educational level, current smoking status, and medical history (diabetes, dyslipidemia, and stroke). Model 2: based on model 1; additional adjustment was made for practice-level characteristics, including geographic location, number of in-service physicians, educational level of physicians, age of physicians, and patient volume. Model 3: based on model 2; additional adjustment was made for treatment fidelity. Owing to missing data on treatment fidelity for 77 cases, the total number of cases included in the analysis was 4863. Treatment intensification was more likely in the CDSS group than in the usual care group (OR, 6.87 [95% CI, 4.90-9.64]) and was also higher in practices located in the northern region and those with a greater proportion of more senior physicians, whereas other practice characteristics were not significantly associated with treatment intensification (eTable 3 in Supplement 2). Larger MORs in the CDSS group were observed in the northern region, in strata with fewer and more senior physicians, lower physician educational attainment, and higher patient volume (eTable 4 in Supplement 2). BP Control A total of 2335 patients (50.6%) achieved BP control at their final visit. The crude BP control rate was 52.4% in the CDSS group (1118 of 2134) and 49.1% in the usual care group (1217 of 2478) (P = .03). After adjustment for patient and practice characteristics, the adjusted BP control rates were similar between groups: 49.3% (95% CI, 43.5%-55.1%) in the CDSS group and 47.7% (95% CI, 42.1%-53.4%) in the usual care group, with an adjusted OR of 1.07 (95% CI, 0.85-1.34; P = .58) (Table 3). The adjusted BP control rates and MORs in the CDSS group did not change significantly after additional adjustment for adherence to the CDSS recommendation in the model (eTable 5 in Supplement 2). Table 3. BP Control by the Intervention GroupModelBP control rate (95% CI), %Odds ratio (95% CI)P valueCDSS (n = 2134)Usual care (n = 2478)Model 0: unadjusteda53.7 (49.4-57.9)50.9 (46.8-55.0)1.12 (0.88-1.42).35Model 1: adjusted for patient characteristicsb50.3 (44.7-55.9)48.2 (42.8-53.6)1.09 (0.86-1.37).47Model 2: adjusted for patient and practice characteristicsc49.3 (43.5-55.1)47.7 (42.1-53.4)1.07 (0.85-1.34).58Model 3: adjusted for patient and practice characteristics, and treatment fidelityd46.4 (39.2-53.9)NANANAAbbreviations: BP, blood pressure; CDSS, clinical decision support system; NA, not applicable. a Model 0: no covariates were included; practices were modeled as random effects. b Model 1: based on model 0; individual patient-level covariates were included as fixed effects, including sex, age, classes of antihypertensive medication used, educational level, current smoking status, and medical history (diabetes, dyslipidemia, and stroke). c Model 2: based on model 1; additional adjustment was made for practice-level characteristics, including geographic location, number of in-service physicians, educational level of physicians, age of physicians, and patient volume. d Model 3: based on model 2; additional adjustment was made for treatment fidelity. Owing to missing data on treatment fidelity for 8 cases, the total number of cases included in the analysis was 2126. BP Control by the Intervention Group Abbreviations: BP, blood pressure; CDSS, clinical decision support system; NA, not applicable. Model 0: no covariates were included; practices were modeled as random effects. Model 1: based on model 0; individual patient-level covariates were included as fixed effects, including sex, age, classes of antihypertensive medication used, educational level, current smoking status, and medical history (diabetes, dyslipidemia, and stroke). Model 2: based on model 1; additional adjustment was made for practice-level characteristics, including geographic location, number of in-service physicians, educational level of physicians, age of physicians, and patient volume. Model 3: based on model 2; additional adjustment was made for treatment fidelity. Owing to missing data on treatment fidelity for 8 cases, the total number of cases included in the analysis was 2126. Treatment Intensification Score and BP Change Among patients with uncontrolled BP at baseline, the median treatment intensification score was −0.25 (IQR, −0.50 to 0) in the CDSS group and −0.50 (IQR, −0.75 to −0.30) in the usual care group, with a mean difference of 0.22 (95% CI, 0.17-0.28; P < .001) between groups (Figure 1; eFigure 2 in Supplement 2). The treatment intensification score was negatively correlated with systolic BP change in the overall cohort (ρ = −0.32; P < .001) (Figure 2). Each 0.22-point increase in the treatment intensification score was associated with a mean change of −3.8 mm Hg (95% CI, −4.1 to −3.5 mm Hg) in systolic BP. Box Plots Showing the Distribution of Treatment Intensification Scores by Intervention Group and Blood Pressure (BP) Control Status The light blue boxes indicate the interquartile range; the horizontal line within the light blue boxes indicates the median; the diamond within the light blue boxes indicates the mean; the whiskers indicate the range of nonoutlier values; and the circles indicate outliers. CDSS indicates clinical decision support system. Line Graph Showing Association Between Treatment Intensification Score and Systolic Blood Pressure (BP) Change The shaded area indicates the 95% CI. Discussion This post hoc secondary analysis of a cluster randomized clinical trial found that implementation of a guideline-based CDSS was associated with higher rates of treatment intensification in primary care settings. The enhancement of treatment intensification did not translate into a comparable improvement in BP control compared with usual care. In further analyses, greater treatment intensification was associated with larger reductions in systolic BP. This study extends the prior literature by providing evidence on the potential role of CDSS in promoting treatment intensification and BP control. Similar to prior studies in the US,27,28 the usual care group in our study also exhibited low intensification rates among patients with uncontrolled BP, highlighting the common existence of clinical inertia in routine practice. In this context, CDSS implementation was associated with increased treatment intensification, suggesting that point-of-care decision support may help address this gap in hypertension management. Prior studies have shown that interventions designed to reduce therapeutic inertia, including audit and feedback strategies, reminder tools, and team-based care, can increase treatment intensification29 and may lead to improvements in BP control.30,31,32 Compared with these multicomponent approaches, the CDSS evaluated in this study represents a pragmatic intervention, which is a low-cost, sustainable approach embedded in routine clinical workflows. Although no significant improvement in BP control was observed in the CDSS group, the increase in treatment intensification, together with prior evidence of greater systolic BP reduction in the previous report,17 suggests that CDSS may still represent a practical strategy worth broader implementation in primary care. The observed differences in treatment intensification between groups, targeting the commonly used threshold of 140/90 mm Hg, indicate greater heterogeneity in treatment intensification at this threshold. In the subgroup analysis, greater heterogeneity was also observed in the CDSS group with lower physician educational attainment, fewer and more senior in-service physicians, and higher patient volume. However, these differences were attenuated after adjustment for treatment fidelity, suggesting that differences in adherence to CDSS recommendations may be partly associated with the heterogeneity observed across practices. These findings suggest that although CDSS can facilitate guideline-based treatment decisions, the extent to which recommendations are adopted in practice may depend on local clinical environments and operational constraints. Additional organizational strategies, such as targeted training, workload redistribution, and workflow optimization,33,34 may be necessary to achieve more uniform and sustained adoption of digital decision support across diverse primary care environments. Despite the enhancement in treatment intensification, no significant improvement in BP control was observed in the CDSS group compared with the usual care group. Several factors may explain this finding. First, although higher treatment intensification scores were associated with greater BP reduction, consistent with prior evidence supporting the benefits associated with treatment intensification,35 the overall increase in treatment intensity may not have been sufficient to substantially improve BP control at the population level. Additional efforts incorporating the use of CDSSs are needed. Second, this apparent discrepancy may partly reflect a statistical aggregation phenomenon similar to the Simpson paradox,36 whereby treatment intensification tends to occur more frequently among patients with higher baseline BP at each visit, potentially attenuating improvements in overall BP control at the population level. In addition, a large proportion of patients had moderately elevated BP levels, which may have limited the magnitude of medication escalation in routine clinical practice. Differences in the uptake of or adherence to CDSS recommendations across clinical settings may also have been associated with the extent to which increased treatment intensification translated into improved BP control. Our findings highlight the potential role of CDSSs in promoting more guideline-concordant antihypertensive treatment in routine primary care. By enhancing treatment intensification, the use of CDSSs may help address therapeutic inertia, a well-recognized barrier to effective hypertension management. The implementation of a CDSS was associated with greater reductions in systolic BP compared with usual care. These findings suggest that CDSSs alone may not be sufficient to achieve large improvements in BP control at the population level. Additional strategies—such as improving physician engagement with decision support recommendations, addressing workflow barriers,37,38,39 and integrating CDSSs with broader implementation or care-delivery interventions—may help maximize their impact in routine clinical practice.33 Limitations This study has several limitations. First, although treatment intensification and BP outcomes were assessed using data from a cluster randomized clinical trial, residual confounding from unmeasured factors cannot be ruled out. Specifically, the structural characteristics of practices, such as academic affiliation, were not collected in the trial dataset and therefore were not included in this analysis. Second, the treatment intensification score reflects the frequency rather than the magnitude of medication adjustments, which may underestimate the true intensity of treatment escalation. Third, behavioral changes in the usual care group cannot be excluded; better performance because of participation in the trial may have underestimated the effectiveness of CDSS use. Conclusions In this post hoc secondary analysis of a cluster randomized clinical trial, we found that implementation of a CDSS in primary care settings was associated with increased antihypertensive treatment intensification. However, the increase in treatment intensity observed with CDSS implementation did not translate into significant improvements in overall BP control. Additional efforts may be required to optimize the use of decision support tools in clinical practice to achieve sustained improvements in hypertension control.