Remote follow-up by pharmacists for blood pressure control in patients with hypertension: a systematic review and a meta-analysis of randomized controlled trials

Hypertension is a major cause of cardiovascular diseases. Several recent studies reported that pharmacists’ remote follow-up reduced hypertension patients’ blood pressure (BP). This meta-analysis aims to verify whether remote follow-up by pharmacists improves BP levels and reveal the factors that make the intervention effective. The search, conducted using PubMed/Medline, Embase, and Cochrane Library from June to July 2023, targeted articles published between October 1982 and June 2023, using terms including “pharmacist”, “hypertension”, and “randomized controlled trial (RCT)”. The inclusion criteria were: (a) RCTs involving hypertension patients with or without comorbidities, (b) pharmacists using remote communication tools to conduct follow-up encounter during the intervention period, (c) reporting systolic blood pressure (SBP) at baseline and during intervention. SBP was the primary outcome for the meta-analysis. Thirteen studies (3969 participants) were included in this meta-analysis. The mean difference of SBP between intervention group and control group was − 7.35 mmHg (P < 0.0001). Subgroup analyses showed the greater reduction of SBP in the “regularly scheduled follow-up cohort” (− 8.89 mmHg) compared with the “as needed follow-up cohort” (− 3.23 mmHg, P < 0.0001). The results revealed that remote follow-up by pharmacists reduced SBP levels in hypertension patients and scheduled remote follow-up may contribute to the effectiveness.


Effect of remote follow-up for BP control
The forest plot of comparison between intervention group (IG) and control group (CG) at the final point of the intervention period in 13 included studies is shown in Fig. 2. The mean difference in SBP between IG and CG was − 7.35 mmHg (95% Confidence Interval [CI] − 9.10 to − 5.59 mmHg, P < 0.0001), and there was a high degree of heterogeneity (χ 2 = 44.09,df = 12, I 2 = 73%).
Figure 1.Flow diagram of screening.Duplicate records, non-English records, conference reports, reviews, meta-analyses, research protocols, and reports about non-randomized controlled trial (RCT) trials were excluded from resulting records.The inclusion criteria for the meta-analysis were as follows: (a) RCTs involving hypertension patients with or without coexisting chronic conditions, (b) pharmacists using remote communication tool(s) to follow up with subjects during the intervention period, (c) studies reporting systolic blood pressure at baseline and during the intervention period.In addition, the studies with inadequate values for standard deviation or 95% confidence interval was excluded.www.nature.com/scientificreports/

Regularity of follow-up
The mean differences in SBP between IG and CG were − 8.89 mmHg (95% CI − 10.11 to − 7.66 mmHg, P < 0.0001) and − 3.23 mmHg (95% CI − 5.72 to − 0.74 mmHg, P = 0.01) in the "regularly scheduled follow-up cohort (RFC)" and the "as needed follow-up cohort (AFC), " respectively (Fig. 3).In addition, there was a significant subgroup difference between RFC and AFC (P < 0.0001).Regarding heterogeneity, significant reductions were observed in both subgroups (P = 0.20, I 2 = 28% in RFC, and P = 0.19, I 2 = 37% in AFC) compared with the overall population (P < 0.0001, I 2 = 73%).Because there were two studies 19,27 with large differences in baseline SBP, we repeated the analysis with these two studies removed.Even after excluding these studies, the SBP reduction effect by pharmacist remote followup interventions and subgroup differences between RFC and AFC remained significant (Supplementary Fig. 1).www.nature.com/scientificreports/

Type of communication tools
The mean difference in SBP between IG and CG in the "telephone tool cohort (TTC)" and "other communication tool cohort (OCC)" were − 8.04 mmHg (95% CI − 9.85 to − 6.22 mmHg, P < 0.0001) and − 5.49 mmHg (95% CI − 10.32 to − 0.67 mmHg, P = 0.03), respectively (Fig. 4).There was no significant subgroup difference between TCC and OCC (P = 0.33).Regarding heterogeneity, the I 2 of each subgroup was not significantly different from that of the overall group.www.nature.com/scientificreports/

Communication with physician
The mean difference in SBP between IG and CG in the "physician communication cohort (PCC)" and "no physician communication cohort (NPC)" were − 6.56 mmHg (95% CI − 8.84 to − 4.28 mmHg, P < 0.0001) and − 8.96 mmHg (95% CI − 11.50 to − 6.42 mmHg, P < 0.0001), respectively (Fig. 5).No significant subgroup difference was observed between PCC and NPC (P = 0.17).Regarding heterogeneity, the I 2 of each subgroup was not significantly different from that of the overall group.

Risk of bias assessment
The quality of included studies assessed by the Risk of Bias 2 (ROB2) tool is shown in Table 2.Because the number of excluded participants was not described in the study by Solomon and colleagues, it was rated high risk for missing the outcome data 21 .Due to insufficient randomization information and variability in baseline values, the study by Zaragoza-Fernandez et al. 27 was rated high risk.www.nature.com/scientificreports/A funnel plot was created using mean differences in SBP at the final point of the intervention period in the 13 included studies (Fig. 6).Egger's test for a regression intercept yielded a p value of 0.87, indicating no evidence of publication bias.For the three subgroup analyses, we considered the number of studies in each subgroup to be insufficient to statistically assess publication bias.Result of trim-and-fill analysis is also shown in Supplementary Table 1.There was no sign of publication bias (Mean differences of SBP and 95% CI were slightly different between using Review Manager and using Stata).

Sensitivity analysis
Results of sensitivity analyses are shown in Supplementary Table 2.The results were not substantively different under any conditions examined.

Discussion
Because of the recent COVID-19 pandemic, the attention to remote patient encounters and follow-ups has increased more than ever.The present study revealed that remote follow-up by pharmacists improved SBP levels compared with usual care in patients with hypertension.In the intervention group receiving pharmacists' remote follow-up, regularly scheduled follow-up was found to be an important factor in the success of the intervention compared with as needed follow-up.A previous meta-analysis about pharmacist interventions in hypertension have revealed that interventions performed more often than once a month tended to be more effective than interventions performed less than once a month 28 .The same meta-analysis included trials outside of remote interventions, thus differing from the present study.However, our results were consistent with the previous study findings in that pharmacist interventions were useful for improving SBP among people with hypertension.Although our study did not evaluate the number of follow-ups performed, we saw that scheduling regular followup decreased SBP more effectively than scheduling follow-up as needed.Such a finding is novel and supports implementation of intentionally scheduling remote follow-ups by pharmacists for patients with hypertension.Regular follow-up with expected resource needs and devoted appointment time might benefit both patients and clinicians to achieve the treatment goals of chronic diseases such as hypertension.
In the present study, remote interventions by pharmacists using only telephone were not inferior to those using at least one other communication tool such as web communications or text messages.A network metaanalysis comparing the antihypertensive effects of interventions using various communication tools such as telephones, websites, Short Message Service (SMS) and smartphone-application software (apps) found that the combination of two or more tools was most effective, though the second most effective group used the telephone alone in lowering SBP 29 .The interventions in this meta-analysis were not delivered solely by pharmacists as in our study.In the present study, it was not possible to compare different types of communication tools because only two cohorts were formed due to small sample size, and one of the groups contained various types of communication tools used by pharmacists.But the result that remote interventions using telephone only were not inferior to those using other communication tools in lowering SBP was consistent with past study findings.In other words, pharmacists who do not have communication tools other than the telephone can provide effective interventions for hypertension patients if they follow-up regularly according to the planned schedule.At the same time, telephone calls can be inconvenient for some patients because they must be available at the time of the call.Although there might be a time lag from sending a message to checking and responding to the message, smartphone apps, emails, or text messages may be more convenient ways to communicate.Non-telephone tools were used for remote follow-up in three studies, but none of these examined age-related differences on intervention effectiveness.Therefore, it was not possible to assess statistically the impact of age-related differences in ability to use communication tools on intervention effectiveness except for telephone.Further studies are needed to evaluate the ability to use various communication tools by older patients, and when pharmacists follow up remotely with patients having hypertension, it would be best to use communication tools that fit the individual patient's lifestyle and ability best.
Previous meta-analyses comparing pharmacist-led interventions and collaborative interventions by pharmacists and other healthcare professionals for hypertension patients showed that pharmacist-led interventions www.nature.com/scientificreports/tended to be more effective.However, these intervention methods were not limited to remote follow-ups, as in the present study 28 .Thus, our study findings add to the existing evidence on pharmacist interventions and are consistent with previous study findings in which pharmacist-led interventions were effective regardless of collaboration with other professionals.However, the quality and frequency of pharmacists' reports to physicians in PCC were not evaluated in our study.To enhance the effects of remote follow-up, the quality and timing of communication with physicians may be evaluated in future studies.One of the limitations of the study is that the research outcome was limited to SBP as diastolic BP (DBP) values were not available in all included studies.The timing of SBP measurements differed among the included studies.The DBP outcome and timing of BP measures should be evaluated in future studies.We could not evaluate the impact of comorbidities and age on the intervention because the SBP of patients with each comorbidity was not available from the included RCTs.In AFC, follow-up was carried out as needed, so it is possible that not all subjects received remote follow-up by pharmacists during the study period.The impact of COVID-19 could not be assessed in the current study because study periods indicated by the included articles were prior to the beginning of the COVID-19 pandemic.

Conclusion
The present meta-analysis revealed that remote follow-up by pharmacists reduced SBP in patients with hypertension.In addition, regularly scheduled follow-up contributed to the success of remote follow-up compared with as needed follow-up.Higher quality studies are needed to identify the ideal combination of remote follow-up communication tools and methods that affect BP reduction.

Data search and study selection
Based on Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2020 guideline 30 (PRISAMA2020 check list: Supplementary Table 3, PRISAMA2020 abstract check list: Supplementary Table 4), the search was conducted using PubMed/Medline, Embase, and Cochrane Library from June to July 2023.We searched for available articles published between October 1982 and June 2023.Searched terms included "pharmacist, " "hypertension, " and "randomized controlled trial" (Terms used for the search and the PubMed search details are shown in Supplementary Table 5).The search strategy is summarized in Fig. 1.Duplicate records, non-English records, conference reports, reviews, meta-analyses, research protocols, and reports about non-RCT trials were excluded from resulting records.Two reviewers independently assessed the articles for eligibility and data extraction and resolved disagreements by consensus.The inclusion criteria for the meta-analysis were as follows: (a) RCTs involving hypertension patients with or without coexisting chronic conditions (e.g., diabetes, CKD, CVD, stroke, human immunodeficiency virus [HIV] infection), (b) pharmacists using remote communication tool(s) to follow up with subjects during the intervention period, (c) studies reporting SBP at baseline and during the intervention period.Extracted data included study setting, characteristics of participants, intervention periods, types of remote communication tools used (e.g., telephone, text message, web communications, mail), number of study subjects and SBP.Finally, we excluded studies with inadequate values for Standard Deviation (SD) or 95% CI.

Outcome and data analysis
We selected SBP as the primary outcome of this meta-analysis because a reduction in SBP lowered the risk of cardiovascular events in the previous research.For instance, a 5 mmHg and 10 mmHg reduction in SBP decreased the risk of developing cardiovascular events by 10% and 20%, respectively 3,5 .The subjects receiving remote followup were categorized into the "IG", while those receiving usual in-person follow-up categorized into the "CG".We conducted the meta-analysis to elucidate the regularity of follow-up, the types of communication tools, and communication with physicians.We used the software Review Manager version 5.4 (The Cochrane Collaboration, London, UK).The mean difference, SD, and 95% CI were used to estimate effects.The calculator of Review Manager was used when it was necessary to calculate SD from the mean value of SBP and 95% CI.If the differences between two 95% CI values and the mean value differed by 0.1 from each other, the smaller value of 95% CI was adopted.The meta-analysis was undertaken using the random-effects model with the results presented in a forest plot.Statistical heterogeneity was evaluated by the I 2 statistic.Since the present study was a literature review and meta-analysis of published data, no ethical or human subject protection evaluation was required.

Assessment of risk of publication bias
Two reviewers independently assessed the risk of bias with any disagreement resolved by consensus.Cochrane's ROB2 tool was used to assess the risk of bias 31 .This tool contains the following six assessment domains: (a) randomization process; (b) deviation from the intended interventions; (c) missing outcome data; (d) measurement of the outcome; (e) selection of the reported results; and (f) overall risk of bias.Each domain was ranked "low risk of bias, " "some concerns, " or "high risk of bias." In addition, a funnel plot was constructed and Egger's test 32 and trim-and-fill analysis 33 were conducted to detect the presence of potential publication bias in this random-effects meta-analysis model (Restricted Maximum Likelihood method) using statistical software Stata /MP version 18.0 (Stata Corp LLC, College Station, USA).

Sensitivity analysis
To assess data robustness, a sensitivity analyses was conducted by using Review manager.In the sensitivity analyses, we evaluated whether the results were affected by (1) excluding the study with the highest number of participants, (2) excluding the study in which the intervention reduced SBP the most, (3) excluding studies with high bias, (4) changing measurement time from the final point to the earlier time of intervention period (There were two studies in that SBP was measured at multiple times during the intervention period, and SBP was measured twice in both studies during the intervention period) and ( 5) changing from the random-effects model to the fixed-effect model.

Remote follow-up success factor
We conducted three subgroup analyses to identify factors contributing to success of remote follow-up by pharmacists for BP improvement.

Regularity of follow-up
Thirteen studies were divided into two groups according to the following criteria.The studies having specific contact time or frequency of interventions were classified as "RFC", while the studies without specific contact time or frequency of interventions were classified as "AFC".RFC also contained the studies in which they conducted both regular and as needed follow-up.

Types of communication tools
Thirteen studies were divided into two groups according to the following criteria.The studies that used telephone only were classified as "TCC", while the studies that used various communication tools were classified as "OCC" (Supplementary Table 1).OCC also contained the studies in which both telephone and other communication tools were used.

Communication with physician
Thirteen studies were divided into two groups according to the following criteria.In "PCC", there were descriptions in the articles that pharmacists reported information from the patient encounters or their recommendations to physicians, while there was no description about them in the "NPC".

Figure 2 .
Figure 2. Forest plot of comparison between the intervention and control groups in 13 studies at the final point of the intervention period.A meta-analysis was conducted using the random-effects model.The total number of patients was 3969.The mean difference in SBP between the intervention and control groups was − 7.35 mmHg (95% CI − 9.10 to − 5.59 mmHg, P < 0.0001).SBP systolic blood pressure, SD standard deviation, CI confidence interval.

Figure 3 .
Figure 3. Forest plot of comparison between the "regularly scheduled follow-up cohort" and the "as needed follow-up cohort" at the final point of the intervention period.A meta-analysis was conducted using the random-effects model.The total number of patients and studies was 3969 and 13, respectively.The mean difference of SBP was − 8.89 mmHg (95% CI − 10.11 to − 7.66 mmHg, P < 0.0001) and − 3.23 mmHg (95% CI − 5.72 to − 0.74 mmHg, P = 0.01) in RFC and AFC, respectively.SBP systolic blood pressure, SD standard deviation, CI confidence interval, RFC regularly scheduled follow-up cohort, AFC as needed follow-up cohort.

Figure 4 .
Figure 4. Forest plot of comparison between the "telephone tool cohort" and "other communication tools cohort" at the final point of the intervention period.Meta-analysis was undertaken with random-effects models.The total number of patients and studies was 3969 and 13, respectively.The mean difference of SBP in TTC and OCC was − 8.04 mmHg (95% CI − 9.85 to − 6.22 mmHg, P < 0.0001) and − 5.49 mmHg (95% CI − 10.32 to − 0.67 mmHg, P = 0.03), respectively.SBP systolic blood pressure, SD standard deviation, CI confidence interval, TTC telephone tool cohort, OCC other communication tools cohort.

Figure 5 .
Figure 5. Forest plot of comparison between "physician communication cohort" and "no physician communication cohort" at the final point of the intervention period.(A) meta-analysis was conducted using the random-effects model.The total number of patients and studies was 3969 and 13, respectively.The mean difference of SBP in PCC and NPC was − 6.56 mmHg (95% CI − 8.84 to − 4.28 mmHg, P < 0.0001) and − 8.96 mmHg (95% CI − 11.50 to − 6.42 mmHg, P < 0.0001), respectively.SBP systolic blood pressure, SD standard deviation, CI confidence interval, PCC physician communication cohort, NPC no physician communication cohort.

Figure 6 .
Figure 6.Funnel plot using mean differences of SBP at the final point of the intervention period in 13 studies.Mean diff.Mean difference of SBP, SBP systolic blood pressure.

Table 1 .
Characteristics of studies and patients.SD standard deviation, CI confidence interval, SBP systolic blood pressure, USA The United States of America, IG intervention group, CG control group, NA not available, SMS short message service, CKD chronic kidney disease, CVD cardiovascular diseases, DM diabetes mellitus, HIV human immunodeficiency virus.* 1 IG had been subdivided into two groups with separate baseline.* 2 Results after 9 months were excluded in the present study because outcomes after 9 months were measured separately for minority and non-minority groups.* 3 SBP measured at time 2 (about 15 min after arrival of patients to the clinic) was used for the present analysis.

Table 2 .
Risk of bias assessment.The studies were assessed by following six domains; "Randomization process", "Deviation from the intended interventions", "Missing outcome data", "Measurement of the outcome", "Selection of the reported result", and "Overall".