Validation of Blood Pressure Devices as per 2020 World Health Organization Technical Specifications: A Scoping Review of Global Literature

Publications between January 1, 2000, and December 21, 2021, were included to identify the latest policy-relevant research literature. Primary research published in English as journal articles, thesis/dissertation, books, organizational reports, and case reports that provided primary data on validation of BPMDs were included. Qualitative studies, perspectives, commentaries, opinion papers, letters to the editor, review articles, patents, conference abstracts, guidelines, and secondary research were excluded. In addition, modeling studies that were based on secondary data sources were excluded.

The Population, Concept, Context notation of Joanna Briggs Institute²⁸ was used to define the eligibility criteria for studies.

Electronic searches were conducted on MEDLINE (via PubMed), Web of Science, Scopus, EMBASE, CINAHL, CENTRAL, ProQuest, and dabl Educational Trust website. Population, concept, context guided the 3 domains for search, and each domain was combined with the Boolean operator AND. Keywords within each domain were combined using the Boolean operator OR. The Dabl Educational Trust website was designed to provide updated information to consumers on validation results of BP monitors.²⁹ A search strategy was developed on PubMed and subsequently translated to other databases. The electronic search strategies for all databases can be accessed on Open Science Framework.³⁰ The date of the last search was December 21, 2021.

Retrieved reports were exported to PICO Portal³¹—an online review platform that facilitated study selection and preliminary data extraction for our review. The portal facilitated automatic deduplication of records. Screening was performed in 2 stages: title-abstract and full-text screening. A screening protocol was drafted and revised following multiple internal team discussions. The reviewers performing the screening (K.S., P.K., and R.J.) were trained and calibrated in the screening process for both the stages of screening by S.M. and E.R. Given our scope and objectives, and to minimize the risk of excluding potentially relevant studies, we were overinclusive in the title-abstract stage. Full texts were retrieved by S.M. and E.R. References to validation reports in the dabl website were screened for eligibility. Three reviewers (K.S., P.K., and R.J.) screened the title-abstracts and full texts in duplicate. Two reviewers (S.M. and E.R.) resolved conflicts through discussion and reaching consensus where required.

Data extraction was first done by 4 trained reviewers in single. Three of these reviewers (P.K., K.S., and R.J.) extracted data for peer-reviewed articles, and 1 reviewer (R.R.) extracted data for the website reports. The 4 reviewers subsequently cross-checked the data extracted by the other reviewers. Two senior reviewers (S.M. and E.R.) verified 10% of the extracted data randomly for quality (eg, accuracy, completeness) and flagged errors or conflicts, if any, in the verified data set. Disagreements were resolved via discussion and consensus or consultation with the senior review authors.

Data were exported from the PICO portal to Microsoft Excel. Subsequently, data were cleaned, managed, and analyzed descriptively using frequencies and percentages on Microsoft Excel (S.M., E.R., and R.R.). Geographical locations were mapped to create a visual summary on Tableau Desktop Public Edition, V.2021 (S.M. and R.R.).³³ The results have been summarized narratively and supplemented with tables and figures in the Supplemental Material. The scoping review has been reported in line with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for Scoping Reviews Checklist.³⁴

Of the total 11 825 search results, there was a total of 4126 duplicates. Following deduplication, 7133 titles and abstracts were found to be irrelevant and excluded after title and abstract screening. Of the 566 full-text records (eligible for full-text screening), 28 reports could not be retrieved. At the full-text screening stage, of 538 full-text articles, 280 articles were excluded because of non-English language (n=1), study population not humans (n=6), study designs such as qualitative studies, narrative reviews, modeling studies of secondary data (n=69), and studies that were nonvalidation studies for automated/semiautomated BPMDs (n=203). Therefore, a total of 258 studies from electronic databases were included in the review. From the additional resources (dabl website and backward citation screening), a total of 264 full texts were identified, and 260 full texts were screened for eligibility. Of these, 249 full texts were excluded due to not being in English (n=4), publication date before 2000 (n=16), and 229 full texts were excluded as these were equivalence studies or trials (trials that assess similarity between 2 interventions) and not clinical validation studies. Finally, a total of 269 studies (258 from electronic databases and 11 from additional resources) were included in the review (Table S1). The study flow and reasons for exclusion are displayed in Figure S1. The characteristics of included studies can be accessed on Open Science Framework.³⁰ As shown in Figure S2, 2001 had the minimum number of studies (n=4). The maximum number of studies (n=21 each) were published in 2007 and 2008.

We identified 89 BPMD manufacturers and 251 BPMD models from the included studies. Omron (29%; 59 models), Microlife (10%; 12 models), and A&D Company (8%; 20 models) were the top 3 BPMD manufacturers. The device manufacturer was not specified in 36 (13%) validation studies. Complete lists of identified BPMD manufacturers and BPMD models are provided in Tables S2 and S3. Around 91% and 87% of the devices being validated were automated and oscillometric BPMDs, respectively.

Of 269 validation studies, 189 studies mentioned only one country as the country of population recruitment, in 8 studies, multicountry recruitment was performed, and 72 studies did not clearly specify the country of recruitment. The distribution of studies for country of recruitment is given in Figure S3, with 40 studies reporting China as the country of population recruitment for validation studies. This was followed by the United Kingdom (n=23) and the United States (n=21). Among the validation studies reporting multicountry recruitment, 2 studies were conducted in the United Kingdom and South Africa. While most validation studies reported the country from which the participants were recruited, the settings of validation were unclear. For the purposes of our review, we assumed that the country of recruitment was the country where the BPMD was validated.

China (20%), United Kingdom (13%), and the United States (12%) were the top 3 countries where validation studies were undertaken (Figure S3). A heat map representing the 6 WHO regions (based on frequency distribution) among the included validation studies is provided in Figure S4. One-third of the validation studies were conducted in the WHO region of Europe (n=84; 31%), followed by Western Pacific (n=60; 22%) and the Americas (n=34; 13%). Three studies were multiregion validations. The WHO region of validation was unclear in one-fourth of the studies (Figure S4).

Approximately 45% (n=140) and 38% (n=118) of the BPMDs were validated for clinical and home use, respectively (Figure S5); some devices were validated for use in >1 setting. The other settings of BPMD use reported by study authors included research/epidemiological purposes (n=13), ambulatory (n=10), prehospital (n=10), patient use or monitoring (n=4), community/low resource (n=5), and nonmedical public use (n=1). The settings of use were unclear for 46 BPMDs.

We retained the term ambulatory use as reported in included studies or for monitors worn by the patient throughout the day. These could indicate BPMDs that are programmed to measure BP at various intervals throughout the day and night over a 24-hour period while the patient is performing routine activities.^35,36 Home monitoring or self-measurement referred to BP monitors that were used for self-measurement at home or work, typically by the patients or lay population. Prehospital use was captured as reported in the included studies and typically referred to the ambulance transport of critically ill patients. Similarly, patient use was captured as reported in studies, with no elaboration on the definitions including that for the device applicability and did not include the terms clinical, office, home, or self use in the study results, discussion, or concluding remarks of the study.

The applicability of the validated BMPDs for use in the target population was unclear for 6 BPMDs. Poor reporting practices in the included studies made determining exactly how many BPMDs were applicable among the general adult or patient population particularly challenging. Unclear or incomplete information on what the study set out to do versus what was finally achieved related to the population or setting of actual use is one such example or suboptimal reporting practices. Nonetheless, 139 BMPDs were applicable for use in the adult population (excluding pregnant women); 58 of these BPMDs were reported to be applicable specifically among the general adult population (Table S4). Twenty-five BPMDs were applicable for use in infants, children, and adolescents, whereas 21 BPMDs were applicable for use in pregnancy or preeclampsia. Specific information including whether patients or nonpatients were recruited were unclear for 126 BPMD validations. These numbers are not mutually exclusive as there was a mix of patient and nonpatient population in the included studies. The applicable arm circumference typically ranged from 14 to 52 cm.

The identified studies used a plethora of validation protocols for the various BPMDs (Figure S6). The three most commonly used protocols were the ESH-IP 2002, ESH-IP 2010, and the modified BHS 1993 protocol. The ESH-IP 2002 was used in 77 (29%) validation studies, where 72 studies exclusively used ESH-IP 2002 and 5 studies used ESH-IP in combination with other validation protocols. ESH-IP 2010 was used in 69 (26%) validation studies (as a single protocol in 66 studies and in combination with other protocols in 3 studies), and the modified BHS 1993 was used in a total of 69 (25%) validation studies (as the only protocol in 44 studies and in combination with other protocols in 25 studies). The AAMI/ESH/ISO universal standard (ISO 81060-2:2018) was used in 7 studies. Around 36 validation studies used >1 protocol for validation of the BPMDs. Nearly 9% (n=29) of the devices being validated were reported by study authors as not fulfilling the validation criteria (Table S3). The types and numbers of different validation protocols across studies are given in Figure S6.

A relative majority of the studies reported funding (partial/complete) for the studies; of these, 72% received support from the BPMD manufacturers. Over a fifth (22%) of manufacturers donated or loaned devices for validation. Eleven percent of the studies declared potential conflicts of interest.

We included English language studies published since 2000, excluded conference abstracts, and did not contact study authors for additional information due to resource constraints. Since we did not include other gray literature sources, non–peer-reviewed studies, for example, those conducted by contract research organizations may not have been included. However, we assess minimal risk of missing potentially relevant studies for the following reasons: we adhered to a standardized review methodology, incorporated evidence-informed search strategies, and included 7 databases and the dabl educational website (though it does not have supervision or formal endorsement by a scientific organization). Furthermore, we incorporated quality assurance measures to minimize the exclusion of potentially relevant studies, such as multiple rounds of team training and calibration and quality checks by experienced review authors. Finally, the assessment of protocol violations/adherence was beyond the scope of this review.

BP measurement using a validated BPMD is vital to control hypertension—a global public health issue and an independent risk factor for many noncommunicable diseases. Validation studies are invaluable to assess the accuracy of a BPMD across patient populations. While most of the BPMDs included in the review fulfilled the validation criteria, there are considerable gaps that BPMD research and policy can fill as suggested below.

Efforts are needed on 2 particular fronts considering the speculation on the quality of published reports and the challenges in validating BPMDs across protocols. One is on the research front: more equity-focused and contextual research is required to validate BPMDs for and from resource-constrained and primary care settings, specific target populations and conditions, and a range of arm circumferences. Furthermore, we call for improvement in the quality of reporting the validation studies to improve transparency and comparability. Second is the push for a potential strategy to accelerate the adoption of the AAMI/ESH/ISO universal standard (ISO 81060-2:2018) for BPMD validation (research). Tying both these efforts would be the adoption of an agenda at the policy, decision-making, and funding levels, ideally concurrently, to facilitate such an adoption and build research capacity, especially in resource-constrained settings. Finally, this may be best achieved by initiating a dialogue with all related stakeholders with the aim of making BPMD research, regulation, and practice more equitable.

Table of Contents

Figures S1–S6

Tables S1–S4