This meta-analysis study was conducted and reported following the Preferred Reporting Items for Systematic Reviews and Meta-analyses.14 The protocol was registered in PROSPERO (registration number CRD42021262062).

A systematic review was conducted using the PubMed and Cochrane databases to identify studies on the prevalence of hypertension among Portuguese adults. The search strategy used the following terms: (((population-based)) AND ((hypertension)) AND (prevalence)) AND (Portugal).

In addition to conducting a database search, a manual search was performed, which included reviewing reference lists from articles, unpublished data from the “b-on” platform (covering Portuguese university studies), and reports from the Statistics Portugal and the Public Health National Institute (INSA). This search took place from June 2021 to June 2022 and was updated in June 2025 to meet reviewers’ suggestions. The authors were contacted directly for further clarification if the articles were unavailable or contained incomplete data.

Studies were included if they met the following criteria:

• •

  Population-based cross-sectional or cohort studies

• •

  Other studies reporting hypertension prevalence in Portugal

• •

  Participants aged 15 years or older

• •

  Studies published between 2000 and 2020.

Studies were excluded if:

• •

  Non-probability sampling was used to identify and enroll participants

• •

  Studies investigating specific populations (high-school students, university students, specific clinical conditions other than hypertension).

Two independent reviewers screened titles and abstracts for eligibility (JP, VP). In the second stage, full-text articles were reviewed. In case of disagreement, a third reviewer (RM) resolved the discrepancies. Data from 11 eligible studies15–25 were extracted and organized in a Microsoft Excel spreadsheet. Data from eight14–16,18,21–24 out of eleven studies were included in the overall prevalence of hypertension meta-analysis. The remaining three studies enrolled participants of specific age strata and were included in the following subgroup analysis by age groups.17,19,20

Hypertension was defined based on the criteria presented by individual studies as described in Table 1. In four out of eight studies used for the primary analysis, blood pressure (BP) was objectively measured, and hypertension was defined as mean systolic BP ≥140 mmHg and/or diastolic BP ≥90 mmHg and/or the use of blood pressure-lowering medication (referred to as MH). In the remaining four studies, hypertension was self-reported and not ascertained by objective measurement (SRH).

Two independent reviewers (JP and VP) conducted a quality assessment of the studies. A third reviewer (RM) was consulted to resolve any disagreements that arose. The methodological quality of the individual studies was evaluated using the JBI critical appraisal checklist for studies reporting prevalence data.26 This checklist includes nine questions, which we categorized into three domains: participants (questions 1, 2, 4, and 9), outcome measurement (questions 6 and 7), and statistics (questions 3, 5, and 8).

Studies were rated based on the following criteria:

• •

  Low risk of bias: Achieving the highest possible score in all three domains

• •

  Medium risk of bias: Achieving the highest possible score in two out of three domains, with an overall total score of seven or eight points

• •

  High risk of bias: Scoring less than seven points or having only one or none of the domains achieve the maximum score

Among the eligible studies, five were categorized as having a low risk of bias and seven as having an intermediate risk of bias, as shown in Supplementary Figure S2. Details are presented in Supplementary Table S1.

Data from included studies were summarized in Table 1, which presents key characteristics such as sample source, sampling strategy, sample size, study design, control for confounders, and participant demographics. A quantitative meta-analysis was performed when at least three studies were available for analysis. Pooled estimates and their 95% confidence intervals (CIs) were calculated using a random-effects model due to the anticipated heterogeneity of the studies. Heterogeneity was assessed using Cochran's Q test and the I2 statistic. Additionally, considering that I2 statistics are often high in prevalence studies, even when the tau-squared value is minimal, a Radial Plot (Galbraith Plot) was constructed to assess the studies’ precision and the presence of outliers. Key variables to allow sensitivity analyses were collected through the data extraction phase, such as sex, age of participants, data collection period, and hypertension measurement method (MH vs. SRH).

Subgroup analyses were conducted based on these variables of interest when available. Trends in hypertension prevalence were analyzed by comparing studies from the first (2000–2010) and second (2011–2020) decades of the 21st century. Meta-regression analyses were also conducted, including the year of publication as a covariate.

Sensitivity analyses were conducted to evaluate the effects of small-study bias, the substantial influence of studies on the meta-analysis results, and the presence of outliers through a leave-one-out analytic strategy. In addition, a pre-specified sensitivity analysis considering age and the method used to assess hypertension as potential confounders was conducted since these factors may be a significant source of heterogeneity between studies. All analyses were performed using the STATA v.18.0 software.

A total of 40 studies were identified through a systematic search of the PubMed and Cochrane databases. Five reports, one thesis, one scientific poster, and 27 manuscripts were retrieved through manual searches, including government reports from INE and INSA, as well as reference lists from relevant articles. A total of 74 papers were identified; 38 were excluded after screening titles and abstracts, and 25 more were excluded after full-text evaluation based on the predefined eligibility criteria. For transparency, a table detailing the reasons for excluding non-selected full-text papers is provided in the Supplementary Materials (Table S2). At the end of the systematic review process, 11 population-based studies14–24 were found to be eligible. Eight studies were used to assess the overall prevalence of hypertension, and four provided data for additional subgroup analyses. Figure 1 presents a flow diagram illustrating the review process.

Figure 1.

Flow diagram of the systematic review process.

The eight studies included to assess the overall prevalence of hypertension comprised a total of 131782 participants, with sample sizes ranging from 254218 to 415732 participants. Data were collected between 1998 and 2020 from participants aged 15 and older. Female participation ranged from 52.6% to 55.1% across the studies. Geographic coverage included one local study, three from mainland Portugal, and four that included data from all regions of Portugal, including both the mainland and the islands. Stratification by age, sex, and hypertension definition (SRH vs. NH) was available in most studies. Details are presented in Table 1.

As stated, three studies included specific age-targeted populations17,19,20 and were incorporated into the subgroup analysis by age stratum. The study conducted by Perdigao et al.17 enrolled participants >40 years of age, while Rodrigues et al.19 included participants aged ≥65 years. Pereira et al. enrolled participants older than 18 years; however, they oversampled individuals >40 years. This created a potential selection bias toward a higher prevalence of hypertension. These three studies were, therefore, only included in the age strata subgroup analysis.

The pooled prevalence of hypertension from 2000 to 2020 was 31% (95% CI: 25%–37%), with rates ranging from 20% to 42%. This variability reflects differences in study populations and data collection methods, highlighting the expected high heterogeneity across studies, as shown in Graph 1. However, as shown in Graph SG1 (Supplementary Material), the Galbraith Plot shows studies clustered at the far right margin and within the control lines, indicating that the included studies are precise, although some heterogeneity remains.

Although the studies were precise and no clear outliers were identified, a visual difference in prevalence rates between MH and SRH studies can still be observed. This suggests that measurement bias should be examined in a further sensitivity analysis.

An additional “leave-one-out” sensitive analysis was performed, showing no significant differences in the prevalence of hypertension by omitting one study at a time from the meta-analysis (Supplement Graph SG2). This indicates no individual study had a substantial influence on the meta-analysis results.

The prevalence of hypertension varied greatly based on the method used to assess the outcomes. In the four studies where BP was objectively measured (MH: 4 studies; n=16195), the overall pooled prevalence of hypertension was 38% (95% CI: 33%–43%). In contrast, the meta-analysis from studies relying on self-reported hypertension (SRH: 4 studies; n=115587) demonstrated a statistically significant lower pooled prevalence of hypertension of 24% (95% CI: 21%–27%) compared to MH. Details are shown in Graph 2.

Another potential source of heterogeneity is the age of enrolled participants. To address this issue, we extended the systematic review criteria to allow the inclusion of studies enrolling specific targeted age populations, such as younger and older adults only. By adding three more studies according to this criterion, a total of seven studies provided stratified data to compare hypertension prevalence by age group. The overall prevalence of hypertension in these groups is depicted in Graph SG3. Considering the previously demonstrated heterogeneity secondary to measurement bias, only MH studies were used to pool hypertension prevalence by age subgroups. Graph 3 illustrates the results described below:

• •

  15–34 years (n=16420): 11% (95% CI: 5%–18%)

• •

  35–64 years (n=28263): 44% (95% CI: 39%–50%)

• •

  65 years and older (n=15373): 77% (95% CI: 72%–81%)

A meta-regression analysis including age categories as a categorical covariate showed a clear increase in hypertension risk with advancing age. The risk was 3.9 times higher (95% CI, 2.63–5.75; p<0.001) in individuals aged 35–64 years, and 7.02 times higher (95% CI, 4.91–10.04; p<0.001) in adults aged 65 and older, compared to adults under 35 years (Supplementary Table S3).

The overall pooled prevalence rates from 2000 to 2010 and 2011 to 2020 were compared to assess trends in hypertension prevalence between the first and second decades of the 21st century. By pooling data from all eight studies in the meta-analysis, no difference in hypertension prevalence was observed across decades (Graph SG 4.1).

By focusing the analysis exclusively on MH studies and stratifying the data by age subgroups – the main sources of heterogeneity – a significant decreasing trend in hypertension prevalence was observed among both young and older adults from 2011 to 2020 compared to the period from 2000 to 2010. This is illustrated in Graph 4. In younger adults, the pooled estimate of hypertension prevalence declined from 17% (95% CI 12%–22%) to 6% (95% CI 5%–8%), p for trend >0.001. In older adults, hypertension rates decreased from 80% (95% CI 77%–84%) to 73% (95% CI 70%–77%), with a p value for trend=0.01. However, among middle-aged adults (35–64), a non-significant 5% reduction was observed, p value for trend=0.29.

Notably, heterogeneity dropped significantly after accounting for this relevant source of bias (age and method used to define hypertension), as τ2 statistics after stratification ranged from 0.00 to 0.01, and I2 statistics also declined, as shown in Graph 3.

Out of the eight studies, six provided sex-specific data on hypertension prevalence; allowing for a meta-analysis by sex subgroups. No significant differences were found in the overall hypertension prevalence between males and females. The pooled prevalence of hypertension was 33% among females (n=40336, 95% CI: 28%–38%) and 34% in males (n=35648, 95% CI: 24%–46%). Details are depicted in the Supplementary Graph SG 5.1. Considering these observed differences, Graph SG5 illustrates the prevalence of hypertension by sex in studies in which hypertension was measured (MH). In males, the pooled prevalence was 43% (95% CI 38%–48%), and in females, 36% (95% CI 32%–40%), p for group comparison=0.03.

A key strength of this study is its inclusion of population-based research with rigorous sampling methods, providing reliable and representative estimates of hypertension prevalence in Portugal. The quantitative analyses also provide an accurate summary of hypertension prevalence, giving valuable insights into sex and age-specific trends. We were able to examine trends in the prevalence of hypertension in Portugal over the past 20 years. Concerns have been raised regarding the criteria for hypertension used in population-based studies, particularly after comparing prevalence estimates based on hypertension obtained through SRH with those based on MH. Prevalence is likely to be underestimated when SRH is used as the chosen method for identifying cases. However, because blood pressure was measured during a single office visit, the prevalences of hypertension reported in studies measuring hypertension may be overestimated due to the white-coat effect. As a result, detailed sensitivity analyses were conducted.

It is important to acknowledge the study's limitations. While the overall sample size is robust, the number of studies included in some analyses was lower than the recommended minimum for meta-analysis36 (fewer than nine studies), which may affect the precision of certain subgroup analyses. Furthermore, heterogeneity statistics presented in Graph 1 suggest that true prevalence rates vary substantially across studies. Even though this level of heterogeneity is typical in prevalence meta-analyses due to differences in populations, methodologies, and settings among studies, some caution is required when interpreting the results presented in this study. Nevertheless, the Galbraith Plot (Supplementary Material SG1) suggests that the studies analyzed in this review are precise, especially because they have large sample sizes. Due to high precision, I2 becomes inflated even when results from different studies are homogeneous. Since I2 quantifies the fraction of the total variability caused by heterogeneity among studies, it can be inflated when within-study variance is low (low τ2), as it usually occurs when compiling studies with high precision. In such cases, even small values of τ2 can lead to a high I2. However, sensitivity analyses were conducted to consider the main sources of heterogeneity and mitigate this issue.